Engineering inorganic lead halide perovskite deposition toward solar cells with efficiency approaching 20%

Inorganic perovskite materials have gained tremendous attention due to their superior chemical and thermal stability than organic‐inorganic hybrid perovskites. In the past years, substantial research efforts have been devoted to developing uniform and pin‐hole free inorganic perovskite films with high electronic quality. As a result, power conversion efficiency of inorganic perovskite solar cells (PSCs) has boosted to over 19%, which presents a promising potential for technology commercialization. Herein, we give a comprehensive review of the recent progress on state‐of‐the‐art inorganic cesium lead halide based PSCs, particularly on the perovskite deposition approaches. We show a clear roadmap to fabricate high electronic quality inorganic perovskite films by tuning the precursor crystallization kinetics, performing post‐deposition treatments, and passivating surface/interface defects. Inorganic perovskite films prepared by these approaches exhibit not only high crystallinity, favorable morphology, and low trap densities but also improved phase stability. The advanced deposition approaches lay the foundation for further improving the performance and long‐term operational stability of inorganic perovskite photovoltaics.


INTRODUCTION
In recent years, perovskite solar cells (PSCs) have obtained great progress. The certified power conversion efficiency (PCE) has reached 25.5%, which is comparable to commercial photovoltaic technologies such as crystalline silicon, cadmium telluride (CdTe), and copper indium gallium selenide (CIGS) solar cells. [1][2][3][4][5][6][7] Perovskite refers to a class of materials with the formula of ABX 3 , in which A is monovalent organic cations such as methylammonium (CH 3 NH 3 + , MA + ), formamidinium (NH 2 = CHNH 2 + , FA + ) and inorganic cations such as Cs + , B is divalent metal cations such as Pb 2+ and Sn 2+ , and X is halide anion such as I -, Brand Cl -. [8] The most-investigated perovskite materials remain a major limitation to survival under moisture, thermal, and UV irradiation stress due to the intrinsic volatility of organic cations. [9][10][11][12] To mitigate the moisture-induced degradation pathway, many strategies have been reported, e.g., optimization of crystallization process and film morphology, substitution of the small organic cation with bulky or long-chain organic cations at A site, deposition of a hydrophobic charge transfer layer or a surface modification layer, and employment of encapsulation, etc. [13][14][15][16][17][18][19] Significant stability enhancement has been realized in different research labs. On the other hand, thermal instability is still a huge challenge for organic perovskites due to the high volatility nature of organic species at elevated temperature (e.g., 85 • C for MAPbI 3 ). [20][21][22] Complete substitution of the organic cations with inorganic cation, i.e., cesium (Cs + ) to form CsPbI 3 prevents the degradation pathways due to the much stronger Cs-I ionic bond in inorganic perovskite than the H-I bond in organic perovskite [23,24]. As a result, the inorganic perovskites show substantially enhanced stability against heat and illumination. In addition, partial substitution of Iwith Brto form the halide alloy allows stabilization of the inorganic perovskite in the photoactive phase at the device operating temperature and broadening its optical bandgap. [25] This opens up the application of inorganic perovskite as absorbers of the top cells in tandem and triple-junction solar cells. With the great research F I G U R E 1 PCE evolution of inorganic PSCs. Note: dimethyl sulfoxide (DMSO), solvent-controlled growth (SSG), phenyltrimethylammonium bromide (PTABr), choline iodine (CHI), phenyltrimethylammonium chloride (PTACl) efforts made in the community, inorganic PSC has reached over 19% efficiency and enhanced stability in a very short period ( Figure 1). [26][27][28][29][30][31][32][33][34] Quality of inorganic perovskite films, such as the crystallinity, surface roughness, film thickness, defect level and density, etc., determine the carrier transport and recombination kinetics and therefore the device performance. [35][36][37][38] Besides, phase stability of inorganic perovskite films determines the device long-term operational stability. [8] These properties can be adjusted by the development of appropriate perovskite film deposition approaches. In this review, we aim to give a comprehensive overview of the recent advancement of deposition methods for inorganic perovskite films with favorable optoelectronic properties. First, we briefly introduce the fundaments of CsPbX 3 . Then, we systematically review the representative deposition methods for four inorganic perovskite systems, i.e., CsPbI 3 , CsPbI 2 Br, CsPbI 2 Br 2 and CsPbBr 3 with the following two focuses: i) preparation of thick inorganic perovskite films with low defect density and ii) stabilization of the inorganic perovskite in the photoactive phase. Specifically, the effect of tuning the film crystallization and growth kinetics, defect passivation, interface optimization, post-deposition treatment, and two-dimensional passivation are discussed in detail. Finally, we conclude the review with perspectives for the future development of inorganic PSCs toward an efficiency beyond 20%.

FUNDAMENTS OF INORGANIC PEROVSKITES
For the inorganic perovskite of CsPbX 3 (X refers to iodine or bromine), the three-dimensional network of PbX 6 octahedra is formed by Pb and X. Cs ions are occupied at the octahedral voids between halogens. [39] The Goldschmidt tolerance factor is used to estimate the phase stability according to Equation (1) where r A , r B and r X are the radii of A, B, and X ions, respectively. [40,41] Empirically, the stable cubic phase could be obtained when the t is between 0.9 and 1. If the t value is smaller than 0.9, the PbX 6 titling will induce perovskite structure distortion. Meanwhile, there will be non-perovskite phases when the t value is smaller than 0.8 or larger than 1. Take CsPbI 3 as an example, four complex phases exist, which are cubic (α-, Pm3m), tetragonal (β-, P4/mbm), orthorhombic (γ-, Pbnm) and the non-perovskite (δ-, Pnma) phases. [27,33,[42][43][44][45][46] Due to the small Cs ionic radii, CsPbI 3 exhibits a small Goldschmidt tolerance of 0.81. As a result, CsPbI 3 is thermodynamically stable in the yellow δ-phase at room temperature. [47,48] The δ-CsPbI 3 has a wide bandgap of 2.82 eV and low carrier mobility, which are unsuitable for photovoltaic application. Upon high-temperature annealing at 350 • C δ-CsPbI 3 transfers to cubic α-CsPbI 3 with a favorable bandgap of 1.73 eV. The α-CsPbI 3 converts to the tetragonal β phase at 260 • C, and subsequently the orthorhombic γ phase at 175 • C during the cooling process. [49] The β-CsPbI 3 and γ-CsPbI 3 are both black phases but tend to convert to the δ phase when exposed to humidity at room temperature. [31,50,51] Due to the strong relationship between perovskite crystal structure and device performance, stabilization of CsPbI 3 in the black α phase is essential to obtain high efficiency and stable inorganic PSCs. By partially replacing Iwith Br -, the Goldschmidt tolerance factor of CsPbX 3 slightly increases, indicating improved phase stability. [52] The representative inorganic perovskite, i.e., CsPbI 2 Br (bandgap 1.9 eV), indeed exhibits improved phase stability at room temperature, but still suffers from phase transition under high humidity exposure. [53][54][55][56][57] Theoretical calculations demonstrated that CsPbI 2 Br exhibits a stable alloying phase due to strong coulomb interactions, which is beneficial for suppressing phase segregation. [58,59] Further increase Brconcentration in the X cite results in CsPbIBr 2 and CsPbBr 3 with enlarged bandgaps of ∼2.05 and ∼2.3 eV, respectively. The wide bandgaps dramatically constrain the absorption edges of absorbers, leading to a low photovoltaic performance for single-junction solar cells. [60][61][62][63][64][65] On the other hand, these perovskites are promising top cell candidates for tandem and triple-junction perovskite-based solar cells.
To address these challenges of inorganic perovskite, it is urgent to develop new perovskite deposition approaches to simultaneously increase solar cell efficiency and long-term operational stability. In the following section, we have summarized the deposition methods for a series of perovskite systems, i.e., CsPbI 3 (Section 3.2), CsPbI 2 Br (Section 3.3), CsPbIBr 2 (Section 3.4) and CsPbBr 3 (3.5), respectively.

Methodologies for inorganic perovskite deposition
The preparation of high electronic quality perovskite thin films lays the foundation for the realization of highperformance PSCs. Similar to the hybrid organic-inorganic perovskite materials, all-inorganic perovskite films can be prepared by the solution processing techniques, including one-step and two-step spin-coating and blade coating methods ( Figure 2). The solution processing technique is a relatively facile approach to fabricate perovskite films with a low manufacturing cost. In general, the precursors are dissolved in a proper solvent to form a solution, which is spin-or blade-coated on a substrate and followed by a thermal annealing process. Using the perovskite quantum dots (QDs) precursor is an alternative method. The dispersions of CsPbX 3 QDs are applied as "inks" to fabricate perovskite films by spin-casting at room temperature. [66] Compared with the solution processing method, QDs deposition is almost unaffected by external conditions, and allows multiple deposition steps to adjust film thickness. In addition, vapor deposition techniques including co-evaporation and sequential deposition, are effective strategies to prepare high-quality all-inorganic perovskite films. The vapor deposition techniques have demonstrated special advantages in obtaining uniform and smooth perovskite films over large area with great repeatability, and are especially suitable for the perovskite materials that show low solubility.

CsPbI 3
CsPbI 3 is thermodynamically stable in a non-perovskite (δ) phase at room temperature and the black cubic (α) phase at a temperature over 350 • C. [42] In 2014, Choi et al. reported the first CsPbI 3 PSCs by substitution of MA + from Cs x MA 1−x PbI 3 , resulting in a low PCE of 0.09% (see Table 1). [26] Snaith and co-workers introduced a small amount of hydroiodic acid (HI) into the CsPbI 3 precursor solution to prepare black phase films. These films exhibited a stable α phase with small grains at room temperature, yielding an improved PCE of 2.9%. [ [74] precursor to mitigate the black phase transition to γ-CsPbI 3 . By employing this bication EDA stabilization method, a high efficiency of 11.8% was reached. In 2018, Zhao and co-workers developed an organic salt post-treatment on CsPbI 3 films and improved the CsPbI 3based PSCs efficiency ( Figure 4a). [71] The large phenylethylammonium (PEA) cations were physically absorbed on the CsPbI 3 films without forming a 2D perovskite capping layer by ion exchange (Figure 4b-e). The organic cations exhibited effective defect passivation, which was responsible for improved phase stability and moisture resistance. Accordingly, PSC with PEAI post-treatment obtained a champion PCE of 13.5%. Motivated by this work, a series of organic salts such as phenyltrimethylammonium bromide (PTABr), choline iodine (CHI) and phenyltrimethylammonium chloride (PTACl) were successfully employed to continuously improve the recording-PCE of CsPbI 3 -based PSCs up to 19.03%. [32][33][34] Based on the fact that the residual DMSO in the as-coated film increases the mass transport and diffusion, Wang et al. developed a solvent-controlled growth (SCG) method to prepare high crystallization and pinhole-free α-CsPbI 3 films (Figure 5a-c). [42] The CsPbI 3 -based PSCs showed a PCE of 15.7% and over 500 h stability under continuous light soaking. To improve the phase stability at room temperature, Bai et al. constructed a heterojunction between 0D Cs 4 PbI 6 and 3D CsPbI 3 (Figure 5d-f). [72] The 0D Cs 4 PbI 6 , located at the grain boundaries of CsPbI 3 , stabilized the black phase and passivated the surface defects, leading to an impressive PCE of 16.39%.
The long-term stable CsPbI 3 was synthesized by using a polymer-induced surface passivation engineering (Figure 6a-d). [70] A thin poly-vinylpyrrolidone (PVP) layer enhanced electron cloud density and lower surface energy of CsPbI 3 , therefore stabilized it in the α-phase. An ultra-long carrier diffusion length of over 1.5 μm was obtained which contributed to improved charge collection efficiency and PCE. Moreover, a mediator-antisolvent strategy (MAS) was demonstrated combining phenyl-C61-butyric acid methyl ester (PCBM) in chlorobenzene (CBZ) antisolvent and methylammonium iodide (MAI) mediator. This strategy allowed synthesizing high crystalline and pinhole-free CsPbI 3 films (Figure 6e-g). [73] PCBM in CBZ antisolvent greatly reduced the CsPbI 3 grain sizes. Meanwhile, MAI in perovskite precursor served as a crystallization mediator and subsequently induced oriented crystal growth. Taking advantage of the film morphology and electronic quality, the devices exhibited a significant high PCE of 16.04% and maintained 95% of their initial efficiency for over 1000 h storage in N 2 condition. In view of processing upscalability, all-printable CsPbI 3 -based PCSs were reported by Liu and co-workers. They utilized multifunctional molecular additive, i.e., Zn(C 6 F 5 ) 2 , to bridge SnO 2 and perovskite, which not only improved the perovskite crystallinity but also F I G U R E 3 a) Schematic mechanism for stabilization of CsPbI 3 in the γ-phase, b) XRD patterns of freshly prepared and aged CsPbI 3 samples. The aged sample was stored in an ambient condition for one month. c) Gibbs free energy of CsPbI 3 polymorphs and bulk δ-CsPbI 3 as a function of surface area, d) J-V curve of the best PSCs. Reproduced with permission from Ref. 43 Copyright 2018 American Chemical Society. e) UV-vis spectra of CsPbI 3 films prepared from PbI 2 +CsI and PbI 2 ⋅xHI +CsI, f) XRD patterns of CsPbI 3 ⋅xEDAPbI 4 , g) photovoltaic performance of PSCs with CsPbI 3 ⋅xEDAPbI 4 . Reproduced with permission from Ref. 29 Copyright 2017 American Association for the Advancement of Science allowed a favorable energy level alignment (Figure 6h-j). [74] As a result, a PCE of about 19% was obtained with increased open-circuit voltage.

CsPbI 2 Br
CsPbI 2 Br, with a bandgap of approximately 1.93 eV, is a popular composition for PSCs considering the efficiency and stability. The CsPbI 2 Br film is more stable in the α phase compared with the pure CsPbI 3 at room temperature, although the transition to the δ phase still occurs in the high humidity environment. [52,96] Such a bandgap makes CsPbI 2 Br a promising top cell candidate for tandem and triple-junction solar cells.
Although solubility of bromide salt is low in the common solvents, e.g., N,N-dimethylformamide (DMF), solution processing was used because of the relatively low bromide content in CsPbI 2 Br. Snaith et al. first obtained high-quality CsPbI 2 Br films via a one-step spin-coating method. The PSCs achieved a PCE of 9.8%. [52] However, CsPbI 2 Br shows limited solubility in DMF (0.4 M). Therefore thick-ness of this CsPbI 2 Br film is only 150 nm. McGehee et al. employed DMSO, a solvent with stronger polarity, to fabricate the invert-structured CsPbI 2 Br PSCs, and achieved a champion PCE of 6.69%. [97] Soon after, Park et al. demonstrated that the formation of black polymorph is critical to both solar cell efficiency and phase stability. [98] By carefully tuning the annealing condition, the PSCs presented an efficiency of 10.7% and an open-circuit voltage of 1.23 V. Later on, Chen and co-workers systematically studied the effects of solvent composition and precursor concentration on film electronic quality and thickness (Figure 7a). [99] They showed that a suitable solvent composition with 40% DMSO is favorable for the preparation of a pinhole-free and surface-smooth perovskite film with improved crystallinity. Chen et al. modulated the film growth kinetics via optimizing solvent composition (DMF: DMSO volume ratio) to enhance mass transport during film growth and achieved a stabilized PCE of 14.31% (see Table 2). [78] Zheng and co-workers developed a hot-casting method to deposit CsPbI 2 Br films ( Figure 7b). They casted the perovskite precursor solution on a hot substrate (maintained at 55 • C) and annealed the sample at a low-temperature to obtain CsPbI 2 Br films with large F I G U R E 4 a) Schematic drawing of the post-treatment process for fabrication of CsPbI 3 , b) illustration of PEAI post-treatment, c) SEM images of a pristine CsPbI 3 film and an IPA washed xPEAI-CsPbI 3 film, d) UV-vis spectra and XRD patterns of pristine and aged CsPbI 3 and PEA + -CsPbI 3 films. The aged films were exposed in an ambient atmosphere for 0.5 h under 85%-90% relative humidity (RH) at room temperature, e) chemical structure of PEAI, f) J-V characteristics of PSCs with and without PEAI post-treatment. Reproduced with permission from Ref. 71  The solvent engineering method is commonly used to prepare high electronic quality organic-inorganic hybrid perovskite films. Upon dripping the antisolvent on the precursor film, solubility of the solute in solution rapidly decreases, accelerating the nucleation and later on the crystal growth. This method can be applied to prepare CsPbI 2 Br films. Dong et al. used a low boiling point green EA solvent to fabricate inorganic perovskite film. [100] The EA-treated film presented compact morphology and a large grain size. Soon after, the authors developed an antisolvent assisted multi-step deposition strategy for fabrication of CsPbI 2 Br film with a pure phase, high crystallinity and full coverage. Different antisolvents were introduced in the PbI 2 precursor via pro-cessing to construct porous PbI 2 (DMSO) films. [92] The porous film was coated with multiple layers of CsBr to ensure sufficient reaction between the PbI 2 and the CsBr. PSCs using this film achieved a champion PCE of 10.21% (Figure 8a). Li et al. reported a multi-step annealing approach to precisely control the crystal growth kinetics, involving a gradient thermal annealing (GTA) process and IPA antisolvent post-treatment (Figure 8b,c). [88] The CsPbI 2 Br film showed an averaged grain size of one-micron and reduced defect density. The multi-step annealing approach led to a high PCE of 16  precursor solution. They fabricated thick and pinhole-free CsPbI 2 Br films with large crystalline grain size and high homogeneity. Devices with the perovskite films achieved a maximum PCE up to 14.81%. [82] Recently, Hu and coworkers developed a pre-annealing strategy for perovskite film formation, which allows to fine-tune the perovskite film nucleation and crystallization kinetics (Figure 8d-e). [84] Fabrication of CsPbI 2 Br films using highly crystalline CsPbI 2 Br NCs solution is another effective approach. The surfactants capping the nanocrystals (NCs) surfaces allow the NCs dispersing homogeneously in a nonpolar solvent. This allows fabricating uniform perovskite films independent of substrates. Yang et al. prepared the CsPbI 2 Br layers with the NC solution (Figure 9a-c). [101] A solvent soaking process in anhydrous isopropanol (IPA) was applied to the NC film to not only remove the ligands but also allow the NCs to reassemble and form a continuous film. PSCs with absorber prepared by this method achieved a PCE of 12.02%, a V oc of up to 1.32 V, and a V oc deficit of 0.5 eV. It worth noting that the whole device fabrication process was conducted under ambient air conditions. Moreover, vacuum deposition has also been demonstrated to be an effective method for fabrication of perovskite films. The method offers several advantages such as high film uniformity, good batch to batch reproducibility, and is suit-able for integration into industrial facilities.  Lin and co-workers first applied a co-deposition technique to fabricate CsPbI 2 Br films (Figure 9d-e). [95] The CsPbI 2 Br films presented a substantially small crystalline domain size of 100 nm. After annealing, the grain size becomes as large as 3 μm. As a result, the vacuum-deposited CsPbI 2 Br PSCs achieved a remarkable PCE of 11.8%, a high V oc of 1.13 V and negligible J-V hysteresis.
Zhao et al. developed a blade coating technique to fabricate the large area CsPbI 2 Br film (Figure 9f-h). [94] The results indicate that both Benard-Marangoni instability and the moisture attack can be eliminated at a moderate processing temperature, leading to the formation of a high-quality film. The blade-coated device efficiencies reach 14.7% on an area of 0.03 cm 2 cell and 12.5% on an area of 1.0 cm 2 , respectively.

CsPbIBr 2
CsPbIBr 2 , with a bandgap of 2.05 eV, appears in red color in the cubic phase and nearly transparent in the orthorhombic phase. Due to high bromide content, the CsPbIBr 2 devices exhibit much higher stability than the CsPbI 3 devices but suffer from a lower PCE. The one-step and two-step  (Figure 10a). [114] They kept the substrate at 75 • C during the film deposition and performed a post-annealing at 250 • C to achieve the films with a grain size reaching 500-1000 nm. PSCs based on the optimal film achieved a remarkable PCE of 4.7% without an HTL (see Table 3). Soon after, they also developed a spray assisted solution strategy to fabricate high-quality CsPbIBr 2 film, which overcame the low solubility issue of the bromide ion in the precursor solution (Figure 10b). [115] The best-performing device fabricated at the optimized conditions achieved a stabilized efficiency of 6.3% and negligible hysteresis. Song et al. employed the antisolvent and organic ion surface passivation strategies to precisely control the growth of CsPbIBr 2 crystals (Figure 10c-f). [112] A pure phase CsPbIBr 2 film of full coverage and high crystallinity with preferable (100) orientation was successfully obtained by introducing diethyl ether as the antisolvent, followed by guanidinium surface passivation. The optimal CsPbIBr 2 film showed large grains with an average size of 950 nm, few grain boundaries, and high hydrophobicity. The device based on this film exhibited a PCE of 9.17%.
In addition, one-step spin-coating method is also recognized as an effective and convenient method for CsPbIBr 2 thin film fabrication. Zhang et al. demonstrated a light processing technology that enabled pure-phase CsPbIBr 2 films with large grains, high crystallinity with [99] grains orientation, and favorable electronic structure> (Figure 11a). [106] The resulting carbon-based, all-inorganic planar cells showed a PCE of 8.60% with the V oc of 1.283 V. Que et al. reported a pre-heating assisted one-step spin-coating method (Figure 11b). [108] During spin-coating, the high-temperature substrate accelerates volatilization of the solvent molecule. As a result, CsPbIBr 2 films show complete coverage and  (Figure 11c). [107] Microscopy and spectroscopy characterization indicated that the CsPbIBr 2 films were composed of high-crystallinity, (100)-oriented, micrometer-sized crystalline grains. As a result, the cost-effective, carbonbased all-inorganic planar cells fabricated by this route delivered an optimized PCE of 9.16% and a stabilized PCE of 8.46% in ambient condition.

CsPbBr 3
CsPbBr 3 shows a direct bandgap of 2.25-2.37 eV, which varies with different fabrication approaches. [122] The absorption edge of the CsPbBr 3 film is shorter than 540 nm, and the characteristic absorption peak is located at 520 nm. Tuning the ratio between CsBr and PbBr 2 further increases the bandgap from 2.3 to 4.0 eV because the phase transition occurs from the cubic perovskite structure to the derivative phases (Cs 4 PbBr 6 and CsPb 2 Br 5 ). Notably, this compound is among the most stable lead halide perovskite materials under ambient condition. The solution-processed method was usually employed to prepare CsPbBr 3 films due to low cost. Meanwhile, vapor-based deposition approaches, including co-evaporation and sequential deposition, were also investigated. These approaches allow for the fabrication of thick CsPbBr 3 films that are technically challenging for solution-processed approaches because of limited precursor solubility. Usually, DMF, DMSO, γ-butyrolactone (GBL), and a combination of mixed solvents are used to dissolve perovskite precursor materials. However, due to the poor solubility of CsBr, it is difficult to prepare CsPbBr 3 precursor with a molar ratio of 1:1.  that the maximum concentration of CsPbBr 3 precursor solution was only 0.4 M in the mixed DMF and DMSO solvent. It was difficult to form a thick and full-coverage CsPbBr 3 film by the one-step spin-coating method. [42] To overcome the insolubility issue of CsBr, Zhong et al. used ionic liquid methylammonium acetate (MAAc) and cesium acetate (CsAc) to increase the concentration of the CsPbBr 3 precursor (i.e., 1.0 M) and tailor the crystallization kinetics. CsPbBr 3 films with large sized crystalline grains, high uniformity and coverage were formed (Figure 12a) (see Table 4). [118] The champion CsPbBr 3 PSCs achieved an efficiency of 7.37%. Meanwhile, the two-step solution-coating method was developed to prepare CsPbBr 3 films. For example, Tang and coworkers spin-coated the PbBr 2 and CsBr solution sequentially (Figure 12b). [123] The CsBr solution was spin-coated six times to realize an optimized material composition. The all-inorganic perovskite solar cell reached a PCE of 9.72 %.
Vapor deposition method was also employed to fabricate CsPbBr 3 film. Qi et al. controlled the ratio of CsBr and PbBr 2 to fabricate the perovskite derivative phases (CsPb 2 Br 5 /Cs 4 PbBr 6 ) via a vapor growth method (Figure 12c). [102] Upon post-annealing, the perovskite derivative phases, acting as nucleation sites, transformed into the pure CsPbBr 3 phase via a crystal rearrangement process. The method effectively retarded the perovskite  (Figure 1d). [124] By tuning the thickness of CsBr, a series of phase conversion from CsPb 2 Br 5 to CsPbBr 3 and to Cs 4 PbBr 6 could be precisely controlled. The optimized CsPbBr 3 PSC delivered a PCE of 10.45%.

CONCLUSIONS AND PERSPECTIVES
Within a decade of intensive research and development, perovskite solar cells have witnessed substantial progress. The organic-inorganic hybrid perovskite, being the most successful perovskite system, has reached a certified efficiency of 25.5%, which surpasses all the other thin-film solar cell tech-nologies i.e., CIGS, CdTe, amorphous Si, and is on the way to catch up with crystalline Si and GaAs. The outstanding PCE has attracted considerable attention not only in the research labs but also in the photovoltaics companies for various applications. Despite the impressive performance, organicinorganic hybrid PSCs still present significant challenges in terms of stability. The relatively weak bonding between organic cation and halide ion and the volatile nature of the organic cations makes organic-inorganic hybrid perovskite intrinsically unstable when exposed to oxygen, moisture, illumination and temperature. By substitution of organic cations with Cs + , the inorganic perovskite, i.e., CsPbX 3 offers the promise for thermodynamical stability, therefore mitigating many external impact-induced degradation pathways.
In the past years, much research effort has been devoted to the development of deposition approaches for achieving high electronic quality CsPbX 3 perovskite films. Efficiency of CsPbX 3 PSCs has boosted to over 19% (Figure 13a) with    V oc deficit (E g -eV oc ) approaching 0.5 eV (Figure 13b). Nevertheless, many problems still exist, deteriorating the device efficiency, stability and upscalability. The following investigation directions have been proposed to further unleash the efficiency potential of CsPbX 3 beyond 20%, improve the long-term operational stability and large-area processability (Figure 13c). I) Development of advanced inorganic perovskite deposition methods. So far, precisely control the composition and structure of inorganic perovskite films at the microscopic scale is still a challenge. Investigation of chemically coordinated additives, such as low-dimensional perovskite, ammonium salts and organic passivation molecule, would be beneficial for adjusting crystal nucleus formation and grain growth kinetics, functioning grain boundaries and protecting layers against external impact. In addition, a systematical study of the solvent composition will further provide insights into solution-processed inorganic perovskite formation kinetics and allow fine-tuning the solvent drying and crystal growth.
II) Design rationale interfaces between perovskite and charge transfer layers. Energy mismatch between inorganic perovskite and charge transfer layer results in an energy barrier for carriers extracting and increases interface recombination. A recent report has shown that an optimized interface with favorable energy alignment between CsPbI 2 Br and P3HT can effectively decrease interface recombination and increase the device open-circuit voltage. [84] Different inorganic perovskite materials, i.e., CsPbI 3 , CsPbI 2 Br, CsPbIBr 2 , and CsPbBr 3 , show different band structures. Therefore, it is of significant importance to develop a series of new electron or hole transfer layers that are suitable for different inorganic perovskite materials from an energy level alignment point of view.
III) Strategies to improve device stability. So far, stabilization of inorganic perovskite in the photoactive phase over the device lifetime is still a challenge. [125] To unleash the stability potential, it is urgent to gain fundamental understandings of the material chemistry and device physics. From this viewpoint, the partial incorporation or doping of suitable metal ions might improve the intrinsic stability, which in turn allows all-inorganic PSCs to withstand severe environmental conditions. Furthermore, the preparation of inorganic perovskite films with large grains and dense morphology, the construction of continuous organic or inorganic transport layer with high moisture resistance and optimization of encapsulation methods are also critical strategies to improve the stability of all-inorganic PSCs.
IV) Strategies to improve processing upscalability. At present, there is still a considerable efficiency gap between the inorganic PSCs and large-area solar modules. Along with the continuous improvement of device efficiency and stability, more and more researches should be done in these fields. The knowledge obtained on the vacuum-and solutionprocessed methods provides significant insight for the development of large-scale preparation of inorganic perovskite films. Moreover, to further achieve ultra-uniform fabrication of perovskite thin films on large scale, it is necessary to understand in depth the mechanism of the solution-to solid-phase nucleation and crystal growth in the future.

A C K N O W L E D G E M E N T S
F.-Z.Q. and M.-H.L. contributed equally to this work. This work was supported by funding from the Chinese Academy of Sciences, Beijing Natural Science Foundation (2202030) and Energy Materials and Optoelectronics Unit of Songshan Lake Materials Laboratory.

C O N F L I C T O F I N T E R E S T
The authors declare that there is no conflict of interest.