Dr. Rajni Chahal, a postdoctoral research associate in the Chemical Sciences Division at Oak Ridge National Laboratory (ORNL), will give a seminar titled "Machine Learning-Accelerated Modeling and Simulation for Structure-Property Prediction in Materials."

Abstract: A deep understanding of structure-property relationships is indispensable for advancing materials design and development. As such, traditional experimental approaches are often hindered by time constraints, high costs, and harsh operating conditions. For instance, in nuclear reactors, high temperatures and radiation exposure inhibit fundamental insights into molten salts used for cooling the reactor core. This presentation explores how machine learning-enhanced predictive computational modeling and simulation can overcome these challenges. Specifically, I will highlight my work on developing and utilizing machine learning interatomic potentials (MLIPs) to accelerate the prediction of structure, properties, and chemical transformations in molten salts and polymers by up to 10,000 times. The developed MLIP models are validated using ab initio simulations and experimental data when predicting structure as well as transport, thermophysical, and mechanical properties of these materials. These results demonstrate that machine learning-accelerated modeling and simulation are powerful tools for screening and discovery of materials for several engineering and clean energy applications. Finally, I will discuss key challenges in developing and validating machine learning materials models, including strategic data sampling, model trustworthiness, and energy efficiency, along with potential solutions such as uncertainty quantification approaches.

Biography: Dr. Rajni Chahal is a Postdoctoral Research Associate in the Chemical Sciences Division at Oak Ridge National Laboratory (ORNL). Her research philosophy centers on developing efficient computational models rigorously validated by experiments and leveraging computational insights to guide and better interpret experiments. As such, her research combines multiscale material modeling and simulation with enhanced sampling methods, uncertainty quantification, machine learning, and informed experiments to enable fundamental insights into structure, properties, and chemical transformations in materials relevant to aerospace and energy applications. Before joining ORNL, she was a Postdoctoral Researcher in the Chemical and Nuclear Engineering Department at the University of Massachusetts Lowell, where she led the development of machine learning interatomic potentials for molten salts. She earned her Ph.D. in Aerospace Engineering from The University of Texas at Arlington in 2020, focusing on multiscale modeling of carbon nanotube reinforced polymer composites. She received her Bachelor’s degree in Mechanical Engineering from Dr. B.R. Ambedkar National Institute of Technology, Jalandhar, India (2015). Her work includes high performance computing, development of new computational tools for materials modeling and simulation, model validation and benchmarking, enhancing the accuracy and trustworthiness of ML predictions, and creating ML workflows to reduce data requirements, computational costs, and energy consumption in their development. Her research interests span mechanical and aerospace materials, chemical separations, energy storage and conversion, pyroprocessing, and heat transfer in advanced nuclear reactors. Outside of her research and mentorship activities, Rajni enjoys cooking, hiking, and spending time with her family. She also has over 15 hours of glider-flying experience as a student pilot.