Activation enthalpy for kink-pair nucleation on dislocations: Comparison between static and dynamic atomic-scale simulations

We show in the case of a high-Peierls-stress Lomer dislocation in an aluminum crystal that the dependence of the kink-pair activation enthalpy on the stress obtained from static nudged elastic band method calculations agrees with that extracted from dynamical, constant strain-rate simulations. In order to perform the dynamical simulations, we first propose flexible boundary conditions to replace the rigid conditions that are usually applied. This removes the spurious forces on the dislocation that arise because of the mismatch between the elastic strain imposed by the rigid conditions and the plastic strain associated with the dislocation motion. Second, we present a statistical analysis to rigorously extract enthalpy-stress relations from dynamical simulations. We find that the activation enthalpy becomes zero for a stress (which we call the Peierls stress for kink nucleation) smaller than that required to move athermally a rigid straight dislocation (called here the Peierls stress for rigid motion). This effect may explain the discrepancy often reported in the literature between the Peierls stress predicted by atomistic calculations, determined on short two dimensional dislocations, i.e., the Peierls stress for rigid motion, and the Peierls stress extracted from experiments, which corresponds to that when kink pairs form on three dimensional dislocations without the help of thermal fluctuations, i.e., the Peierls stress for kink nucleation.