We report the interplay between In incorporation and strain relaxation kinetics in high-In-content InxGa1-xN (x = 0.3) layers grown by plasma-assisted molecular-beam epitaxy. For In mole fractions x = 0.13–0.48, best structural and morphological qualities are obtained under In excess conditions, at In accumulation limit, and at a growth temperature where InGaN decomposition is active. Under such conditions, in situ and ex situ analyses of the evolution of the crystalline structure with the layer thickness point to an onset of misfit relaxation after the growth of 40 nm, and a gradual relaxation during more than 200 nm, which results in an inhomogeneous strain distribution along the growth axis. This process is associated with a compositional pulling effect, i.e., indium incorporation is partially inhibited in presence of compressive strain, resulting in a compositional gradient with increasing In mole fraction towards the surface.
References
The measurement of the substrate temperature by RHEED via the Ga(In) desorption time provides good reproducibility and is reliable. The determination of the actual temperature value was performed using a calibrated pyrometer, and measuring the Ga/In desorption (Ga in the 750–700 °C range, and In in the 720–590 °C range) from a GaN substrate just after closing and baking the PAMBE system, i.e., with the pyrometer window just cleaned. Repeating the measurements several times, our error bars in the actual temperature values are no larger than ±10 °C.
The incorporation limit temperatures reported in this paper are about 10–60 °C higher than in previous reports (Refs. 6, 17–20). This disagreement is attributed to the fact that Fig. 2(a) presents In concentrations measured in samples ≈250 nm thick, whereas Refs. 17 and 18 report estimations based on RHEED measurements at the beginning of the InGaN growth on GaN, and Refs. 19 and 20 refer to samples that are 40–120 nm thick. As concluded in this paper, misfit relaxation can lead to an enhancement of the In incorporation that can reach 20% in thick layers. Furthermore, precise decomposition temperature must be taken with caution, since it can depend on the overall metal-to-nitrogen ratio and on the growth rate, as described in Ref. 20.
Relaxation is defined as R = (a − aGaN)/(a0 − aGaN), where a is the measured average in-plane lattice parameter, a0 is the average lattice parameter of the relaxed InGaN layer, and aGaN is the lattice parameter of the GaN substrate.
