Introduction

Presently, several groups have achieved power conversion efficiency (PCE) exceeding 26% on emerging metal halide perovskite solar cells (PSCs), which makes it one of the most promising photovoltaic technologies1,2,3. It is encouraging that perovskite/Si tandem solar cells (TSCs) demonstrated a tremendous PCE of 34.6%4. In recent years, inverted PSCs witnessed remarkable advancements in PCE and stability5,6,7,8,9,10,11,12. However, the commercial deployment of single junction and tandem PSCs is hindered by their poor long-term operational stability. Soft ionic lattice properties of perovskites make them suffer from poor intrinsic and extrinsic stabilities13,14. The extrinsic stabilities can be overcome by developing advanced encapsulation technology15. However, the intrinsic instability induced by ion migration and deep-level defects is challenging to address8,16. The defects within perovskite films can provide pathways for ion migration17. It was reported that the defect density at the interface of perovskite films is 1 ~ 2 orders of magnitude larger than that in the bulk of the films18,19. Trap-assisted nonradiative recombination, also called Shockley-Read-Hall (SRH), would reduce not only PCE but also long-term durability20,21. Moreover, ion migration is much faster at grain boundaries (GBs) and interface than in the bulk of perovskite films19,22. Therefore, it is of significant importance to enhance the PCE and intrinsic stability of PSCs by passivating interfacial defects and inhibiting interfacial ion migration via rational interface engineering.

The interface between the perovskite and electron transport layer (ETL) plays a critical role in realizing high-performance inverted PSCs. The C60 and its derivatives (e.g., PC61BM) are commonly used as ETL in inverted PSCs8,23. Unfortunately, it was reported that the perovskite/ETL interface in inverted PSCs usually suffers from severe nonradiative recombination losses resulting from minority carriers and trap states.24,25 At this interface, there are usually various defects simultaneously encompassing positively charged defects (e.g., undercoordinated Pb2+ and halide vacancies) and negatively charged defects (such as cation vacancies and PbI3−).26 To effectively heal these harmful defects, a variety of interface materials have been developed, mainly involving low-dimensional perovskites27,28,29, Lewis bases30,31,32, organic cations33, and organic salts8,24,34,35,36. Compared with other interface materials, organic salts exhibit great potential in minimizing nonradiative recombination and suppressing ion migration due to their ability to simultaneously passivate positively and negatively charged defects8,24,35,36. For example, Sargent et al.24 synergistically used two types of organic ammonium salts propane-1,3-diammonium iodide (PDAI2) and 3-(methylthio)propylaminehydroiodide (3MTPAI) to passivate the defects at upper interface in inverted PSCs, which achieved a certified quasi-steady-state PCE of 25.1%. To maximize the potential of organic salt modifiers, the rational design of organic anions and cations is of extreme importance. In terms of anions, halide anions (i.e., F-, Cl−, Br−, and I−) are frequently employed24,35,36. However, nonhalide anions possess the advantages of adjustable structure and wide varieties as compared to halide anions.8,37 The organic anions containing -SO3−38 and -COO−8 have been certified to be capable of effectively passivating undercoordinated Pb2+ and halide vacancy defects and thus suppressing ion migration. Yang et al. demonstrated that p-toluenesulfonate stabilizes the perovskite interface by neutralizing negative work function shifts, reducing halide migration, and preventing charge accumulation, thereby enhancing the performance and stability of PSCs39. Generally, cations can only passivate negatively charged defects through ionic bonds or electrostatic interaction. Our groups have revealed that multiple active site molecules are more effective in passivating trap states than single active site molecules3,40,41,42. In order to increase the functions of cations and strengthen their interaction with perovskites, additional effective functional groups should be incorporated to functionalize cations. For instance, the introduction of Lewis base groups (e.g., -C=O and -SH) in organic cations should be able to passivate both negatively charged cation vacancy defects by ionic bond or hydrogen bond and positively charged undercoordinated Pb2+ and halide vacancy defects by coordination bond. Finally, the influencing mechanisms of the steric hindrance and the number of hydrogen atoms on cations on defect passivation remain blurry.

In this work, we proposed an effective defect passivation strategy by rational design of hydrogen atoms and steric hindrance of amino acid benzyl ester organic cations in nonhalide ammonium salts. The three amino acid salts, benzyl glycinate p-toluenesulfonate (BGTS), L-valine benzyl ester p-toluenesulfonate (VBETS), and L-leucine benzyl ester p-toluenesulfonate (LBETS), which have the same p-toluenesulfonate (TS-) nonhalide anion and different amino acid benzyl ester cations, were employed to passivate the upper surface of perovskite films in inverted PSCs. We found that the defect passivation effect of TS- anions greatly depended on the steric hindrance induced by cation size. The only difference between the three cations (i.e., BG+, VBE+, and LBE+) is whether there is isopropyl or isobutyl substituent on BG+. We revealed that the number of hydrogen atoms on cations exhibited a profound influence on defect passivation. The defect passivation effect of cations was determined by the balance between the number of hydrogen atoms and cation sizes. The multiple different active sites (TS-, -NH3+, and, -C=O) enabled simultaneous passivation of multiple defects, mainly including undercoordinated Pb2+, halide vacancy, and cation vacancy defects, which minimized interfacial nonradiative recombination losses. Through comprehensive consideration, VBETS was the most successful in passivating interfacial defects due to its appropriate number of hydrogen atoms and cation size. We also demonstrated that the VBETS can be used to passivate the defects of conventional and wide band gap (WBG) perovskite films, which confirmed the universality of this passivation strategy. The passivation of VBETS is generally suitable for improving the efficiency of single-junction and tandem PSCs.

Results

Theoretical screening of organic cations

Figure 1a shows the chemical structures and electrostatic potential (ESP) of three sorts of cations and TS- -anions. All organic salts have TS-, -NH3+, and -C=O active sites, which should be able to simultaneously manage various positively and negatively charged defects. Density functional theory simulation was performed to investigate the defect passivation effect of three organic nonhalide ammonium salts for the surface of perovskite films. As illustrated in Fig. 1b and Supplementary Figs. 1–3, we explored the passivation effect of different organic salts for various defects, including Pb2+, substituted I- (PbI), I- vacancy (VI), and FA+ vacancy (VFA). It was found that the three salts can passivate these defects, but VBETS and LBETS were better than BGTS regardless of defect types, which could be because of the introduction of alkyl substituent. For positively charged PbI (Supplementary Fig. 1a–c) and VI (Supplementary Fig. 2a–c), the defect passivation effect increased in the order of BGTS, VBETS, and LBETS, which is consistent with the cation size order of BG+ < VBE+ < LBE+. In our previous work, we have uncovered that the size of imidazolium cations can markedly impact the defect passivation effect of BF4- anions43. The large-sized cation would weaken the Coulomb force between the cation and anion due to large steric hindrance, heightening the interaction of anions with perovskites43,44,45.

Fig. 1: Theoretical screening of organic cations.
figure 1

a The chemical structures and electrostatic potential of BG+, VBE+, LBE+, and TS-. b The binding energy of BGTS, VBETS, and LBETS with FAPbI3 containing PbI anti-site, VI, and VFA defects. c–e The binding energy of (c) BG+, (d) VBE+, and (e) LBE+ cations with FAPbI3 containing VFA defects. f Binding energy of An+, MA+, EA+, BG+, VBE+, and LBE+ with FAPbI3 containing VFA defects. g–i Charge density difference of (g) BG+, (h) VBE+, and (i) LBE+ with FAPbI3 surface encompassing VFA defects (cyan indicates a decrease in charge density, yellow indicates an increase in charge density).

Full size image

Regarding negatively charged VFA defects (Supplementary Fig. 3a–c), the defect passivation order of BG+ < LBE+ < VBE+ was observed. This means incorporating an alkyl substituent can promote the defect passivation of cations for VFA defects. From the perspective of steric hindrance, the smaller cation size is more beneficial for passivating VFA defects. However, this is the opposite of our experimental results. Therefore, we speculated that the other factor exerts a highly positive role in passivating VFA defects, which counteracts the negative effect of steric hindrance. Subsequently, we focused on delving into the passivation effect of three different cations for the VFA defects. Figure 1c–e presents that the binding energies of BG+, VBE+, and LBE+ cations with FAPbI3 perovskites containing VFA defects are −4.62 eV, −5.15 eV, and −4.91 eV, respectively. This indicates that the cations’ passivation order agrees with the salt molecule passivation order. Obviously, the VBE+ showcased the optimal passivation effect for VFA. To study the reason behind the phenomenon, we systematically compared the binding energies of NH4+ (An+), CH3NH3+ (MA+), CH3CH2NH3+ (EA+), BG+, VBE+ and LBE+ with FAPbI3 perovskites containing VFA defects (Fig. 1f and Supplementary Fig. 4a–f). It was found that the binding energy gradually increased as the number of hydrogen atoms on alkyl increased. We inferred that hydrogen atoms on alkyl had strong electrostatic interaction with VFA defects and/or [PbI6]4- octahedron, which was further verified by the charge density difference of BG+, VBE+, and LBE+ with FAPbI3 surface encompassing VFA defects (Fig. 1g–i and Supplementary Fig. 5a–c). It can be concluded that the defect passivation effect of cations was determined by the compromise between the number of hydrogen atoms and steric hindrance. The proper number of hydrogen atoms and steric hindrance are necessary for achieving optimal defect passivation.

Experimental screening of organic cations via investigating interactions

It needs to be pointed out that the 1.58 eV-Cs0.05(FA0.95MA0.05)0.95Pb(I0.95Br0.05)3 perovskite was used to investigate the interaction of modifiers with perovskite and the effect of interface modification on the optoelectronic properties of perovskite films. We further experimentally analyzed the chemical interactions of three organic ammonium salts with the perovskite. As shown from the X-ray photoelectron spectroscopy (XPS) in Fig. 2a, the BGTS, VBETS, and LBETS modified perovskite films exhibited S 2p characteristic peaks, indicating that all modification molecules have been successfully introduced into the upper surface of the perovskite films. At the same time, VBETS exhibited the maximum peak shift, demonstrating the maximum variation in the chemical environment and binding energy of -SO3- with the perovskite. As shown in XPS results in Fig. 2b, c, the binding energy of both Pb 4f and I 3d peaks of the modified film by the three organic ammonium salt molecules shifted to the lower binding energy compared to the control film, which is attributed to the interaction of -C=O in cations and -S=O in anions with undercoordinated Pb2+ and/or VI defects. It revealed that BGTS, VBETS, and LBETS exhibited peak shifts of 0.4 eV, 0.6 eV, and 0.5 eV, respectively. The largest peak shift of 0.6 eV indicates that VBETS had the best passivation effect for Pb-related defects through the synergistic coordination of -C=O and -S=O in anions and cations due to appropriate cation size and steric hindrance. The shift of the I 3d binding energy peak indicates that hydrogen bonds were formed between -NH3+ in the three organic ammonium salt molecules and I− in [PbI6]4− in perovskite, which is conducive to filling the VFA vacancy defects in octahedral crystals and stabilizing perovskite crystals.46 Additionally, the electrostatic interaction of the hydrogen atoms on the alkyl group with I- on [PbI6]4− octahedron should also be responsible for the shifted binding energy of I 3d peaks, which has been confirmed by theoretical calculation results. As shown in the Fourier transform infrared spectroscopy (FTIR) in Fig. 2d-f, the C=O peak was shifted from 1749 cm−1 of BGTS to 1751 cm−1 of BGTS+PbI2, from 1748 cm−1 of VBETS to 1735 cm−1 of VBETS+PbI2, and from 1746 cm−1 of LBETS to 1741 cm−1 of LBETS+PbI2, suggesting the coordination bonds between C=O and Pb2+. In a similar way, the shift of S=O from 1033 cm−1 of BGTS to 1038 cm−1 of BGTS+PbI2, from 1038 cm−1 of VBETS to 1022 cm−1 of VBETS+PbI2, and from 1035 cm−1 of LBETS to 1025 cm−1 of LBETS+PbI2 affirmed the coordination interaction of S=O with Pb2+. In short, combining XPS and FTIR confirmed the interaction of C=O and S=O with Pb2+.

Fig. 2: Chemical interactions and perovskite film quality.
figure 2

a XPS spectra of S 2p in the control and BGTS, VBETS, or LBETS-modified perovskite films. b, c XPS spectra of (b) Pb 4f and (c) I 3d in control and BGTS, VBETS, or LBETS modified perovskite films. d-f FTIR for (d) BGTS and BGTS+PbI2, (e) VBETS and VBETS+PbI2, and (f) LBETS and LBETS+PbI2. g, h Steady-state PL (g) and TRPL (h) spectra of the control, BGTS, VBETS, and LBETS-modified perovskite films on a bare glass substrate. i I-V curves of the devices based on ITO/SnO2/perovskite (BGTS, VBETS, and LBETS)/C60/Ag structure under dark.

Full size image

Characterization of perovskite films

The above experimental and theoretical results certified the chemical interaction between organic salt modifiers and perovskites as well as their passivation ability for various defects. Then, we conducted a qualitative and quantitative analysis of the defects through spectroscopic and electrical means. A 475 nm laser excition was employed to measure the steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) of perovskite films on bare glass. As shown in Fig. 2g, the PL intensities of the samples gradually increased according to this order of the control, BGTS, LBETS, and VBETS-modified perovskite films, indicating the lowest trap density in the VBETS-modified sample. The TRPL results in Fig. 2h show that the average lifetimes of the control, BGTS-, VBETS-, and LBETS-passivated perovskite films were 362.05 ns, 840.99 ns, 1346.85 ns, and 1057.46 ns, respectively. Among all samples, the VBETS-modified perovskite film had the longest average carrier lifetime and strongest PL intensity, indicating its excellent defect healing function, which minimized nonradiative recombination losses. Much reduced nonradiative recombination should be conducive to enhancing open-circuit voltage (VOC) and fill factor (FF) as well as reinforcing device stability.

To further evaluate the passivation effects of different modification molecules from a quantitative perspective, we calculated the trap state density of perovskite films by the space charge limited current (SCLC) method. According to the formula of Nt = 2ε0εrVTFL/qL2,47 the value of VTFL is positively correlated with the value of trap state density. SCLC results based on electron transporting-only devices (ITO/SnO2/perovskite/C60/Ag) revealed that the VTFL of control, BGTS, VBETS, and LBETS-modified devices are 1.03 V, 0.75 V, 0.51 V and 0.71 V (Fig. 2i), while the results based on hole transporting-only devices (ITO/NiOx/perovskite/Spiro-OMeTAD/Ag) revealed that the VTFL was 0.67 V, 0.66 V, 0.53 V, 0.58 V (Supplementary Fig. 6). After BGTS, VBETS, and LBETS passivation, the electron defect density was reduced from 13.2×1015 cm−3 to 9.6×1015 cm−3, 6.5×1015 cm−3, and 9.1×1015 cm−3, and hole defect density was decreased from 8.6 × 1015 cm−3 to 8.5×1015 cm−3, 6.8 × 1015 cm−3, and 7.4×1015 cm−3, indicating that VBETS can efficiently passivate the multiple defects on the surface of perovskite films.

From the perspective of macroscopic thin film and crystal morphology, the effects of BGTS, VBETS, or LBETS modification on the morphology of 1.58 eV-Cs0.05(FA0.95MA0.05)0.95Pb(I0.95Br0.05)3 perovskite films were analyzed by scanning electron microscopy (SEM) in Fig. 3a. There are many white PbI2 phase and pores on the surface of the pristine perovskite film, while the white PbI2 phase and pores were reduced after BGTS, VBETS, and LBETS modification48. The same results were obtained in the SEM images of wide bandgap 1.66 eV-Cs0.05MA0.15FA0.8Pb(I0.76Br0.24)3 perovskites (crystallization on a monocrystalline Si substrate) for the reduction of excess PbI2 phase impurities after the modification of BGTS, VBETS, or LBETS (Supplementary Fig. 7a–d). Among them, the VBETS-modified perovskite film exhibited the flattest surface morphology with lowest root mean square roughness of 15.22 nm, lower than that of 19.68 nm, 17.34 nm and 16.04 nm for control, BGTS, and LBETS modified perovskite films, as exhibited in the atomic force microscope (AFM) images in Supplementary Fig. 8a–d This could be ascribed to the strongest interaction of VBETS with perovskites. The flat and uniform perovskite film will promote electron extraction at the perovskite/ETL interface and reduce nonradiative recombination losses49.

Fig. 3: Surface potentials and energy level of perovskite films.
figure 3

a SEM images of the control, BGTS, VBETS, and LBETS-modified 1.58 eV-Cs0.05(FA0.95MA0.05)0.95Pb(I0.95Br0.05)3 perovskite films. b LBIC mapping images of the control, BGTS-, VBETS-, and LBETS-modified 1.58 eV-Cs0.05(FA0.95MA0.05)0.95Pb(I0.95Br0.05)3 based PSCs, the top row is fresh perovskite films and the bottom row is perovskite films after aging for 5 days at room temperature under RH of 60 ± 10%. c KPFM images of the control, BGTS, VBETS, and LBETS-modified 1.58 eV-Cs0.05(FA0.95MA0.05)0.95Pb(I0.95Br0.05)3 perovskite films. d Tangent lines in KPFM images to demonstrate changes in surface potential. e UPS spectra of the control, BGTS, VBETS, and LBETS-modified perovskite films. f Energy level diagram of the control and BGTS, VBETS, and LBETS-modified perovskite films.

Full size image

The defect distribution and density were further characterized by laser beam-induced current (LBIC) imaging technology, which used a laser beam to scan the surface of the PSCs to reveal the photocurrent mapping and generate the internal trap defect distribution of the device on a macroscopic scale50. As shown in Fig. 3b, the LBIC results show that the control film exhibited very poor film uniformity and low photon current response, while all modified devices presented enhanced film uniformity and current. VBETS-modified devices exhibited the highest current due to their best defect passivation effect. Further, the surface current of the control devices decreased significantly after aging at room temperature under relative humidity (RH) of 60 ± 10% for 5 days, while the surface current of the VBETS-modified PSCs decayed slightly. Moreover, the slowest decay was found in the VBETS-modified devices.

Kelvin probe force microscopy (KPFM) measurement was carried out to gain insights into the surface potentials of the perovskite films. As shown in Fig. 3c, d, the average surface potential of the control, BGTS, VBETS, and LBETS-modified perovskite films was 196.64 mV, 22.92 mV, -233.22 mV and -53.46 mV, respectively. It was revealed that the VBETS-modified perovskite film exhibited the lowest surface potential with the largest difference up to 430 mV compared with the control perovskite film. As shown from the ultraviolet photoelectron spectroscopy (UPS) (Fig. 3e) and corresponding bandgap structure schematic diagram (Fig. 3f), we can conclude that the Fermi energy levels of the BGTS, VBETS, and LBETS modified perovskite films shifted upward compared with the control perovskite film. Particularly, VBETS-modified perovskite films exhibited a most obviously shifted Fermi energy, indicating that VBETS modification induced more n-type characteristics33. This is related to the reason that VBETS passivated the electronic defects on the surface of the perovskite and formed a back surface field with the bulk phase of the perovskite. The formed back surface field is in the same direction as the built-in electric field of the device, thus making the stronger built-in potential (Vbi) of 0.95 V, which is higher than 0.87 V for control devices from the Mott-Schottky plots (Supplementary Fig. 9a, b), thus promoting electron transport and extraction51. Actually, organic cation-induced n-type doping and back electric field at perovskite/ETL interface have been reported to improve the PCE and stability of inverted PSCs33.

Investigation of photovoltaic performance

To assess the photovoltaic performance of PSCs, we fabricated p-i-n type NiOx-based inverted PSCs device, where BGTS, VBETS, and LBETS were used to modify the upper surface of perovskite films. Specifically, to verify the universality of our surface passivation strategy, we adopted conventional bandgap perovskites of 1.53 eV and 1.58 eV perovskite, as well as a WBG 1.66 eV perovskite as light-harvesting materials. The comparative analysis of the optimized concentrations of three passivating agents of BGTS, VBETS, and LBETS for modifying the perovskites is located at 0.5 mg/mL in an isopropanol solution (Supplementary Fig. 10a–c). The cross-sectional SEM image of the VBETS-modified 1.53 eV-perovskite based PSC device is shown in Supplementary Fig. 11, revealing the good crystallinity and clear functional layer interface. Supplementary Fig. 12a–d shows statistical data of all photovoltaic parameters (PCE, VOC, JSC, and FF) from 1.53 eV PSCs. VBETS-modified PSCs exhibited higher average PCE values of 24.93%, respectively, higher than 23.01% for the control devices. Figure 4a and Supplementary Fig. 13 present the typical J-V curves of the best-performing control BGTS, VBETS, and LBETS-modified 1.53 eV PSC devices in reverse scan (RS) and forward scan (FS) testing modes. It was revealed that the control, BGTS, VBETS, and LBETS-modified PSCs delivered a champion PCE of 23.44%, 23.82%, 25.26% and 24.77%, respectively. We also achieved an attractive certified PCE of 25.15% (JSC of 25.68 mA cm−2, a VOC of 1.191 V, and an FF of 82.28%) in RS and a certified value of 24.95% (JSC of 25.75 mA cm−2, a VOC of 1.184 V, and an FF of 81.84%) in FS (Supplementary Figs. 14–20). To the best of our knowledge, our obtained certified PCE value of 25.15% is the record result reported for the inverted PSCs based on the vacuum flash method in ambient conditions for perovskite deposition (Supplementary Table 1). In addition, the hysteresis index (HI = (PCEReverse-PCEForwad)/PCEReverse) was 4.39%, 2.73%, 1.15%, and 2.26% for the control, BGTS, VBETS, and LBETS-modified PSCs, respectively (Supplementary Table 2). The smallest hysteresis for the VBETS-modified PSCs could have resulted from reduced defect density and facilitated interfacial electron extraction, thus suppressing interfacial charge accumulation. We also prepared PSC devices based on 1.53 eV perovskite in an N2 atmosphere using VBETS as a passivator. The J-V curve results (Fig. 4b) show that the VBETS-modified device can achieve champion efficiency values of 26.28% (JSC of 26.13 mA cm−2, a VOC of 1.191 V, and an FF of 84.45%) in RS mode and 26.23% (JSC of 26.13 mA cm−2, a VOC of 1.191 V, and an FF of 84.28%) in FS mode, which are much higher than the efficiency values of 24.64% (JSC of 25.71 mA cm−2, a VOC of 1.178 V, and an FF of 81.36%) and 24.57% (JSC of 25.71 mA cm−2, a VOC of 1.178 V, and an FF of 81.13%) of the control device.

Fig. 4: Device performance and long-term stability.
figure 4

a–c Champion J-V curves of the best-performing control and VBETS-modified (a) 1.53 eV (prepared in the ambient conditions), (b) 1.53 eV (prepared in the glovebox with 99.999% N2 conditions), and (c) 1.66 eV perovskite-based PSC devices in reverse and forward scan mode. d The column-shaped statistical chart of PCE, VOC, JSC, and FF parameters for control devices and VBETS-modified devices based on 1.53 eV, 1.58 eV, and 1.66 eV perovskite as photoactive layer Specifically, * represents that perovskite devices are prepared in N2 glove boxes, while the others are prepared under air ambient conditions. e The EQE spectra and integrated JSC of the champion VBETS-modified PSCs with 1.53 eV, 1.58 eV, and 1.66 eV perovskites. f J-V curves of the 1.58 eV-perovskite-based PSC modules with VBETS modification. The inset is the photograph of a VBETS-modified perovskite module with an aperture area of 32.144 cm2. g Comparison of the historical PCEs of the PSC module with an aperture area exceeding 30 cm2. h Cross-sectional SEM image for exhibiting the device structure of perovskite/HJT crystalline silicon TSCs modified with VBETS. i J-V curves for the TSCs without and with VBETS in reverse and forward scan mode. j MPPT stability curves of the single-junction PSCs without and with VBETS.

Full size image

Our developed passivation approach is also effective for the 1.58 eV and WBG 1.66 eV-based PSCs prepared by using vacuum flash technology in ambient conditions. As is seen in Supplementary Fig. 21, the champion J-V curves indicate that VBETS passivation leads to an increased PCE from 21.69% to 23.52% for the 1.58 eV PSCs. Further, VBETS modification helped to achieve an enhanced PCE from 20.79% to 21.75% in the WBG 1.66 eV-based PSCs (Fig. 4c), accompanied by increased VOC and FF together with slightly improved JSC. This indicates that our surface passivation method suits both conventional bandgap and WBG 1.66 eV-PSCs. The statistical chart of JSC, FF, VOC, and PCE (Fig. 4d and Supplementary Figs. 12, 22–23) of the devices with different components reveals that VBETS modification could help to improve all the photovoltaic performance in PSCs, proving the universality and versatility of VBETS passivator. The integrated JSC from the EQE of devices modified by VBETS all show a certain degree of increase, regardless of the type of perovskite system (Supplementary Figs. 24–26). Specifically, the integrated JSC values of the champion PSCs with band gaps of 1.53 eV, 1.58 eV, and 1.66 eV are 25.74 mA cm−2, 24.56 mA cm−2, and 20.42 mA cm−2, respectively (Fig. 4e). We then fabricated a large-area module by vacuum flash method to verify the effectiveness and universality of VBETS passivation, where 14 subcells were connected in series. The champion VBETS-modified PSC module with an aperture area (including dead area) of 32.144 cm2 achieved a PCE of 21.00% (Fig. 4f). Calculating by the active area of 30.408 cm2 (which accounts for 94.6% of the aperture area), we obtained an efficiency of 22.20%. According to the statistics, these results are all the highest efficiency reported for the large-area modules with an area over 30 cm2 (Fig. 4g and Supplementary Table 3).

To investigate the charge carrier recombination lifetime (τr) and carrier transport lifetime (τt) of PSC devices, transient photovoltage (TPV) and transient photocurrent (TPC) measurements were performed on the control and VBETS-modified devices.52 Supplementary Fig. 27 shows that the VBETS-modified device exhibited a τr value of 2.59 µs, more than twice as long as that of the control device (1.24 µs). In addition, it can be seen that VBETS-modified devices exhibited a τt value of 0.50 µs (Supplementary Fig. 28), which is dramatically shorter than that of 1.66 µs for the control devices. As shown in Supplementary Fig. 29, the -3dB bandwidth (f-3dB) of VBETS-modified devices was 0.40 MHz, which is much larger than the 0.10 MHz of the control devices, indicating that VBETS-modified devices have faster charge carrier transport and light response53. After VBETS modification, the reduced defect density and improved energy band alignment should account for promoted carrier transport and extraction. Electrochemical impedance spectroscopy (EIS) measurements (Supplementary Fig. 30) further revealed the ameliorated charge transport and inhibited nonradiative recombination following VBETS passivation. VOC as a function of light intensity measurements (Supplementary Fig. 31) revealed that the ideal factor (1.24) of the VBETS-modified device is closer to 1 than that of the control device (1.43), indicating the reduced SRH recombination associated with the trap defects54. We also performed PL and TRPL of the glass/perovskite (1.53 eV, 1.58 eV and 1.68 eV)/C60 samples to gain deeper insights into the carrier extraction capabilities in PSC devices (Supplementary Fig. 32a–f). The modification of VBETS can reduce PL intensity and decay lifetime for all bandgap perovskite samples, which is indicative of facilitated electron extraction and transfer at perovskite/C60 interface.

Based on WBG 1.66 eV perovskite as the top-subcells with VBETS passivation, we fabricated perovskite/silicon TSCs, where the bottom cell is 1.1 eV crystalline silicon (Si) heterojunction (HJT) solar cells. The SEM cross-sectional view and overall structural schematic diagram of the TSCs are shown in Fig. 4h and Supplementary Fig. 33. Precisely, the magnetron sputtering method was applied on a heterojunction Si substrate to prepare transparent ITO as the top transparent conductive electrode. As is exhibited in Fig. 4i, the champion efficiency of the VBETS-passivated tandem device was achieved at 30.98% (with a VOC of 1.890 V, a JSC of 20.23 mA cm−2, and an FF of 80.76%) in the RS, and 30.56% (with a VOC of 1.895 V, a JSC of 20.23 mA cm−2 and an FF of 79.73%) in the FS. These efficiency values are much higher than that of 28.69 % (with a VOC of 1.816 V, a JSC of 20.44 mA cm−2, and an FF of 77.29%) in RS and 28.01% (with a VOC of 1.797 V, a JSC of 20.44 mA cm−2, and an FF of 76.25%) in FS for the control tandem solar cells. The effective enhancement of TSC devices comes from the improved Voc, demonstrating the effective role of VBETS interface defect passivation in suppressing nonradiative recombination. The EQE spectra in Supplementary Fig. 34 revealed that the integrated JSC of the VBETS-modified WBG PSCs and filtered SHJ devices were 20.28 mA cm−2 and 20.13 mA cm−2, respectively, which are consistent with the JSC values from J-V curves.

Study of long-term stability

We assessed the operational stability of single-junction PSCs and perovskite/Si TSCs without and with VBETS passivation. We conducted the maximum power point tracking (MPPT) of the control and VBETS-modified devices under continuous 100 mW cm−2 white LED light irradiation at room temperature of 25 ± 5 °C and stored in a 99.99% nitrogen environment. As for the 1.53 eV-Cs0.05MA0.05FA0.9PbI3-based single-junction PSCs, encapsulation using glass front with UV-curable adhesive encapsulant for sealing was applied, as is shown in Fig. 4j, after 4000 h of continuous light exposure, the VBETS-modified device could retain 90.8% of the initial PCE, while the control device dropped to 70.9% under the same conditions. The improved light stability of VBETS-modified devices is mainly associated with reducing trap defects at surface and grain boundaries. The TSCs were then subjected to stability measurements at room temperature and under white light illumination with internal cyclic J–V tests with a scanning interval of 124.8 mins. As illustrated in Supplementary Fig. 35, the VBETS-modified TSCs could maintain 94.2% of the initial PCE value after 1030 h of aging. The overall output characteristics and stability of the device are higher than those of the control TSC device. It was revealed that VBETS modification can enhance the operational stability of single junction and TSCs due to reduced surface defect density by synergistic passivation of nonhalide anion and organic cations.

Meanwhile, the effect of VBETS modification on the long-term stability of PSCs was investigated systematically. The encapsulated PSCs are stored in ambient conditions (RH = 35 ± 5%, Temperature = 25 ± 5 °C). Supplementary Fig. 36a–d compares the results of the long-term stability of the 1.53 eV, 1.58 eV, 1.66 eV-based single-junction PSCs and perovskite/Si TSCs without and with VBETS passivation. It can be observed that, whether in single-junction or tandem devices, the VBETS-modified devices exhibit higher stability than the control devices. The 1.66 eV-PSCs and TSCs degrade at faster rates, correlating with the inherently poor stability of the WBG perovskite with a high bromide (Br) ratio. The contact angles of the control, BGTS, VBETS, and LBETS-modified perovskite films are 30.1°, 31.4°, 43.3°, and 45.1°, respectively. VBETS and LBETS, due to their longer cationic alkyl chains, exhibit stronger hydrophobic characteristics, which in turn contribute to improved moisture resistance of the films and enhanced device stability (Supplementary Fig. 37a–d).

In summary, we have developed a universal passivation strategy for organic ammonium salt molecules containing nonhalide organic anions, minimizing energy loss at the upper interface in inverted PSCs through the rational design of the number of hydrogen atoms on cations and steric hindrance. The synergistic effect of anions and cations enabled the simultaneous passivation of positively charged and negatively charged defects and the modulation of interface bands. VBETS possessing an appropriate number of hydrogen atoms and cation size exhibited the best effect in defect passivation and energy band modulation. The universality of this passivation strategy was confirmed by using different bandgap perovskites. Finally, the VBETS-modified inverted PSCs based on conventional 1.53 eV bandgap perovskite yielded a PCE of 25.26% prepared in ambient conditions and 26.28% prepared in an N2-preparing atmosphere. The perovskite/Si TSCs coupled with VBETS passivation demonstrated a promising PCE of 30.98%. This work highlights the critical role of the number of hydrogen atoms and steric hindrance in designing multisite nonhalide ammonium salts to improve the PCE and stability, which lays the groundwork for the development of perovskite photovoltaics.

Methods

Materials

Formamidine iodide (FAI), methylammonium iodide (MAI), methylamine hydrochloride (MACl), methylammonium bromide (MABr) were purchased from Greatcell Solar Materials. Lead chloride (PbCl2 99.5%) were purchased from Xi’an Bright Optoelectronics Technology Co., Ltd. Lead iodide (PbI2, 99.999%), cesium iodide (CsI, 99.99%) and lead bromide (PbBr2 99.999%) were purchased from Chengdu Alpha Metal Materials Co., Ltd.. N,N-dimethylformamide (DMF, 99.9%), Dimethyl sulfoxide (DMSO) and 1-methy-2-pyrrolidinone (NMP, 99.9%) were purchased from MACKLIN. [2-(9H-Carbazol-9-yl)ethyl]phosphonic Acid (2PACz, >98.0% HPLC) and [4-(3,6-Dimethyl-9H-carbazol-9-yl)butyl]phosphonic Acid (Me-4PACz, > 99.0% HPLC) was purchased from TCI; BCP, PC61BM, C60 were purchased from Advanced Election Technology. The three amino acid salts, benzyl glycinate p-toluenesulfonate (BGTS), L-valine benzyl ester 4-toluenesulfonate (VBETS), and L-leucine benzyl ester p-toluenesulfonate salt (LBETS) was obtained from MACKLIN.

Nickel(II) nitrate hexahydrate (Ni(NO3)2·6H2O) was purchased from Sigma-Aldrich for synthesizing the NiOx nanoparticles. All chemicals were used without further purification. NiOx nanoparticles were synthesized as follows: Firstly, 3 g of Ni(NO3)2·6H2O was dissolved in 50 mL of deionized (DI) water. The solution was stirred at room temperature for 30 min. Subsequently, a 1.0 M NaOH solution was slowly added with continuous magnetic stirring until the pH reached approximately 10. After an additional hour of stirring, a green precipitate was obtained via centrifugation at 9000 rpm for 6 min. The precipitate was washed three times with DI water. The collected green solid was frozen for 3 h, then freeze-dried at 5 Pa for 15 h, and finally calcined at 270 °C for 2 h to produce NiOx nanoparticles. Prior to use, the NiOx was dispersed in DI water by sonication for 10 min at a concentration of 30 mg/mL.

Preparation of perovskite precursor solutions and film deposition

1.53 eV-Cs0.05MA0.05FA0.9PbI3: The stoichiometric Cs0.05MA0.05FA0.9PbI3 precursor solution was prepared by dissolving 20.784 mg of CsI, 247.68 mg of FAI, 12.72 mg of MAI, 15 mg of MACl, 11.124 mg of PbCl2, and 785.55 mg of PbI2 in 1 ml mixed solvent of DMF/DMSO (4:1, by volume). For the 1.53 eV- Cs0.05MA0.05FA0.9PbI3 film crystallization, the perovskite films were deposited by spin-coating the perovskite precursor solution on glass/ITO/NiOx/Me-4PACz substrate at 1000 rpm for 10 s and 5000 rpm for 30 s. The wet perovskite film was quickly put into a sample chamber connected to vacuum-pumping instrumentation (Internal space is 12 cm × 12 cm × 2.1 cm). By opening the valve connecting the specimen chamber to the pump system, the perovskite film was immediately exposed to low pressure maintained at 10 Pa for 30 s, followed by full pressurization by admitting ambient air into the specimen chamber. Subsequently, the perovskite film was annealed at 100 °C for 15 min in the air (RH = 30 ± 10%, Temperature = 25 ± 5 oC) for full crystallization. For devices prepared in N2 environment, all operating procedures are carried out in a glove box.

1.58 eV-Cs0.05(FA0.95MA0.05)0.95Pb(I0.95Br0.05)3: 718.335 mg PbI2, 28.2 mg PbBr2, 19.5 mg CsI, 8 mg MABr, 238.2 mg FAI, and 15.2 mg MACl were dissolved in 1 ml DMF and DMSO mixed solvent with a volume ratio of 4:1 to form 1.4 M stoichiometric solution. For the 1.58 eV perovskite film crystallization, the perovskite precursor solution of 100 μl was rotated for 10 s and 40 s at 2000 rpm and 5000 rpm, respectively, and 160 μl anti-solvent CB was rapidly dropped into the solution at 45 s. The precursor perovskite film was transferred to the hot plate and annealed at 100 °C for 30 min.

1.66 eV-WBG Cs0.05MA0.15FA0.8Pb(I0.76Br0.24)3: 18.2 mg of CsI, 192.6 mg of FAI, 536 mg of PbI2(10% of excess), 131 mg of PbBr2, and 23.5 mg MABr were dissolved in 1 ml DMF and DMSO mixed solvent with a volume ratio of 4:1 to form 1.4 M stoichiometric solution. For the 1.66 eV-Cs0.05MA0.15FA0.8Pb(I0.76Br0.24)3 film crystallization, the perovskite solution of 100 μl was rotated for 5 s and 30 s at 1000 rpm and 5000 rpm, respectively, and 200 μl anti-solvent CB was rapidly dropped into the solution at 20 s. The precursor perovskite film was transferred to the hot plate and annealed at 100 °C for 15 min in the air (RH = 30 ± 10%, Temperature = 25 ± 5 oC) for full crystallization.

Device fabrication

Fabrication of small-area PSCs

First, ITO substrate was etched by laser etching technology. Ultrasonic cleaning of ITO glass after etching was carried out with glass cleaning agent, deionized water, and ethanol in sequence, and each step took 15 min. The wettability of the ITO substrate was enhanced by UV-ozone modification for 15 min. Then, the water-based ink of NiOx nanoparticles (30 mg NiOx nanoparticles dispersed into 1 ml deionized water) was coated on ITO glass at 4000 rpm for 30 s and annealed at 150 °C for 10 min at room temperature. After annealing, the ITO/NiOx substrate was quickly transferred into an N2-filled glove box for subsequent preparation. 2PACZ (1 mg/ml dissolved in IPA, ultrasonic bath for 10 min) or Me-4PACz solution (0.5 mg/ml dissolved in IPA, ultrasonic bath for 10 min) was coated at 4000 rpm for 20 s on ITO/NiOx substrate and then annealed at 150 °C for 10 min. The preparation and crystallization processes of various types of perovskites are described in the previous experimental section. For BGTS, VBETS, and LBETS-modified perovskite film, spin coating 0.5 mg/ml IPA solution of BGTS, VBETS, and LBETS on the crystallized perovskite film at 5000 rpm for 30 s, and annealing at 100 °C for 5 min. Then, for 1.58 eV-Cs0.05(FA0.95MA0.05)0.95Pb(I0.95Br0.05)3 and WBG 1.66 eV- Cs0.05MA0.15FA0.8Pb(I0.76Br0.24)3-perovskite-based solar cells, spin-coating PC61BM-solution (23 mg PC61BM dissolved in 1 ml CB) for 30 s at a speed of 1500 rpm. For the 1.53 eV-Cs0.05MA0.05FA0.9PbI3-based solar cells and modules, about 30 nm C60 was thermally evaporated on the perovskite films under a high vacuum 8 × 10−4 Pa; Next, spin-coating of BCP solution (0.5 mg BCP dissolved in l ml IPA; filtered with a 0.22 μm PTFE filter) on PC61BM layer at a speed of 5000 rpm for 30 s. Finally, the 80 nm Cu, Ag, or Au electrode is thermally evaporated at the top. For sealing the devices for MPP stability testing, a UV-curable adhesive (Eversolar AB-341, Everlight Chemical) was applied over the active area of the perovskite solar cells, followed by the placement of a highly transparent glass panel. The device was then compressed and cured under ultraviolet light for 3 min.

Fabrication of large-area PSC modules

The preparation process of 1.53 eV-Cs0.05MA0.05FA0.9PbI3 perovskite-based large-area modules is similar to that of small-area devices. Specifically, it is needed to evenly disperse 1 mL of Cs0.05MA0.05FA0.9PbI3 precursor solution on the surface of the ITO substrate, after which is needed to rest for 10 s to achieve automatic dispersion of the perovskite precursor solution on the substrate for the following spin-coating process at 2500 rpm for 20 s. The assembly of subcells in the modules involved a three-stage laser etching technique (P1, P2, P3) on a 6*10 cm2 ITO substrate, creating 14 subcells. Initially, P1 targets the ITO substrates for etching, achieving a width of 30 ± 3 µm using a 1064 nm, nanosecond laser. This step involves single optical path etching, ensuring that the insulation resistance between adjacent conductive layers post-etch exceeds 10 MΩ. Importantly, the P1 process leaves no residual TCO film within the etched tracks and does not harm the underlying glass. The second etching step, P2, processes layers of ITO/NiOx/Me-4PACz/Perovskite/C60/BCP, etching to a width of 50 ± 5 µm using a 532 nm picosecond laser. The P2 laser meticulously avoids penetrating the bottom TCO layer, maintaining a precise 30 µm gap between P1 and P2 etchings. P3, the final stage, is employed for etching Au electrodes, matching the width specifications of P1 (30 ± 3 µm) using a 532 nm, picosecond laser. Similar to P2, P3 does not penetrate the bottom TCO layer, preserving a 30 µm interval between the P2 and P3 etchings.

Fabrication of tandem solar cells

The TSC device is built on a HJT silicon substrate. The wafer surface was finely polished for optimal quality. In the fabrication of a tandem device, HJT silicon substrates were initially sectioned into 12 mm × 12 mm squares using a laser. Subsequently, a perovskite sub-cell was constructed atop the HJT Si sub-cell, employing a fabrication technique similar to that used in the creation of perovskite single-junction devices. The thicknesses of the C60/BCP as electron transporting layers and the perovskite absorbers were meticulously adjusted through the spin-coating or evaporation duration. The standard active area for these tandem devices was maintained at 1.0 cm2.

Device characterization

The J-V characteristics were obtained using a Keithley 2400 source meter in a controlled environment simulating AM 1.5 solar irradiance (100 mW cm−2), facilitated by a standard xenon lamp solar simulator (7ISO503A, SOFN INSTRUMENTS). For small area devices, these measurements were conducted either via a forward scan, ranging from −0.1 to 1.4 V, or a reverse scan, from 1.4 V down to −0.1 V. The J-V measurements for the solar cells were conducted in ambient air conditions. The external quantum efficiency (EQE) was determined using an EQE measurement system from EnliTech, Taiwan. All assessments were performed at room temperature in an ambient atmosphere, and the solar cells were not encapsulated during these measurements.

Certification of PSCs

The PSC devices were sent to the Tianjin Institute of Metrological Supervision and Testing, China, for official certification. Specifically, we fill the prepared solar cell samples for certification testing with nitrogen, then store them with a vacuum sealer and always place them in a dark sealed box. During the certification test, the certification agency recalibrates the effective active area of our solar cells to ensure accurate evaluation of data such as photocurrent density. The test environment and test methods are recorded in the certification data in Supplementary Figs. 14–20.

Characterization of device stability

The operational MPPT stability of the encapsulated devices was assessed under continuous one-sun conditions using a white LED lamp without a UV filter in a 99.99% nitrogen environment. These tests were conducted using an MPP tracking system (YH-VMPP-16, Yanghua, Suzhou). Specifically, 16 solar cell samples were simultaneously placed in aging equipment. The environment around solar cells is maintained at around 25 °C through a refrigeration system. Aging equipment is tracked and tested by continuously scanning and analyzing the maximum power point of the device. The interval between each test is 8.01 h. White LED light sources have a light intensity energy of approximately 100 mW cm−2. The cyclic J-V stability of tandem devices is obtained through interval testing of their J-V curves. The scanning interval is 124.8 min. The recording was done using the Keithley 2400 digital source meter, and the data was obtained using the Zeal Young software K2400 Experimental Platform-Professional2450 V3.5. The long-term stability was conducted in ambient conditions at a relative humidity (RH) of 35 ± 5% and a temperature of 23 ± 5 °C.

Characterization morphology of perovskite films

Field-emission SEM (JEOL 7610 F) was used to scan the surface morphology of the film and the cross-sectional view of the device at 3 kV.

Photoluminescence and ultraviolet-visible absorption measurements

The SSPL and TRPL were measured by a fluorescence spectrometer (FLS1000, Edinburgh Instruments) with an excitation wavelength of 475 nm and an emission wavelength of ~800 nm. The ultraviolet-visible absorption spectra of the films were measured using a Shimazu UV-1900 spectrophotometer.

Atomic force microscope measurements

The Asylum Research MFP-3D was used to measure AFM morphology and KPFM in tapping mode.

X-ray photoelectron spectroscopy and ultraviolet photoelectron spectroscopy measurements

XPS and UPS spectra were collected through Thermo ESCALAB XI+. Fourier transform infrared (FTIR) spectroscopy was recorded using the FTIR-805 spectrometer of Tianjin Gangdong SCI.&TECH. CO,.LTD.

Laser beam-induced current LBIC measurements

LBIC was tested by LSD4 of ENLITECH, Taiwan. LBIC was used to map the local performance of photovoltaic devices by scanning a focused 405 nm light beam across the surface, generating electron-hole pairs and a measurable photocurrent. This creates a spatial map that reveals variations in photoelectric efficiency and identifies defects.

Mott–Schottky and electrochemical impedance spectroscopy measurements

The Mott-Schottky measurements were performed using the AMETEK VersSTAT 3 F electrochemical workstation at a fixed frequency of 1 kHz. The applied bias voltage range is 0 V ~ 1.5 V. EIS measurements are made in the dark, with frequencies ranging from 0.1 Hz to 1,000,000 Hz, using the same instruments as the Mott-Schottky measuring instruments.

Transient photocurrent and photovoltage measurements

TPC/TPV was tested by PD-RS of ENLITECH, Taiwan.

Density functional theory calculation

The DFT calculations are performed using a Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional. Projector augmented-wave (PAW) pseudopotentials are employed with a plane-wave cutoff energy of 400 eV. The DFT-D3 method is used for describing van der Waals interactions. The FAPbI3 (001) surface is modelled by a slab consisting of seven atomic layers and a vacuum gap of 40 Å. The bottom two atomic layers in the slab are fixed during structural relaxation. The convergence criteria for the atomic force is set to 0.02 eV/Å. The binding energy of the ligand with the FAPbI3 surface is defined as Eb = E(slab@ligand)–E(slab)–E(ligand). The electrostatic potentials (φ) of the passivators are calculated using the Gaussian 09 package at the B3LYP/def2TZVP level with DFT-D3.

Reporting summary

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