Engineering of the Electron Transport Layer/Perovskite Interface in Solar Cells Designed on TiO2 Rutile Nanorods

The engineering of the electron transport layer (ETL)/light absorber interface is explored in perovskite solar cells. Single‐crystalline TiO2 nanorod (NR) arrays are used as ETL and methylammonium lead iodide (MAPI) as light absorber. A dual ETL surface modification is investigated, namely by a TiCl4 treatment combined with a subsequent PC61BM monolayer deposition, and the effects on the device photovoltaic performance were evaluated with respect to single modifications. Under optimized conditions, for the combined treatment synergistic effects are observed that lead to remarkable enhancements in cell efficiency, from 14.2% to 19.5%, and to suppression of hysteresis. The devices show JSC, VOC, and fill factor as high as 23.2 mA cm−2, 1.1 V, and 77%, respectively. These results are ascribed to a more efficient charge transfer across the ETL/perovskite interface, which originates from the passivation of defects and trap states at the ETL surface. To the best of our knowledge, this is the highest cell performance ever reported for TiO2 NR‐based solar cells fabricated with conventional MAPI light absorber. Perspective wise, this ETL surface functionalization approach combined with more recently developed and better performing light absorbers, such as mixed cation/anion hybrid perovskite materials, is expected to provide further performance enhancements.


Introduction
In the past few years, organic-inorganic hybrid halide perovskite materials have come into focus due to their unique optoelectronic properties such as strong visible light absorbance and high carrier mobility. [1][2][3][4][5] The implementation of this class of materials into solar cells has led to a new generation of rapidly developing be the incomplete filling of pores with the perovskite. This can affect the light absorbance, the charge transport and interfacial transfer, and thus the overall cell efficiency.
In this context, rutile TiO 2 NRs provide better properties, i.e., an open structure for an effective infiltration of the perovskite and a single crystalline nature. The latter can, for example in contrast to anodically grown polycrystalline TiO 2 nanotubes, provide enhanced electron transfer properties [10,12] and thus, in principle, better cell performances.
Nevertheless, aside from the ETL morphology, key to the device efficiency is also to optimize the ETL/perovskite interface properties, e.g. by passivation of the TiO 2 surface defects and trap states. [4,19,20] To this end, titanium tetrachloride (TiCl 4 ) treatment, has been reported as a most efficient approach to improve TiO 2 electron transport ability: it forms at the treated surface a nm-thick layer of TiO 2 anatase nanoparticles (NPs, with size in the 5-10 nm range) that typically favors a more efficient charge extraction across the ETL/perovskite interface, thereby limiting photon losses ascribed to charge recombination in the light absorber. [3,21,22] An alternative approach is to introduce an additive between the ETL and perovskite layer. Phenyl-C61-butyric acid methyl ester (PC 61 BM), a thin organic surface modifier, was recently implemented in perovskite cells' technology as an efficient interlayer to passivate surface defects in both the metal oxide ETL and perovskite light absorber. This can substantially reduce the device hysteresis and improve the cell efficiency. [23,24] Herein, we discuss the synergistic effects in cell performance enhancement achieved by surface modification of TiO 2 NR ETLs by TiCl 4 treatment combined with PC 61 BM monolayer deposition. Such an engineering of the TiO 2 ETL/perovskite interface leads to a remarkable increase of the cell efficiency up to 19.5% for our champion device (17.9% average PCE). Our results suggest that on the one hand, the TiCl 4 surface treatment allows for improving the surface contact of the ETL with the perovskite and thus increases the cell performance due to a more efficient charge transfer. On the other hand, the PC 61 BM monolayer, by passivating surface defects at the TiO 2 ETL/ perovskite interface, enhances the efficiency even further and leads to a nearly complete suppression of the device hysteresis.

Results and Discussion
To optimize the PSCs, we engineered the ETL/perovskite interface using hydrothermally grown TiO 2 NRs modified by TiCl 4 and/or PC 61 BM treatments. The TiO 2 NR arrays were decorated with different loadings of TiO 2 anatase NPs; for this the NRs were treated in TiCl 4 solutions of different concentrations, i.e., 50, 100, and 200 × 10 −3 m. After TiCl 4 treatment and annealing, the ETL were used to fabricate the PSCs as outlined in the Experimental Section. Temperature and time of the TiCl 4 treatment were kept constant for all the samples. Figure 1 and Figure S1 in the Supporting Information show cross-sectional and top view scanning electron microscope (SEM) images, respectively, of vertically aligned TiO 2 NRs grown on fluorine-doped tin oxide (FTO) slides, before and after decoration (surface modification) with different NP loadings. Figure 1a shows the bare NRs (as grown). The length of the NRs is in average 550-600 nm. It is clear from the SEM images b-d) that when increasing the concentration of the TiCl 4 solution from 50 to 200 × 10 −3 m, the TiO 2 NP loading consistently increases as well as the average TiO 2 NP size. The TiO 2 NP forms at the NR surface a conformal layer, the thickness Adv. Funct. Mater. 2020, 30,1909738  of which scales with the TiCl 4 solution concentration (this is well in line with previous work [25] ). The NP seem to form also at the free FTO surface (i.e., FTO surface not coated by NRs).
The X-ray diffraction (XRD) patterns of pure and TiCl 4treated NRs are compiled in Figure 2. In general, NRs' features such as diameter, length, and density could be varied by changing growth parameters, such as growth time and temperature, initial reactant concentration, acidity, or additives. Nevertheless, the growth mechanism remains based on the epitaxial relation between the FTO substrate and the growing rutile TiO 2 phase. Precisely, the small lattice mismatch between FTO and rutile TiO 2 phases plays a key role in driving the nucleation and epitaxial growth towards single-crystalline rutile TiO 2 NRs. [26,27] The XRD data in Figure 2 show that, upon NR growth, peaks appear that agree well with the tetragonal rutile phase. However, compared to common powder diffraction patterns (randomly oriented rutile TiO 2 crystals), the intensity of the (101) diffraction peak (at 36.0°) is relatively high, while, e.g., the peak ascribed to the (110) reflection (at 27.36°, typically a main reflection for a polycrystalline rutile specimen) is absent. This indicates that the NRs are highly oriented with respect to the substrate, supporting the epitaxial growth on FTO. Previous work, [26,27] where similar experimental conditions were used to grow TiO 2 NRs, proved by high-resolution transmission electron microscopy (HR-TEM) analysis the single-crystalline nature of such rutile TiO 2 NRs.
No information on the crystallographic features of the decorated NPs could be obtained by XRD, probably owing to the low loading (below XRD detection limits). However, based on the adopted experimental conditions and previous literature, the deposited NPs are supposed to be composed of anatase phase. [25,28,29] In the case of dual modification of the ETL, the PSCs were constructed with the following architecture: FTO/TiO 2 NRs/ TiO 2 NPs/PC 61 BM/MAPbI 3 /PDCBT/MoO x /Ag. An SEM image of the cross-sectional cut of a typical device is provided in Figure 3. We also fabricated reference cells either without PC 61 BM or without TiCl 4 treatment.
The device performance of the PSCs was assessed by measuring their J-V curves under AM 1.5G irradiation (100 mW cm −2 ); data are compiled in Figure 4a. The results for the average photovoltaic parameters of each cell are summarized in Table 1. According to the results, the pristine TiO 2 NR-based device shows the lowest performance with a power conversion efficiency (PCE) of 12.2%, a short-circuit current density (J SC ) of 21.1 mA cm −2 , an open-circuit voltage (V OC ) of 0.85 V, and a fill factor (FF) of 69%. The relatively poor performance of this device has to be ascribed to a large density of trap states at the surface of the unmodified TiO 2 NRs. [30] The TiCl 4 and PC 61 BM treatments are expected to influence the charge extraction across the TiO 2 ETL/perovskite interface. Figure 4a shows the J-V curves for devices (champion cells) fabricated by different ETL treatments, while in Figure 4b-e we have compiled a statistical evaluation of the photovoltaic parameters for devices based on differently treated ETLs. The photovoltaic parameters of these cells are also summarized in Table 1.
We first explored the effect of TiCl 4 and PC 61 BM treatments separately. For the devices fabricated from ETLs treated with only TiCl 4 (100 × 10 −3 m), the cell performance increases in average from 12.2% (reference cell with pristine NRs) to 13.7%. This is caused by a slight increase of the J SC and, more importantly, by a significant improvement of the V OC (from 0.85 V for the reference device to 0.94 V). Here the deposited TiO 2 NP layer plays a key role, i.e., it provides a charge blocking effect by forming a barrier layer at the FTO/perovskite interface, [31] thereby aiding in suppressing detrimental charge recombination. It has been also reported that TiCl 4 treatment can also effectively reduce the density of surface defects in metal oxide ETLs. [11,22,25] To prove the synergistic effects arising from decorating the NRs by the TiCl 4 treatment, we fabricated control devices without NRs and by a direct TiCl 4 treatment of the FTO substrates. With this treatment, the FTO slides result covered in a homogeneous fashion by TiO 2 NPs, the size of which is in average comparable to that of NPs decorated on the NRs (see the SEM of a typical TiCl 4 -treated FTO slide in Figure S2 in the Supporting Information). As illustrated in Figure S3 (and Table S1) 30,1909738   in the Supporting Information, the devices produced from TiO 2 NP-decorated FTO show a clearly worse performance (PCE of the champion cell: 10.6%) compared to cells fabricated from pristine TiO 2 NRs (14.2%). This confirms not only the benefit of using the NR arrays as nanostructured ETL but also the synergistic effects arising from their surface modification with TiO 2 NPs.
Adv. Funct. Mater. 2020, 30,1909738  For devices constructed from ETLs treated with PC 61 BM only, the PCE reached 14.9% and the V OC increased to 1 V. PC 61 BM binds to the hydroxyl groups at the TiO 2 surface through the carboxylic acid anchoring group, while the fullerene moiety faces the perovskite side. As reported by Wojciechowski et al., the V OC and PCE enhancement can be ascribed to an interfacial alteration of the TiO 2 /perovskite junction, i.e., PC 61 BM passivates or inhibits the formation of trap states at the interface (sub-bandgap states originated by under-coordinated surface Ti(IV) ions), on the TiO 2 surface through the carboxylic acid anchoring group, and on the perovskite side by the fullerene moiety. [32][33][34][35] This in turn prevents interfacial electrostatic barriers and provides swift charge (electron) mobility across the ETL/perovskite interface. [33] It is worth pointing out that for both single modifications (either TiCl 4 treatment or PC 61 BM deposition) the cell performance enhancement should be attributed mainly to an evident increase of the V OC , while the J SC improves only to a minor extent and the FF remains virtually unchanged.
We then investigated the dual modification of TiO 2 NR ETLs, i.e., by TiCl 4 treatment combined to a subsequent PC 61 BM deposition. Upon dual modification, we observed a maximized improvement of the device performance. Such an improvement shows a clear dependence on the TiCl 4 concentration and is particularly pronounced when treating the TiO 2 NRs with a 100 × 10 −3 m TiCl 4 solution. Under optimized conditions, the cells delivered superior photovoltaic parameters with a V oc as high as 1.09 V along with improved J sc of 22.8 mA cm −2 and FF of 74%, resulting in an average PCE of 17.9%. The champion device achieved a J SC of 23.2 mA cm −2 , a V OC of 1.1 V, an FF of 77%, and a PCE of 19.5%. To the best of our knowledge, this is the highest cell performance reported for TiO 2 rutile NRbased devices fabricated with methylammonium lead triiodide (MAPbI 3 ) light absorber (see Table S2 in the Supporting Information for a summary of the highest cell performance values reported in the literature). Perspective wise, we expect further enhancements in cell efficiency by replacing MAPI with more recently developed and better performing light absorbers, such as mixed cation/anion hybrid perovskite materials. [36] Aside from the beneficial effect of the PC 61 BM interlayer and the blocking effect of NPs (outlined above), an additional reason for the performance enhancement can be associated with the more negative conduction band (CB) edge of anatase NPs with respect to the CB minimum of the rutile NRs: as demonstrated by Yang et al., [31] such energetic situation favors efficient transfer of photo-generated electrons from MAPbI 3 to the CB of anatase (NPs) and then onward to the CB of rutile NRs. In addition, the TiCl 4 treatment increases the roughness of the NR, providing a nanostructured interfacial contact between the ETL and the perovskite which can thereby sustain a more efficient charge separation and transfer. [37] Nevertheless, a further increase of NPs loading, i.e., a thicker TiO 2 NP layer, was found to cause a decline in cell efficiency ( Figure 4 and Table 1). The charge transport across the NP layer may occur via "random walk", through the NP and across the NP boundaries. The thicker the NP layer, the larger the number of diffusion paths from a statistical point of view, and hence the less efficient the charge transport. This decreases the charge carrier collection efficiency, [21,25] as suggested by the dramatic drop of J SC and FF (this aspect is discussed also below along with the results of impedance spectroscopy, IS).
A summary of the effect of the TiCl 4 and PC 61 BM treatments is outlined in Figure 4b-e. The characterization of 20 devices fabricated for each condition proves the consistency of the statistical distribution and tendency of the performance parameters. Overall, one can conclude that each parameter is maximized by combining the TiCl 4 and PC 61 BM treatments, and a most optimized TiCl 4 concentration is 100 × 10 −3 m.
The V OC , as mentioned above, is significantly affected by the ETL/perovskite interface modification. The average value increases from 0.85 V for the reference device to 1.09 V for the dual modification. This result may be ascribed to the reduction of the electronic disorder at the ETL/perovskite interface. [3] In general, the "electronic disorder" can be interpreted as a measure of the distribution of the electronic density of states (DOS). The electronic states at the ETL/perovskite interface can feature a broad distribution, i.e., can be subject to a large energy disorder originating from structural and chemical inhomogeneity, low crystallinity, random molecular orientation, interactions with neighboring molecules, and presence of impurities, among others. Higher disorder, e.g., in the titania ETL can increase the number of trap sites in the device structure, which eventually leads to electron-hole recombination and short charge carrier lifetimes. Here, the TiCl 4 treatment is proven to passivate the titania surfaces, lowering the DOS and reducing the disorder level, which in turn lowers the interfacial charge transfer resistance, enhances the open circuit voltage and thus maximizes the device performance. [3,38] The increase of current density for devices based on dual modification is supported by a clear enhancement of the device external quantum efficiency (EQE), as shown in  respectively (these data are consistent with the values derived from the J-V curves). Besides, the FF enhancement (from 69% for the reference device to 74% measured by ETL dual modification) can be ascribed to a decrease of the charge transport resistance and to a better filling of the perovskite in the ETL structure. [3] Figure S4a in the Supporting Information shows the current density delivered by the best performing device: a stable current density of ≈21.8 mA cm −2 is tracked under steady-state conditions; this result is in agreement with the J-V curves and EQE data. We evaluated the stability for a device fabricated with the following cell architecture: FTO/TiO 2 NRs/(100 × 10 −3 m) TiO 2 NPs/PC 61 BM/MAPbI 3 /PDCBT/MoO x /Ag. The photovoltaic performance was assessed for the as-fabricated device as well as after storing the device for 7 and 14 d. The results, provided in Figure S4b in the Supporting Information, show that J SC drops to some extent over time, suggesting that the perovskite is undergoing partial degradation as also reported in previous work. [39] On the contrary, the V OC remains almost unaltered. In this context, we anticipate the use of more stable mixed cation/anion perovskite materials [10] to reduce efficiency losses ascribed to perovskite decomposition. Figure 5b shows the effect of the combined TiCl 4 treatment and PC 61 BM modification on cell hysteresis with respect to a reference device (constructed from pristine NRs). Severe loss in cell performance were measured by sequential forward and reverse bias scans for the device fabricated with unmodified NRs, while the hysteresis is considerably suppressed for the device modified with both TiCl 4 and PC 61 BM. The influence of the interface engineering on the open-circuit voltage is evident: for pristine NR, there is a considerable difference between the V OC measured in forward and reverse scans, while after the dual surface modification the difference becomes negligible. Previous work demonstrated that trap states at the TiO 2 ETL surface (i.e., at the ETL/perovskite interface) are the major cause for hysteresis. [33,40,41] Our results suggest that the dual ETL modification leads to an effective passivation of such trap states, thereby improving the charge mobility across the ETL/perovskite interface.
UV-vis diffuse reflectance and steady-state photoluminescence (PL) measurements were conducted to further study the light absorbance feature and charge transfer properties of the perovskite/ETL architectures based on various surface engineering methods. The data are compiled in Figure 6 (the optical features of differently treated ETLs are reported as reference data in Figure S5 in the Supporting Information). The UV-vis spectra (Figure 6a) show a higher absorbance for the perovskite deposited on ETLs modified by TiCl 4 and PC 61 BM treatments, particularly for NRs treated under optimized conditions (100 × 10 −3 m TiCl 4 solution). This can be ascribed to enhanced internal light scattering effects induced by the NR surface roughness, which seems to increase the overall photon harvesting by the perovskite. At the same time, the SEM images in Figure S6 in the Supporting Information and the grain size distribution statistics shown in Figure S7 in the Supporting Information confirm that the optical properties of devices fabricated from differently treated ETLs cannot be ascribed to the perovskite crystallographic features. The grain size statistics of MAPbI 3 layers grown on NRs appears in fact not to be affected by different ETL surface treatments. Moreover, the XRD patterns in Figure 2 reveal that the crystallographic properties of the perovskite grown on unmodified and modified ETLs do not feature remarkable differences.
The PL spectra of perovskite layers formed on differently treated ETLs are compiled in Figure 6b. Steady-state PL can provide a measure of the quality of the perovskite film (e.g., to assess the effects of defect passivation) or of the quenching efficiency (e.g., by charge extraction). The results point to a more efficient charge extraction for the modified ETL/perovskite interface compared to perovskite layers grown on pristine (unmodified) NRs. The highest PL (i.e., worst quenching yield) is measured for the perovskite layer grown on pristine NRs. After TiCl 4 or PC 61 BM modification of the ETL, the PL intensity substantially decreased. Precisely, the most effective Adv. Funct. Mater. 2020, 30,1909738  PL quenching is reached by a TiCl 4 treatment with TiCl 4 concentration of 100 × 10 −3 m combined with PC 61 BM modification; the results are well in line with the data discussed above and further prove that TiO 2 NRs modified by a combined TiCl 4 and PC 61 BM treatment can lead to a most efficient extraction of photo-excited electrons from the light absorber. [41] Solid-state IS was performed to determine the charge recombination properties of reference, PC 61 BM-and/or TiCl 4 -treated devices. [42] The results are reported in Figure 6. The IS measurements were carried out under light illumination, in a highfrequency range (1 MHz to 1 Hz), and at an applied bias of 0.8 V. The semicircle in the Nyquist plots (see in Figure 6c a magnified view of the spectra of different devices) is assigned to the parallel combination of the recombination resistance and the geometric capacitance of the devices. For data analysis, a constant phase element (Q) instead of an ideal super-capacitor element was used, in order to improve the quality of the fitting. The equivalent electric parameters were extracted accordingly and are compiled in Table S3 in the Supporting Information. From a diagnostic point of view, in previous work, it has been reported that a high recombination resistance (i.e., low recombination of generated charges) results typically in highly efficient PSCs. [10] In line with the photovoltaic parameters, we observed high recombination resistance for cells fabricated from ETLs modified by dual TiCl 4 and PC 61 BM treatment. This once again confirms the positive effects of the dual surface treatment in suppressing detrimental charge recombination phenomena. According to the impedance results, the TiO 2 NPs (formed by TiCl 4 treatment) seem to have a considerable effect in decreasing the charge recombination rate. This is more likely caused by the TiO 2 NP blocking layer formed at the free FTO surface, which prevents leakage current that could originate from a direct perovskite/FTO contact. Here, we should mention that the growth of TiO 2 NRs can be ascribed to heterogeneous nucleation (of TiO 2 seeds on the FTO substrate) driven by lower degrees of super-saturation. [27] The NR coverage (or NR density Adv. Funct. Mater. 2020, 30,1909738  at the FTO surface) can be determined by the competition between seed nucleation and NR growth rates, which in turn is affected by the experimental conditions (e.g., Ti precursor and HCl concentrations). The experimental conditions here adopted lead to rutile TiO 2 NRs with an open structure, i.e., the NR density leaves empty space for the perovskite filling. While this on the one side is beneficial for an effective infiltration and crystallization of the perovskite within the ETL, the free FTO surface left behind negatively affects the cell functionality (perovskite/ FTO contact) and has therefore to be coated (blocking layer) via an optimized TiCl 4 treatment. Besides, by increasing the concentration of the TiCl 4 solution, the modified ETLs show a gradually higher charge recombination resistance. This indicates that the TiCl 4 modification is beneficial as it forms at the NR surface a TiO 2 NP buffer interface that enhances the charge transfer.

Conclusion
We systematically explored the effects of multiple ETL surface treatments on the efficiency of MAPbI 3 PSCs. Our results showed that also in the case of mesoporous nanostructured ETLs the open-circuit voltage, and therefore the overall cell efficiency, can be largely improved by an adequate engineering of the interface between the ETL and the perovskite absorber. We observed that a combined TiCl 4 and PC 61 BM treatment is a pre-requisite to fabricate efficient PSCs from ETLs consisting of arrays of single crystalline TiO 2 NRs. The dual treatment led to a champion efficiency of 19.5% and enabled a nearly complete suppression of the device hysteresis. The performance enhancement originates from a more efficient charge transfer across the ETL/perovskite interface, which is caused by the passivation of defects and trap states at the TiO 2 ETL surface. This ETL surface engineering approach can potentially provide a further cell performance enhancement if combined to better performing, mixed cation/anion hybrid perovskite materials.

Experimental Section
Preparation of Rutile TiO 2 Nanorod ETL: Laser patterned FTO substrates (7 Ω m −2 , Solaronix) were cleaned by ultrasonication in acetone first, then in isopropanol, and were then dried in a stream of nitrogen. TiO 2 NRs were grown directly on the FTO surfaces by a hydrothermal process. [43] Typically, 1 mL of titanium isopropoxide (Sigma-Aldrich) was added to an equal volume (50 mL: 50 mL) of a mixture of DI water and HCl (37%, Sigma-Aldrich) in a 250 mL Teflonlined autoclave. This mixture was stirred for 15 min. Then, the pre-cleaned FTO was immersed in the solution facing down. The hydrothermal synthesis was performed at 180°C for 2.5 h. Finally, the as-prepared NRs were annealed at 450 °C for 1 h in air furnace. For the TiCl 4 treatment, experimental conditions reported elsewhere were adopted. [44] briefly, the TiO 2 NRs were immersed in aqueous TiCl 4 solutions of different concentrations (50, 100, and 200 × 10 −3 m) at 70 °C for 30 min. The samples were then heat-treated at 450 °C for 10 min to crystallize the deposited NPs into anatase phase.
Device Fabrication: The as-prepared FTO supported NR layers (ETL) were moved to a nitrogen-filled glovebox and processed as substrates to fabricate the PSCs. PC 61 BM (99.5%, Solenne) solution (10 mg mL −1 in chlorobenzene, Sigma-Aldrich) was spin-coated on the substrate followed by heating at 110 °C for 5 min. For the preparation of the perovskite solution, 710 mg of PbI 2 (Lumtec) and 240 mg of CH 3 NH 3 I (Lumtec) were dissolved in a mixture of DMF (dimethylformamide; Sigma-Aldrich) and DMSO (dimethyl sulfoxide; Sigma-Aldrich) at 40 °C. The perovskite precursor was deposited on the PC 61 BM layer by spin-coating at a speed of 4000 rpm for 15 s followed by blowing off the leftover solvent with a N 2 stream. The samples were then annealed on a hot plate at 110 °C for 10 min. The hole transport material solution, consisting of a 10 mg mL −1 solution of poly[5,5′-bis(2-butyloctyl)-(2,2′-bithiophene)-4,4′-dicarboxylate-alt-5,5′-2,2′-bithiophene] (PDCBT, 1-Material) in chlorobenzene, was spin-coated on the perovskite layer at 2000 rpm for 30 s and annealed at 70 °C for 3 min. Finally, a 15-nm-thick MoO x (Alfa Aesar) and 120-nm-thick Ag films were evaporated on PDCBT by using a thermal evaporation system. A mask was used to pattern the deposited MoO x and Ag films.
Characterization: Current density-voltage (J-V) and steady-state current density (J MPP ) characterization of the devices was carried out under simulated solar light illumination (AM1.5G) with a light intensity of 100 mW cm −2 (calibrated with a Si-cell). The EQE spectra were recorded using an Enli Technology (Taiwan) EQE measurement system (QE-R). Prior to measurement, the light intensity at each wavelength was calibrated by using a standard single-crystal Si photovoltaic cell.
The morphology and crystalline structure of the nanostructured TiO 2 ETLs and perovskite layers were characterized by a field-emission SEM (Fe-SEM, S4800, Hitachi) and by XRD (X'pert Philips MPD diffractometer, using a Panalytical X'celerator detector and graphite monochromized Cu Kα radiation, λ = 1.54056 A Å) respectively. The perovskite grain size distribution was determined by using the software Image J.
UV-vis light absorption spectra were recorded using an Avantes spectroscopy system (AvaLight-DH-S-BAL Balanced Power and AvaSpec-ULS2048L StarLine Versatile Fiber-optic Spectrometer) equipped with an integrating sphere (a BaSO 4 standard white board was used as reference).
Solid-state IS measurements were carried out under 0.1 sun illumination, at 0.8V and in a frequency range from 1 MHz to 1 Hz, by using of an Agilent HP 4192A impedance analyzer. Steady-state PL was measured using an Argon-ion 488-nm laser excitation source, a Horiba monochromator, and a Silicone detector.

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