Structural and Quantitative Investigation of Perovskite Pore Filling in Mesoporous Metal Oxides

In recent years, hybrid organic–inorganic perovskite light absorbers have attracted much attention in the field of solar cells due to their optoelectronic characteristics that enable high power conversion efficiencies. Perovskite-based solar cells’ efficiency has increased dramatically from 3.8% to more than 20% in just a few years, making them a promising low-cost alternative for photovoltaic applications. The deposition of perovskite into a mesoporous metal oxide is an influential factor affecting solar cell performance. Full coverage and pore filling into the porous metal oxide are important issues in the fabrication of highly-efficient mesoporous perovskite solar cells. In this work, we carry out a structural and quantitative investigation of CH3NH3PbI3 pore filling deposited via sequential two-step deposition into two different mesoporous metal oxides—TiO2 and Al2O3. We avoid using a hole conductor in the perovskite solar cells studied in this work to eliminate undesirable end results. Filling oxide pores with perovskite was characterized by Energy Dispersive X-ray Spectroscopy (EDS) in Transmission Electron Microscopy (TEM) on cross-sectional focused ion beam (FIB) lamellae. Complete pore filling of CH3NH3PbI3 perovskite into the metal oxide pores was observed down to X-depth, showing the presence of Pb and I inside the pores. The observations reported in this work are particularly important for mesoporous Al2O3 perovskite solar cells, as pore filling is essential for the operation of this solar cell structure. This work presents structural and quantitative proof of complete pore filling into mesoporous perovskite-based solar cells, substantiating their high power conversion efficiency.


Introduction
Hybrid organic-inorganic perovskites have recently emerged as a promising light harvester material in solar cells due to their distinctive chemical and physical properties and simple synthesis routes.Hybrid perovskites are described by the general formula ABX 3 (A = organic group, B = Pb or Sn, and X = I, Br, Cl); methylammonium iodide (CH 3 NH 3 I = MAI) is the most investigated organo-metal perovskite in the field of perovskite-based solar cells [1][2][3][4][5].
Miyasaka et al. first used MAPbI 3 in a dye-sensitized solar cell (DSSC) configuration on mesoporous TiO 2 , achieving a power conversion efficiency (PCE) of 3.8% [6].However, these cells were unstable in the liquid electrolyte and suffered from degradation.Replacement of the liquid electrolyte with a solid-hole conductor material (HTM) solved the immediate instability problem of these solar cells, and increased their PCEs to over 10% [7].To date, a PCE of 22.1% was demonstrated for perovskite-based solar cells [8].PCEs of 7.6% and 10.9%, respectively, were first reported for mixed halide perovskite deposited into mesoporous Al 2 O 3 and TiO 2 [9], demonstrating the ability of perovskite to efficiently transfer electrons.Therefore, perovskite-based solar cells are not restricted to mesoporous structures, and planar configuration was also demonstrated with high PCEs [10,11].In addition, it has been shown that perovskite can be used both as a hole conductor and a light harvester, lowering the cost and simplifying the fabrication process.These HTM-free perovskite solar cells have PCEs of 10%-12.8%[12][13][14][15][16][17].
When perovskite was first introduced into solar cells, it was initially deposited via a solution process deposition technique into the mesoporous metal oxide film using a mixture of PbI 2 and CH 3 NH 3 I in a common polar solvent.However, this technique resulted in wide variations in the performance of the photovoltaic (PV) solar cell devices due to the uncontrolled precipitation of the perovskite, which produced large morphological variations [18].As a result, several deposition techniques have been developed for perovskite-based solar cells, including vapor deposition, vapor assisted solution process (VASP), spray deposition, and two-step deposition [19][20][21][22][23].
Graetzel and co-workers first introduced the two-step deposition technique in a sensitized mesoscopic perovskite solar cell [20].In this technique, a solution of PbI 2 in dimethylformamide (DMF) was dropped and spin-coated onto the metal oxide nanoparticles (NPs) and subsequently transformed into perovskite by dipping the electrode into a solution of CH 3 NH 3 I.The two-step deposition technique produced high efficiency solar cells, demonstrated good control over the perovskite morphology, and increased reproducibility of performance [20,21].
Moreover, perovskite pore filling into the scaffold's metal oxide layer has an impact on the PV performance.Snaith et al. [24] estimated the perovskite pore filling fractions into the metal oxide layer.They have found that complete perovskite pore filling improves the electron transport rate, photovoltage, and reduces recombination rate.This finding was also supported by Wu et al. and Nanova et al., who showed the effect of the TiO 2 morphology on the PV performance [25,26].
Due to the importance of the pore filling, in this work we provide direct observation of perovskite penetration into the metal oxide pores.We performed a structural and quantitative study of perovskite pore filling into mesoporous metal oxide.Cells without HTM were investigated; the perovskite was deposited using the two-step deposition technique.The use of HTM-free solar cells eliminates the filtration of the hole conductor into the metal oxide pores and enables thorough investigation, avoiding undesirable end results.Energy dispersive X-ray spectroscopy in transmission electron microscopy (EDS-TEM), TEM, scanning electron microscopy (SEM) and X-ray diffraction (XRD) were used to investigate the pore filling.This study provides better understanding of the formation and the percolation of perovskite within the porous metal oxide.

Material Synthesis
TiO 2 paste Dyesol, DSL-90-T was diluted with ethanol in 1:4 ratio by volume, then spin coated at 2000 rpm for 10 sec on top of the compact TiO 2 layer and annealed at 500 • C for 30 min.In the case of Al 2 O 3 electrode, 50 nm particle size Al 2 O 3 purchased from Sigma-Aldrich was diluted with isopropanol in 1:14 ratio by volume.The deposition and annealing conditions were the same as for TiO 2 .
CH 3 NH 3 I was synthesized as described previously [27], by reacting 30 mL of methylamine (40% in methanol, TCI) and 32.3 mL of hydroiodic acid (57 wt% in water, Aldrich) in a 250 mL round bottom flask at 0 • C for 2 h with stirring.The precipitate was recovered by putting the solution on a rotavap and carefully removing the solvents at 50 • C. The yellowish raw product of methylammonium iodide (CH 3 NH 3 I) was washed with ethanol by stirring the mixture for 30 min.Then, the mixture was filtered and washed three times with diethylether.After filtration, the solid was collected and dried at 70 • C in a vacuum oven for 24 h.

Device Fabrication
SnO 2 :F (FTO) conducting glass (15 Ω•cm −1 Pilkington) was etched with zinc powder and HCl (37%) to obtain the desired electrode pattern.The FTO glass was then washed with ethanol, acetone, deionized water, and finally was sonicated for 15 min.A compact layer of TiO 2 (blocking layer) was then spin coated on to the FTO substrate using a solution of titanium diisopropoxidebis(acetylacetonate) (TiDIP, 75% in isopropanol, Aldrich) in ethanol and then annealed at 500 • C for 35 min.The TiO 2 solution or the Al 2 O 3 solution were spin coated subsequent to TiCl 4 treatment for 30 min at 70 • C and annealing at 450 • C for 30 min.
The synthesis of the CH 3 NH 3 PbI 3 on the TiO 2 or Al 2 O 3 surface was carried out by a two-step deposition technique.First, PbI 2 was dissolved in DMF (0.1M) and dropped onto the TiO 2 or Al 2 O 3 film and spin coated, followed by annealing at 70 • C for 30 min.In the second step, the cell was dipped into methylammonium solution (0.66M).Following the dipping step, the samples were annealed at 70 • C for another 30 min.Finally, the back contact was deposited by evaporating 70 nm of gold in vacuum of 5 × 10 −6 Torr.
Structure and composition of HTM-free perovskite solar cells were analyzed with scanning electron microscopy (SEM) and scanning transmission electron microscopy (STEM), both equipped with energy dispersive X-ray spectroscopy (EDS).EDS analysis was performed on top-view samples in the FEI HR SEM Sirion (FEI, Hillsboro, OR, USA) equipped with an Oxford INCA450 EDS solid state detector (Oxford Instruments NanoAnalysis, High Wycombe, UK).TEM analysis and STEM analysis were performed on cross-sectional TEM lamellae in FEI (S)TEMTecnai F20 G 2 (FEI, Hillsboro, OR, USA) equipped with EDAX EDS detector (EDAX AMETEK BV, Tilburg, The Netherlands).TEM lamellae were prepared with a FEI Helios Dual Beam focused ion beam (FIB, FEI, Hillsboro, OR, USA).

X-Ray Diffraction
X-ray diffraction measurements were performed on the D8 Advance diffractometer (Bruker AXS, Karlsruhe, Germany) with a secondary graphite monochromator, two Soller slits, and a 0.2 mm receiving slit.XRD patterns within the range 3 to 75 2q were recorded at room temperature using CuKa radiation (λ = 1.5418Å) with the following measurement conditions: tube voltage of 40 kV, tube current of 40 mA, step-scan mode with a step size of 0.02 • 2θ and counting time of 1 s/step.TOPAS-v.3 software was used for structure refinement and crystallite size calculation (TOPAS V3, General Profile and Structure Analysis Software for Powder Diffraction Data, User's Manual, Bruker AXS, Karlsruhe, Germany, 2003).

Results and Discussion
The structures of the discussed perovskite-based solar cells are presented in Figures 1A,B and 2B, composed of FTO-coated glass/compact TiO 2 layer/mesoporous metal oxide/CH 3 NH 3 PbI 3 (MAPbI 3 )/Au.The penetration of perovskite into the mesopores of two metal oxides-TiO 2 , and Al 2 O 3 -was studied with TEM on cross-sectional lamellae prepared by FIB.Cross-section TEM images of the devices are presented in Figure 1A,B.The thicknesses of metal oxides-which demonstrate high PCEs that we studied in this work-are based on our previous reports [15,28].The current voltage curves of this study's cells are shown in Figure 2A, the hole conductor free TiO 2 / MAPbI 3 cell has V oc of 0.85 V, current density (Jsc) of 18 mA/cm 2 , fill factor (FF) of 56% with efficiency of 8.5%.The hole conductor free cell based on Al 2 O 3 /MAPbI 3 shows V oc of 0.88 V, Jsc of 13.9 mA/cm 2 , FF of 17%, with efficiency of 2.1%.The photovoltaic results demonstrate that both cells are working, which strengthens the discussion in this manuscript.It is important to note that these cells are hole conductor free; in particular, in the case of the Al 2 O 3 -based cell, the perovskite is the only charge transporter, as will be discussed below.The s-shape in the current-voltage curve for the Al 2 O 3 -based cell may be due to charge accumulation in this specific device structure.The exact reason is out of the scope of this work.24) Å. Calculated unit cell parameters are in a good agreement with the same reported by Lilliu et al. [29,30].Crystallite size calculated using the Scherrer equation is equal to 356 nm.This is in a good agreement with the character of electron diffraction patterns obtained from the same sample (see Figures 5 and 6).The graphical result of Rietveld refinement of XRD pattern acquired from the sample shown in Figure 3B is presented in Figure 2C.According to Rietveld refinement results, the synthesized CH3NH3PbI3 has a tetragonal perovskite structure with I4/mcm space group and unit cell parameters a = b = 8.8695( 16) Å, c = 12.6532(24) Å. Calculated unit cell parameters are in a good agreement with the same reported by Lilliu et al. [29,30].Crystallite size calculated using the Scherrer equation is equal to 356 nm.This is in a good agreement with the character of electron diffraction patterns obtained from the same sample (see Figures 5 and 6).[29,30].Crystallite size calculated using the Scherrer equation is equal to 356 nm.This is in a good agreement with the character of electron diffraction patterns obtained from the same sample (see Figures 5 and 6).
Elemental composition of the samples was analyzed with EDS in SEM and TEM. Figure 3A It is important to note that in both cases, large perovskite crystals of 100-200 nm can be observed, while in the case of the Al2O3/perovskite, the coverage seems to be slightly better, although the PV performance is lower.Additional elemental analysis of the perovskite sample was done by X-ray photoelectron spectroscopy (XPS), presented in Figure 4.The XPS shows the presence of Pb, I, C, and N with a ratio of 3.6 for I/Pb-further support of the successful deposition of stoichiometric CH3NH3PbI3 perovskite.The quantification results of the XPS analysis are provided in Table S1.Additional elemental analysis of the perovskite sample was done by X-ray photoelectron spectroscopy (XPS), presented in Figure 4.The XPS shows the presence of Pb, I, C, and N with a ratio of 3.6 for I/Pb-further support of the successful deposition of stoichiometric CH 3 NH 3 PbI 3 perovskite.The quantification results of the XPS analysis are provided in Table S1.To investigate filling of pores by perovskite within the mesoporous metal oxides, we used STEM-based energy-dispersive X-ray (STEM-EDS) analysis.Figures 5A and 6A show the location of the EDS line scan performed in the STEM-EDS line scans of TiO2/perovskite and Al2O3/perovskite solar cells, respectively.The line scanning started in the perovskite layer and ended deep in the metal oxide film.
The line profiles of elemental compositions of the two cells are shown in Figures 5B and 6B.(Figures S1 and S2 demonstrate additional STEM-EDS line scans of the two interfaces, which show the reliability of the results) The data demonstrate the presence of Pb and I throughout the perovskite and the metal oxide film, indicating that the CH3NH3PbI3 perovskite penetrates into the metal oxide films (e.g., TiO2 and Al2O3).As seen, the intensities of PbL and IL signals are higher at the perovskite layer, and decrease at the perovskite/metal oxide interface and at the metal oxide film.Simultaneously, a sharp increase in the TiK and AlK signals indicates the positions of perovskite/metal oxide interfaces in both samples.Within the metal oxide films, both TiK and AlK signals are changing following the mesoporous structure of oxides: the maxima appear at the locations of oxide grains, while the minima appear at the pores.It is worth noting that within the oxide layer, PbL and IL signals (i.e., the perovskite markers) also follow the mesoporous structure, while their signal changes are anti-phase to those of oxide markers: maxima of PbL and IL appear at pores' locations (marked with arrows for clarity in Figures 5B and 6B).
This observation is direct proof of complete filling of the oxide pores with perovskite containing both Pb and I.In particular, the complete pore filling in the case of mesoporous Al2O3 documents the reason why this solar cell functions successfully with scaffold layer.It was previously reported [10] that the Al2O3 functions as a scaffold, so that charges are transported through the perovskite, which requires a complete pore filling of the Al2O3 film.
Although both Pb and I were found in the perovskite and in the metal oxide layers of TEM samples, their Pb:I ratio measured within the perovskite was about 1:1.This value differs significantly from the stoichiometric value of 1:3 observed in SEM samples.We relate this phenomenon to the preferential loss of iodine during the FIB lamella preparation (as mentioned, the samples show 1:3 ratio in the EDS-SEM analysis and high crystalline structure in the XRD spectrum, where the lamella preparation was not required).Moreover, the ratio of 1:1 for Pb:I also excludes the possibility of PbI2 residue in the pores.To investigate filling of pores by perovskite within the mesoporous metal oxides, we used STEM-based energy-dispersive X-ray (STEM-EDS) analysis.Figures 5A and 6A show the location of the EDS line scan performed in the STEM-EDS line scans of TiO 2 /perovskite and Al 2 O 3 /perovskite solar cells, respectively.The line scanning started in the perovskite layer and ended deep in the metal oxide film.168 The line profiles of elemental compositions of the two cells are shown in Figures 5B and 6B.The structure of the perovskite layer was characterized by selected area electron diffraction in TEM.The patterns obtained in TiO2 and Al2O3 samples are presented in Figures 5C and 6C, respectively.
The d-spacing measured on these patterns are shown in Figures 5D and 6D, along with reported structural data, which shows the crystalline tetragonal perovskite I4/mcm with unit cell parameters a = b = 8.87 Å and c = 12.67 Å [31][32][33].Both patterns are of ring-type, which indicates the presence of a large number of randomly-oriented nanoscale crystallites in each of the perovskite layers.The strongest diffraction is observed around inter-planar spacing of 3.17-3.2Å in both patterns, which reasonably matches the reported data for the tetragonal I4/mcm (a = b = 8.87 Å, c = 12.67 Å) perovskite [25,28,31].However, the sets of diffraction reflexes appearing in both patterns do not coincide.Based on these two observations, we suggest that while the major part of nanocrystalline perovskite ensembles has the crystalline structure of the tetragonal perovskite, each of the analyzed perovskite layers contains "occasional" nanocrystals with different structures.We relate the observed structural features of perovskite to the ion beam damage at the sample preparation in FIB.The changes in the elemental composition observed with EDS should necessarily be accompanied by structural changes.To conclude, perovskite layers of TiO2 and Al2O3 cells are structurally similar to the tetragonal perovskite, although their electron diffraction data document different stages of structural decomposition.

Conclusions
In this work, we performed a quantitative and structural study of hybrid perovskite pore filling into mesoporous metal oxides, TiO2 and Al2O3.STEM-EDS characterization on cross-sectional FIB lamellae the maxima appear at the locations of oxide grains, while the minima appear at the pores.It is worth noting that within the oxide layer, PbL and IL signals (i.e., the perovskite markers) also follow the mesoporous structure, while their signal changes are anti-phase to those of oxide markers: maxima of PbL and IL appear at pores' locations (marked with arrows for clarity in Figures 5B and 6B).
This observation is direct proof of complete filling of the oxide pores with perovskite containing both Pb and I.In particular, the complete pore filling in the case of mesoporous Al 2 O 3 documents the reason why this solar cell functions successfully with scaffold layer.It was previously reported [10] that the Al 2 O 3 functions as a scaffold, so that charges are transported through the perovskite, which requires a complete pore filling of the Al 2 O 3 film.
Although both Pb and I were found in the perovskite and in the metal oxide layers of TEM samples, their Pb:I ratio measured within the perovskite was about 1:1.This value differs significantly from the stoichiometric value of 1:3 observed in SEM samples.We relate this phenomenon to the preferential loss of iodine during the FIB lamella preparation (as mentioned, the samples show 1:3 ratio in the EDS-SEM analysis and high crystalline structure in the XRD spectrum, where the lamella preparation was not required).Moreover, the ratio of 1:1 for Pb:I also excludes the possibility of PbI 2 residue in the pores.
The structure of the perovskite layer was characterized by selected area electron diffraction in TEM.The patterns obtained in TiO 2 and Al 2 O 3 samples are presented in Figures 5C and 6C, respectively.
The d-spacing measured on these patterns are shown in Figures 5D and 6D, along with reported structural data, which shows the crystalline tetragonal perovskite I4/mcm with unit cell parameters a = b = 8.87 Å and c = 12.67 Å [31][32][33].Both patterns are of ring-type, which indicates the presence of a large number of randomly-oriented nanoscale crystallites in each of the perovskite layers.The strongest diffraction is observed around inter-planar spacing of 3.17-3.2Å in both patterns, which reasonably matches the reported data for the tetragonal I4/mcm (a = b = 8.87 Å, c = 12.67 Å) perovskite [25,28,31].However, the sets of diffraction reflexes appearing in both patterns do not coincide.Based on these two observations, we suggest that while the major part of nanocrystalline perovskite ensembles has the crystalline structure of the tetragonal perovskite, each of the analyzed perovskite layers contains "occasional" nanocrystals with different structures.We relate the observed structural features of perovskite to the ion beam damage at the sample preparation in FIB.The changes in the elemental composition observed with EDS should necessarily be accompanied by structural changes.To conclude, perovskite layers of TiO 2 and Al 2 O 3 cells are structurally similar to the tetragonal perovskite, although their electron diffraction data document different stages of structural decomposition.

Conclusions
In this work, we performed a quantitative and structural study of hybrid perovskite pore filling into mesoporous metal oxides, TiO 2 and Al 2 O 3 .STEM-EDS characterization on cross-sectional FIB lamellae showed the presence of Pb and I inside the pores.SEM and XRD showed the highly crystalline structure of the tetragonal perovskite with a 1:3 ratio of Pb:I.Moreover, submicron structure of the perovskite was observed by XRD and by selected electron diffraction pattern via TEM.This work demonstrates direct proof of perovskite pore filling into mesoporous metal oxides.These results support previous assumptions about complete pore filling in mesoporous perovskite-based solar cells.The observations reported in this work are in particular important for mesoporous Al 2 O 3 perovskite solar cells, as pore filling is essential for the operation of this solar cell structure.

Figure 2 .Figure 1 .
Figure 2. (A) Current-voltage curves of the study's solar cells; (B) Schematic illustration of the solar cell; (C) Graphical representation of Rietveld refinement results.Peak positions of CH3NH3PbI3, cassiterite, and gold are shown by vertical bars.FTO: SnO2:F.

Figure 2 .Figure 2 .
Figure 2. (A) Current-voltage curves of the study's solar cells; (B) Schematic illustration of the solar cell; (C) Graphical representation of Rietveld refinement results.Peak positions of CH3NH3PbI3, cassiterite, and gold are shown by vertical bars.FTO: SnO2:F.
,C show the EDS spectra acquired at top-view samples of the TiO 2 /perovskite and the Al 2 O 3 /perovskite solar cells in SEM.They reveal the presence of Pb, Ti, Al, O, Ca, Si, and I. Calcium and silicon signals are contributions of the FTO-coated glass.Quantitative analysis of the EDS spectra shows that the atomic ratios of Pb/I are 10:34 for TiO 2 /perovskite and 4:11 for the Al 2 O 3 /perovskite solar cells.The observed ratios are very close to the value of 1:3 that indicates successful deposition of stoichiometric CH 3 NH 3 PbI 3 perovskite in both cases.Figure 3B,D show the SEM images of the areas from which EDS spectra were acquired.It is important to note that in both cases, large perovskite crystals of 100-200 nm can be observed, while in the case of the Al 2 O 3 /perovskite, the coverage seems to be slightly better, although the PV performance is lower.Crystals 2016, 6, x FOR PEER REVIEW 5 of 10 are contributions of the FTO-coated glass.Quantitative analysis of the EDS spectra shows that the atomic ratios of Pb/I are 10:34 for TiO2/perovskite and 4:11 for the Al2O3/perovskite solar cells.The observed ratios are very close to the value of 1:3 that indicates successful deposition of stoichiometric CH3NH3PbI3 perovskite in both cases.Figure 3B,D show the SEM images of the areas from which EDS spectra were acquired.

Figure 3 .
Figure 3. Energy dispersive X-ray spectroscopy (EDS) spectra acquired in scanning electron microscopy (SEM) at the top view samples of the TiO2/ perovskite (spectrum (A) and image (B)) and the Al2O3/perovskite (spectrum (C) and image (D)) solar cells.Scale bars are 5 μm.

Figure 3 .
Figure 3. Energy dispersive X-ray spectroscopy (EDS) spectra acquired in scanning electron microscopy (SEM) at the top view samples of the TiO 2 /perovskite (spectrum (A) and image (B)) and the Al 2 O 3 /perovskite (spectrum (C) and image (D)) solar cells.Scale bars are 5 µm.

Crystals 2016, 6 ,
x FOR PEER 6 of 9 To investigate filling of pores by perovskite within the mesoporous metal oxides, we used STEM 165 based energy-dispersive X-ray (STEM-EDS) analysis.Figures 5A and 6A show the location of EDS 166 line scan performed in the STEM-EDS line scans of TiO2/ perovskite and Al2O3/perovskite solar cells, 167 respectively.The line scanning started in the perovskite layer and ended deep in the metal oxide film.

169(
Figures 1S and 2S demonstrate additional STEM-EDS line scans of the two interfaces, which show 170 the reliability of the results) The data demonstrate presence of Pb and I throughout the perovskite 171 and the metal oxide film, indicating the CH3NH3PbI3 perovskite penetrates into the metal oxide films preparation was not required.)Moreover, the ratio of 1:1 for Pb:I also excludes the possibility for PbI2 192 residue in the pores. 193

Figure 5 .
Figure 5. (A) STEM image of a cross-sectioned TiO2/perovskite solar cell.(B) EDS line profiles of PbL, IL

Figure 5 .
Figure 5. (A) Scanning transmission electron microscopy (STEM) image of a cross-sectioned TiO 2 /perovskite solar cell; (B) Energy dispersive X-ray spectroscopy (EDS) line profiles of PbL, IL, and TiK signals along the line shown on the STEM image; (C) Electron diffraction pattern acquired at the CH 3 NH 3 PbI 3 region; (D) d-spacing between the adjacent lattice planes measured on the pattern C.

Figure 6 .
Figure 6.(A) STEM image of a cross-sectioned Al2O3/perovskite solar cell; (B) EDS line profile of PbL, IL, and AlK signals along the line shown on the STEM image; (C) Electron diffraction pattern acquired at the CH3NH3PbI3 (MAPbI3) region; (D) d-spacing between the adjacent lattice planes measured on the pattern C.

Figure 6 .
Figure 6.(A) STEM image of a cross-sectioned Al 2 O 3 /perovskite solar cell; (B) EDS line profile of PbL, IL, and AlK signals along the line shown on the STEM image; (C) Electron diffraction pattern acquired at the CH 3 NH 3 PbI 3 (MAPbI 3 ) region; (D) d-spacing between the adjacent lattice planes measured on the pattern C.

Figure S1 :
Scanning transmission electron microscopy (STEM) image of a cross-sectioned TiO 2 /perovskite solar cell and the energy dispersive X-ray spectroscopy (EDS) line profiles of PbL, IL, and TiK signals along the line shown on the STEM image.MAPbI 3 : CH 3 NH 3 PbI 3 ; Figure S2: STEM image of a cross-sectioned Al 2 O 3 /perovskite solar cell and the EDS line profile of PbL, IL, and AlK signals along the line shown on the STEM image; Table