Accelerating Solid-State Material Development Through AI-Driven Autonomous Synthesis Laboratories

Many technological advancements, including the development of novel electrode and electrolyte materials for energy storage, rely on the creation of new solid-state materials. Identifying promising material design strategies is only the initial step; synthesizing these materials poses a greater and more time-consuming challenge. To streamline the design–make–measure cycle, we have established an autonomous solid-state synthesis laboratory driven by AI and automation. This lab autonomously generates synthesis recipes from a comprehensive archive of historical literature, conducts experiments through robotic systems and automated instruments, and applies machine learning for data interpretation. Active learning algorithms guide the experimental direction to optimize results. This presentation will detail a case study on determining the solubility limit of fluorine in disordered rocksalt materials using the autonomous lab. Employing a Bayesian optimization workflow, we can minimize experimental trials and maximize information gain using an entropy search algorithm. This integrated approach not only refines existing synthesis methods but also advances the exploration and development of a wide array of oxide-based powder materials.