Speaker
Description
Understanding processes underlying the evolution of materials (aging, defect formation, chemical reactions, catalysis) requires simulation methods that can describe the atomistic kinetics of materials over time scales of seconds and more. While off-lattice Kinetic Monte Carlo simulations, such as the kinetic Activation-Relaxation Technique [1], have already demonstrated their ability to access these time scales for complex materials, these tools have remained somewhat marginal, however, due to a number of challenges. First, the low reliability of empirical potentials for the description of defects and alloys, especially of their associated activated states that control their kinetics. And, second, codes that are still very much experimental and unstable, making them hard to use, except by a few experts.
In this talk, I'll present recent advances to redesign the kART algorithm into a quick, versatile, stable, and easy-to-use Python code, pyKMC, using tools such as LAMMPS ([3]), the pARTn plugin ([4]), IRA ([5]), an efficient shape matching algorithm, and integrate on-the-fly machine-learned potential [6], in addition to discussing some recent applications of these methods.
References:
[1] El-Mellouhi et al. Physical Review B , 3202 (2008); Béland et al., Physical Review E 84, 67(2011)
[2] Thompson et al., Comp Phys Comm 1, 8171 (2022)
[3] Poberznik et al., Computer Physics Communications 5, 8961 (2024)
[4] M. Gunde et al., Software Impact, 12, 100264 (2022)
[5] Sanscartier et al., The Journal of Chemical Physics, 158, 244110 (2023)