Speaker
Description
We investigate the mechanisms that govern the thermal stability of two technologically relevant systems: magnetic skyrmions, envisioned as information carriers in novel spintronics devices, and nanopillars, as used in magnetic tunnel junctions (MTJs) for data storage, and currently envisaged as stochastic p-bits for unconventional computing schemes. To do so, we combine the Kramers framework and a path sampling scheme in atomistic simulations.
In skyrmions, we find that their internal modes of deformation lead to a stabilizing entropic contribution to the Arrhenius prefactor. On the other hand, in nanopillars, the domain-wall mediated magnetization reversal yields a large, destabilizing entropic contribution to the prefactor, which scales like an exponential of the activation energy. Our simulations show good agreement with experimental measurements carried out on 50 nm diameter perpendicular MTJs, with mean dwell times down to the nanosecond. We explain this observation with very low Arrhenius prefactors in the femtosecond range. In both cases, we challenge the common assumption of a constant, 1 ns attempt time in these systems, and underline the importance of the activation entropy in transitions involving solitonic textures.
