Relationship between the Annealing Temperature and the Presence of PbI2 Platelets at the Surfaces of Slot-Die-Coated Triple-Halide Perovskite Thin Films

We investigated triple-halide perovskite (THP) absorber layers with 5 mol % MAPbCl3 added to the double-halide perovskite (Cs0.22FA0.78)Pb(I0.85Br0.15)3. As a deposition method, a highly scalable printing technique, slot-die coating, with a subsequent annealing step was used. We found a strong power conversion efficiency (PCE) dependence of the corresponding solar cells on the annealing temperature. The device performance deteriorated when increasing the annealing temperature from 125 to 170 °C, mainly via losses in the open-circuit voltage (Voc) and in the fill factor (FF). To understand the mechanisms behind this performance loss, extensive characterizations were performed on both, the THP thin films and the completed solar-cell stacks, as a function of annealing temperature. Correlative scanning electron microscopy analyses, i.e., electron backscatter diffraction, energy-dispersive X-ray spectroscopy, and cathodoluminescence, in addition to X-ray diffraction and photoluminescence, confirmed the presence of PbI2 platelets on the surface of the THP thin films. Moreover, the area fraction of the PbI2 platelets on the film surface increased with increasing annealing temperature. The deteriorated device performance when the annealing temperature is increased from 125 to 170 °C is explained by the increased series resistance and increased interface recombination caused by the PbI2 platelets, leading to decreased Voc and FF values of the solar-cell devices. Thus, the correlative analyses provided insight into microscopic origins of the efficiency losses.


■ INTRODUCTION
Triple-halide perovskites (THPs) 1 are applied as top-cell absorber layers in perovskite-silicon tandem solar cells (PSTSCs).Developed as a wide-band-gap semiconductor, 2 it can reach the optimum band-gap energy for PSTSC top cells of about 1.7 eV 3 without the efficiency-limiting, light-induced effect of halide phase segregation. 4,5ecently, a record power conversion efficiency (PCE) for PSTSCs of 33.7% was achieved. 6However, this record was realized by solar cells fabricated on small device areas via laboratory-scale processes, such as spin-coating.While this technique has shown to be an excellent method to fabricate solar cells on about 1-cm 2 -large substrates, it is rather unsuitable for upscaling, i.e., for the transfer to industrial processes for solar modules.Printing techniques such as slotdie coating can be regarded as promising candidates for the industrial deposition of halide perovskites.
A previous work by Xu et al. 7 investigated the effect of different annealing temperatures on slot-die-coated THP thin films and solar cells.The aim was to find optimal process parameters and to achieve a suitable crystallization of THP thin films without the application of antisolvents.Therefore, N 2 quenching was used to dry the deposited ink, followed by an annealing step at various temperatures.It revealed an increasing amount of PbI 2 with increasing annealing temperatures detected by X-ray diffraction (XRD); however, these PbI 2 agglomerates were not further investigated, and their impact on the device performance is not discussed in detail.
Therefore, in the present contribution, we continue the work of Xu et al. 7 by analyzing a series of THP thin films, annealed at temperatures ranging from 100 to 170 °C.Furthermore, completed solar cells from these stacks were investigated.Using various characterization methods in scanning electron microscopy (SEM) in addition to XRD and photoluminescence (PL), it was possible to identify the PbI 2 precipitates as homogeneously distributed secondary phases and to verify that, indeed, PbI 2 forms platelets at the THP thin-film surface only.This detailed microscopic insight provided the possibility of proposing a specific model of how the PbI 2 precipitates affect the photovoltaic parameters of THP solar cells.

■ EXPERIMENTAL SECTION
Material Synthesis and Device Fabrication.The THP composition studied in the present work was 5 mol % MAPbCl 3 alloyed to a double-halide perovskite of stoichiometry (Cs 0.22 FA 0.78 )-Pb(I 0.85 Br 0.15 ) 3 .The THP deposition process consisted of slot-die coating and subsequently N 2 quenching by a N 2 knife for controlling thin-film drying, which acts as a replacement of the widely applied antisolvent dripping.The N 2 quenching followed an annealing step at temperatures of 100, 125, 150, 160, and 170 °C for 20 min.
The THP thin films in the THP/self-assembled monolayer (SAM)/indium−tin oxide (ITO)/glass and Ag/SnO 2 /C 60 /LiF/ THP/SAM/ITO/glass stacks were prepared via very similar processes.For complete solar-cell stacks, LiF and C 60 were thermally evaporated followed by the deposition of SnO 2 .The Ag back-contact was deposited by thermal evaporation.The overall structures of both sample types are depicted in Figure 1.Further details about the materials synthesis, ink preparation, and thin-film deposition are given in the Supporting Information and by Xu et al. 7 Solar-Cell and Thin-Film Characterization.The secondary electron (SE) images and cathodoluminescence (CL) mappings were acquired with a Zeiss Merlin scanning electron microscope at an acceleration voltage of 5 kV.The beam current was limited to 50−100 pA to reduce the electron-beam damage of the THP thin films, which can be detected well by reduced luminescence signals.The pixel size of the CL images was set to 27 nm, and the CL intensity was measured with a photomultiplier tube (300 μm dwell time and 1 V gain).In order to obtain monochromatic CL images, 700 ± 50 and 500 ± 50 nm band-pass filters were applied.It was not possible to acquire hyperspectral images by the charge-coupled device due to the rapidly progressing beam damage and the necessarily longer integration time in this measurement mode.
The electron backscatter diffractometry (EBSD) and energydispersive X-ray spectroscopy (EDX) measurements were performed by using a Zeiss Ultra Plus scanning electron microscope equipped  with Oxford Instruments Symmetry EBSD and Ultim Extreme EDX detectors.The EBSD measurements were conducted at an acceleration voltage of 15 kV.The samples were tilted to an angle of 70°during the measurements.The step size was 50 nm.EDX spectra and elemental distribution maps were acquired at 3.5 kV on nontilted samples.
The solar cells with an active area of about 0.16 cm 2 were characterized in a sun simulator under 1-sun illumination through the glass substrate.The previous study of Xu et al. 7 already reported about XRD and PL measurements, which are used and extended in the present work for further analyses and discussion (included in the Supporting Information).

■ RESULTS AND DISCUSSION
Analyses of THP Solar Cells.J−V characteristics of 258 solar cells in total were acquired, and the corresponding solarcell parameters, i.e., the photoconversion efficiency (PCE), fill factor (FF), short-circuit current density (J sc ), and open-circuit voltage (V oc ), were determined.The data of the 100, 150, and 170 °C samples were extracted from the work by Xu et al. 7 and added to those of the 125 and 160 °C samples measured for the present study.
In the previous study by Xu et al., 7 the solar cell with the THP film annealed at 150 °C performed best.However, as shown in Figure 2, it was found in the present work that the maximum PCE peak is found for the 125 °C sample with a value of 19.9% (J−V measurement shown in Figure S1).The PCE increases from about 15.4% at 100 °C and decreases with increasing annealing temperature for temperatures >125 °C.This maximum value at 125 °C is close to the record PCE of 20.3% reported for single-junction solar cells with slot-diecoated THP absorbers. 1The FF as well as the V oc values exhibit an increase from 100 to 125 °C followed by a steady decrease for annealing temperatures ranging from 125 to 170 °C.The solar cells annealed at 125 °C exhibited maximum values of FF = 80% and V oc = 1.23 V (not in the same solar cell).On the other hand, the median values of J sc remain within the interval between about 19.5 and 20 mA/cm 2 .Therefore, the PCE decrease visible between 125 and 170 °C is mainly due to decreases in FF and V oc .
The irregular parameters of the 100 °C samples indicate insufficient crystallization at such low temperatures.This  assumption is further strengthened and discussed in the following results and discussions.
Microscopic Analyses of the THP Thin Films.For correlation of the solar-cell parameters and efficiencies with the THP thin-film properties on the submicron scale, various techniques in SEM were applied to THP/SAM/ITO/glass stacks.The SE images of all five samples (Figure 3) show contiguous THP thin films.The sizes of the crystallites increase with increasing annealing temperature, as was already reported for the 100, 150, and 170 °C samples in the previous study by Xu et al. 7 While SE images are suitable for estimations of grain sizes, EBSD maps provide accurate grain-size and grain-orientation distributions.Figure 4a shows EBSD orientation maps in which different colors correspond to different crystal orientations.The black regions in the maps are those on which the EBSD patterns were not able to be indexed by the software; this may be due to shading via surface roughness and the 70°sample tilt during the measurement or due to grain sizes smaller than the EBSD detection limit.
To analyze the effect of surface roughness of the THP layers with PbI 2 secondary phases and its effect on the solar-cell performance, we first acquired cross-sectional SEM images (Figures S2 and S3).These images depict the film surface under a small viewing angle and do not exhibit any strong roughnesses.Moreover, it may be assumed that the film roughness affects the coverage of the C 60 contact layer.However, cross-sectional SEM images and EDX measurements of the completed solar-cell stack (Figure S2) depict a continuous C 60 layer.It should be noted that, in the present work, the C 60 layer thickness was about 23 nm, and other publications reported that full coverage of the C 60 layer can be achieved already at a thickness of 2.3 nm 8 and that even an ultrathin C 60 film of about 1 nm thickness can still be functional in terms of efficient solar-cell operation. 9he THP thin films exhibit a preferred orientation perpendicular to the {101} crystal planes with no significant differences for the five different temperatures.The increase in the grain size with increasing annealing temperature is visible as well (Figure 4).The corresponding measured grain sizes are provided in Table S1.
From the EBSD data sets, also the phase distributions can be extracted (Figure 4b).Apparently, the thin films consist of the THP phase with fractions of the PbI 2 precipitates.With increasing annealing temperature, an increase in the area fraction of PbI 2 crystallites can be detected.The presence of PbI 2 was already discovered by Xu et al. 7 and was unexpected because no excess in PbI 2 was provided during the THP thinfilm synthesis.While this previous study provided first hints of an increasing amount of PbI 2 by XRD and grazing-incidence XRD, the electron microscopy analyses in the present work directly mapped the distributions of PbI 2 on the THP thin films.
For confirmation of the phase distributions obtained by EBSD, EDX measurements were applied (Figure 5).However, it was not possible to identify the PbI 2 precipitates unambiguously in the EDX elemental distribution maps.In these maps, regions were present that exhibited enhanced Pb and I signals, as expected for the PbI 2 precipitates.The differences in the Pb and I signals between the THP matrix and these regions were too small to be easily detectable in the maps.This fact, also given that rather low electron-beam energies of 3.5 keV lead to low information depths of the measurements, indicates that the extensions of the precipitates in the direction perpendicular to the substrate are very small, about 10−20 nm, which gives rise to the conclusion that the  PbI 2 precipitates are present at the surface as platelets.To highlight the compositional differences between the precipitates and the matrix, the N (blue) and I (yellow) signals are superimposed in Figure 5a.From the regions marked by white circles, EDX spectra were extracted from the THP matrix and the Pb-rich and I-rich regions (Figure 5b).From these two spectra, it is apparent that the precipitates are depleted in N and Br and enriched in Pb and I, as expected for PbI 2 .
CL analyses were able to effectively map the PbI 2 phases on top of the THP thin films.The main CL peak energy of 1.69 eV (734 nm), expected for the THP phase (Figure S4), agrees well with PL measurements (Figure S5a), and measured peak energies of PbI 2 at 2.43 eV (511 nm) are in the expected range as well. 1 By the superimposition of monochromatic CL images acquired using band-pass filters of 750 ± 50 and 500 ± 50 nm, detailed false-color maps of the THP and PbI 2 phase distributions were obtained (Figure 6).These CL images reveal the area coverage of the PbI 2 precipitates on top of the THP thin films.The two phases were very distinguishable, which allowed for a quantification of the results employing image segmentation and further analyses of the PbI 2 coverage.
The cross-sectional CL image in Figure 7, in which once more the PbI 2 (yellow) and THP (red) phase distributions are depicted superimposed on the SEM image, shows the PbI 2 covered surface of the thin film without any PbI 2 precipitates detectable within the bulk of the THP thin film.This result confirms that PbI 2 precipitates form as platelets only on the film surface during the annealing process.If PbI 2 precipitates were existent in considerable amounts in the bulk, it would be easily revealed by microscopic investigations in the cross section, as shown by other reports. 10However, no PbI 2 was detected in the THP bulk by EDX measurements on crosssectional specimens prepared from solar-cell stacks (Figure S2).
Influence of PbI 2 Coverage on FF and R s .A common reason for low FF values is the high series resistance R s of the solar cell.The increasing area fraction of PbI 2 can increase R s , which strongly influences the FF. 11The shunt resistance was analyzed as well; however, the determined median values of 0.6−2.4kΩ•cm 2 did not indicate any shunting or other deterioration of the photovoltaic performance (Figure S6b).
Figure 8a shows the PbI 2 thin-film coverage plotted as a function of the median series resistance of the measured solar cells.The data of the coverage were extracted from the CL images by image segmentation, while the R s data originated from the measured J−V curves of the solar cells.The viewgraph in Figure 8a reveals a quasilinear correlation between PbI 2 and the median R s , for all temperatures except for 100 °C.This linear relationship agrees with an assumption of a decrease in the conductor cross section with increasing PbI 2 coverage.The coverage is about 3% at 100 °C and reaches about 22% at 170 °C.R s exhibits a higher value of 7 Ω•cm 2 at 100 °C, decreases to its minimum of 5 Ω•cm 2 at 125 °C, and increases to 11 Ω•cm 2 at 170 °C.To check the R s values for consistency, FF and R s were plotted (Figure 8b) and should follow the approximately linear behavior as described by Green's empirical expression: where FF 0 corresponds to a fill factor without the influence of R s .For large characteristic resistances R V J ch oc sc = compared with R s it can be approximately described as linear, as shown in Figure 8b.The 100 °C sample and its median R s stand out due to its insufficient film crystallization, which makes it a poor semiconductor without the influence of PbI 2 secondary phases.Influence of PbI 2 on V oc .V oc decreases as well with an increase in the annealing temperature (Figure 2b). Figure 9a indicates a linear correlation between the thin-film coverage by PbI 2 and a decrease in the median V oc values.The 100 °C data point is an outlier again.The linear regression reveals a potential PbI 2 -free median V oc of about 1.25 V. Describing the exact correlation between recombination and the amount of   PbI 2 on the surface of the THP thin film is more complex than referring only to the correlation between R s and PbI 2 .This is because PbI 2 may have passivating properties 13,14 in appropriate amounts and at corresponding locations, and complete removal may potentially reduce surface passivation or even generate new recombination centers.Additionally, the bandgap energy decreases with increasing annealing temperature, as indicated by mean measured PL peak energies ranging from 1.67 to 1.65 eV (Figure S5).This result can be attributed to a loss of chlorine, which leads to a further decrease of V oc .Such an issue is independent of PbI 2 and difficult to avoid but plays only a minor role in the temperature-dependent V oc decrease.
Figure 9b shows the mismatch between quasi-Fermi-level splitting (QFLS) and the measured V oc , indicating increasing recombination losses with increasing THP thin-film coverage by PbI 2 platelets.The effective lifetimes obtained from timeresolved photoluminescence (TRPL) measurements (Figure S7) on the THP thin films without an electron-transport layer (ETL; Table S2) are about the same order of magnitude (about 300 ± 100 ns) for the entire temperature range.Thus, we neglected any strong contributions of the THP bulk properties to the performance losses found with the annealing temperature.Rather, elevated recombination rates at the THP/ C 60 interface in connection with the PbI 2 platelets must be considered.
Formation of PbI 2 and Its Effects on the Solar Cell Performance.The formation of PbI 2 in slot-die-coated THP films is not a result of differences in the chemical compositions already in the inks but of compositional changes occurring in the fabricated THP films after the annealing process.The elevated annealing temperatures of 100−170 °C, compared with lower temperatures in antisolvent-quenched spin-coating film depositions, lead to an increased evaporation rate of organic compounds in the THP ink during the annealing step and finally to a collapse of the ABX 3 perovskite structure.
In the past, when investigating the synthesis of the mixedhalide perovskite MAPbI 3−x Cl x was very popular, a material that in most cases did not contain chloride in detectable amounts, it was found that annealing temperatures above 100 °C led to significant losses of MACl. 15,16The same can also be assumed for FACl in THPs containing FA and Cl.A study conducted by Song et al. 17 showed how volatile compounds such as FACl and MACl influence the stoichiometric ratios in gas-quenched (without the application of antisolvents) perovskite absorber layers by different annealing parameters.It is assumed that FACl/MACl losses lead to Pb-rich surfaces, eventually resulting in the formation of PbI 2 on the film surfaces.While Song et al. 17 correlated the thin-film stoichiometry with the device performances, these authors did not link the impaired device performances and stabilities with the amount of PbI 2 .Guo et al. 18 explained the presence of PbI 2 at the film surface by the migration of excess PbI 2 to material interfaces.
An indication for chloride losses in the THP films studied in the present work was found by means of PL.The PL peak energies (Figure S5a) decreased from 1.67 to 1.65 eV for annealing temperatures increasing from 150 to 170 °C.However, this difference is not sufficient to explain the performance losses of the corresponding solar cells.
Several authors reported that an increase of PbI 2 on the surface of perovskite thin films can be beneficial for the solarcell performance, 13,14,19−22 due to passivation of the interfaces 13,14 but also due to important influences on the thin-film formation and crystallization. 22In this study, the detrimental effects of PbI 2 clearly predominate.One reason for this fact is that the solar-cell structure in the present work differs from most of these reports.In studies and publications dealing with regular solar-cell structures, PbI 2 was generally situated between the halide−perovskite absorber layer and hole-transport layer.
For integration of the THP solar cell as the top cell in a tandem device, 23 the "inverted" perovskite solar-cell structure was used in the present work.In this inverted structure, PbI 2 is located at the interface between THP and the ETL, as shown in Figure 10a.The detrimental effect of PbI 2 in this structure is a consequence of its elevated conduction band edge compared with those of perovskite compounds usually applied in solar cells.−27 Nevertheless, even the alignment of the perovskite−PbI 2 heterojunction is a matter of debate. 22We assume an electron-blocking behavior by forming a type 1 alignment 26 or type 2 alignment (Figure 10) with hole-injecting properties. 25Both alignment types are detrimental in inverse THP solar-cell structures.
There are indeed reports claiming beneficial effects even at the absorber−ETL interface; however, they are about the passivation of TiO 2 , which was not applied in the present work. 13Instead, we integrated LiF-passivated C 60 as a selective ETL.The interface between this ETL and PbI 2 is not yet well investigated.−31 We have not analyzed the long-term stability, but we assume a stability-harming effect of PbI 2 in our samples.
It should be noted that the detrimental effect of PbI 2 is not an issue exclusively in THP solar cells.It can be generalized to all perovskite-type solar cells with inverted cell structure, containing an absorber with a conduction band edge of several 100 meV lower than that of PbI 2 24 and with a suitable amount of PbI 2 present at the absorber−ETL interface.
Correlation of the Microstructure and Solar-Cell Performances.The experimental results revealed the 100 °C sample as an outlier in the measurements of the solar-cell parameters.Treating it as an outlier is based on the results and conclusions of XRD (Figure S8) and EBSD measurements (Figure 4).In the XRD patterns, the (100) Bragg diffraction peak of the THP at around 14.2°of the 100 °C sample exhibits the largest full width at half-maximum compared with those of the other samples.Additionally, the large fraction of nonindexed areas in EBSD mappings are an indication for regions with grain sizes below the detection limit of the EBSD system.Such small grain sizes correspond to high grain-boundary densities with the result of lower charge-carrier mobilities and increased recombination losses.Thin-film formation at lower temperatures, such as 100 °C, is usually promoted and enhanced by the application of antisolvents in spin coating, but for large-area depositions and in future roll-to-roll processes, antisolvents cannot be sufficiently supplied. 7,32Annealing temperatures of 100 °C are not sufficient to form grains of suitable sizes, and even leftovers of intermediate phases could be speculated; therefore, the 100 °C samples are excluded from following analyses and discussions.The inability to use antisolvents makes the annealing temperature one of the most important parameters for perovskite thin-film crystallization in slot-die coating.We consider an annealing temperature of 125 °C as the lower limit, where a desired crystallization can be expected without the application of antisolvents.
The goal of the correlation between the THP thin-film microstructure measured by techniques of electron microscopy and the solar-cell performance parameter is to understand the mechanisms behind the evolution of solar-cell parameters with increasing annealing temperatures.The electron microscopy analyses revealed two main trends: temperature-dependent grain sizes and temperature-dependent area fractions of PbI 2 platelets on the surfaces of the THP thin films.Furthermore, the solar-cell parameters exhibit the following trends with increasing annealing temperature: no significant change in the J sc values and decreases in V oc , FF, and (as a result of these trends) PCE.
The measured increase in the grain size is considered to be a beneficial influence on the solar-cell performance because grain boundaries are commonly known as locations of increased nonradiative recombination. 33Grain boundaries can scatter charge carriers and, finally, decrease their mobility.Larger grain sizes are equivalent to a lower grain-boundary density in the bulk material, and as a result, the bulk recombination decreases.Lower recombination rates and high charge-carrier mobility lead to a higher amount of collectable charge carriers and finally should increase J sc , which we did not find.The same positive effects are expected for the V oc value because it is a parameter connected to recombination processes in solar cells.An increase in recombination losses with increasing annealing temperature can be regarded as the main reason for low V oc values.Therefore, another presumably PbI 2 -related mechanism rules the recombination processes, opposing the beneficial effect of increasing grain sizes.
Analyses of the PL measurements provided values for QFLS in the THP thin films.Such QFLS values represent the upper limit of qV oc .A mismatch between QFLS and qV oc indicates losses due to recombination processes, which seem to increase strongly with increasing annealing temperature (Figure 9b).The reason for the increasing recombination rates might be the opening of new recombination pathways by PbI 2 (Figure 10b).Holes can enter PbI 2 and potentially recombine with blocked electrons in the THP or electrons in the C 60 ETL.
The blocking of electrons by PbI 2 explains the increase in R s with increasing PbI 2 coverage.The linear correlation between the PbI 2 coverage and R s allows a linear approximation of a hypothetical R s of PbI 2 -free solar cells (Figure 8).As a result, a PbI 2 -free R s of 2.5 Ω•cm 2 was determined.The measured values for FF and PCE in Figure 2 can be recalculated with the hypothetical PbI 2 -free R s using Green's equation (eq 1) to model the potential of complete PbI 2 removal.
The simulated effect of PbI 2 removal is shown in Figure 11.It reveals an overall increase of FF (Figure 11a), median values exceeding 80%, and median PCEs reaching almost 20% (Figure 11b), as well as a flattening of the decreasing trend with increasing annealing temperatures.The comparison with measured mean PCE values exhibits that all, but especially the 170 °C sample, would benefit from such a removal.The potential gain of PbI 2 removal regarding the PCE is at least 1%.In a similar way, a PbI 2 -free V oc of more than 1.24 V can be determined and would double the PCE gain to more than 2%.Attempts to remove PbI 2 have already been done by polishing or chemical washout, 34,35 but complete removal has not been achieved yet.
Notably, the presented model requires the entire removal of PbI 2 , which is not easily done.Such a removal may even lead to a lack of PbI 2 and the potential opening of new recombination pathways because the presence of PbI 2 has been reported to enhance passivation. 13,14Another potential effect missing in the present model is a presumable increase in J sc with increasing annealing temperature due to the increase in the grain size, reducing grain-boundary recombination and due to enhanced collection by reducing the PbI 2 interface recombination.

■ CONCLUSION
The present study showed that the PCE of solar cells with slotdie-coated THP absorbers is limited by increased area fractions of PbI 2 precipitates on the THP thin-film surface with increased annealing temperature.By means of correlative electron microscopy, we demonstrated that, while a temperature of 100 °C is not sufficient without the application of antisolvents, the samples annealed at temperatures above 125 °C revealed a correlation between the PbI 2 coverage and photovoltaic parameters.The stronger coverage of the THP thin-film surface by PbI 2 precipitates causes a significant increase in the series resistance R s of the corresponding solar cells, which again is responsible for the decreasing FF values.Moreover, the PbI 2 precipitates give rise to recombination pathways at the interfaces, leading to V oc losses.These detrimental effects of PbI 2 are even more effective in inverted perovskite solar-cell structures due to its accumulation at the THP−ETL interface.We estimated that complete removal of PbI 2 by chemical polishing or washing will lead to an absolute PCE gain of at least 1%.This removal would prevent higher series resistances and recombination pathways induced by the PbI 2 precipitates and allow enhanced device performance at higher annealing temperatures via larger average grain sizes.

Figure 1 .
Figure 1.Schematic of the THP solar-cell structure and images of the THP solar-cell specimen made for determination of the solar-cell parameters and the THP thin-film specimen with an open THP surface for electron microscopy analyses.

Figure 2 .
Figure 2. Photovoltaic parameters of the solar cells analyzed in the present work: (a) PCE, (b) FF, (c) J sc , and (d) V oc of solar cells annealed at temperatures between 100 and 170 °C.For PCE, FF, and V oc , the median values exhibit local maxima at 125 °C annealing temperature and a decrease for higher temperatures.

Figure 3 .
Figure 3. SE images of samples annealed at 100−170 °C showing contiguous THP thin films with increasing crystallite sizes for increasing temperature.

Figure 4 .
Figure 4. EBSD analyses with (a) the orientation−distribution maps (given by false colors; see the legend with orientations for the pseudocubic THP/trigonal PbI 2 phases) and (b) the phase distribution maps of the identified THP and PbI 2 phases.

Figure 5 .
Figure 5. (a) Composite map superimposing the elemental distributions of I (yellow) and N (blue) obtained by EDX (I M and N K X-ray lines) to confirm the presence of PbI 2 phases.Two different and significantly varying regions in the EDX map were selected (white circles), and the corresponding EDX spectrum 1 (gray filled) and spectrum 2 (red line) extracted and depicted in part b.These spectra revealed that, compared with the THP matrix, the assumed PbI 2 precipitates are depleted in N and Br and enriched in Pb and I, as expected for PbI 2 .

Figure 6 .
Figure 6.CL composite maps in plane view consisting of the 750 ± 50 nm band-pass filter signal corresponding to the THP luminescence (red) and the 500 ± 50 nm band-pass filter signal corresponding to PbI 2 (yellow).

Figure 7 .
Figure 7. Cross-sectional CL composite map consisting of the 750 ± 50 nm band-pass filter signal including the THP luminescence (red) and the 500 ± 50 nm band-pass filter signal for PbI 2 (yellow).

Figure 8 .
Figure 8.(a) Estimated PbI 2 coverage as a function of the median series resistance R s with a guide for the eyes (dashed line) showing a theoretical PbI 2 -free R s of 2.5 Ω•cm 2 at the x intercept and (b) median FF as a function of the median R s .Data points are labeled with their corresponding annealing temperatures in degrees Celsius.The 100 °C line is illustrated in light gray to mark it as an outlier.

Figure 9 .
Figure 9. (a) Estimated PbI 2 coverage as a function of the median V oc .The dashed line corresponds to a guide for the eye, pointing to a hypothetical V oc without any PbI 2 coverage.(b) Mismatch between QFLS and median V oc as a function of the PbI 2 coverage for the annealing temperatures 100, 150, 160, and 170 °C (data for 125 °C were not available), indicating increasing recombination losses with increasing amounts of PbI 2 .

Figure 10 .
Figure 10.(a) Schematic of PbI 2 platelets situated at the THP/C 60 interface in the ITO/SAM/THP/C 60 stack, a part of an inverted THP solar cell, and (b) schematic band diagram depicting the detrimental effects of PbI 2 at the THP/C 60 interface on the charge-carriertransport and recombination processes.

Figure 11 .
Figure 11.Simulated data of (a) FF and (b) PCE with the subtracted influence of PbI 2 by recalculating FF and PCE using eq 1 with an approximated median R s for a 0% PbI 2 thin-film coverage of 2.5 Ω• cm 2 .The median gain in the PCE was calculated to be about 1−2%.For comparison, the measured median values are included in red.