Autonomous basin climbing method with sampling of multiple transition pathways: application to anisotropic diffusion...

This paper presents an extension of the autonomous basin climbing (ABC) method, an atomistic activation-relaxation technique for sampling transition-state pathways. The extended algorithm (ABC-E) allows the sampling of multiple transition pathways from a given minimum, with the additional feature of identifying the pathways in the order of increasing activation barriers, thereby prioritizing them according to their importance in the kinetics. Combined with on-the-fly kinetic Monte Carlo calculations, the method is applied to simulate the anisotropic diffusion of point defects in hcp Zr. Multiple migration mechanisms are identified for both the interstitials and vacancies, and benchmarked against results from other methods in the literature. The self-interstitial atom (SIA) diffusion kinetics shows a maximum anisotropy at intermediate temperatures (400~700 K), a non-monotonic behavior that we explain to originate from the stabilities and migration mechanisms associated with different SIA sites. The accuracy of the ABC-E calculations is validated, in part, by the existing results in the literature for point defect diffusion in hcp Zr, and by benchmarking against analytical results on a hypothetical rough-energy landscape. Lastly, sampling prioritization and computational efficiency are demonstrated through a direct comparison between the ABC-E and the activation relaxation technique.