Optimization of the GaN epilayer quality using in-situ reflectance measurements

Abstract

Although a tremendous amount of work has been done these last years on the nitride semiconductor system, a lot is still to be understood regarding the growth mechanisms of GaN. The standard GaN MOCVD growth process includes the low temperature deposition of a nucleation layer, followed by an anneal at high temperature, and the GaN layer is then deposited. The number of process parameters which can be used to tune the growth is very large (temperatures, times, thicknesses, molar flow rates and ratios …) and, due to the coupling between them, the role of each one is not clearly understood. In this paper, we present systematic series of growth experiments, where in-situ reflectance monitoring was used and correlated to ex-situ optical characterization of the samples by photoluminescence at low temperature (2K). Here, we demonstrate that the nucleation layer and its annealing have a determining effect. The nucleation layer growth temperature was not found to be a very sensitive parameter, while the amount of re-crystallization is. Surprisingly, the amount of ammonia present in the gas phase has a determining effect on the recrystallization behavior of the nucleation layer. Another interesting point is the sensitivity versus growth temperature for the main GaN layer, which was found to affect the initial stages of the growth in a drastic manner when changed by only 5°C. In-situ reflectance allowed us to tune precisely our process and to obtain GaN layers with 500 cm2/Vs electron mobility at room temperature and photoluminescence fwhm of 1.7 meV at 2K for the donor-bound exciton.

Using optical reflectance to measure GaN nucleation layer decomposition kinetics

“…Sumiya et al found that adding H 2 to the annealing ambient enhanced the evaporation and reduced the NL surface coverage [37]. Finally, no change or influence in the NL evolution was observed for changes in the NL growth T [35], and similar results were reported in Section 4.2.…”

“…This is particularly true if different NL thicknesses are optimized, because the thicker NLs will likely have faster decomposition rates but require more time for recrystallization compared to thinner NLs [32]. The P at which the NL is annealed has been shown to influence the reflectance waveform [34], leading to rougher NLs at high P. The relative H 2 , N 2 , and NH 3 flows during annealing also influence the NL evolution [35][36][37]. For example, Halidou et al have shown that when NLs are annealed in H 2 and NH 3, a dramatic decrease in the optical reflectance is observed, indicating a rough morphology.…”

Infrared thermography applied to the validation of thermal simulation of high luminance LED used in automotive front lighting

“…The adopted solution is to take advantage of the very small roughness induce on such micro-electronic manufactured devices. Indeed, Ruffenach-Clur et al [32] have focused on process parameters involved in the GaN MOCVD growth process. At the end of the process, the rms rugosity Rq of the GaN sample was found to be 177nm.…”