Growth of high quality gallium nitride thin films by RF-plasma assisted molecular beam epitaxy

High quality gallium nitride thin films were grown on sapphire substrates by rf-plasma assisted molecular beam epitaxy. The methodology of the optimization of the two important growth parameters, the III/V flux ratio and the growth temperature, is discussed. Yellow luminescence (YL), peaking at ∼2.2 eV, is a universal feature in undoped and n-doped GaN films grown by various growth techniques. The intensity can vary over a wide range with good samples exhibiting almost no YL. Our results show that YL related defects can be significantly reduced by rapid thermal annealing. Annealing at temperatures higher than 800°C results in material degradation. In addition, YL can be partly reduced when GaN films were grown with indium surfactant. Improved surface morphology is obtained which is attributed to the enhanced two-dimensional growth by the application of In surfactant. Defect analysis and off-normal channeling studies show that the defect density is reduced when samples were grown with In surfactant.; Having optimized the growth conditions, the intensity of YL is four orders of magnitude lower than that of the bandedge emission, indicative of excellent material quality in terms of optical property. The material's electrical property is further improved by the deposition of intermediate-temperature GaN buffer layers (ITBL). The Hall mobilities of the sample grown without ITBL and the one grown with an ITBL of 800 nm thick are 87 and 390 cm2V −1s−1, respectively. A systematic shift in the photoluminescence peak position, following the same trend as the Hall mobility, indicates the relaxation of residual strain in the top GaN layers. A maximum Hall mobility of 460 cm2V−1s −1 can be obtained using an ITBL of 800 nm thick and further optimizing the growth conditions for the low-temperature buffer layer. A model has been presented to account for the observed G-R noise in the samples grown with ITBL of different thickness. Detailed numerical evaluation indicates a reduction in the trap density by over an order of magnitude with the use of ITBL.