Synergetic Effect of Aluminum Oxide and Organic Halide Salts on Two‐Dimensional Perovskite Layer Formation and Stability Enhancement of Perovskite Solar Cells

Long‐chain organic halide salts are widely used in perovskite‐based optoelectronic devices for surface passivation owing to their capability to interact with the surface defects of perovskites. Here, aluminum oxide (AlOx) is introduced via atomic layer deposition onto octylammonium iodide (OAI) to exploit the benefits of organic halide salts without generating undesired defects. The devices incorporating AlOx on OAI‐treated perovskite (OAI/AlOx) show enhancement in both device performance and photo‐stability compared to those with only treatment. A diffusion of aluminum from AlOx into the perovskite through surface characterization contributes to a uniform photo‐generated carrier transport in both the surface and the bulk of the perovskite absorber. In addition, it is revealed that light‐induced two‐dimensional perovskite formation on OAI/AlOx. This may be ascribed to preventing the loss of OA cations due to the presence of AlOx, leading to a decrease in the number of iodine anions which suppresses the light‐induced degradation of corresponding devices. Consequently, the devices show over 24% efficiency and retain their efficiency over 1000 hours under continuous light illumination.


Introduction
Organic-inorganic halide perovskite solar cells (PSCs) have attracted interest in the solar cell research community due to their excellent optoelectronic properties including long carrier diffusion length, [1] high absorption coefficient, [2] and easily tunable bandgaps. [3]Although PSCs have shown power conversion efficiencies (PCE) as high as 25.7%, [4] defects within the perovskite absorber and at the perovskite/hole transport layer (HTL) interface hinder their further development toward commercialization as they induce non-radiative recombination of photo-generated carriers [5] and severe ion migration, accelerating the degradation of such devices. [6]ne strategy to tackle this issue is to post-treat organic halide salts on the surface of the perovskite as a passivation medium which can for instance reduce the number of surface defects.Jiang et al. revealed that fluorinated phenethylammonium iodide can inhibit the formation of iodine vacancies which act as non-radiative recombination centers. [7]It was also demonstrated that prophylammonium iodide can reduce the defect density in perovskites and enhance the moisture stability of PSCs. [8]espite organic halide salts' beneficial impact on device performance, detrimental effects originating from their incorporation in PSCs have also been reported. [9]9b] These undesired defects accelerate ion migration, resulting in the photo-induced degradation of the devices.9a] It was also reported that aggravate ion migration can occur in octylammonium iodide (OAI)-treated PSCs under illumination. [10]Although they can enhance device performance, through these collective results, it is understood that using organic halide salts is not enough for improving device stability.
One strategy to improve the stability of the devices is to insert metal oxide layers deposited by atomic layer deposition (ALD) at a suitable interface in the PSCs. [11]This has been widely reported to improve device stability, especially against humidity. [12]as et al. demonstrated that inserting ALD aluminum oxide between the perovskite film and the HTL, 2,2′,7,7′-tetrakis(N,N-dip-methoxyphenylamine)-9,9′-spirobifluorene (spiro-OMeTAD), can induce self-healing of the perovskite and suppress ion migration toward the HTL under illumination. [13]However, direct deposition of metal oxide layers using ALD onto the perovskite can be challenging owing to unfavorable interactions between the perovskite absorber and the ALD precursors such as trimethylammonium (TMA) [14] and water molecules. [15]This leads to reduced device performance despite improved stability.
Herein, we introduced aluminum oxide (AlO x ) by ALD onto OAI-deposited formamidinium lead iodide (FAPbI 3 ) PSCs to exploit the beneficiary effects of organic halide salts and inorganic metal oxides via ALD.The introduction of the AlO x layer not only helps improve the device's performance but also its photo-stability.In terms of the origin of improving device performance, surface analysis revealed the formation of non-stoichiometric aluminum oxide resulting in homogeneous carrier photo-generation confirmed by Kelvin probe force microscopy (KPFM).We further explored the origin of the improved photo-stability in the AlO x on OAI-treated PSCs.Based on a com-prehensive analysis, it was found that AlO x suppressed the formation of -phase perovskite in the OAI-treated perovskite and triggered the light-induced 2D perovskite formation.We propose that AlO x can prevent the loss of octylammonium (OA) cations during prolonged illumination, unlike in a merely OAItreated perovskite, enabling this light-induced 2D perovskite formation mechanism.Therefore, this 2D formation mechanism contributes to the reduction of volatile materials originating from OAI, preventing light-induced degradation.This work shines light on the beneficial role of AlO x by ALD when deposited onto long-chain organic halide salts-passivated perovskites.

Results and Discussion
We first investigate the effect of different passivation strategies on the photovoltaic performance of PSCs having FAPbI 3 as the absorber.The device configuration is presented in Figure 1a.Two types of passivation layer combinations were used: (1) treatment of the perovskite surface with OAI (denoted by OAI) and (2) deposition of AlO x by ALD onto the OAI (denoted by OAI/AlO x ).A detailed description of the fabrication procedures is provided in the Experimental Section.
Figure 1b shows the illuminated current density-voltage (J-V) curves of the champion PSCs without and with the passivation layers (OAI or OAI/AlO x ).The photovoltaic (PV) parameters are summarized in Table S1, Supporting Information.It was found that the highest performance was achieved by OAI/AlO x -treated devices.The statistics of the PV parameters for different device configurations collected on seven cells are shown in Figure 1c (see also Table S2, Supporting Information).The average PCE of the devices without a passivation layer (defined as control) is 21.54% which was increased to 23.21 and 23.75% in the OAItreated devices and in the OAI/AlO x -treated devices, respectively.The average short-circuit current density is almost unchanged across the three device configurations, which is consistent with the external quantum efficiency spectra of the devices as shown in Figure S1 and Table S3, Supporting Information.On the other hand, the average open-circuit voltage (V oc ) improved in the OAIand OAI/AlO x -treated devices to 1.14 and 1.15 V, respectively, as compared to the control devices (1.10 V).The average fill factor (FF) for the OAI-treated devices increased to 80.76%, showing 3.1% absolute enhancement with respect to the control, and further improved after introducing the AlO x , reaching 82.06%.
To understand the reason for improved V oc which is related to the energy level in the devices, [16] ultraviolet-visible (UV-vis) absorption (Figure S2, Supporting Information), and ultraviolet photoelectron spectroscopy (UPS) measurements (Figure S3, Supporting Information) were carried out.As shown in Figure S3, Supporting Information, there is an upward shift of the Fermi level for the OAI-treated perovskite, and the Fermi level shifts even further toward the conduction band minimum (CBM) for the OAI/AlO x -treated perovskite.The shift of Fermi level for OAI and OAI/AlO x is ascribed to the alternation of surface properties resulting from the introduction of the functional group of OAI. [10]However, it is important to note that the reduction in surface hole traps also contributes to the shift of the surface work function of perovskite, which is linked to changes in charge carrier density. [17]Figure S4 and Table S5, Supporting Information, through pulse-voltage space charge limited current (PV-SCLC) presents that the OAI/AlO x -treated sample presents a lower density of hole traps compared to other conditions.This suggests that the further shift of the Fermi level toward the CBM can be attributed not only to the inherent properties of OAI but also to the effective passivation of hole traps at the crystal boundary or interface. [18]In addition, the valance band maximum (VBM) values are −6.48,−6.33, and −6.03 eV for the control, the OAIdeposited perovskite, and the OAI/AlO x -treated perovskite, respectively, indicating a reduction of VBM values.Consequently, improved energy level alignment of VBM between the perovskite with Spiro-OMeTAD (−5.22 eV) in the OAI/AlO x configuration leads to enhanced V oc . [19]To understand the mechanism behind the enhancement of FF, an analysis of series resistance (R s ) and shunt resistance (R sh ) was conducted. [20]It is generally known that lower R s associated with increased charge transportation [21] and higher R sh related to the reduction of leakage current [22] are required to obtain higher FF. [23]Higher R sh and lower R s in the case of OAI/AlO x-treated devices were observed as shown in Table S2, Supporting Information, indicating reduced carriers' recombination as compared to that of the control sample. [24]urthermore, the PSCs of the control, OAI-, and OAI/AlO xtreated perovskite devices were measured under continuous illumination with the intensity of 1 SUN for 1000 h for long-term photo-stability.As shown in Figure 1d, a transient decline in performance during the initial few hours was observed at all sample conditions, with a more pronounced drop observed in the case of PSCs deposited with organic-halide salts.This behavior can potentially be attributed to cation migration from post-treatment of the organic salts, leading to partial degradation of the interface.However, notable differences in PCE trends were observed among the sample's conditions after 180 h.The control device demonstrated an 80% of retention of its initial PCE after a partial recovery, while the OAI/AlO x -treated devices exhibited a PCE value that rebounded to 95% of the initial PCE value after 1000 h.In contrast, the OAI-treated devices experienced a continuous decrease in PCE value, reaching 60% of initial PCE at 1000 h.Additionally, as shown in Figure S5, Supporting Information, 90% of initial PCE was retained for OAI/AlO x -treated PSCs , whereas there was a continuous decrease in PCE for OAI-treated PSCs after 1872 h under 85°C and 85% relative humidity.From device performance results, it is noted that the introduction of AlO x on the OAI-deposited perovskite results in both improved device performance and stability.
To unveil the interaction between the surface of the perovskite and post-treated materials, Fourier transforms infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS) measurements were implemented.As shown in Figure 2a, the intensity of the peak at 1712 cm −1 , corresponds to C═N stretching vibration of formamidinium (FA) cations. [25]The peak intensity at 1712 cm −1 slightly increased after the OAI and OAI/AlO x treatment.This indicates the possibility of compensation for decomposed FA or protection of FA due to post-treatment. [26]Furthermore, the peaks at 2851 , 2923, and 2959 cm −1 highlighted by the pink dotted line, were identified for the OAI-and OAI/AlO xtreated perovskite, supporting the existence of octylammonium (OA) cations on the surface, [27] which is consistent with the additional peak of C─N bonding detected in the case of the OAI-and OAI/AlO x -treated perovskite as shown in Figure 2b.In addition, the Pb 4f 5/2 and Pb 4f 7/2 peaks located at 143.1 and 138.25 eV shifted to lower binding energies of 142.92 and 138.0 eV, respectively, which was indicative of the interaction between nitrogen with lone electron pair [28] in the OAI molecule and uncoordinated lead at the surface of the perovskite.Furthermore, the peak intensity of metallic lead at 141.59 and 136.51 eV decreased after treatment of passivation (OAI-and OAI/AlO x ) as shown in Figure 2c.The specific peak at 400.4 eV, indicating the bonding of C═NH 2 + from FA cations [29] also shifted to 400.25 eV in the case of passivated perovskite (OAI and OAI/AlO x ) in N 1s spectra (Figure 2b).In addition, it was found that the N─H stretch vibration peaks (3387 and 3254 cm −1 ) [30] (Figure 2a) and the C─H bending vibration (1465 cm −1 ), [31] which originates from FAPbI 3 (Figure 2d) , shifted toward higher wavenumber in FTIR results (i.e., 3387→3401, 3254→3266, and 1465→1473 cm −1 ).These results related to the shift of peaks after post-treatment indicate a coordination reaction between the perovskite and post-treated materials. [32]o further investigate the formation of the metal oxide layer by ALD depending on different surfaces of perovskite, we also characterize the low wavenumber rages for FTIR spectra shown in Figure 2d, and Al 2p and O 1s for peaks of XPS spectra shown in Figures 2e and 2f, respectively.As shown in Figure 2d, we found a peak located at 667 cm −1 , corresponding to Al─O bonding (see the purple rectangular box). [33]However, the peak at 1522 cm −1 was not detected for OAI/AlO x -treated perovskite (see the orange rectangular box) .Given that the signal at 1522 cm −1 could originate from the stoichiometric aluminum oxide (Al 2 O 3 ), [34] the formation of Al 2 O 3 was not preferred when ALD was used on the OAI-treated perovskite.In Figure 2e, although the weak peak was located at 74.5 eV for Al 2 O 3 , originating from Al─O of Al 2 O 3 , [35] the distinct peak at 73.1 eV was observed for OAI/AlO x, which is attributed to AlO x suboxides. [36]As shown in Figure 2f, the peak in agreement with O─H bonding was present at 532.62 eV [37] for the control.It was shifted toward lower binding energy of 532.52 eV and the intensity of this peak has reduced for both OAI and OAI/AlO x -treated perovskite.These results indicated that the reduced alcohol group is due to the OA-based post-treatment.However, for only OAI/AlO x -treated perovskite, the binding energy of 532.4 eV, corresponding to Al─O, [38] and additional bonding located at 531.3 eV were observed.The lower binding energy peak at 531.3 eV could be attributed to AlO x of either deficient aluminum or oxygen. [39]Therefore, from the XPS results, the introduction of OAI can change the surface of the perovskite, resulting in non-stoichiometric aluminum oxide (AlO x ) formation during the ALD process.
To investigate the origin of this formation deeply, we conducted time-of-flight secondary-ion mass spectrometry measurement as shown in Figure 2g.The intensity of aluminum ions at the surface is higher due to the ALD process after the introduction of AlO x .Interestingly, aluminum ions seem to be diffused toward the bulk of the perovskite, resulting in a deficiency of aluminum elements at the surface.Hence, non-stoichiometric aluminum oxide formation from XPS results might be attributed to the diffusion of aluminum ions into the bulk perovskite as shown in Figure 2h.
As in the abovementioned results, it was found that aluminum ions arising from ALD diffused into the bulk for the OAI/AlO x-treated perovskite.To investigate the correlation between this observation and the electronic properties of the  OAI/AlO x -treated perovskite, KPFM was measured.In detail, contact potential difference (CPD) images of all samples were measured in the dark and under 750 and 400 nm illumination conditions, respectively (see the Experimental Section and Figure S6a, Supporting Information) to observe the influence of surface and bulk-related defects.As shown in Figure S6b,c, Supporting Information, it was noted that blue light (400 nm) can generate photo-induced carriers near the top surface (penetration depth ≈95 nm), while red light (750 nm) can produce photo-excited carriers in the bulk (penetration depth ≈300 nm). [40]Topography images for all samples acquired by KPFM measurement are shown in Figure S7a-d, Supporting Information.In addition, by subtracting CPD maps in the dark (CPD dark ) from ones under illumination (CPD light ) at 750 and 400 nm, surface photovoltage (SPV) could be obtained as shown in Figure S7d-i, Supporting Information, individually.
Figure 3a shows the full width at half maximum (FWHM) for each sample from SPV distribution curves (Figure S7j, Supporting Information).The FWHM value represents excess charge carrier homogeneity and smaller FWHM value implies better charge carrier homogeneity.As presented in Figure 3a, FWHM values of SPV distribution across the surface for the OAI-treated andcontrol samples show comparable FWHM values when illuminated by 750nm.Meanwhile, FHWM values at 750 nm for OAI/AlO x -treated perovskite have reduced, indicating enhanced excess charge carrier homogeneity.This could be attributed to uniform charge transport due to diffused aluminum cations as we observed in the above section for OAI/AlO x -treated sample, resulting in an increase in FF [42] for the corresponding device configuration.Note that the FWHM results' trend at 400 nm of incident light is similar to that in the case of 750 nm of incident light (i.e., FWHM values are comparable for control and OAItreated perovskite, while they decrease for OAI/AlO x -deposited perovskite).These results imply that both the surface and bulk excess charge carriers are more uniformly generated and seperated for the OAI/AlO x -deposited sample.
In Figure 3b, we compared the average values of SPV distribution as a function of wavelengths of incident light wavelength to deeply investigate the passivation effect.The OAI/AlO x -treated perovskite exhibits the lowest SPV values under illumination at 750 nm (0.355 V), following the control (0.393 V) and the OAItreated sample (0.469 V).Given that applying incident light with 750 nm on perovskite not only produces photo-generated carriers but also affects ions' migration in the bulk perovskite. [43]Nevertheless, given the trend of defects density from PV-SCLC (Table S5, Supporting Information) and FWHM of SPV (Figure 3a), it is suspected that OA from OAI was not consumed for passivating defects. Under illumination at 400 nm, SPV values of the OAI-treated sample an average SPV value of 0.173 V and the control sample also exhibited the similar SPV at 400 nm (0.193 V).In the case of the OAI/AlO x -treated sample, SPV values at 400 nm (0.315 V) are the largest compared to those of the control and OAI-treated sample.Nonetheless, considering that post-treatment (OAI, OAI/AlO x ) is enhancing hole extraction from UPS results (Figure S3, Supporting Information) and relatively less uniform charge separation across the surface (OAI) (Figure 3a), more photo-generated carriers from the OAI sample's bulk can be transported compared to the control, contributing to improving Voc.Meanwhile, OAI/AlO x -treatment could significantly reduce interfacial recombination [46] except for improving hole extract toward hole transport, which is in good agreement with much improved Voc [47] of the PSCs with OAI/AlO x (see Figure 1c).
Furthermore, to investigate the improvement in photo-stability for OAI/AlO x -treated devices as shown in Figure 1d, we utilized a customized degradation chamber as shown in Figure S8, Supporting Information.And then, KPFM measurements were implemented for the light-exposed samples.All topography and corresponding CPD images are in Figures S9 and S10, Supporting Information.
As shown in Figure 4a, in terms of the control, the formation of small grain clusters was observed (marked by white circles) under illumination for 182 h, which were also observed in scanning electron microscopy (SEM) images of the control sample (Figure S11, Supporting Information).Light exposure can degrade the initial perovskite morphology, inducing these small grain clusters. [48]The grains highlighted by red circles in Figure 4a show higher CPD values than those of the surrounding (see Figure S10, Supporting Information), which is attributed to the formation of decomposed hexagonal phases [49] (see Figure S12a, Supporting Information).Furthermore, the work function was reduced after illumination for the control sample (Figure 4b).The reduction in work function can be attributed to the generation of negatively charged iodine interstitial defects, [50] which have a lower formation energy. [51]This leads to a more n-type behavior at the surface and a corresponding decrease in work function.Consequently, a negative work function shift (that is, more n-type) under illumination could be responsible for lowering the activation energy of halide migration, which causes the determinantal effect on PSC stability. [10]For the OAI-treated perovskite, the average size of grains has reduced after light exposure as shown in Figure 4c without the formation of small grain clusters at specific locations on the surface of the sample.Furthermore, as shown in SEM images (Figure S11, Supporting Information), pinholes were also detected for the OAI-treated perovskite.Slightly reduced peak intensity of lead (II) iodide (PbI 2 ) and the additional peak which indicates -perovskite were detected from X-ray diffraction (XRD) patterns (see Figure S12b, Supporting Information).It was indicated that the severe decomposition of perovskite on the surface and the formation of volatile vapor in the bulk is due to light exposure. [52]The value of the work function reaches around 5.65 eV after illumination as shown in Figure 4d, which is consistent with the values of iodine vacancies (V I ). [53]During illumination, organic-halide salts (in our case: OAI) could be decomposed by light and evaporated, resulting in decreased amounts of OAI on perovskite.Consequently, pinholes could be formed at the surface owing to volatile gas from the decomposition of the OAI-treated perovskite.
Interestingly, as illustrated in Figure 4e, the morphology changes for OAI/AlO x -treated perovskite were different from the control and OAI-deposited perovskite.From XRD patterns in Figure S12c, Supporting Information, the -perovskite phase is rarely found.Therefore, it is hard to tell that only the formation of the -perovskite phase of OAI/AlO x contributes to morphological changes.Structure change of the capping layer on the perovskite could be possible under illumination, [54] which leads to morphology changes of perovskite.In addition, it was reported that the increased work function of post-treated alkylammonium chloride perovskite is associated with (e.g., formation of 2D perovskite), [55] which is similar to our observation as shown in Figure 4f.Therefore, we hypothesize that one of the possible reasons for different morphology changes in the case of OAI/AlO x is related to the formation of 2D perovskite induced by light unlike the aggressive generation of V I due to volatile iodide vapor from the perovskite for OAI-treated perovskite.
To confirm this assumption, we measured grazing-incidence wide-angle X-ray scattering (GIWAXS) with incident angles of 0.4°as shown in Figure S13a, Supporting Information.Penetration depth at the incident angle of 0.4°is 141 nm as shown in Figure S13b, Supporting Information.Processed GIWAXS images at the incident angle of 0.4°by the software are in Figure Figure 5. Grazing-incidence wide-angle X-ray scattering (GIWAXS) measurement.a) peak intensity at q z = 0.82 Å −1 (i.e., -4H phase) [56] and at q z = 0.84 Å −1 (i.e., −2H phase), [57] and b) that at q z = 0.46, 0.60, and 0.77 Å −1 (2D perovskite) [27] for sample conditions as a function of light exposure time at incident angles (i) of 0.4°.Note that Figure 5 is extracted from line profiles as shown in Figure S15, Supporting Information.Note that q z denotes the scattering vector of the vertical direction and we plot the intensity of each curve at certain q z positions.
As shown in Figure 5a, for the control, the intensity of peaks that indicate hexagonal −4H phase and hexagonal −2H, respectively, is higher compared to that for OAI and OAI/AlO xtreated perovskite in the dark.In addition, the intensity of q z = 0.84 Å −1 is much larger than that of q z = 0.82 Å −1 under illumination.This indicates severe transformation from the −4H phase (q z = 0.82 Å −1 ) to the −2H phase (q z = 0.84 Å −1 ) in the case of the control sample.For the OAI-deposited perovskite, the peak intensity of polytype phases was significantly reduced in the dark, indicating that OAI can suppress the formation of the polytype phases initially.However, the peak intensity of polytypes increases under illumination, especially showing a much larger increase in the −4H phase than in −2H phase.This indicates that although OAI treatment can suppress the formation of polytype phases, the formation of polytype phases under illumination cannot be mitigated.For OAI/AlO x -treated perovskite, it was observed that the peak intensity of polytype phases is also reduced compared to that of the control sample.In addition, there is no significant increase in peak intensity for polytype phases under illumination, which is consistent with XRD in Figure S12, Supporting Information.This indicates OAI-AlO x treatment can assist in the mitigated formation of polytypes under illumination.Next, considering that new observed peaks at q z ≈ 0.45, 0.61, and 0.78 Å −1 , which are consistent with (OA) 2 PbI 4 , [27] the correlation between 2D formation and surface treatment strategies can be determined by plotting these peak intensities as a function of illumination exposure time as shown in Figure 5b.As depicted in Figure 5b, the control and OAI-treated perovskite does not show a variation of 2D perovskite peak intensities as a function of illumination time, whereas in OAI/AlO x -treated perovskite these 2D perovskite peaks exhibit an increase in intensities under illumination.This shows that the formation of 2D perovskite under illu-mination conditions is expedited in the case of OAI/AlO x -treated conditions.
Furthermore, to determine the change of crystal orientation under illumination, the integrated plane diffraction at q r = 1.0 Å −1 (a (001) plane of perovskite) was plotted as a function of azimuthal angle, , with incident angles ( i ) of 0.4°in Figure S15b, Supporting Information.
Except for the decrease in intensity at  = 0, there are no significant changes in either isotropic regions or different tilt angles of  in the case of the control under illumination.This indicates reduced crystallinity, [58] resulting from the change of perovskite -phase in bulk [58] during illumination.The peak intensity at vertical orientation ( = 0°) increases before illumination in the bulk for the OAI-treated perovskite compared to that of the control sample.One of the possible scenarios is the diffusion of OA cations into the bulk, contributing to preferred vertical orientation in the bulk given that OA cations can anchor each perovskite crystal, resulting in preferred (001) orientation. [59]However, under illumination for 40 h, volatile OA cations can be released [54] from the bulk, resulting in a considerable intensity drop as observed in Figure S15b, Supporting Information.For 182 h illumination, although a little restoration was detected due to possible interaction between the perovskite and concentrated OA cations or concentrated iodine anions which can promote the formation of the preferred crystallinity of perovskite as the interaction of perovskite, [60] the portion of this beneficial impact for a preferred orientation is smaller than that of a severe decrease in preferred orientation under illumination.For the OAI/AlO xtreated perovskite, an increase in the intensity at  = 0°compared to that of the control was also detected before light exposure as shown in Figure S15b, Supporting Information.After 40 h illumination, the extent of decrease in the peak for OAI/AlO xtreated perovskite was less than that of OAI-treated perovskite in the bulk.This indicates that the extent of releasing OA cations in the bulk can be relatively reduced in the case of the OAI/AlO xtreated perovskite, unlike the OAI-treated perovskite, resulting in maintained vertical orientation of the perovskite.Furthermore, the considerable peak intensity at  = 0°for the OAI/AlO x -treated perovskite was recovered after 182 h of illumination compared to the OAI-deposited perovskite.
From these results, we can conclude the following regarding the effect of AlO x on the OAI-treated perovskite: (1) AlO x works as a template to prevent the loss of interaction between OAI materials and perovskite due to light soaking effects at the surface, (2) since AlO x has the role of protecting the layer from the removal of OA cations in the bulk, additional formation of 2D perovskite under illumination can be induced, (3) Remaining OA cations and diffused aluminum cations can synergize restoration of the perovskite crystallites orientation.
From the acquired results, the proposed light degradation mechanism could be summarized as shown in Figure 6.Lightinduced degradation mechanism of FAPbI 3 was reported [61] as the following.
where I o i , I + i , and I − i are neutralized iodine interstitial, positively charged iodide interstitial, and negatively charged interstitial iodine, respectively.e and h are photo-generated electrons and holes.FA + is formamidinium.V 2− pb and V + I are lead vacancies and iodine vacancies, respectively.
Mechanism (1) shows that neutral interstitial iodine could form I 2. Generated electrons and holes also react with iodine defects to neutralize them, producing I 2. These reactions accelerate the perovskite degradation explained by mechanism (3).
Figure 6a indicates the OAI-treated perovskite degradation under illumination.The mechanisms (1-3) also exists in the case of OAI treatment.However, OA cations and iodine anions could be diffused into bulk.In addition, under illumination, these diffused cations and anions react with photo-generated electrons and holes, respectively, resulting in the formation of volatile OA and I 2 as outlined in mechanisms (4)- (6).Even though restructuring of FAPbI 3 may occur partially confirmed by the above GI-WAXS results, mechanism (1) indicates that degradation of the perovskite is the major light-induced mechanism.
Mechanism ( 7) << Mechanism (1) .Figure 6b presents degradation under illumination for the AlO x on OAI-treated perovskite.Mechanisms arising from FAPbI 3 or OAI mentioned above are still progressing under the light.Interestingly, GIWAXS also confirmed distinct 2D formation in AlO x on the OAI-treated perovskite as expressed by mechanism (8).This light-induced 2D formation allows consuming OA cations and iodine anions.Consequently, this formation contributes to the suppression of volatile vapor generation and reduced ion migration that accelerates light-induced degradation.

Conclusion
We introduced AlO x by ALD on OAI-deposited perovskite (OAI/AlO x -treatment), showing not only enhanced device performance (higher V oc and FF) but also improved stability of devices.
To better understand the mechanism of OAI/AlO x -treatment which leads to improved device performance, surface analyses are carried out.It was unveiled that the formation of nonstoichiometric AlO x forms on OAI-deposited perovskite as well as uncoordinated Pb on the surface was passivated, confirmed by FTIR and XPS results.By using wavelength-dependent KPFM, it was found that homogeneous photo-induced carriers were generated at 750 (bulk) and 400 (surface) nm for the OAI/AlO x -treated sample, which may be attributed to diffused aluminum cations originating from non-stoichiometric AlO x formation.Next, the reason for the improved light stability of OAI/AlO x was investigated.It was found that transitions of morphology and crystallinity under illumination were observed by using XRD, SEM, and KPFM.Furthermore, the inhibition of the -phase perovskite formation was observed after the introduction of AlO x on OAItreated perovskite.GIWAXS measurements were implemented to further understand the light-induced crystallinity change under illumination.This showed not only suppression of the phase formation, but also resulted in light-induced 2D perovskite formation for the OAI/AlO x -treated sample, which may be ascribed to mitigated generation of volatile iodide and organics arising from excess iodine and OA cations.Hence, our suggested approach using AlO x on organic halide salts-treated perovskite can provide insights to improve both device performance and device operational stability.
Fabrication: As transparent electrodes, glass/FTO (fluorine-doped tin oxide on glass, Philkington) substrates were prepared by cleaning with a special detergent followed by ultra-sonication in deionized water, acetone, and isopropyl alcohol.After drying, a compact TiO 2 blocking layer (c-TiO 2 ) was first deposited onto precleaned a FTO substrate by spray pyrolysis, on a hotplate kept at 450 °C, using an airbrush.The solution used in the spray pyrolysis was 10 mL titanium diisopropoxide bis(acetylacetonate) 75 wt% in isopropanol with 100 mL ethanol.And then, mesoporous TiO 2 (mp-TiO 2 ) films were spin-coated onto the FTO/c-TiO 2 substrate at 500 rpm for 5 s, and 2500 rpm for 50 s using the paste.The films were calcined at 500 °C for 1 h to remove the organic part used as the binder.The perovskite solution was prepared by dissolving 800 mg of formamidinium lead iodide (FAPbI 3 ), and 40 mg of methylammonium chloride (MACl) in an N,N-dimethyl formamide/dimethyl sulfoxide (8:1 v/v) mixed solvent.The FAPbI 3 perovskite solutions were spin-coated onto the FTO/c-TiO2/mp-TiO 2 substrates at 500 rpm for 5 s, 1000 rpm for 8 s, and 5000 rpm for 12 s, and the ethylacetate in the final spin-stage was dripped onto the substrate during spin coating.After that, the substrates were dried on a hotplate at 150 °C for 10 min.Spiro-OMeTAD solutions were prepared in chlorobenzene (100 mg/1.1 mL) with additives of 23 μL of Li-TFSI in acetonitrile (540 mg mL −1 ), 39 μL of 4-tert-butylpyridine (t-BP), and 10 μL of tris(2-(1H-pyrazol-1-yl)−4-tert-butylpyridine)cobalt(III) tri[bis(trifluoromethane)sulfonimide] (Co(III) TFSI) in acetonitrile (376 mg mL −1 ).Spiro-OMeTAD solutions were dynamically spin-coated on FTO/c-TiO 2 /mp-TiO 2 /FAPbI 3 substrates at 3000 rpm for 30 s. Finally, a metal electrode consisting of Au (80 nm) with an area of 0.16 cm 2 was deposited by thermal evaporation in a vacuum for all devices.In terms of PV-SCLC measurements, hole-only devices were fabricated with the following configuration: Glass/Au/PEDOT:PSS/Perovskite with or without OAI; OAI-AlO x /Sprio-OMeTAD/Au.The deposition procedures for each layer were kept consistent with those used for solar cells, except for the specific device configuration.
Surface Treatment and ALD Process: In the case of OAI, 10 mm in chloroform solution was treated on the perovskite surface.OAI solutions were spin-coated onto the FTO/c-TiO 2 /mp-TiO 2 /FAPbI 3 substrate at 5000 rpm for 30 s. Substrates were post-annealed on a hotplate at 100 °C for 3 min.ALD AlO x was grown at 100 °C in a commercial ALD system (NCD, Lucida D100).The precursors used were TMA and deionized H 2 O for the aluminum and oxygen sources, respectively.Each ALD cycle consisted of a dose of TMA for 0.5 s, followed by N 2 purge of 15 s, followed by a dose of H 2 O for 0.5 s, and then N 2 purge of 15 s for four cycles.To determine the thickness of four cycles AlO x, 400 cycles AlO x on Si sample was measured by ellipsometer, showing 39.3 nm.By extrapolation, four cycles AlO x was ≈3.93 Å. Topography (Figure S16, Supporting Information) for four cycles AlO x by ALD is presented in the Supporting Information.
SEM: SEM images were obtained by using FEI NovaSEM 230, operated at 15 kV.
XRD: XRD patterns were carried out by using PANalytical Empyrean instrument with Cu K radiation ( = 1.5418Å) at 45 K and 40 mA.
GIWAXS measurements were implemented at beamline 8-ID-E [62] of the Advanced Photon Source, Argonne National Laboratory.The used photons of energy of 10.91 KeV were applied to the sample with the incident angle ( i ) of 0.4°for the exposure time of 10 s on a Pilatus 1M detector with pixel size 172 μm positioned 217 mm from the sample for GIWAXS.The beam size was 10 μm vertically and horizontally 200 μm.The Matlab package GIXSGUI [63] was used to apply corrections for the detector nonuniformity, X-ray polarization, and geometrical solid-angle considerations, as well as for subsequent data reduction.The sample and flight path were maintained under vacuum (10 −3 Torr) during the measurement.
XPS and UPS: UPS was implemented by using ESCALAB250Xi system (Termo Scientific) with He I (21.2 eV) source.XPS was performed by using the same instrument for UPS with Al K X-ray source.Acquired binding energies from XPS were calibrated by C 1s (284.8 eV).
KPFM: In order to investigate electrical properties with regard to samples (no light degraded), KPFM was conducted using AIST-NT SmartSPM 1000 in nitrogen conditions.The platinum tip with a 30 nm of radius (HQ:NSC35/PT) was used for KPFM measurement.To characterize ion migration behavior, tip bias was applied when KPFM measurements were carried out in the dark condition.
In the case of measuring SPV in KPFM, an external incident light source (FemtoPower1060, NKT photonics Inc.) with a pulse width of 200 fs, a high-repetition-rate of 80 MHz, and bandwidth of 25 nm was used.It was applied to the surface of the sample at an angle of 30°to avoid unexpected shading due to the probe.The wavelength of 400 and 750 nm with the intensity of 50 SUN was utilized to examine wavelength-dependent properties.
FTIR: FTIR spectra of fresh and degraded samples were obtained by using an FT-IR spectrometer (Spectrum Two, PerkinElmer) with attenuated total reflectance mode from 450 to 4000 cm −1 .
UV-vis Absorption Spectra: The absorbance, transmittance, and reflectance of samples whose configuration was the Glass/FTO/ETL/Perovskite were conducted by using a UV-vis spectrophotometer (Lamda, 1050, Perkin Elmer).
External Quantum Efficiency Measurement and Internal Quantum Efficiency: External quantum efficiency measurement (EQE) and internal quantum efficiency (IQE) measurements were performed with a quantum efficiency measurement tool (Newport, Model Quantx-300) at room temperature in air, where IQE = EQE/(1 − R).J-V Measurements: Photocurrent density versus voltage (J-V) characteristics were measured using a Keithley 2420 sourcemeter.The standard 100 mW cm −2 (1 SUN) illumination was generated by a Newport Oriel Class A 91195A solar simulator using a 450 W Xe-lamp (Oriel) with an AM 1.5 G filter, while the light intensity was calibrated by a Si-reference cell certified by NREL.The J-V curves were measured from 1.5 to −0.2 V along the reverse scan direction, with a step voltage and scan speed fixed at 10 mV and 150 mV s −1 , respectively.All devices were measured with a metal mask with an active area of 0.094 cm 2 .
MPPT Test for Light Stability: Operational stability tests were performed by maximum power point tracking (MPPT) using a Keithley 2420 source meter with a 450 W Xe-lamp (Oriel) using an AM 1.5 G filter.The light intensity was calibrated by a Si-reference cell certified by NREL.While the temperature was not controlled during light soaking, ambient temperature and humidity were maintained at 25 °C and 30% relative humidity, respectively.
Damp Heat Test (85 °C and 85% Relative Humidity): A climatic test was conducted in a chamber (C 4-340 E series, Votschtechnik), and was carried out in a chamber set to a constant temperature (85 °C) and constant humidity (85%).The efficiency of the PSCs was measured under illumination at AM 1.5G after removing the devices from the chamber and cooling them down to room temperature.
PV-SCLC: PV-SCLC was conducted instead of continuous voltage SCLC to mitigate the impact of ion re-distribution that makes interpretation difficult during SCLC.The hole trap density of each sample condition can be extracted from PV-SCLC measurements using the following equation. [64]t = 2V TFL  0 ed 2 (9) where V TFL is the trap-filled voltage, and d is the film thickness (≈600 nm). is the relative dielectric permittivity; used 47; [65]  0 is the vacuum permittivity (8.8542 × 10 −14 F cm −1 ); e is the elementary charge (1.602 × 10 −19 C).

Figure 1 .
Figure 1.a) Schematic illustration of the device structure (configuration: AlO x on the OAI-deposited perovskite).b) The illuminated J-V curves of the champion PSCs.c) The statistics of device parameters with different configurations.d) Maximum power point tracking measurements of the encapsulated control device and the PSC with OAI and OAI/AlO x , under illumination with an intensity of 100 mW cm −2 (1 SUN) in ambient conditions to determine the long-term photo-stability.

Figure 2 .
Figure 2. Fourier transform infrared spectroscopy spectra of a) in the wavenumber range of from 3500 to 1650 cm −1 and d) from 1600 to 600 cm −1 , respectively.X-ray photoelectron spectroscopy spectra: b) N 1s; c) Pb 4f; e) Al 2p; f) O 1s. g) Depth profile of positive elements in the control (top), the OAI (middle), and the OAI/AlO x (bottom) from time-of-flight secondary ion mass spectrometry.h) Schematic illustration of aluminum cations diffusion in the case of the OAI/AlO x -treated perovskite.

Figure 3 .
Figure 3. Surface photovoltage (SPV, CPDlight-CPDdark) measurement at 750 and 400 nm using Kelvin probe force microscopy.a) The full width at half maximum (FWHM) values of the SPV distribution extracted from Figure S7j, Supporting Information for the control, the OAI-deposited perovskite, and the OAI/AlO x -treated perovskite.b) The average values of the SPV distribution with different incident light wavelengths (750 and 400 nm) extracted from Figures S7d-f and S7g-i, Supporting Information, respectively.

Figure 4 .
Figure 4. Topography images of a) the control, c) the OAI-treated perovskite, and e) the OAI/AlO x -treated perovskite after 182 h of a white LED at a light intensity of 100 mW cm −2 (1 SUN).The scale bar is 2 μm.Work function distribution of b) the control, d) the OAI-treated perovskite, and f) the OAI/AlO x -treated perovskite extracted from CPD images in Figure S10, Supporting Information.

Figure 6 .
Figure 6.Proposed degradation mechanism as a function of light exposure time for a) OAI and b) OAI/AlO x -treated samples.Note that the dashed line indicates the boundary between the surface and the bulk of perovskite.