Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

MM: Fachverband Metall- und Materialphysik

MM 37: Functional and Complex Materials

MM 37.1: Vortrag

Freitag, 21. März 2025, 12:00–12:15, H22

Unveiling Material Dynamics with Machine-Learned Interatomic Potentials — •Ferenc Tasnadi, Bobur Mukhamedov, Amanda Ehn, Florian Trybel, and Igor A. Abrikosov — IFM Linköping University

Machine-learned interatomic potentials (MLIPs) have revolutionized our ability to understand the properties of materials with complex dynamical processes. In this work, we present an active learning MLIP strategy to: (i) investigate the elasticity of alloys near dynamical instability [1] and (ii) explore dynamical bond disorder in high-pressure synthesized PN2 [2]. bcc-Ti-based alloys have wide industrial applicability, ranging from low-modulus biomedical implants to high-strength GUM metals. The low-temperature dynamical instability of bcc-Ti can be tailored through the addition of bcc stabilizers (Nb, Ta, Zr, V), resulting in anomalous mechanical properties such as elinvar behavior or high anisotropy with low modulus in Ti-Nb-Zr and Ti-Zr-Sn alloys. Pnictogen compounds (Group 15 elements) exhibit fascinating chemistry due to the variety of bonding configurations, as demonstrated in the recently studied P-N system [2]. Long-timescale MLIP-driven simulations reveal that N-N distances connecting the P-N octahedra vary dynamically between single-bonded and non-bonding configurations. The results are compared with experimental observations. If time permits, we will demonstrate how MLIP-driven molecular dynamics simulations can offer deeper insights into materials exhibiting Peierls instability. [1] New J. Phys. 22 113005 (2020); J. Vac. Sci. Technol. A 42, 013412 (2024). [2] Chem. Eur. J. 2022, 28, e202201998.

Keywords: dynamical instability; bond disorder; Peierls instability; machine learned interatomic potential