472740 Describing the Diverse Geometries of Gold from Nanoclusters to Bulk − a First-Principles Based Hybrid Bond Order Potential
In this talk, we introduce a unified ab initio-based hybrid bond order potential (HyBOP), which captures the competition between short-range interactions (bond directionality) and long-range dispersions to accurately predict the size dependent diverse geometries of gold clusters, surfaces and bulk. Our work involves an important methodological development: we employed genetic algorithms to train our HyBOP potential against a large DFT dataset comprising (a) energies of a thousand 13-atom Au clusters, (b) equation of state for various bulk polymorphs of gold, and (c) low-index surfaces of bulk face-centered gold. Our newly developed HyBOP accurately predicts (a) evolution of various structural motifs in Au clusters with increasing size, (b) critical size for transition from planar to globular structures (i.e., dimensionality changes), (c) global energy minimum configurations at various cluster sizes, and (d) thermodynamics, structure, elastic properties, and energetic ordering of bulk condensed phases as well as surfaces, in excellent agreement with DFT calculations and spectroscopic experiments. As a representative test, using long time HyBOP based MD simulations, we identified the atomic scale mechanisms governing the coalescence of icosahedral and planar clusters into pyramids.
See more of this Group/Topical: Computational Molecular Science and Engineering Forum