High resolution transmission electron microscopy of dislocation core dissociations in gold and iridium

The occurrence of brittle fracture in iridium has attracted significant attention in recent years and is thought to be related to the energetics of the dislocation core, in particular the extremely high unstable stacking energy. Although it is not experimentally possible to measure the unstable stacking energy, first-principles calculations have been used to predict both this and the stacking-fault energy. These calculations suggest that, despite large differences in stacking-fault energy and elastic constants, gold and iridium exhibit similar dissociation behaviours, with screw dislocations in both metals dissociated by approximately 1 nm. In the current study, high-resolution transmission electron microscopy (HRTEM) has been utilized to observe experimentally the arrangement of atomic columns surrounding dislocation cores. Deviations from perfect lattice sites have been measured, and experimental observations quantified through comparisons with image simulations. In the case of screw dislocations, the displacement field of atomic columns relative to a perfect lattice was used to determine the extent of in-plane lattice distortion. Through comparison with simulated displacement maps, this allowed the screw dissociation width in both gold and iridium to be measured as 0.8 nm. Direct comparisons of simulated and experimentally obtained images were used to characterize the core structures of 60° dislocations, which were found to be dissociated by 3.25 nm (gold) and 1.25 nm (iridium). The stacking-fault energy for gold (33 mJ m−2), as calculated from the present high-resolution measurements, is in good agreement with previous weakbeam studies. Finally, weak-beam observations of dissociated dislocations in iridium agree well with HRTEM measurements and yield a stacking-fault energy of 420 mJ m−2. For gold and iridium, both high-resolution and weak-beam measurements of dissociation distance agree with the orientation dependence of stacking-fault width as predicted by anisotropic elasticity.