In this study, we integrate machine learning/deep learning models into theoretical and simulation frameworks to predict and analyze optical, thermal, magnetic, and molecular dynamics properties in metallic glasses, oxides, polymers, phosphors, perovskites, and thermoelectric materials. Machine learning models are constructed to predict the melting temperatures, glass transition temperatures (Tg), emission peak positions, phase transition energy levels, bandgap, Curie temperatures, thermoelectric figure of merit (ZT), and other properties from their chemical compositions. Our approach, despite its simplicity, provides predictions with higher accuracy compared to prior research. This approach proves particularly beneficial for predicting properties of novel materials not yet synthesized. The predicted-Tg values from simulations and AI are integrated into the Elastically Cooperative Nonlinear Langevin Equation theory to determine the temperature dependence of structural relaxation time of amorphous materials. All our calculations show good agreement with experimental data and prior simulations without any adjustable parameters. Beyond 'forward prediction' (predicting material properties based on chemical composition), our developed models can be developed to perform 'inverse design' (suggesting chemical compositions to achieve desired material properties).  

Dr. Anh D. Phan is now a senior researcher, group leader, and lecturer of Phenikaa University (Vietnam). He has received his B.S. (2005-2009) at Hanoi National University of Education (Vietnam) in Physics, MSc (2011-2012) at University of South Florida (USA) in Applied Physics and Ph.D. (2014-2018) at University of Illinois Urbana-Champaign (USA) in Physics. Dr. Anh D. Phan's current research interest mainly focuses on developing theoretical approaches to study properties of plasmonic nanostructures and amorphous materials. He has published more than 60 leading peer-reviewed papers including 1 PRL, 1 in PNAS, and 1 in Nature Physics, with citations ~1336 times and h-index of 22 (Google Scholar).