While Density Functional Theory (DFT) has been an increasingly useful tool in many applications of materials science, its delocalized nature of description of d and f electrons can cause significant inaccuracy in predicting electronic properties for materials containing heavy elements; such as the bandgap. A common way to correct this problem is to add a corrective potential barrier, known as the Hubbard U potential. However, determining the parameters for this potential could itself require a series of expensive first-principles calculations, especially when no experimental band gaps values are available for empirical calibration. In this work, we outline a machine learning (ML) framework for predicting the Hubbard U potential for inorganic solid state systems containing a single transition metal element. For this purpose, we generate a training set of over 200 periodic structures and the corresponding U values for the constituent transition metal sites. The training-set Hubbard U data is generated based on density functional perturbation theory calculations. We then utilize descriptors for inorganic periodic systems to build ML models to predict Hubbard U values for novel structures outside the training set. The most important set of descriptors for predicting the Hubbard U potential for the examined materials systems are highlighted, with an analysis of outstanding trends and variations of its value across the d-block elements.

Prediction of Hubbard U parameters of transition metal-containing compounds using machine learning

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.