Heterogeneous catalysis is an inherently multi-scale process that ultimately connects the behavior of electrons to the global-scale production of chemicals. Understanding how these processes interact is a never-ending challenge, but recent research has shown that the application of machine learning and artificial intelligence models is a promising strategy for discovery of novel catalytic materials and advancing fundamental insight at the interface between chemistry and physics.
Andrew J. Medford,
Georgia Tech
This talk will present progress in the application of machine learning from opposite ends of the multi-scale spectrum. At the scale of electrons, the talk will introduce the use of machine learning approaches to establish a new paradigm of exchange-correlation functional design that uses “multipole features” to provide flexibility between the solid-state and molecular electronic environments that arise in solid-gas/liquid interfaces of heterogeneous catalysis.
At the scale of reactors, the use of “kinetics informed neural networks” will be presented as a route to directly analyze large volumes of transient kinetic and spectroscopic data to extract rate parameters that can help elucidate intrinsic kinetics and reaction mechanisms.
The talk will demonstrate how these fundamentally different approaches have complementary strengths and weaknesses, indicating that a combination of methods will ultimately be required to understand the complex multi-scale processes involved in heterogeneous catalysis.
A.J. Medford is an associate professor in the School of Chemical and Biomolecular Engineering at Georgia Tech. He attended North Carolina State University as an undergraduate and subsequently spent a year as a Fulbright fellow at the Technical University of Denmark before attending Stanford University where he received his Ph.D in Chemical Engineering.
His past research has spanned a wide range of applications including lithium-ion battery electrodes, polymer solar cells, data science, and catalysis. His thesis research focused on developing computational tools for analyzing trends in catalysis under the guidance of Prof. Jens Nørskov, and as a postdoc he worked with Prof. Surya Kalidindi on data infrastructure for materials science.
As a faculty member, his group’s research lies at the intersection of catalysis and surface science, computational chemistry, and machine learning, and he has received several research awards including the NSF CAREER Award and the Early Career award from the ACS CATL division. He also helped develop the “Data Science for the Chemical Industry” online certificate program at Georgia Tech, which has been recognized through the Georgia Tech Curriculum Innovation Award and the AIChE Himmelblau Award for Innovations in Computer-Based Chemical Engineering Education.