October 13
@
4:00 PM
–
5:00 PM
Professor Ying Diao
Department of Chemical and Biomolecular Engineering
University of Illinois, Urbana-Champaign
AI-guided Closed-loop Learning for Functional Molecule Discovery
Closed-loop discovery of functional molecules represents a new paradigm to new materials discovery. However, so far, AI-driven closed-loop discovery approaches have rarely yielded new chemical insight. I am leading a team to address this challenge under the Molecule Maker Lab Institute, one of the first AI Institutes in US. We connect autonomous synthesis, autonomous device printing and testing, and machine learning in a closed loop to drive molecular design of light harvesting small molecules towards high photostability. We combined blind Bayesian Optimization (BO) with physics-based ML models to expand the classical “closed loop optimization” method for a more informative “Closed-Loop Transfer” (CLT) paradigm capable of making scientific discoveries We uncovered a first principles understanding of the molecular determinants of photostability in which the triplet excited state manifold plays a critical role against literature findings. We further applied this strategy to understanding molecular design rules underpinning chiral emergence of conjugated polymers. Such understanding laid the foundation to developing next generation chiral electronics spanning solar cells, conductors, spintronics and beyond. At present, we are working to partner Large Language Model with our closed-loop learning approach towards developing a chemical language model for organic electronics.