Introduction

Recently, perovskite solar cells (PSCs) have made an efficiency breakthrough beyond 26%, as is comparable to that of the silicon-based counterparts1,2,3. Despite the rapid evolution in efficiencies, further improvement in device performance is still required to make this technology more competitive. Currently, the primary challenges in enhancing the photovoltaic performance of PSCs are predominantly associated with severe non-radiative charge recombination losses at the interfaces between perovskite and charge transport layers4,5,6,7,8. Intensive efforts have been made to minimize non-radiative recombination at such interfaces including interface chemical passivation, energy level optimization, and 3D/2D heterojunction strategies9,10,11,12,13,14,15,16,17,18,19,20,21,22. It still remains a daunting challenge to achieve high performance in both normal and inverted PSCs since the charge recombination behavior is highly related to the local chemical environment.

Introducing a dipole interlayer formed by orderly arrangement of molecules at the perovskite/hole transport layer (HTL) interface has been proven to be an effective strategy to manipulate the interfacial energy level alignment23,24,25,26,27. The ordered electrical dipole moment modifies the vacuum level bending across the interface, leading to a shift in the energy levels at the perovskite/HTL interface28,29. This shift reduces the interfacial energy offset for hole transport, enhances the hole extraction, and meanwhile suppresses non-radiative charge recombination, ultimately improving the photovoltaic performance of PSCs26,30,31. However, the typical method to induce an ordered surface dipole moment requires a construction of a self-assembled monolayer (SAM), which can hardly form a full surface coverage and mostly induces a relative weak dipole moment32,33,34,35. Increasing the thickness of the SAM can enhance the interfacial dipole moment accumulation. However, an excessively thick layer may adversely impair the performance of PSCs36. As a result, the effect of traditional SAM strategies is limited to an unfavorable upper regime.

Fortunately, ferroelectric materials can enable a large dipole moment accumulation in the bulk owing to its spontaneous polarization, which has been demonstrated to play a dominant role in enhancing carrier separation and transfer dynamics28,37,38,39,40,41. Specifically, in molecular ferroelectric materials, the polarization orientations within the microstructures of domains can be altered by adjusting the external electric field directions. This alteration results in a decreased deficit in open-circuit voltage (VOC)38,42,43,44. Thus, improving the interfacial properties to realize efficient multi-structured PSCs via introducing a ferroelectric molecule with orientational polarization presents to be highly promising. Besides, the designable Lewis base/acid substituents in molecular ferroelectric may readily facilitate the regulation of the defect states45,46,47.

Herein, we explored the effect of ferroelectric SAMs, i.e., the 1-adamantanamine hydroiodide (ADAI), on performances of PSCs through combined theoretical and experimental efforts. We found that ADAI can form a dipole layer, and maintain its ferroelectric performance at the perovskite/HTL interface due to its strong chemical interactions with the perovskite film, which significantly improves the interfacial energy level alignment. The improved energy level alignment between perovskite and HTL mitigates the non-radiative charge recombination and enlarges the fill factor (FF) and VOC of PSCs. The corresponding Power Conversion Efficiencies (PCEs) of normal (n-i-p) and inverted (p-i-n) PSCs are promoted to 25.13% and 25.59%, respectively, exhibiting a great generality of our strategy. This work highlights the use of ferroelectric SAM for manipulating the spontaneous polarization and modifying carrier dynamics of perovskite, which is of great significance for highly efficient PSCs with enhanced operational stability.

Results

Surface properties

A ferroelectric molecule, 1-adamantanamine hydroiodide (ADAI), known for its self-assembling properties48, was used to modify the interfacial properties of PSCs, by constructing a dipole layer at the interface between perovskite and HTL. The ADAI demonstrates strong ferroelectric properties at temperatures as high as 373 K, as evidenced in prior research49. ADAI belongs to the orthorhombic crystal system, featuring a polar point group mm2 and ferroelectric space group Pmn21, showcasing pronounced polarity and exceptional ferroelectric characteristics. The electric ferroelectric effect is generated due to intrinsic disorder in the ferroelectric phase, leading to the formation of a permanent ferroelectric moment of 10.05 D. This basic information of ADAI was illustrated in Fig. 1a, b, and more detailed information about ADAI can be found in Supplementary Note 1. Furthermore, the ammonium group on this molecule can form strong chemical interaction with the perovskite, where the ammonium group will interact with the iodide via hydrogen bond (Fig. 1c). As a result, this molecule is selected in our work to explore the synergistic effect of SAM and ferroelectric on the interfacial optoelectronic properties of PSCs. After depositing it on the perovskite layer, dramatic ferroelectric property has been observed with the piezoelectric force microscopy (PFM) characterizations. As shown in Figs. 1d–g, S1 and S2, the ADAI-modified perovskite surface shows the typical ferroelectric domain structure, delivering the switchable ferroelectric property with 180° phase shift after switching, which could not be observed in the pure perovskite film (Fig. S3).

Fig. 1: Ferroelectric perovskite solar cells: from molecule to device.
Fig. 1: Ferroelectric perovskite solar cells: from molecule to device.
Full size image

a Molecular structure of ADAI. b Packing diagram of ADAI along the a-axis at 298 K. c Schematic diagram of the chemical interaction between ferroelectric SAM and PbI octahedron. d–f Surface properties of ferroelectric SAM on perovskite films. Out-of-plane (OP) PFM phase (d), amplitude (e), and topography (f) images showing the typical domain structure. g PFM switching spectroscopy showing piezoelectric phase (up) and amplitude (down) loops for a selected point indicated by black point marked in (e). h, i Surface potential map of perovskite film without and with molecular ferroelectric ADAI. j Schematic energy structure of perovskite/Spiro-OMeTAD (HTL) without and with molecular ferroelectric ADAI. Note, a rigid energy shift of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of the Spiro-OMeTAD was assumed caused by the change in EVac after ADAI modification.

Moreover, due to the ADAI modification, the domain structure of perovskite films is observed in the PFM phase image (Fig. 1d), and the corresponding domain walls (dark line contrast) can be clearly discerned in the amplitude image (Fig. 1e). By overlaying the PFM mappings on a film morphology (Figs. 1f and S2), it can be verified that the domain structure has trivial correlation with the sample topography, excluding the effect of morphological factors on the PFM measurement and reconfirming the formation of ferroelectric structure on perovskite layer with ADAI deposition. Additionally, the in-plane PFM amplitude and phase images of the ADAI-modified perovskite films have also been measured and are shown in Fig. S4, which also clearly shows a polydomain structure. Furthermore, a comparison of the out-of-plane and in-plane PFM images suggests that the orientation of polarization varies from grain to grain. Further, Fig. 1g shows the local PFM switching spectroscopy measured at the point marked by the sign of “black point” in Fig. 1e, revealing the ferroelectric polarization switching process of perovskite film with ADAI modification. Figure S3a–c shows a typical PFM imaging of the control film, and in sharp contrast, no ferroelectric domain structure is observed at all in these images. Polarization switching behavior was not observed by the PFM switching spectroscopy (Fig. S3d), either. To further visualize the polarization switching of ferroelectric domains, we performed local domain manipulation experiments before and after voltage pulses were applied (Fig. S5). We found that the ferroelectric polarization domain structure of perovskite film with ADAI modification was reversed after loading an external +20 V pulse for 1 s duration (Fig. S5). Again, all the above results prove the existence of ferroelectric response in perovskite films with ADAI modification.

Besides the apparent ferroelectric phenomenon with depositing the ADAI layer, it is susceptible to form a permanent dipole moment at the perovskite/HTL interface, which includes the ordered self-assembly of ADAI and the spontaneous polarization. To confirm this point, the Kelvin probe force microscopy (KPFM), the ultraviolet photoelectron spectroscopy (UPS, Figs. S6 and S7), and UV-vis absorption (Fig. S8) measurements were performed. Firstly, the surface electronic chemical potentials for the perovskite films with/without an ADAI ferroelectric layer were also measured using KPFM. We found the surface potential (SP) of the perovskite surface increases to 0.90 from 0.12 V due to the ADAI dipole layer, as displayed in Fig. 1h, i (more details in Fig. S9). The alignment of molecular dipoles is responsible for the increase in SP, and the higher SP also indicates a decreased work function30,50. Furthermore, the introduction of ADAI results in a more uniform SP distribution, facilitating advantageous conditions for local carrier transport. The SP variation across the perovskite/perovskite & ADAI interface, as depicted by the white dotted lines in Fig. S10b, intuitively reflects the significant change in SP upon ADAI incorporation. The substantial increase in SP at the interface provides compelling evidence for the successful formation of a heterojunction induced by the presence of ADAI. This distinct SP variation highlights the crucial role played by ADAI in modifying the interfacial properties of the perovskite, consequently influencing charge transfer dynamics and mitigating non-radiative recombination losses. The energy level alignments of perovskite films with/without ADAI ferroelectric layer are calculated and presented in Fig. 1j, the Evac of ADAI-modified film downward shifts 0.31 eV extracted from Fig. S6, reflecting a reduction in the work function and the formation of negative dipole capping layer6,30,51,52. Furthermore, we found that the energy mismatch Φb between the valence band maximum of the perovskite and the HOMO of Spiro-OMeTAD decreased from 0.47 to 0.01 eV after introducing ADAI, and this results in largely more aligned energy levels for hole transfer, fostering advantages for hole extraction and the suppression of non-radiative charge recombination25,53. Also, the VB of the perovskite film is increased by 0.35 eV, suggesting the formation of a band-bending structure, which blocks the electron by forming a back-scattering electric field. The modification of the perovskite film’s work function and energy levels by ADAI can be attributed to the dipole moment induced by its ferroelectric polarization. The resulting negative dipole capping layer alters the SP, leading to a more favorable energy level alignment with the HTL, ultimately enhancing device performance. The optical absorption spectra of the films with and without ADAI modification show similar band edge absorption at ~800 nm for all thin films. The band edges, determined from the Tauc plots (inset of Fig. S8), exhibit only a minor increase from 1.54 to 1.55 eV, indicating that the surface modification of ADAI has a minimal impact on the band structure of the bulk perovskite phase.

Fig. S11a, c shows top-view scanning electron microscopy (SEM) images of perovskite films before and after capping ADAI ferroelectric layer. After capping ADAI, the average grain sizes are enlarged and the ADAI capping layer is uniformly distributed upon the perovskite film, indicating the formation of a thin layer of ADAI on the surface. The surface-sensitive grazing-incidence wide-angle X-ray scattering measurements are also conducted to trace the surface morphology of perovskite films without/with an ADAI ferroelectric layer, as shown in Fig. S12. For the ADAI-treated perovskite films, the perovskite peaks remained in their positions while new signals that correspond to the ADAI appeared, suggesting that the self-assembled ADAI monolayer was successfully deposited on the surface of the perovskite film with good orientation, rather than forming any low-dimension perovskite structures, as reported in previous work47. From the cross-sectional image of the device, it is further demonstrated that the ADAI layer facilitates to improve the morphology of perovskite film, as shown by the well-connected grains with fewer grain boundaries (GBs) observed in the perpendicular direction after the introduction of the ADAI capping layer (Fig. S11). From X-ray diffraction patterns (Fig. S13), a significant enhancement in the (110) plane peak intensity at ~14° with a narrower full width at half maximum of 0.18° was observed for ADAI-modified films. These observations suggest that the ADAI-modified perovskite films exhibit enhanced crystallinity and improved structural quality, corroborating the findings from the SEM analysis.

Constrained by the limited penetration depth inherent in UPS and KPFM techniques (~ a few nanometers), we subsequently employed cross-sectional KPFM scanning to meticulously profile and scrutinize variations in electric potential distribution throughout the entire device (Fig. S14). Notably, two distinct potential steps manifested at the interfaces of the HTL-perovskite and perovskite-electron-transport layer (ETL), while no potential steps were observed within the perovskite layer. The perovskite-ETL junction exhibited comparable strength in both control and ADAI-modified cells, substantiating the identical nature of the perovskite-ETL interfaces in these two cells. Otherwise, the introduction of ADAI led to an enhanced variation in potential at the HTL-perovskite interface. The elevated variation in potential can be ascribed to the arrangement of ADAI within the upper stratum of the perovskite film, resulting in a desirable energy level alignment for efficient removal of photo-generated charge carriers with minimal recombination54,55.

To elucidate the chemical interactions between ADAI and the perovskite film, X-ray photoelectron spectroscopy (XPS) measurements were conducted. As shown in Fig. S15, after ADAI modification, the Pb 4f peaks, initially located at 138.32 and 143.19 eV in the control film, exhibited a red shift to 138.07 and 142.93 eV, respectively. Similarly, the I 3d5/2 and I 3d3/2 peaks shifted to lower binding energies by 0.34 and 0.33 eV, respectively. The observed redshifts in the binding energies of the Pb 4f and I 3d peaks indicate significant changes in the electrostatic environment surrounding the [PbI6]4− octahedron56. Specifically, the observed shift results from the contraction of the perovskite crystal lattice and the formation of ionic bonds between C-NH3+ and [PbI6]4− octahedron as the ADAI fills the A-site cation vacancies57,58. Consequently, the decrease in A-site cation vacancies aligns the perovskite film more closely with its optimal stoichiometry.

Additionally, analysis of the Pb 4f and I 3d core-level energy spectra reveals that the Pb:I ratio in the control perovskite film is ~1:1.43, suggesting a significant iodine deficiency at the surface that may result in considerable point defects. With the ADAI modification, the Pb:I ratio markedly increased to 1:3.22, implying that the I− ions in ADAI likely serve to passivate the iodine vacancies, thereby enhancing the stoichiometric balance in the perovskite surface composition59.

Furthermore, two prominent shoulder peaks are observed on the low binding energy side of the main peak, which are associated with metallic Pb0 on the pristine perovskite surface. The presence of Pb0, functioning as a deep-level trap site, facilitates non-radiative recombination, which is, however, eliminated after applying the ADAI for surface modification60,61. This can be ascribed to the iodine compensation from ADAI and also verifies its chemical passivation. Overall, the XPS results indicate that the ADAI capping layer on the surface of the perovskite film effectively reduces the defect density at the perovskite/HTL interface by filling the vacancies of A-site cations and halogens to reduce the carrier non-radiative recombination.

Photovoltaic performance

To explore the effect of the ADAI ferroelectric dipole layer on device performances, we fabricated PSC devices with a typical structure of glass/ITO/SnO2/perovskite/(ADAI)/Spiro-OMeTAD/Au (see more details in Experimental Section). The relationship between device performance and ADAI concentration (see Fig. 2a, Table 1, Fig. S16, and Supplementary Table 1) reveals a significant increase in PCE upon treating perovskite films with a low concentration of ADAI solution. Notably, the highest PCE was achieved at an ADAI concentration of 6 mg/mL. Under these optimal conditions, the device with the ADAI ferroelectric dipole layer exhibits an improved PCE of 25.13% compared to the control device, mainly attributed to VOC improvement from 1.13 to 1.19 V and JSC from 24.58 to 25.07 mA cm−2. Additionally, under these optimal conditions, the ADAI ferroelectric dipole layer exhibits a thickness of ~20 nm according to the results from the step-profiler. However, the device performance starts to decline at higher ADAI concentrations, probably due to the electrical insulating nature of ADAI. From the external quantum efficiency (EQE) spectra (Fig. S17), the integrated JSCs are 24.36 and 25.03 mA cm−2 for control and ADAI-modified devices, respectively, which are equivalent to the values acquired from the J–V curves. Besides, the statistical distribution of the device parameters collected from 20 devices is shown in Figs. S18 and S19 and the averaged PCE is 24.59% for target PSCs, which is 3.14% higher than that of the control devices. The stabilized power outputs at the maximum power point (MPP) for the PSCs are plotted in Fig. S20, showing that the PCEs of the related devices are stabilized at 25.02 % and 21.89% under continuous 1 equivalent sun illumination for 300 s with negligible fluctuation. To further examine the scalability of the target device, we up-scaled the device's active area to 1 cm2 and the resultant target device delivered the best PCE of 23.52%, showing a considerable improvement above the control cell (20.05%). We attribute the slight increase in JSC to the improved charge extraction with the reduced interfacial hole extraction barrier. The significant increase in VOC and FF can be attributed to the effective passivation of defects and GBs on the surface of perovskite films. This passivation process plays a crucial role in suppressing non-radiative charge recombination associated with defects, contributing to the improved performance of the perovskite films.

Fig. 2: The performance of the devices.
Fig. 2: The performance of the devices.
Full size image

a J–V curves of the champion control and ADAI-modified normal PSCs. b J–V curves of the champion control and ADAI-modified normal PSCs with an active area of 1 cm2. c J–V curves of the champion control and ADAI-modified inverted PSCs. d Operational stability of encapsulated normal PSCs under constant 1-sun illumination in air with 50% relative humidity.

Table 1 Summary of the normal device performances
Full size table

However, with the ADAI dipole layer at the cathode, we find the performance is reduced compared to the control device, with a VOC of 1.09 V, JSC of 24.70 mA cm−2, FF of 74.77%, and PCE of 20.15% for the champion device (Fig. S21). Considering that the direction of the dipole moment would be different at different sides of the perovskite layer, the deposition of ADAI at the bottom of perovskite film will play an opposite role compared to that on top surface, which well explains the reduced efficiency.

To further explore the working mechanism of ADAI, we applied it in an inverted PSC structure. The device architecture is glass/ITO/NiOX/MeO-4PACz/ADAI/perovskite/C60/BCP/Ag (Fig. S22). Typical current-voltage (J–V) curves for the inverted PSCs were measured in an active area of 0.06 cm2. The control PSC delivers a PCE of 23.02%, with VOC of 1.15 V, JSC of 24.78 mA cm−2, and FF of 80.66%. However, with an ADAI dipole layer underneath the perovskite layer, the device performance improves a lot, with a VOC of 1.20 V, a JSC of 25.95 mA cm−2, and an FF of 82.29% for a PCE of 25.59% (certified 25.36% in Chinese National Photovoltaic Industry Metrology and Testing Center, Fig. S23). The accuracy of the JSC derived from the J–V measurement is further confirmed by EQE results (Fig. S24). To our best knowledge, it is rare to realize over 25% efficiency in both normal and inverted PSCs with such a facial interface engineering. Further, the sharp contrast between the effect of the ADAI layer underneath or on top of the perovskite (Figs. S21 and S25) in either normal or inverted device structures further confirms the importance of ferroelectric polarization direction in causing the variation of device performance. Compared with the conventional devices, the inverted devices incorporating the ferroelectric interlayer demonstrate a more pronounced enhancement in JSC from 25.07 to 25.95 mA cm−2. This improvement can be attributed to the synergistic effect of enhanced carrier dynamics and film morphology, which will be discussed in detail in the following sections.

Considering that the passivation effect with the ammonium salt is typically observed in previous work, similar effect is also expected for ADAI surface modification62,63. To disentangle the effect of dipole moment and surface passivation, we studied the individual effect of dipole induction and passivation, via comparing it with butylamine iodide (BAI), decylamine iodide (DAI), and phenylethylamine iodide (PEAI) for perovskite top surface treatment. We find their influence on the device performance is inferior to ADAI, as shown in Fig. S26, which suggests that the role of dipole moment is more significant compared to the passivation effect, and the dipole moment from ferroelectric ADAI overwhelming that of the specific chemical interaction at the interface, i.e., ammonium-perovskite.

To further disentangle the effect of ferroelectricity from the molecular passivation effect, we also compared ADAI with adamantanamine hydrochloride (ADACl). From the PFM measurements shown in Fig. S27, there is no discernible piezo-response on the ADACl film, and this result is consistent with its centrosymmetric crystalline structure. The PSCs with ADACl modification deliver an increased VOC from 1.126 to 1.148 V, and the device efficiency increases from 22.5% to 23.3% (Fig. S28, Supplementary Table 2), as consistent with Liu’s work64. Considering that ADACl has no ferroelectricity effect, the enhancement in device performance is attributed to the passivation effect, which is associated with the amine group, rather than a ferroelectric dipole. Although the ADACl increases the device’s VOC and efficiency, the ADAI is much better in enhancing the device's performance (VOC of 1.19 V, and efficiency of 25.13%). The distinct comparison affirms the ferroelectricity-induced dipole plays a key role in suppressing the charge recombination and promoting the interfacial charge transfer.

For the normal structure, with ADAI capping on perovskite top surface, the devices exhibit a wonderful storage stability in the N2 glovebox (H2O < 10 ppm, temperature: 25 ± 5 °C) for 4000 h, retaining 96% of their initial PCE in contrast to only 80% of the control devices (Fig. S29). When being operated under continuous MPP tracking, the encapsulated target devices demonstrate 700 h for 90% of the efficiency maintenance. By contrast, the control PSCs decay to 90% of their initial PCE just after 150 h. The improved device stability with ADAI deposition may be attributed to the chemical passivation of ADAI (Fig. 2d). As supported by XPS characterization results, the chemical passivation of ADAI involves several key aspects, including the filling of A-site cation vacancies, the passivation of iodine vacancies, and the inhibition of metallic Pb0 formation.

Carrier dynamics

Further, we comprehensively studied the carrier dynamics in normal PSCs without or with the ADAI ferroelectric layer. According to previous report, the variation of VOC is related to the non-radiative charge recombination, which can be numerically characterized with the photoluminescence (PL) and electroluminescence (EL) measurements30. As shown in Fig. 3a, higher fluorescence intensity was observed in the steady-state PL spectra, indicative of a high-quality perovskite film with fewer defects and less non-radiative recombination65. Specifically, a five-fold increase in PL intensity was noted in the ADAI-modified perovskite film compared to the control film. This suggests that the ADAI ferroelectric layer significantly hinders the trap-assisted recombination process, reducing non-radiative losses and improving the quality of the perovskite film66. Besides, charge carrier lifetimes, determined by fitting time-resolved photoluminescence (TRPL) spectra (Fig. 3b), were notably extended in the perovskite film with the ADAI ferroelectric interlayer compared to the control. This provides further confirmation that the introduction of the ADAI ferroelectric interlayer effectively suppresses defect-induced non-radiative recombination processes67,68,69.

Fig. 3: Carrier dynamics characterization of the devices.
Fig. 3: Carrier dynamics characterization of the devices.
Full size image

a The steady states fluorescence spectroscopy of the control and ADAI-modified perovskite film on glass. b The time-resolved fluorescence spectroscopy of the control and ADAI-modified perovskite film on glass. c EQEEL spectra of the control and ADAI-modified devices operating in light-emitting diode mode. Dark J–V curves of the hole-only (d) control and (e) ADAI-modified devices. f The dynamic evolution of the bleaching recovery of the control and ADAI-modified perovskite films on glass. g Voc versus light-intensity curves of the control and ADAI-modified devices. h Jph–Veff curves of the control and ADAI-modified devices. i The photocurrent decay process of the control and ADAI-modified devices.

To elucidate the mechanism behind the enhanced photovoltaic performance of ADAI-based inverted PSCs, we also investigated the PL and TRPL spectra of perovskite films deposited on glass/(ADAI) and ITO/NiOx/(ADAI) substrates. Firstly, a glass/(ADAI)/perovskite structure was examined, with the excitation light directed towards the glass side. The results presented in Fig. S30 revealed that the ADAI-bottom-modified perovskite exhibits a higher PL intensity and longer carrier lifetime compared to the unmodified counterpart. These findings suggest that the incorporation of ADAI effectively passivates the buried interface defects, enhances the quality of the perovskite bottom surface, and suppresses non-radiative recombination within the perovskite films.

As for the structure of ITO/NiOx/(ADAI)/perovskite, when excited from the top surface, the perovskite film grown on NiOX/ADAI exhibited a higher PL intensity compared to the film on pristine NiOX (Fig. S31a), suggesting improved crystallinity and reduced trap density in the ADAI-modified perovskite film. The morphology of perovskites atop ADAI was further confirmed by SEM (Fig. S32). Upon excitation from the bottom surface near the HTL, the ADAI-modified perovskite film demonstrated more pronounced PL quenching (Fig. S31b), indicative of enhanced carrier extraction by ADAI compared to pure NiOx. TRPL analysis revealed a shorter decay lifetime for the perovskite at the NiOx/ADAI interface (89 ns) compared to that at the pure NiOx interface (106 ns) (Fig. S31c), corroborating the PL results and confirming the enhanced carrier extraction capability of the ADAI-modified perovskite film. Overall, ADAI enhances the interfacial properties of the perovskites fabricated atop and promotes the formation of high-quality perovskite layers. Consequently, this ensures better photovoltaic performance of inverted PSCs.

We also test the charge recombination properties at the real operation of PSCs, which also work as light-emitting diodes (Fig. 3c). The normalized EL spectra of PSCs with and without the ADAI ferroelectric layer are similar, whereas the EQEEL values of PSCs largely varied from 2.91% (Control) to 11.29% (with ADAI ferroelectric layer), confirming the mitigated non-radiative recombination and photovoltage loss upon introduction of ADAI dipole layer. The improved ELQE contributes to a reduction of non-radiative charge recombination of ~35 meV (detailed discussion in Supplementary Note 2), which is consistent with the variation in VOC.

Further, the trap densities are quantified by the space-charge-limited current measurements based on the hole-only device (ITO/NiOX/perovskite/Spiro-OMeTAD/Au) and electron-only device (ITO/SnO2/perovskite/PCBM/Ag). Notably, the hole-only and electron-only devices with the ADAI dipole layer exhibit a reduced trap density (Ntrap,hole = 0.7 × 1015 cm−3, Ntrap, electron = 3.7 × 1015 cm−3) than the control device (Ntrap,hole = 2.7 × 1015 cm−3 to Ntrap = 6.7 × 1015 cm−3) shown in Figs. 3d, e and S33 and Supplementary Note 3. Generally, the lower trap density, signifying fewer scattering centers for carriers and efficient charge transport, can result in an enhanced Voc and FF in ADAI-modified PSCs and thus improved photovoltaic performance.

To characterize the interfacial charge recombination and transfer, the ultrafast transient absorption (TA) measurements were performed on perovskite films with and without ADAI layers, based on the glass/perovskite and glass/perovskite/Spiro-OMeTAD structures (Figs. 3f and S34). The recovery kinetics of the transient bleaching signal were obtained by monitoring the photo-bleach at 780 nm as a function of delay time. As depicted in Fig. 3f, the ADAI-modified film exhibited a slower decay rate than the control, indicative of fewer trap states capturing or scattering carriers, further underscoring the efficacy of ADAI in reducing defect concentration70,71. Furthermore, Spiro-OMeTAD, as an HTL, was applied through spin-coating on the perovskite films to explore the impact of ferroelectric molecular surface modification on carrier extraction (Fig. S34). In contrast to the glass/perovskite structure, these modified structures displayed accelerated decay processes, revealing that a thin ADAI capping layer on the perovskite film surface contributes to efficient charge extraction30.

Light-intensity dependent optoelectronic response measurements were conducted to probe the charge recombination/extraction behavior for both PSCs with and without the ADAI ferroelectric layer (Figs. 3g and S35). The relationship between Voc and light intensity (Plight) is elucidated by Voc∝nkT/e ln(Plight), with k, T, and e representing the Boltzmann constant, temperature in Kelvin, and elementary charge, respectively. The calculated slope is 1.84 kBT/e for the control PSCs, decreasing to 1.41 kBT/e for the ADAI-modified perovskite. This decrease implies a significant suppression of monomolecular Shockley–Read–Hall recombination in the ADAI-modified PSCs72,73. Furthermore, the increased α values of PSCs with the ADAI dipole layer also indicate the enhanced charge transfer through the introduction of the ADAI ferroelectric layer (Fig. S23)74.

Fig. S36 illustrates the capacitance measurements. The Mott–Schottky analysis reveals that the ADAI with good ferroelectric properties plays a crucial role in reconfiguring the interfacial energy band structure, leading to an enhanced Vbi and consequently, improved efficiency in charge collection30,75. Besides, an improvement in the Vbi was also observed in the ADAI-modified inverted device, suggesting that a similar efficiency enhancement mechanism operates in both normal and inverted device architectures (Fig. S37). Furthermore, the photocurrent density (Jph)-effective internal voltage (Veff) measurement was carried out to study the charge collection efficiency of PSCs without/with the ADAI ferroelectric layer (Fig. 3h and Supplementary Note 5). For the PSCs with ADAI capping layer, the calculated exciton dissociation efficiencies (Pdisss) and the calculated charge collection efficiencies (Pcolls) are 99.6% and 93.3%, while 99.3% and 89.9% for the control device, which further confirmed the potential of ADAI ferroelectric layer in adjusting the carrier dynamics.

Transient photocurrent (TPC) and transient photovoltage (TPV) decay measurements are conducted to further study the effect of ADAI molecular layers on the photo-carrier dynamics. The TPC decays are examined under pulsed light at short-circuit conditions, and the photo-generated carriers should be swiftly transported and extracted by the electrode. As displayed in Fig. 3i, the carrier transfer time (τt) is accelerated from 1.05 to 0.56 μs with the introduction of ADAI ferroelectric layer, which indicates its better charge extraction efficiency, consistent with the TAS and light-intensity dependent JSC measurements32. This observation aligns closely with the observed increases in Jsc and FF in the devices. Besides, based on the TPV decay measurements (Fig. S38), the ADAI-treated device also exhibits a prolonged carrier recombination lifetime (τr) (4.95 μs) than the control devices (3.85 μs), confirming the prominently reduced defect density is consistent with previous results76.

First-principles calculations

First-principles calculations were conducted to investigate the mechanism of the ADAI surface modification at the molecular level. The schematic diagram is shown in Fig. S39. Three analog molecules, i.e., BAI, PEAI, and DAI, were selected for comparison, and the calculation details are provided in the Experimental Methods section in SI. Five atomic surface models of the FAPbI3 perovskite films were built: one pristine FAPbI3 slab, and the other four with adsorption of ADAI, BAI, PEAI, and DAI molecules, respectively (Fig. 4a). Stable configurations were obtained after surface relaxation, and the dipole state for ADAI molecule on the surface was defined by comparing diverse spatial orientations (Figs. S39S40); their electronic structures were then evaluated by computing the total density of states (DOS) spectra (Fig. 4b). It can be discovered that the adsorption of the ADAI on the FAPbI3 surface induced a localized surface state within the bandgap, close to the bottom of the valence band, in contrast to the bare surface without adsorption. This additional surface state could excite more electrons on the valence band, thereby enhancing the conduction of the FAPbI3 film. It should also be noted that although the BAI, PEAI, and DAI are polar molecules like ADAI, no surface state occurred between the valence and conduction bands when they are adsorbed. As a result, it can be concluded that the formation of the surface states is related to the strong dipole moment of the ADAI molecules anchored on the perovskite surface, as is further attributed to the intrinsic ferroelectric polarization of ADAI molecule and formation of a built-in electric field to facilitate charge transfer as shown in Figs. 4b and S4149. We also computed the DOS spectra of the FAPbI3 perovskite thin film with the surface adsorption of ultrathin ferroelectric oxide layer PbTiO3 that has been reported for the preparation of high-performance PSCs (Fig. S42)77. Interestingly, we found that the adsorption of inorganic ferroelectric PbTiO3 resulted in the same surface states as the adsorption of ADAI, which contributes to enhanced charge transfer and improved conduction of the FAPbI3. This finding further verifies the mechanism by which ADAI enhances device performance and highlights its potential as a viable alternative to traditional inorganic ferroelectric materials. To delve deeper into the impact of ADAI surface modification on the electrical properties, the changes in work function of the FAPbI3 film from adsorption of different molecules were calculated and compared. The comparative analysis, as depicted in Fig. S43, indicates that FAPbI3 perovskite films with ADAI molecule adsorption exhibit a more significant reduction (−1.00 eV) in work function as compared to the films adsorbed with the other three molecules. This observation aligns well with the experimental measurements, confirming the notable influence of ADAI on the work function of the FAPbI3 film.

Fig. 4: First-principles calculations of the FAPbI3 perovskite thin films with and without adsorbates.
Fig. 4: First-principles calculations of the FAPbI3 perovskite thin films with and without adsorbates.
Full size image

a The optimized atomic models of the FAPbI3 film surface with no adsorption, ADAI, BAI, PEAI, and DAI molecule adsorption, respectively. b The calculated total density of states spectra corresponding to surface models of (a). The black arrow marked the surface state in the bandgap. Schematic in the black box on the right shows the ferroelectric polarization displacement of ADAI unit cell.

Discussion

In summary, we have developed the strategy of using ferroelectric self-assembly interlayer, e.g., ADAI to modify the perovskite surface. The deposition of ADAI on perovskite form strong chemical interactions to induce the oriented monomolecular dipole moment, and moreover, allow the propagation of the dipole moment over a large thickness due to the ferroelectric nature. Coupled with the photovoltaic device, the strong electric dipole of ADAI layer improves the alignment of the device energy levels and promotes carrier extraction. As a result, the champion device based on normal solar cell with ADAI modification delivers a significantly improved PCE of 25.13% on an active area of 0.06 cm2 and 23.5% over a larger area of 1 cm2, compared to that without any modification. Moreover, we also reported the ADAI-modified inverted device with improved PCE up to 25.59% on an active area of 0.06 cm2 (certified PCE of 25.36%), among the highest inverted PSCs processed with two-step processing. In addition, the ADAI-modified device aging in the glovebox or continuous illumination conditions shows excellent long-term stability. Therefore, this study showcases a facial way of optimizing the interface and presents a promising platform for high-performance PSCs toward commercialization.

Methods

Materials

All commercial materials were used as received without further purification. NiOx, Lead iodide (PbI2), 1-adamantanamine hydroiodide (ADAI), and formamidinium iodide (FAI) were purchased from MaiTuoWei Co., Ltd. Methylammonium iodide (MAI), BAI, methylammonium chloride (MACl), Bis(trifluoromethane)sulfonimide lithium salt (LiTFSI) and 4-tert-butylpyridine were all purchased from Xi’an Yuri Solar Co., Ltd. 2,2′,7,7′- tetrakis-(N, N-di-4-methoxyphenylamino)-9,9′spirobifluorene (Spiro-MeOTAD) was purchased from Youxuan. Dimethylformamide (DMF), dimethyl sulfoxide (DMSO), isopropanol alcohol (IPA), and chlorobenzene (CB) were purchased from Sigma Aldrich. SnO2 colloid precursor (tin (iv) oxide, 15% in H2O colloidal dispersion) was purchased from Alfa Aesar. ITO substrate was purchased from Hua Nan Xiang Cheng Co., Ltd.

Regular PSCs fabrication

Patterned ITO glass was cleaned by sequential ultra-sonication with detergent, deionized water, acetone, and isopropanol (IPA), respectively. Before use, the ITO substrates were further cleaned with ultraviolet ozone for 30 min. Then the ETL was formed by deposing a thin layer of SnO2 colloids (3%, diluted by water) at 3000 rpm for 30 s, and subsequently annealed at 150 °C for 30 min in ambient air. Besides, for better surface wetting, the substrate was cleaned with ultraviolet ozone for 10 min.

The perovskite film FA1-xMAxPbI3 was deposited using a two-step spin-coating method. For depositing of the PbI2 films, 1.5 M of PbI2 in mixed solvents of DMF and DMSO with a volume ratio of 9:1 was spin-coated onto SnO2 at 1500 rpm for 30 s, followed by annealed at 70 °C for 1 min in N2 atmosphere, and then cooled to room temperature. For the second step, a solution containing FAI (90 mg), MAI (6.39 mg), MACl (9 mg), and isopropanol (IPA) (1 ml) was spun on the PbI2 film at 2000 rpm for 30 s in the N2 atmosphere, then annealed at 150 °C for 15 min in an ambient air atmosphere (40–50% humidity). For the ADAI passivation layer, the ADAI /isopropanol solution with different concentrations was spin-coated onto the FAMAPbI3 perovskite surface (5000 rpm, 30 s), followed by annealing at 100 °C for 3 min.

After the preparation of perovskite films, the Spiro-OMeTAD solution, which consisted of 72.3 mg Spiro-OMeTAD, 17.5 μL bis(trifluoromethane) sulfonimide lithium salt (LiTFSI) stock solution (520 mg LiTFSI in 1 ml acetonitrile), 28.8 μL 4-tertbutylpyridine, and 1 ml CB, was deposited on top of the perovskite layer at a spin rate of 3000 rpm as the HTL. Finally, a 60 nm thick Au electrode was thermally evaporated to complete the fabrication of small devices.

Inverted PSCs fabrication

Patterned ITO glass was cleaned by sequential ultra-sonication with detergent, deionized water, acetone, isopropanol (IPA), respectively. Before use, the ITO substrates were further cleaned with ultraviolet ozone for 30 min. Then the HTL was formed by deposing a thin layer of NiOx colloids (20 mg/ml, dissolved by water) at 4000 rpm for 30 s, and subsequently annealed at 115 °C for 15 min in ambient air. Then, a thin layer of Me-4PACZ (0.5 mg/ml, dissolved by water) was spined on the NiOx layer at 4000 rpm for 30 s and subsequently annealed at 100 °C for 10 min in N2 glovebox. For the ADAI modification layer, the ADAI /isopropanol solution with different concentrations was spin-coated onto the HTL surface (5000 rpm, 30 s), followed by annealing at 100 °C for 3 min.

The perovskite film FA1-xMAxPbI3 was deposited using a two-step spin-coating method. For depositing of the PbI2 films, 1.5 M of PbI2 in mixed solvents of DMF and DMSO with a volume ratio of 9:1 was spin-coated at 1500 rpm for 30 s, followed by annealed at 70 °C for 1 min in N2 atmosphere, and then cooled to room temperature. For the second step, a solution containing FAI (90 mg), MAI (6.39 mg), MACl (9 mg), and isopropanol (IPA) (1 ml) was spun on the PbI2 film at 2000 rpm for 30 s in the N2 atmosphere, then annealed at 150 °C for 15 min in an ambient air atmosphere (40–50% humidity). For ETL, 25 nm C60 was thermally evaporated on the perovskite films, followed by 5 nm BCP evaporation as a hole-blocking layer. Finally, 100 nm Ag electrode was evaporated to complete the fabrication of small devices.

Reporting summary

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