Towards the Atomic-Level Analysis of Ground-State Electronic Structures of Solid Materials via Prediction from Core-Loss Spectra

Local electronic structure at the ground state is essential for understanding the stability and properties of materials. Core-loss spectroscopy using electron or X-ray provides the insights into the local electronic structure, but directly observable information is limited to the partial density of state (PDOS) of the conduction band at the excited state. To overcome this limitation, we developed a machine learning (ML) approach by creating a database of Si-K core-loss spectra and corresponding ground-state PDOS for various silicon structures. Using this database, we constructed an ML model capable of predicting atom-specific ground-state PDOS of the valence and conduction bands from Si-K edges. Our model demonstrated the ability of ML to extract the complex correlation between ground-state PDOS and Si -K edges. This study provides crucial insights into achieving atomic-level analysis of ground-state electronic structures, paving the way for a deeper understanding of stability and properties of materials.