University of Maryland,
Keywords: perovskites, AFM, functional imaging, photoluminescence, machine learning
Summary:The extraordinary power conversion efficiency of perovskite photovoltaics (currently > 23%) is hampered by material’s dynamic processes that occur when the perovskites are exposed to water, oxygen, temperature, bias, and light. Therefore, determining the effect of these stress factors, both individually and in combination, is critical to overcome device degradation. To resolve the contribution of each intrinsic and extrinsic parameter on materials’ optical and electrical responses we realize advanced scanning probe methods based on photoluminesce (PL) microscopy and atomic force microscopy (AFM). We investigate a series of hybrid perovskites, including MAPbI3, MAPbBr3, CsxFA1−xPb(IyBr1−y)3 , and triple cation Cs-mixed. Through environmental PL microscopy we identify a humidity-induced PL hysteresis that strongly depends on the Cs/Br ratio. Using fast Kelvin-probe force microscopy we quantify a dynamic open-circuit voltage (Voc) response as a function of perovskite chemical composition and illumination treatment, as will be discussed in details during the presentation. The individual and collective effects of the five aforementioned parameters on perovskites’ ability to recover are further investigated using a machine learning (ML) approach. Our functional imaging platform, combined with ML, can be expanded to test the stability of emerging perovskites, including Pb-free options, and novel perovskites for light absorbing and emitting applications.