The heterogeneous nucleation of threading dislocations on partial dislocations in III-nitride epilayers

J Smalc-Koziorοwska,G P Dimitrakopulos,I Belabbas,Ph Komninou,Ø Prytz,J Moneta,P Chatzopoulou,C Bazioti,I G Vasileiadis

doi:10.1038/s41598-020-74030-y

Abstract

III-nitride compound semiconductors are breakthrough materials regarding device applications. However, their heterostructures suffer from very high threading dislocation (TD) densities that impair several aspects of their performance. The physical mechanisms leading to TD nucleation in these materials are still not fully elucidated. An overlooked but apparently important mechanism is their heterogeneous nucleation on domains of basal stacking faults (BSFs). Based on experimental observations by transmission electron microscopy, we present a concise model of this phenomenon occurring in III-nitride alloy heterostructures. Such domains comprise overlapping intrinsic I1 BSFs with parallel translation vectors. Overlapping of two BSFs annihilates most of the local elastic strain of their delimiting partial dislocations. What remains combines to yield partial dislocations that are always of screw character. As a result, TD nucleation becomes geometrically necessary, as well as energetically favorable, due to the coexistence of crystallographically equivalent prismatic facets surrounding the BSF domain. The presented model explains all observed BSF domain morphologies, and constitutes a physical mechanism that provides insight regarding dislocation nucleation in wurtzite-structured alloy epilayers.

Highlights

III-nitride compound semiconductors are breakthrough materials regarding device applications
threading dislocation (TD) that emanate from basal stacking faults (BSFs) often form inverse half loops indicative of a heterogeneous nucleation phenomenon
It can be seen that the TD emanates from a region comprising not a single I2 BSF, but an overlap of two I1 BSFs with parallel stackings, as shown in the inset of Fig. 1c

Summary

Introduction

III-nitride compound semiconductors are breakthrough materials regarding device applications. Based on experimental observations by transmission electron microscopy, we present a concise model of this phenomenon occurring in III-nitride alloy heterostructures Such domains comprise overlapping intrinsic I1 BSFs with parallel translation vectors. Such TDs have a-type (1/3 < 1120 >) Burgers vectors and appear in very high densities (typically 108–1010 cm-2)[1] Their elimination has great technological importance for (opto)electronic device performance, as they affect diverse phenomena like parasitic luminescence2,3 segregation[4,5,6], electron mobility[7,8,9] and device d egradation[10,11]. The I2 BSF would eventually be annihilated through successive dislocation reactions, and the remaining TDs be connected to a hexagon of 60° misfit dislocation segments This model lacks an explanation of how the a-type half-loops are introduced in the first place, given the lack of resolved shear stress. These BSFs can be delimited by either 1/6 < 2023 > Frank-Shockley PDs or prismatic stacking faults (PSFs). I1 BSF superposition leads to formation of closed domains in the form of distorted hexagonal prisms, eliminating the Scientific Reports | (2020) 10:17371

Methods

Results

Discussion

Conclusion