MLTB: Enhancing Transferability and Extensibility of Density Functional Tight Binding Theory with Multi-body Interaction Corrections
Daniel J. Burrill, Chang Liu, Michael G. Taylor, Marc J. Cawkwell, Nicholas Lubbers, Danny Pérez, Enrique R. Batista, Ping Xiong Yang · ChemRxiv · 2024
We present a hybrid semi-empirical density functional tight binding (DFTB) models with a machine learning neural network potential as a correction to the repulsion term. This hybrid model, termed MLTB, employs the standard self-consistent charge (SCC) DFTB formalism as the baseline, enhanced by the HIP-NN potentials as an effective many-body correction for the short range pairwise repulsive interactions. The MLTB model demonstrates significantly improved transferability and extensibility compared to standalone the SCC-DFTB and HIP-NN models. This work provides a practical computational framework for developing reliable SCC-DFTB models with additional many-body corrections that more closely approach the DFT-level accuracy. We illus- trate this method with the development of an accurate model for the thorium-oxygen system, applied to the study of its nanocluster structures, (ThO2)n.