2021/09/15 by Pedram Tavadze, Reese Boucher, Tavadze, Pedram +17 · 1 citation
Physics and Astronomy · Materials Science · #Rare-earth and actinide compounds #Machine Learning in Materials Science #Advanced Chemical Physics Studies
paper · pdf · doi:10.48550/arxiv.2109.07617
Density-functional theory is widely used to predict the physical properties\nof materials. However, it usually fails for strongly correlated materials. A\npopular solution is to use the Hubbard corrections to treat strongly correlated\nelectronic states. Unfortunately, the exact values of the Hubbard U and J\nparameters are initially unknown, and they can vary from one material to\nanother. In this semi-empirical study, we explore the U and J parameter\nspace of a group of iron-based compounds to simultaneously improve the\nprediction of physical properties (volume, magnetic moment, and bandgap). We\nused a Bayesian calibration assisted by Markov chain Monte Carlo sampling for\nthree different exchange-correlation functionals (LDA, PBE, and PBEsol). We\nfound that LDA requires the largest U correction. PBE has the smallest\nstandard deviation and its U and J parameters are the most transferable to\nother iron-based compounds. Lastly, PBE predicts lattice parameters reasonably\nwell without the Hubbard correction.\n