Abstract
In this work we use X-ray diffraction to analyze the microstructures of two III-nitride sample systems. The first part of sample systems is cubic-GaN epilayers MBE-grown on GaAs(001) of assorted Ga flux ratio from 0.8, 1.0, 1.1, to 1.2; the second part is the InxGa1-xN-GaN multiple quantum wells on sapphire, varied with the fractional molar value of In- composition: x = 0.1~0.25, nominally. Starting from the focus of polytype GaN as revealed by conventional X-ray diffraction, we derive a correlation of the light-emitting efficiency with different Ga flux ratio to the crystalline quality of microstructures. It is simply confirmed that the Ga-poor sample of Ga flux ratio 0.8 shows worse quality both on the emission peak of 380-nm photoluminescence and on the diffraction width of zinc-blende crystalline structures. While, after a rigorous examination on the purposed cubic-GaN films from Ga-poor to Ga-rich, we do not resolve any difference in the lattice constants. On the other side, the lattice constant c of the wurtzite phases mixed in the GaN film gradually expands as the Ga flux ratio getting increased. We conjecture that, in Ga-rich samples, extra Ga atoms would occupy the interstitial sites in wurtzite domains to play a role as impurity-like defects. Consequently, as if the wurtzite domains had a better crystalline quality, the overall thin film would suffer more stress and yield a worse efficiency in the light emission. In view of the orientation spheres, the geometric correlation and domain distributions of the primary and twined wurtzite phases are properly reconstructed with respect to the cubic-GaN lattice. We conclude that the origin of wurtzite domains is due to the stacking fault and twin bands prevailing in the GaN film. Following after the study of GaN epilayers, but taking advantage of bright and well-collimated synchrotron X-rays, we apply high-resolution X-ray diffraction to characterize the superlattice structure of InxGa1-xN multiple quantum wells. Preliminary results have been done to precisely determine the thickness of periodic InxGa1-xN-GaN bilayers and to assure the indium composition as compared with the nominal value x. All of our analyses are straightly based upon a linear regression from the peak positions of superlattice reflections, which is less dependent of the details presumed in a strain-layer model.