Role of dislocation locking and unlocking in the yield strength anomaly of γ -TiAl revealed by machine-learning moment tensor potential

The yield stress anomaly (YSA) is a commonly observed phenomenon in ordered intermetallics in which the yield stress increases with temperature over an intermediate temperature range. Understanding this effect is key to engineering the properties of intermetallic materials used in high temperature, high stress applications, such as jet engines. However, the fundamental mechanisms of YSA in γ-TiAl are still not understood. In this work, we leverage a newly developed machine learning interatomic potential based on the Moment Tensor Potential formalism to perform molecular statics and dynamics simulations of dislocations in γ-TiAl. We observe that, analogous to the Kear–Wilsdorf locks in Ni3Al, ⟨0̄11] screw superdislocations in γ-TiAl can undergo locking through cross-slip, transforming a glissile planar core into a sessile non-planar core. Analysis of the transition state provides evidence that this mechanism is consistent with the YSA. The frequency of cross-slip is found to increase at elevated temperatures, consistent with a thermally-activated locking mechanism. This core transformation involves the conversion of a complex stacking fault in the planar core to a more stable superlattice intrinsic stacking fault in the non-planar core. We also find that the locking mechanism can be modified in off-stoichiometric γ-TiAl, suggesting alloying strategies to tune the YSA.