Green Solvent Ethanol‐Based Inks for Industrially Applicable Deposition of High‐Quality Perovskite Films for Optoelectronic Device Applications

Incontrovertibly there is an increasing demand for the development of benign inks suitable for fabrication of high‐performing perovskite‐based thin film functional layers. Nevertheless, most reported perovskite precursors rely on the use of highly toxic solvents such as acetonitrile, 2‐methoxyethanol, dimethylformamide, and many others. Hence, there is a strong imperative for the development of novel and greener inks, which will facilitate smoother commercialization of technologies based on functional perovskite films. Therefore, four perovskite precursors are studied, some of which consist of up to 90% ethanol. All inks are developed to fulfill the requirements of a high‐throughput deposition compatible with roll‐to‐roll techniques at room temperature, assisted by an air knife for instant solvent removal. Two of the inks are particularly suitable for the fabrication of high‐quality and densely packed multi‐crystalline (CH3NH3)PbI3 layers, as confirmed by numerous nanoscale spectroscopic and material characterization techniques. Additionally, large‐area photoluminescence (PL) imaging is demonstrated to improve the quality of the deposited perovskite films, with a route to enhance deposition uniformity when upscaling for manufacture. The genuine potential of the developed greener perovskite inks is demonstrated with the fabrication of solar cells with power conversion efficiencies above 19.5%.


Introduction
Novel thin-film photovoltaic (PV) technologies have been developed during the last decades with a focus on improving efficiency, DOI: 10.1002/smtd.202300564stability, and sustainability while reducing the carbon footprint and lowering the cost of generated electricity. [1]A great example of such an emerging PV technology is the perovskite-based solar cells (PSCs), which can potentially reduce the watt-peak (Wp) price of generated electricity even further than silicon. [2]This is ascribed to a tremendous increase in efficiency of PSCs (26.0% and 33.7% for single-and multi-junction cells in June 2023, respectively) and due to the possibility of significantly lower production costs compared to traditional inorganic PVs. [3]mong many available perovskite materials, lead halide-based semiconducting perovskites show outstanding optoelectronic properties such as high light absorption coefficient, long carrier diffusion lengths, and a tuneable bandgap. [4]uch materials have extensively been used for the fabrication of small-area PSCs, which are commonly prepared by the spin-coating deposition technique.Although high power conversion efficiencies (PCEs) can be achieved with the spin-coating process, this deposition technique is not scalable and leads to a large volume of material waste.Besides, perovskite precursors developed for spin-coating often exhibit rheology characteristics that are not suitable for highthroughput manufacturing with roll-to-roll (R2R) compatible deposition techniques (blade-or slot-die coating, spray coating, ink-jet printing), which require perovskite precursors with tailored properties designed for large-area thin film deposition. [5]ence, there is demand for the development of novel recipes for perovskite-based coatings, which can be valuable to different technologies such as solar PV, light-emitting diodes (LEDs), lasers, and photodetectors. [6]or spin-coated devices, DMF, dimethyl sulfoxide (DMSO), butyrolactone (GBL), or N-N-methyl-2-pyrrolidone (NMP) are commonly used as a main solvent agent.On the other hand, ACN, 2-ME, or tetrahydrofuran (THF) are often employed for large-area deposition techniques such as bar-, blade-, slot-die-, or spray coating. [7]The latter solvents exhibit low boiling points and high vapor pressures but low polarities, which impedes the easy dissolution of materials used for perovskite precursors, particularly lead salts.Therefore, methylamine (MA) gas blowing has been reported to successfully dissolve lead-based perovskite precursors in solvents with low polarity. [8]This originates from the fact that solid MAPbI 3 can be dissolved in MA gas through the interactions between the electron-donating nitrogen group in MA and the cations of Pb 2+ and MA + in MAPbI 3 forming complexes based on Lewis acid-base interactions (MAPbI 3 ⦁xMA).The obtained cation-Lewis base adduct causes a solid-to-liquid phase transition and the dissolution of MAPbI 3 in gas.This phenomenon was applied for the preparation of inks based on pure ACN which were then used for the successful fabrication of largearea perovskite thin films. [9]9b] This is also true when such a liquid (MAPbI 3 ⦁MA adduct) is either dissolved or formed in solvents such as ACN. [10]9a] Although many high-performance perovskite layers fabricated by scalable deposition techniques have been reported already, highly toxic solvents were used in most cases. [11]Only a limited number of studies address the need for reduction or substitution of toxic solvents with less toxic alternatives. [12]Priya et al. demonstrated the use of a MAPb(I 1-x Cl x )-based perovskite ink in a mixed solvent system consisting of ACN as the main toxic solvent and EtOH.The perovskite ink was used primarily for the fabrication of spin-coated PSCs, although a simplistic small-area bar coating with a pipette tip was also shown. [13]Recently, Seok et al. developed a FAPbI 3 -based ink using EtOH as the main green solvent in combination with DMA (2:1 in volume ratio), which allowed for the fabrication of small-area spin-coated PSCs showing remarkable efficiencies above 25%. [14]Clearly, there is a lack of studies on using low-toxicity perovskite precursors for the deposition of large-area coatings by means of using scalable deposition methods viable for the commercialization of the PSCs technology.
In this study, we present a new methodology for the development and preparation of MA-mediated green inks suitable for large-scale preparation of MAPbI 3 films.We simplify the ink preparation process by using a MA solution in EtOH for the dissolution of the MAI and PbI 2 perovskite precursors instead of the MA gas-blowing method.We prepared four perovskite precursors by substituting the toxic solvent ACN with EtOH by 30%, 50%, 70%, or 90% related to the total volume of the solvents in the ink (hereafter ink E30, E50, E70, or E90, respectively) and demonstrate the preparation of perovskite films from all inks by using the scalable deposition technique bar-coating.Based on our extensive preliminary characterization of the developed inks, we shortlist inks E50 and E70 for the demonstration of a simple, reproducible, and environmentally friendly route for the deposition of high-quality perovskite films at room temperature.The spatial uniformity, defect density, and deposition reproducibility of these perovskite layers were furthermore characterized with large-area photoluminescence (PL) and electroluminescence (EL) imaging techniques.Finally, the perovskite precursor ink E50 was chosen and used for the fabrication of PSCs showing PCEs peaking at 19.5%.

Optimisation of the Bar-Coating Deposition Process
We chose the air knife-assisted bar-coating method for the largearea deposition and screening of our perovskite inks due to its suitability for transfer to an R2R-compatible deposition technique such as slot-die coating.A schematic representation of the coating set-up is shown in Figure 1a and the working principle is explained in the supporting information.To achieve reproducible film characteristics over large coating areas, it is essential that the deposited layers show pinhole-free coverage of the substrate with large perovskite grain sizes, comparable crystallinity, and uniform thickness across the complete length of the deposited film.This can be achieved by designing a precursor solution with the correct rheology, that is tailored to the deposition process and can be further tuned by careful optimization of the coating and drying process parameters.Although various parameters such as the precursor ink concentration and volume, the surface tension, coating bar diameter, coating gap and velocity, and curvature radius of the downstream meniscus are needed for calculating the resulting dry layer thickness after deposition, it is possible to present the layer thickness as a function of the coating velocity solely. [15]Therefore, under the assumption that none of the parameters mentioned above change, the deposited layer thickness will be constant if the coating velocity does not change.However, as shown in Figure 1b for ink E50, we observed a thickness gradient of perovskite-based films when deposited at a constant coating velocity of 12 mm s −1 .As visualized in Figure 1b, the highest layer thickness was observed at the start of the coating (424.3 ± 7.4 nm) and the thickness gradually decreased to its lowest value at the end of the coating (338.7 ± 8.1 nm) where the perovskite ink has almost been completely used for the deposition of the wet film.This is ascribed to the slow reduction of the ink volume captured in the coating gap between the bar and the substrate and the resulting reduction of the curvature radius of the downstream coating meniscus.To overcome this drawback and improve the coating homogeneity, we varied the coating velocity from 12 to 14 mm s −1 in 0.5 mm s −1 steps after every 10 mm, which resulted in a significant improvement of the thickness homogeneity, as shown in Figure 1c.
As described in a previous publication from our group, a nucleation-driven (ND) film formation mechanism for perovskite layer is only possible if a rapid solvent removal during the drying process is carried out.In particular, a high-quality smooth perovskite thin layer with a densely packed crystal grain structure and no voids or holes between them is achieved by inducing the nucleation step with a high number of nuclei at the start of the drying step, followed by a controlled crystal growth process. [16]9a] Due to the low boiling points and high volatile pressure of the solvent system of most inks developed in this study, the solvent removal has to be carried out shortly after the deposition of the perovskite precursor (app.3 s for ink E30; 5 s for inks E50 and E70; and 6-7 s for ink E90).In contrast, an excessive and unnecessary delay in gas blowing  c) Thickness mapping of the deposited dry perovskite films on 50 mm × 50 mm substrates with a constant or variable coating velocity.d-i) PL study of the delay time "t" and pressure "P" of the gas-blowing step after the deposition of MAPbI 3 thin films from ink E50.The gas-blowing conditions are shown in the PL images.j-k) Comparison of the PL spectra of ink E50 and ink E70, which were dried at optimized conditions.All samples were deposited using identical coating parameters.7e] Our observations show that this happens if the time delay is above 5 seconds for all inks apart from ink E90 (time delay of 6-7 s).In the extreme case of a longer delay time (above 10 seconds), the nucleation process has been completed before the gas blowing has started.This leads to the formation of non-uniform, highly rough, and low-quality perovskite thin films.On the contrary, a too-early gas blowing can induce supersaturation and nucleation of perovskite crystals at the film surface, which can lead to trapped solvents underneath and create voids and pinholes in the layer.Hence, we used large-area PL imaging to study the spatial uniformity and deposition reproducibility of the developed perovskite inks by fabricating samples with different air knife delay times and blowing pressures from ink E50. [17]As shown in Figure 1d-i, all deposited layers exhibit high uniformity.A closer look at the films dried immediately after deposition (t = 0) indicates the formation of wrinkles parallel to the coating direction.We speculate that these small coating imperfections are caused by the immediate solvent removal step, which does not allow for the deposited wet film to level out on top of the substrate surface after the coating process.As evident from Figure 1d,f,h, the wrinkle formation is independent of the blowing pressure.In contrast, the films deposited with a delay time of 5 seconds (Figure 1e,g,i,) are wrinkles-free and show very high spatial uniformity regardless of the blowing pressure used.In order to avoid the potential formation of pinholes or voids in the film due to trapped solvents and to stay in the safe operational window of our gas-blowing process, we decided to use a delay time of 5 seconds and a blowing pressure of 2 bar for the direct comparison between ink E50 and ink E70.It is evident from the high film uniformity shown in Figure 1j,k, that the developed deposition process is suitable for both inks.Besides the characterization of the spatial film uniformity, the large-area PL mapping can also be used to study the deposition reproducibility of the fabricated films or completed working devices. [18]It can be seen in Figure S1 (Supporting Information) that our ink E50 is suitable for batchto-batch fabrication of samples with comparable properties.

Characterisation of Perovskite-Based Layers
The quality of the prepared perovskite layers was thoroughly investigated using nanoscale characterization methods.Scanning electron microscopy (SEM) was performed on the MAPbI 3 perovskite films coated on glass/ITO substrates, to carefully study the crystal grain size and packing density.According to the topography, SEM scans in Figure 2a,b, densely packed crystal grains with similar grain sizes were observed for ink E50 and ink E70, respectively.This was achieved by tailoring the ink rheology, incorporating EtOH as the solvent with lower vapor pressure compared to ACN, and fine-tuning the fabrication conditions (bar coating parameters and the gas blowing step).Moreover, a precisely timed rapid solvent evaporation, triggering a suitable nucleation step, led to the formation of homogenously distributed single crystal grains with no visible presence of voids throughout the thickness of the MAPbI 3 film.This observation is confirmed by the cross-sectional SEM images in Figure S2a,b (Supporting Information) for ink E50 and E70, respectively.These findings are in good agreement with large-area PL mapping results for the samples based on these inks.We also studied perovskite films obtained from ink E30 and E90.As shown in Figure S2c,d (Supporting Information), similar characteristics were observed for ink E30 and E90, although the MAPbI 3 crystals grain sizes appear much smaller compared to those of ink E50 and E70.7e] Based on the high toxicity of ink E30 and the lowquality perovskite films obtained from ink E90, only ink E50 and E70 were shortlisted for further characterization and potential use for the fabrication of PSCs.Atomic force microscopy (AFM) technique was used to study the surface topography of the barcoated MAPbI 3 layers obtained from ink E50 (Figure 2c) and ink E70 (Figure 2d).The smaller grain sizes observed for ink E70 compared to ink E50 are in line with the SEM results shown in Figure 2a,b, which also correspond to the slightly reduced average RMS surface roughness measured for ink E70 compared to ink E50 (13.9 nm vs 16.2 nm, respectively).This low surface roughness is beneficial and necessary for achieving intimate contact at the interface between the perovskite layer and any subsequentially deposited layer on top in order to reduce the contact resistance at the interface crucial for device optimization and operations.We also studied the topography of MAPbI 3 films coated on top of glass/ITO/Cu:NiO x as the quality of the perovskite films depends on the surface quality and energy of the underlaying layer.The choice of Cu:NiO x as the hole transport layer (HTL) in our p-i-n PSCs is based on previous studies. [19]According to the images presented in Figure S3 (Supporting Information), a significant decrease in surface roughness was observed for the films from both inks, in comparison to the films coated on bare glass/ITO.Similarly, to the results presented in Figure 2c,d, ink E70 exhibited a lower RMS value compared to ink E50 (9.6 nm vs 6.0 nm, respectively).
The developed inks and fabrication procedures were furthermore employed for the deposition of perovskite layers on flexible PET/ITO substrates.The bar-coated PET/ITO/MAPbI 3 samples displayed smooth and mirror-like surfaces, which are shown in Figure 2e,f.This points out that our room-temperature coating process is well-suited for the fabrication of devices on flexible substrates by delivering the same high-quality MAPbI 3 perovskite layers.The quality of the obtained films was further investigated using SEM (see Figure 2g,h).The results indicate that regardless of the rigidity of the substrate, the proposed materials and methods can successfully provide densely packed crystallized perovskite thin layers, devoid of defects or voids and pinholes.
The best possible performance of a perovskite thin film employed in optoelectronic devices can only be ensured by obtaining a superior purity with high crystallinity for the perovskite material.Any impurity can result in forming trap states in the bulk, surface, and grain boundaries of the perovskite layer.This eventually will lead to recombination of photo-generated charge carriers, lowering the overall efficiency of the perovskite device.Using Xray diffraction (XRD), we investigated the purity and crystallinity of our developed perovskite thin films, based on both ink E50 and ink E70.The results strongly confirm the high purity of the films with the characteristic peaks of MAPbI 3 perovskite.9a] The position of the peak, along with their corresponding Miller indices are represented in Figure 3a,b.High-intensity XRD peaks indicate the superior crystallinity of MAPbI 3 film with no undesirable secondary phases within the film.The XRD measurements were performed on both samples at four different points of the film coated on a 5 cm × 5 cm glass substrate to further verify the full coverage of the high-purity perovskite thin film over the whole coating area.The significantly strong peaks, including <110>, <220>, and <330> in all the measured samples clearly indicate the preferred crystal orientation in the <110> direction, which is known to facilitate out-ofplain transport of charge carriers, which is beneficial to various thin film optoelectronic devices.
The optoelectronic properties of the perovskite thin films based on the developed EtOH-based inks deposited on ITO and ITO/Cu:NiO x were studied using spectral PL measurements (Figure 3c).Focusing on the PL spectra of the MAPbI 3 films on ITO, both films showed high PL intensity, further supporting the high quality of the films and their relatively low defects inside the perovskite film.It is worth noting that a level of PL intensity quenching is expected for these samples due to the direct contact of perovskite and ITO, which cause non-radiative recombination via the traps, and electron transfer from MAPbI 3 to the Fermi level of ITO. [20]A sharper PL quenching was observed when MAPbI 3 thin films were coated on top of Cu:NiO x HTL.At least three main parameters can control the intensity of a PL spectrum for a MAPbI 3 film: the quality of the film itself, the quality of the interface, and the charge extraction from the underlying charge transfer layer.According to AFM results, it is suggested that the quality of the MAPbI 3 films for both inks did not drastically change when coated either on ITO or ITO/Cu:NiO x .This suggests the level of bulk and surface non-radiative recombination can be assumed approximately similar in the two groups of samples.However, the underlying Cu:NiO x can actively affect both the quality of the interface and the extraction of photogenerated holes from the MAPbI 3 film.These two parameters work in opposite directions when having an impact on the PL intensity.Therefore, the decrease or increase in PL intensity alone is not sufficient to confirm the quality of the interface or the efficiency of the charge extraction.To further delve into this matter, time-resolved PL (TRPL) measurement has been performed on all four samples from two inks (Figure 3d).The main fitted data for the decay curves, obtained from the following equation is represented in Table S1, Supporting Information. (1) The decay curves for all samples start with an initial sharp drop (up to ≈20 ns), followed by reaching a plateau.Maintaining the PL intensity at the plateau is mainly attributed to the quality of the interface between the perovskite and the underlying film. [21]nterestingly, an increase was observed in the calculated lifetime of photogenerated charges when MAPbI 3 is coated on Cu:NiO x for both inks (in comparison to the samples without Cu:NiO x ).This can be directly associated with a higher interface quality for the samples with Cu:NiO x .Regarding this finding, along with the observed drop in the steady-state PL intensity of Cu:NiO x -based samples, it can be concluded that the presence of Cu:NiO x as the hole selective layer could significantly enhance the charge extraction from the MAPbI 3 while improving the interface quality.

Characterisation of PSCs
To investigate the quality of the developed perovskite inks and their applicability for the fabrication of optoelectronic devices via scalable deposition methods, PSCs were fabricated.In agreement with the experimental results obtained during the initial characterization of the developed perovskite films, ink E50 was chosen for screening.This originates from higher quality of perovskite films observed for ink E50 compared to ink E70.The structure of the fabricated devices is shown in the cross-sectional SEM image in Figure 4a.It can be seen that the bar-coated MAPbI 3 perovskite layer was sandwiched between the Cu:NiO x as the HTL and PC 60 BM/BCP as the electron transport layer (ETL) forming a p-i-n device architecture. [19]The photovoltaic parameters of all fabricated devices (n = 15; each device was fabricated on a separate substrate) were measured once weekly for four weeks.Figure 4b shows the characteristic photocurrent density-voltage (J--V) plots of the champion device for each week.These device parameters in addition to the averaged parameters calculated on week 3 are furthermore summarised in Table 1.It is evident from the measurement data, that during the storage procedure, the device characteristics undergo a steady increase.This effect is known from the literature and studies, suggesting that it is caused by a re-organization of the perovskite films, which improves the crystallinity and leads to a released residual stress  b) Hysteresis index was calculated based on the reported method. [22]d lattice distortion; a spontaneous coalescence of crystallites in the perovskite films into larger ones; a migration of ions leading to passivation of trap states such as gradual diffusion of Na + ions from the indium tin oxide (ITO) electrode. [23]The J-V plots in Figure 4b and the calculated hysteresis index in Table 1 are also indicative of a negligible hysteresis for the fabricated PSCs, which points to an optimized device structure with an excellent material selection in addition to the use of a correct measurement protocol. [22]Due to the low hysteresis index of our devices, only the data obtained from the reverse direction J-V scans of the PSCs was used in the box plots summarizing the device open circuit voltage (V OC , Figure 4c), short circuit current density (J SC , Figure 4d), fill factor (FF, Figure 4e), and PCE (Figure 4f).It can be seen that all devices exhibited highly reproducible V OC values with a comparable average statistical V OC of ≈1.12 V.It is evident from Figure 4d and Table 1, that there is a gradual J SC increase from an average value of 16.54 ± 0.55 mA cm −2 in week 0 up to 21.20 ± 0.44 mA cm −2 in week 3, respectively.A similar trend is observed for the FF of the devices (see Figure 4e), which shows a slight improvement from an average value of 75.9 ± 1.2% in week 0 to 78.6 ± 0.9% in week 3.We speculate that the improvement of both characteristics is due to an improved crystallinity of the perovskite layer and passivation of trap states as discussed above.As a result, the average PCE of the devices improves from 14.12 ± 0.53% in week 0 to 18.74 ± 0.47% in week 3.It is worth mentioning the low standard deviation of the measured data and therefore, the excellent experimental reproducibility, which is due to the quality of the developed perovskite ink and the precise control over our large-scale applicable deposition process.The champion device was measured in week 3 (see J-V characteristics in Figure 4b) and it exhibited a PCE of 19.56%.In Figure 4 g,f, PL and EL imaging results are presented for one of the devices fabricated in this work.For this characterization, the device was not measured with an aperture mask, which allowed for characterizing the complete photoactive area defined by the overlap of the electrodes (≈ 0.25 mm 2 ).The device demonstrates strong PL and EL signals, as expected considering the efficiencies reported in Table 1.The EL and PL measurements were acquired after the J-V testing of the samples, hence the squarelike pattern in the middle of the device is due to the masking with this aperture mask.This also indicates that the devices are affected by previous and consequent measurements (please note that the J-V measurements were carried out in air without any encapsulation).Non-uniformities similar to the defect visible at the bottom of the device in Figure 4 g affect the electrical performance, which is visualized by the reduced intensity in the EL image.Such local defects can be observed with PL imaging during film fabrication (see Figure 1d-k as a reference) and will eventually affect the performance of finished devices, as illustrated by this example device.Therefore, it is possible to further enhance the performance of such samples by improving the uniformity of the complete device stack and reducing defects in finished PSCs.

Conclusion
In this study, we show the development and screening of MAPbI 3 perovskite precursor inks, based on the green solvent EtOH, which was used in combination with ACN.Four ink formulations were prepared (inks E30, E50, E70, and E90 containing 30%, 50%, 70%, and 90% EtOH by vol. with respect to the total volume of the solvents in the ink) and characterized, after being deposited via a scalable deposition technique, bar coating.Full optimization of the coating and drying conditions was carried out, which led to the deposition of perovskite thin layers with superior quality and purity suitable for optoelectronic thin film device structures.These were deposited at room temperature on rigid or flexible substrates and instantly dried with an air knife gas-blowing process step as shown in Video S1 (Supporting Information) "S2S coating" in the supporting information.We found that the timing and pressure of the nitrogen gas blowing step have a significant impact on the quality of the final film.Moreover, the optimized conditions for the bar coating process are transferable to a roll-to-roll compatible large-scale deposition method such as slot-die coating as we demonstrate in the Video S2 (Supporting Information) "R2R coating" in the supporting information.The deposition reproducibility and spatial uniformity of the perovskite films were studied and shown with large-area PL imaging.This was further confirmed with surface morphology and crystallographic studies conducted by means of using AFM, SEM, and XRD.Small-area PL and TRPL were also carried out to analyze the charge carrier generation statistics, lifetime, and transfer to the adjacent HTL.Finally, the optoelectronic performance of one of the developed MAPbI 3 precursor inks was examined by means of fabricating PSCs.Average device PCEs of 18.74 ± 0.47% and a champion PCE of 19.56% were measured.PL and EL characterization were additionally used to visualize the photoactive area of one device, allowing for a clear identification of defects and layer inhomogeneities, paving a path for further optimization of the perovskite inks and their deposition procedure.Overall, we demonstrate the scaled-up fabrication of high-quality halide perovskite thin layers and show a path toward the widespread application and industrial manufacturing of various perovskite-based optoelectronic devices.
Preparation of Perovskite Precursor Inks: ACN/EtOH-based perovskite inks (with concentration of 1 m) were prepared by adding 1 mmol of MAI and 1 mmol of PbI 2 into the solvent mixture, consisting of MA solution in EtOH (33 wt.%) and ACN, with the volume ratios of 3/7, 5/5, 7/3, and 9/1 for the inks E30, E50, E70, and E90, respectively.Additionally, 24 mol% of DMSO was added to the mixture.Yellow ink was achieved after 5 min of ultrasonic treatment.The inks were filtered using a 0.45 μm PTFE micro filter before deposition.
Fabrication of Perovskite Thin Film Samples: Bar-coated perovskite samples were prepared on ITO-coated glass-and PET substrates, using a slot-die coater from Ossila, which was adapted to function as a bar coater.A smooth stainless-steel bar was used for the coating process.The gap between bar and the substrate was set by using micrometer adjustment screws.An air knife (Exair Super Air Knife) was positioned at the downstream side of the coating bar with a height of 17 cm from the substrate.Nitrogen gas with pressures ranging from 1 to 3 bar was used as the feed gas.All substrates were cleaned sequentially for 5 min in an ultrasonic bath using Helmanex III detergent, deionized water, acetone, and isopropanol and were transferred to an ozone chamber for 10 min before film deposition.The coating width and length were defined based on the substrate size.A coating gap of 150-200 μm was introduced between the bar and the substrates.The ink was pipetted in the coating gap until the whole gap was filled with the ink.For 5 cm × 5 cm glass substrates, 40 μL of the MAPbI 3 precursor was used.The coating process was carried out at room temperature in a nitrogen-filled glove box with a variable velocity from 12 to 14 mm s −1 .An air knife blowing nitrogen was started 5 s after the coating process has finished and an optimized gas pressure of 2 bar was used.The perovskite films were then annealed at 100 °C for 10 min.
Device Fabrication: PSC devices were fabricated on 20 mm × 20 mm × 0.9 mm pre-patterned ITO glass substrates.The ITO substrates were sonicated with detergent, deionized water, acetone, and IPA.They were further cleaned by UV-Ozone treatment for 10 min prior to use.All devices had the inverted architecture of glass/ITO/Cu:NiO x /MAPbI 3 /PC 60 BM/BCP/Ag.The Cu:NiO x nanoparticle (NP) solution was prepared based on the method explained in the previous report. [19]The Cu:NiO x NP solution was spin-coated at 2000 rpm for 20 s, followed by an annealing step at 120 °C for 20 min in air.The MAPbI 3 perovskite thin films were bar coated on top of the Cu:NiO x using the process explained above.PC 60 BM thin film was obtained through spin-coating at 2500 rpm for 30 s in a nitrogen-filled glove box from a 30 mg mL −1 solution in CB.Subsequently, a 0.5 mg mL −1 BCP solution in IPA was spin-coated at 5000 rpm for 25 s.Finally, 100 nm of Ag was thermally evaporated as top electrode using a shadow mask which defines devices with active areas of 0.25 cm 2 .
Device Characteristics: PANalytical X'Pert Pro powder diffractometer in a Bragg-Brentano geometry using a Cu K target (45 kV, 30 mA) was used for the X-ray diffraction crystallographic measurements.SEM images were collected using a Jeol JSM-7800F field emission gun electron microscope, with a working distance of ≈10 mm, an acceleration voltage of 5 kV, and a probe current of the order of 9 μA.AFM images were obtained using Bruker Dimension Edge in a tapping mode.The film thickness was measured by a surface profilometer (Dektak XT).PL and TRPL were obtained using Picoquant FluoTime 300 fluorescence Spectrometer.The J-V curves of the PSCs were measured in air and at room temperature with no encapsulation.The simulated light was one-sun AM 1.5G illumination (100 mW cm −2 ) delivered by a solar simulator (Enlitech, SS-F5-3A).The light intensity was calibrated by using a standard monocrystalline silicon solar cell with a KG-5 filter and a reference cell purchased from Fraunhofer ISE CalLab (ISE001/013-2018).The J-V characteristics and MPP tracking were performed outside the glovebox at the lab condition by using a Keysight B2901A source meter under simulated one-sun AM 1.5G illumination (100 mW cm −2 ) with a AAA steady solar simulator (Enlitech, SS-F5-3A).Before J-V measurements, the simulator was cautiously calibrated by using a standard monocrystalline silicon solar cell with a KG-5 filter same as previously reported, to ensure the accuracy of the J SC measured from J-V scans, a mask with an aperture area of 0.09 cm 2 was employed during the measuring process.The sweeping conditions are reverse scan (1.20 V → −0.02 V, scan rate 40 mV s −1 , and no delay time), forward scan (−0.02V → 1.20 V, scan rate 40 mV s −1 ), with no delay time), and a reference cell purchased from Fraunhofer ISE CalLab (ISE001/013-2018).All devices were measured both in the reverse scan (1.20 V→ − 0.20 V, step 0.02 V, delay time 100 ms) and forward scan (− 0.20 → 1.20 V, step 0.02 V, delay time 100 ms) without any pre-light soaking and pre-bias process.To ensure accuracy, a mask with an aperture area of 0.09 cm 2 was employed during the measuring process).The devices were stored in a nitrogen-filled glove box between the weekly measurements.
18b] 450 nm illumination was used, at an equivalent of ≈300 W m −2 , unless mentioned otherwise.A Hamamatsu ORCA Flash 4.0 CMOS camera was used for capturing images, a 700 nm long pass filter was used for all measurements.A GaAs-calibrated reference cell was used to ensure irradiance intensity stability and image calibration. [24]The processed images were normalized by exposure time when appropriate.Irradiance non-uniformities at the sample plane were <5% (measured up to a 200 mm × 200 mm area).PL images were acquired after 1 min waiting time for each sample, to al-low samples to stabilize in case of metastability effects.EL imaging was applied with a Keithley 2401, at 1 V, measurements were acquired after 10 s, to allow the sample current to stabilize.The sample used for the PL and EL characterization in Figure 4g,h was encapsulated with a glass slide and a UV-curable epoxy prior to the characterization.

Figure 1 .
Figure 1.a) Schematic illustration of the air knife-assisted bar-coating set-up used in this study.The insert image shows the coating bar and respective coating gap (CG) spreading the ink on top of a substrate.b,c) Thickness mapping of the deposited dry perovskite films on 50 mm × 50 mm substrates with a constant or variable coating velocity.d-i)PL study of the delay time "t" and pressure "P" of the gas-blowing step after the deposition of MAPbI 3 thin films from ink E50.The gas-blowing conditions are shown in the PL images.j-k) Comparison of the PL spectra of ink E50 and ink E70, which were dried at optimized conditions.All samples were deposited using identical coating parameters.

Figure 3 .
Figure 3. XRD spectra of MAPbI 3 thin films.a) Sample prepared with ink E50.b) Sample prepared with ink E70.Each sample was measured at 4 locations (S-1 to S-4).PL and TRPL measurements.c) PL spectra of ink E50 and ink E70 deposited on top of different materials.d) Corresponding TRPL measurements of the films shown in a).

Figure 4 .
Figure 4. a) Cross-sectional SEM image of the fabricated PSCs showing the device structure glass/ITO/Cu:NiO x /MAPbI 3 /PC 60 BM/BCP/Ag.Ink E50 was used for device fabrication.b) Photocurrent density-voltage curves of the champion device measured on the day of fabrication (week 0) and weekly for three weeks of storage.Box plots showing the average values of c) V OC , d) J SC , e) FF, and f) PCE for all devices during each week of measurements.g) PL image (50 ms exposure time) of a PSCs device on week 3. h) Corresponding EL image (300 ms exposure time at 1 V, 18 mA).
a) Average data based on 15 individual devices, fabricated on separate substrates;