Events

Materials discovery has traditionally relied on intuition-driven experiments, but computational methods have transformed how we predict material properties and identify promising candidates. Despite these advances, synthesizing predicted materials remains a bottleneck, and characterizing their properties is often time-intensive and costly. In this talk, I will explore how computational tools based on quantum chemistry and machine learning can address these challenges. Starting with ab-initio phase diagrams calculated using zero-temperature bulk thermodynamics, I will discuss how computational methods are evolving to describe phase equilibria and reactions under synthesis-relevant conditions. Case studies from my research will demonstrate how these approaches can efficiently optimize synthesis pathways for energy-related materials. These studies integrate closely with experiments, using feedback from in-situ X-ray diffraction to understand reaction outcomes. I will also highlight how machine learning can automate structural and compositional analysis from these measurements, a key step toward enabling self-driving laboratories. Finally, I will outline the limitations of current methods and emphasize the need for improved models of synthesis kinetics and automated characterization workflows.

Nathan Szymanski is a postdoctoral researcher in Chemical Engineering and Materials Science at the University of Minnesota. He earned dual Bachelor of Science degrees in Physics and Applied Mathematics from the University of Toledo before pursuing his Ph.D. in Materials Science and Engineering at UC Berkeley under the guidance of Professor Gerbrand Ceder. During his Ph.D., Nathan developed computational methods to streamline materials synthesis and characterization, combining quantum chemistry calculations with machine learning to identify promising battery materials, design synthesis pathways, and interpret experimental data. He was an NSF Graduate Research Fellow and received the Didier de Fontaine Award, which recognizes the top graduate student in computational materials research at Berkeley. Nathan currently works with Professor Chris Bartel at Minnesota on the development of generative AI models and the use of thin-film deposition as a route to synthesize predicted materials.