Composition Engineering of All‐Inorganic Perovskite Film for Efficient and Operationally Stable Solar Cells

Cesium‐based inorganic perovskites have recently attracted great research focus due to their excellent optoelectronic properties and thermal stability. However, the operational instability of all‐inorganic perovskites is still a main hindrance for the commercialization. Herein, a facile approach is reported to simultaneously enhance both the efficiency and long‐term stability for all‐inorganic CsPbI2.5Br0.5 perovskite solar cells via inducing excess lead iodide (PbI2) into the precursors. Comprehensive film and device characterizations are conducted to study the influences of excess PbI2 on the crystal quality, passivation effect, charge dynamics, and photovoltaic performance. It is found that excess PbI2 improves the crystallization process, producing high‐quality CsPbI2.5Br0.5 films with enlarged grain sizes, enhanced crystal orientation, and unchanged phase composition. The residual PbI2 at the grain boundaries also provides a passivation effect, which improves the optoelectronic properties and charge collection property in optimized devices, leading to a power conversion efficiency up to 17.1% with a high open‐circuit voltage of 1.25 V. More importantly, a remarkable long‐term operational stability is also achieved for the optimized CsPbI2.5Br0.5 solar cells, with less than 24% degradation drop at the maximum power point under continuous illumination for 420 h.


Introduction
Since the first report in 2009, [1] organic-inorganic hybrid perovskite solar cells (PSCs) have attracted tremendous attention due to their superior optoelectronic properties. [2][3][4][5][6] The 10 years' developments have witnessed a huge advancement of PSCs, with certified power conversion efficiency phase) under ambient conditions at room temperature, hindering the long-term application. [13,14] Many efforts have been devoted to stabilizing the α-phase of CsPbI 3 for photovoltaic applications, such as crystallization engineering, [15] additive engineering, [16][17][18][19][20][21][22] interface engineering, [22][23][24][25][26] compositional optimization, [27][28][29][30] dimensional engineering, [31,32] and synthesis of nanocrystals. [33] However, the small Goldschmidt tolerance factor t (≈0.8) of CsPbI 3 is the root cause, leading to the unstable cubic lattice structure at room temperature. Instead, CsPbI x Br 3−x mixed halide perovskites, which are formed by partially replacing iodine (2.2 Å) with the smaller bromine halide ion (1.96 Å), could provide much better structural stability, making them as a new research hotspot for perovskite solar cells. [34,35] Although the CsPbI x Br 3−x perovskites show better thermal and structural stability, their photostability is still a critical issue to be solved. It was reported that under illumination, photoinduced phase segregation of the halide anions could be occurred in the mixed halide all-inorganic perovskites, [36][37][38] and therefore to better evaluate the overall performance of the PSCs, their operational stability should be studied. [39][40][41] Several reports revealed that ion migration is initiated by the halide defects at the grain boundaries (GBs) of the polycrystalline perovskite film and eventually results in Br-rich and I-rich perovskite phases. [42][43][44][45] Therefore, in order to improve the operational stability of mixed-halide inorganic PSCs, it is urgently needed to improve the film quality and reduce the density of defects. One common strategy used for grain boundary passivation is the introduction of small amount of excess PbI 2 in the perovskite films, which significantly improves the PSC performance. [46][47][48] Yang and co-workers first reported that passivation of MAPbI 3 with PbI 2 could improve the efficiency of PSCs. [46] Seok and co-workers also demonstrated that the incorporation of excess PbI 2 could be beneficial to passivate the trap states of MAPbI 3 film. It was claimed that excess PbI 2 could act as a passivation layer between the GBs, affecting defect states, dangling bonds, ion migration, etc. [42] Although excess PbI 2 has been widely considered as an effective trap-state passivation method, its effects on the all-inorganic PSCs was rarely investigated.
Stability test is important for evaluating the overall performance of perovskite solar cells. However, in the literatures, the stability test of PSCs was performed under different conditions, ranging from placing unencapsulated devices in the dark under uncontrolled ambient conditions and periodically measuring their PCE, to studies that were under controlled atmosphere, illumination, temperature, and load. [39] Different experimental conditions influence the ageing behavior considerably, making it difficult to compare these results. Many research groups have investigated the differences in ageing behavior with respect to different conditions and found that ageing at the maximum power point (MPP) represents the closest condition resembling that of a working solar cell. Therefore, using MPP tracking to study the operational stability is the most appropriate and reliable way to report stability of PSCs. [39][40][41] Here, the effect of excess PbI 2 on device performance and operational stability is systematically investigated by incorporating different amount of PbI 2 into the all-inorganic CsPbI 2.5 Br 0.5 perovskite precursor. It is found that the presence of a small amount of excess crystalline PbI 2 in all-inorganic CsPbI 2.5 Br 0.5 perovskite has great benefits in improving the overall performance and device stability. The optimized PSCs was achieved by adding 1.02 mmol PbI 2 in the precursor solution, which produces a champion PSC with a PCE of 17.1% and a high V OC of 1.25 V, which is among the highest efficiencies reported for this type of all-inorganic PSCs. In addition, the devices also show excellent operating stability and maintain 76% of their initial efficiency after continuous illumination at the MPP for over 400 h. To the best of our knowledge, this is the first demonstration of operationally stable all-inorganic mixed-halide PSCs. Our results demonstrate that adding moderate amount of PbI 2 in all-inorganic PSCs can induce higher film crystallinity, lower defects density, lower charge recombination, and rapid charge transport, suggesting that such a simple composition engineering strategy can effectively improve both the efficiency and stability of all-inorganic PSCs.

Results
We began our research by evaluating the photovoltaic performance of the PbI 2 -passivated CsPbI 2.5 Br 0.5 PSCs. The perovskite films were deposited by the one-step spin-coating method from the precursor with a recipe of x mmol PbI 2 (x = 1.00, 1.02, 1.04, 1.06), 0.5 mmol CsBr, and 0.5 mmol CsI in a mixture solvent of dimethyl sulfoxide (DMSO) and dimethylformamide (DMF). For ease of description, the films with different amount of PbI 2 are abbreviated as 1.00-PbI 2 , 1.02-PbI 2 , 1.04-PbI 2 , and 1.06-PbI 2 . All the devices were fabricated with the planar architecture of indium tin oxide (ITO)/SnO 2 /ZnO/perovskite/Spiro-OMeTAD/ MoO 3 /Ag. Figure 1a shows the representative current density-voltage (J-V) characteristics of the champion devices based on 1.00-PbI 2 (control) and 1.02-PbI 2 (optimized) perovskite precursors, and the photovoltaic performances are summarized in Table 1. The control device shows a relatively low efficiency of 15.2%, with open-circuit voltage (V OC ), current density (J SC ), and fill factor (FF) of 1.19 V, 16.00 mA cm −2 , and 79.2%, respectively. By contrast, the optimized 1.02-PbI 2 device achieves a higher PCE of 17.1%, with V OC , J SC , and FF of 1.25 V, 16.50 mA cm −2 , and 82.8%, respectively. The increased PCE of the optimized devices mainly comes from the improvements of V OC and FF. The significantly improved champion PCE with a low V OC -deficits of 0.57 V demonstrates the effectiveness of excess PbI 2 in fabricating high-efficiency all-inorganic PSCs. The J-V characteristics and photovoltaic performances of the PSCs with the different amount of PbI 2 are shown in Figure S1 and Table S1 in the Supporting Information. When the amount of PbI 2 increases to 1.04 and 1.06, the device performance decreases due to the fact that the excess PbI 2 may hinder the charge carrier transport at the GBs. The corresponding external quantum efficiency (EQE) curves are illustrated in Figure 1b. The integrated current density matches well with the J SC values observed from the J-V scan. The higher EQE value at 550-640 nm is attri buted to the increased carrier collection efficiency close to the hole transporting layer (HTL)/ anode interface and the reduced charge carrier recombination, caused by the smoother and more uniform surface of 1.02-PbI 2 perovskite film (main reason, Figure S2, Supporting Information) and the defects passivation of PbI 2 at the GBs. [49] The J SC evolution under 1 sun illumination was further evaluated and depicted in Figure 1c. The 1.02-PbI 2 device presents a better photostability during the test period, whereas the control device shows a fast degradation with time and reduces by over 40% after 90 min illumination. The observation suggests the enhanced trap suppression effect by felicitously introducing the excess PbI 2 in all-inorganic CsPbI 2.5 Br 0.5 perovskite. The improved trap-mediated recombination and limited ion migration benefit the stable power output under the long-term operational condition. The improved stability was further testified by the long-term operational stability test, as shown in Figure 1d. The 1.02-PbI 2 device retains 76% of their initial efficiency after continuous power output at the MPP for 420 h with a constant load of 1.08 V (the voltage at the MPP from initial J-V curve). While the control device shows a poor operational stability, which loses over 90% of their initial efficiency within 300 h with a constant load of 1.02 V. The lifetime to 80% of the initial value (t 80 ) was extracted from the degradation curves. [50] The detailed quantitative analysis demonstrates a t 80 lifetime as high as 260 h of 1.02-PbI 2 device, which significantly outperforms the 33 h for control device. The correlated changes in device performances imply that the adding excess PbI 2 is highly effective for synergistically enhancing the PCE and stability of CsPbI 2.5 Br 0.5 devices. We attribute the key effect provided by the excess PbI 2 to the formation of extra PbI 2 at the GBs, which efficiently passivates the traps and reduces trap-mediated recombination. In addition, extra PbI 2 also physically limits the ion migration at the GBs and is beneficial to achieve stable power output under the long-term operational condition, which will be discussed thoroughly below.

Discussion
To explore the underlying mechanisms of the significant improvements in device efficiency and operational stability of the PbI 2 -passivated devices, we first estimated the optical and structural properties of the excess PbI 2 -incorporated perovskite films. Figure    effect plays negligible role on the final device performance. The plane-view scanning electron microscopy (SEM) images are illustrated for the representative control and optimized 1.02-PbI 2 samples in Figure 2a. The 1.02-PbI 2 film exhibits the significantly enlarged grain sizes with fewer GBs with respect to the control film, indicating the positive effects of excess PbI 2 in regulating the crystallization dynamics. [51,52] When the amount of PbI 2 was further increased, 1.04-PbI 2 and 1.06-PbI 2 samples with larger grains showed obvious bright PbI 2 phase on the perovskite surface. [53,54] All PbI 2 phase forms around the perovskite grain boundary on the surface, similar to the previously reported works on organic-inorganic hybrid perovskite solar cells. [42,46,47,55,56] The X-ray diffraction (XRD) patterns are shown in Figure 2b. Both the films present two typical peaks of α-phase CsPbI 2.5 Br 0.5 phase located at 14.2° and 28.8° for the (100) and (200) crystal planes, respectively. The unchanged peak positions indicate the excess PbI 2 in precursor has not been incorporated into the perovskite crystal lattice during the film formation, leading to a negligible influence of phase composition. Meanwhile, the strong diffraction peak at 12.6° indicates the coexistence of PbI 2 in 1.02-PbI 2 film. The narrower full width at half maximum of diffraction peaks and corresponding enhanced crystallinity of 1.02-PbI 2 film are in good agreement with the SEM results. The crystal texture was further assessed using grazing incidence wide-angle X-ray scattering (GIWAXS) analysis ( Figure S4, Supporting Information, and Figure 2c). Both samples exhibit the similar features of the 3D perovskite structure represented by the broad Debye-Scherrer rings. The intensity versus q z curve for the scattering features is shown in Figure S5 in the Supporting Information, the increased intensity of diffraction peaks in the 1.02-PbI 2 case indicated that there is more oriented crystal presented in the perovskite film. A small PbI 2 diffraction peak at q z = 0.935 Å −1 observed in PbI 2 -excess sample is in accordance with the XRD results. The observations above suggest that the excess PbI 2 doping could effectively influence the crystal nucleation behavior in the precursor solutions and slow down the growth process of perovskite films, which may be due to the promotion of the formation of Lewis base adduct of PbI 2 (DMSO) in the precursors. [57] It is well known that PbI 2 could easily combine with DMSO solvent to form PbI 2 -DMSO complex. [58,59] Thus, excess PbI 2 forms a larger concentration of PbI 2 (DMSO) intermediate phase in the perovskite precursor, retarding the fast reaction between PbI 2 , CsI, and CsBr to form CsPbI 2.5 Br 0.5 phase during the spin-coating process and then resulting in less crystal nucleus. When annealing the precursor film, the more PbI 2 (DMSO) adduct will slow down the crystallization of perovskite and give more time for crystal nucleus growth, inducing the inorganic perovskite grains to grow larger and obtaining the high-quality CsPbI 2.5 Br 0.5 films with better orientation. After annealing, the residual PbI 2 located at the GBs is expected to passivate the defects, thus contributes to the improved optoelectronic properties. The enhanced film quality cooperated with the PbI 2 -passivation effect benefit the PSCs achieving the longterm operational stability. The steady-state photoluminescence (PL) and time-resolved PL (TRPL) spectra were carried out to investigate the charge transport property and recombination behavior of the effect  of excess PbI 2 . As displayed in Figure 3a, the 1.02-PbI 2 film exhibits the stronger PL intensity without significant peak shift, indicating the effective suppression of the nonradiative recombination and negligible influence in phase composition of the excess PbI 2 within perovskite films. Through fitting the TRPL curves by single-exponential function, the 1.02-PbI 2 film exhibits the doubled carrier lifetime of 1.24 ns in contrast to 0.68 ns of control film (Figure 3b and Table S2, Supporting Information). The results demonstrate the excess PbI 2 could effectively inhibit the nonradiative recombination and elongated the carrier lifetime in CsPbI 2.5 Br 0.5 films.
To further quantitatively study the defect-state density of perovskite films with different amount of PbI 2 , we fabricated the electron-only device with structure of ITO/SnO 2 /ZnO/ perovskite/[6,6]-phenyl-C 61 -butyric acid methyl ester (PCBM)/ Ag and evaluated the defect-state density via the trap-filled limit voltage (V TFL ) using Equation (1) where ε 0 is the vacuum permittivity, ε r is the relative dielectric constant (ε r = 35), [60,61] V TFL is the onset voltage of the trap-filled limit region, q is the elemental charge, and L is the thickness of the CsPbI 2.5 Br 0.5 film (350 nm). From Figure 3c,d, the defect-state density is estimated to be 1.42 × 10 16 cm −3 and 9.49 × 10 15 cm −3 for the 1.00-PbI 2 and 1.02-PbI 2 perovskite films, respectively. The results of 1.04-PbI 2 and 1.06-PbI 2 cases are also shown in Figure S6 in the Supporting Information. Among all the cases, a lowest trap density was observed in the 1.02-PbI 2 film. The decrease in defect-state density could be attributed to the enhanced crystallinity and the effective defect passivation of the excess PbI 2 at the GBs. However, the further increased concentration causes the unwished defects around the GBs on the perovskite surface, resulting from the aggregation of the residual PbI 2 observed for samples with 1.04-PbI 2 and 1.06-PbI 2 . The lowest trap density in 1.02-PbI 2 perovskite film would promote effective charge transport and suppress the halide ion migration under illumination, which leads to the improvement in efficiency and operational stability of the PSCs.
Combining our characterization results, we speculated that the residual PbI 2 is mainly existed at the GBs, which plays a crucial role in passivating defects and reducing carrier recombination in the absorber films. The mechanism schematic is summarized in Scheme 1. Due to the solution-based deposition processes are far from thermodynamic equilibrium, a  large number of defects at the GBs coexist in the polycrystalline CsPbI 2.5 Br 0.5 film. These defects acting as undesirable carrier recombination centers would hinder the charge extraction and collection, leading to increase energy loss, and even accelerate the degradation process under external loading condition. By contrast, the slight excess PbI 2 in optimized film could not only improve the nucleation and growth process to obtain the highquality perovskite film but also serve as the passivator between the adjacent perovskite grains, hindering the charge recombination. It is known that the energy offset between the molecules and perovskite also plays an important role in trap state passivation and intergranular carrier transport. [62] The conduction band minimum (CBM) and the valence band maximum (VBM) for CsPbI 2.5 Br 0.5 are −3.48 and −5.30 eV, respectively. And the CBM and VBM for PbI 2 are −3.40 and −5.80 eV, respectively. Therefore, the type I band alignment forms between PbI 2 and CsPbI 2.5 Br 0.5 perovskite ( Figures S7 and S8, Supporting Information). PbI 2 is well known as a p-type semiconductor with a large bandgap. The n-p junction should thus form at the CsPbI 2.5 Br 0.5 -PbI 2 interface when excess p-type PbI 2 is introduced. From the energy band structure, the VBM of p-type PbI 2 has a disparity of 0.5 eV with the VBM of CsPbI 2.5 Br 0.5 . Such a large energy disparity between the perovskite and PbI 2 would form a barrier for charge injection of the holes from CsPbI 2.5 Br 0.5 into PbI 2 and thus should not behave as the bridge for intergrain carrier transport. Therefore, the carrier blocking effect of PbI 2 between perovskite grains is conducive to the extraction and collection of charge carriers generated in the perovskite grains by the charge transporting layers, leading to reduced charge recombination in passivated devices.
Having confirmed the influences of incorporation of excess PbI 2 on the device performance, crystal quality, and optoelectronic properties, we further investigated the device physics to gain better insight into the passivation effects in charge transport kinetics. As illustrated in Figure 4a, the light-dependent V OC results present the corresponding slope of 1.76 K B T/q for the 1.02 PbI 2 -based device, which is smaller than that of control device (2.48K B T/q). In general, the deviation in slope from the value of K B T/q to 2K B T/q indicates the trap-assisted Shockley-Read-Hall dominants in device. [63] The reduced slope indicates that trap-assisted charge recombination process is effectively suppressed through the inclusion of excess PbI 2 . The light-dependent J SC (J SC ∝ I α ) were also conducted and displayed in Figure S9 in the Supporting Information. The exponent α close to 1 indicates the fully suppressed space charge effect in a solar cell owing to the absence of interfacial barrier or carrier imbalance. [64,65] The 1.00 PbI 2 -based device exhibits an α = 1.012 with respect to an α = 1.001 of the 1.02-PbI 2 device. This result suggests that the proper PbI 2 additive in precursor could effectively suppress the charge recombination and improve electron extraction capability from the perovskite layer, benefiting the charge collection inside the CsPbI 2.5 Br 0.5 PSCs.
As mentioned earlier, the excess PbI 2 in CsPbI 2.5 Br 0.5 perovskite films is closely related to the enhanced device performance, especially for the improved V OC and FF. A series of device characterizations were then carried out for the detailed understand of such improvements. The charge-recombination lifetime and charge-extraction time of the CsPbI 2.5 Br 0.5 PSCs with different amount of PbI 2 were compared based on transient photovoltage (TPV) and transient photocurrent (TPC) measurements (Figure 4b,c). We observe a longer chargerecombination lifetime (7.00 µs) and shorter charge-extraction time (0.77 µs) for the 1.02-PbI 2 device, while those of the 1.00-PbI 2 device are 3.81 and 1.85 µs, respectively. The results indicate that the excess PbI 2 could suppress charge recombination and enhance extraction efficiency owing to the enhanced crystallinity and reduced trap states, benefiting the improvements of device parameters.
The transport and recombination resistances within different devices were further extracted from the Nyquist plots of the electrochemical impedance spectra (EIS), as shown in Figure 4d. [66,67] The Nyquist plots typically can be demarked into two main regions: at low and high frequencies. Normally, the low-frequency region represents the recombination process, including recombination chemical capacitance (C rec ) and the recombination resistance (R rec ). The high-frequency region represents the transport process, including the transport chemical capacitance (C trans ) and the transport resistance (R trans ). [68] The detailed parameters are shown in Table S3 in the Supporting Information. By contrast, the extracted R trans values decrease from 53.8 to 33.4 Ω for the control and 1.02-PbI 2 devices, respectively, accompanied with an increase in the R rec from 3200 to 11 300 Ω. The decreased R trans and increased R rec indicate that the enhanced charge transport properties and effectively suppressed recombination rate are achieved in the PbI 2 -passivation case. In addition, frequency-dependent capacitance under dark condition was obtained, as shown in Figure 4e. The capacitance plateau in middle frequency (10 3 -10 5 Hz) is known as a result of electric polarization, caused by dipole polarization or geometrical capacitance of the perovskite. Whereas the capacitance in low frequency (<10 3 Hz) is due to electrode polarization, which is in relation to the defect response or ion accumulation in the perovskite. [11,69] According to the profiles, the control and 1.02-PbI 2 devices present the same capacitance plateaus (≈10 nF) in middle frequency, suggesting the same perovskite geometrical capacitance. However, the higher defect density in the 1.00-PbI 2 film leads to the much larger capacitance in low frequency in contrast to the 1.02-PbI 2 sample, which is believed to ensure the long-term operational stability of devices under illumination.
We experimentally confirmed the synergetic improvements from PbI 2 -passivation in trap density, charge separation, and transfer. Finally, the effect of excess PbI 2 on the built-in potential was then investigated by capacitance-voltage (C-V) measurements. Figure 4f gives the Mott-Schottky (MS) plots 1/C 2 versus voltage obtained from the C-V measurements. The relationship between the capacitance and bias voltage can be summarized by the Mott-Schottky Equation (2) [24] 1 2 2 where C is the capacitance under applied voltage (V), V bi is the built-in potential at equilibrium, q is the elemental charge (1.60 × 10 −19 C), A is the active area, ε r is the relative permittivity (= 35), ε 0 is the vacuum permittivity (8.85 × 10 −12 F m −1 ), and N is the carrier concentration. From Equation (2), the slope gives the carrier concentration of the device (−2/qεN). [24] Based on the results, the slope of the 1.02-PbI 2 device (−15.36) exhibits an over twofold increase with respect to the control device (−7.48), illustrating the much enhanced charge extraction and lower charge density within the perovskite film due to the passivation of PbI 2 . Moreover, V bi extracted from the intercepts of the linear region with the X-axis [70] shows a similar trend with the V OC values from the J-V curves. The relatively higher V bi of the 1.02-PbI 2 PSCs indicates the increased driving force for the separation of photogenerated carriers and the efficient suppression of electron-hole recombination, which is attributed to the improved V OC and FF. Furthermore, we tested the electroluminescence (EL) spectra of the devices, operating as light-emitting diodes (LEDs) in the dark and under the forward voltage bias. Obviously, both devices show the visible EL and a same emission peak located at 680 nm ( Figure S10, Supporting Information). Meanwhile, the 1.02-PbI 2 device displays a stronger EL intensity, further indicating the nonradiative losses in the perovskite layer have been significantly suppressed by PbI 2 passivation.

Conclusion
In summary, we systematically studied the influences of PbI 2passivation in mixed-halide CsPbI 2.5 Br 0.5 PSCs. The morphology and structural characterizations highlight the negligible influence of the excess PbI 2 on the final phase composition, while the induced effect of the PbI 2 additive significantly regulates the nucleation process and produces the high-quality perovskite film with improved grain sizes and crystallinity. The residual PbI 2 at the GBs could serve as a passivator, greatly passivating the trap state, reducing the carrier recombination loss, inhibiting ion segregation. Moreover, the carrier blocking effect of PbI 2 between perovskite grains facilitates the extraction and collection of charge carriers by the charge transporting layers and thus the reduced charge recombination. As a result, the optimized CsPbI 2.5 Br 0.5 PSCs delivered an excellent PCE of 17.1% with a high open-circuit voltage of 1.25 V, in which the V OC -deficits reduce to 0.57 V. More importantly, an excellent operational stability was observed for the mixed-halide CsPbI 2.5 Br 0.5 PSCs with excess PbI 2 decoration, which 1.02-PbI 2 device maintains 76% of its initial efficiency after continuous power output at the MPP for 420 h under continuous illumination. This is the first report that achieves the long-term operational stability of the CsPbI x Br 3−x PSCs under continuous illumination at the MPP. Our findings show that composition engineering is an effective strategy to improve all-inorganic PSCs and provide important insights to further improve the efficiency and operational stability of allinorganic PSCs and perovskite optoelectronic devices.
PSCs Fabrication: Glass/ITO substrates were cleaned by successive sonication in deionized water, detergent, and isopropanol for 30 min each and then dried in an oven. The glass/ITO substrates were treated under oxygen plasma for 4 min before use. The SnO 2 layer was prepared by spin-coating the diluted SnO 2 colloid precursor with deionized (DI) water (1:6.5) onto ITO substrates at 3000 rpm for 30 s. The ZnO nanoparticles (5 mg mL −1 ) were spun coated onto ITO/SnO 2 substrates at 2500 rpm for 30 s, followed by thermal annealing in air at 150 °C for 30 min. The perovskite precursor solution was prepared by dissolving x mmol PbI 2 (x = 1.00, 1.02, 1.04, 1.06), 0.5 mmol CsBr, and 0.5 mmol CsI in a mixture of DMSO and DMF (9:1 v/v) (1 mL) and stirred overnight in a glovebox. The perovskite precursor solution was spun onto ITO/SnO 2 /ZnO substrates at 1500 rpm for 15 s and 5000 rpm for 30 s. Subsequently, the substrates were thermally annealed by a two-step annealing process at 45 °C for 60 s and 300 °C for 60 s. The Spiro-OMeTAD HTL was spun on the perovskite film at 4000 rpm for 30 s, which was prepared by dissolving 72.3 mg Spiro-OMeTAD, 28.8 µL of tBP, and 17.5 µL of a LiTFSI stock solution (520 mg mL −1 in acetonitrile) and 29 µL of an FK209 stock solution (300 mg mL −1 in acetonitrile) in 1 mL chlorobenzene. Finally, a 12 nm MoO 3 and 100 nm silver (Ag) were thermally evaporated under vacuum, respectively. A nonrefractive mask was employed to determine the effective device area of 0.04 cm 2 .
PSCs Characterization: The current density-voltage (J-V) characteristics were measured under a computer controlled Keithley 2400 sourcemeter under 1 sun, AM 1.5G solar simulator (Taiwan, Enlitech SS-F5). The light intensity at each wavelength was calibrated by a standard silicon solar cell (certified by National Renewable Energy Laboratory) before testing, giving a value of 100 mW cm −2 . The EQE spectra were performed on a commercial EQE measurement system (Taiwan, Enlitech, QE-R). The UV-vis absorption spectra were recorded on an HP 8453E spectrophotometer. The bandgap (E g ) was estimated by using UV-vis absorption spectrum and the equation (Ahν/K) 2 = hν − E g . The VBMs of perovskite and PbI 2 films were measured by photoelectron yield spectra (AC-3). The CBMs were calculated by E CBM = E g + E VBM . The SEM images were obtained with a Hitachi FE-SEM S-4800 field emission scanning electron microscope (Japan) based on the sample structure of ITO/SnO 2 /ZnO/perovskite. Based on the data acquired in the range of 10°-60° in θ to 2θ mode, the perovskite crystal structure was investigated using an X-ray diffraction (PANalytical X'pert PRO) equipped with a Cu-K α X-ray tube, using ITO/SnO 2 /ZnO/perovskite as samples. The PL spectra were recorded with an FLS920 spectrofluorometer (Edinburgh) and an HP 8453 spectrophotometer. The defect-state density was extracted by the dark current-voltage characteristics with an electron-only device with architecture of ITO/SnO 2 /ZnO/perovskite/PCBM/Ag. TPV, TPC, EIS, and C-V measurements were carried out with a Paios 4.0 measurement instrument (FLUXiM AG, Switzerland). The EIS spectra were measured at an applied bias of V OC with a frequency range from 10 MHz to 10 Hz and oscillating amplitude of 10 mV under illumination with white LED array (720 W m −2 integrated power). The EL measurement was recorded by a commercialized system (XPQY-EQE-350-1100, Guangzhou Xi Pu Optoelectronics Technology Co., Ltd.).

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