AI-Guided Autonomous Materials Discovery for Organic Photovoltaics

In this work, we are pursuing an AI-guided, closed-loop approach to discover new organic photovoltaics (OPVs) with high device performance. Our work integrates autonomous synthesis, automated materials characterization, and AI-based molecular prediction methods in a closed-loop manner to discover new high-performance OPV molecules. We use a Bayesian optimization framework in which physicochemical descriptors of OPV candidates guide the search through a large molecular combinatorial space while maintaining a customizable tradeoff between exploitative and explorative sampling. Candidate OPV molecules suggested by the AI framework are prepared via automated synthesis methods using a “Lego-like” molecular building block-based approach relying on an iterative Suzuki cross-coupling reaction scheme. Following synthesis, the physical properties of candidate OPV molecules are characterized using an automated workflow, with experimental results passed back to the Bayesian optimizer for subsequent rounds of closed-loop materials discovery. Using this approach, our team has uncovered new high-performing OPV molecules after multiple rounds of AI prediction, synthesis, and materials characterization. Overall, our work aims to fill key knowledge gaps in understanding how molecular structure and properties encode OPV device performance. Moreover, our work is advancing the science of closed-loop autonomous discovery by learning how to synergistically integrate AI, automated synthesis, and automated testing and characterization into a common workflow. We aim to meet the “10-10” target (10% efficiency and 10-year stability for OPV materials) to make organic photovoltaics a commercial reality for next-generation energy capture applications.