Modulating CsPbl3 crystallization by using diammonium agent for efficient solar cells

Cesium lead triiodide (CsPbI3) perovskite receives tremendous attention for photovoltaic applications, owing to its remarkable thermal stability and optoelectronic properties. However, realizing the CsPbI3 perovskite with high black‐phase stability and optoelectronic properties remains a significant challenge, which largely affects the photovoltaic performance of perovskite solar cells (PSCs). Herein, aromatic ammonium agents are used to modulate the crystallization of the CsPbI3 perovskite to improve its black‐phase stability and optoelectronic properties for efficient PSCs. Systemically experimental studies and comprehensively theoretical calculations are performed, which reveal that histammonium dihydrochloride (HACl2) could strongly couple with the perovskite during its crystallization, leading to faster nucleation and slower perovskite growth, and thus modulating the crystallization dynamics of the perovskites. Moreover, the residual diammonium cations (HA2+) distributed at the grain boundaries and on the surface of the perovskites can effectively passivate defects through electrostatic interactions, substantially suppressing trap‐assisted nonradiative recombination, and prompting more matched perovskite surface energetics. Consequently, the photovoltaic performance of CsPbI3 PSCs is largely improved because of a combination of enhanced crystallinity and optoelectronic properties of the perovskites. This work offers a new avenue to prepare inorganic perovskites with high optoelectronic properties for photovoltaics.

9][20][21] In the last couple of years, even though the PCE of inorganic PSCs was largely improved, 22,23 their efficiency still largely lags behind the hybrid PSCs because of the large open-circuit voltage deficit, 24,25 which could be mainly assigned to the unsatisfied quality of inorganic perovskites, such as low black-phase stability, numerous defects, and poor crystallinity.
Various strategies, such as compositional engineering, [26][27][28] interface engineering, [29][30][31] and crystal secondary growth, 32,33 were explored to improve the photophysical properties of CsPbI 3 perovskites and thus the efficiencies of PSCs.8][39][40] Therefore, different organic ammonium halides, such as phenylethylammonium iodide 37 or phenyltrimethylammonium iodide, 22 were explored as crystallization agents to modulate crystallization process during perovskite film formation and simultaneously passivate defects within the perovskite films, thereby improving the photophysical properties of the perovskites, and promoting the efficiency and stability of PSCs.However, the ammonium groups could directly interact with the perovskite crystals, and the hydrophobicity of organic groups generally leads to orientation stacking, forming layered structures.The layered structures could inhibit charge transport and might deteriorate surface passivation. 413][44][45] Whereas, the ammonium agents are limited to monoammonium-based agents, and the diammonium-based agents were rarely studied for CsPbI 3 perovskites.Besides, a fundamental understanding of the crystallization kinetics of CsPbI 3 perovskites can offer important insights into the crystallization modulating of CsPbl 3 perovskites for high-performance PSCs.
In this work, aromatic ammonium chlorides, phenylethylammonium chloride (PEACl) and histammonium dihydrochloride (HACl 2 ), were applied to modulate the crystallization of CsPbI 3 perovskites.Comprehensive theoretical calculations and extensive experimental studies were conducted, which revealed that HACl 2 with two ammonium groups (─NH 3 + and imidazolium) could strongly couple with the perovskites, allowing for modulation of the crystallization dynamics of perovskite films.Meanwhile, the residual diammonium HA 2+ cations distributed at the grain boundaries and on the surface of the perovskites can effectively passivate defects through electrostatic interactions and prompt more matched perovskite surface energetics.Consequently, the HACl 2 -based PSC yielded an efficiency of 19.67%, which was greatly improved compared with the control PSC, with an efficiency of 17.02%.The charge carrier dynamics were systemically studied to figure out the causes, leading to the improved performance in HACl 2 -based PSCs, which shows that the high performance in HACl 2 -based PSCs corresponded to suppressed nonradiative recombination as the results of improved crystallization of the perovskites.

Structure and morphology of perovskites
Rapid nucleation and slow crystal growth processes are highly needed to modulate the crystallization of perovskites, realizing high-quality perovskite films for PSCs.In this work, aromatic ammonium chlorides, PEACl and HACl 2 , were used as crystallization agents in the perovskite precursor solutions to modulate the perovskite crystallization.The details are demonstrated in the Experimental Section.Notably, considering the positive effect of chloride ions in improving the crystallization of perovskites, [46][47][48] chlorine ions were selected as anions for the ammonium salts.Figure S1 shows the molecular structures of PEACl and HACl 2 .To explore the influences of different structures of ammonium on the electronic structure, electrostatic potential (ESP) maps of phenylethylammonium (PEA + ) and histammonium (HA 2+ ) were calculated.As displayed in Figure 1A, the positive charges mainly locate on the ammonium groups, which could couple with [PbI 6 ] 4− octahedra frameworks to passivate Cs + vacancy defects.Different from the PEA + cation, the diammonium HA 2+ cation contains the dual functionality of a primary ammonium and an imidazolium group, indicating that HA 2+ could act as a "double linker" for adjacent perovskite crystals.
The ammonium chloride agents may affect the crystallinity of perovskites, which largely affects the performance of PSCs.As such, x-ray diffraction (XRD) patterns of perovskite films were measured to study the effects of aromatic ammonium chlorides on the crystallinity of perovskite films.For comparison, the perovskite film without any additives in the precursor was also prepared.As depicted in Figure 1B, these perovskite films exhibit two strong diffraction peaks at 14.3 • and 28.9 • , corresponding to the (110) and (220) planes of the tetragonal β-phase CsPbI 3 perovskite, respectively. 49,50Notably, no additional peak and the shift in peaks were observed, suggesting that the organic ammonium cations and Cl − ions did not incorporate into the perovskite lattice and no 2D perovskites were formed.However, PEACl-and HACl 2 -based perovskite films show higher diffraction intensity than the control sample in the XRD patterns, which reveals that the perovskite film with improved crystallinity was obtained after introducing the PEACl or HACl 2 into the precursor solutions.
To gain more details on the crystallinity of the perovskites affected by the ammonium chloride agents, the grazing-incidence wide-angle x-ray scattering (GIWAXS) analysis was further conducted.Figure 1C shows the one-dimensional (1D) GIWAXS profiles along the q z direction integrated from the 2D scattering patterns that are presented in Figure 1D-F, corresponding to the control, PEACl-and HACl 2 -based perovskite films, respectively.It can be seen that these 2D scattering patterns present diffraction rings at q = 10.3 nm −1 , which originates from the (110) plane of the CsPbI 3 perovskites.Meanwhile, these 2D scattering patterns are intensive around the in-plane (q xy ) or out-of-plane (q z ) directions, suggesting the formation of perovskites with a perpendicular and parallel-oriented structure.However, after incorporating ammonium chlorides into the perovskites, especially in HACl 2 -based perovskite film, the diffraction intensity of the (110) plane was obviously increased, which results from improved crystallization and orientation of the perovskites, in line with the XRD results (Figure 1B).With these ammonium chloride agents, highly oriented crystal grains with vertical growth of perovskites could be well prepared on the substrates.][53] The improved crystallization and orientation of perovskites may also affect the film morphology, and the pinholes in the perovskite films generally provide pathways for contact between the electron transport layer and hole transport layer, leading to current leakage.Therefore, scanning electron microscope (SEM) images of these perovskite films were measured.These perovskite films possess close-packed grains with pinhole-free morphology (Figure 1G-I), which is favorable for minimizing current leakage.Charge carrier recombination at the grain boundaries of the perovskites generally results in serious energy losses of PSCs, and the perovskite with a smaller grain size generally leads to more grain boundaries. 54,55eanwhile, the defects at the grain boundaries could also facilitate ion migration, which is disadvantageous for the photovoltaic performance of PSCs.After incorporating the ammonium chlorides into the perovskites, the perovskite films with a larger grain size were prepared.The average grain sizes of the perovskite films were increased from approximately 400 nm for the control perovskite film to approximately 550 and 600 nm for PEACl-and HACl 2based perovskite films, respectively.For the HACl 2 -based perovskite films, the influence of the agent concentration on the film morphology was also studied, which exhibits an optimal concentration of 0.2 mg mL −1 (Figure S2).Besides, the control perovskite film presents stacked grains with plentiful grain boundaries and disarranged orientation, while PEACl-and HACl 2 -based perovskite films show more orderly crystal orientation with less grain boundaries accompanied by similar film thicknesses of approximately 400 nm (Figure 1J-L).A highly vertical orientation of grains was obtained for the HACl 2 -based perovskite film.The SEM results are in agreement with the above XRD and GIWAXS results, indicating that improved crystallinity and ordered orientation are realized for the perovskite films with the ammonium chloride agents.Therefore, the ammonium chlorides could not only benefit the perovskite crystallization but also lower the grain boundaries.
The evolution of perovskite crystallization could be derived from the regulation of crystallization kinetics, which might be caused by Cl − ions.Thus, to rule out whether the Cl − ions lead to the improved crystal quality of the perovskites, another ammonium chloride, ethylenediammonium dihydrochloride (EDACl 2 ), was used as the additive to modulate the crystallization of perovskites.The XRD pattern and SEM image of the EDACl 2 -based perovskite film are displayed in Figure S3, which suggests that the crystallinity of EDACl 2 -based perovskite film did not enhance, exhibiting similar results in comparison with the control film.Therefore, according to these important results, it can be deduced here that ammonium cations play a crucial role in regulating the crystallization kinetics of perovskites.

Crystallization dynamics of perovskites
To fundamentally understand the effect of ammonium chlorides on subsequently the crystallization process of CsPbI 3 perovskites, in situ light absorption spectra were recorded during the annealing process of CsPbI 3 perovskite films.As depicted in Figure 2A-C, as the annealing time increases, the absorption at the wavelength of longer than 500 nm is gradually enhanced, which could be derived from the phase transition of the CsPbI 3 perovskite from the intermediate DMAPbI 3 and Cs 4 PbI 6 phase and enhanced crystallinity.To quantify the difference among these films, the absorption intensity at the wavelength of 700 nm as a function of annealing time was further extracted to account for the crystallization dynamics, which is plotted in Figure 2D.It can be found that the crystallization of perovskites could be divided into two stages: nucleation (Stage I: 0-4 min) and growth (Stage II: 5-9 min) processes, and subsequently the crystallization process was completed cross over the film thickness.The control film presents a slow nucleation process with a corresponding rate (n N ) of 0.00073, and a relatively rapid growth process with a corresponding rate (n G ) of 0.099 was observed, which may account for the smaller grain size of the control film.The PEACl-and HACl 2 -based film demonstrates a faster nucleation process with corresponding rates (n N ) of 0.0074 and 0.0203, respectively.Furthermore, the growth rates of PEACl-(n G = 0.081) and HACl 2 -based (n G = 0.076) perovskite films were decreased compared with that of the control film, suggesting that ammonium agents in the perovskite precursor delay the crystal growth process.Overall, the ammonium chlorides modulated CsPbI 3 crystallization realizing accelerated nucleation and retarded crystal growth, finally leading to high-quality polycrystalline films being obtained.
The high-quality perovskites generally result in high optoelectric properties, and thus the optoelectric properties of CsPbI 3 perovskite films without and with ammonium chlorides as crystallization agents were further characterized.Due to the enhanced crystallinity, the HACl 2based perovskite film shows the strongest absorption in the visible wavelength region (Figure S4).The Urbach energy (E U ) that is generally applied to describe the energetic disorder of perovskite films was also investigated.As displayed in Figure S5, the HACl 2 -based perovskite film represented the lowest value of E U compared with the control and PEACl-based perovskite films, suggesting that the good crystallized film with a decreased density of sub-bandgap states was obtained, which is in line with structure and morphology characterizations.
The steady-state photoluminescence (PL) spectra of the perovskite films coated on the quartz glass are exhibited in Figure 2E.The characteristic emission peak at 733 nm agrees well with the light absorption results.The HACl 2 -based perovskite film presents a superlative PL intensity, suggesting that the nonradiative recombination was significantly suppressed.Meanwhile, the perovskite films coated on the fluorine-doped tin oxide (FTO)/TiO 2 substrates were also investigated to analyze the charge transport at the perovskite/TiO 2 interface.Figure S6 shows that the PL intensity of the HACl 2 -based perovskite film is quenched in comparison with that of the control film, suggesting more charge carriers could be transported from the HACl 2 -based perovskite film to the TiO 2 layer.The improved charge transport at the HACl 2 -based perovskite/TiO 2 interface may be caused by the effectively passivated buried interface of the perovskite film, which will be thoroughly discussed in the next section.
The time-resolved PL decay measurement can be applied to reveal the dynamic of photogenerated charge carriers in the perovskite films.Thus, the transient PL spectra of the perovskite films without and with the ammonium chlorides as crystallization agents were measured.Figure 2F exhibits the time-resolved photoluminescence (TRPL) spectra of these perovskite films coated on the quartz substrates, and the details of the fitting parameters are listed in Table S1.The fitted results disclose that the charge carrier lifetime in the HACl 2 -based perovskite film is the longest (τ ave = 7.28 ns) compared with the control (τ ave = 5.99 ns) and PEACl-based (τ ave = 7.01 ns) perovskite films.The increase of charge carrier lifetime in the HACl 2 -based perovskite film implies reduced defects being realized, which can be derived from ameliorated morphology and high crystallinity of the HACl 2 -based perovskite film.

Mechanism of modulating crystallization
To deeply understand the crystallization kinetics of the perovskite induced by the ammonium chlorides, firstprinciples density functional theory (DFT) calculations were performed to theoretically study the intermolecular interaction.Theoretical models of PbI 2 coupling with CsI, PEACl, and HACl 2 were constructed and the interaction energies (E inter ) between them were calculated.The geometrical structure with charge distribution and the energetics of these complexes are displayed in Figure 3A.The E inter of CsI-PbI 2 , PEACl-PbI 2 , and HACl 2 -PbI 2 are −1.51,−1.59, and −1.61 eV, respectively, showing relatively lower values of E inter in PEACl-PbI 2 or HACl 2 -PbI 2 compared to CsI-PbI 2 .Considering the more energetically favorable configuration for the complexation of PbI 2 with HACl 2 , a stronger interaction may be conducted between PbI 2 and HACl 2 .
To elucidate the above calculation and link the theoretical results with the crystallization of perovskites, x-ray photoelectron spectroscopy (XPS) was measured to gain more information regarding the chemical coupling between the ammonium chlorides and perovskites.The full XPS spectra of the perovskites are displayed in Figure S7.As depicted in Figure 3B, the N 1s peak at the binding energy of 402.16 eV is observed in both PEACl-and HACl 2 -based perovskite films, while there is no such peak in the control sample, which suggests that the ammonium (C-NH 3 + ) was anchored on the perovskite surface.As the ammonium cations did not integrate into the perovskite lattice according to XRD results, the ammonium cations are supposed to exist at the grain boundaries and on the surface of the perovskites.Figure 3C displays the Pb 4f peaks of the perovskite films, in which two peaks at 137.94 and 142.81 eV can be assigned to the signals of Pb 4f 5/2 and Pb 4f 7/2 , respectively.After incorporating the ammonium chlorides into the perovskites, the Pb 4f peaks move to 138.07 and 142.93 eV for the PEACl-based perovskite film and 138.34 and 143.19 eV for the HACl 2 -based perovskite film, indicating chemical interaction between the ammonium chlorides and [PbI 6 ] 4− octahedra frameworks of the perovskites.The I 3d peaks also show a similar trend to the Pb 4f peaks (Figure 3D).The larger movement of the Pb 4f and I 3d peaks in HACl 2 -based perovskite film may suggest a stronger interaction between the HACl 2 and perovskites, probably originating from the dual functionality of primary ammonium and imidazolium groups, which is in agreement with the results of the DFT calculation.Notably, no characteristic Cl 2p signals were detected in these samples, which may be attributed to anion exchange between HACl 2 and DMAI during the annealing process in consideration of the more volatile property of DMACl than DMAI (Figure S8). 56ccording to the above results, we can infer that by incorporating HACl 2 into the perovskite precursor, the HACl 2 -perovskite adducts could be formed due to the robust interaction between the HACl 2 and CsPbI 3 perovskite, which facilitates crystal nucleation in the subsequent crystallization process, as schematically described in Figure 3E.Meanwhile, the spacer diammonium HA 2+ located between nucleation sites could slow the crystal growth of the perovskites, eventually leading to improved crystallinity being realized in HACl 2 -based perovskite films.The nucleation and crystal growth processes of the perovskites can be illustrated with the LaMer graph (Figure S9), which displays a faster nucleation rate due to the preformation of colloidal complexes and a retarded crystal growth owing to the strong interaction between PbI 2 and HACl 2 for HACl 2 -based perovskite film.Besides, the residual diammonium HA 2+ cations distributed at the grain boundaries and on the surface of perovskites may be favorable to defect passivation.
To obtain more insight into the coupling between these ammonium chlorides and perovskites, DFT simulations were further performed.The CsI-terminated perovskite with a Cs + vacancy defect (V CS ) was constructed as a reference, and PEA + and HA 2+ were adopted to passivate the V CS .Figure 4A depicts the structural relaxation models of the CsPbI 3 perovskite slab containing a V CS without and with agent coupling.It can be found that the ammonium group of PEA + cation could occupy the V CS , leaving the benzene ring of PEA + points outside of perovskite.As divalent HA 2+ cation contains a primary ammonium group and an imidazolium group, both sides of HA 2+ could occupy the V CS , realizing two surface configurations (HA-I: couping with primary ammonium; HA-II: couping with imidazolium).The binding energies of PEA, HA-I, and HA-II configurations were calculated to be −4.66,−5.16, and −5.04 eV, respectively.The binding energies of HA-I and HA-II configurations were lower than that of the PEA + coupled configuration, suggesting that the HA 2+ may prefer to occupy V CS and shows more effective defect passivation.Meanwhile, as displayed in Figure S10, compared with the perovskite coupled with the PEA + , the HA 2+ -based systems show a less disordered lattice structure, which indicates that a superior defect passivation of HA 2+ was realized compared with the PEA + -based system.
The differential charge density distributions of the perovskite slab coupled with PEA + or HA 2+ were also plotted to further understand the electrostatic interactions and charge transfer on the perovskite surface.As presented in Figure 4b-d, charge transfer was hindered by the ammonium group in PEA + -based system, while excellent interfacial electron transfer was achieved in both HA-I and HA-II configurations, which is conducive for charge carrier transport between the adjoining perovskite grains.Moreover, the perovskite coupled with the ammonium chlorides could also affect the formation of defects, and thus the formation energies of iodine vacancy defect   S2).As expected, after being coupled with PEA + or HA 2+ , the formation energies of V I and Pb I were largely increased.The positive formation energies of V I and Pb I indicate that the defects are not easily made at the surface of the perovskites; in other words, the formation of surface defects of the perovskites coupled with PEA + or HA 2+ was substantially restrained.Therefore, theoretical results demonstrate that the diammonium HA 2+ cation could efficiently passivate V CS with both the ammonium and imidazolium groups and restrain the formation of surface defects without affecting the charge transfer in the perovskites, resulting in improved optoelectronic properties.

Hot carrier dynamics
The ammonium chlorides can also influence the charge carrier dynamics within the perovskite films, especially the hot carriers.To examine such a hypothesis, transient absorption (TA) measurement was further performed to investigate the charge carrier dynamics within the perovskite films induced by these ammonium chlorides.information on the influence of ammonium chlorides on charge carrier recombination, the time evolution derived from the pseudo-color TA spectra is depicted in Figure S11.These samples exhibit similar features, in which the negative ΔA valley located at approximately 705 nm can be detected.
The PB signal at the hot carrier cooling stage could offer more details of the carrier generation and recombination within the perovskites.Owing to the low exciton binding energy of the perovskite, the photoinduced carriers would generate promptly within 1 ps forming hot carriers, and then lose their excess energy until reaching thermal equilibrium through optical phonon emission.Figure 5D-F shows the normalized TA spectra at the first 10 ps of control, PEACl-and HACl 2 -based perovskite films, respectively, which exhibits the evolution of hot carriers cooling.
According to the cooling curves, the energy loss rate per carrier (J) could be gained by applying the following equation,  = −1.5B  C

𝑑𝑡
, where k B is the Boltzmann constant and T C is the carrier temperature. 57As the high-energy bleach tail could be described by the Maxwell-Boltzmann distribution function, ( f − ∕ B  C ), the value of T C could be derived from the fitting function.The fitted results are presented in Figure 5G.As depicted in Figure 5H, the energy loss rate of hot carriers is decreased after modulation of the perovskites with ammonium chlorides, and the HACl 2 -based perovskite possesses the slowest energy loss rate.Such a tardy energy loss rate has a high potential to take advantage of the excess hot carrier energy for further improving the photovoltaic performance of PSCs, realizing enhanced V oc of PSCs. 58Bimolecular recombination and trap-assisted monomolecular recombination would subsequently happen after hot carrier cooling.Figure 5I shows the TA kinetics of control, PEACl-and HACl 2 -based perovskite films at 705 nm, and the details of the fitting parameters are listed in Table S3.In comparison with the control perovskite film with a charge recombination lifetime of 107.85 ps, the HACl 2 -based perovskite film presents a longer recombination lifetime of 195.21 ps, demonstrating suppressive nonradiative recombination.Therefore, we could conclude here that the charge carrier cooling and recombination dynamics could be substantially ameliorated by modulating the perovskites with ammonium chlorides.With HACl 2 as crystallization agents in the perovskites, the charge trapping within the perovskite film is significantly reduced and thus the adverse effect of bulk and surface defects on hot carrier cooling is mitigated, which is conducive to increasing the charge carrier extraction of PSCs.

Photovoltaic performance
The perovskite film with improved crystallization and optoelectronic properties is favored for the fabrication of efficient PSCs.Therefore, to investigate the inorganic perovskites with ammonium chlorides as crystallization agents, PSC with a device architecture of glass/FTO/TiO 2 /perovskite/2,2',7,7'-tetrakis [N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD)/Ag was constructed, as depicted in Figure 6A and Figure S12.TiO 2 and Spiro-OMeTAD were adopted as electron and hole-selective layers, respectively.Considering the ammonium chlorides might influence the energy levels of the perovskites, ultraviolet photoelectron spectrum (UPS) was conducted to figure out the energy levels of the perovskites, which is displayed in Figure S13.The energy difference between the Fermi level (E F ) and the valence band maximum (VBM) decreases from 1.52 eV for the control perovskite film to 1.40 eV for the PEAClbased perovskite film and to 1.23 eV for the HACl 2 -based perovskite film, indicating a less n-type perovskite surface after the modification using ammonium chlorides.The perovskite surface energetics changes could be derived from reduced n-type point defects, such as V I and Pb I , due to increased defect formation energy verified by the above DFT results.Such changes in perovskite surface energetics are conducive to promoting hole extraction and restraining the nonradiative recombination at the perovskite/hole selective layer contact for PSCs. Figure 6B displays the energy levels of each functional layer within the PSCs.It can be seen that the PEACl-and HACl 2 -based perovskite films could form better energy level alignment with the hole selective layers within the PSCs, which is favorable for device performance.Figure 6C depicts the photocurrent density-voltage (J-V) curves of the PSCs, with and without ammonium chlorides as crystallization agents, and the performance parameters are summarized in Table  short-circuit current density (J sc ) of 20.55 mA cm −2 , and a fill factor (FF) of 0.781.After incorporating PEACl into the perovskites, the efficiency of PEACl-based PSC was increased to 18.48%.Remarkably, a high PCE of 19.67% was realized in HACl 2 -based PSCs with V oc of 1.14 V, J sc of 20.86 mA cm −2 , and FF of 0.827.The influence of the agent concentration on the photovoltaic performance of HACl 2based PSCs was also studied, which exhibits an optimal concentration of 0.2 mg mL −1 (Figure S14 and Table S4).
The inorganic PSCs generally show hysteresis when recording the J-V curves under different scanning modes (from V oc to J sc or from J sc to V oc ).Therefore, the J-V curves were simultaneously recorded under different scanning modes (Figure S15 and Table S5), which shows that the hysteresis indexes (HIs) of 15.3%, 6.5%, and 5.2% were obtained for control, PEACl-, and HACl 2 -based PSCs, respectively.The lowered HI in HACl 2 -based PSCs may be primarily derived from the reduced defect density, which is attributed to the improved crystallinity of the perovskites.
The steady-state power output of PSCs measured at the maximum power point is presented in Figure 6D, which shows that the PSCs exhibit stable efficiencies under continuous illumination.Figure 6E displays the external quantum efficiency (EQE) spectra and integrated photocurrent density of the PSCs.The integrated J sc calculated from the EQE spectra are 19.67,20.35, and 20.48 mA cm −2 for control, PEACl-, and HACl 2 -based PSCs, respectively, which is in good agreement with the J sc obtained from the J-V measurements.The statistical parameters from 20 devices are depicted in Figure 6F and Figure S16, which shows that these PSCs exhibit good reproducibility, and HACl 2 -based PSCs possess relatively higher average PCEs.The improved photovoltaic performance in HACl 2based PSCs primarily resulted from enhanced FF and V oc , which is presumably because of reduced charge carrier recombination induced by improved crystallization and optoelectronic properties of HACl 2 -based perovskites.

Charge carrier extraction and recombination
For a deeper understanding of the causes inducing an enhanced photovoltaic performance in ammonium-based PSCs, the charge carrier collection and recombination within the PSCs were comprehensively studied.The high built-in electric field within PSCs is favorable to widening the depletion region and driving the extraction of photoinduced charge carriers for suppressing charge carrier recombination.To verify such speculation, the built-in potential (V bi ) of PSCs was extracted from the Mott-Schottky curves using the following equation 59 : where C, A, e, ε, ε 0 , and N A are the measured capacitance, device area, elementary charge, relative dielectric constant, vacuum permittivity, and carrier density, respectively.As shown in Figure 7A, the V bi of the HACl 2 -based device is 0.99 V, which is greatly enhanced in comparison with the control PSC (V bi = 0.83 V) and PEACl-based PSC (V bi = 0.93 V), which to some extent leads to enhanced V oc of HACl 2 -based PSCs.
Transient photovoltage (TPV) characterization offers a practical method to study the charge carrier recombination within the PSCs. Figure 7B shows the TPV curves of control, PEACl-and HACl 2 -based PSCs, which were fitted using a bi-exponential function to extract the charge lifetime (Table S6).The HACl 2 -based PSC shows the longest charge lifetime (358.44 μs) in comparison with that of the control PSC (246.87 μs) and PEACl-based PSC (106.89μs).The long charge lifetime within the HACl 2 -based PSC demonstrates the restrained recombination of photogenerated charge carriers, which suggests that the trap-assisted recombination is substantially confined that could explain the enhanced V oc and FF in HACl 2 -based PSCs.
Electrochemical impedance spectroscopy (EIS) measurement was further conducted to investigate the charge transport within the PSCs. Figure 7C depicts the Nyquist plots of the EIS results, and the fitting results are listed in Table S7.The HACl 2 -based PSC exhibits the lowest series resistance (R s ) of 13.86 Ω and transfer resistance (R tr ) of 145.7 Ω, and the largest recombination resistance (R rec ) of 6114 Ω among these PSCs.It is speculated that the efficient charge transport could be derived from the diminished defect density.For quantitatively evaluating trap state density (N t ) of perovskite films, dark J-V curves were conducted for the electron-only device with a structure FTO/TiO 2 /perovskite/6,6-phenyl-C61butyric acid methyl ester (PC 61 BM)/Ag.The N t can be gained by applying the following equation 60 : where e, ε 0 , ɛ, V TFL , and L are the elementary charge, vacuum permittivity, the relative dielectric constant, trapfilled limit voltage, and the thickness of the perovskite film, respectively.As depicted in Figure 7D, the V TFL of the control device is 0.35 V, corresponding to the N t of 2.65 × 10 15 cm −3 .Obviously, the V TFL and N t were simultaneously decreased after incorporating the PEACl or HACl 2 into the perovskites.The N t of PEACl-and HACl 2 -based devices are 2.04 × 10 15 and 1.81 × 10 15 cm −3 with the V TFL of 0.27 and 0.24 V, respectively.The lowest N t of the HACl 2 -based device can be ascribed to the robust interaction between HACl 2 and perovskites, which is conducive to regulating the crystallization of the perovskite films.
Electroluminescent quantum efficiency (EQE EL ) of a solar cell in the dark condition operated as a light-emitting diode (LED) under bias voltage can be adopted to quantitatively evaluate the nonradiative recombination losses in the solar cell.In principle, the V oc loss resulting from the nonradiative recombination is inversely proportional to the EQE EL of the PSCs according to the reciprocal principle. 61,62Thus, the electroluminescence (EL) spectroscopy and the EQE EL of the PSCs were measured by operating the PSCs as LEDs under forward biases.As shown in Figure 7E, EL spectra at 735 nm are observed and the EL intensity of the HACl 2 -PSC is the strongest among these devices, indicating that nonradiative recombination in the HACl 2 -based PSC is dramatically restrained.As presented in Figure 7F, the HACl 2 -based PSC yields an EQE EL of 0.0363% under an injection current density of approximately 20.8 mA cm −2 , which is much higher in comparison with that of the control PSC (EQE EL = 0.0018%) and the PEACl-based PSC (EQE EL = 0.0042%).According to these important results, we determined the reduced nonradiative recombination loss (ΔV OC,nonrad ) contributing to the V OC increase (ΔV OC ) based on the following formula 63 : where k is the Boltzmann constant, T is the Kelvin temperature, q is the elemental charge, EQE Target and EQE Control are the EQE EL value of the target and control PSCs as the injection current density under the dark condition is equal to the photocurrent density under the illumination, respectively.The ΔV OC,nonrad values were determined to be 22 and 75 mV for PEACl-and HACl 2 -based PSCs, respectively.The reduced ΔV OC,nonrad of PEACl-and HACl 2 -based PSCs are in good agreement with the increased V OC , which indicates that the enhancement of V OC is primarily afforded by the mitigation of nonradiative recombination losses.Overall, the diminished defect density coupled with restrained charge recombination resulting from the improved crystallization of perovskites is responsible for the high photovoltaic performance of the HACl 2 -based PSCs.Moreover, the device stability is vital to understanding their overall performance, and thus the device stability was further considered.Because the inorganic CsPbI 3 perovskite film is hypersensitive to humidity, the stability of the perovskite films was studied when the control, PEACland HACl 2 -based perovskite films were aged in the air with relative humidity (RH) of 55% ± 5%.The light absorption spectra of the perovskite films were recorded along with exposure time.As depicted in Figure S17, the intensity of absorption at 700 nm for the control film decreases to approximately 68% of its original intensity after exposure for 12 h, while the PEACl-and HACl 2 -based films show no significant difference, indicating that PEACl or HACl 2 could improve the phase stability of perovskites because of strong coupling of PEACl and HACl 2 with the perovskites with reduced defects.The storage lifetime of unencapsulated devices was also recorded in the air condition (25% ± 5% RH).As depicted in Figure S18, the HACl 2based PSCs maintain approximately 83% of their original efficiencies with a 500 h storage, while the control PSCs only keep approximately 56% of their initial efficiencies.Therefore, by modulating perovskite films with ammonium chlorides, the phase stability of perovskites and the device stability of PSCs were enhanced entirely.

CONCLUSION
In conclusion, the crystallization of inorganic CsPbI 3 perovskites was modulated using ammonium chlorides as additives for efficient solar cells.Comprehensive experimental studies and detailed theoretical calculations disclose that the HACl 2 could strongly couple with PbI 2 in the precursor solution and subsequently accelerate the nucleation and slower crystal growth of the perovskites, finally resulting in ameliorated crystallization and optoelectronic properties of perovskite films being realized.Additionally, residual diammonium HA 2+ cations distributed at the grain boundary and on the surface of perovskites could effectively passivate defects with its dual functionality of primary ammonium and imidazolium group through strong electrostatic interactions, suppressing trap-assisted nonradiative recombination and producing more matched perovskite surface energetics.Consequently, the photovoltaic performance of PSCs was greatly enhanced, and a PCE of up to 19.67% was obtained in HACl 2 -based PSCs compared with that of 17.02% for control PSCs.The improved efficiency of HACl 2 -based PSCs was derived from substantially attenuated charge carrier recombination induced by ameliorated crystallization and optoelectronic properties of perovskite films.This work paves an advanced way for regulating the crystallization of inorganic perovskites or the perovskite with other compositions for efficient PSCs as well as other optoelectronic devices.

A C K N O W L E D G M E N T S
This work was supported by the National Key Research and Development Program of China (grant number 2022YFB3807200), the National Natural Science Foundation of China (grant numbers 52372169 and 51872014), the Recruitment Program of Global Experts, and the "111" project (B17002).This work was also supported by the HPC of Beihang University.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare they have no conflict of interest.

F
I G U R E 1 (A) Molecular structure and electrostatic potential maps of the ammonium cations.(B) XRD patterns and (C) 1D GIWAXS intensity profiles along the q z axis of perovskite films.2D GIWAXS patterns of (D) control, (E) PEACl-and (F) HACl 2 -based perovskite films.Top-view SEM images of (G) control (H) PEACl-and (I) HACl 2 -based perovskite films.Cross-section SEM images of (J) control, (K) PEACland (L) HACl 2 -based perovskite films.

F I G U R E 2
In situ light absorption spectra of (A) control, (B) PEACl-and (C) HACl 2 -based perovskite films as a function of annealing time at 180 • C. (D) The evolution of absorption intensity of control, PEACl-and HACl 2 -based perovskite films at 700 nm.(E) Steady-state PL spectra and (F) TRPL spectra of control, PEACl-and HACl 2 -based perovskite films.

F
I G U R E 3 (A) The geometrical structure and energetics models of PbI 2 coupling with CsI, PEACl, and HACl 2 .Core-level XPS spectra of (B) N 1s, (C) Pb 4f and (D) I 3d of control, PEACl-and HACl 2 -based perovskite films.(E) Schematic illustration of the interaction between perovskite and HACl 2 .

F
I G U R E 4 (A) Theoretical models of the perovskite slab containing a Cs + vacancy without and with agent passivation.Differential charge density distribution of the perovskite slab coupled with (B) PEA + , (C) primary ammonium of HA 2+ , and (D) imidazolium HA 2+ .The upper and lower images depict the side view and top view of the theoretical model, respectively.(E) The defect formation energies in the perovskite slab containing a Cs + vacancy without and with the agent coupling.

(
V I ) and Pb-I antisite defect (Pb I ) were further calculated (Figure4E and Table Figure 5A-C presents pseudo-color TA spectra for the control, PEACl-and HACl 2 -based perovskite films.A sharp photo-bleaching signal (PB, ΔA < 0) with carrier decay could be clearly observed at approximately 705 nm, resulting from the prevailing band-filling effect and subsequent carrier recombination.To obtain more detailed F I G U R E 5 Typical pseudo-color TA spectra of (A) control, (B) PEACl-and (C) HACl 2 -based perovskite films.Normalized TA spectra at the first 10 ps of (D) control, (E) PEACl-and (F) HACl 2 -based perovskite films.(G) Hot carrier temperature (T C ) at the first 10 ps for control, PEACl-and HACl 2 -based perovskite films.(H) Energy loss rate as a function of T C of control, PEACl-and HACl 2 -based perovskite films.(I) TA kinetics of control, PEACl-and HACl 2 -based perovskite films.

F
I G U R E 6 (A) Device structure of the PSC.(B) The energy-level diagram of each functional layer within the PSCs.(C) The J-V curves, (D) stabilized output curves, (E) EQE spectra, and (F) statistical PCE of control, PEACl-and HACl 2 -based PSCs.

F
I G U R E 7 (A) Mott-Schottky curves, (B) TPV curves and (C) the EIS Nyquist plots of control, PEACl-and HACl 2 -based PSCs.(D) SCLC curves of the electron-only device prepared with the control, PEACl-and HACl 2 -based perovskite films.(E) EL spectra of control, PEACl-and HACl 2 -based PSCs under 2 V bias voltage.Inset shows a photograph of the HACl 2 -based PSC working as a LED under 2 V bias voltage.(F) EQE of EL of control, PEACl-and HACl 2 -based PSCs operated as LEDs.