All materials were purchased and used without undergoing any further purification process. 2,2′,7,7′-tetra(N,N-di-tolyl)amino-9,9-spiro-bifluorene (Spiro-TTB) and C₆₀ were purchased from Luminesce Technology Corp. Lead(II) iodide (PbI₂) was purchased from Sigma-Aldrich. Cesium chloride (CsCl), formamidine bromide (FABr), formamidine iodide (FAI), methylamine hydrochloride (MACl) and 4-fluorobenzylamine hydroiodide (F-PMAI) 2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP) were purchased from Xi’an Polymer Light Technology in China. Ethanol (99.5% anhydrous) and isopropyl alcohol (IPA) (99.5% anhydrous) were purchased from Alfa Aesar.

The current density-voltage (J-V) curves of single-junction perovskite solar cells and TSCs were measured using a Keithley 2400 Source meter, under AM 1.5G illumination (100 mW cm⁻²) with a xenon-lamp-based solar simulator (Enli. Tec., Taiwan). The area of the cells was determined by the aperture shade masks. The devices were measured at a scan rate of 20 mV s⁻¹ and a delay time of 20 ms. For single-junction cells, the bias scan was from 1.2 to 0 V for reverse scan and from 0 to 1.2 V for forward scan. For tandem cells, the bias scan was from 2 to 0 V for reverse scan, and from 0 to 2 V for forward scan. The external quantum efficiency spectra (EQE) were measured by the Enli Tec (Taiwan) system. Space charge limited current (SCLC) measurement was taken in the dark at a scan rate of 20 mV s⁻¹ with a delay time of 20 ms. The bias scan ranged from 0 to 5 V. Electrical impedance spectroscopy (EIS) measurement was taken by Metrohm Autolab (PGSTAT204). Mott-Schottky plot was obtained by Metrohm Autolab (PGSTAT204). The scanning electron microscopy (SEM) images of top views and cross sections were measured by FEINanoSEM650. X-ray diffraction (XRD) was measured by a multi-functional diffractometer (Rigaku, ATX-XRD) with Cu Kα radiation (λ = 1.5405 Å). X-ray photoelectron spectrometry (XPS) and ultraviolet photoelectron spectrometry (UPS) analysis were conducted with a photoelectron spectrometer (Thermo Scientific ESCALAB250Xi, UK). Photoluminescence (PL) and time-resolved photoluminescence (TRPL) spectroscopy were measured by a PL spectrometer FLS980 (Edinburgh Instruments), and a pulsed laser with a wavelength of 475 nm was used as the excitation source. Kelvin probe force microscopy (KPFM) and conductive atomic force microscopy (C-AFM) measurements were taken by Bruker Dimension Icon. The Fourier transform infrared (FTIR) was measured by Thermo Nicolet Nexus intelligent Fourier transform infrared spectrometer. Transient photovoltage (TPV) decay were recorded in an electrochemical workstation (IM6ex, Zahner, Germany) under ambient condition. The contact angles of the samples were performed with an automatic contact angle measuring apparatus (Dataphysics OCA-20, Germany). The time-of-flight secondary ion mass spectroscopy (ToF-SIMS) was measured by ToF-SIMS 5 in Germany GmbH. The PL mapping measurements were taken by using a Q2 multi-dimensional confocal fluorescence microscopy imaging system, equipped with an excitation wavelength 488 nm. The grazing-incidence wide-angle X-ray scattering (GIWAXS) was measured by Xeuss 2.0, using a copper target with a light tube power of 30 W and wavelength of 1.5 Å.

We added F-PMAI to the organic salt solution to prepare the perovskite solar cell, as shown in Fig. S1. We deduce that the molecular can interact with the uncoordinated lead and form hydrogen bonds with the organic cations [29,30]. To verify the above-mentioned interactions, we first performed XPS to study the surface chemistry state of perovskite films (Fig. S3a). A clear F 1s peak at 687.5 eV, which belongs to F-PMA⁺, is observed in the doped film whereas it was absent in the control film in the XPS spectra (Fig. S3b), suggesting the successful introduction of F-PMAI into the perovskite. The Pb 4f peaks move toward lower binding energy after the addition of F-PMAI (Fig. 1a), indicating the intermolecular interaction between F-PMAI and PbI₂. This may be from the interaction between electronegative F⁻ and uncoordinated Pb²⁺ [29]. We further analyzed the XPS spectra of I 3d and also found a lower binding energy shift (Fig. S3c). Such a shift is caused by the bonding between NH₃⁺ from F-PMAI and I⁻ from [PbI₆]⁴⁻ octahedra [31,32]. Besides, we observed that the C=O peak (284.8 eV) originating from external oxygen/moisture was significantly suppressed after F-PMAI addition (Fig. S3d), and the hydrophobicity of perovskite film is also enhanced (Fig. S4), suggesting that the hydrophobic benzene ring with F ending groups structure is benefit to mitigating the degradation of perovskite [33]. To further investigate the interaction between F-PMAI and organic species in perovskites, we performed FTIR spectrometry of F-PMAI, FAI and F-PMAI/FAI mixture, respectively. As shown in Fig. 1b, the N-H stretching peak in F-PMAI and FAI is 3322 and 3325 cm⁻¹, respectively, indicating almost the same position. However, it significantly shifted from 3325 to 3334 cm⁻¹ for F-PMAI/FAI, and the C=N character stretching peak in FAI also shifted slightly from 1723 to 1722 cm⁻¹ on adding F-PMAI, suggesting that the F-PMAI molecules form hydrogen bonds with organic cation FA⁺. The hydrogen bond consists of an electrostatic interaction between the hydrogen from organic salt and electronegative F (N-H···F) [30], which can effectively stabilize organic cations and reduce the generation of A-site vacancies.

SEM and XRD were performed to reveal how F-PMAI affects perovskite film quality. The PbI₂ and CsCl were co-evaporated to form inorganic precursor films (Fig. S5a, b) and exhibit lead iodide (PbI₂)-dominated crystalline structure, as confirmed by the XRD peak near 12.5° (Fig. S5c). Before forming the final perovskite films, three stages including organic salts deposition, pre-annealing and air-annealing were gone through. As illustrated in Fig. 1c, the air-annealing final perovskite film grains with F-PMAI were significantly enlarged than the control film. We also found perovskite films without and with F-PMAI exhibit different morphologies at organic salts deposition and pre-annealing stages as shown in Fig. S6. The control film shows a compact and randomly distorted morphology after drop-coating organic salts. Subsequently, the film further crystallized during the pre-annealing process and surface morphology has hardly changed (Fig. S6a), while for the additive film shows porous morphology after the addition of organic salts, and crystals began to precipitate in Fig. S6b. The difference in morphology demonstrated the much slower phase conversion for additive film, and the time of color change from yellow to black gave another evidence (Fig. S7). Notably, the peak area ratios of (001) PbI₂/(100) perovskite for the additive film are larger than the control film at the stage of organic salt deposition and pre-annealing as shown in Fig. S8, indicating that a slower crystalline process for the additive film. This delayed nucleation kinetics can be attributed to the hydrogen bond interaction between F⁻ and FA⁺, thereby slowing down the reaction of PbI₂ and FAI. Moreover, we found the peak positions of additive perovskite films are similar to those of control films, indicating that the F-PMAI does not enter the lattice and change the lattice parameters of the perovskite (Fig. 1d). There are no other diffraction peaks at low angles (2θ < 10°). Therefore, the addition of F-PMAI does not lead to the formation of 2D perovskite phase. The absence of 2D perovskite is beneficial for charge carrier transport and extraction throughout the perovskite absorber. ToF-SIMS were performed to observe the distribution of F-PMAI molecules in perovskite films, based on the structure ITO/Spiro-TTB/perovskite (F-PMAI)/C₆₀/ALD SnO₂ (Fig. 1e). Interestingly, a rising peak of F signal at the perovskite and Spiro-TTB interface is observed. This result suggests that the F-PMAI molecules are mostly extruded to the buried interface. This is related to the stronger Lewis acid-base interactions between the F-PMAI and Spiro-TTB. In short, the F-PMAI with large bulk cation cannot incorporate into the lattice, and be repelled to bottom and top interface, severing as a defect passivator via passivating undercoordinated Pb²⁺/I⁻ and A site vacancies defects. Hence, the schematic diagram of the F-PMAI interaction mechanism is shown in Fig. S9. F-PMA⁺ passivates uncoordinated Pb²⁺/I⁻ on the perovskite, inhibiting the production of deep-level defects and reducing the non-radiative recombination centers. F⁻ interacts with organic cations to form hydrogen bonds. In addition, I ions in F-PMAI also possibly fill the iodine vacancies [34]. Above, we propose that the F-PMAI retards crystallization by hydrogen bond interaction, and the larger steric hindrance of F-PMAI hinders the immediate reaction between the organic salt (e.g., FAI) and the precursor film (e.g., PbI₂), improving the grain size. In terms of additive film, a higher critical Gibbs free energy (ΔG_c) to nucleate contributes to fewer crystal nuclei and growth in more room into larger grains. This agrees with the appearance of a porous morphology at the film initial growth stage (Fig. S6b). The slow crystallization in additive film can provide a longer diffusion distance and self-assembly time between precursor salt ions and organic molecules, conducive to the formation of more complete reactions and larger grains as shown in Fig. S10.

In Fig. 1d, we notice an obvious (111) facet preferred orientation exhibited in the perovskite film with F-PMAI. The intensity ratio of (111)/(100) perovskite in the XRD pattern has been calculated as shown in Fig. S11. With the increase of F-PMAI concentration, the preference degree of (111) orientation also increases (Fig. S12). To further investigate the effect of the F-PMAI dopant on orientational preference of perovskite film, we performed GIWAXS measurements. The 2D diffraction patterns are shown in Fig. 2a, b. For the control perovskite film, the (100) plane showed the out-of-plane direction and other strong orientations, indicating irregular crystal orientation, in agreement with the above XRD and SEM discussed results. In contrast, the preferred orientation of (100) planes in the additive perovskite films changes from the out-of-plane (~ 90°) to ~ 45°-oriented direction, according to the integrated GIWAXS intensity plot azimuthally along the ring at a q value approximate to 10 nm⁻¹ (Fig. S13). Simultaneously, the (111) planes preferentially orientate in the out-of-plane direction (Fig. 2b) [35]. This may be related to the interaction between F-PMAI and perovskite, resulting in a substantial crystal facet rotation along (111) out-of-plane directions [36]. We further conducted density functional theory (DFT) calculations to calculate the adsorption energy (E_ads) of F-PMAI molecules on the (100) and (111) facets of perovskite, respectively. Shown in Fig. 2c, the calculated results indicate that the F-PMAI molecules have higher interaction energy with (111) than (100) facets (|E_ads, ₍₁₁₁₎| >|E_ads, ₍₁₀₀₎|), suggesting that the additive molecules are preferred to be adsorbed on (111) facets. The preferred adsorption of F-PMAI molecules can reduce the surface energy and facilitate the crystal growth of the (111) facet, leading to the perovskite film with exposed (111) facets [27].

Figure 2d shows the schematic illustration of the perovskite crystal growth process. After spin-coating organic salts on the inorganic precursor film, compared with the control, the film with F-PMAI initially forms fewer nuclei. In the subsequent pre-annealing stage, due to the presence of the bulky F-PMAI and its hydrogen bond interaction with the FA⁺, the organic salts react slowly with the inorganic salts in the precursor film, forming large-sized perovskite grains and preferred adsorption on the (111) facet to form (111) facet dominated perovskite crystal, which leads to grains stacking along the out-of-plane direction. During the subsequent air thermal annealing, the hydrogen bonds were broken, and the large F-PMAI molecules were repelled to the bottom and little to the top surface and the grain boundaries, severing as defects passivator. In this case, F-PMAI changes the nucleation and growth pathways to give a strong growth orientation perpendicular to the substrate surface, as shown in the cross-sectional SEM images of Fig. S10b, which promises to improve the film crystal quality, reduce film defects, facilitate the transport of charge carriers and achieve higher device performance.

We measured the PL and TRPL to explore how F-PMAI affects perovskite film quality. Perovskite film with F-PMAI exhibits stronger PL intensity and longer carrier lifetime from 252.20 to 536.38 ns than control film (Fig. S14a, b and Table S1), indicating better crystal quality and suppression of non-radiative charge recombination in film with F-PMAI. Furthermore, the PL mapping test was used to better understand the effect of F-PMAI on charge carrier extraction and transport dynamics. Perovskite film with F-PMAI deposited on ITO substrate exhibited a higher PL intensity than control film (Fig. 3a, b). The results show that F-PMAI significantly decreased defect density and suppressed trap-assisted non-radiative recombination. This is further confirmed by the densities of trap states (N_t) of perovskite film using SCLC measurement, with the structure of ITO/SnO₂/perovskite with or without F-PMAI/C₆₀/Ag for electron-only device, as shown in Fig. 3c. Compared to the V_TFL value of the control film (1.26 V), the film with F-PMAI has a smaller V_TFL value of 0.88 V. The N_t of samples without and with F-PMAI is 2.60 × 10¹⁶ and 1.81 × 10¹⁶ cm⁻³, respectively [32]. The lower trap density can be attributed to the decreased grain boundaries and associated defects by enlarged grain size and the reduced undercoordinated Pb²⁺/I⁻ defects via passivation of F-PMAI. Considering that F-PMAI is distributed at both bottom and top of the perovskite film, we also investigated the charge extraction between perovskite film and charge transport layer. Perovskite film with F-PMAI deposited on ITO/Spiro-TTB (hole transport layer (HTL)) substrate and ITO/perovskite with F-PMAI/C₆₀ (electron transport layer (ETL)) both exhibit a PL quenching than control film, indicating enhanced charge transfer from perovskite to HTL/ETL, respectively (Fig. S15) [36,37]. It is expected to improve interfacial contact and effectively passivate interface defects. Moreover, the impact of F-PMAI on charge carrier dynamics was investigated using C-AFM. Compared to the control film, the dark current is significantly and uniformly enhanced in the perovskite film with F-PMAI, with the average current value increasing from 96.4 to 113 pA (Fig. 3d, e). This enhancement indicates more efficient charge transport, which can be attributed to the superior electrical properties of (111) facets as reported previously [38]. The KPFM was measured to evaluate the distribution of the surface potential of perovskite films (Fig. S16a, b), which shows a decrease of work function (WF) in the perovskite film with F-PMAI, indicating a more n-type nature. This change matches well with the results obtained from UPS, showing the Fermi level (E_F) closer to the conductive band edge (Fig. S17). Figure 3f shows the energy-level scheme of perovskite films without and with F-PMAI. The more n-characteristic film surface stems from the improved crystal quality and modified surface termination after adding F-PMAI, enabling more efficient electron extraction between the perovskite and ETL [39,40].

We fabricated CsFAMAPbI_3−xBr_x PSCs based on inverted p-i-n architecture ITO/Spiro-TTB/perovskite/C₆₀/BCP/Ag to evaluate the effects of the additive on device performance. The preparation process of perovskite solar cells shown in Fig. S1 and the other layers are prepared by a thermal evaporation process. The optimized concentration of F-PMAI is 1.5 mol% (Fig. S18). Increasing the concentration of F-PMAI enhances the (111) orientation, but it can also cause a large number of F-PMAI to hinder sufficient reaction and accumulate in the interface, leading to the precipitation of PbI₂ as shown in Fig. S12. These features ultimately lead to a decrease in device performance, the PV parameters of perovskite solar cells are shown in Fig. S18. Encouragingly, F-PMAI substantially improved the open-circuit voltage (V_OC) and fill factor (FF) of devices. A champion device using F-PMAI exhibits V_OC up to 1.15 V, a gain of 50 mV compared with the control device in Fig. 4a and Table 1. Accompanied by a noticeable FF improvement (77.43% to 85.48%), the champion F-PMAI device achieved a high PCE of 22.74%. The integrated current density with F-PMAI and without F-PMAI from the EQE is 22.4 and 22.2 mA cm⁻² (Fig. 4b), respectively, which well match with the J_sc extracted from J-V curves. The hysteresis index (HI) is defined as ((PCE_reverse- PCE_forward)/PCE_reverse × 100%), decreasing from 7% in the control to 2% in the additive device. The stabilized power output (SPO) of the F-PMAI device is 21.2% at the maximum power point (MPP) based on a V_max of 0.98 V (Fig. 4c). We investigated the light stability of PSCs with F-PMAI using an LED solar simulator with an intensity equivalent to 1-sun illumination at 25 °C in ambient conditions. The unencapsulated F-PMAI device maintained 95% of initial PCE after 600 h under continuous light soaking (Fig. S19). This reduced hysteresis and increased photostability may benefit from reduced defects and suppressed ion migration, so we calculated the ion migration activation energy (E_a) using temperature-dependent conductivity (Fig. S20). The additive perovskite film is found to show a higher E_a of 262 meV compared with the control (199 meV). This indicates that ions are harder to migrate in the perovskite layer with F-PMAI than in the control [27,36,38,41].

To further investigate the reason for the improvement in device efficiency, we systematically studied the charge carrier behavior and defect physics of devices without and with F-PMAI. First, we performed capacitance-voltage (C-V) measurements to directly compare the built-in potential (V_bi) of devices. The control device has a V_bi of 0.97 V, while the additive device has a higher V_bi of 1.00 V, in agreement with the V_oc enhancement shown in Fig. 4d. The larger the V_bi is, the more favorable for carrier separation and transfer, which reduces the charge accumulation at the perovskite/C₆₀ interface, reducing the hysteresis [32]. We attribute the increased potential to defect passivation and improved energy level at the perovskite/C₆₀ interface. The TPV measurement result in Fig. 4e shows that the carrier lifetime significantly increases from 2.3 to 3.45 ms after adding the F-PMAI, which further proves the suppressed carrier non-radiative recombination [42,43]. To identify the effect of F-PMAI additive on FF of the device, we investigated the electrical conductivity of perovskite film without and with F-PMAI by vertical dark J-V measurements with a structure of ITO/perovskite/Ag (Fig. 4f). Adding the F-PMAI additive resulted in higher electrical conductivity and yielded a higher FF [44]. We calculated the ideally factor (m) of the device in the dark J-V curves (Fig. 4g). The value reduces from 1.48 in the control to 1.35 in the device with F-PMAI. This suggests suppressed non-radiative recombination [45], which is consistent with the increased recombination resistance (R_rec) in the EIS (Fig. 4h). It is well known that the R_rec values can reflect the degree of charge recombination in a device, which is typically correlated to the defect state density of the perovskite film. In general, a larger R_rec signifies a more difficult charge carrier recombination and results in a higher V_OC [46]. To further elucidate the charge carrier recombination dynamics of PSCs, we conducted light intensity-dependent V_OC measurement (Fig. 4i). Compared with the slope of 1.88 KT/q for the control device, the slope decreases to 1.61 KT/q for the additive device, revealing that the defect-assisted recombination is effectively suppressed due to reduced trap density [28]. The above calculation methods can be found in Supporting Information. The above results show that F-PMAI improves the perovskite film crystalline quality, passivates interface defects, effectively suppresses non-radiative recombination, and enhances carrier transport properties of devices, which leads to improved J_SC, V_OC and FF.

Inspired by the substantial enhancement in single-junction devices using F-PMAI, we used our additive perovskite films to fabricate monolithic perovskite/silicon TSCs. The Si bottom cells are industrially feasible double-side fully textured silicon solar cells. The detailed fabrication process is described in the experimental section, and the schematic diagram of the tandem structure is illustrated in Fig. 5a. The enlarged grain size can also be found in the tandem devices with F-PMAI (Figs. 5b and S21). Energy-dispersive X-ray spectroscopy (EDS) mapping shows that F-PMAI molecules are distributed on the bottom and top of perovskite films, suggesting similar crystal and passivation effects can also be produced on textured silicon bottom cells (Fig. S22). Therefore, this additive can also be explored for improving TSCs efficiency as shown in Fig. S23, F-PMAI significantly enhances the V_OC and FF of TSCs as well as improves PCE. Encouragingly, TSC with F-PMAI reaches a reverse-scan PCE of 30.05% with a V_OC of 1.81 V, a J_SC of 20.01 mA cm⁻², and an FF of 82.91% as shown in Fig. 5c and Table 2. The efficiency represents the best result reported to date for industrially feasible textured silicon tandem solar cells based on nanocrystalline silicon tunneling junction (nc-Si:H) (Table S2). The HI decreases from 13% in the control to 2% in the additive TSCs, attributed to the improved crystal quality and defect passivation in the perovskite films with F-PMAI. MPP tracking yields a steady-state PCE of 29.4% up to 1500 s (Fig. 5d). The EQE spectra show integrated J_SC of 20.1 and 19.9 mA cm⁻² for top and bottom subcells, respectively, basically consistent with the J-V results (Fig. 5e).

Long-term stability is important to the industrialization of perovskite photovoltaics. Therefore, we investigated the effects of F-PMAI on the long-term stability of TSCs. The thermal stability of unencapsulated tandem devices was performed on a hot plate at 85 °C in a dark nitrogen-filled glovebox. The control device reduced to less than 80% of initial efficiency after 24 h and fluctuated, whereas that of the F-PMAI device showed a very small PCE decay and retained 90% of initial PCE after 1650 h, indicating improved thermal stability (Fig. 5f). We further tracked the long-term light stability of tandem devices in Fig. 5g, which shows that the PCE of the unencapsulated devices with F-PMAI still retained its initial PCE after 700 h tracking in ambient conditions (RH: 30%-40%, 25 °C). The enhanced stability of TSCs with F-PMAI is attributed to better perovskite film quality and all-around defect passivation. If more efficient encapsulation techniques and optimal perovskite compositions are used, further improved efficiency and stability of TSCs can be expected to meet the requirements for practical applications.