Determining the rate of thermally activated processes for crystal dislocations using advanced sampling or minimum-energy-path methods is challenging because dislocations are extended defects involving a large number of atoms. In this talk, I will describe our progress in modeling two thermally activated dislocation processes: cross slip in face-centered cubic (FCC) metals and dislocation nucleation in diamond cubic (DC) semiconductors. A modified string method is used to stabilize the minimum energy path relaxation under high stresses where the final state (B) has much lower energy than the initial state (A). The modified string method is also stable enough to cope with the rough energy landscape of the semiconductor. However, the minimum energy path obtained at zero temperature would lead to a reaction rate that is generally too low compared with experiments (and sometimes even direct Molecular Dynamics simulations). I will discuss the possible reasons causing such a discrepancy.
Copyright. All Rights Reserved.