Do moving basal dislocations in sapphire (α-Al2O3) have non-stoichiometric cores?

An idealized crystal structure for sapphire (α-Al2O3) (perfect oxygen hcp packing, flat cation planes perpendicular to [0 0 0 1]) has been used by Kronberg [Acta. Metall. 5 (1957)] and many others over the past 50 years to describe basal slip and basal twinning at the atomic level. However, it was recognized a decade ago [Bilde-Sørensen et al., Acta Mater. 44 2145 (1996); Pirouz et al., Acta Mater. 44 2153 (1996)] that the actual structure of sapphire allows much simpler atomic mechanisms to be postulated for basal slip and basal twinning. These models are supported by convincing arguments derived from chemical and structural considerations. Recently, a climb-dissociated basal dislocation in the boundary of a manufactured bicrystal was observed by atomic resolution transmission electron microscopy [Shibata et al., Science 316 82 (2007)]. The images were interpreted as indicating non-stoichiometric charged dislocation cores and it was inferred that, during dislocation motion on the basal plane, the basal dislocations had to move according to a variant of Kronberg's mechanism. This conclusion is difficult to reconcile, with (i) the models based on the actual structure [Bilde-Sørensen et al., Acta Mater. 44 2145 (1996); Pirouz et al., Acta Mater. 44 2153 (1996)], (ii) weak beam TEM images [Lagerlöf et al., in Proceedings of the Electron Microscopy Society of America, edited by G.W. Baily (San Francisco Press, 1982), p.554], which contradict important implications of this variant of Kronberg's model, (iii) implications concerning dislocation motion in ionic materials, and (iv) the possibility that interface dislocations can be subject to entirely different constraints than apply to gliding lattice dislocations.