Physics Colloquium with Dr. Stephen Irle

Harnessing the Power of Supercomputers and AI for Molecular and Materials Design

with Dr. Stephen Irle
Oak Ridge National Laboratory

ABSTRACT: In the chemical and materials sciences, one of the holy grails is the rational design of molecular compounds and materials with desirable chemical or physical properties. Examples are nuclear materials able to withstand extreme environments, molecules and materials for energy-efficient solar power conversion, and compounds capable to remove undesirable green-house gases or industrial waste products either by secure storage or conversion into value-added compounds. The current state-of-the-art for the discovery of novel molecular structure or materials can still be characterized as an Edisonian “needle in the haystack” approach, increasingly assisted over the past 20 years by post-discovery rationalization of the target properties via experimental atomic structure observation and complimented by computational investigations focusing on the essential details of the candidate compounds in question. Trial-and-error however is error-prone, costly, and vastly inefficient to explore the immensely wide range of chemical structure and element combinations. Therefore, we developed a workflow that integrates computational chemical and materials sciences methodologies with machine learning approaches for the inverse design of molecules and materials with user-defined target properties. It is envisioned that our workflow will function as a “virtual lab” in which simulations will be capable of predicting, within matters of days, chemical composition and structure that can be suggest with high confidence for experimental validation.

BIOGRAPHY: Stephan Irle has performed research in computational chemistry and materials sciences in Germany, Austria, the United States, and Japan. His specialty is the quantum chemical study of complex systems. Target areas are soft matter and biosimulations, excited states of large molecules, separations, and catalysis. Complementary studies of physicochemical properties, theoretical spectroscopy, and the development of scalable algorithms including approximate quantum chemical methods on classical as well as quantum computing platforms, and machine learning-based solutions for inverse design problems accompany this research. Dr. Irle has more than 300 publications in peer-reviewed journals and conference proceedings, 14 book chapters, and co-authored 2 books. Dr. Irle received a B.S. and M.S. in Chemistry, both from the University of Siegen in Germany. He received his Ph.D. in Chemistry from the University of Vienna in Austria.

With refreshments from Panera Bread