Marina Leite1
University of California, Davis1
Marina Leite1
University of California, Davis1
Metal halide hybrid organic-inorganic perovskites display near-optimal optical properties for optoelectronic devices, including photovoltaics and light-emitting diodes. Yet, this class of material has not been commercialized due to often-seen degradation. Overall, establishing the correlation between structure and optoelectronic performance is essential and requires the implementation of in situ characterization techniques that can assess material-light interactions. At the nanoscale, we realize atomic force microscopy (AFM) methods that enable us to resolve how grains and their boundaries are affected by photo-induced processes, such as ion migration. Through Kelvin-probe force microscopy we quantify spatial variations in photo-voltage. In the realm of material stability, we use high-throughput and automated optical measurements, such as steady-state photoluminescence (PL) and PL lifetime to monitor the effects of environmental stressors in hybrid perovskites. Here, we submit the material to distinct values of temperature and relative humidity and monitor, <i>in situ</i>, their optical response. We implement machine learning neural networks to forecast material behavior. The vast hyper-parameter space of hybrid perovskites’ chemical composition is benefiting tremendously from artificial intelligence methods. To illustrate how one can gain understanding about the physical behavior of perovskites, we will showcase baseline models that apply machine learning, encompassing how to predict device power output as a function of temperature and how to find hidden correlations between property (electrical conductance) and dark-field optical images of material morphology.