Elucidating Structure Formation in Highly Oriented Triple Cation Perovskite Films

Abstract Metal halide perovskites are an emerging class of crystalline semiconductors of great interest for application in optoelectronics. Their properties are dictated not only by their composition, but also by their crystalline structure and microstructure. While significant efforts are dedicated to the development of strategies for microstructural control, significantly less is known about the processes that govern the formation of their crystalline structure in thin films, in particular in the context of crystalline orientation. This work investigates the formation of highly oriented triple cation perovskite films fabricated by utilizing a range of alcohols as an antisolvent. Examining the film formation by in situ grazing‐incidence wide‐angle X‐ray scattering reveals the presence of a short‐lived highly oriented crystalline intermediate, which is identified as FAI‐PbI2‐xDMSO. The intermediate phase templates the crystallization of the perovskite layer, resulting in highly oriented perovskite layers. The formation of this dimethylsulfoxide (DMSO) containing intermediate is triggered by the selective removal of N,N‐dimethylformamide (DMF) when alcohols are used as an antisolvent, consequently leading to differing degrees of orientation depending on the antisolvent properties. Finally, this work demonstrates that photovoltaic devices fabricated from the highly oriented films, are superior to those with a random polycrystalline structure in terms of both performance and stability.


Introduction
Over the last decade, significant advances have been made in the design of their composition, 3 passivation of defects, 4 interfacial engineering, 5 and control over the layer microstructure. 6 Many different strategies have been employed for controlling the microstructure of the perovskite layer.1] Alternatively, Lee et al. demonstrated that the microstructure of the perovskite active layer evolves when chlorine-containing precursors such as lead chloride (PbCl2) or methylammonium chloride (MACl) are added to the perovskite solution, leading to significantly larger grain sizes. 12reover, the use of additives also proved effective in controlling the microstructure of perovskite layers.Notable examples of such additives are thiourea, 13 ammonium hypophosphite (NH4H2PO2) 14 and hypophosphorous acid. 15Finally, the surface properties of the substrate on top of which the perovskite precursor solution is deposited can also impact on the resultant microstructure.Non-wetting surfaces have been shown to lead to the formation of larger grains, 16 but on the other hand might also result in microstructural defects such as pinholes and nanovoids. 17 addition to the size of the perovskite grains, recent studies suggest that their relative orientation with respect to the substrate and each other might impact on the photovoltaic performance. 18For example, Yang et al. reported that the addition of caffeine into the perovskite solution results in a preferential orientation of the perovskite grains along the (110) planes.The authors suggest that this preferential orientation improves charge transport in the device, leading to enhanced photovoltaic performance. 19However, the impact of grain orientation could not be disentangled in this case, since the addition of caffeine also led to increase in grain size and defect passivation, which also result in improved device performance.
Interestingly, the work examining the impact of chlorine also reported a preferential orientation along the (110) upon the addition of MACl into the perovskite precursor solution. 12Yet, also in this case, the impact of change of orientation could not be decoupled from the change in microstructure.On the other hand, spectroscopic studies suggest that neither the size, nor the orientation of perovskite grains impacts their optoelectronic properties, 20 leaving the question of the consequences of crystalline orientation for device performance unanswered.
Importantly, the formation of perovskite films occurs via crystalline intermediate phases, often containing high boiling point solvent molecules in the crystal lattice. 213] Understanding the formation mechanisms of such intermediates, and the development of strategies to control them can enable precise structural engineering of the deposited perovskite layers. 246][27] Indeed, such characterizationalthough experimentally complexhas already led to significant insights.For example, Qin et al. identified three clear stages of film formation of mixed perovskites, and demonstrated that annealing has to take place in the second stage, in order to avoid the formation of undesirable phases. 28Huang and co-workers employed in situ X-ray diffraction (XRD) to investigate the crystallization processes in FAPbI3 and demonstrated the presence of multiple solvent-coordinated intermediate phases. 29While these examples illustrate the efficacy of in-situ characterization for the study of perovskite crystallization processes, to the best of our knowledge, these techniques were not yet applied to the study of orientation control.
To examine the impact of orientation on the photovoltaic performance, it is thus important to not only isolate the orientational variation from microstructural changes, but also investigate the temporal evolution of structure formation, thus elucidating the mechanism that triggers orientational preference.In our previous work, we observed that the former can be made possible in case of perovskite layers fabricated via the antisolvent engineering route. 30ecifically, we observed that the use of alcohols as antisolvent leads to highly oriented films, while other antisolvents largely lead to a random grain orientation.A similar observation was later reported by Wang et al, who observed preferred orientation of perovskite layers fabricated using isobutanol (IBA) as an antisolvent. 31The authors suggested that the polarity of the antisolvent molecule led to a different orientation of formamidinium (FA + ) molecules in an IBA-DMSO-FA + complex as compared to the DMSO-FA + complexes formed when using a non-alcoholic antisolvent.Importantly, the authors observed an improved photovoltaic performance for the oriented perovskite layers.While these results are highly promising, many questions regarding the structure formation of oriented perovskite films and the impact on the photovoltaic performance remain open.For example, it remains unclear which characteristics of the alcoholic antisolvents impact the orientation of the perovskite layers and how the relative degrees of orientation impact the performance and stability of perovskite solar cells.
To address these questions, we investigate the temporal evolution of crystallization in triple cation perovskite films deposited by antisolvent engineering method.In short, in this method the perovskite thin film is formed by spin-coating the perovskite solution (in a 4:1 mixture of N,N-dimethylformamide (DMF) and dimethyl sulfoxide (DMSO)) on to the substrate, during which an antisolvent is dripped onto the substrate, triggering crystallization.Once the spincoating procedure ends, the crystallization is completed by thermal annealing (Figure 1a).To probe the structure formation, we employed in-situ grazing-incidence wide-angle X-ray scattering (GIWAXS) performed during the fabrication of the perovskite layers.Such in-situ techniques proved to be highly effective in studying the crystallization processes of perovskite films, revealing both crystallization kinetics and growth mechanisms. 32We examine the structure formation of perovskite films fabricated using three different alcoholic antisolvents comparison to films fabricated using a non-alcoholic solvent.The chemical structures of the examined antisolvents, namely butanol (BuOH), isopropanol (IPA), isobutanol (IBA) and trifluorotoluene (TFT) are shown in Figure 1b.Our measurements reveal the presence of a short-lived, highly oriented intermediate species that templates the growth of the oriented perovskite layer.Finally, we compare the performance and stability of the fabricated perovskite solar cells, revealing that both these factors are correlated with the degree of crystal grain orientation.

Microstructure Characterization
In our previous work we reported that the use of alcohols as antisolvents may lead to microstructural defects due to the extraction of organic halides during the antisolvent application step. 30Specifically, this may occur if the antisolvent is extruded slowly and the interaction time of the antisolvent with the spinning substrate and thinned precursor solution is not short enough.To avoid this, all films in this study were fabricated by extruding the antisolvents rapidly.To ensure no microstructural defects were formed, the films were examined using scanning electron microscopy (SEM).SEM images confirmed that polycrystalline and pinhole-free perovskite films were fabricated using each of the antisolvents (Figure 2).This has been further corroborated via cross-sectional SEM imaging that confirmed that all antisolvents led to the formation of compact perovskite layers without any pin-holes or nanovoids at the buried interfaces (Supplementary Information Figure S1).The grain size is similar in all the films with grains ranging from 50 to 300 nm in diameter.Interestingly, the change in the relative orientation in these films can be observed already via SEM.While grains in perovskite films fabricated using TFT exhibit various edges of crystal facets, grains on the IPA, IBA and BuOH films often display a flat surface facing upward with concentric edges around them, suggesting crystal planes parallel to the film surface.Some of these edges have been highlighted in the high magnification images shown in Figure 2 for clarity.

Structural Characterization
To examine in detail the evolution of the crystalline structure during the formation of the films, in-situ GIWAXS characterisation was performed on a bespoke setup in which a spin-coater was integrated into the synchrotron beamline and the reciprocal space maps were recorded as a function of time for each of the investigated antisolvents.The evolution of these maps as videos can be found as Supplementary Note 1.
In Figure 3 we present the GIWAXS maps at important time points during the deposition procedure for samples with IPA as an antisolvent, which will allow us to track the evolution of different structures formed during the film formation.At the first frame (15s), taken after the perovskite precursor solution was dispensed, but prior to the start of spin-coating, no crystalline features are observed.The spin-coating procedure was started at 37s, and shortly afterwards (50s), the solution is thinned down, making it possible to observe the reflections associated with the ITO substrate, which is marked in a dashed line (q = 2.15 Å -1 ).We note that the feature at 0.49 Å -1 originates from the kapton window of the experimental setup and is present on all images independent from the sample properties.The antisolvent is dispensed at 67s, at which point immediately two crystalline species can be observed.The first, marked in yellow circles, leads to a strong signal at qz = 0.54 Å -1 , which we assign to a highly oriented intermediate phase that templates the oriented growth of the perovskite, since -as will be shown in the following -it is only observed in the case of the alcoholic antisolvents.The intermediate species is very short lived and is observed only for 2 seconds under the applied preparation methods.The second species, marked in orange circles is a hexagonal phase of the triple cation perovskite. 33 is noteworthy that already at this stage the hexagonal phase exhibits a clearly preferred orientation, since distinct diffraction features are observed, rather than full diffraction rings.
Shortly after the spin-coating has finished (100s), we observe a co-existence of the hexagonal and cubic phases (marked in black circles) of the perovskite layer. 34The latter also exhibits a highly oriented structure, evidenced by distinct diffraction features.Due to instrumental limitations, annealing could only commence roughly 2 minutes after the completion of spincoating.Approximately 100s after spin coating stopped (199s), we no longer observe a hexagonal phase of the perovskite, but instead detect the formation of a known (MA)2Pb3I8‧2DMSO intermediate (pink circles). 35This intermediate remains for the first 20s of annealing, but is eliminated after 90s of annealing, at which point a small contribution associated with phase separated PbI2 can be observed (red circle) alongside highly oriented features of cubic perovskite.To compare the structure evolution for the different antisolvents, we focus our attention on the templating species and hexagonal and cubic phases of the perovskite.Figure 4 displays the intensity evolution of these three species as a function of time and the final GIWAXS maps obtained post annealing for each of the films.In case of the TFT antisolvent, no templating species is observed and the formation of the hexagonal perovskite phasewhich, in contrast to the alcohols, shows significantly less orientation -occurs once the antisolvent is dispensed.
After an initial increase, this phase is decreased with an increasing intensity of the cubic phase.
Once annealed, only cubic phase features remain, with a largely random orientation, evidenced by the Debye ring shape of the GIWAXS pattern.On the other hand, in the case of all three of the alcoholic antisolvents, a short-lived templating species is observed immediately once the antisolvent is dispensed, which in all cases appeared at qz = 0.54 Å -1 .This finding suggests that the structure of this species is independent of the specific alcohol used, which implies that the antisolvent is not incorporated into that crystalline structure.To compare the degrees of orientation between the different samples, we performed angular integration along the (100) reflection as well as X-ray diffraction (XRD) measurements that enable us to compare the intensity of the (111) reflections (q = 1.72 Å -1 , 2Θ = 24.46°).The angular profiles confirm that no preferential orientation in the case of the TFT samples, but a clearly preferred orientation for all the alcoholic antisolvents with increased intensity at approximately 15°, 55° and 77° (Figure 5a).The distribution of intensities shows a dependency on the choice of antisolvent, with IPA leading to particularly oriented films with the strongest intensity at 55° in comparison to that at 15° and 77°.Similar observations can be made by examining the XRD patterns (Figure 5b).Very clearly, the (111) intensity is strongest in the IPA fabricated samples, although it is very prominent also in the other samples fabricated with alcoholic antisolvents.To investigate whether there are differences in the vertical distribution of the crystalline species in the fully fabricated, annealed perovskite layers, we performed angular dependent GIWAXS measurements.At this stage, the films consist almost exclusively of the cubic perovskite phase and PbI2, the vertical evolution of which is shown in Supplementary Information Figure S5.
The results reveal that the vertical distribution of the PbI2 is impacted by the choice of antisolvent: while alcoholic antisolvents lead to increased amounts of PbI2 in the bulk of the films.In the films made using TFT, its distribution is homogenous throughout the layers.7][38] The potential impact of the differences in the PbI2 distribution on the device performance will be discussed in Section 2.4.

Proposed Mechanism for Structure Formation
As mentioned above, our previous observation that the short-lived templating structure appears at qz = 0.54 Å -1 regardless of the type of alcohol used suggests that the alcoholic antisolvent is not integrated into this crystalline structure, indicating that it consists of the precursors and/or solvents present in the wet perovskite film being spin-coated.6][37][38][39][40][41] To gain further insights into the species that are integrated into the short-lived intermediate phase, we examined the structure formation in MAPbI3 and FAPbI3 films fabricated with IPA as an antisolvent.Interestingly, GIWAXS measurements revealed that only the latter composition exhibited a templating structure (Supplementary Figure S6).This observation suggests that FA molecules are incorporated into the templating structure, since the high degree of orientation depends significantly on the FA content.Drop-casting a highly concentrated solution of FAI and PbI2 in a molar 1:1 ratio in pure anhydrous DMSO resulted after gentle drying at 60°C in a pale yellow, crystalline film that exhibits a series of intense reflections that are in a good agreement with those observed by GIWAXS for the templating structure (see Supplementary Figure S7).The crystalline film is highly ordered, which can also be observed via optical microscopy (Supplementary Figure S8).We note that single crystal structure characterisation and in-plane diffraction experiments failed due to the high sensitivity of these crystals, that converted to brown perovskites rather rapidly under light or X-ray exposure, which is typical for perovskite intermediates that incorporate solvent molecules.Our experiments suggest a composition of FAI-PbI2-x‧DMSO considering the 1:1 FAI to PbI2 ratio we used.Examining the literature reveals that an intermediate with this composition has been proposed by Ren et al, 42  Hansen defined the "Hansen space" as the three-dimensional coordinate space (ΔD, ΔP, ΔH).
The closer two molecules are to each other in this Hansen space, the more likely it is that they are capable of dissolving in each other.Table 1 lists the Hansen solubility parameters for the host solvents (DMF and DMSO) and the four antisolvents used in this study.Based on these parameters, we can calculate the distance between the corresponding coordinates in the Hansen space for each of the antisolvents with respect to the host solvents, defined as RA(DMF) and RA(DMSO).By examining these distances, we observe that the interaction of DMF with the alcoholic antisolvents is far stronger to that of DMSO, evidenced by the smaller values of RA(DMF).This is due to the stronger interaction via hydrogen bonds that can form between the alcohols and the polar DMF host solvent.
On the other hand, due to the absence of a hydroxyl group in TFT, only the dispersive and dipolar interactions determine the interaction among the solvents.These interactions are very similar for both host solvents, resulting in an equally good extraction of both DMF and DMSO by TFT.This is further illustrated by calculating the RA(DMF)/RA(DMSO) ratio, which is significantly smaller for the alcohols than for TFT.This difference in solvent interaction suggests that alcohols preferentially extract DMF from the precursor solution as it is more soluble in them, resulting in a local enrichment of DMSO on the substrate during the antisolvent treatment.The high DMSO concentration, in turn, enables the formation of the highly oriented FAI-PbI2-xDMSO intermediate that templates the crystallisation of the perovskite and consequently its orientation.Employing TFTor other non-alcoholic solvents -as antisolvent does not lead to a DMSO enriched environment and the DMSO-intermediate does not form, thus leading to a lack of preferred orientation in the final perovskite film.Taken together with the results of our previous studies, 30,44 the proposed mechanism adds an additional consideration into the selection of an antisolvent for perovskite film fabrication, resulting in three different factors that impact film formation: (1) The solubility of the perovskite precursors in the antisolvent: in case the chosen antisolvent can easily dissolve some of the perovskite precursors, its application may lead to an irreparable alternation of the intended film stoichiometry.This can be largely avoided by applying the antisolvent very fast, 30 or modifying its deposition strategy from pipetting to spraying. 38) The miscibility of the antisolvent with the host solvents: certain antisolvents exhibit a very poor miscibility with the host solvents DMF and DMSO.In this case, the extraction The very similar VOC of the devices suggests that the changes in the vertical distribution of the PbI2 observed in the angular dependent GIWAXS measurements are not the cause of the improved photovoltaic performance.At the same time, it is interesting that the highest performance average was achieved for samples fabricated using IPA, which also exhibited the highest degree of preferred orientation.Considering that the improvement is associated with an enhancement in the photocurrent and not in the other solar cell parameters, it is unlikely that it originates from a change in the optoelectronic properties of the layers.Indeed, UV-vis absorption and photoluminescence measurements (Supplementary Figure S12) are similar between all the measured samples.This suggests that the change in orientation mainly impacts the charge transport properties, which appears to be enhanced in the case of the highly oriented films.
To gain initial insights into the degradation behaviour of the different devices, their performance was remeasured 23 days later after being stored unencapsulated in the dark in ambient air.The results are presented in Supplementary Figure S13.We observe that the degradation in performance is more severe for TFT-based devices, in comparison to that of those made using alcoholic antisolvents.This is an initial indication that the latter exhibit a slower degradation process, and considering that all other parameters in the device fabrication were kept identical, we preliminarily associate this suppression of degradation with the higher degree of orientation in the perovskite active layers.Monitoring the performance evolution of the devices under continuous illumination (Supplementary Figure S14a) reveals that TFT based devices exhibit a significantly stronger burn-in (more than 15% of initial performance) than devices made using alcoholic antisolvents (approximately 5%).This observation is in agreement with recent reports that suggest that highly oriented films result in a superior stability under operational conditions. 46 -48When exposed to thermal stress, however, the devices exhibited identical degradation dynamics (Supplementary Figure S14b).These results suggest that orientation plays a significant role in determining the degradation dynamics of perovskite solar cells, nevertheless, a comprehensive study of the impact of orientation on degradation mechanisms of perovskite films is a topic of future investigation and is beyond the scope of the current work.

Conclusion
To summarize, we investigated the film formation processes that govern the growth of highly oriented triple cation perovskite films fabricated by alcoholic antisolvents.By monitoring these processes by in-situ GIWAXS, we uncovered the presence of a highly oriented intermediate species that templates the growth of the perovskite layers.We identify this species to be a FAI-PbI2-x‧DMSO complex that is formed due to a strong interaction of the alcoholic antisolvents with the DMF host solvent of the perovskites solution, which results in its preferential extraction during the antisolvent application step.We find that films with stronger degree of orientation result in higher photovoltaic performance and stability when incorporated in solar cells, highlighting the importance of developing strategies to control the orientation of polycrystalline perovskite thin films.For the electron transport layer (ETL), PCBM was dissolved in anhydrous chlorobenzene (CB) in an amber vial with a concentration of 20 mg mL −1 .To ensure the dissolution, the mixture was stirred in a nitrogen filled glovebox overnight with a magnetic stirring bar at 70 °C.

Materials
Afterwards, the solution was filtered through a 0.45 μm PTFE syringe filter.As a hole-blocking layer (HBL), Bathocuproine (BCP) was deposited by means of a 0.5 mg mL −1 solution in In-situ GIWAXS characterization: The in-situ GIWAXS measurements were performed at beamline P08 at PETRA III (DESY, Hamburg). 49with a photon energy of E = 18 keV and a Perkin Elmer XRD 1621 flat panel detector at a distance of 750 mm.The angle of incidence during in-situ characterization was 0.5° to probe the bulk features of the thin films.To control the application of antisolvents during the experiments, a remote-controlled dispensing system with an attached syringe pump was built into the measurement chamber.Both the spin-coater and the syringe pump are integrated as devices in the beamline control software which allows for electronic synchronization of the spin-coating procedure, antisolvent dispensing and GIWAXS measurement within an error of approximately 1 s.Diffraction intensities were calculated by integrating peak intensities over the entire peak area and applying a baseline correction.This radial integration was performed using the software ImageJ.
Scanning electron microscopy: SEM measurements were performed on perovskite films on glass/ITO/MeO-2PACz in vacuum.The pre-patterned ITO stripe was used to ground the samples with silver paste to avoid sample charging.In a ZEISS GeminiSEM 500 the InLens and HE-SE2 detectors were utilized to yield images with a 5 nm resolution with electrons of 1.5 kV landing energy.

UV-vis absorption:
A Jasco V-770 Spectrophotometer was used to determine the spectral absorption of the perovskite films.The samples were glass substrates coated with perovskite and a pure glass substrate was used as reference.The spectrum was measured form 850 nm to 600 nm with a step size of 1 nm.
Photoluminescence and PLQE: To measure PL and PLQE, the samples were fixed in the beam path of a 532 nm laser operated at 5 mW inside a calibrated Labsphere 6 inch QE sphere integration sphere and measured utilizing an Ocean Optics QE65 Pro spectrometer, following the procedure described by De Mello et al. 50During the measurement, the integration sphere was flushed with nitrogen in order to prevent oxygen or water molecules in ambient air from interacting with the perovskite surface.The samples were glass substrates coated with perovskite.

Figure 1 .
Figure 1.(a) Schematic illustration of the antisolvent engineering method for perovskite layer deposition.(b) Chemical structures and illustration of the corresponding film orientation of the antisolvents used in this study.

Figure 2 .
Figure 2. Scanning electron microscopy images collected via the secondary electron detector of triple cation perovskite films fabricated using different antisolvents.Exemplary grain edges have been highlighted to show the difference in orientation.

Figure 3 :
Figure 3: GIWAXS data taken during different time points during film formation using IPA as an antisolvent.The evolution for the other antisolvents can be seen in the Supplementary Information as Figures S2-S4.

Figure 4 :
Figure 4: Temporal evolution of the intensity of the templating species, the hexagonal and cubic perovskite phases and the final GIWAXS map obtained post annealing for each the investigated antisolvents by integrating the relevant diffraction rings for each of the tracked species.

Figure 5 .
Figure 5. (a) Angular profile along the (100) reflection of the GIWAXS maps shown in Figure 4. (b) XRD measurements on perovskite samples fabricated using different antisolvents.* For clarity of the graph, the TFT curve was normalized to the highest peak of BuOH instead of its own highest peak.
yet its complete crystalline structure has not been reported by the authors.The absence of DMF in the drop-casting experiments proves that solely DMSO is incorporated into the crystal lattice of the observed intermediate upon treatment with alcoholic antisolvents.This suggests that alcoholic antisolvents are preferentially removing DMF from the DMF:DMSO host solvent mixtures used in the film fabrication.This hypothesis is supported by considering the Hansen solubility parameters of the solvents involved in the film fabrication process.Hansen solubility parameters, established by Charles M. Hansen in 1967, 43 are defined as follows: ΔD -The energy from dispersion forces between molecules ΔP -The energy from dipolar intermolecular force between molecules ΔH -The energy from hydrogen bonds between molecules.

:
Pre-cut glass 12×12 mm 2 substrates with a pre-coated central stripe of indium tin oxide (ITO) by Psiotec Ltd. were used as a substrate for device fabrication.Perovskite precursor solution was created with PbI2 and PbBr2 from TCI, CsI from abcr and MAI (CH3NH3I) and FAI (HC(NH2)2I) from GreatcellSolar Materials.PCBM was purchased from Lumtec and MeO-2PACz from TCI. IBA was purchased from Alfa Aesar, EtOH from ACROS Organics and BCP, TFT and all solvents from Sigma Aldrich.The materials, solvents and solutions were stored in a dry nitrogen atmosphere except for PCBM and BCP, which were stored in ambient air.Silver pellets for thermal evaporation of the top contact were purchased from Kurt J. Lesker Company.Solution preparation: MeO-2PACz was used to form a hole-transport layer (HTL).It was dissolved in anhydrous EtOH and the solution was sonicated for 15 min at 30 °C to 40 °C.The 1 mmol L −1 solution for spin coating was diluted from a 10 mmol L −1 stock solution.The perovskite precursor solutions were prepared in a sequential solution method to keep a precise stoichiometry at 1.2 mol L −1 of precursors for Cs0.05(MA0.17FA0.83)0.95Pb(I0.9Br0.1)3 in a 4 : 1 mixture of DMF and DMSO by volume with 1 % excess of PbI2 and 0.25 % ionic liquid ([BMP] + [BF4] − ) as additive.In the first step, the component salts were weighed into adequate vials.Then the inorganic salts, CsI, PbI2 and PbBr2, were dissolved in dimethylsulfoxide (DMSO) in the first case and a 4:1 mixture by volume of anhydrous N,N-dimethylformamide (DMF) to DMSO in the two latter cases at 180°C.After the salts had dissolved completely and the solutions had cooled down, the CsI and PbBr2 solutions were added to the PbI2 solution in a volume ratio of 0.05:0.15:0.85 to obtain a 1.2 mol L −1 inorganic stock solution of Cs0.05PbI1.75Br0.3 with 1% excess of PbI2.In a molar ratio of 0.95:1 the inorganic stock solution was added into vials with correctly weighed amounts of FAI and MAI.Then, the solution from the MAI vial was added into the FAI solution in a volume ratio of 1:5 MAI to FAI, yielding a 1.2 mol L −1 solution of Cs0.05(MA0.17FA0.83)0.95Pb(I0.9Br0.1)3with 1% excess of PbI2.Finally, the appropriate amount of this solution is transferred to a vial with the ionic liquid [BMP] + [BF4] − , to yield a 0.25% concentration of the organic liquid in the resulting solution.

Table 1 :
Hansen parameters of the perovskite solvents and antisolvents used in this study.