Composition mapping in InGaN by scanning transmission electron microscopy.

We suggest a method for chemical mapping that is based on scanning transmission electron microscopy (STEM) imaging with a high-angle annular dark field (HAADF) detector. The analysis method uses a comparison of intensity normalized with respect to the incident electron beam with intensity calculated employing the frozen lattice approximation. This procedure is validated with an In(0.07)Ga(0.93)N layer with homogeneous In concentration, where the STEM results were compared with energy filtered imaging, strain state analysis and energy dispersive X-ray analysis. Good agreement was obtained, if the frozen lattice simulations took into account static atomic displacements, caused by the different covalent radii of In and Ga atoms. Using a sample with higher In concentration and series of 32 images taken within 42 min scan time, we did not find any indication for formation of In rich regions due to electron beam irradiation, which is reported in literature to occur for the parallel illumination mode. Image simulation of an In(0.15)Ga(0.85)N layer that was elastically relaxed with empirical Stillinger-Weber potentials did not reveal significant impact of lattice plane bending on STEM images as well as on the evaluated In concentration profiles for specimen thicknesses of 5, 15 and 50 nm. Image simulation of an abrupt interface between GaN and In(0.15)Ga(0.85)N for specimen thicknesses up to 200 nm showed that artificial blurring of interfaces is significantly smaller than expected from a simple geometrical model that is based on the beam convergence only. As an application of the method, we give evidence for the existence of In rich regions in an InGaN layer which shows signatures of quantum dot emission in microphotoluminescence spectroscopy experiments.