Exploring the Crystal\nStructure and Electronic Properties\nof γ‑Al<sub>2</sub>O<sub>3</sub>: Machine Learning Drives\nFuture Material Innovations

For decades, researchers have struggled to determine\nthe precise\ncrystal structure of γ-Al<sub>2</sub>O<sub>3</sub> due to its\natomic-level disorder and the challenges associated with obtaining\nhigh-purity, high-crystallinity γ-Al<sub>2</sub>O<sub>3</sub> in laboratory settings. This study investigates the crystal structure\nand electronic properties of γ-Al<sub>2</sub>O<sub>3</sub> coatings\nunder the influence of an external electric field, integrating machine\nlearning with density functional theory (DFT). A potential 160-atom\nsupercell structure was identified from over 600,000 γ-Al<sub>2</sub>O<sub>3</sub> configurations and confirmed through high-resolution\ntransmission electron microscopy and selected area electron diffraction.\nThe findings indicate that γ-Al<sub>2</sub>O<sub>3</sub> deviates\nfrom the conventional spinel structure, suggesting that octahedral\nvacancies can reduce the system’s energy. Under an external\nelectric field, the material’s band structure and density of\nstates (DOS) undergo significant changes: the bandgap narrows from\n3.996 to 0 eV, resulting in metallic behavior, while the projected\ndensity of states (PDOS) exhibits peak broadening and splitting of\noxygen atom PDOS below the Fermi level. These alterations elucidate\nthe variations in the electrical conductivity of alumina coatings\nunder an electric field. These findings clarify the mechanisms of\nγ-Al<sub>2</sub>O<sub>3</sub>’s electronic property modulation\nand offer insights into its covalent and ionic mixed bonding as a\nwide-bandgap semiconductor. This discovery is essential for understanding\ndielectric breakdown in insulating materials.