Self‐Formation of SnCl2 Passivation Layer on SnO2 Electron‐Transport Layer in Chloride–Iodide‐Based Perovskite Solar Cell

The phenomenon of the self‐formation of a passivation layer at the interface of the perovskite/electron‐transport layer (ETL) is observed. FA0.6MA0.4PbI3−xClx perovskite thin film is deposited on a SnO2 nanoparticle thin‐film ETL. It is observed from the depth‐resolved spectroscopy that the Sn2+ ion migrates toward the perovskite layer within the ETL. At the same time, Cl− ion also migrates toward ETL within the perovskite layer. This unique ion migration phenomenon leads us to conclude that a passivating SnCl2 layer is formed at the perovskite/ETL interface. It is found that this SnCl2 layer at the interface works as a passivation layer like Al2O3. There is a significant effect of this self‐formed passivating layer behind the improvement of the device's efficiency and stability. It is believed that this SnCl2 passivation layer helps to reduce the recombination loss at the interface and boosts the performance of the perovskite solar cell (PSC). The perovskite/hole‐transport layer is also passivated with octylammonium bromide. Finally, the PSC offers a photoconversion efficiency (PCE) of 20.81% under 1 sun and AM1.5 G condition. Again, it maintains more than 80% of PCE under open‐air room conditions, white light emitting diode, and 85 °C continuous heating for more than 12 h without encapsulation.


Introduction
Perovskite solar cells (PSC) are emerging as a promising candidate for future solar energy harvesting.Besides gradually achieving efficiency comparable with the current market available silicon solar cells, another major advantage of PSCs is solution processability.Due to the solution process of the perovskite active layer, there are inherent unavoidable pinholes which is a big issue at present.These pinholes generate defect states at the interface with charge-transport layers (CTLs) which are detrimental to device performance because they work as trap states for photogenerated electrons.Therefore, finding a suitable interfacial semiconductor that is both environment-friendly and cheap is necessary to passivate interfacial defect states.What if any interfacial defects are selfhealed with a self-formed interfacial passivation layer?Here, we will present such a self-formed interfacial passivation layer between the perovskite and electrontransport layer (ETL).
PSCs' efficiency mostly depends on the efficient photogenerated charge extraction and transport toward the electrodes through CTLs.For this reason, interfaces of perovskite with the CTLs need careful treatment to reduce the interfacial defects.Nonradiative recombination at the interfacial defect states is the biggest challenge for efficient charge extraction by the CTLs.Especially the front-side interface between the perovskite and CTL is very crucial as most of the visible wavelengths of radiation are absorbed here. [1]At the same time, these interfaces play a crucial role in the stability of the device. [2]Generally, weak secondary chemical bonds like hydrogen bonds and van der Waals interaction are found between the perovskite layer and CTLs. [3]he weak nature of the bonding between the perovskite and CTL makes the interfaces vulnerable to more nonradiative recombination losses. [4]n recent years, stannic oxide (SnO 2 ) ETL is becoming popular in n-i-p-type PSCs. [5][8][9] Different methods have been applied with SnO 2 ETL to reduce the interfacial recombination losses and thus improve the charge transport properties, such as doping with different atoms to control the energy-level alignment [10] and surface passivation at the interface to effectively reduce the defect states and the nonradiative recombination. [11]Again, chloride-iodide perovskite is considered more favorable than its counterpart bromide-iodide perovskite for PSC. [12]There are some advantages with chloride-iodide perovskite such as better crystallization, almost no bandgap change, and long electron diffusion length while doping with chlorine (Cl). [12]ecently, there have been some breakthroughs concerning the efficiency and stability of PSCs that are related to chloride-iodide perovskite or Cl treatment (mainly to improve crystallinity and morphology) of iodine (I)-based perovskite and SnO 2 ETL-based PSC.Yan et al. [13] fabricated a PSC (p-i-n) with MACl-treated Ibased perovskite and SnO 2 ETL.The PSC offered more than 23% photoconversion efficiency (PCE) and maintained it for more than 1500 h at 85 °C.Similarly, Yang et al. [14] observed PCE of more than 25% with improved stability.The PSC (n-i-p) offered 92% of its initial efficiency after 5000 h.
It is now well established that chloride (Cl À ) and iodide (I À ) ions in chloride-iodide perovskite migrate toward the CTLs.Here, to mention that according to some recent works, a chemically interconnected interface could be formed between perovskite and SnO 2 .For example, Min et al. [15] reported that a coherent interlayer formed between Cl-treated perovskite (FAPbI 3 ) and Cl-bonded SnO 2 (n-i-p PSC).They found a crystalline perovskite (FASnCl x ) phase as an atomically coherent interlayer between the perovskite and SnO 2 .This interlayer between the perovskite and SnO 2 reduces the interfacial charge recombination with minimal contact resistance and results in PCE.It was also found in some of the earlier work on PSC (n-i-p) by Snaith et al. [16] that a passivation layer of Al 2 O 3 between TiO 2 and perovskite (MAPbI 2 Cl) reduces the recombination loss and enhances the PCE.In another work (p-i-n-type PSC) by Marshall et al. [17] they mixed extra stannous chloride (SnCl 2 ) in perovskite (CsSnI 3 ) precursor.The device was made without hole-transport layer (HTL) and phenyl-61-butyric acid methyl ester (PCBM ETL).They found that SnCl 2 formed as a very thin layer (approximately 1 nm) on the surface of the perovskite thin film and guessed a similar SnCl 2 layer formation between the CsSnI 3 and indium tin oxide (ITO).SnCl 2 is a high bandgap semiconductor and works as a passivation layer.This layer is considered a reason for low recombination at the perovskite/ITO interface and as a result high PCE in that device.At the same time, SnCl 2 at the interface of perovskite/PCBM improved the device stability.
Due to halogen ion migration toward the CTLs in chlorideiodide perovskite, there are atomic vacancies and defects, especially in the vicinity of the grain boundaries in the bulk perovskite.Therefore, both bulk and surface passivation are mandatory simultaneously.Recently, Arafat et al. [18] applied octylammonium (OA) halides on Cs 0.07 Rb 0.03 FA 0.765 MA 0.135 PbI 2.55 Br 0.45 perovskite surface.Here, they found that the smaller halide anion diffuses into the bulk perovskite, while the long ligand organic cation remains on the surface.The halide anions heal the bulk defects, and the organic cations heal the surface defects.In their study, OA chloride showed the best performance among different OA halides.It is well established that the stability of PSC is better for bromide-iodide perovskite compared to chloride-iodide one. [19]Therefore, OA bromide (OABr) should be the most effective for chloride-iodide perovskite.At the same time, long OA cation works as a 2D capping layer on top of the 3D perovskite. [20] this report, first, we optimize the SnO 2 nanoparticle percentage and Cl percentage in the perovskite to get the control device.We carry out a broad range of analyses to find out the possible causes of the high performance.From the analysis, we find that ion migration happens not only within the chloride-iodide perovskite layer but also in the SnO 2 layer.We see that Cl À and Sn 2þ ions migrate toward the perovskite/ ETL interface.This phenomenon indicates the self-formation of the SnCl 2 passivation layer at the interface.Then we passivate the perovskite/HTL interface with OABr to get the best performance from our proposed PSC.The passivation layers have a significant effect on the efficiency and stability of PSC.In our case, the SnCl 2 is self-formed and helps to passivate the defects and voids and strain or stress at the interface of perovskite/ETL.Again, the bromine (Br) doping makes the perovskite active layer more robust to moisture and oxygen which helps in lengthening the stability time.The 2D capping layer formed by the long cation OA at the interface of perovskite/HTL also helps to passivate the defects and voids there.We propose an n-i-p PSC with a PCE of 20.81% and stability, showing more than 85% of its initial PCE over 650 h in ambient open-air conditions.The other key photovoltaic parameters are a fill factor (FF) of 75.70%, an open-circuit voltage of 1.17 V, and a short-circuit current of 23.56 mA cm À2 .

Results and Discussions
In this article, we fabricate chloride-iodide PSC (n-i-p) with SnO 2 nanoparticles thin film as ETL and Spiro-MeOTAD as HTL.The perovskite we used for the fabrication of the device is FA 0.6 MA 0.4 PbI 3Àx Cl x , where FA represents formamidinium and MA represents methylammonium.MA þ ions are very sensitive to moisture, and we partially replaced MA þ ions with FA þ ions. [21]Again, the high content of FA þ ions makes the perovskite crystal structure unstable. [22][25][26][27] Therefore, we adopt the FA þ and MA þ ions ratio as 60% and 40% for our PSC and change the ratio of the I and Cl in our perovskite.We vary the Cl concentration from 5% to 15% (5%, 10%, and 15%) and find different combinations of perovskites (FA 0.6 MA 0.4 PbI 2.85 Cl 0.15 , FA 0.6 MA 0.4 PbI 2.7 Cl 0.3 , and FA 0.6 MA 0.4 PbI 2.55 Cl 0.45 ).We measure the PCE for each case and sort out the optimum candidate.Before that, we investigate the optimum SnO 2 nanoparticle concentration for the ETL.
In our literature review, we found that 5-10% Cl concentration in chloride-iodide perovskite should give the best efficiency in the case of chloride-iodide PSC. [12]Therefore, first, we arbitrarily consider a 10% Cl sample (FA 0.6 MA 0.4 PbI 2.70 Cl 0.30 ) and vary the SnO 2 nanoparticle concentration for ETL.We use 1.875-15% (1.875%, 3.75%, 7.5%, and 15%) SnO 2 nanoparticle concentration and find the best PCE for the 7.5% SnO 2 nanoparticle sample.Figure 1a shows the J-V curves for the variation of the concentration of SnO 2 nanoparticles in ETL.We also statically analyze the individual parameters of V oc (Figure S1a, Supporting Information), J sc (Figure S1b, Supporting Information), FF (Figure S1c, Supporting Information), and PCE (Figure S1d, Supporting Information) (for statical data analysis we use 24 PSCs of each type).The average and champion values with a standard deviation of each key parameter for different SnO 2 nanoparticle concentrations in ETL are listed in Table S1, Supporting Information.From these analyses, it is clear that every parameter except J sc shows the highest values for the 7.5% SnO 2 nanoparticle sample.Even in the case of J sc, the 7.5% SnO 2 nanoparticle sample gives very close values to the highest condition of the 3.75% SnO 2 nanoparticle sample.Now, considering the 7.5% SnO 2 nanoparticle concentration for ETL, we vary the concentration of Cl doping in the chlorideiodide perovskite active layer.As we already consider a 10% Cl-doped perovskite active layer for the optimization of SnO 2 nanoparticle ETL, we then make devices with Cl concentration higher (15% Cl) and lower (5% Cl) values than 10% Cl sample.The results show that among those three concentrations, 10% Cl is the optimum one.Figure 1b shows the comparison of J-V curves for different concentrations of Cl.The 10% Cl sample shows the highest values in each key parameter like J sc (Figure S2b, Supporting Information), FF (Figure S2c, Supporting Information), and PCE (Figure S2d, Supporting Information).Only V oc (Figure S2a, Supporting Information) of the 10% Cl sample shows a lower value than the 15% Cl sample.The higher value of V oc is due to the bandgap broadening of the 15% Cl sample compared to the 10% Cl sample.The average and champion values with a standard deviation of each key parameter for different Cl concentrations in perovskite (for statical data analysis we use 24 PSCs of each type) are tabulated in Table S2, Supporting Information.Finally, we passivate the surface of a 10% Cl sample (that means 10% Cl in the perovskite active layer and 7.5% SnO 2 nanoparticle ETL) with OABr.Our perovskite consists of halogen elements of I and Cl only.Br doping in iodide perovskite gives better stability.Thus, we need to dope our perovskite with Br in the bulk.Again, 2D perovskite shows better stability compared to 3D perovskite. [20]Thus, we need to passivate the perovskite/HTL interface with some 2D perovskite that has a long cation chain.Here, OABr is a nice candidate for both bulk and surface passivation.The passivation of OABr increases the V oc significantly as shown in Figure 1b.From Figure 1b, we see that the J sc and FF remain almost the same.As a result, the PCE improved significantly as manifested in Figure 1b.The corresponding forward and reverse of J-V curves for 5-10% Cl concentrations and OABr passivation are shown in Figure S3a-d, Supporting Information.
Besides PCE, another important issue with PSC is stability.Three external factors are detrimental to PSC's stability.The three factors are moisture, heat, and UV light. [28]We take a half-day stability test for the proposed PSC as illustrated in Figure 1c for testing mutual moisture, heat, and white light emitting diode (LED) light (0.5 sun) effect.For this, we put our sample with no encapsulation on a hot plate at 85 °C under white LED light (0.5 sun) and a normal ambient laboratory environment for more than 12 h.After about 12 h, we get about 40% PCE for the 10% Cl sample and about 80% for OABr-the sample of respective initial values.We take a month-long stability test for the proposed PSC as presented in Figure 1d only for testing the moisture effect.For this, we put our sample in a normal ambient laboratory environment with no heat, light, and encapsulation for more than 560 h.From the data, we see that our proposed device is quite stable to moisture.After about 650 h, we find more than 80% PCE for the 10% Cl sample and more than 85% for the OABr sample of respective initial values.
The dark current measurements show very similar values at AE1 V for all the different Cl and OABr samples as shown in Figure 2a.In every case, the forward bias voltage current is higher (about one to two orders) than the reverse bias voltage current.In the case of a 15% Cl sample, the forward bias current shows significantly lower values.In dark conditions, there should be only room-temperature thermally excited electrons in the conduction band (CB).Thus, a lower dark current value is the desired condition.At the same time, the forward bias current should be greater than the reverse bias current.The lower forward bias dark current indicates that there are crystalline and morphological defects in the 15% Cl sample.Crystalline and morphological defects produce trap states.
The room-temperature thermally excited electrons are trapped in these trap states.
The external quantum efficiency (EQE) measurements in Figure 2b manifest more than 90% for the 5% and 10% Cl and OABr samples.The EQE value of the 15% Cl sample is more than 70%.The EQE of 5% and 10% Cl and OABr samples is almost flattened for the whole wavelength range, while the EQE of the 15% Cl sample shows substantially lower values, especially in the long wavelength range.If we further compare the EQE results of 5% and 10% Cl and OABr samples, we see a descending pattern for OABr, 10%, and 5% Cl conditions, respectively.This indicates the better crystalline and morphological features of the OABr sample compared to the 10% and 5% Cl samples.A higher amount of Cl might cause these crystalline and morphological imperfections.We know that the radius of Cl is much smaller than I.In I-based perovskite, the octahedron is formed by six I-atoms and one lead (Pb) atom.It is known that the lattice constants increase with the bigger size of halide anions and vice versa. [29]Due to the radius mismatch of Cl and I, while doping in I-based perovskite, deformation occurs in the octahedron.Due to local stress or strain, the Pb-Cl/Pb-I bonds could be broken, and the Pb 2þ /I À /Cl À ions be detached.As the Cl À ion is the smallest one, it can be easily detached and local atomic vacancies are created which ultimately results in crystalline and morphological imperfections.That is why the higher amount of Cl addition also causes the phase segregation of the mixed chloride-iodide perovskite into two individual phases of MAPbI 3 and MAPbCl 3 . [12]32][33] Due to the separation of the high bandgap phase, the absorption of the 15% Cl sample in the long wavelength range is low.In addition, here from the EQE data, we also determine the bandgaps of our different samples (see Figure S4a, Supporting Information) as follows: 5% Cl sample (820 nm, 1.51 eV), 10% Cl sample (815 nm, 1.52 eV), 15% Cl sample (813 nm, 1.53 eV), and OABr sample (813 nm, 1.53 eV).These results show that with increasing Cl quantity the bandgap is increasing and due to Br diffusion in the bulk the bandgap is also increased.The low absorption causes the lower value of EQE in the long wavelength range for the 15% Cl sample.To observe the absorption property, we take the absorption spectra of those samples as depicted in Figure 2c.The absorption spectra of those samples manifest that the absorption of the 15% Cl sample is low in the long wavelength region.In addition, here from the absorption data, we also calculate the Tauc plot to determine the bandgaps of our different samples (see inset of Figure S4b, Supporting Information) as follows: 5% Cl sample (1.54 eV), 10% Cl sample (1.54 eV), 15% Cl sample (1.56 eV), and OABr sample (1.54 eV).These results also indicate that with increasing Cl quantity the bandgap is increasing and due to Br diffusion in the bulk the bandgap is also increased.We also calculate the integrated J sc from the EQE and in each case, the final J sc value perfectly matches with their corresponding J sc values and is presented in Figure 2b.
We examine the crystalline properties of different perovskite samples with different Cl concentrations and samples with OABr passivation.Figure 2d shows the X-Ray diffraction (XRD) patterns which explain the crystalline features of all samples.The peak at 12.60°corresponds to PbI 2 and in the case of the 15% Cl sample, this peak shifts to 12.20°. [34]Here, this peak is broadened and divided into two peaks: one with high intensity and another one with less intensity.The intensity of this peak relation to PbI 2 is higher for the 15% Cl sample compared to the other two samples of 5% and 10% Cl.This is the indication of higher I À ion migration toward the surface.The higher amount of Cl addition deforms the PbI 6 octahedral framework and as a result, free ions like (Pb 2þ , I À , and Cl À ) are created.The free Pb 2þ and I À ions migrate toward the surface and combine to form PbI 2 .Surprisingly, this peak is absent in the OABr sample.OA is a large organic cation, and it is not able to diffuse into the bulk perovskite.It remains on the surface of bulk perovskite and reacts with the PbI 2 on the surface to form a 2D perovskite.The peak at 14°corresponds to both tetragonal MAPbI 3 or cubic FAPbI 3 .The peak at 28°represents the characteristic peak of tetragonal MAPbI 3 . [35,36]The peaks at 14°, 25°, 30.50°, and 32°i ndicate the cubic FAPbI 3 . [37]The peaks relating to MAPbI 3 or FAPbI 3 are highly intense in the 15% Cl sample compared to the other samples.Another peak at 15.40°corresponds to the cubic MAPbCl 3 /FAPbCl 3 . [38,39]Interestingly, this peak corresponding to cubic MAPbCl 3 / FAPbCl 3 at 15.40°is only visible for the 15% Cl sample.This peak indicates that there is phase segregation in the 15% Cl sample.On the other hand, the intensity of peaks corresponding to MAPbI 3 /FAPbI 3 is higher in the 5% Cl sample than in the 10% Cl sample and lowest in OABr sample.That means a 5% Cl sample has more MAPbI 3 / FAPbl 3 phases.We can now conclude that as the intensity of all the characteristic peaks relating to MAPbI 3 or FAPbl 3 or MAPbCl 3 or FAPbCl 3 is the least in number in the 10% Cl and OABr sample, the mixed halide phase forms here.We also witness that the width of all the peaks related to MAPbI 3 /FAPbl 3 and MAPbCl 3 /FAPbCl 3 is broadened in the 15% Cl sample compared to the other samples.This is the indication of deformed and less compact crystallinity of 15% Cl sample due to the addition of a higher amount of Cl.Besides, the characteristic peak of SnO 2 is at 26.10°and we do not see any peaks there as it was a very thin layer. [40]There are peaks from ITO with very low intensity.[43] Thus, it is clear that we get characteristic peaks from the ITO (front electrode), while the XRD is taken from the opposite surface.This indicates that the X-Ray penetration depth is more than 400 nm.We also found a peak around 19.80°which should present the characteristic peak of SnCl 2 . [44]We also compare this peak with SnCl 2 data (code 15 452-ICSD) from ICDD.
We also carry out the electrochemical impedance spectroscopy (EIS) of two samples with 0% and 10% Cl concentrations.There are generally two important regions of frequency in the EIS of perovskite.The high-frequency region is related to electron transport, while the low-frequency region is related to ion migration.To compare the EIS among different samples, we chose 1 mV as the bias voltage and experimented in dark condition.We choose to measure in dark condition because we only want to investigate the ion migration property.In dark condition, there are no photogenerated electrons, and it is expected only ion migration with bias voltage.From the EIS of our samples (see Figure S4c, Supporting Information), we see that there are two loops.A smaller loop is situated on the higher frequency region and represents the impedance faced by electrons.A larger loop is situated on the lower frequency region and represents the impedance faced by ions.The real axis stands for the resistance and the imaginary axis stands for reactance (here capacitive reactance).The resistance (high-frequency R H and low-frequency R L ) and capacitance (high-frequency C H and low-frequency C L ) values are summarized in Table S3, Supporting Information.R H is lower for the 10% Cl sample, while the R L is equal for both samples.That means the 10% Cl sample poses better crystalline and morphological properties.We observe that the reactance values decrease with increasing Cl quantity.Thus, capacitance values increase in 10% Cl sample.This is due to the self-formation of SnCl 2 at the perovskite/SnO 2 interface.Therefore, with Cl addition, ion migration happens, and ions accumulate at the perovskite/SnO 2 interface which results in a high capacitive effect.
We take different microscopy images to explore the morphological properties of different samples with different Cl quantities.The Kelvin probe force microscopy (KPFM) images of different samples show contact potential differences.It is found that the contact potential difference gradually increases from 5% Cl to 15% Cl sample.The contact potential differences are 0.0625 V (Figure 3a), 0.075 V (Figure 3b), and 0.095 V (Figure 3c) for 5%, 10%, and 15% Cl samples, respectively.If we see the spatial contact potential difference for all the samples, there is almost no visible difference for the 5% and 10% Cl samples.In the case of a 15% Cl sample, the spatial contact potential difference is visible.There are lots of maxima and minima contrast in the scanned region, which interprets the phase segregation of perovskite into the iodide phase and chloride phase.At the same time, to further explore the phase segregation phenomena, we have done photoluminescence (PL) spectra of different samples presented in Figure S4d, Supporting Information.From the PL spectra, it is found that the PL peaks are at 800 nm (1.55 eV) for 5% and 10% Cl samples.However, we have got two PL peaks at 790 nm (1.57eV) with higher intensity and 630 nm (1.97 eV) with lower intensity.These two different peaks must be from two different phases of MAPbI 3 and MAPbCl 3 .Therefore, it is obvious that a higher amount of Cl addition is creating phase segregation.This is a clear evidence of phase segregation when an excess amount of Cl is added to the I-based perovskite.Atomic force microscope (AFM) images of different samples illustrate the surface roughness as shown in Figure 3d-f.The mean roughness of these three samples is 37.89, 31.99, and 46.30 nm for 5%, 10%, and 15% Cl samples, respectively.From here we see that the 10% Cl sample has the least roughness factor.We also observe that the AFM produces crispier images of 5% Cl and 10% Cl samples than 15% Cl sample.
The scanning electron microscopy (SEM) images further demonstrate the surface morphology of our samples.From the SEM images, as given in Figure 4a-f, we see that the morphological aspects are very similar for all the samples.For the 15% Cl sample, there are visible grain boundaries compared to the other two samples.Here, the grain boundaries work as defects in the thin film and produce trap states.Among the three samples, the 10% Cl sample shows the best morphological property.The 5% Cl sample has a similar morphology as like 10% Cl sample with pinholes in different areas.The corresponding cross sections of different samples are presented in Figure 4d-f.If we closely inspect the interface of perovskite/Spiro-OMeTAD, we see that there are grain boundaries in the 15% Cl sample.More close inspection reveals that there are also pinholes at the interface of the 5% Cl sample and a very smooth interface in the 10% Cl sample.The AFM image of the OABr treated chloride-iodide film is shown in Figure S5a, Supporting Information.We find that the mean roughness of the film is further reduced to 26.26 nm.This is due to surface defects passivation by the long organic ligand of OA.We further take an SEM image of the same sample and get a much smoother surface (see Figure S5b, Supporting Information).We use focused ion beam milling of this sample to see the cross section more clearly and find a clearer SEM image of different layers of the device as shown in Figure S5c, Supporting Information.It is noticed that there are light-colored substances in all the samples.This might be due to the degradation of samples by oxygen or moisture from the surroundings.
To investigate the phenomena of ion migration, we carried out energy-dispersive X-Ray spectroscopy (EDS) from the surface of the different samples.The EDS taken from the surface of the samples reveals the atomic percentage of different elements especially Pb, I, and Cl (see Figure S6, Supporting Information and Table S4, Supporting Information).[47] The diffusion of I À ions is much higher compared to Pb 2þ and Cl À ions toward the surface of all the samples.We see that the number of Pb 2þ and I À ions coming out to the surface is much lower for the 10% Cl and OABr samples.These results ensure the optimum compact PbI 6 octahedral framework for 10% Cl and OABr samples.The Cl À ions diffusion toward the surface of 5% and 10% Cl and OABr samples is quite low while of 15% Cl sample is surprisingly high.Because of the radius mismatch of Cl À and I À ions, the higher concentration of Cl À ions deforms the PbI 6 octahedral framework and escapes from the octahedron.In this way, atomic vacancies are created, and those vacancies are the pathways for ion migration.
We also dig out the device to investigate the interface of perovskite/SnO 2 .We first performed a time-of-flight secondary ion mass spectroscopy (TOF-SIMS) depth profile analysis of the 10% Cl sample (Figure 5a).From the analysis we see that both Pb and I atoms count sharply degrade after 100 s of sputtering time.At the same time, the oxygen (O) atom count starts to increase after 100s and becomes saturated.Because Pb and I atoms are components of the perovskite and O is the component of SnO 2 , it could be claimed that the perovskite/SnO 2 interface is situated between 100 s and 150 s.It is expected that the I atoms migrate toward the perovskite/Spiro-OMeTAD interface, and the Cl atoms migrate toward the perovskite/SnO 2 interface.Here at the surface, we see that the I-atom count is higher than in the bulk, whereas the Cl-atom count is lower than in the bulk.On the opposite side of the thin film, we see that the I-atom count is the same as the bulk, while the Cl-atom count is higher than the bulk.It is now clear that the I atoms migrate toward the surface and Cl atoms migrate toward the opposite.Cl and Sn atoms show a very interesting atomic quantity profile.Cl and Sn atoms count gradually increases and then gradually decreases and shows peak value at the perovskite/SnO 2 interface.To further verify this phenomenon, we performed the X-Ray photoelectron spectroscopy (XPS) depth profile analysis to find out and examine the interface of perovskite/SnO 2 (Figure 5b).The XPS also shows a similar atomic concentration profile of TOF-SIMS.The amount of Pb and I atoms remains almost constant until the 600 s.After 600 s, the amount of Pb and I atoms shows a sharp degradation.On the other hand, after 600 s, the quantity of O atoms shows a sharp upgradation and tends to be constant after 1400 s.Therefore, the perovskite/SnO 2 interface is situated between 600 s and 800 s.The remaining Cl and Sn atoms also show similar peak profiles at the perovskite/SnO 2 interface, just as we see in the case of TOF-SIMS.Here, the most surprising phenomenon is the Sn atomic migration toward the interface of perovskite/SnO 2 .This Sn atomic migration is evidence of the self-formation of a SnCl 2 passivation layer at the interface.If we see the depth-resolved binding energies of Sn, Cl, I, and O (Figure S7, Supporting Information), we see that the binding energies of both the elements (Sn and Cl) shift toward higher energy values with depth increase and again shift toward lower energy values.This supports the possible formation of a SnCl 2 passivation layer.At the same time, the binding energy variation of I and O seems to be unchanged with the depth variation.I and O do not tend to be bonded with other elements.
At this stage, we describe the possible phenomena behind the self-formation of SnCl 2 at the interface of perovskite/SnO 2 .It is well known that O vacancy is one of the most common defects in metal oxide (MO). [48]The concentration of O vacancy also depends on the morphology of the MO.Kar et al. studied different morphology types of SnO 2 like nanoparticles, nanospheres, and nanorods and found that nanoparticles exhibit a higher concentration of O vacancy compared to those nanospheres and nanorods. [49]Usually, the O vacancy is generated at the surface of SnO 2 . [50]Due to O vacancies at the surface of SnO 2 nanoparticles, there are many dangling bonds with local stress and therefore free Sn 2þ ions.In the perovskite bulk side,  due to a radius mismatch between Cl and I, there is a local stress or strain in the octahedron of PbX 6 (X = Cl or I).Thus, there are many free Cl À and I À ions.Again, due to the solution process, there are lots of atomic vacancies in the bulk of both sides.The mobile Sn 2þ , Cl À , and I À ions diffuse toward the interface through these atomic vacancies.The attractive coulombic force works between Sn 2þ and Cl À or I À atoms.The electronegativity of Cl is higher than I. Thus, the coulombic attraction between Sn 2þ and Cl À overcomes the coulombic attraction between Sn 2þ and I À .Thus, Sn 2þ and Cl À migrate toward the interface of perovskite/SnO 2 and there is a possible formation of SnCl 2 .
There is an accumulation of Sn 2þ and Cl À ions at the interface which generates a self-electric field.The free I À ions are repelled by the electric field move toward the opposite direction and accumulate near the opposite interface.Now we see that the device performance of our designed PSC depends on the self-formation of the SnCl 2 passivation interlayer at the perovskite/SnO 2 interface.In this respect, we want to mention the pioneering work by Snaith et al. [16] They fabricated n-i-p PSC with an Al 2 O 3 passivation layer to help the photogenerated electrons remain in the perovskite layer until they are collected by the ETL.Al 2 O 3 passivating layer efficiently reduced the photogenerated electron-hole recombination in their device.
Similarly, we expect this self-formed SnCl 2 passivation interlayer to play an efficient role in reducing the electron-hole recombination at the interface of perovskite/SnO 2 .This interfacial passivation layer should reduce the back-recombination of extracted photogenerated holes at the interfacial region and at the same time should not reduce the photogenerated electron transport toward ITO.Again, in the n-i-p type device, most of the visible wavelength radiation is absorbed near the vicinity of the perovskite/SnO 2 /ITO interface.Due to the high amount of photogenerated carriers at the interface, there is the possibility of electron-hole recombination at the perovskite/SnO 2 interface.That is why this self-formation of the SnCl 2 passivation layer helps to reduce the interfacial recombination and thus improve performance.This finding is just the counterpart of the work on p-i-n type HTL-free PSC by Marshall et al. [17] In their work, they claimed a formation of the passivating SnCl  According to them, the VB of SnCl 2 is about 6.6 eV below the vacuum level and the bandgap is 4.15 eV.After calculating the CB of SnCl 2 , they found that the CB of SnCl 2 is also higher than the CB of CsSnI 3 .Therefore, SnCl 2 at the interface of CsSnI 3 / ITO worked as an insulating interlayer for electrons.They explained that the improved efficiency of their device is due to the efficient hole transport through the SnCl 2 layer from CsSnI 3 to ITO.In our case, we fabricate a n-i-p type-PSC and obtain a self-formed passivating SnCl 2 layer at the perovskite/SnO 2 interface.From the UPS analysis of our 10% Cl sample, we find the VB at 5.3 eV from the vacuum level as shown in Figure S8a, Supporting Information.From the absorption spectra, we calculate the Tauc plot and find the bandgap of 10% Cl sample is 1.54 eV (see Figure S8b, Supporting Information).Thus, the CB of SnCl 2 % and 10% Cl sample should be 2.45 and 3.75 eV, respectively.[53][54][55] According to the above findings, we draw the band structure and alignment of SnO 2 /SnCl 2 /FA 0.6 MA 0.4 PbI 2.7 Cl 0.3 which is depicted in Figure 5d.From the structure, we see that the photogenerated holes face an extra barrier to go through SnO 2 toward ITO.While the photogenerated electrons can transmit to SnO 2 by electron tunneling effect through very thin SnCl 2 .It is expected that there is only one monolayer of SnCl 2 as the self-formed SnCl 2 is forming due to ion migration.

Conclusion
In this experiment, we witness a unique phenomenon of the self-formation of the passivating SnCl 2 layer at the interface of perovskite/SnO 2 .It is found that Cl À ions migrate toward the interface of perovskite/SnO 2 .It is also found a surprising fact of Sn 2þ ion migration toward the interface of perovskite/SnO 2 .This unique phenomenon is confirmed by TOF-SIMS and XPS depth profiling.We believe that this ionic migration results in the self-formation of a monolayer of the SnCl 2 passivation layer at the interface of perovskite/SnO 2 .The SnCl 2 passivating layer helps to reduce the back-recombination of photogenerated electrons and holes, which ultimately increases the photogenerated current.The self-formed SnCl 2 is formed by ionic bonds and creates a strong interfacial bonding between perovskite and SnO 2 .
In our experiment, we optimize the SnO 2 nanoparticle concentration in ETL and Cl concentration in the active layer.We find the optimum value of SnO 2 nanoparticle concentration is 7.5% and Cl concentration is 10%.We also passivate the bulk perovskite with OABr.Here, OABr serves three purposes: first, it passivates bulk perovskite; second, it passivates the surface of the perovskite with the long organic OA ligand; and third, Br passivation in the bulk contributes to the enhanced stability of our proposed PSC.We observe a champion PCE of 20.81% and it can retain more than 80% of its initial PCE after 12 h in an open-air room-temperature laboratory environment under white LED and 85 °C continuous heating and with no encapsulation.

Figure 1 .
Figure 1.Different J-V characteristics of chloride-iodide PSCs for a) different concentrations of SnO 2 nanoparticle in ETL and b) different amounts of Cl doping concentrations/OABr passivation on perovskite active layer.Stability comparison among different chloride-iodide PSCs with 10% Cl and OABr-passivated 10% Cl samples: c) mutual moisture, heat, and white LED light (0.5 sun) effect and d) moisture effect.
2 layer at the CsSnI 3 /PCBM ETL interface.Here, they used an extra 10 mol % of SnCl 2 with the perovskite precursor and found a very thin layer of SnCl 2 at the surface.They guessed the same possible formation of the SnCl 2 layer at the interface of ITO/no HTL/CsSnI 3 .The ultraviolet photoelectron spectroscopy (UPS) data they got show a deep valence band (VB) of SnCl 2 compared to the VB of CsSnI 3 as well as the work function of ITO.

Figure 5 .
Figure 5.The elemental depth profile analysis of perovskite film with 10% Cl sample by a) TOF-SIMS and b) XPS.Here, the sample structure is ITO/SnO 2 (7.5%)/FA 0.6 MA 0.4 PbI 2.7 Cl 0.3 .c) Schematic diagram of our proposed PSC structure.d) Band structure and band alignment of the heterostructure of SnO 2 /SnCl 2 /FA 0.6 MA 0.4 PbI 2.7 Cl 0.3 .