Robot-Accelerated Exploration of Perovskite Crystallization and Reaction Networks

Hybrid inorganic/organic materials offer modular platforms for generating libraries of custom materials, whose properties can be tailored to applications in lasers, photovoltaics, and other optoelectronic devices. The large number of components in these complex materials, and the numerous parameters that specify their synthesis, result in an experimental space that is challenging to explore with conventional laboratory techniques. I will discuss the development of data-enabled strategies for discovering new materials through the combination of robotic synthesis, physical models, and machine learning. I will discuss our Robot-Accelerated Perovskite Investigation and Discovery (RAPID) workflow, which has performed over 15,000 distinct reactions. We used these large datasets to train machine learning classification models for the crystallization of metal halide perovskites and to validate new software pipelines for experimental design and capture. We demonstrate how such robotic techniques have accelerated the application of data approaches such as meta-learning, active learning, and model fusion to perovskite crystallization. Finally, we discuss how expansive robotic datasets allowed us to uncover the dual roles of water in perovskite crystallization and to elucidate the reaction networks that govern chemical transformations of perovskites.