Perovskite solar cells (PSCs) have shown great commercial potential, due to their high power conversion efficiency (PCE)^[1,2], low cost and solution- and air-based fabrication processes^[3–7]. One of the challenges in the commercialization of PSCs is to scale them from the laboratory level to large area fabrication^[8]. As the most commonly employed fabrication process in laboratory, spin coating has a low utilization of raw materials and cannot obtain uniform films in a large area^[9–11]. With the advantages of fast deposition speed, high raw material utilization and compatibility with nonplanar surfaces, spray-coating has been demonstrated as a promising scalable film deposition process for the large area fabrication of PSCs^[12,13].

Spray coating PSCs with uniform perovskite films, controllable morphology and reproducible device performance has been the primary focus of research, and many approaches (e.g., hot-air blowing treatment^[14], plasma treatment^[15], low-vacuum treatment^[16], anti-solvent bath treatment^[17], and air knife treatment^[18]) have been developed to this end. These post-treatment processes inevitably complicate the device preparation, increase cost, and fail to achieve high device performance due to the presence of intrinsic defects within the spray-coated perovskite layer^[19]. The defects in perovskite structures, including vacancies, interstitials, and anti-site defects, are non-radiative recombination centers that impair device performance^[19–21]. Among them, vacancy defects such as cation (MA⁺/FA⁺) vacancies and halogen (I⁻) vacancies are most likely to occur due to their lowest formation energy^[22]. Moreover, cation vacancies can induce other negatively charged defects, such as undercoordinated I⁻ and Pb-I anti-defects ( PbI3−), further increasing the defect density in perovskite films^[23]. Furthermore, vacancy defects provide channels for ion migration, which is considered to be one of the main reasons for device hysteresis^[24–26].

Additive engineering is an efficient method to passivate defects in perovskite films^[27]. Zhaoet al. reported that the O atom in polyethylene glycol (PEG) additive can form a hydrogen bond with MA⁺ in perovskite to stabilize MA^+[28]. Similarly, Liet al. used dipentaterythritol pentaacrylate (DPPA) additive to passivate vacancy defects, and the O and hydroxyl (–OH) groups of DPPA are suggested to form hydrogen bonds with FA⁺ and I⁻ in perovskites respectively, thereby inhibiting the generation of vacancy defects^[29]. The F atom of NaF has also been demonstrated to be able to form strong hydrogen bonds with MA⁺/FA⁺, thus passivating cation vacancies^[30]. As a result, the incorporation of additives in perovskites has been found to reduce or eliminate device hysteresis, especially potassium salts such as KI, KI₃ and KPF₆, which is attributed to the effective inhibition of ion migration by K^+[31–35]. Additive engineering, without introducing any additional processing steps, is simple and convenient, and fully conforms to the characteristics of high throughput and fast deposition of spray-coating.

In this work, we employed (4-methoxyphenyl) potassium trifluoroborate (C₇H₇BF₃KO) as an additive to passivate defects in one-step spray-coated (FAPbI₃)_x(MAPbBr₃)_1–x perovskite films to realize efficient devices without hysteresis. It is shown that hydrogen bonds can form between F in BF₃⁻ of C₇H₇BF₃KO and H in the amino group of the perovskite cation MA⁺/FA⁺, thereby stabilizing MA⁺/FA⁺, inhibiting the occur of A-site cation vacancies, reducing non-radiative recombination and improving the performance of the device. Moreover, K⁺ in C₇H₇BF₃KO and the reduced MA⁺/FA⁺ vacancies due to hydrogen bonding interactions can suppress migration, thereby eliminating hysteresis. Ultimately, we obtained hysteresis-free spray-coated device with a maximum PCE (PCE_max) ofca. 19.5% from both forward and reverse scans. Our work demonstrates that defect passivation by additives engineering is an effective way to further improve the performance of spray-coated devices, paving the way for large-scale fabrication.

Lead(II) iodide (PbI₂) was purchased from TCI. (4-methoxyphenyl) potassium trifluoroborate (C₇H₇BF₃KO) was purchased from Macklin. Thiourea (CH₄N₂S),γ-butyrolactone (GBL), tin(II) chloride dihydrate (SnCl₂·2H₂O) and 4-tert-butylpyridine (TBP) were purchased from Aladdin. L-α-phosphatidylcholine (L-α-P), bis(trifluoromethane)sulfonimide lithium salt (Li-TFSI) and other solvents were purchased from Sigma-Aldrich. Other materials were all purchased from Xi’an Polymer Light Technology and used as received.

The ITO glass was ultrasonically cleaned with deionization, ethanol and isopropanol in sequence for 15 min, and treated with UV-Ozone for 15 min. SnCl₂·2H₂O and CH₄N₂S were dissolved in deionized water in a molar ratio of 1 : 1 to obtain a SnO₂ solution with a concentration of 45 mg/mL. The SnO₂ solution was spin-coated onto ITO glass at 3000 rpm, then annealed at 200 °C for 1 h, and then treated with UV-Ozone for 40 min. PbI₂ (835 mg), FAI (238 mg), MABr (34 mg) and MACl (19 mg) were dissolved in a mixed solvent of DMSO (1950μL), GBL (1950μL) and IPA (100μL). 0.02 wt% L-α-P and 0.5 wt% C₇H₇BF₃KO were added to the perovskite precursor solution. Perovskite films were fabricated by Siansonic UC342 spray-coater and detailed spraying parameters are: the nitrogen flow pressure is 0.5 MPa, the substrate temperature is 60 °C, the flow rate is 0.7 mL/min, the height and moving speed of nozzle is 60 mm and 900 mm/min, the ambient relative humidity is less than 40% and the ambient temperature is about 25 °C. Then, the perovskite wet films were annealed at 150 °C for 25 min. In an N₂-filled glovebox, the Spiro-OMeTAD solution (95 mg/mL in CB) containing 30μL TBP, 30μL FK209 (400 mg/mL in acetonitrile) and 11.5μL Li-TFSI (400 mg/mL in acetonitrile) were spin-coated onto spray-coated perovskite films at 4000 rpm. Ultimately, the Ag electrodes were evaporated onto Spiro-OMeTAD and the fabrication of devices is completed.

Optical and AFM images were obtained by BX51 microscope (Olympus, Japan) and atomic force microscopy (NT-MDT, Russia), respectively. The light intensity of the Newport 3A solar simulator was calibrated with NREL certified silicon solar cells to 100 mW/cm², andJ–V curves were obtained by usingJ–V scanning software of Ossila Ltd. (UK) and 2612B source meter unit of Keithley (USA). EQE measurements were carried out by EQE system (Zolix, China).¹H NMR spectra were measured by¹H NMR spectrometer (DRX500, Bruker, Germany). XRD spectra were obtained by X-ray diffractometer (D8 Advance, Bruker, Germany). UV–vis absorption was measured by UV–vis spectrophotometer (Hitachi U-3900H, Japan). PL spectra were obtained by PL microscopic spectrometer (Flex One, Zolix, China). TRPL decay curves were obtained through a time-correlated single photon counting spectrofluorometer (PicoQuant, Germany). Mott-Schottky plots and EIS measurements were conducted through ModuLab XM electrochemical workstation (AMETEK, UK). Equivalent circuit simulations were carried out by the ZView software package (Scribner, USA).

The (FAPbI₃)_x(MAPbBr₃)_1–x perovskite active layers were fabricated by one-step spray-coating in air (Fig. 1(a)), using precursors described in the experimental section, where surfactant L-α-phosphatidylcholine (L-α-P) was incorporated because it is critical to adjust the surface energy to receive uniform perovskite films during spray coating^[36,37]. However, serious hysteresis exist in the L-α-P modified PSCs^[37]. We have demonstrated that a secondary additive KI can help to reduce hysteresis of spray-coated PSCs, although it cannot completely eliminate hysteresis^[37]. C₇H₇BF₃KO was chosen in this work as a defect passivator for the spray-coated PSCs (Figs. 1(b) and1(c)). The morphology of spray-coated perovskite films with or without the presence of C₇H₇BF₃KO were first examined and shown inFig. 2. The optical images (Figs. 2(a) and2(e)) show that the grain sizes of the reference and C₇H₇BF₃KO based perovskite films are similar of over 100μm in diameter. However, further morphology characterization using atomic force microscopy (AFM) in smaller scale shows that the spray-coated perovskite films are hierarchical, with the large grains observed inFigs. 2(b)–2(d) and2(f)–2(h) being composed of small grains ofca. 400 nm in diameter, which is consistent with our previous report^[37]. The introduction of C₇H₇BF₃KO brings no apparent changes to the size of both large- and small- grains. However, the C₇H₇BF₃KO based perovskite film (Fig. 2(h)) shows a smoother surface with a lower root-mean-square (RMS) roughness of 18.8 nm, compared to 27.0 nm for the reference film (Fig. 2(d)), which is conducive to improving the interface contact.

The X-ray diffraction (XRD) patterns inFig. 3(a) show that, compared with the reference perovskite film, the C₇H₇BF₃KO based perovskite film displays stronger diffraction peak intensities at 13.9°, 28.1°, 31.5°, 40.2° and 42.8°, which correspond to the (110), (220), (310), (224) and (330) crystal plane of perovskite, respectively. The enhancement of peak intensity represents the increased crystallinity, which is conducive to enhance the light absorption of perovskite films and improve the performance of the device^[38]. There are no new peaks or peak shifts, indicating that the perovskite phase structure remains consistent and C₇H₇BF₃KO cannot enter the perovskite lattice^[39]. The UV−vis absorption spectra (Fig. 3(b)) confirm the slightly better absorption ability of the C₇H₇BF₃KO based perovskite film, especially at the wavelength of 450–550 nm. The steady-state PL spectra is shown inFig. 3(c), the PL intensity of C₇H₇BF₃KO based perovskite film is significantly enhanced compared with the reference film, indicating that C₇H₇BF₃KO effectively passivates the defects and non-radiative recombination is significantly suppressed^[40,41]. The time-resolved photoluminescence (TRPL) decay curves (Fig. 3(d)) of spray-coated perovskite films show bi-exponential decays. The decay curves are fitted by the equation:i =A₁exp(−t/τ₁)+A₂exp(−t/τ₂), whereτ₁ is the fast decay time associated with trap-assisted non-radiative recombination,τ₂ is the slow decay time related to bimolecular radiative recombination,A₁ andA₂ are decay amplitudes^[42,43]. The detailed fitting parameters are shown inTable 1. With the incorporation of C₇H₇BF₃KO, the values ofτ₁ andτ₂ both increased, and the average PL lifetime was improved from 201 to 294 ns. This longer carrier lifetime further demonstrates the lower trap density and improved quality of spray-coated perovskite films after passivation by C₇H₇BF₃KO^[44].

The hydrogen bonding interactions between C₇H₇BF₃KO and the cation MA⁺/FA⁺ in perovskite were investigated through¹H nuclear magnetic resonance (¹H NMR) spectra. As shown inFigs. 3(e)–3(f), after the addition of C₇H₇BF₃KO, the chemical shifts (δ) of H of –NH₂⁺ in FAI and –NH₃⁺ in MABr move from 8.78 and 7.67 ppm to 8.67 and 7.56 ppm respectively as a result of the formation of hydrogen bonds, which agrees with literature reports^[28,29]. The H of –OCH₃ in C₇H₇BF₃KO has no chemical shift, indicating that O is not involved in the formation of hydrogen bonds. Therefore, we think that the F in BF3− can form hydrogen bonds with the H on the amino group in MA⁺/FA⁺, which can effectively stabilize organic cations, reduce the generation of A-site vacancies, thereby improving the performance of the spray-coated devices^[30].

The PSC devices fabricated in this work have an n-i-p structure of ITO/SnO₂/Perovskite/Spiro-OMeTAD/Ag. The effects of C₇H₇BF₃KO on the photovoltaic performance of spray-coated PSCs are shown inFig. 4(a) andTable 2. The optimization process of the additive amount of C₇H₇BF₃KO is shown in Table S1, from which the optimum content of 0.5 wt% of C₇H₇BF₃KO can be determined. The PCE_max of the reference device is 16.3% with a fill factor (FF) of 67.1%, a short-circuit current density (J_SC) of 21.9 mA/cm² and an open-circuit voltage (V_OC) of 1.11 V. After the addition of 0.5 wt% C₇H₇BF₃KO, the average PCE (PCE_ave) of PSCs increases significantly from 14.7% to 17.6%. The PCE of C₇H₇BF₃KO based champion device reaches 19.5%, with the improvement mainly attributing to enhanced FF andV_OC, reaching 77.8% and 1.14 V, respectively. The steady-state power output of the spray-coated devices at the maximum power point (MPP) is shown in Fig. S1. The PCE of the reference device shows a gradual decrease with time that finally stabilizes at 15.2%, which is likely caused by hysteresis^[45], whilst the PCE of the C₇H₇BF₃KO based device stabilizes at 18.9% for the whole duration of the MPP testing. The hysteresis indices (HI = (PCE_reverse −PCE_forwad)/PCE_reverse) for the reference and C₇H₇BF₃KO based devices are 8.6% and 0.5%, respectively, indicating that hysteresis is negligible with the presence of C₇H₇BF₃KO. This is attributed to the effect of K⁺ and hydrogen bond-stabilized MA⁺/FA⁺, which effectively suppresses ion migration and eliminates hysteresis^[31–35].Fig. 4(b) shows the external quantum efficiency (EQE) spectra of the champion devices. The corresponding integratedJ_sc values of the reference and C₇H₇BF₃KO based devices are 21.3 and 21.5 mA/cm² respectively, which are close to the values obtained from theJ–V curves, demonstrating the accuracy ofJ–V measurements.

We then studied the optoelectronic properties of PSCs with and without C₇H₇BF₃KO to explainJ–V performance improvement. The built-in potential (V_bi) of PSCs can be obtained from Mott–Schottky curves (Fig. 4(c)), the value ofV_bi for the reference and C₇H₇BF₃KO based PSCs are 1.13 and 1.17 V, respectively. The enhanced built-in potential provides a larger driving force for charge transport and collection, resulting in reduced carrier recombination and higherV_OC^[46,47]. In addition, the darkJ–V curves of devices are shown inFig. 4(d), compared with the reference PSC, the dark current density of C₇H₇BF₃KO based PSC is reduced by one order of magnitude. The reduction of leakage current density indicates a lower defect density that contributes to a higher FF, which further confirms the passivation effect of C₇H₇BF₃KO^[48,49].

We then carried out space charge limited current (SCLC) measurements (Fig. 4(e)) based on electron-only devices with a structure of ITO/SnO₂/perovskite/PC₆₁BM/Ag to evaluate the trap density of devices with and without C₇H₇BF₃KO. The trap density can be estimated by the equationn_t = 2εε₀V_TFL/eL², whereV_TFL is trap-filled limit voltage defined as the intersection voltage between the Ohmic-type response region and the trap-filled limit region,ε₀ is the vacuum permittivity,ε represents the relative dielectric constant,L means the film thickness (ca. 600 nm), ande refers to elementary charge of the electron^[50–52]. Compared with the reference device, the C₇H₇BF₃KO based device exhibits a reducedV_TFL from 1.66 to 1.18 V, and the calculated trap densities are 1.63 × 10¹⁶ and 1.16 × 10¹⁶ cm⁻³, respectively, indicating that the defects of PSCs are reduced after the introduction of C₇H₇BF₃KO. Therefore, the non-radiative recombination is effectively suppressed and the power conversion performance of PSC is enhanced^[53].

Finally, the electrochemical impedance spectroscopy (EIS) was measured under dark conditions at a voltage bias of 1 V to evaluate charge transport in PSCs. The Nyquist plots are shown inFig. 4(f), and the series resistance (R_s) and recombination resistance (R_rec) can be obtained by fitting with the inserted equivalent circuit model and the fitting parameters are shown inTable 3. After the incorporation of C₇H₇BF₃KO, the PSC displays a decreasedR_s from 156 to 136 Ω cm², and an increasedR_rec from 15 886 to 18 972 Ω·cm², indicating more efficient charge transport and decreased carrier recombination, which resulted in improved device performance^[50,54].

In conclusion, we have employed C₇H₇BF₃KO as an additive to passivate defects in spray-coated perovskite films for high photovoltaic performance. The F of C₇H₇BF₃KO forms hydrogen bonds with MA⁺/FA⁺ in perovskite to stabilize A-site cations and suppress the formation of cation vacancies, thereby reducing defect density and improving devices performance. Further, K⁺ of C₇H₇BF₃KO and reduced cation vacancies as a result of hydrogen bonding interactions can inhibit ion migration and eliminate hysteresis. The passivation of C₇H₇BF₃KO enables us to fabricate efficient and hysteresis-free spray-coated PSCs with a PCE_max of 19.5% in air. Our work demonstrates that simple and convenient additive engineering has huge potential for improving the quality of spray-coated perovskite films, providing a feasible scheme to further improve the optoelectronic properties of spray-coated PSCs.

The authors acknowledge funding supports from the National Natural Science Foundation of China (51861145101).

Supplementary materials to this article can be found online athttps://doi.org/10.1088/1674-4926/43/9/092201.