16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Research Article www.advenergymat.de Solution Processable Direct Bandgap Copper-Silver- Bismuth Iodide Photovoltaics: Compositional Control of Dimensionality and Optoelectronic Properties Narendra Pai, Manjunath Chatti, Sebastian O. Fürer, Andrew D. Scully, Sonia R. Raga, Nitish Rai, Boer Tan, Anthony S. R. Chesman, Zhou Xu, Kevin J. Rietwyk, Saripally Sudhaker Reddy, Yvonne Hora, Gaveshana A. Sepalage, Nadja Glück, Monica Lira-Cantú, Udo Bach,* and Alexandr N. Simonov* 1. Introduction The search for lead-free alternatives to lead-halide perovskite photovoltaic Solar energy conversion technologies materials resulted in the discovery of copper(I)-silver(I)-bismuth(III) halides already produce 4% of electricity world- exhibiting promising properties for optoelectronic applications. The present wide, and this contribution will continue work demonstrates a solution-based synthesis of uniform CuxAgBiI4+x thin to proliferate to support our transition films and scrutinizes the effects of x on the phase composition, dimension- toward sustainable energy. This progress ality, optoelectronic properties, and photovoltaic performance. Formation builds upon developments of new photo- of pure 3D CuAgBiI5 at x = 1, 2D Cu2AgBiI6 at x = 2, and a mix of the two at voltaic devices, like lead-halide perov- skite solar cells, that have piqued strong 1 < x < 2 is demonstrated. Despite lower structural dimensionality, Cu2AgBiI6 interest from both industry and academia has broader optical absorption with a direct bandgap of 1.89 ± 0.05 eV, a as a competitive alternative to the crys- valence band level at -5.25 eV, improved carrier lifetime, and higher recombi- talline silicon technology that currently nation resistance as compared to CuAgBiI5. These differences are mirrored dominates the market. However, chal- in the power conversion efficiencies of the CuAgBiI5 and Cu2AgBiI6 solar lenges in demonstrating the sufficient long-term stability of perovskite solar cells cells under 1 sun of 1.01 ± 0.06% and 2.39 ± 0.05%, respectively. The latter in operation and toxicity of Pb2+ are yet to value is the highest reported for this class of materials owing to the favorable be solved, which motivates the search for film morphology provided by the hot-casting method. Future performance similarly effective but more stable and improvements might emerge from the optimization of the Cu2AgBiI6 layer lead-free light harvesters.[1] thickness to match the carrier diffusion length of ≈40–50 nm. Nonencapsu- While the research on the homovalent lated Cu2AgBiI6 solar cells display storage stability over 240 days. substitution of lead(II) with cations like Sn2+ and Ge2+ in the perovskite structure N. Pai, S. O. Fürer, N. Rai, B. Tan, K. J. Rietwyk, S. S. Reddy, Y. Hora, N. Pai, S. O. Fürer, K. J. Rietwyk, U. Bach G. A. Sepalage, N. Glück, U. Bach ARC Centre of Excellence for Exciton Science Department of Chemical Engineering Monash University Monash University Clayton, Victoria 3800, Australia Clayton, Victoria 3800, Australia M. Chatti, A. N. Simonov E-mail: udo.bach@monash.edu School of Chemistry N. Pai, S. S. Reddy, N. Glück, U. Bach Monash University The Australian Centre for Advanced Photovoltaics (ACAP) Clayton, Victoria 3800, Australia Monash University E-mail: alexandr.simonov@monash.edu Clayton, VIC 3800, Australia A. D. Scully, A. S. R. Chesman CSIRO Manufacturing The ORCID identification number(s) for the author(s) of this article Clayton, Victoria 3168, Australia can be found under https://doi.org/10.1002/aenm.202201482. S. R. Raga, M. Lira-Cantú © 2022 The Authors. Advanced Energy Materials published by Institut Català de Nanociència i Nanotecnologia Wiley-VCH GmbH. This is an open access article under the terms of Universitat Autònoma de Barcelona the Creative Commons Attribution-NonCommercial-NoDerivs License, Bellaterra, Barcelona 08193, Spain which permits use and distribution in any medium, provided the original Z. Xu work is properly cited, the use is non-commercial and no modifications Monash Centre for Electron Microscopy or adaptations are made. Monash University Clayton, VIC 3800, Australia DOI: 10.1002/aenm.202201482 Adv. Energy Mater. 2022, 12, 2201482 2201482 (1 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de is making outstanding progress in power conversion efficiency extreme pressures[14] and ionic substitution[15] produced some (PCE) terms, the resulting compounds are even less stable than positive results, including a decrease in Eg and an increase in their Pb-based counterparts.[2] In contrast, heterovalent metal visible light absorption. However, it has proven challenging substitution produces materials with excellent stability, yet their to translate these improvements to operational photovoltaic optoelectronic properties are significantly less favorable for photo- devices.[16] Even when the solar cells based on modified Cs2Ag- voltaic applications.[3] For example, relatively stable bismuth(III)- BiBr6 could be produced, the reported improvements did not based compounds with a generic formula A3Bi2X9 (where enable PCEs at the level comparable to that of lower-dimen- A+ = CH3NH3+ or Cs+, and X− = Cl−, Br− or I−) exhibit lower struc- sional bismuth perovskites.[16a,c] Moreover, none of the reported tural dimensionality with wide bandgaps (Eg > 2 eV) and higher M-site modification strategies resulted in a strong direct excitonic binding energies (over 0.1 eV) as compared to lead(II) bandgap transition predicted by theoretical studies.[9b,17] perovskites based on the same A+ and X− components, and there- Many of the performance-limiting band-structure problems fore struggle to demonstrate high photovoltaic performance.[4] of Cs2AgBiBr6 stem from the influence of Br p-orbitals on the Aiming to address this, research on replacing alkaline-earth valence band maxima (VBM) and conduction band maxima A+ cations with coinage metal cations, viz. Cu+ and Ag+, has (CBM)[16d] and are therefore expected to be resolved upon demonstrated a structurally and electronically 3D AaBbXa+3b transition to the iodide counterpart, viz. Cs2AgBiI6. Another (A = Cu+ or Ag+; B = Sb3+ or Bi3+; X = Cl−, Br− or I−) family constraint that needs to be resolved is the lower electronic of pnictogen halides with direct band transitions that outper- dimensionality of Cs2AgBiBr6, which arises due to the isolation form conventional bismuth iodide perovskites.[5] Although the of the [AgBr6] octahedra by the adjacent [BiBr6] octahedra and relatively wide bandgaps of such compounds (Eg > 1.85 eV) results in the inability to connect VBM (contributed by Ag 4d were initially a limitation, bandgap engineering strategies have and Br 5p orbitals) and CBM (contributed by Bi 6p orbitals) in been introduced to produce better-performing photovoltaic three dimensions.[10] The iodide-based double perovskites were devices. As a result, almost a fivefold increase in the PCE from predicted to be devoid of these problems and to exhibit a direct ≈1.2 to 5.5% for solar cells based on silver bismuth iodides band transition with improved dimensionality,[18] but Cs2AgBiI6 has been achieved within only a few years of research.[5a,b,6] is, unfortunately, thermodynamically unstable[19] and decom- However, Kulkarni et al. have demonstrated recently that the poses into a lower-dimensional Cs3Bi2I9.[20] Thus, no success Ag+ ion migration induces device degradation in the highest- has been achieved in synthesizing Cs2AgBiI6 so far, except in performing silver-rich AgaBibIa+3b (a > b) compositions.[7] the form of a colloidal nanocrystal dispersion in toluene.[21] Replacing Ag+ with Cu+ was suggested to reduce the ion migra- A possible strategy to avoid phase decomposition of Cs2AgBiI6 tion and associated issues, but introduced new problems, is to replace large Cs+ (174 pm) with cations that have smaller including phase instability. As a result, the overall progress ionic radii.[22] While this approach did not produce positive in the photovoltaic performance improvements has been sig- results when using monovalent cations neighboring cesium, nificantly slower for the copper-based compounds of this class viz. Rb+ (156 pm) or K+ (146 pm), the substitution of Cs+ with (best PCE < 1.2 % reported so far).[6b,8] Cu+ resulted in a stable Cu2AgBiI6 compound.[23] Unlike corner- Concurrently, computational studies by Savory et al. and Cai shared lead-halide perovskites, Cu2AgBiI6 was reported to exhibit et al. revealed that replacing two adjacent Pb2+ with monovalent structural similarities with the edge-shared AaBibIa+3b (A = Cu+ and trivalent cations could produce 2 × 2 × 2 supercell perov- or Ag+) materials. When exploring the CuI-BiI3 system, Sansom skites with a generic A2M1+M3+X6 formula, broadening oppor- et al.[24] observed a metastable behavior of CuBiI4, which decom- tunities for the design of new photovoltaic materials.[9] Among posed back to CuI and BiI3 under ambient conditions even this family of compounds, Cs2AgBiBr6 received specific atten- without any irradiation. However, once AgI was introduced, a tion due to its high stability and a range of advantageous proper- stable phase of Cu2AgBiI6 was formed and found to have a trig- ties like small carrier effective mass (< 0.5m0), moderate charge onal unit cell of space group R3m forming a CdCl2-type rhom- carrier mobility (1 cm2 V−1 s−1), and long carrier lifetime.[10] The bohedral layered structure. It was also postulated that the intro- latter parameter reported for Cs2AgBiBr6 single-crystals exceeds duction of Cu+ reduces Ag+ ion migration.[25] Thin films of the 1 µs, which theoretically corresponds to a very long carrier dif- Cu2AgBiI6 material showed promising optoelectronic properties, fusion length over 1 µm (calculated using the carrier mobility including an absorption coefficient over 1 × 105 cm−1 within a value noted above). However, this material also presents a 375–600 nm wavelength range, an exciton binding energy of broad range of limitations, including an indirect wide bandgap 0.025 eV and a direct bandgap. Although the Eg value of 2.06 eV (>2 eV), downshifted valence band edge (below -5.6 eV; herein- was still relatively high, it was also improved compared to after vs vacuum), shallow absorption edge with Urbach energy ≈2.3 eV for Cs2AgBiBr6. However, a proof-of-concept Cu2AgBiI6- of 0.07 eV, and high trap-state density mediating charge-carrier based solar cell delivered a low PCE of ≈0.4%, emphasizing the trapping and nonradiative recombination.[11] Moreover, Snaith need for further research toward understanding the effects of and co-workers reported that the Cs2AgBiBr6 thin-films exhibit the active layer morphology and composition along with the very short, trap-restricted carrier lifetimes below 100 ns, which device architecture on the photovoltaic performance.[23] result in short electron and hole carrier diffusion lengths Further exploration of the CuI-AgI-BiI3 system by Sansom (≤30 and ≈150 nm, respectively).[12] As a result, Cs2AgBiBr6- et al. revealed a 3D spinel CuAgBiI5 composition, which is based solar cells do not operate efficiently in the visible and structurally close to spinel CuBiI4 and Bi-rich AgaBibIa+3b (a < b) infrared regions and have failed to produce short-circuit cur- compounds. It was reported that the large trigonal unit cell with rent densities above 6 mA cm−2 under 1 sun irradiation.[12,13] a 3D spinel octahedral motif had no significant effects on the Attempts to improve the properties of Cs2AgBiBr6 by applying optoelectronic capabilities of Cu2AgBiI6 compared to CuAgBiI5. Adv. Energy Mater. 2022, 12, 2201482 2201482 (2 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de This is distinct to the dimensional confinement effects observed a temperature-assisted hot-casting strategy with the solutions in lead-halide and heterovalent lead-free perovskites.[24,26] and substrates preheated at different temperatures within a Overall, our current fundamental understanding of the cor- 60–120 °C range for 5 min and then immediately used for spin- relations between dimensionality, structure, composition, and coating. This method avoids the use of antisolvents and facili- optoelectronic properties of the promising Cu-Ag-Bi-I mate- tates control over crystallinity, morphology, and composition of rials is far from complete. Moreover, no photovoltaic devices the produced coatings. Scanning electron microscopic (SEM) based on CuAgBiI5 have been reported yet, while those based analysis showed that the best quality films with a minimal con- on Cu2AgBiI6 delivered only low power conversion efficien- centration of pin-holes and impurities are obtained with 0.45 M cies below 1%.[27] Addressing this, we present herein a study [CuxAgBiI4+x] precursor concentration and a DMSO:DMF ratio of the compositional tunability of optoelectronic characteristics of 3:1 vol (Figure S4, Supporting Information). and associated photovoltaic performance for the CuxAgBiI4+x To further improve the quality of the films, which is critical system. One feature of the present work is the first demon- for optoelectronic applications, the temperatures of the solu- stration of a solvent-based hot-casting method to prepare thin tion and substrate were optimized (Figure S5, Supporting CuxAgBiI4+x films with improved morphology and surface cov- Information). As determined by SEM, the best results were erage, which allowed us to achieve the highest photovoltaic obtained with the solution and the mTiO2|cTiO2|FTO support performance for these materials reported so far. Stability under both preheated to 100 °C (Figure S5, Supporting Information). ambient storage conditions on a timescale of months is also Preheating controls the rate of crystallization and facilitates presented. complete removal of DMSO from the films, i.e. avoids trapping of the solvent with a high boiling point (189 °C), which can induce the formation of pin-holes during postannealing. The 2. Results and Discussion latter was undertaken herein in two steps, viz. 80 °C followed by 150 °C to slow down the crystallization and minimize the All materials investigated herein were synthesized by casting loss of volatile iodide. The optimized hot-casting assisted spin- their precursor solutions in mixed dimethylsulfoxide (DMSO) coating approach produces dark shiny films with visually uni- + dimethylformamide (DMF) solvent system onto a substrate form substrate coverage (Figure S6, Supporting Information). of interest, following the procedure summarized in Figure S1 When formed, the investigated materials exhibit high thermal (Supporting Information). Using this solvent-based synthesis stability with the decomposition temperature above 250 °C strategy to produce McAgBiI4+c materials with M = Rb+ and K+ (Figure S7, Supporting Information). was unsuccessful and resulted in phase segregation (Figure S2, Another advantage of the hot-casting method is homoge- Supporting Information). These results are consistent with pre- neous nucleation of the target material throughout the support vious reports.[24] provided by preheating the latter, in contrast to conventional At the same time, the solvent-based procedure introduced antisolvent- and gas-assisted strategies that facilitate nucleation herein was found to be effective for the fabrication of the in the topmost layers of the spin-coated film. Thus, the temper- CuxAgBiI4+x films from the solutions containing equimolar ature-assisted approach enables crystallization throughout the concentrations of bismuth(III) and silver(I) iodides along with whole surface and volume of the mesoporous TiO2 scaffold and different amounts of copper(I) iodide. To overcome the solu- decreases the contribution of voids and SEM-bright, viz. poorly bility limitations of AgI and CuI in DMSO + DMF, the pre- conducting impurities (Figure 1a–d). Furthermore, adjusting cursor solutions were heated to 80 °C and stirred for one hour, temperatures used for the hot-casting procedure provides a cooled down to ambient temperature, and filtered before depo- technically simple means for controlling the impregnation and sition. The resulting homogenous red-colored solutions were crystallization of the precursors inside the uneven mesoporous spin-coated utilizing gas-assisted, anti-solvent, and hot-casting scaffold aiming for the formation of flat and uniform films strategies. Films were deposited onto fluorine-doped tin(IV) (Figure S5, Supporting Information). It was additionally noted oxide (FTO) coated glass modified with a compact titania layer that the film morphology improves with the amount of copper (c-TiO2; ≈15 nm thickness) and then with a mesoporous TiO2 in CuxAgBiI4+x, with the best films obtained at 1.5 ≤ x ≤ 2.0, framework (m-TiO2; ≈140 nm thickness; 30 nm average particle while excessive amounts of Cu (x > 2.0) induce undesirable size), except for the glass-supported samples for the X-ray dif- agglomeration of grains (Figure S6, Supporting Information). fraction (XRD) and photoluminescence spectroscopic (PL) anal- Probing the structure of the CuxAgBiI4+x films produced by ysis. The experimental section provides further details on the the optimized solution-based hot-casting method with X-ray preparation and characterization procedures. diffraction revealed that the materials present a composition- dependent combination of two R-3m space group phases with edge-sharing octahedra[24] (Figure 2a). When using a precursor 2.1. Film Morphology, Composition, and Dimensionality solution with x = 1, a rhombohedral CuAgBiI5 with 3D spinel octahedra was a dominant phase. Increasing the Cu:Au ratio The use of conventional deposition methods, either with the beyond 1 resulted in the emergence of the Cu2AgBiI6 CdCl2- gas flowing or different antisolvents applied to the spin-coated type layered phase and a progressive decrease in the intensity of solution, produced heterogeneous coatings with a high concen- the CuAgBiI5 diffraction signals (Figure 2a). It is worth noting tration of pin-holes (Figure S3, Supporting Information), which that no admixtures of CuI or AgI were detected in all samples could not be improved by adjusting the precursor concentra- examined herein, contrasting the results typically obtained for tions and the DMSO:DMF ratio. To resolve this, we employed Ag-rich or Bi-rich silver bismuth iodides.[5a,6a,7] Furthermore, Adv. Energy Mater. 2022, 12, 2201482 2201482 (3 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de Figure 1. a) Top-view and b) side-view SEM images of the CuxAgBiI4+x films with x = 1 (left) and 2 (right) deposited via regular antisolvent(chlorobenzene)- assisted spin-coating and hot-casting onto m-TiO2|c-TiO2|FTO substrates. Cross-sectional micrographs were recorded for complete devices and were manually colored to highlight the spiro-OMeTAD (purple), CuxAgBiI4+x (dark orange), CuxAgBiI4+x + m-TiO2 (teal) and c-TiO2 (brown) layers on the FTO support. unlike the case of AgBiI4, which demonstrates cubic (Fd3m) unambiguously demonstrate a compositionally-controlled and CdCl2-type R-3m space grouped twinned structures, crystal transformation from 3D CuAgBiI5 to 2D Cu2AgBiI6 (Figure 2b twinning was not observed for the CuxAgBiI4+x system herein, and Figure S8, Supporting Information). which either contained pure (x = 1 and x ≥ 2) or two mixed X-ray photoelectron spectroscopic (XPS) analysis confirmed phases (1 < x < 2) (Figure 2a). Relative intensities of XRD peaks the presence of copper, silver, bismuth, and iodine species on for individual phases (x = 1 and 2) were different from those the film surfaces, as well as the change in atomic ratios of the reported by Sansom et al.[23] for materials synthesized by high- elements corresponding to the transformation from CuAgBiI5 temperature fusing, reflecting different crystal orientations in to Cu2AgBiI6 (Figure 3a–d and Figure S9, Supporting Informa- thin-film materials, which are often affected by the support.[28] tion). While Ag+, Cu+ and I− were detected as the only states This prevented conventional fitting of the XRD profiles with the of these elements, which did not change with increasing x, the tabulated data. It was also noted that the diffractograms for the dominating Bi3+ signals in the Bi 4f spectra (164.6 and 159.4 eV) CuxAgBiI4+x films could not be fitted by linear combinations of were accompanied by low-intensity signals with binding energies the data recorded herein for CuAgBiI5 and Cu2AgBiI6, which close to metallic bismuth (Figure 3c and Figure S10, Supporting indicates that the structural features of one phase are affected Information). The appearance of these “Bi0” species could be by intermixing with another. Most importantly, the XRD data due to a high concentration of interstitial defects, viz. absence of Adv. Energy Mater. 2022, 12, 2201482 2201482 (4 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de Figure 2. a) XRD patterns of the CuxAgBiI4+x films on glass with x = 1 (light blue), x = 1.25 (blue), x = 1.5 (green), x = 1.75 (purple), x = 2 (orange), and x = 2.25 (red); black curves show XRD patterns for CuAgBiI5 (2107463 entry ID) to Cu2AgBiI6 (2107468 entry ID) deposited by Sansom et al.[23,24] to the Cambridge database. b) Structures of 3D CuAgBiI5 (x = 1) and 2D Cu2AgBiI6 (x = 2). some of the BiI bonds typically observed in BiI3.[25] It is also absorption as compared to Cs2AgBiBr6 (Figure 4a and Figure S12, vital that XRD did not provide any evidence for significant crys- Supporting Information). Analysis of materials obtained by tallographic defects in the materials examined herein, suggesting reacting [KI + AgI +BiI3] and [RbI + AgI + BiI3] produced optical that the “Bi0” species might be predominantly surface-confined. absorption with several peaks between 500 and 550 nm, prob- The surface composition of the CuxAgBiI4+x materials with ably contributed by the lower-dimensional bismuth perovskites x = 1 and 2 derived from XPS was in satisfactory agreement (Figure S12a, Supporting Information), confirming the XRD with the expected stoichiometry (Figure 3e and Table S1, Sup- results on the phase segregation. porting Information). It is also noted that essentially the same Cu2AgBiI6 films demonstrated broader absorption than elemental ratios were measured for the samples deposited by CuAgBiI5 with no well-defined peaks (Figure 4a). The optical conventional antisolvent-assisted spin-coating with poor mor- absorption covered almost the entire visible spectra with a phology (Figure S11, Supporting Information). In addition, cut-off around ≈680 and 700 nm for CuAgBiI5 and Cu2AgBiI6, bulk metal ratios determined by inductively coupled plasma respectively. Slightly broader absorption was reported for optical emission spectroscopy (ICP-OES) also conformed to the CuAgBiI5 by Sansom et al. for the materials obtained by high- expected CuAgBiI5 and Cu2AgBiI6 compositions (Figure 3e). temperature synthesis.[24] Herein, improved absorption is dem- Overall, physical characterization demonstrates that the onstrated by 3D CdCl2-type CuAgBiI5 at shorter wavelengths employed hot-casting method enables formation of the (<480 nm) alongside a small unknown peak at ≈420 nm CuxAgBiI4+x materials with phase composition and dimension- (Figure 4a and Figure S12b, Supporting Information). It is ality directly controlled by the amount of Cu introduced. The additionally noted that absorption coefficients measured herein films produced uniformly cover the titania-based substrate as are slightly higher than those reported before,[24] which can required for optoelectronic applications. be attributed to improvements in the film compactness and slightly lower direct bandgap values. Examination of the optical absorption edges in the (αhv)n–hv 2.2. Optoelectronic Properties Tauc coordinates revealed that CuxAgBiI4+x are direct bandgap photoactive semiconductors with Eg of 1.94 ± 0.05 eV for Effects of the phase composition and dimensionality on the CuAgBiI5 and 1.89 ± 0.05 eV for Cu2AgBiI6 (Figure 4a). optical properties of CuxAgBiI4+x were analyzed and compared Photoelectron spectroscopy in the air (PESA) studies also to those of the well-explored double perovskite, Cs2AgBiBr6. produced slightly different valence band edge (EVB) values The light-harvesting capability was significantly improved of –5.16 ± 0.05 eV for x = 1 and –5.25 ± 0.05 eV at x = 2 for CuAgBiI5 and Cu2AgBiI6 with red-shifted and broadened (Figure 4b and Figure S13, Supporting Information). These Adv. Energy Mater. 2022, 12, 2201482 2201482 (5 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de states reported for CuxAgBiI4+x materials,[24] one can conclude that the A-site cationic (Cs+ to Cu+) and X-site anionic (Br− to I−) modification collectively induce several important changes, viz. transition from indirect to direct band semiconductor with con- siderably improved light absorption and upshifted valence band. The valence band edges were slightly-off for CuAgBiI5 (-5.16 eV) and closely aligned for Cu2AgBiI6 (-5.25 eV) against the highest occupied molecular orbital (HOMO) of the con- ventional hole transport material spiro-OMeTAD (−5.20 eV).[29] This can be resolved by employing other hole transport mate- rials and modifying the band level of spiro-OMeTAD by deeper doping with spiro-OMeTAD(TFSI)2 (Figure 4b and Figure S13, Supporting Information). Calculated conduction band edges (ECB) were -3.22 and -3.35 eV for CuAgBiI5 and Cu2AgBiI6, respectively, leaving a significant gap from the conduction band edge of the conventional TiO2 electron transport mate- rial (−4.20 eV),[29] signifying possible trap-state recombination during the charge transfer. Like many bismuth-based lead-free perovskite-inspired materials, CuxAgBiI4+x demonstrates weak photoluminescence (PL), which was very similar for x = 1, 1.5, and 2 (Figure 4c and Figure S14a, Supporting Information). The PL signal decay was found to decay mono-exponentially with a carrier lifetime of 1.3 ± 0.1 µs, which was essentially unaffected by x (Figure 4d and Figure S14b, Supporting Information). In contrast, PL decay transients for Cs2AgBiBr6 are best fitted by a biexponen- tial dependence with a prominent short-lifetime (<10 ns) com- ponent (Figure 4d and Figure S14, Supporting Information). No significant differences in the PL properties of CuAgBiI5 and Cu2AgBiI6 were found even at a higher fluence (Figure S15, Supporting Information). These results indicate that the PL properties of the CuxAgBiI4+x films are effectively independent of phase dimensionality. Interestingly, using the hot-casting method, the PL spectra of CuAgBiI5 and Cu2AgBiI6 films deposited on glass substrates show excellent stability for 22 h under ambient laboratory conditions (Figure S16, Supporting Information), contrasting previous reports of a blue shift and intensification of the PL spectra upon exposure to air.[24] It should be noted that all PL data reported in the present work were measured in air, while Sansom et al.[24] conducted their t = 0 measurements in vacuum. No significant change in the PL of CuxAgBiI4+x materials over 22 hours of air exposure (Figure S16, Supporting Information) indicates that passivation of the trap-states that suppress non- Figure 3. High resolution a) Cu 2p + I 3p, b) Ag 3d, c) Bi 4f, and radiative recombination does not occur under these conditions. d) I 3d spectra of CuxAgBiI4+x with x = 1 (light blue) and 2 (orange), and This can be either attributed to the effects of the employed pro- e) corresponding atomic fractions derived from XPS (solid) and ICP-OES cessing procedure and/or rapid air-passivation of the materials analysis (hashed). upon initial exposure to air prior to the PL measurements. Electron-beam-induced current (EBIC) measurements were results indicate that the effects of the lower-dimensional undertaken to investigate the carrier diffusion lengths for the A-site cations on band edges and electronic structure are not CuxAgBiI4+x materials (Figure 4e–g). In these experiments, strong, in contrast to the Bi and I valence states.[4c] Sansom numerous electron–hole pairs are generated in semiconductors et al. also observed similarly small differences in Eg and EVB by a highly energetic electron beam (electron-voltaic effect rather between CuAgBiI5 and Cu2AgBiI6. However, the band level than photovoltaic), which are later extracted and collected at the values were different from those obtained herein, probably electrodes by applying an internal electric field. Thus, imaging due to the use of distinct measurement methods and syn- of the localized charge carrier generations reveals impor- thesis conditions.[24] When comparing the results obtained tant electrical features such as the minority carrier diffusion for the CuxAgBiI4+x system to the optoelectronic structure of under applied voltage, p–n junction formation within devices, Cs2AgBiBr6 (Eg = 2.37 eV, EVB = -5.56 eV) and the density of and recombination centers if the crystal grain boundaries are Adv. Energy Mater. 2022, 12, 2201482 2201482 (6 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de Figure 4. Optoelectronic properties of CuxAgBiI4+x with x = 1 (light blue) and 2 (orange) (data for Cs2AgBiBr6 are provided in purple for comparison): a) UV–vis spectra and Tauc plots (inset) (on mTiO2|cTiO2|FTO); b) valence band edge values (EVB) derived from photoelectron spectra (PESA) (glass- coated samples) along with calculated conduction band edge data compared to energy levels of TiO2 and HTMs; c) photoluminescence spectra, and d) photoluminescence decay measurements (λex = 465.8 nm; glass-coated samples) (inset shows TRPL profile for Cs2AgBiBr6); cross-sectional e) secondary electron image and f) EBIC image under internal electric field for complete Au|spiro-OMeTAD|Cu2AgBiI6|m-TiO2|c-TiO2|FTO devices, and g) EBIC profiles for CuxAgBiI4+x with x = 1 (light blue) and 2 (orange) integrated over relevant areas in the images (exemplified for Cu2AgBiI6 with dashed lines in panels e,f); background shading shows layers within the devices. visible. It can be considered that electrons and holes are gener- Information). The integrated profiles extracted from the EBIC ated as free charge carriers considering that the applied energy images demonstrated the current probe peaking at the center (>0.05 eV) significantly exceeds the exciton binding energy. of the CuxAgBiI4+x layer, indicating the formation of a p–i–n Within complete solar cell devices, internal electric fields and the structure (Figure 4g). The difference in the EBIC current inten- field created by the external bias would contribute to the carrier sity provides a measure of the carrier dynamics. Specifically, we generation minimizing the photoinduced exciton generation. observe a gradual improvement in the EBIC values from the Cross-sectional images and corresponding EBIC profiles of HTM|CuxAgBiI4+x and ETM|CuxAgBiI4+x interfaces toward the the devices reveal localized charge carrier generation within active layer (Figure 4g) and increased intensity around larger the active material (Figure 4e,f and Figure S17, Supporting compared to the less connected smaller grains (Figure 4e,f). Adv. Energy Mater. 2022, 12, 2201482 2201482 (7 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de The latter observation indicates that the grain boundaries and density tracking at a constant potential corresponding to interfaces rather than the layer surface are active channels for the maximal power output in the J–V curves over 300 s for the carrier extraction and collection. The electron diffusion Cu2AgBiI6, which provides a reliable measure of the initial per- lengths (Ln) derived from the EBIC data for CuAgBiI5 and formance of the devices, produced PCE values (2.39 ± 0.05%) Cu2AgBiI6 were 40 ± 1 and 46 ± 1 nm, while the hole carrier similar to those derived from transient J–V measurements diffusion lengths (Lp) were around 40 ± 1 nm and 49 ± 1 nm, (2.35 ± 0.04%). respectively (Figure 4e–g and Figure S18a–c, Supporting Infor- The better performance of 2D Cu2AgBiI6 as compared to mation). Considering the amount of drift in EBIC current while 3D CuAgBiI5 is consistent with the slightly better film quality imaging, these values could be slightly underestimated. How- (Figure 1) on the m-TiO2|c-TiO2|FTO support, broadened optical ever, comparing these values to the active layer thickness of absorption (Figure 4a), better-aligned energy levels (Figure 4b), 120–150 nm above the m-TiO2 network implies that a thinner and slightly improved carrier lifetime (Figure 4d). The PCE and compact active layer could avoid potential recombination and Jsc achieved herein for Cu2AgBiI6 compare well to those in the devices and thereby enhance the efficient extraction of reported for Cs2AgBiBr6 combined with similar ETM and charges. However, attempts to develop thinner layers resulted HTM and are better than those provided by a range of other in non-uniform coverage of the films at this stage (Figure S19, lead-free perovskite-inspired materials (Table S6, Supporting Supporting Information). Information).[6a,13a,27,34] Furthermore, the photovoltaic met- rics presented in this work are the best reported to date for the CuxAgBiI4+x class of materials. As anticipated, the films produced 2.3. Photovoltaic Performance with the solution and substrate temperatures of 100 °C produced the best results among the devices based on the CuxAgBiI4+x Solar cells based on the hot-casted CuxAgBiI4+x films had a films prepared using hot-casting and were also better than those n–i–p architecture C|HTM|CuxAgBiI4+x|m-TiO2|c-TiO2|FTO achieved when using antisolvent-assisted spin-coating (Figure S24, (Figure 1b). The use of c-TiO2|FTO without a mesoporous titania Supporting Information). These results emphasize the impor- layer did not produce high-quality films regardless of the depo- tance of film quality for photovoltaic performance. sition conditions used (exemplified in Figure S19, Supporting Profiles of the incident photon to current conversion effi- Information). Corresponding devices expectedly produced ciency (IPCE) spectra (Figure 5b) were qualitatively consistent much worse photovoltaic performance than the m-TiO2|c- with the optical absorption data (Figure 4a and Figure S25, TiO2|FTO-based ones (Figure S20 and Table S2, Supporting Supporting Information) but quantitatively different, showing Information). It was also found that conventional LiTFSI, tBP, significantly higher IPCE for Cu2AgBiI6 within the whole wave- and FK209 cobalt(III) complex additives to spiro-OMeTAD length range examined. The IPCE recorded for the Cu-rich have a strongly adverse effect on the active layers (Figure S21, material was lower than those reported for the AgaBibIa+3b- Supporting Information), which motivated us to use [spiro- PTAA-based solar cells,[6–8] but comparable to other bismuth- OMeTAD2+][TFSI]2 as a chemically cleaner dopant, following based perovskite-inspired materials.[4c,16a,35] procedures established in our laboratories.[30] The best perfor- Dependencies of the short-circuit current and open-circuit mance was achieved by adding 8 mol% of spiro-OMeTAD2+ voltage on the light intensity were linear in the semi and biloga- with respect to spiro-OMeTAD0, resulting in a HOMO level rithmic coordinates, respectively, confirming the charge extrac- of –5.14 eV. Other HTMs such as PTB7 (HOMO = −5.2 eV),[29] tion capability at varying illumination (Figure 5c). The Jsc plots PTAA (HOMO = −5.1 eV),[6a] P3HT (HOMO = −4.8 eV),[31] and displayed the same slope of ≈1.0 for CuAgBiI5 and Cu2AgBiI6. Fe(bpyPY4)](OTf)2+x (HOMO = −5.4 eV)[32] were also examined, On the other hand, the ideality factors derived from the Voc but all provided inferior results to those obtained with 8 mol.% versus light intensity data were distinctly different, viz. 1.3 and spiro-OMeTAD2+ doped spiro-OMeTAD (Figure S22 and 2.7 for CuAgBiI5 and Cu2AgBiI6, respectively. These values sug- Table S3, Supporting Information). The latter HTM was used gest multiple trapping in band edge states or tail states, thereby for all major experiments discussed below. mediating trap-assisted (SRH) recombination.[36] A recently developed cheap adhesive carbon counter elec- To further understand the performance-limiting parameters, trode[33] was used herein and provided significantly better per- impedance spectroscopy was applied to investigate the charge formance than conventional evaporated gold (Figure S23 and transfer within the devices as a function of the dimensionality Table S4, Supporting Information). This might be possibly and phase composition defined by the change in the copper(I) attributed to the penetration of evaporated gold through the concentration. Experimental data for the devices based on both nanoscale pin-holes within a HTM layer and further through CuAgBiI5 and Cu2AgBiI6 were found to be well described by the active layer stimulating the Ag+ or Cu+ ion migration. a simple Rs[Rrec||CPE] equivalent circuit, where Rs is series Photocurrent–voltage (J–V) curves recorded from open- resistance, CPE is a constant phase element used to model circuit to short-circuit (forward bias) and short-circuit to the capacitive behavior of the device, and Rrec is recombina- open-circuit (reverse bias) at a scan rate of 0.200 V s−1 showed tion resistance (Figure S26, Supporting Information). Hence negligible hysteresis (Figure 5a). A substantial improvement in the charge transfer within the CuxAgBiI4+x-based devices is the open-circuit voltage (Voc) and short-circuit photocurrent den- governed by recombination at one of the critical interfaces, sity (Jsc) was observed for CuxAgBiI4+x-based solar cells with the described by the parallel capacitor and resistor elements. It is increase in x, peaking the performance at x = 2, corresponding worth noting that from Nyquist plots (Figure S26a, Supporting to the pure 2D Cu2AgBiI6 phase (Figure 5a and Table S5, Information), we have found a small inductive loop at low fre- Supporting Information). The quasi-steady-state photocurrent quencies below 10 Hz, which suggests that ion migration might Adv. Energy Mater. 2022, 12, 2201482 2201482 (8 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de Figure 5. Photovoltaic performance of the C|spiro-OMeTAD|CuxAgBiI4+x|m-TiO2|c-TiO2|FTO solar cells (aperture 0.16 cm2) with x = 1 (light blue) and 2 (orange): a) photocurrent density versus voltage curves (sweep rate 0.200 V s−1; 1 sun AM 1.5G irradiation) recorded in the open-circuit to short-circuit (solid) and short-circuit to open-circuit (dashed) (dotted lines are guides to the eye showing zero J and V); b) IPCE spectra (solid) and corresponding integrated photocurrent (dashed); c) dependence of the open-circuit voltage (Voc) and short-circuit photocurrent (Jsc) on the illumination light intensity (note semi-log coordinates for Voc and bilogarithmic coordinates for Jsc), and d) recombination resistance, Rrec extracted from the impedance spectro- scopic analysis (lines are guides to the eye); Nyquist plot with the equivalent circuit used for the fitting and other parameters obtained are shown in Figure S25 and Table S6 in the Supporting Information. still exist in the CuxAgBiI4+x materials, but much slower than in values derived from the steady-state power output measurements the lead-halide perovskites. (Figure 6e). The increase in Voc with x can be attributed to the As expected, the Rrec values were found to decrease with improved film coverage that reduces surface defects and band increasing the applied voltage (Figure 5d, Table S7 and alignment with the interfacial layers and avoids trap-assisted Figure S26b, Supporting Information).[37] Rrec is smaller for the recombination while broadening the light absorption. Higher devices based on CuAgBiI5 as compared to the Cu2AgBiI6-based IPCE of Cu2AgBiI6 promotes improvements in Jsc with x. An ones within the voltage range of 0.4–0.7 V, consistent with the increase in x beyond 2 induces a slight drop in the photovoltaic lower Voc for the 3D material (Figure 5a). This result was cor- performance, which might be due to uneven crystal growth, dete- roborated by the J–V data obtained from the impedance fitting rioration of the film morphology, and the presence of impurities (Figure S27, Supporting Information). The geometric capaci- on the surface (Figure S6, Supporting Information). However, tance was very similar for all devices examined (Figure S26d, a higher fill factor was obtained for the films with x = 2.25 Supporting Information), indicating similar dielectric constants (Figure 6c), probably due to improved interfacial contact. of the active layers with different phase compositions. The best-performing devices based on CuAgBiI5 exhibited The effect of x on the photovoltaic performance of the a Voc = 0.49 V, Jsc = 3.93 mA cm−2, FF = 0.63, and a corre- CuxAgBiI4+x materials was analyzed for 24 devices of each type sponding PCE of 1.21%, and Cu2AgBiI6 exhibited a Voc = 0.71 V, to ensure consistency of the established trends (Figure 6 and Jsc = 5.3 mA cm−2, FF = 0.65, and a corresponding PCE of Table S5, Supporting Information). As x increased from 1 to 2.45%. Considering the S–Q limit, CuAgBiI5 and Cu2AgBiI6 1.75, open-circuit voltage improved steadily though to a small have the potential to provide up to ≈24% and ≈25% PCE at extent from 0.46 ± 0.03 to 0.53 ± 0.06 V, but abruptly changed to the radiative limit, with a Voc of 1.64 and 1.59 V, Jsc of 16.0 and a significantly higher value of Voc = 0.70 ± 0.03 V at x = 2, i.e., 17.2 mA cm−2, and a FF of 0.92. While the FF obtained for the when a pure Cu2AgBiI6 phase was formed (Figure 6a). In con- devices is reasonably good with a <25% loss, there is a sig- trast, the short-circuit current density followed a close to linear nificant loss in Voc and Jsc.[38] This could be due to a range of dependence on x within the 1–2 range changing from 3.5 ± 0.3 factors, including low diffusion length even while having a to 5.2 ± 0.2 mA cm−2 (Figure 6b). The fill factor also improved longer carrier lifetime, implying possibly low intrinsic carrier from 0.59 ± 0.05 at x = 1 to 0.67 ± 0.04 at x = 2 (Figure 6c) and mobilities, flat bands with not favorable electronic structure, did not demonstrate any abrupt changes upon the formation of and higher built-in potential due to improper band alignment a pure 2D structure. These trends were corroborated by the PCE with the charge-transport layers. Most of these constraints can Adv. Energy Mater. 2022, 12, 2201482 2201482 (9 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de be resolved by employing thinner compact layers and utilizing better-aligned charge transport layers, e.g., an ETM with a more positive conduction band. 2.4. Environmental Stability When stored in air under ambient conditions (24 ± 2 °C, rela- tive humidity 10–20%) and diffuse light, the CuxAgBiI4+x films with x ≤ 1.5 (supported on m-TiO2|c-TiO2|FTO) suffer visually detectable degradation, viz. change in color from dark purple to dark red and then to yellow on a timescale of days (Figure 7a). While the XRD patterns of the aged material still exhibited the peaks associated with the copper-silver-bismuth iodide phase(s), an unknown peak at 10.7° emerged, and the intensity of the (101) peak reduced slightly during storage in air (Figure 7a). Contrasting this behavior, no visible changes were observed for the films with x > 1.5 and containing a higher amount of 2D Cu2AgBiI6 (Figure 7b). The significantly higher stability of the 2D structure as compared to the 3D one is consistent with the obser- vations reported by Sansom et al.[24] Interestingly, these differences in the stability of materials were not translated to the long-term storage stability of the non- encapsulated devices under ambient conditions and diffuse light, which did not lose significant performance over 240 days (Figure 7c). During the initial 100 days, both the CuAgBiI5 and Cu2AgBiI6 devices retained over 90% of the efficiency; however, over the next 140 days, the CuAgBiI5 devices lost almost 15% of their initial efficiency. The Cu2AgBiI6 devices retained over 80 % of the initial efficiency after these 8 months. The improve- ment in the performance during the initial 7 days of storage might be associated with the beneficial surface passivation that suppresses the ion-migration-induced defects, although there might be other factors contributing.[39] It was also found that devices with CuAgBiI5 as an active layer demonstrated loss in the performance under continuous irradiation, which was not the case for the Cu2AgBiI6-based solar cells, which were per- fectly stable in operation (Figure 7d). The impressive environmental stability of the devices indi- cates that other cell components provide sufficiently adequate protection of the active layer from the degradation-inducing reactions with the ambient atmosphere. At the same time, the Cu2AgBiI6-based devices with P3HT as a hole-transporting material suffered degradation at the gold-HTM interface, probably due to the Ag+ and Bi+ ion migration (Figure S28, Supporting Information).[40] Importantly, this unfavorable phenomenon was not observed with the carbon electrode-based devices and solar cells with a Au electrode and other HTMs. 3. Conclusions Figure 6. Influence of the composition of CuxAgBiI4+x on the major The present work presents a comprehensive study on the compo- photovoltaic parameters of the C|spiro-OMeTAD|CuxAgBiI4+x|m-TiO2|c- sitional control of the phase, dimensionality, and optoelectronic TiO2|FTO solar cells (aperture 0.16 cm2): a) open-circuit voltage, b) short- properties of the solution-processed CuxAgBiI4+x thin films. circuit current density, c) fill factor, d) power conversion efficiency, and We demonstrate a correlation between x and the ratio of the e) the steady-state power output. Data were derived from a–d) J–V curves and e) quasi-steady-state measurements recorded under 1 sun AM1.5G two major phases—3D CuAgBiI5 and 2D Cu2AgBiI6, and cor- irradiation (as demonstrated in Figure 5) for 24 independent solar cells responding changes in the light absorption. The latter is sig- of each type. nificantly better for CuxAgBiI4+x as compared to the archetypal Adv. Energy Mater. 2022, 12, 2201482 2201482 (10 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de compound. This partially addresses the impossibility of the highly desired but unachievable direct Br−-to-I− anionic modifica- tion in Cs2AgBiBr6. Our results also demonstrate that the change in the dimensionality in the CuxAgBiI4+x system from 3D to 2D decreases the bandgap to expand the light absorption by the material to almost the whole visible spectrum and downshifts the valence band level. Another feature of the present work is the first demonstra- tion of a solution-based method for synthesizing thin films of CuxAgBiI4+x with controlled composition by the hot-casting approach. Hot-casting-assisted spin-coating improves crystal- lization control and produces uniformly covered and well- connected CuxAgBiI4+x films with suppressed surface defects. Solar cells produced using this method demonstrated high storage stability and delivered power conversion efficiencies of up to 1.3% and 2.4% for CuAgBiI5 and Cu2AgBiI6-based devices, respectively. These are the best performances of these materials reported to date. Nevertheless, the advantages of Cu2AgBiI6 compared to Cs2AgBiBr6 demonstrated herein are not as significant as the differences in the optoelectronic prop- erties of the two materials. One aspect that might be optimized in the future is the thickness of the active Cu2AgBiI6 layer to better match the minority carrier diffusion lengths in the range of 40–50 nm measured herein. The use of dopants to improve the carrier mobility and appropriate charge transport layers that could match the positive conduction band of CuxAgBiI4+x is expected to address the voltage and current loss issues. Finally, computational and experimental exploration of related AxBMX4+x compositions might unveil new promising mate- rials, of which compounds with a wider direct bandgap might be used as components of the tandem solar cells. 4. Experimental Section Materials: Copper(I) iodide (≥99.5%), silver(I) iodide (99%), bismuth(III) iodide (99%), anhydrous dimethylsulfoxide (DMSO; 99.9%), anhydrous N,N-dimethylformamide (DMF; 99.8%), anhydrous chlorobenzene (CBz; 99.9%) anhydrous toluene (99.8%), titanium diisopropoxide bis(acetylacetonate) solution (TAA) (75 wt% in isopropanol), lithium bis(trifluoromethane)sulfonimide (LiTFSI; >99%), 4-tert-butylpyridine (t-BP; 98%), tris-2-(1H-pyrazol-1-yl)-4-tert- butylpyridine)cobalt(III) tri[bis(trifluoromethane)sulfonimide] (FK209; 98%), and poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA; Mn = 7000–10   000) were purchased from Sigma-Aldrich. 2,2′,7,7′- Tetrakis(N,N-di-p-methoxy phenylamino)-9,9 spirobifluorene (spiro- OMeTAD) was purchased from Luminescence Technology Corp. 2,2′,7,7′-Tetrakis(N,N-di-p-methoxy phenylamino)-9,9 spirobifluorene bis(trifluoromethane) sulfonimide (spiro-OMeTAD(TFSI)2) and (6,6′-bis(1,1-di(pyridin-2-yl)ethyl)-2,2′-bipyridine)-iron(II/III) Figure 7. XRD patterns and photographs of aged a) CuAgBiI5 and trifluoromethanesulfonate ([Fe(bpyPY4)](OTf)2+x) were synthesized b) Cu2AgBiI6 thin films deposited on glass. c) Evolution of normalized following the reported procedures.[32] Titania particle paste (average PCE of the C|spiro-OMeTAD|CuxAgBiI4+x|m-TiO2|c-TiO2|FTO solar cells size 30 nm; 30 NRD) was purchased from GreatCell Solar Materials (aperture 0.16 cm2) with x = 1 (light blue) and 2 (orange) stored in air Ltd. Absolute ethanol was purchased from UNIVAR solutions. Poly(3- under diffuse light. d) Photocurrent transients recorded at potentials of hexylthiophene-2,5-diyl (P3HT) was purchased from Rieke Metals. Water maximal power derived from J–V data for the devices tested with an aper- purified by reverse osmosis (quoted resistivity 1 MΩ cm) was used in ture of 0.16 cm2. all experimental procedures whenever water is mentioned. High purity N2 (≤ 2 ppm O2, ≤ 0.1 ppm H2O) and Ar (< 3 ppm O2, < 0.2 ppm H2O) were used. Cs2AgBiBr6 double-perovskite, demonstrating how the simul- Fluorine-doped tin (IV) oxide (FTO) coated glass substrates with the taneous A-site cationic and anionic substitution can deliver a size of 2.5 cm × 2.5 cm having a sheet resistance of 15 Ω sq−1 were new photovoltaic material with improved optoelectronic char- purchased from Yingkou Shangneng Photoelectric material Co., Ltd. acteristics while maintaining the high robustness of the parent Microscope-glass substrates of 1–1.2 mm thickness used for the physical Adv. Energy Mater. 2022, 12, 2201482 2201482 (11 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de characterization were purchased from Livingstone International Pty. Ltd. 80 µm. The resulting electrodes were immersed in stirred ethanol for 2 vol% Hellmanex III detergent solution was purchased from Helma 30 min to remove additives in the commercial carbon paste and dried in analytics. Jelcon CH-8 carbon paste was purchased from JuJo Chemicals the ambient environment for 2 h at 65 °C. After cooling, the electrodes Co. Ltd. for the carbon electrode preparation. Silver paste (CP1 30J) from were cut to a size of 0.5 cm × 0.5 cm and applied to four devices ChipQuick and a 12 µm thick domestic grade all-purpose aluminum simultaneously using a conventional hydraulic press (Metalmaster foil from Castaway Caterers Foil Australia were used as purchased to PP-10HD) at 20 MPa for 2 min; to avoid breaking the glass substrate, improve the contact of carbon electrodes.[33] the cell and electrode were sandwiched between ≈ 1.5 mm thick rubber Photovoltaic Device Fabrication: Cleaning of the pre-patterned FTO- (Translucent silicon rubber RS Component Pty Ltd). Gold electrodes coated glass substrates included sequential ultrasonication (Elmasonic were produced by evaporating a 80 nm thick Au layer (DDHightech S 300H bath at an operating power of 1500 W) in a 2 wt% Helmanex GCMO3CR; initial vacuum level 8 × 10-6 Torr) over the HTM layer with surfactant solution, water, and ethanol (15 min in each medium), drying a constant rotation of the stage; customized metal photomask was under N2 flow, and plasma treatment at 1000 mTorr for 10 min (Harrick applied to produce solar cells with an active area of 0.16 cm2. Plasma cleaner PDC-002 at maximum power). A 15 nm thick compact Characterization: All samples used for characterization were deposited titania blocking layer (c-TiO2) was formed on a cleaned FTO substrate by the optimized hot-casting assisted spin-coating method on TiO2 or surface using a custom-made automatic spray pyrolysis system by glass supports as relevant. depositing a 5 vol% TAA precursor in anhydrous isopropanol at 500 °C X-ray diffraction (XRD) patterns were obtained using a D2 Phaser (heated by high-precision hot-plate Harry Gestigkeit GmbH PZ28-3TD) BRUKER diffractometer with a Cu-Kα radiation (λ = 1.5406 Å) source for 30 min. Upon cooling to ambient temperature, cTiO2|FTO substrates operating at 30 kV and 15 mA at a scan rate of 0.5° min-1. For the were covered with a mesoporous titania layer (m-TiO2; thickness analysis, freshly produced films with a thickness of 200 nm on glass ≈ 150 nm) by spin-coating (4000 rpm for 20 s with a Laurell-150 mm slides were employed. Vesta software was used to generate crystal spin-coater) 75 µL of a 30 nm TiO2 particle paste dispersed in ethanol, structures displayed in Figure 2b. followed by drying at 100 °C and sintering at 500 °C in air on a high X-ray photoelectron spectroscopic (XPS) analysis was performed precision hot-plate (Harry Gestigkeit GmbH PZ28-3TD). with a Nexsa Surface Analysis System (Thermo Fisher Scientific) on The CuxAgBiI4+x precursor solutions were prepared by dissolving the 200 nm thick copper silver bismuth iodide films spin-coated onto copper(I), silver(I), and bismuth(III) iodides to achieve the target 2.5 cm × 2.5 cm c-TiO2|FTO substrates. Analysis was undertaken with [CuxAgBiI4+x] concentration of 0.45 m in a DMSO:DMF solvent mixture a monochromated Al Kα X-rays (1486.6 eV) source at a power of 72 W (optimized ratio 3:1 vol) for 1 h under vigorous stirring at 100 °C. The (12 kV × 6 mA) and a hemispherical analyzer operating in a fixed analyzer obtained solutions were filtered once cooled down and were spin-coated transmission mode. During analysis, the base pressure was typically less onto 2.5 cm × 2.5 cm FTO-substrates by a two-step programmed spin- than 5.0 × 10−9 mbar. Survey and high-resolution scans were recorded coating, where initially the solution was spread by spinning at 1000 rpm at analyzer pass energies of 150 and 50 eV, respectively. Survey scans for 10 s, followed by 6000 rpm for 20 s. For the gas-assisted deposition, Ar were performed with a step size of 1.0 eV and a dwell time of 10 ms. gas was flown at 20 mL min−1 in the center of the spun sample after 2 s in High-resolution spectra were recorded with 0.1 eV step size and dwell the second cycle for 15 s. For the antisolvent-assisted deposition, 100 µL of time of 50 ms, typically yielding a full width at half of the maximum of isopropanol, chlorobenzene, diethyl ether, or toluene were decanted into 0.8–0.9 eV for the Ag 3d5/2 peak for the performance tests. A low-energy the center of the substrate 5 s before the end of the spinning. For the dual-beam (ion and electron) flood gun was used to compensate for dynamic hot-casting, the films and substrates were preheated at varying surface charging. Samples were analyzed at a nominal photoelectron temperatures (60 °C ≤ T ≤ 120 °C) before spin-coating and the solution emission angle of 0°. was deposited at 2 s of the initial spinning cycle. Data processing was performed using Thermoscientific Avantage The resultant films were annealed in an N2-filled glove box at 80 °C processing software version 5.9925. All elements present were identified for 2 min, followed by additional heating at 150 °C for 5 min with an IKA from survey spectra. The atomic concentrations of the detected elements RCT basic titanium hot-plate, and then cooled to ambient temperature were calculated using integral peak intensities and the sensitivity factors (24 ± 2 °C) naturally. The 2.5 cm × 2.5 cm CuxAgBiI4+x + mTiO2|c- supplied by the manufacturer. Binding energies were corrected to the C TiO2|FTO films were then modified with a hole-transporting material 1s peak at 284.8 eV (aliphatic hydrocarbon). The accuracy associated with (HTM) layer. Examined HTMs included PTAA, P3HT, spiro-OMeTAD quantitative XPS is ≈ 10–15%. Precision, i.e., reproducibility, depends on doped with FK209 cobalt complex in the presence of LiNTf2 and tBP, the signal: noise ratio and is usually better than 5%. spiro-OMeTAD doped with spiro-OMeTAD(TFSI)2, and [Fe(bpyPY4)] Scanning electron micrographs (SEM) were collected using a (OTf)2+x. PTAA was deposited by spin-coating 25 µL of a 10 mg mL-1 FEI Magellan 400 XHR FEGSEM at a voltage of 3–5 kV and spot size toluene solution at 1800 rpm for 30 s, then annealing at 60 °C. P3HT 2. Freshly deposited CuxAgBiI4+x + m-TiO2|c-TiO2|FTO samples were was deposited by spin-coating 20 µL of a 15 mg mL−1 dichlorobenzene immobilized on a specimen stub using double-sided carbon tape, solution at 2000 rpm for 30 s and subsequently annealing at 135 °C. and the copper tape was employed to create electrical contact for top- Spiro-OMeTAD was deposited by spin-coating 30 µL of a solution view imaging. For side-view imaging, complete Ag|HTM|CuxAgBiI4+x + containing 67.2 × 10-3 m spiro-OMeTAD dissolved in chlorobenzene with m-TiO2|c-TiO2|FTO devices were cracked using a glass cutter and pliers, 20 mmol tBP (28.8 µL), 20 mmol LiNTf2 (17 µL of 520 mg mL−1 solution and samples were immobilized in a TED PELLA, Inc. cross-sectional in acetonitrile), and 160 mmol FK209 (8 µL of 300 mg mL−1 solution side-view holder and coated with iridium to avoid charging. A Bruker in acetonitrile) at 3000 rpm for 30 s. For the spiro-OMeTAD2+-doped Quantax silicon drift detector was used to record energy dispersive X-ray solution, spiro-OMeTAD and spiro-OMeTAD(TFSI)2 with different mole spectra at a voltage of 10 kV and a spot size of 4.5. fractions were mixed in chlorobenzene while the total concentration Inductively coupled plasma optical emissions spectroscopic remained at 67.2 × 10-3 m. This solution was vigorously stirred while (ICP-OES) analysis was undertaken using a Perkin Elmer's Avio 200 ICP heating at 70 °C for 30 min to fully dissolve the chemicals and cooled instrument. Calibration samples were made by diluting a multi-element to room temperature before spin coating. [Fe(bpyPY4)](OTf)2 (21.9 mg) commercial standard solution (Sigma Aldrich). Characterization and [Fe(bpyPY4)](OTf)2.9 (25.6) were dissolved in dichlorobenzene samples were prepared by immersing the CuxAgBiI4+x films deposited (1 mL), and 25 µL of this solution were spin-coated at 2000 rpm for 30 s. onto the glass in 1 mL of 2 wt% HNO3 aqueous solution overnight, The devices were completed by depositing a counter-electrode, either which resulted in complete dissolution of the materials of interest. carbon-based or Au. Adhesive carbon electrodes with a thickness of Thermogravimetric analysis (TGA) was conducted using a Mettler ≈25 µm were prepared following the reported procedure.[33] Briefly, the Toledo Thermogravimetric Analyzer 2 Star System. Nitrogen gas was electrode was deposited onto a copper-coated aluminum substrate by used during the analysis at a 20 mL min−1 rate with a temperature ramp automated doctor blading (MTVmesstechnik CX202) of ≈ 5 mg cm−2 rate of 10 °C min−1 from 25 to 500 °C. Samples were heated in 40 µL of the Jelcon CH-8 paste at a rate of 9 mm s−1 and height adjusted to alumina crucibles. Adv. Energy Mater. 2022, 12, 2201482 2201482 (12 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de UV–vis spectra were recorded for the freshly-prepared CuxAgBiI4+x + where a scan rate of 0.100 V s−1 was used. No significant differences in m-TiO2|c-TiO2|FTO samples in transmission and reflectance modes the data obtained at these two scan rates were observed. The devices using a Perkin Elmer Lambda 1050 spectrophotometer equipped with an were illuminated for 10 s before testing to stabilize them. The steady- integrating sphere. UV-vis spectra of the m-TiO2|c-TiO2|FTO substrates state efficiency was calculated by measuring the photocurrent at a were used to correct the absorption data. The absorption coefficients constant potential matching the highest power voltage in the reverse J–V were calculated using the equation reported in the literature, using the scan. film thickness derived from the SEM cross-section images.[29] The illumination light source was chopped at 8 Hz, and the electrical Photoluminescence measurements demonstrated in Figure 4 signal was collected by a lock-in amplifier (Merlin Radiometric Lock-In and Figure S13 (Supporting Information) were carried out using an Amplifier, Newport) under short-circuit conditions. A calibrated silicon Edinburgh Instruments Ltd. FLSP920 time-correlated single-photon diode (70356-70316NS, Newport) with a known spectral response counting (TCSPC) spectrometer with a 465.8 nm pulsed diode laser spectrum was used as a reference. excitation source (EPL-475, Edinburgh Instruments Ltd.) having a The incident photon to current conversion efficiency (IPCE) spectra pulse width of ≈100 ps and a peak fluence of ≈0.2 nJ cm−2 (2 W cm2). were measured with a Keithley 2400 Source Meter and an Oriel Corner- Decay curves and spectra were measured using a pulse repetition rate stone 260¼ m monochromator under 300 W Newport xenon lamp of 50 kHz and 1 MHz, respectively. Luminescence was detected via a illumination. A calibrated silicon cell (Peccell Technologies) with known grating monochromator and photomultiplier tube. Emission spectra spectral response spectra was used to measure the monochromatic were corrected for the wavelength response of the detection system. photon flow. Excitation and luminescence detection of samples were from the coated A Zahner Zennium electrochemical workstation with a frequency side of samples consisting of spin-coated films on 25 mm × 25 mm response analyzer module was used to acquire impedance spectroscopic glass substrates. data. A sinusoidal potential perturbation with an amplitude of 0.010 V Steady-state photoluminescence measurements displayed in was used to make measurements throughout a frequency range of Figures S14 and S15 (Supporting Information) were carried out using a 2 MHz to 0.1 Hz. Solar cells were illuminated using a white light-emitting Kymera 328i spectrometer with a Zylar 4.2 PLUS sCMOS and 150 lines diode during measurements. The output intensity was calibrated to mm−1 grating for samples mounted in a Nikon Ti2 microscope base with generate the same photocurrent obtained using the Abet Technologies a 130 W mercury lamp. The light was monochromated using a Semrock Sun 3000 class AAA source under 1 sun irradiation. The impedance data bright line filter cube to provide either 405 or 488 nm excitation. Where were modelled using the Z-View software. relevant, spectra intensities were corrected using published efficiency losses for the various optics within the microscope and spectrometer. Samples were exposed to air for 2–4 min before measuring the PL intensity at specific intervals. Supporting Information Photoelectron spectroscopy in air (PESA) measurements were accomplished using a Riken Kekei AC-2 spectrometer with a power Supporting Information is available from the Wiley Online Library or intensity range of 20 to 50 nW. CuxAgBiI4+x films of 200 nm thickness from the author. were deposited onto 2.5 cm × 2.5 cm c-TiO2|FTO substrates for analysis. Electron beam-induced current/scanning electron microscopy (EBIC/ SEM) experiments were performed in a FEI Nova NanoSEM FEG-SEM equipped with a Deben Type 31 SEM specimen current EBIC amplifier at a Acknowledgements relatively low accelerating voltage of 2 kV and beam currents of 30–50 pA The authors are highly grateful to Dr. Jiangfeng Lu (Wuhan University without external bias to achieve a high-resolution probe with a decent of Technology) for invaluable discussions, Mr. Philippe Holzhey (Oxford signal-to-noise ratio while preserving low injection EBIC conditions.[40] University) for assistance with PL measurements, Mr. D. Vowles Surface recombination can occur due to sample preparation and (Monash University) for assistance with EBIC experiments, and to the attachment damage, limiting EBIC observations. As a result, cross- Monash Centre for Electron Microscopy and Monash X-ray Platform section samples were prepared using a high-precision glass cutter and for providing access to their characterization facilities used for some connected by wrapping Kapton tape around the substrate except for the parts of this work. The authors acknowledge funding of this work by imaging region. When applied to carbon electrodes, these preparation the Australian Government through the Australian Centre for Advanced procedures were ineffective, which urged us to undertake experiments Photovoltaics (ACAP), the Australian Renewable Energy Agency with the gold electrode-based solar cells. Because the samples had to (ARENA), and the Australian Research Council through the Centre of be physically placed on the EBIC stage, the measured device cross- Excellence in Exciton Science (CE170100026) and Future Fellowship to sections for each sample preparation procedure were exposed to air ANS (FT200100317). Responsibility for the views, information, or advice for several minutes before loading into the SEM. In addition, because expressed herein is not accepted by the Australian Government. SRR these materials were susceptible to electron beam exposure, each cross- acknowledges the support from “laCaixa” Foundation (ID 100010434; section area could only be scanned once. As a result, the measured EBIC LCF/BQ/PI20/11760024). declined after each initial EBIC scan regardless of the sample. Line/area Open access publishing facilitated by Monash University, as part of intensity gradient profile in arbitrary units proportional to electric current the Wiley – Monash University agreement via the Council of Australian was generated from the obtained EBIC images using ImageJ software. University Librarians. Photovoltaic Measurements: A system developed by Surmiak et al.[41] for high throughput characterization of solar cells was used in the present study. A small amount of solder was added to the device edges to connect the ports and the cell electrodes. The devices were lit by an Abet Technologies Sun 3000 class AAA solar simulator calibrated at AM Conflict of Interest 1.5 (100 mW cm2) intensity through a 0.16 cm2 aperture mask, and J–V The authors declare no conflict of interest. data were recorded using a 16-channel Bio-Logic VMP3 electrochemical workstation. No specific pre-treatment or pre-conditioning procedures were applied, as solar cells did not change their performance while being stored under ambient conditions. Current density (J)–voltage (V) scans Data Availability Statement were recorded in both backward (0.8 to –0.1 V) and forward (–0.1 to 0.8 V) directions with 0.010 V steps and settling periods of 100 ms, viz. at The data that support the findings of this study are available from the nominal scan rate 0.200 V s−1, except for the data in Figures S20 and S22 corresponding author upon reasonable request. Adv. Energy Mater. 2022, 12, 2201482 2201482 (13 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de Keywords [11] a) Z. Li, S. P. Senanayak, L. Dai, G. Kusch, R. Shivanna, Y. Zhang, D. Pradhan, J. Ye, Y.-t. Huang, H. Sirringhaus, R. A. Oliver, CuAgBiI5, Cu2AgBiI6, solar cells, thin film N. C. Greenham, R. H. Friend, R. L. Z. Hoye, Adv. Funct. Mater. 2021, 31, 2104981; b) J. Leveillee, G. Volonakis, F. Giustino, J. Phys. Received: April 30, 2022 Chem. Lett. 2021, 12, 4474; c) H. Lei, D. Hardy, F. Gao, Adv. Funct. Revised: June 1, 2022 Mater. 2021, 31, 2105898. Published online: July 17, 2022 [12] G. Longo, S. Mahesh, L. R. V. Buizza, A. D. Wright, A. J. Ramadan, M. Abdi-Jalebi, P. K. Nayak, L. M. Herz, H. J. Snaith, ACS Energy Lett. 2020, 5, 2200. [13] a) F. Igbari, R. Wang, Z.-K. Wang, X.-J. Ma, Q. Wang, K.-L. Wang, [1] a) A. K. Jena, A. Kulkarni, T. Miyasaka, Chem. Rev. 2019, 119, 3036; Y. Zhang, L.-S. Liao, Y. Yang, Nano Lett. 2019, 19, 2066; b) N.-G. Park, M. Grätzel, T. Miyasaka, K. Zhu, K. Emery, Nat. Energy b) R. L. Z. Hoye, L. Eyre, F. Wei, F. Brivio, A. Sadhanala, S. Sun, W. Li, 2016, 1, 16152; c) N. G. Park, K. Zhu, Nat. Rev. Mater. 2020, 5, 333. K. H. L. Zhang, J. L. MacManus-Driscoll, P. D. Bristowe, R. H. Friend, [2] a) F. Igbari, Z. K. Wang, L. S. Liao, Adv. Energy Mater. 2019, A. K. Cheetham, F. Deschler, Adv. Mater. Interfaces 2018, 5, 1800464; 9, 1803150; b) Z. Xiao, Z. Song, Y. Yan, Adv. Mater. 2019, 31, c) W. Gao, C. Ran, J. Xi, B. Jiao, W. Zhang, M. Wu, X. Hou, Z. Wu, 1803792; c) W. Ke, M. G. Kanatzidis, Nat. Commun. 2019, 10, 965; ChemPhysChem 2018, 19, 1696; d) E. Greul, M. L. Petrus, A. Binek, d) N. Glück, T. Bein, Energy Environ. Sci. 2020, 13, 4691. P. Docampo, T. Bein, J. Mater. Chem. A 2017, 5, 19972. [3] a) G. Volonakis, M. R. Filip, A. A. Haghighirad, N. Sakai, B. Wenger, [14] Q. Li, Y. Wang, W. Pan, W. Yang, B. Zou, J. Tang, Z. Quan, Angew. H. J. Snaith, F. Giustino, J. Phys. Chem. Lett. 2016, 7, 1254; Chem., Int. Ed. 2017, 56, 15969. b) M. Wang, W. Wang, B. Ma, W. Shen, L. Liu, K. Cao, S. Chen, [15] K. z. Du, W. Meng, X. Wang, Y. Yan, D. B. Mitzi, Angew. Chem., Int. W. Huang, Nano-Micro Lett. 2021, 13, 62. Ed. 2017, 56, 8158. [4] a) L. Zhang, K. Wang, B. Zou, ChemSusChem 2019, 12, 1612; [16] a) N. Pai, J. Lu, M. Wang, A. S. R. Chesman, A. Seeber, b) N. C. Miller, M. Bernechea, APL Mater. 2018, 6, 084503; c) F. Ünlü, P. V. Cherepanov, D. C. Senevirathna, T. R. Gengenbach, M. Deo, S. Mathur, T. Kirchartz, A. Kulkarni, J. Phys. D: Appl. Phys. N. V. Medhekar, P. C. Andrews, U. Bach, A. N. Simonov, J. Mater. 2022, 55, 113002; d) N. Tewari, S. B. Shivarudraiah, J. E. Halpert, Chem. A 2020, 8, 2008; b) M. Keshavarz, E. Debroye, M. Ottesen, Nano Lett. 2021, 21, 5578; e) S. Attique, N. Ali, S. Ali, R. Khatoon, C. Martin, H. Zhang, E. Fron, R. Küchler, J. A. Steele, M. Bremholm, N. Li, A. Khesro, S. Rauf, S. Yang, H. Wu, Adv. Sci. 2020, 7, 1903143; J. Van de Vondel, H. I. Wang, M. Bonn, M. B. J. Roeffaers, f) R. L. Z. Hoye, J. Hidalgo, R. A. Jagt, J.-P. Correa-Baena, T. Fix, S. Wiedmann, J. Hofkens, Adv. Mater. 2020, 32, 2001878; c) J. Li, J. L. MacManus-Driscoll, Adv. Energy Mater. 2022, 12, 2100499. J. Duan, J. Du, X. Yang, Y. Wang, P. Yang, Y. Duan, Q. Tang, ACS [5] a) Y. Kim, Z. Yang, A. Jain, O. Voznyy, G. H. Kim, M. Liu, Appl. Mater. Interfaces 2020, 12, 47408; d) H. Lei, D. Hardy, F. Gao, L. N. Quan, F. P. García de Arquer, R. Comin, J. Z. Fan, Adv. Funct. Mater. 2021, 31, 2105898. E. H. Sargent, Angew. Chem., Int. Ed. 2016, 55, 9586; b) I. Turkevych, [17] R. Nie, R. R. Sumukam, S. H. Reddy, M. Banavoth, S. I. Seok, Energy S. Kazaoui, E. Ito, T. Urano, K. Yamada, H. Tomiyasu, H. Yamagishi, Environ. Sci. 2020, 13, 2363. M. Kondo, S. Aramaki, ChemSusChem 2017, 10, 3754; c) V. Pecunia, [18] X.-G. Zhao, J.-H. Yang, Y. Fu, D. Yang, Q. Xu, L. Yu, S.-H. Wei, J. Zhao, C. Kim, B. R. Tuttle, J. Mei, F. Li, Y. Peng, T. N. Huq, L. Zhang, J. Am. Chem. Soc. 2017, 139, 2630. R. L. Z. Hoye, N. D. Kelly, S. E. Dutton, K. Xia, J. L. MacManus- [19] a) Z. Xiao, W. Meng, J. Wang, Y. Yan, ChemSusChem 2016, 9, 2628; Driscoll, H. Sirringhaus, Adv. Energy Mater. 2021, 11, 2003968. b) N. Zarabinia, R. Rasuli, Energy Sources, Part A 2021, 43, 2443. [6] a) N. Pai, J. Lu, T. R. Gengenbach, A. Seeber, A. S. R. Chesman, [20] P. Vishnoi, R. Seshadri, A. K. Cheetham, J. Phys. Chem. C 2021, 21, L. Jiang, D. C. Senevirathna, P. C. Andrews, U. Bach, Y.-B. Cheng, 11756. A. N. Simonov, Adv. Energy Mater. 2019, 9, 1803396; b) Z. Hu, [21] S. E. Creutz, E. N. Crites, M. C. De Siena, D. R. Gamelin, Nano Lett. Z. Wang, G. Kapil, T. Ma, S. Iikubo, T. Minemoto, K. Yoshino, 2018, 18, 1118. T. Toyoda, Q. Shen, S. Hayase, ChemSusChem 2018, 11, 2930; [22] M. Li, F. Li, J. Gong, T. Zhang, F. Gao, W.-H. Zhang, M. Liu, Small c) K. W. Jung, M. R. Sohn, H. M. Lee, I. S. Yang, S. D. Sung, J. Kim, Struct. 2022, 3, 2100102. E. Wei-Guang Diau, W. I. Lee, Sustainable Energy Fuels 2018, 2, 294; [23] H. C. Sansom, G. Longo, A. D. Wright, L. R. V. Buizza, S. Mahesh, d) W.-X. Chai, L.-M. Wu, J.-Q. Li, L. Chen, Inorg. Chem. 2007, 46, B. Wenger, M. Zanella, M. Abdi-Jalebi, M. J. Pitcher, M. S. Dyer, 1042. T. D. Manning, R. H. Friend, L. M. Herz, H. J. Snaith, J. B. Claridge, [7] A. Kulkarni, F. Ünlü, N. Pant, J. Kaur, C. Bohr, A. K. Jena, S. Öz, M. J. Rosseinsky, Sansom, et al., J. Am. Chem. Soc. 2021, 143, M. Yanagida, Y. Shirai, M. Ikegami, K. Miyano, Y. Tachibana, 3983. S. Chakraborty, S. Mathur, T. Miyasaka, Sol. RRL 2021, 5, 2100077. [24] H. C. Sansom, L. R. V. Buizza, M. Zanella, J. T. Gibbon, M. J. Pitcher, [8] a) B. Zhang, Y. Lei, R. Qi, H. Yu, X. Yang, T. Cai, Z. Zheng, Sci. China M. S. Dyer, T. D. Manning, V. R. Dhanak, L. M. Herz, H. J. Snaith, Mater. 2019, 62, 519; b) L. Wang, Y. Bao, S. Wang, F. Wang, C. Xie, J. B. Claridge, M. J. Rosseinsky, Inorg. Chem. 2021, 60, 18154. K. T. Butler, X. Fan, Cryst. Growth Des. 2021, 21, 2850; c) N. Qu, [25] A. Mutlu, C. Zafer, Sol. Energy 2022, 234, 190. Y. Lei, X. Yang, X. Hu, W. Zhao, C. Zhao, Z. Zheng, J. Mater. Chem. [26] D. Marongiu, M. Saba, F. Quochi, A. Mura, G. Bongiovanni, C 2020, 8, 8451. J. Mater. Chem. C 2019, 7, 12006. [9] a) C. N. Savory, A. Walsh, D. O. Scanlon, ACS Energy Lett. 2016, 1, [27] F. Zhang, Z. Hu, B. Zhang, Z. Lin, J. Zhang, J. Chang, Y. Hao, ACS 949; b) Y. Cai, W. Xie, Y. T. Teng, P. C. Harikesh, B. Ghosh, P. Huck, Appl. Mater. Interfaces 2022, 14, 18498. K. A. Persson, N. Mathews, S. G. Mhaisalkar, M. Sherburne, [28] T. C. T. Pham, H. S. Kim, K. B. Yoon, Science 2011, 334, 1533. M. Asta, Chem. Mater. 2019, 31, 5392. [29] N. Pai, J. Lu, D. C. Senevirathna, A. S. R. Chesman, T. Gengenbach, [10] a) A. H. Slavney, T. Hu, A. M. Lindenberg, H. I. Karunadasa, J. Am. M. Chatti, U. Bach, P. C. Andrews, L. Spiccia, Y.-B. Cheng, Chem. Soc. 2016, 138, 2138; b) M. Delor, A. H. Slavney, N. R. Wolf, A. N. Simonov, J. Mater. Chem. C 2018, 6, 2483. M. R. Filip, J. B. Neaton, H. I. Karunadasa, N. S. Ginsberg, [30] B. Tan, S. R. Raga, A. S. R. Chesman, S. O. Fürer, F. Zheng, ACS Energy Lett. 2020, 5, 1337; c) D. Bartesaghi, A. H. Slavney, D. P. McMeekin, L. Jiang, W. Mao, X. Lin, X. Wen, J. Lu, Y.-B. Cheng, M. C. Gelvez-Rueda, B. A. Connor, F. C. Grozema, H. I. Karunadasa, U. Bach, Adv. Energy Mater. 2019, 9, 1901519. T. J. Savenije, J. Phys. Chem. C 2018, 122, 4809; d) E. T. McClure, [31] M. Bernechea, N. C. Miller, G. Xercavins, D. So, A. Stavrinadis, M. R. Ball, W. Windl, P. M. Woodward, Chem. Mater. 2016, 28, 1348. G. Konstantatos, Nat. Photonics 2016, 10, 521. Adv. Energy Mater. 2022, 12, 2201482 2201482 (14 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH 16146840, 2022, 32, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202201482 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advenergymat.de [32] M. K. Kashif, I. Benesperi, R. A. Milhuisen, S. Meyer, J. Hellerstedt, [36] a) N. E. Courtier, Phys. Rev. Appl. 2020, 14, 24031; b) T. Kirchartz, D. Zee, N. W. Duffy, B. Halstead, M. S. Fuhrer, J. Cashion, Y.-B. Cheng, B. E. Pieters, J. Kirkpatrick, U. Rau, J. Nelson, Phys. Rev. B 2011, 83, L. Spiccia, A. N. Simonov, U. Bach, ACS Energy Lett. 2017, 2, 1855. 115209. [33] G. A. Sepalage, H. Weerasinghe, N. Rai, N. W. Duffy, S. R. Raga, [37] E. von Hauff, D. Klotz, J. Mater. Chem. C 2022, 10, 742. Y. Hora, M. Gao, D. Vak, A. S. R. Chesman, U. Bach, A. N. Simonov, [38] a) J. Diekmann, P. Caprioglio, M. H. Futscher, V. M. Le Corre, Adv. Mater. Technol. 2021, 7, 2101148. S. Reichert, F. Jaiser, M. Arvind, L. P. Toro, E. Gutierrez-Partida, [34] a) Z. Zhang, X. Li, X. Xia, Z. Wang, Z. Huang, B. Lei, Y. Gao, J. Phys. F. Peña-Camargo, C. Deibel, B. Ehrler, T. Unold, T. Kirchartz, Chem. Lett. 2017, 8, 4300; b) S. M. Jain, D. Phuyal, M. L. Davies, D. Neher, M. Stolterfoht, Sol. RRL 2021, 5, 2100219; b) W. Shockley, M. Li, B. Philippe, C. De Castro, Z. Qiu, J. Kim, T. Watson, W. C. Tsoi, H. J. Queisser, J. Appl. Phys. 1961, 32, 510. O. Karis, H. Rensmo, G. Boschloo, T. Edvinsson, J. R. Durrant, Nano [39] a) Y. Zhang, Y. Zhu, M. Hu, N. Pai, T. Qin, Y.-B. Cheng, U. Bach, Energy 2018, 49, 614; c) Q. Zhang, C. Wu, X. Qi, F. Lv, Z. Zhang, Y. Liu, A. N. Simonov, J. Lu, J. Phys. Chem. Lett. 2022, 13, 2792; S. Wang, B. Qu, Z. Chen, L. Xiao, ACS Appl. Energy Mater. 2019, 2, 3651; b) Y. Cho, J. Bing, H. D. Kim, Y. Li, J. Zheng, S. Tang, M. A. Green, d) F. Wei, Z. Deng, S. Sun, N. T. P. Hartono, H. L. Seng, T. Buonassisi, A. Wakamiya, S. Huang, H. Ohkita, A. W. Y. Ho-Baillie, ACS Appl. P. D. Bristowe, A. K. Cheetham, Chem. Commun. 2019, 55, 3721. Mater. Interfaces 2021, 13, 39178. [35] a) B. Wang, N. Li, L. Yang, C. Dall'Agnese, A. K. Jena, T. Miyasaka, [40] Barange, V. B. Chu, M. Nam, I.-H. Ahn, Y. D. Kim, I. K. Han, X.-F. Wang, J. Am. Chem. Soc. 2021, 143, 14877; b) B. Wang, N. Li, B. K. Min, D.-H. Ko, Adv. Energy Mater. 2016, 6, 1601114. L. Yang, C. Dall'Agnese, A. K. Jena, S.-i. Sasaki, T. Miyasaka, [41] M. A. Surmiak, T. Zhang, J. Lu, K. J. Rietwyk, S. R. Raga, H. Tamiaki, X.-F. Wang, J. Am. Chem. Soc. 2021, 143, 2207. D. P. McMeekin, U. Bach, Sol. RRL 2020, 4, 2000097. Adv. Energy Mater. 2022, 12, 2201482 2201482 (15 of 15) © 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH