Modification of Hydrophobic Self‐Assembled Monolayers with Nanoparticles for Improved Wettability and Enhanced Carrier Lifetimes Over Large Areas in Perovskite Solar Cells

The development of perovskite solar cells (PSCs) with low recombination losses at low processing temperatures is an area of growing research interest as it enables compatibility with roll‐to‐roll processing on flexible substrates as well as with tandem solar cells. The inverted or p–i–n device architecture has emerged as the most promising PSC configuration due to the possibility of using low‐temperature processable organic hole‐transport layers and more recently, self‐assembled monolayers such as [4‐(3,6‐dimethyl‐9H‐carbazol‐9‐yl)butyl]phosphonic acid (Me‐4PACz). However, devices incorporating these interlayers suffer from poor wettability of the precursor leading to pin hole formation and poor device yield. Herein, the use of alumina nanoparticles (Al2O3 nanoparticles (NPs)) for pinning the perovskite precursor on Me‐4PACz is demonstrated, thereby improving the device yield. While similar wettability enhancements can also be achieved by using poly[(9,9‐bis(3′‐((N,N‐dimethyl)‐N‐ethylammonium)‐propyl)‐2,7‐fluorene)‐alt‐2,7‐(9,9‐dioctylfluorene)]dibromide (PFN‐Br), a widely employed surface modifier, the incorporation of Al2O3 NPs results in significantly enhanced Shockley–Read–Hall recombination lifetimes exceeding 3 μs, which is higher than those on films coated directly on Me‐4PACz and on PFN‐Br‐modified Me‐4PACz. This translates to a champion power conversion efficiency of 19.9% for PSCs fabricated on Me‐4PACz modified with Al2O3, which is a ≈20% improvement compared to the champion device fabricated on PFN‐Br‐modified Me‐4PACz.


Introduction
Achieving net zero carbon emissions has been identified not only as the route toward negating the detrimental impact of climate change, but also as the route for realizing a more sustainable future for humanity. [1] Of the renewable energies expected to contribute toward achieving this target, photovoltaics (PVs) is anticipated to play a major role. [2] Among the existing PV technologies, perovskite solar cells (PSCs) have emerged as a promising candidate for future large-scale deployment owing to their high efficiencies and low fabrication cost. [3] Rapid developments in interlayers, passivation strategies, and bandgap optimization have allowed the realization of a record power conversion efficiency (PCE) of 26.0% for single-junction devices and 33.7% for perovskite-based tandem solar cells (NREL Best Research-Cell Efficiency Chart).
A significant fraction of the record efficiency single junction PSCs has relied on the original n-i-p device architecture. However, there has been a steady growth in interest in the inverted p-i-n device architecture. This has been driven by several factors including the low processing temperatures generally used for the p-i-n architecture, which ensures its compatibility with roll-to-roll processing as well as its relevance for tandem architectures which has recently led to PCEs that exceed the single-junction radiative limit for silicon PVs. [4,5] The development of inverted device architectures has mainly relied on the use of organic hole-transport layers (HTLs) such as poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine (PTAA) and poly[(N,N 0 -bis-(4-butylphenyl)-N,N 0 -bis(phenyl)benzidine] (poly-TPD). [6][7][8] These organic HTLs have resulted in lower recombination losses in PSCs allowing high open-circuit voltages (V OC ) to be realized in inverted devices. [7,9] Recently, Albrecht and co-workers developed a class of self-assembled monolayers (SAMs) that eliminates the requirement of conventionally used transport layers at the anode contact in PSCs. [10] Among these, the SAMs 2-(3,6-dimethoxy-9H-carbazol-9-yl)ethyl]phosphonic acid (MeO-2PACz) and (2-(9H-carbazol-9-yl)ethyl)phosphonic acid (2PACz) have been widely adopted in place of conventional organic HTLs, yielding promising results in perovskite-based single-junction and tandem devices. [11][12][13][14][15][16][17] This was followed by the development of the methyl-substituted SAM, [4-(3,6-dimethyl-9H-carbazol-9-yl)butyl]phosphonic acid (Me-4PACz) ( Figure 1a). [11] Compared to the previously developed SAMs, Me-4PACz has its highest occupied molecular orbital (HOMO) more favorably positioned with respect to the valence band maximum (VBM) of the perovskite. This leads to a more efficient hole extraction at the hole selective contact (HSC)/ perovskite interface compared to MeO-2PACz, 2PACz and even PTAA, which translated to lower V OC losses and improved resistance to light-induced phase segregation. [11] A key challenge associated with the use of organic HTLs in PSCs has been the poor wettability of the perovskite precursor which leads to the formation of perovskite films with pinholes. This has, in the past, been addressed by solvent treating with dimethylformamide or by using poly [(9,9-bis(3 0 -((N,Ndimethyl)-N-ethylammonium)-propyl)-2,7-fluorene)-alt-2,7-(9,9dioctylfluorene)]dibromide (PFN-Br, Figure 1b) as an interfacial compatibilizer. [18][19][20] Another example of an interfacial compatibilizer used with hydrophobic charge transport layers is alumina (Al 2 O 3 ) nanoparticles (NPs), although this has not been widely adopted. [21][22][23][24][25] Similar challenges associated with the wettability of the perovskite have been reported for coating of perovskite precursors on Me-4PACz by several groups which have also been observed by us. [26][27][28] This has led to a lower number of reported studies based on the use of Me-4PACz on PSCs due to poor device reproducibility. One route toward improving the device yield on Me-4PACz is texturing of the underlying substrate as reported by Tockhorn et al. [28] While this approach is compatible with tandem architectures, there is a need for www.advancedsciencenews.com www.solar-rrl.com methodologies that can be implemented on single-junction PSCs.
Here, we introduce interfacial modification with randomly dispersed, solution processable Al 2 O 3 NPs as means of overcoming the wettability challenge of perovskite on the hydrophobic Me-4PACz over large areas (up to 4 cm 2 demonstrated). We show that the Al 2 O 3 NPs act as pinning sites for the perovskite precursor resulting in complete coverage of the perovskite absorber on the SAM. Furthermore, we show that modification with Al 2 O 3 NPs results in perovskites with Shockley-Read-Hall (SRH) recombination lifetimes exceeding 3 μs, which is higher than those on perovskites coated on bare Me-4PACz and on PFN-Br-modified Me-4PACz. We propose that this phenomenon is in accordance with the amphoteric nature of Al 2 O 3 which allows it to behave as both a Lewis acid and a Lewis base. PSCs fabricated on Me-4PACz modified with Al 2 O 3 NPs demonstrate a champion PCE of 19.9% which is a marked improvement in PCE as compared to 16.5% observed on devices with PFN-Br-modified Me-4PACz. We note that following the first submission of this manuscript, Al-Ashouri et al. have reported the addition of 1,6-hexylenediphosphonic acid as a second component into the Me-4PACz solution resulting in similar PCE for triple-cation perovskites reported herein. [29] Furthermore, Peng et al. have also reported the use of Al 2 O 3 NPs where reduced recombination was demonstrated resulting in a PCE of %25% when combined with a perovskite with a narrower bandgap of 1.55 eV. [30] 2. Results and Discussion

Optimization of Al 2 O 3 NP Coverage
We initially characterized the Al 2 O 3 NP dispersion to identify its structure and size distribution. A high-resolution transmission electron microscopy (TEM) image of a representative Al 2 O 3 NP is given in Figure 1c while the related selected area electron diffraction (SAED) pattern is given in Figure 1d in which the crystal planes have been indexed based on the γ-Al 2 O 3 phase. [31] Using dynamic light scattering measurements, we identify the average particle size to be 17.1 AE 3.9 nm ( Figure S1, Supporting Information). We then studied the impact of the NP concentration on the NP coverage on Me-4PACz ( Figure  S2, Supporting Information) and its effect on the PSC performance. While Lee et al. demonstrated that a mesoporous layer of Al 2 O 3 NPs results in a faster charge extraction compared to titania in regular PSCs, [32] the work of Stranks et al. demonstrated that the presence of Al 2 O 3 NPs can result in suboptimal performance. [33] From our initial optimization studies, we identified a 1:1000 dilution of the parent Al 2 O 3 NP dispersion as the optimum concentration for realizing a balance between improving device reproducibility and maintaining device efficiency (Figure 1e, S3, S4, Table S1, Supporting Information). Compact (i.e., pinhole free) perovskite films were achieved over large areas (demonstrated up to 4 cm 2 ), both when processed on Me-4PACz modified under this optimized Al 2 O 3 condition as well as on Me-4PACz modified with PFN-Br, whereas the absence of a surface modifier resulted in incomplete coverage ( Figure 1f ) of the perovskite absorber leading to a high number of shunted devices.

Precursor Retention and Surface Chemistry
To understand the impact of the different surface modifiers on the retention of the perovskite precursor on Me-4PACz, we carried out roll-off angle measurements, as conducted by Tockhorn et al. [28] In our measurements, we observed a roll-off angle of 30.9°for Me-4PACz on indium tin oxide (ITO) which is significantly improved to 37.2°when Me-4PACz is modified with PFN-Br. On the other hand, modification of Me-4PACz with Al 2 O 3 leads to a roll-off angle of only 31.9° (Figure 2a). Although this roll-off angle is not as high as that for PFN-Br, the low density of randomly positioned Al 2 O 3 NPs on the substrate surface is sufficient to pin the precursor on the SAM, thereby allowing a combination of high device yield and excellent optoelectronic properties. We further observe breaking up of the perovskite precursor into smaller droplets on bare Me-4PACz while the precursor is observed to maintain its cohesiveness and show better wettability on Me-4PACz modified with PFN-Br and Al 2 O 3 ( Figure 2b). The enhanced retention can be explained based on the mathematical model developed by Joanny and de Gennes which describes the influence of randomly dispersed particles (or contaminants) on the surface. [34,35] However, we do note that perovskite precursors with their multicomponent nature, in combination with the complex surface chemistry of the underlying substrate, will require the development of more detailed models.
Following the above, we proceeded to understand the surface chemistry of the modified substrates using X-ray photoelectron spectroscopy (XPS) measurements. In the P 2p and N 1s spectra given in Figure S5, Supporting Information, the peaks appearing at binding energies of 133.9 and 400 eV correspond to the P in phosphate group and N in C─N bonds, respectively, of Me-4PACz, [36] confirming the presence of the SAM on the ITO. Further, In 3d 3/2 and In 3d 5/2 signals characteristic to ITO at binding energies of %452 and %445 eV, respectively, can be observed in the In 3d spectra of the Me-4PACz-coated substrates ( Figure S6a, Supporting Information). [37] A probing depth of %5 nm was used in these measurements. Based on the appearance of peaks specific to ITO in SAM-coated substrates at such small probing depths, it is reasonable to assume that a few nm-thick layer of Me-4PACz, i.e., a monolayer, is formed here. The O 1s and In 3d spectra depicted in Figure 2c and S6a, Supporting Information, respectively, show a slight, but noticeable shift in the signals toward higher binding energies in substrates coated with Me-4PACz (in comparison to bare ITO). This is consistent with observations reported in the literature where it has been correlated to a change in the work function due to the introduction of a dipole by the SAM. [37] To evaluate the effect of the Me-4PACz modification on surface roughness and work function (Φ) of ITO, we carried out atomic force microscopy (AFM) and scanning Kelvin probe force microscopy (SKPFM). As shown in Figure 2d and Table 1, Me-4PACz causes the root mean square roughness (R q ) of ITO to increase from %3.6 to %6.6 nm while modification with Al 2 O 3 causes R q to further increase to %17.7 nm. The latter indicates the presence of Al 2 O 3 NPs on Me-4PACz. Figure 2e shows the corresponding contact potential difference (CPD) mapping of the samples, with the Φ values given in Table 1. It is apparent that, while the work function of ITO increases from %4.7 to %5.3 eV when coated with Me-4PACz, it is not altered further upon modification with Al 2 O 3 . Furthermore, it can be inferred from the homogeneity of the CPD map that there is uniform and homogeneous spatial distribution of Me-4PACz on the  substrates, which is favorable toward achieving efficient charge extraction. [38]

Surface and Bulk Structure
To gain an insight into the influence of the Me-4PACz surface modification on the crystallization and the morphology of the perovskite, we studied the films using scanning electron microscopy (SEM) and grazing incidence X-ray diffractometry (GIXRD). Analysis of the scanning electron micrographs (Figure 3a-c) for grain sizes (Figure 3d-f ) shows that perovskite grains exceeding 240 nm in diameter are formed on modified Me-4PACz as compared to 196 nm on bare Me-4PACz. The formation of large grains is considered as favorable for device performance as it reduces the trap density at grain boundaries which act as nonradiative recombination sites for charges carriers. [39] However, we note that these grain sizes may not extend through the film thickness and can vary depending on the nature of the substrate leading to differences in carrier lifetimes as observed in timeresolved photoluminescence (TRPL) measurements which will be discussed later. GIXRD patterns of perovskites formed directly on Me-4PACz (as a reference) and on modified Me-4PACz are depicted in Figure 3g. One of the most notable differences between the spectra is the peak at 30.3°which only appears in the spectra of ITO/Me-4PACz. This peak, which is commonly attributed to ITO, [40] confirms the presence of pinholes/areas not covered by the perovskite when coated directly on Me-4PACz.
The absence of such an intense peak for ITO in the spectra for modified Me-4PACz substrates indicates the achievement of an improved coverage of the perovskite by modification using both PFN-Br and Al 2 O 3 . It is further noted that the peak at 12.7°c orresponding to PbI 2 appears more prominently on ITO/Me-4PACz/PFN-Br and ITO/Me-4PACz/Al 2 O 3 as compared to that www.advancedsciencenews.com www.solar-rrl.com on ITO/Me-4PACz. [41] This indicates the presence of excess PbI 2 in perovskite films formed on modified Me-4PACz. Excess PbI 2 is often incorporated in high-performing PSCs as it is reported to have beneficial effects on grain boundary trap passivation which is also expected to be a contributing factor toward the realization of longer SRH recombination lifetimes in perovskites formed on Me-4PACz/Al 2 O 3 as compared to those on bare Me-4PACz. [42] 2.4. Device Performance To examine the effect of this Al 2 O 3 -based HSC/perovskite interfacial modification on device performance, we fabricated inverted planar PSCs with a configuration of ITO/Me-4PACz/ Cs 0.05 MA 0.15 FA 0.8 Pb(I 0.85 Br 0.15 ) 3 /C60/bathocuproine (BCP)/Ag (Figure 4a). It should be noted here that, as discussed initially, the dewetting of perovskite on Me-4PACz was so severe that it was not possible to achieve working devices on Me-4PACz without any modification. Figure Figure 4c where an improved EQE is observed for the Al 2 O 3 device in the 400-700 nm wavelength range which explains its higher short circuit photocurrent density ( J SC , 23.0 mA cm À2 ) compared to that of the PFN-Br device (20.5 mA cm À2 ). This higher J SC is in agreement with the improved charge carrier lifetimes observed on Al 2 O 3modified Me-4PACz in comparison to those on PFN-Br-modified Me-4PACz in TRPL measurements shown in Figure 5a. We also tracked any changes in the current density of the champion devices at their maximum power point (MPP) for 180 s (Figure 4d). Here, a satisfactorily stable current was observed in both devices during the testing period indicative of sufficient device stability. The statistical distribution of the photovoltaic parameters for Al 2 O 3 devices (N = 33) is given in Figure 4e-h with their average values summarized in Table S2, Supporting information (the corresponding statistical distributions for PFN-Br-modified devices are given in Figure S7, Supporting Information). Across all device parameters, Al 2 O 3 -modified devices show significantly higher average performances as compared to PFN-Br-modified devices. Figure 4i,j depicts the reflectance and EQE spectra as well as photocurrent density losses due to reflectance and electronic effects in champion devices fabricated on Me-4PACz modified with PFN-Br and Al 2 O 3 , respectively. Here, while there seem to be similar losses due to reflectance in the two devices, electronic losses are noticeably lower in the Al 2 O 3 device (%2.0 mA cm À2 ) as compared to the PFN-Br device (%2.7 mA cm À2 ). The higher losses in the PFN-Br-based devices are attributed to the higher nonradiative losses as evidenced through optical and electrical characterization which will be discussed later. We note a similar trend for MAPbI 3 -based PSCs when fabricated on Al 2 O 3 NP-modified Me-4PACz as compared to devices fabricated on PFN-Br-modified Me-4PACz ( Figure S8-S10, and Table S3, Supporting Information).

Optoelectronic Origins of the Observed Performance Improvements
To identify any influence of the surface treatment of Me-4PACz on the recombination characteristics, we conducted both steadystate PL and TRPL measurements on perovskite films coated on Me-4PACz, Me-4PACz/PFN-Br, and Me-4PACz/Al 2 O 3 . TRPL measurements showed a shorter SRH recombination lifetime of %1 μs for perovskites coated on Me-4PACz/PFN-Br ( Figure 5a and Within the field of perovskite optoelectronics, Lewis acids and Lewis bases are well known and widely adopted for passivation of defects in the perovskite absorber. [43] Al 2 O 3 , due to its amphoteric nature, can behave as both a Lewis acid and a Lewis base which enables it to cause defect passivation of perovskites. A typical reaction scheme which allows this to happen is given in Figure 5b(i). [44] During thermal treatment of Al 2 O 3 , any hydroxy (OH) groups bonded to the surface are detached. When this occurs at two neighboring sites, a strained oxygen bridge is formed resulting in a local Lewis acid and Lewis base behavior as shown. [45] In the examples depicted in Figure 5b(ii) and (iii), Pb-I antisite and halide defects are passivated by Lewis acids due to their ability to accept electrons from negatively charged sites while defects such as halide vacancies or unsaturated Pb 2þ defects are passivated by Lewis bases due to their ability to donate electrons to defect sites. [46] Based on the above, we propose that the observed improvements in the SRH recombination lifetimes for perovskites coated on Me-4PACz are due to the passivation of defects at HSC/perovskite interface, facilitated by the amphoteric nature of Al 2 O 3 . We note that the nature of the substrate, i.e., Me-4PACz, Me-4PACz/PFN-Br, or Me-4PACz/Al 2 O 3 , does not affect the bandgap of the perovskite absorber as is evident through the steady-state PL spectra of the films depicted in Figure 5c where no noticeable differences in the peak position or the shape can be observed. Further, the bandgap estimated here is in agreement with the values reported in the literature for this composition. [41] Following the characterization and analysis of the PL characteristics, the V OC behavior of the devices with varying illumination intensity was examined to understand the influence of the hole extraction contact modification on the electronic characteristics of the devices. The diode ideality factors for PSCs based on Me-4PACz/PFN-Br and Me-4PACz/Al 2 O 3 were estimated using the following relationship [47] n ¼ q kT Here, n, k, q, T, and ϕ represent the diode (light) ideality factor, Boltzmann constant, electronic charge, temperature, and light intensity, respectively. Based on a linear fit to a semilogarithmic plot of V OC versus logðϕÞ (Figure 5d  www.advancedsciencenews.com www.solar-rrl.com bimolecular recombination under short-circuit conditions. We also analyzed the electrical characteristics of hole-only devices based on ITO/Me-4PACz/PFN-Br or Al 2 O 3 /perovskite/ Spiro-OMeTAD/Au. Plotting logðJÞ % logðVÞ, we observe three different regions: an Ohmic region characterized by a slope of 1 (or J ∝ V) followed by a rapid rise indicative of a J ∝ V m type behavior ðm > 2Þ and thirdly, the emergence of a quadratic behavior ðJ ∝ V 2 Þ indicative of the Child's regime for both device types (Figure 5e,f ). We observe a reduction in the voltage for the trap-filled limit ðV TFL Þ when Me-4PACz is modified with Al 2 O 3  www.advancedsciencenews.com www.solar-rrl.com NPs compared to modification with PFN-Br. We estimated the trap density ðN t Þ for the two systems using the following relationship [48] Here, ε, ε o , and L represent the relative permittivity (taken as 62), [48] vacuum permittivity, and thickness of the perovskite absorber (%500 nm, Table S5, Supporting Information), respectively. We note here that we did not observe a change in the perovskite thickness irrespective of whether PFN-Br or Al 2 O 3 NPs were used to modify the Me-4PACz surface. Based on the above values, we estimate trap densities of %5.7 Â 10 15 and %3.3 Â 10 15 cm À3 when Me-4PACz is modified with PFN-Br and Al 2 O 3 NPs, respectively, in agreement with our previous observations in TRPL and V OC versus logðϕÞ measurements. This reduction in trap density for perovskites on Me-4PACz modified with Al 2 O 3 NPs in comparison to perovskites on Me-4PACz modified with PFN-Br leads to the higher J SC observed for the former type of devices (as evident in Figure 5).
Finally, we evaluated the influence of the Me-4PACz surface modification on the stability of these devices under ISOS-D-2 and ISOS-D-2-i conditions (i.e., stored under open-circuit conditions in the dark at a temperature of 65°C in ambient and N 2 environment, respectively). [49] Within the duration studied, the PFN-Brmodified devices show a higher degradation in device performance in comparison to Al 2 O 3 NP-modified devices ( Figure S12, Supporting Information). In particular, the PFN-Br-modified devices demonstrate more than 20% drop from the initial PCE, while the Al 2 O 3 NP-modified devices retained more than 90% of the initial PCE under ambient conditions. These preliminary results indicate the promise of Al 2 O 3 NP modification as a strategy toward realizing stable inverted PSCs based on hydrophobic SAMs.

Conclusion
In conclusion, we have demonstrated that the incorporation of Al 2 O 3 NPs at the HSC/perovskite interface is an effective strategy toward achieving better, uniform coverage of perovskites over large areas in Me-4PACz-based inverted PSCs. In comparison to other more commonly used interfacial compatibilizers such as PFN-Br, the use of Al 2 O 3 NPs results in the realization of significantly longer carrier lifetimes, even exceeding those of perovskites coated directly on Me-4PACz, mediated by a reduction in trap density. This, in turn, leads to higher device efficiencies approaching 20%. The use of such insulating NPs as pinning agents for the perovskite precursor is anticipated to not only be beneficial on Me-4PACz, but also on other organic transport layers such as PTAA and poly-TPD allowing the realization of higher efficiencies on many different optoelectronic systems that use perovskites as its active layer.
Cesium iodide (CsI, 99.999%) and aluminum oxide (Al 2 O 3 , 20% v/v in 2-propanol) were purchased from Sigma-Aldrich (UK). Lead(II) iodide (PbI2, 99.99%) and Lead(II) bromide (PbBr2, >98.0%) were purchased from Tokyo Chemical Industry Co., Ltd. (TCI, Japan). Poly [(9,9- Device Fabrication: Spin coating of Me-4PACz on the substrates was carried out immediately following UV-O 3 treatment to ensure adhesion of the SAM to the substrate. 50 μL of the solution was spread on the substrate and after a waiting time of 5 s, it was spun at 3000 rpm for 30 s. The films were then annealed at 100°C for 10 min. For PFN-Br-based modification, 30 μL of the PFN-Br solution was dispensed on a substrate spinning at 5000 rpm, at 5 s from the spin start time while the overall spin coating duration was kept at 30 s. For Al 2 O 3 -based modification, 40 μL of the solution was deposited by spin coating at 4000 rpm for 30 s. The spin coating of perovskite was carried out immediately following the surface modification of Me-4PACz using PFN-Br or Al 2 O 3 . For MAPbI 3 -coated substrates, 35 μL of the precursor was spread on the substrate followed by spinning at 4000 rpm for 30 s. 600 μL of diethylether, the antisolvent, was dropped at 7 s from the spin start time. CsFAMA perovskite precursor was spin-coated using an antisolvent-assisted method and a two-step spin program. 50 μL of the perovskite precursor was spread over the substrate prior to starting the spinning process. The substrates were first spun at 2000 rpm for 10 s and then at 3500 rpm for 30 s. At 10 s from the end of the spin program, 120 μL of CB was dispensed on the substrate as the antisolvent. The substrates were then transferred to a hot plate at 100°C and annealed for 30 min. C60 (20 nm), BCP (7 nm) and Ag (100 nm) were thermally evaporated in an Angstrom EvoVac system. For devices fabricated for stability testing, Cu (100 nm) was used in place of Ag. For deposition of Cu, the samples deposited with C60 and BCP were taken out of the glove box and loaded to an evaporator (Moorfield) placed outside the glove box.
Current (I)-Voltage (V) Characteristics: I-V characteristics of the fabricated solar cells were evaluated using an Enlitech SS-F5-3A (Class 3A) solar simulator with a Keysight 2901A source measure unit acting as www.advancedsciencenews.com www.solar-rrl.com the electrical load. The calibration of the simulator was carried out using a KG-5 filtered Si diode. A mask with 0.09 cm 2 aperture area was used to define the active area of the device. All devices were measured without any encapsulation under ambient conditions at a temperature of %25°C, light intensity of 100 mW cm À2 (AM1.5G), and a relative humidity of 30-35%. No preconditioning of the cells was carried out. External and Internal Quantum Efficiency (EQE, IQE): EQE measurements of the fabricated devices were carried out using a Bentham PVE300 system. All measurements were carried out under ambient conditions. Reflectance measurements were carried out using an integrating sphere incorporated to the same system. All devices were measured without any encapsulation under ambient conditions at a temperature of %25°C and a relative humidity of 30-35%.
Stability Testing: For stability testing, devices were encapsulated with a proprietary encapsulant developed in-house. For ISOS-D-2 testing, samples were stored at 65°C in the dark under ambient conditions with a relative humidity of %35%. For ISOS-D-2-i testing, samples were stored at 65°C in the dark in a N 2 glove box.
SEM: SEM images of the perovskite films were obtained using a TESCAN FERA3 dual beam/focused ion beam SEM under an accelerating voltage of 5 kV. Grain size analysis was carried out using ImageJ software.
GIXRD: GIXRD measurements were taken using a Panalytical X'pert Pro diffractometer using a GI stage with an incident angle of 1°using a Cu Kα 1 X-ray source driven at 45 kV.
XPS: XPS spectra were obtained on a Thermo Fisher Scientific Instruments K-Alphaþ spectrometer consisting of a monochromated Al Kα X-ray source (hν = 1486.6 eV) with a spot size of %400 μm radius. A pass energy of 200 eV was used for acquisition of the survey spectra. For obtaining high resolution core level spectra, a pass energy of 50 eV was used for all elements. For correction of possible charging effects that can occur during acquisition, the obtained spectra were charge referenced against the C 1s peak (285 eV). Fitting of the spectra was carried out using the manufacturers' Avantage software.
AFM: A commercial AFM system (Park Systems, NX10) was used for the KPFM measurements. Nitrogen gas was used for the ambient condition during the measurements and for cleaning the surface of the samples before the measurements. KPFM images were taken on the sample surface using a platinum-coated conductive probe (NSC15/Pt, force constant k = 40 N m À1 ). For calibration, the work function of highly oriented pyrolytic graphite (HOPG) was measured using the same probe. During the measurements, DC bias was applied toward the tip. The KPFM images were taken with a scan direction from bottom to top and a scan rate of 0.8 Hz.
TEM and SAED: The Al 2 O 3 NPs were diluted in 2-propanol and sonicated for 30 min. A 10 μL volume from the dispersed solution was dropped onto a Holey carbon support film and left to dry naturally. The structural analysis was carried out in a field emission transmission electron microscope (TEM) model Talos F200i, Thermo Scientific, using a 200 keV beam. Data analysis was done using the ImageJ software.
Steady-State and TRPL Spectroscopy: TRPL measurements were carried out using a Picoquant FT300 spectrophotometer. A 640 nm excitation source was used for both steady-state and TRPL measurements.
Roll-Off Angle Measurements: Roll-off angle (θ RA ) measurements were carried out using a Krüss DSA 100 drop shape analyzer equipped with tilt table. A single 10 μL droplet was deposited onto the test substrate (at a tilt angle of 0°) using the equipped dosing module and a Hamilton 1000 gastight syringe equipped with a 0.97 mm inner diameter dosing needle supplied by Weller. The substrate was tilted at a rate of 5°min À1 and the position of the deposited droplet was recorded as a function of time. θ RA was determined for a given substrate-liquid system as the tilt angle corresponding to the initial movement of the three-phase point (3PP).
Thickness Measurements: Thickness measurements were carried out using a DektakXT profilometer.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.