Organic Ligands Armored ZnO Enhances Efficiency and Stability of CsPbI2Br Perovskite Solar Cells

Abstract Inorganic perovskite solar cells (PSCs) have witnessed great progress in recent years due to their superior thermal stability. As a representative, CsPbI2Br is attracting considerable attention as it can balance the high efficiency of CsPbI3 and the stability of CsPbBr3. However, most research employs doped charge transport materials or applies bilayer transport layers to obtain decent performance, which vastly complicates the fabrication process and scarcely satisfies the commercial production requirement. In this work, all‐layer‐doping‐free inorganic CsPbI2Br PSCs using organic ligands armored ZnO as the electron transport materials achieve an encouraging performance of 16.84%, which is one of the highest efficiencies among published works. Meanwhile, both the ZnO‐based CsPbI2Br film and device show superior photostability under continuous white light‐emitting diode illumination and improved thermal stability under 85 °C. The remarkable enhanced performance arises from the favorable organic ligands (acetate ions) residue in the ZnO film, which not only can conduce to maintain high crystallinity of perovskite, but also passivate traps at the interface through cesium/acetate interactions, thus suppressing the photo‐ and thermal‐ induced perovskite degradation.


Introduction
Perovskite optoelectronic devices have skyrocketed tremendous progress over the past decade. As one of most significant applications, hybrid organic-inorganic perovskite solar cells (PSCs) have achieved 25.2% certified power conversion efficiency (PCE) due to the excellent optoelectronic properties of perovskite DOI: 10.1002/advs.202000421 semiconductors. [1] However, the instability of MA + and FA + (MA = methylamine, FA = formamidine) organic cations of hybrid perovskite impedes the commercial process of PSCs as the photovoltaic modules require high thermal stability during long-term solar irradiation. Thus, recent researches have introduced alkali metal cesium Cs + into the crystal lattice to stabilize the perovskite structure, [2][3][4] however this can only alleviate rather than eliminate the thermal instability. Therefore, all inorganic perovskites are introduced to suffer extreme thermal condition, which can hold stable perovskite phase over 300°C. [5,6] CsPbBr 3 is a very stable perovskite material, PSCs of which have demonstrated nearly no degradation under 90−95% relative humidity (RH) or 100°C over 800 h, but the wide bandgap (≈2.3 eV) of CsPbBr 3 makes it hardly to absorb sunlight over 550 nm, thus achieving unsatisfied photovoltaic performance. [7] Though CsPbI 3 with a bandgap of ≈1.73 eV can extend the absorption spectrum to 720 nm to receive enhanced current density and PCE, the heavily deflected Goldschmidt tolerance factor (0.81) of CsPbI 3 make it easy to form undesired non-perovskite -phase from -phase in ambient atmosphere, even in glovebox. [8] Considering adding smaller radius Br − into the I-based perovskite film can improve phase stability, which have been widely employed in organic-inorganic hybrid PSCs, many researches have attempted to change the ratio of I − and Br − of inorganic PSCs to balance performance and stability. [9][10][11][12] As a prominent representative, CsPbI 2 Br, with reduced phase transition temperature and desired bandgap (≈1.92 eV) between CsPbBr 3 and CsPbI 3 , perform decent PCE and stability by various methods since 2016. [12][13][14][15][16][17][18][19] Hayase and co-workers reinforced the phase stability and performance of PSC by partially substituting Pb with germanium (Ge) to form CsPb 0.8 Ge 0.2 I 2 Br, presenting excellent stability in ambient atmosphere. [13] Li and co-workers combined gradient thermal annealing and green anti-solvent treatment to precisely control the morphology of CsPbI 2 Br, obtaining a high PCE of 16.07% and excellent ultraviolet (UV) stability in N 2 atmosphere. [14] Cao and co-workers employed dual interfacial organic molecules to passivate trap state and restrain the photo-induced halide segregation of CsPbI 2 Br, whilst enhancing the PCE up to 16.2% and the long-term photostability. [15] Yang and co-workers used oleylammonium to induce the secondary grain growth of perovskite film, enlarging grain sizes up to 4 µm and maintaining 90% of PCE under 85°C after 500 h. [16] For the aforementioned attempts to achieve high performance and stability, most of them employ titanium dioxide (TiO 2 ) or tin dioxide (SnO 2 ) as the electron transport materials (ETMs), which is not as ideal as zinc oxide (ZnO) due to the more matched energy level of ZnO with perovskite, as well as the higher electron mobility and transparency of ZnO ETM comparing to TiO 2 and SnO 2 . [20,21] However, ZnO rarely attracts the attention of researchers in PSCs though it has been widely applied in organic solar cells, [22][23][24] because ZnO can deprotonate MA + and FA + cations and therefore induce the degradation of hybrid perovskites. Although incorporating inorganic cations (potassium, rubidium, Cs + ) into the organic/inorganic hybrid perovskites or pretreating ZnO materials can stabilize perovskite to some extent, the unsatisfied stability and complicated fabrication processes are not favorable during commercialization. [25][26][27][28] After substituting all organic cations with Cs + to form inorganic perovskite, organic components are absent to allow ZnO to degrade. Cao and co-workers [29] applied a SnO 2 /ZnO bilayer ETL to promote desirable cascade energy levels with CsPbI 2 Br, devices of which exhibit improved stability and open-circuit voltage (V OC ) (1.06 to 1.23 V), demonstrating that ZnO is a promising ETM for inorganic PSCs. Besides, state-of-the-art inorganic PSCs utilized doped HTMs (Spiro-OMeTAD or PTAA) or bilayer ETMs to obtain high efficiency, however the introduction of dopants will induce the fast degradation of devices and the bilayer ETMs will complicate the fabrication process. [14][15][16]30,31] Therefore, it is urgent to demonstrate high-efficiency and stable inorganic PSCs with simplified preparation procedures.
Herein, we have fabricated all-layer-doping-free planar heterojunction inorganic PSCs with a structure of Glass/ITO/ ETM/CsPbI 2 Br/polyTPD/MoO 3 /Ag by employing organic ligands armored ZnO as the ETM, which remarkably outperform the TiO 2 -and SnO 2 -based devices. After optimizing thermal annealing temperatures to control the content of organic ligands, champion device of ZnO-based inorganic PSCs exhibits a high PCE of 16.84%, which is one of the highest photovoltaic performance of CsPbI 2 Br-based inorganic PSCs. Upon a continuous white light LED illumination, ZnO/perovskite film can maintain integrity over 100 h in ≈10% RH and the ZnO-based device with encapsulation can maintain 90% PCE over 200 h. Besides, the ZnO-based device can maintain 86% PCE after 400 h continuous thermal annealing under 85°C. We attribute the enhanced performance, photostability and thermal stability to the proper content of acetate ions in the ZnO film, which can not only conduce to maintain excellent crystallinity of perovskite but also passivate perovskite film at the interface through cesium/acetate interactions that were confirmed through the comprehensive analysis of XPS, FTIR, 1 H NMR and theoretical calculation. Our work provides a promising ETM candidate and facile procedure to fabricate high performance and stable inorganic PSCs.

Results and Discussion
We employed a two-step annealing process to fabricate smooth perovskite film using TiO 2 , SnO 2 , and ZnO as ETMs, with the ZnO based device structure was shown in Figure 1a. As recorded in Figure S1 (Supporting Information), we measured the performance of inorganic PSCs using different ETMs under AM 1.5G simulated sun light, and obtained improved PCE from 11.6% (TiO 2 ) and 12.4% (SnO 2 ) to 16.84% (ZnO) from reverse scans, derived from the elevated V OC (from 1.01 and 1.09 to 1.24 V), short-circuit current (J SC ) (from 15.2 and 15.0 to 16.5 mA cm −2 ) and fill factor (FF) (from 75.5% and 75.8% to 82.1%). Admittedly, there are hysteresis in all these PSCs, agreeing well with literature reports. [16,30,[32][33][34] The enhanced V OC arises from more matched energy level alignment between ZnO (−4.31 eV) and CsPbI 2 Br (−4.16 eV) compared to TiO 2 (−4.43 eV) ( Figure S1a, Supporting Information), which is conducive to decrease energy barrier and facilitate electron transport at the ETMs/perovskite interface. [29,35] Besides the quality of the perovskite layer grown on these ETMs and the interfacial charge transport efficiency, the higher electron mobility and transparency of ZnO also contribute to promote J SC and FF, which have been widely reported in many publications. [20,36,37] The performance of those TiO 2 and SnO 2 based PSCs can be improved upon further bulk and interfacial treatments. [18] Afterward, SnO 2 /CsPbI 2 Br and ZnO/CsPbI 2 Br films were selected to be illuminated under simulated sunlight at 20-30% RH to evaluate the photostability of perovskite films, and the corresponding evolution of measured PL were recorded in Figure S2a,b, Supporting Information. The continuous increase of PL intensity and emission wavelength in SnO 2 /CsPbI 2 Br and ZnO/CsPbI 2 Br films indicate photo-induced halide phase  separation of CsPbI 2 Br, which has been widely observed in mix-halide PSCs, and the ambient humidity (20-30% RH) may amplify the evolution process. [38][39][40][41][42] However the emission wavelength of ZnO/CsPbI 2 Br film only increases from 646 to 653 nm, which is much lower than the SnO 2 /CsPbI 2 Br film (646 to 663 nm) ( Figure S2c, Supporting Information), suggesting that the ZnO film can effectively slowdown the halide phase separation of CsPbI 2 Br therefore contribute to enhanced photostability.
In order to clarify the reasons of the simultaneously enhanced performance and photostability of ZnO-based device, we altered the ZnO annealing temperature to obtain different performance of devices and the corresponding device J-V curves from reverse scans are illustrated in Figure 2a. The device with 100°C annealed ZnO gives an unsatisfied performance of 5.91%, a V OC of 1.08 V, a J SC of 11.14 mA cm −2 and an FF of 43.1%. After elevating the annealing temperature toward ZnO, device PCE first increases to 15.32% and 16.84% upon annealing at 130 and 150°C then decreases to 13.84% upon annealing at 200°C, and all photovoltaic parameters vary with PCEs including V OC (1.19, 1.24, and 1.14 V), J SC (15.72, 16.54, and 15.16 mA cm −2 ) and FF (81.7%, 82.1%, and 79.9%). Figure 2b illustrates the external quantum efficiency (EQE) of four kinds of devices, and the relevant integrated J SC values are 10.80, 15.02, 16.06, and 14.62 mA cm −2 respectively. The difference of J SC obtained from J-V curves and EQE are all below 5%, confirming the accuracy of current density from J-V curves. The detailed photovoltaic parameters are summarized in Table 1. Figure 2c shows the J-V curves of champion device from the reverse scan with a PCE of 16.84% and from the forward scan with a PCE of 13.91%, which is the highest performance in inorganic perovskite based on CsPbI 2 Br devices (see summary in Table S2, Supporting Information). Admittedly, the hysteresis of CsPbI 2 Br remains outstanding and urges further investigation via bulk or interfacial modifications to reduce defect and balance charge extraction.
We continued to measure the transmittance and X-ray diffraction (XRD) of ZnO films annealed at 100, 130, 150, and 200°C, and find that there are no obvious differences, [20] meantime the conductivity (3.58 × 10 −3 , 3.93 × 10 −3 , 4.00 × 10 −3 , and 3.49 × 10 −3 mS cm −1 ) and electron mobility (4.39 × 10 −3 , 4.36 × 10 −3 , 4.10 × 10 −3 , and 4.05 × 10 −3 cm 2 V −1 s −1 ) are also similar, so the variations of efficiency are not strongly related to these parameters of the ZnO film ( Figure S3, Supporting Information). Then we proceeded to deposit ZnO with the same thickness on KBr pellets to measure the infrared absorption spectrum after annealing at 100, 130, 150, and 200°C, therefore we can probe the relative amount of organic ligands by comparing the absorption intensity. As shown in Figure 2d, the peaks at 3405 (-NH 2 ) and 1340 (-OH) nm −1 can represent ethanolamine, 1570 (COO − ) nm −1 can represent CH 3 COO − (Ac − ). [43] With the increase of annealing temperature, the amount of organic compounds in ZnO thin films decreases dramatically, indicating less organic ligands could interact with perovskite. Thereafter, we used scanning electron microscope (SEM) to observe the morphology of perovskites deposited on different temperature annealed ZnO. In Figure S4a-d, Supporting Information, perovskite cast on 100°C annealed ZnO exhibits lots of pinholes, probably arising from the secondary volatilization of organic ligands during the perovskite crystallization process (annealing at 240°C). After improving the annealing temperature of ZnO, pinholes of perovskite films are obviously reduced by casting on ZnO annealed at 130°C then completely eliminated by casting on ZnO annealed at 150 and 200°C. Therefore less organic ligands of ZnO guarantees the formation of smooth and compact perovskite films, which can tremendously reduce pinhole-induced film defects. The XRD spectra of perovskite films grown on these ZnO are shown in Figure S4e (Supporting Information), from which we can identify stronger diffraction peaks of perovskite cast on ZnO annealed at 150°C, which corresponds to higher crystallinity. The enhanced crystallinity of CsPbI 2 Br perovskite will improve the light absorption ability and lead to higher J SC , a result that is consistent with our J-V measurements presented in Figure 2a and Table 1. X-ray photoelectron spectroscopy (XPS) is a powerful tool to probe the interaction between different thin films. [15,44,45] Hence we fabricated films of CsPbI 2 Br, ZnO and ZnO/CsPbI 2 Br to investigate interactions between organic ligands and CsPbI 2 Br. Figure 3a,b shows the Cs 3d and O 1s data of CsPbI 2 Br, ZnO, and ZnO/CsPbI 2 Br films. After the deposition of an ultrathin CsPbI 2 Br film (≈7 nm) on ZnO, the Cs 3d peak shifted toward a lower binding energy compared to the pristine CsPbI 2 Br film, indicating that electron density around Cs atoms is increased (Figure 3a). Meantime O 1s of ZnO and ZnO/CsPbI 2 Br films show different peak shape in Figure 3b. Due to the multi-sources of the oxygen element in the ligand containing ZnO film, we have differentiated peaks of O 1s into Zn-O, oxygen-deficient (O v ) and C=O components (representing CH 3 COO − ) from the lowest to the highest binding energy, and the corresponding differentiated peaks were plotted in Figure 3c,d. Both intensities of C=O and O v of the ZnO/CsPbI 2 Br film become weaker than those in pure ZnO film excluding Zn-O, indicating that partial content of CH 3 COO − has been volatilized and more Zn atoms are bound to O atoms during the perovskite annealing process, which agrees well with the secondary volatilization to induce pinholes in perovskite films observed in SEM images. C=O moves toward a higher binding energy, indicating that the electron density around C=O is decreased. [20,46,47] Therefore, the fact that Cs atoms and C=O move toward opposite directions in binding energy suggests that electrons of O in CH 3 COO − can donate to Cs to form strong Cs-Ac interaction to passivate traps of perovskite, which has been reported in previous publication. [48] Whereas the peak positions of Pb 4f exhibit no shift between ZnO/CsPbI 2 Br and pure CsPbI 2 Br films ( Figure S5a, Supporting Information), indicating that the ZnO film has negligible passivation effect to Pb atoms. Figure 2d also shows the existence of -NH 2 and -OH groups in the ZnO film, whilst the -OH group cannot be recognized as the passivation units due to its electron withdrawing character, XPS spectrum does show weak interactions between -NH 2 and perovskite (see Figure S5b, Supporting Information). However the XPS peak intensity of -COO-group is dozens of times stronger than that of -NH 2 group, suggesting that the passivation effect of the trace amount of NH 2 groups is a secondary factor. From the above information, we can conclude that CH 3 COO − can form Cs-Ac interactions and effectively passivate traps of Cs atoms in inorganic CsPbI 2 Br perovskite (as schemed in Figure 1b), contributing to higher performance. Critically, excess organic ligands will create pinholes in perovskite films, only applying a 150°C annealing of ZnO can we develop pinhole-free and highly passivated perovskite films.
To further verify the mechanism of interaction between acetate ion ligands in ZnO film and CsPbI 2 Br, we performed Fourier transform infrared (FTIR) spectra and 1 H nuclear magnetic resonance ( 1 H NMR) of pure Zn(Ac) 2 , Zn(Ac) 2 /PbI 2 , and Zn(Ac) 2 /CsBr mixtures. As shown in Figure 4a, the peaks of Zn(Ac) 2 in 1560 and 1445 cm −1 represent asymmetric and symmetric stretching of the carboxylate ion (−COO − ). After adding CsBr into Zn(Ac) 2, the strong asymmetric stretching peak shifts to 1591 cm −1 and the symmetric stretching peak shifts to 1393 cm −1 , indicating strong interactions between Ac − and Cs atom. Whilst the minor shift of these peaks in the Zn(Ac) 2 /PbI 2 mixture suggests negligible interactions between Ac − and Pb atom. [48] The result is also confirmed by 1 H NMR. In Figure 4b, all samples are calibrated by the peak of dimethyl sulfoxide-d6 (DMSO-d6) locating at 2.50 ppm, with peaks in 3.30-3.50 ppm are contributed by water. After mixing PbI 2 and CsBr with Zn(Ac) 2 , the original peak of Ac − from Zn(Ac) 2 shifts from 1.81 to 1.80 and 1.76 ppm respectively, indicating that Ac − is more likely to interact with Cs rather than Pb atoms. [49] We continued to calculate the formation energy of perovskite film with and without the presence of Ac − bonding by the density functional theory (DFT) method, with detailed calculation information presented in the Supporting Information. In Figure 4c, the formation energy of I vacancy in CsPbI 2 Br is −59.9 eV, which retains after bonding Ac − with the Pb atom (Figure 4d), but turns more negative to −62.6 eV with the formation of Cs-Ac bonding (Figure 4e). This suggests that the Cs-Ac bonding can help to form more stable CsPbI 2 Br structure than the I vacancy and Pb-Ac bonding, which is consistent with previous report that the Ac − can strongly stabilize CsPbI 2 Br through Cs-Ac interactions. [48] The comprehensive analysis of XPS, FTIR, 1 H NMR and theoretical calculation therefore confirm that Ac − is more easily to interact with Cs atoms rather than Pb atoms, which will effectively passivate traps of Cs atoms in inorganic CsPbI 2 Br perovskite to impact on device performance.
We proceeded to probe charge recombination dynamics by using steady-state photoluminescence (PL). Figure 5a shows that the PL spectra of ZnO/CsPbI 2 Br films, and the 150°C annealed ZnO based film shows the strongest PL quenching, meaning the most effective electron extraction from perovskite to ZnO. In order to quantize this process, time-resolved photoluminescence (TRPL) was employed to record carrier lifetime of perovskite films. Figure 5b illustrates the TRPL curves of ZnO/CsPbI 2 Br films, with elevated annealing temperature of ZnO from 100 to 130, 150 and 200°C, corresponding lifetime firstly decreases from 41.9 to 15.0 and 8.9 ns then slightly increases to 9.2 ns, again confirming that the 150°C annealed ZnO based perovskite film have the fastest electron transport ability and least recombination. [50] Impedance spectroscopy measurements were utilized to investigate bulk and interface charge transport properties. Figure  S6a-d (Supporting Information) illustrate the impedance plots of ZnO-based (100, 130, 150, and 200°C annealing) devices under 1 sun illumination at a series of applied voltage, and the bias loading mode is recorded in Figure S6f (Supporting Information). All impedance plots can be well fitted by the equivalent circuit inserted in Figure 5c, and relevant fitting parameters R rec , R s , and CPE represent recombination resistance, series resistance and constant phase element respectively. [51] Figure 5c shows the decreased R rec of all devices along the increased applied voltage, and 150°C annealed ZnO based device has the largest R rec than all other devices, which indicates the maximally declined recombination in 150°C annealed ZnO based device, in accordance with the electron transport and efficiency results. This result could be also related to the reduced trap density at the 150°C annealed ZnO/CsPbI 2 Br interface compared to more recombination centers in 100, 130, and 200°C annealed ZnO based perovskite films. Thus we fabricated electron-only devices with the structure of ITO/ZnO/CsPbI 2 Br/PC 61 BM/Ag to quantitatively compute the trap density. In Figure 5d, J-V curves of double logarithmic coordinates can be divided into three parts, Ohmic, trap-filling limited and space charge limited current (SCLC) regions, and the transition point between Ohmic and trap-filling limited regions represents the trap-filled limit voltage (V TFL ), from which we can calculate trap density (n t ) from V TFL = en t L 2 /2 0 , where 0 and are the vacuum permittivity and the relative dielectric constant of CsPbI 2 Br respectively, L is the thickness of the CsPbI 2 Br film. [52] The calculated trap density of 100, 130, 150, and 200°C annealed ZnO based device is 3.46 × 10 16 , 1.57 × 10 16 , 9.44 × 10 15 , and 2.89 × 10 16 cm −3 respectively. The reduced trap density of 150°C annealed ZnO based device could be attributed to the effective interfacial passivation of the CsPbI 2 Br film by the CH 3 COO − ligands in ZnO film.
We continued to probe the thermal stability by recording insitu XRD at an annealing temperature of 150°C first for 30 min and then increased to 250°C for another 120 min. Figure 6a,b illustrates the XRD original data of CsPbI 2 Br film and its mapping figure, and the characteristic diffraction peaks of cubic structure CsPbI 2 Br at 14.8°and 29.7°represent to the (100) and (200) crystal planes respectively. [33,53] Figure 6c plots the peak values of CsPbI 2 Br (100) and (200) planes, from which we can find that the CsPbI 2 Br film exhibit almost identical intensity of various diffraction peaks, even elevated to a temperature of 250°C, indicating the excellent thermal stability of CsPbI 2 Br. Then we also measured in-situ high temperature XRD of organic-inorganic hybrid perovskite MAPbI 3 under the same condition. Figure 6d,e shows the XRD original data of MAPbI 3 and its corresponding mapping figure, and the characteristic diffraction peaks of MAPbI 3 at 14.3°, 28.5°, and 32.1°represent to the (110), (220), and (310) crystal planes respectively, and the diffraction peak at 12.8°can be attributed to PbI 2 . [51] We can find that all diffraction peaks of MAPbI 3 became weaker and the peak of PbI 2 got stronger during the heating process. For a clearer comparison, the strongest peak of MAPbI 3 (110) and the peak of PbI 2 are recorded in Figure 6f. During the first 30 min annealing under 150°C, there are not obvious change of diffraction peaks, but the intensity of PbI 2 increases and MAPbI 3 (110) decreases continuously when the annealing temperature was elevated to 250°C. Therefore the organic-inorganic hybrid perovskite is less thermally stable than the inorganic CsPbI 2 Br. Afterward we fabricate full devices of MAPbI 3 and CsPbI 2 Br to track the thermal stability and performance evolution of MAPbI 3 and CsPbI 2 Br PSCs under 85°C in N 2 glovebox ( Figure S7, Supporting Information). MAPbI 3 -based device is almost dead after 6 h and CsPbI 2 Br-based device can maintain 75% of the initial performance over 20 h, consistent with the in-situ high temperature XRD, and the quick decreased performance may be associated with the destruction of interface between polyTPD and silver electrode.
We subsequently applied continuous 100 mW cm −2 white light LED on ETMs/CsPbI 2 Br films under ≈10% RH in air to observe the film degradation. As shown in Figure 7a, the CsPbI 2 Br film on SnO 2 began to decompose apparently after 15 h and nearly disappeared after 39 h. The TiO 2 /CsPbI 2 Br film became transparent after 60 h and the ZnO/CsPbI 2 Br film showed little decomposition after 100 h. This result demonstrates the superior photostability of the ZnO/CsPbI 2 Br film, consistent with the suppression function of ZnO film in photoinduced halide phase separation of CsPbI 2 Br. We also tracked the performance evolution of encapsulated CsPbI 2 Br devices under continuous 100 mW cm −2 white LED illumination (Figure 7b). The 150°C annealed ZnO based device kept 90% of the initial performance after 200 h illumination, much higher than the TiO 2 -and SnO 2 -based devices (below 50%). Finally, we tracked the thermal stability of CsPbI 2 Br PSCs using different ETMs under 85°C in N 2 glovebox. Similar to the aforementioned film thermal stability, all devices using Ag electrode exhibit quick performance reduction, although the ZnO based CsPbI 2 Br PSC show the best thermal stability (Figure 7c). However, the overall thermal stability of CsPbI 2 Br PSCs can be enhanced through the replacement of Ag electrode with Au electrode to inhibit the reaction of Ag with perovskite, which has been well reported in the literature. [54] As shown in Figure 7d, with Au as the top electrode, the 150°C annealed ZnO based device maintained 86% of the initial performance after 400 h continuous thermal annealing under 85°C, much higher than the TiO 2 -(67%) and SnO 2 -(71%) based devices.

Conclusions
We have fabricated all-layer-doping-free planar heterojunction inorganic CsPbI 2 Br PSCs using organic ligands (acetate ions) armored ZnO as ETM, acquiring an encouraging photovoltaic performance of 16.84%, which is among the highest PCE of inorganic CsPbI 2 Br PSCs. Meantime, ZnO can remarkably suppress the photoinduced phase separation of CsPbI 2 Br film compared to SnO 2 , with the ZnO/CsPbI 2 Br film can maintain integrity in ≈10% RH over 100 h and the encapsulated ZnO-based PSCs can maintain 90% PCE over 200 h, upon continuous 100 mW cm −2 white light LED illumination. The ZnO-based PSCs can maintain 86% PCE over 400 h continuous thermal annealing under 85°C. The enhanced performance, photostability and thermal stability arise from the proper amount of acetate ions residue in the ZnO film by optimizing the thermal annealing temperature, which can not only maintain excellent crystallinity of CsPbI 2 Br but also effectively passivate perovskite through the formation of cesium/acetate interactions that are confirmed through the comprehensive analysis of XPS, FTIR, 1 H NMR, and theoretical calculation. Our work provides a simple and convenient method to fabricate high efficiency and stable inorganic CsPbI 2 Br PSCs.
Device Fabrication: Unless otherwise stated, all of fabrication were according to the previous work [35,55] SnO 2 solution was prepared by mixing SnO 2 precursor (Alfa Aesar) and deionized water at a ratio of 1:6, and was stirred at least 2 h before use, then was spin-coated on ITO whereafter annealed at 150°C for 30 min to form a compact SnO 2 layer. ZnO precursor was prepared by dissolving 200 mg Zinc acetate and 56 µL ethanolamine in 2 mL 2-methoxyethanol, and was stirred at least 12 h before use, then was spin-coated on ITO and annealed at several temperature (100, 130, 150, and 200°C) for 30 min to obtain ZnO layer (40 nm) with different amount of organic ligands residue. 1 m PbI 2 :CsBr (1:1) dissolving in DMF:DMSO (1:9) was deposited on metal oxide at 1500 rpm 10 s then 4000 rpm 30 s, then was blown by fan for 60 s, finally was annealed at 50°C for 30 s and 240°C for 60 s successively to form CsPbI 2 Br film.
Characterizations: Unless otherwise stated, all of characterization methods and techniques can be found in the previous work. [35,51] Briefly, the device J-V characterization was performed under AM 1.5G (100 mW cm −2 ) using a Newport 3A solar simulator in air at room temperature. The light intensity was calibrated using a standard silicon reference cell certified by the National Renewable Energy Laboratory (NREL, USA). J-V characteristics were recorded using J-V sweep software developed by Ossila Ltd. (UK) and a Keithley (USA) 2612B source meter unit. An aperture mask was placed over the devices to accurately define a test area of 2.12 mm 2 on each pixel and to eliminate the influence of stray and wave guided light. The J-V scan rate was 50 mV s −1 . External quantum efficiency (EQE) was measured with a Zolix (China) EQE system equipped with a standard Si diode. The steady-state PL (Flex One, Zolix, China) and time-resolved PL (Pico-Quant, Fluo Time 300, Germany) spectra were obtained utilizing 532 and 520 nm CW laser respectively. The in-situ high temperature XRD data were obtained using X-ray diffractometer (D8 DISCOVER, China). X-ray photoelectron spectra were acquired by X-ray photoelectron spectrometer (ES-CALAB 250Xi, America). Infrared absorption was measured using Fourier transform infrared spectroscopy (spectrum two, PerkinElmer, America), mixtures of Zn(Ac) 2 /PbI 2 and Zn(Ac) 2 /CsBr were obtained by dissolving in DMSO then drying in vacuum oven at about 100°C. NMR spectra were measured in DMSO-d6 solvent using a BRUKER DRX500 spectrometer with tetramethylsilane (TMS) as the internal standard.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.