The dissociation of (a+c) misfit dislocations at the InGaN/GaN interface

In hexagonal materials, ( a+c ) dislocations are typically observed to dissociate into partial dislocations. Edge ( a+c ) dislocations are introduced into (0001) nitride semiconductor layers by the process of plastic relaxation. As there is an increasing interest in obtaining relaxed InGaN buffer layers for the deposition of high In content structures, the study of the dissociation mechanism of misfit ( a+c ) dislocations laying at the InGaN/GaN interface is then crucial for understanding their nucleation and glide mechanisms. In the case of the presented plastically relaxed InGaN layers deposited on GaN substrates, we observe a trigonal network of ( a+c ) dislocations extending at the interface with a rotation of 3° from <1 $\overline{1}$00> directions. High-resolution microscopy studies show that these dislocations are dissociated into two Frank–Shockley 1/6<2 $\overline{2}$03> partial dislocations with the I₁ BSF spreading between them. Atomistic simulations of a dissociated edge ( a+c ) dislocation revealed a 3/5-atom ring structure for the cores of both partial dislocations. The observed separation between two partial dislocations must result from the climb of at least one of the dislocations during the dissociation process, possibly induced by the mismatch stress in the InGaN layer.