Introduction

Perovskite-based tandem solar cells have attracted considerable attention in recent years due to their higher theoretical efficiencies compared to single-junction counterparts. Recently, advancements have reinforced this promise, achieving record efficiencies of 34.8% for perovskite/silicon tandem solar cells, 30.1% for perovskite/perovskite tandem cells, and 26.4% for perovskite/organic tandem cells1,2. These breakthroughs underscore the critical role of high-performance wide-bandgap (WBG) perovskites in both laboratory research and industrial applications. Significant progress in WBG perovskites has been driven by additive and interface engineering3. For instance, the employment of additives, such as rubidium thiocyanate, has achieved a power conversion efficiency (PCE) of 24.30% in a 1.66 eV perovskite4; furthermore, tailored two-dimensional perovskite layers were introduced as passivation layers in 1.77 eV perovskites, demonstrating a PCE of 20.5%5. However, these methods rely on solution processes with the usage of low-volatile and toxic solvents, such as N, N-dimethylformamide (DMF) and dimethyl sulfoxide (DMSO), posing scalability challenges for industrial production and significant safety concerns6,7,8,9. Addressing these limitations is vital for the sustainable and scalable fabrication of WBG perovskites, enabling their broader adoption in high-performance tandem solar cell technologies.

Vacuum-assisted hybrid deposition for WBG perovskites, where the inorganic parts are evaporated, followed by solution-processing of organic parts, has been widely recognized for its advantages. This method leverages the advantages of the evaporation method, which is convenient for scale-up and conformal growth. Additionally, the solution process is employed in the second step to enable high-quality perovskite, while avoiding the use of toxic solvents. Recently, many advances have been made in this field, demonstrating it as an effective method in fabricating tandem devices. To regulate the perovskite crystallization process in the hybrid-deposited perovskite film, Chin et al. introduced 2,3,4,5,6-pentafluorobenzylphosphonic acid in the hybrid deposition processing, which alleviated recombination loss in the film and achieved a certified PCE of 31.25% in perovskite/silicon tandem device10. To conduct strain management in hybrid deposited ~1.68 eV perovskite, an all-inorganic perovskite (CsPbCl3) buffer layer was incorporated at the buried interface11. Up to now, the record PCE of hybrid-deposited perovskite/Si tandem has surpassed 32%11. Moreover, an impressive 26.3% efficiency has been reported for the large area (aperture area of 16 cm2) hybrid-deposited perovskite/Si tandem device12, highlighting the scalability of this method.

However, most research about hybrid-deposited perovskite mainly focused on the bandgap of ~1.7 eV, which is aiming to provide perovskite subcells for perovskite/silicon tandem solar cells. As for the further widened bandgap (~1.8 eV) perovskite that is adapted as a subcell in perovskite/organic or perovskite/perovskite tandem device, the vacuum-assisted hybrid deposition is rarely explored. Moreover, the deposition process for wide bandgap (>1.8 eV) perovskites remains heavily reliant on solution-based methods, with limited research focused on developing scalable and efficient ~1.8 eV perovskites13. This lack of scalable pathways for ~1.8 eV perovskites poses a significant challenge to the advancement of perovskite/organic and perovskite/perovskite tandem solar cells.

To address these challenges, we introduced alkylammonium salt (n-propylamine hydrochloride, PACl) to regulate the crystal growth of WBG perovskites. Lead halide-based inorganic materials were co-deposited onto the substrates via thermal evaporation to create an inorganic template layer (ITL). Subsequently, organic cations were dissolved in ethanol and spin-coated onto the ITL, facilitating the formation of the photoactive perovskite phase. In-situ synchrotron grazing-incidence wide-angle X-ray scattering (GIWAXS) confirmed that PACl contributes to enhanced face-on stacking of the (100) crystal plane. The nanoscale transient photovoltage (TPV) and transient photocurrent (TPC) mapping indicated that the proportion of carriers with extended diffusion length increased by over 20%. Along with the reduced defect density and improved carrier transfer of perovskite, we obtained an efficiency of 17.48% and an open-circuit voltage (Voc) of 1.315 V. The 1.84 eV perovskite was employed as a front cell and integrated with an organic subcell to develop two-terminal (2 T) perovskite/organic tandem solar cells. The tandem device achieved a maximum efficiency of 26.46%, with a certified record efficiency of 25.82% over an active area of 0.05 cm2. Furthermore, it retained 90% of its initial efficiency during 400 h of maximum power point (MPP) tracking under ISOS-L-1 protocols.

Results and discussion

To achieve a perovskite film with a bandgap of around 1.8 eV, a high ratio of bromine should be introduced into the perovskite to increase the bandgap14,15. Here, we selected a three-source combination of PbI2, PbBr2 and CsBr to prepare the ITL (Fig. 1a, details about the evaporation method are shown in the method section). The ITL was deposited on the substrates, and its thickness was determined by scanning electron microscopy (SEM) to be around 160 nm (Supplementary Fig. 1a). After that, the organic cations, which were dissolved in ethanol, were spin-coated on the ITL. The wet film was pre-annealed at 70 °C inside the glovebox to facilitate solvent evaporation and then annealed at 120 °C in ambient air to form the perovskite phase. To regulate the perovskite crystallization process, cation solutions with and without PACl (10 mol%) were prepared and studied, referred to as target and control, respectively. SEM measurements were carried out to investigate the effect of PACl on film morphologies. Cross-sectional SEM analysis revealed that both the control and target perovskite films had a thickness of ~340 nm. However, the target perovskite films containing PACl exhibited enhanced grain quality, characterized by fewer grain boundaries in the vertical direction compared to the control films. This indicates a significant improvement in the crystallinity of the perovskite material (Fig. 1b, c). Both control and target conditions delivered dense perovskite films with a surface grain size of about 200–300 nm (Supplementary Fig. 1b, c).

Fig. 1: Hybrid deposited WBG perovskite films.
figure 1

a Schematic of the fabrication process of hybrid-deposited WBG perovskite, cross-section SEM images of b control and c target WBG perovskite film. In-situ synchrotron GIWAXS of (100) plane of d control and e target WBG perovskite film during the annealing process, GIWAXS of f control and g target WBG perovskite film after the annealing process. h the calculated binding energy of PA molecule on (100) and (110) plane, i schematic of crystal growth during annealing.

Full size image

XRD measurements were conducted to verify the perovskite phase transition. Both the control and target perovskite films displayed ideal perovskite phase diffraction patterns (Supplementary Fig. 2a, b). Using the standard perovskite diffraction pattern16, we calculated the texture coefficient (TC) of the perovskite films (Supplementary Fig. 2c, d)17. The TC (100) + TC (200) for the target perovskite was determined to be 2.44, higher than the 2.32 observed in the control perovskite. This result suggests that PACl enhances the (100) and (200) planes compared to the control film (Supplementary Fig. 2). To gain deeper insights into how PACl influences the crystallization process of perovskite, in-situ synchrotron GIWAXS measurements were further carried out (Fig. 1d–g). Different from the control film, which always kept a disordered (100) plane during annealing, the perovskite film with PACl gradually achieved a highly oriented (100) plane with the face-on direction after the annealing process. The highly oriented face-on stacking film will reduce the formation of defects and facilitate carrier transport18,19. In addition, 1H nuclear magnetic resonance (NMR) was used to trace the presence of PACl at different stages. The 1H NMR of pure PACl was shown in Supplementary Fig. 3a20. When PACl was introduced into the perovskite, three distinct alkyl hydrogen signals (one methyl and two methylene) were detectable in the film after spin-coating. These signals persisted after pre-annealing at 70 °C but disappeared following annealing at 120 °C, indicating that PACl escaped from the film during high-temperature annealing (Supplementary Fig. 3b). To further explore the mechanism of PA molecules in the process of film during the annealing process. Density functional theory (DFT) calculations were carried out to explore the interaction between PACl and perovskite (Supplementary Data 1, 2). As shown in Fig. 1h, the binding energy of PA on (100) plane is higher than that on (110) plane, indicating a more stable interaction between PA and (100) plane18. Therefore, the occupation of PA on the (100) planes makes it more likely to be exposed during the crystallization process with ordered face-on stacking.

Combining the experimental and theoretical results, we conclude that the PACl molecules regulate the perovskite crystallization during the annealing process. During this stage, the rapid crystal ripening process occurs, leading to disordered grains due to the uncontrolled coalescence of different nuclei. However, with the addition of PACl, the (100) plane is occupied by the PACl molecule, rendering the coalescence process of (100) plane. Moreover, the existence of PACl on the (100) plane makes it easier to align horizontally with the substrate18. In summary, PACl promotes the dominant growth and face-on stacking of the (100) plane, subsequently vaporizing during the annealing process. The schematic of crystal growth during annealing is shown in Fig. 1i.

After gaining insight into the growth mechanism of the perovskite, photoluminescence (PL) measurement was carried out to characterize the quality of the perovskite film. As shown in Fig. 2a, the PL emission peaks of both control and target film were around 675 nm, indicating the bandgap of perovskite was around 1.84 eV. The target film showed enhanced PL intensity than that of the control. Confocal PL gave the same trend of intensity, and the target film yielded better PL uniformity (Fig. 2b). Time-resolved PL (TRPL) results show that target film exhibits better carrier lifetime when compared to control film (Fig. 2c). The enhanced uniform PL intensity and extended lifetime indicate a reduction of non-radiative recombination sites in the target film18,21, which aligns with the decrease in defect state density from 0.86 × 1015 cm−3 for the control (without PACl) film to 0.65 × 1015 cm−3 for the target (with PACl) film (Supplementary Fig. 4)22,23.

Fig. 2: Characterization of WBG perovskite films.
figure 2

a PL spectra, b confocal PL mapping, and c TRPL of control and target perovskite film, AFM topography of d control and g target perovskite film and corresponding TP-AFM resolved mapping areas (100 × 100 nm2, red boxes marked), TPV (charge-carrier recombination lifetime, τr) mapping of e control and h target perovskite film. Diffusion length (LD) mapping of f control and i target perovskite film (The data analysis areas for TPV and LD correspond to the red boxes area marked in the AFM).

Full size image

To further clarify the effect of crystal improvement on carrier dynamics, the transient photo-response AFM (TP-AFM) was applied to perform transient photovoltage (TPV) and transient photocurrent (TPC) to implement nanoscale photo dynamics mapping24,25 (Fig. 2d–i and Supplementary Fig. 5). Firstly, AFM measurements were performed on the control and target film respectively, and then a region of 100 ×  100 nm2, including both crystals and grain boundaries, was selected for scanning to obtain TPV and TPC results. The target film shows an increased recombination lifetime (τr) in both crystal and grain boundary regions than the control film. In detail, the statistical results show that more than 86% of τr in the control film is below 0.36 µs, while only 18.5% of τr is below 0.36 µs in the target film (Supplementary Fig. 6a, b). Meanwhile, target film shows a reduced transport time (τt), over 54.2% of τt were lower than 0.34 µs, and the ratio is higher than that of the control (46.8%) (Supplementary Fig. 6c, d). With improved recombination lifetime and transport time, the proportion of carriers with extended diffusion length (LD) increased by over 20%. More than 66.8% of LD reached 233 nm in the target film compared to 43.7% for the control film (Supplementary Fig. 6e, f). Nanoscale photo dynamics mapping further determined that PACl acted as a positive role in improving perovskite carrier transport, and the boost could be attributed to the high crystallization and enhanced orientation of perovskite. We further assessed film stability in an air environment. As shown in Supplementary Fig. 7, significant PbI2 appears after 2 h (20 °C, 50% RH) in the control film, with δ phase emerging after 10 h. In contrast, the target film shows only minor PbI2 at 10 h and δ phase after 35 h, indicating that improved orientation and crystallization delay the perovskite degradation process in air26.

With encouragement from enhanced orientation of perovskite, improved carrier transport, and lower density state, we fabricated single-junction WBG perovskite solar cells. The device structure was Glass/ITO/Me-4PACz/Perovskite/C60/SnOx/Ag. The target device with 10 mol% PACl addition achieves a champion power conversion efficiency (PCE) at 17.48% under reverse scan (RS) with 1.315 V of Voc, 82.33% of fill factor (FF) and 16.14 mA/cm2 of short-circuit current density (Jsc), which is better than the control device which obtains an efficiency of 16.64% with 1.284 V of Voc, 80.73% of FF and 16.05 mA/cm2 of Jsc (Fig. 3a). Both devices showed low hysteresis, as shown in Supplementary Fig. 8. Maximum power point tracking (MPPT) was carried out to verify the stable efficiency of devices. As shown in Fig. 3b, the control and target devices maintained stable efficiency at AM 1.5 G light intensity within 600 s, which was 16.1 and 17.0% respectively. EQE measurement shows that both devices exhibit good visible light response. The control and target devices obtained integrated current densities of 15.72 and 15.85 mA/cm2, respectively, which confirms the reliability of the J-V measured current density (Fig. 3c). By differentiating the EQE curve in the first order, we determined that the bandgap of each perovskite was maintained at 1.84 eV (Supplementary Fig. 9), consistent with the PL results. Further, high-resolution EQE (HR-EQE) measurements were performed to calculate Urbach energy (EU), the EU of the target device is 19.59 meV, which is lower than the 20.15 meV of the control device (Supplementary Fig. 10). This indicates the electronic quality and energy order are improved around the sub-bandgap region23. Devices with different concentrations of PACl also were fabricated to clarify the optimal conditions of PACl (Fig. 3d–g and Supplementary Fig. 11). Based on the parameter statistics of single-junction devices, the efficiency improvement of the target device mainly comes from Voc and FF, consistent with the improved carrier dynamics and defect property.

Fig. 3: Single-junction WBG perovskite solar cells.
figure 3

a J-V scans of champion control and target perovskite solar cells (active area = 0.05 cm2), b 600 s MPP tracking (AM 1.5 G) of control and target perovskite solar cells, c EQE curves and integrated Jsc of control and target perovskite solar cells, statistical distributions of device parameters for control and target perovskite solar cells, d PCE, e Voc, f FF, and g Jsc (16 devices for each. The data points are marked with circles).

Full size image

With the improved performance of 1.84 eV perovskite, we applied it as a subcell in tandem solar cells. Considering the current matching of the tandem device13,27, the 1.84 eV perovskite was integrated with an organic subcell in the tandem solar cells. Before that, we first completed a single-junction organic solar cell with PM6:Y18 as the donor and acceptor28,29 (Supplementary Fig. 12). Supplementary Fig. 13 shows the statistical parameters based on 16 single-junction organic solar cells. The champion cell achieved an efficiency of 17.30% with 0.84 V of Voc, 74.30% of FF and 27.73 mA/cm2 of Jsc (Supplementary Fig. 14a). The EQE curve shows the optical response range of the organic solar cell is 300–1000 nm and the integrated current reaches 27.15 mA/cm2 (Supplementary Fig. 14b). Further, we calculated the first derivative of EQE to determine the bandgap of the organic subcell is 1.38 eV (Supplementary Fig. 14c). Then, the perovskite/organic tandem solar cells with architecture of ITO/Me-4PACz/Perovskite/C60/SnOx/ITO/MoOx/Organic/PDINN/Ag were fabricated. Figure 4a shows the cross-sectional SEM image of a perovskite/organic tandem device. The thickness of the perovskite active layer and the organic active layer is 340 and 100 nm, respectively. The champion device of the perovskite/organic tandem cells achieve an efficiency of 26.46% with 2.120 V of Voc, 82.08% of FF and 15.21 mA/cm2 of Jsc (Fig. 4b). Based on the efficiency statistics of 16 tandem devices, it shows a normal distribution at 25.91% (Fig. 4c). EQE measurement of perovskite/organic tandem cells indicates the integrated Jsc for the perovskite and organic subcell are 14.97 and 14.93 mA/cm2 respectively, which shows good current matching and consistent with J-V measured current density (Fig. 4d). As shown in Fig. 4e, the certified device obtained an efficiency of 25.82% with 2.125 V of Voc, 80.87% of FF and 15.03 mA/cm2 of Jsc. (The original certification report is shown in Supplementary Fig. 15). We compiled the reported efficiencies of perovskite/organic tandem solar cells over the past three years2,23,27,30,31,32,33,34,35,36,37,38,39,40,41,42 (Fig. 4f and Supplementary Table 1). It shows that our result represents one of the record-certified efficiencies among reported perovskite/organic tandem devices. We also performed long-term MPPT for the tandem solar cells following the ISOS-L-1 protocols. After 400 h of measurement, the tandem device maintained 90% of its initial efficiency (Fig. 4g), demonstrating strong stability under operational conditions41,42.

Fig. 4: Perovskite/organic tandem solar cells.
figure 4

a Cross-sectional SEM image of perovskite/organic tandem cells, b J-V curves of champion perovskite/organic tandem solar cell measured in the laboratory (active area = 0.05 cm2), c statistical distributions of efficiency of perovskite/organic tandem solar cells with 16 devices, d EQE curves and integrated Jsc for perovskite and organic subcell in tandem device, respectively, e J-V curves of the device certified by SIMIT, f summary of the reported efficiency of perovskite/organic tandem cells in recent 3 years, g long-term MPP tracking of perovskite/organic tandem cells.

Full size image

In conclusion, we successfully implemented a hybrid deposition method for 1.84 eV perovskite, offering a sustainable solution for the green fabrication of WBG perovskites. The alkylammonium salt (PACl) effectively regulated the interactions between cations and lead halides, enabling the dominant face-on stacking growth of the (100) crystal plane. This approach resulted in single-junction 1.84 eV perovskite solar cells with an efficiency of 17.48%. When incorporated into perovskite/organic tandem solar cells as the front cell, it achieved a power conversion efficiency of 26.46% (certified at 25.82%), representing one of the record efficiencies for perovskite/organic tandem devices. The tandem device maintains 90% initial efficiency after 400 h MPPT under ISOS-L-1 protocols. This provides an effective design idea for the future scale-up development of perovskite-based tandem solar cells.

Methods

Materials

Anhydrous isopropanol (IPA), chloroform, methanol and ethanol were purchased from Sigma-Aldrich. Lead iodine (PbI2) and cesium bromide (CsBr) were purchased from Alfa Aesar. Lead (II) bromide (PbBr2), 4-(3,6-dimethyl-9H-carbazol-9-yl) butyl] phosphonic acid (Me-4PACz), 1-chloronaphthalene and propylamine hydrochloride (PACl) were purchased from Tokyo Chemical Industry. Formamidinium iodide (FAI), methylammonium chloride (MACl), and ethane-1,2-diammonium iodide (EDAI) were purchased from Greatcell Solar Materials. C60 was purchased from Luminescence Technology. PM6, Y18 and PDINN were purchased from Solarmer Material Inc. Lithium fluoride (LiF) was purchased from Kurt J. Lesker.

Single-junction hybrid-deposited WBG perovskite device fabrication

Indium tin oxide (ITO)-covered glass substrates (2.5 cm × 2.5 cm) were cleaned by sonication with deionized water, acetone, and isopropanol for 10 min each. Then the substrates were dried with nitrogen flow and then treated with ultraviolet ozone (UVO) for 5 min. After that, Me-4PACz (1.0 mg/mL in IPA) was spin-coated on ITO substrates at 3000 rpm for 30 s in a nitrogen glovebox, followed by annealing at 100 °C for 5 min.

Then, the substrates were transferred into a vacuum chamber for the co-evaporation of PbI2, PbBr2 and CsBr inorganic layer. The PbI2, PbBr2, and CsBr were placed into three separate crucibles, and each source has a dedicated quartz crystal monitor (QCM) sensor. The temperature near the substrate was kept at room temperature (25 ± 2 °C) during the evaporation process. The substrate holder is about 45 cm away from the source. The source temperatures of PbI2, PbBr2, and CsBr were set to 370 ± 5 °C, 360 ± 5 °C, and 510 ± 5 °C, respectively, to achieve a rate ratio of 1:1.2:0.1. Another QCM sensor was set near the substrate holder to monitor deposition rate near the substrate. The process ended when the thickness of the inorganic layer on the substrate accumulated to 160 nm.

Following, the substrates were taken out to spin-coat the cation solution on them. For the control condition, 100 µL of a mixture solution of 0.3 mmol FAI and 0.06 mmol MACl (dissolved in 1 mL ethanol) were spin-coated on the ITL at 3000 rpm for 30 s in glovebox. For the target condition, another 0.03 mmol (10 mol%) PACl was mixed with FAI and MACl when preparing the cation solution. Then the film was annealed at 70 °C for 1 min in a glovebox and then transferred to ambient air (35–65% RH) and annealing at 120 °C for 8 min. The passivation was achieved by spin-coating EDAI (0.5 mg/mL in IPA) at 3000 rpm for 30 s in a glovebox and annealing at 100 °C for 5 min. Then, 20 nm of C60 was thermally evaporated at a rate of 0.2 A/s to build electron transport layer. Another 20 nm tin oxide (SnOx) was deposited in the atomic layer deposition (ALD) system at 100 °C by using precursors of tetrakis(dimethylamino) tin(iv) (TDMASn) and deionized water. During the process, the TDMASn was held at 70 °C and water (H2O) was held at room temperature. The deposition setting of TDMASn/Purge 1/H2O/Purge 2 were 0.4 s/12 s/0.15 s/11 s, 20 nm of SnOx was formed after 120 cycles. Finally, 80 nm of silver (Ag) was deposited by thermal evaporation at the rate of 2.0 A/s to build the electrode.

Organic single-junction device fabrication

About 30 nm MoOx was thermally evaporated on top of ITO and the substrates were brought back to the nitrogen-filled glovebox where PM6:Y18 solution (D/A = 1:1.2 in weight, with a PM6 concentration of 8 mg/mL and 0.5% 1-chloronaphthalene (v/v) as additive in chloroform) was spin-coated at 3000 rpm for 30 s and annealed at 100 °C for 5 min. Next, PDINN was deposited by spin-coating its 1 mg/mL solution in methanol at 3000 rpm for 30 s. Finally, the film was transferred to the thermal evaporator system to deposit 80 nm of Ag as the top electrode to finish the tandem device fabrication process.

Perovskite/organic tandem device fabrication

After completing deposition of SnOx, the substrates were transferred to the sputter system, where ~5 nm ITO was sputtered. After then, 30 nm MoOx was thermally evaporated on top of ITO and the substrates were brought back to the nitrogen-filled glovebox where PM6:Y18 solution (D/A = 1:1.2 in weight, with a PM6 concentration of 8 mg/mL and 0.5% 1-chloronaphthalene (v/v) as additive in chloroform) was spin-coated at 3000 r.p.m. for 30 s and annealed at 100 °C for 5 min. Next, PDINN was deposited by spin-coating its 1 mg/mL solution in methanol at 3000 rpm for 30 s. Finally, the film was transferred to the thermal evaporator system to deposit 80 nm of Ag as the top electrode to finish the tandem device fabrication process. To reduce the incident light loss of the tandem device, we evaporated 90 nm of LiF on the glass side as an anti-reflection layer.

Material characterizations

SEM was characterized by Hitachi Regulus 8230 at 2 kV accelerating voltage under SE mode. 1H NMR was measured by a 400 MHz Bruker. For the sample preparation of 1H NMR: 100 μL of DMSO-d6 was dripped onto four substrates respectively for each stage, then recycled the DMSO-d6 from the four substrates together. The solution collected from each stage was used for ¹H NMR measurement. XRD is recorded on an X-ray diffractometer (Malvern Panalytical) with Cu Kα radiation (λ = 0.1542 nm). PL spectra were recorded on a Renishaw Raman and micro-PL system. TRPL measurements were conducted on a PicoQuant FluoTime 300 spectrometer. Confocal PL mapping was measured on a Nikon A1 confocal microscope equipped with a 633 nm continuous-wave laser. Synchrotron GIWAXS patterns were recorded at the NCD-SWEET beamline at the ALBA synchrotron light source (Spain): an in vacuo undulator gap was optimized to produce an X-ray beam of 12.95 keV, which was further monochromatized using a Si (111) channel-cut monochromator. An array of Be lenses was employed to collimate the beam, obtaining a final beam size of 30 × 80 µm2 (V × H) at the sample position. The 2D patterns were collected using a Rayonix LX255-HS with a binning mode of 2 × 2, resulting in a pixel map of 2880 × 960 (V × H) pixels with a pixel size of 88.54 × 88.54 µm2. The reciprocal space |q| was calibrated using Cr2O3 from NIST as a calibration standard, obtaining a sample to detector distance of 210 mm. The sample was measured at different incident angles below 1° in a scanning fashion to optimize the recorded signal. The sample temperature was controlled using a Linkam stage. During the in-situ measurements, the sample was under continuous N2 gas flow for 1 min and then turned to air flow for 8 min. Data were processed using a home-made Python routine based on pyFAI. For nanoscale-resolved TPV/TPC mapping, an Agilent 5500 conductive atomic force microscope was used to measure local TPV and TPC decays in contact mode. TPV and TPC scans were obtained using a Pt/Cr coated silicon conductive tip (Budget Sensors with a force constant of 0.20 N/m and a resonance frequency of 14 kHz) at different locations on the perovskite film. Transient TPV and TPC data were acquired by an Agilent MSOX4154A oscilloscope. Charge carriers were generated using a green laser pulse (model: MGL-I-532 DPSS) at 532 nm wavelength. To receive the transient signal from the AFM controller, a Breakout Box (model: Agilent N9447A) was connected to the oscilloscope. Oscilloscope input impedances of 1.0 MΩ for TPV and 50 Ω for TPC were used for collecting transient decays. A mono-exponential decay model was used to calculate the carrier τr and τt of the film:

$$Y\left(t\right)={y}_{0}+{Y}_{0}{e}^{-(t-{t}_{0})/\tau }$$
(1)

where Y0 is steady state photovoltage or photocurrent, \({y}_{0}\) and \({t}_{0}\) are offset constants and τ is time constant that represents charge-carrier lifetime or transport time. LD was obtained by the following equation which was used in typical thin film solar cells for calculating the diffusion length:

$${L}_{D}=L\sqrt{\frac{\tau_{{\rm{r}}}}{2.35\tau_{{\rm{t}}}}}$$
(2)

where L is the thickness of perovskite film.

PV measurements

Completed devices were masked with metal aperture masks (0.05 cm2 for J–V measurement by a Keithley 2400 source meter under simulated 1-sun AM 1.5 G illumination (100 mW cm−2) with the solar simulator (Enli Technology Co., Ltd.) in a nitrogen-filled glovebox. EQE measurements were conducted using a Bentham PVE300-IVT system, and the bias illumination obtained by 500 nm short-wave pass filters and 800 nm long-wave pass filters were used for the measurements of the bottom subcell and front subcell, respectively. For the HR-EQE tests, the light was chopped at 137 Hz and coupled into a Bentham monochromator. The monochromatic light spot was focused onto the active area of the perovskite solar cell, and its current under short-circuit conditions was fed to a current preamplifier (Stanford SR 570) before it was analyzed with a lock-in amplifier (Stanford SR830 DSP). The time constant of the lock-in amplifier was chosen to be 1 s, and the amplification of the preamplifier was increased to resolve low photocurrents. The intensity of the LED laser used was calibrated with built-in silicon and germanium diodes before measurements. The long-term MPPT was conducted by MPPT-4B source measure unit system (Shenzhen Lancheng Technology Ltd.) under 100 mW cm−2 white LED in ambient atmosphere.

Computational details

All the spin-polarized DFT calculations were performed with the Vienna Ab initio Simulation Package (VASP, 5.4.4 version) under the framework of the projector augmented wave method. The generalized gradient approximation (GGA) of the Perdew–Burke–Ernzerhof (PBE) functional was employed as the exchange-correlation functional. PBE + D3 method for the Van der Waals (vdW) correction was also included. The valence wave functions were expanded by plane wave with a cutoff energy of 400 eV, while all the structures were optimized until the force on each atom was less than 0.02 eV Å−1. The neural molecule PA was optimized in a 25 Å × 25 Å × 25 Å unit cell. To calculate the binding energy, the optimized molecules were further introduced to the (100) and (110) surfaces of organic-inorganic hybrid wide bandgap perovskites. The binding energy was calculated according to the following equation:

$${{{{\rm{E}}}}}_{{binding}}={{{{\rm{E}}}}}_{{perovskites}+{PA}}-{{{{\rm{E}}}}}_{{perovskites}}-{{{{\rm{E}}}}}_{{PA}}$$
(3)

Reporting summary

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