Theoretical study of GaN molecular beam epitaxy growth using electron cyclotron resonance nitrogen plasma

Wenning Fu, University of Nevada, Las Vegas
Rama Venkat, University of Nevada Las Vegas, Department of Electrical & Computer Engineering
Meyya Meyyappan, University of Nevada, Las Vegas

Journal of Vacuum Science and Technology B: Volume 19, Issue 5: Pages 1803-1807


III–V nitrides (GaN, InN, and AlN) are intensely researched for optoelectronic applications spanning the entire visible spectrum. In spite of the realization of commercial devices and advances in processing of materials and devices, the understanding of the processing and epitaxial growth of these materials is incomplete. In this study, a rate equation based on physically sound surface processes to investigate the molecular beam epitaxial growth of GaN using an electron cyclotron resonance (ECR) plasma source is proposed. A surface riding layer of Ga and N plasma species is included in the model. The surface riding species are allowed to undergo several physical and chemical processes. Rates of all surface processes are assumed Arrhenius type. In the ECR plasma, the flux and reactivity of the active nitrogen are modeled based on plasma dynamics and used in our study. The necessary model parameters, which are unknown, were found by fitting results from simulation to experimental values. As the ECR power increases, the flux of active nitrogen and all other N species, which are by-products, increase almost linearly. Thus the Ga to active N flux ratio increases and hence Ga incorporation rate increases and saturates at a maximum rate. Results of growth rate versus temperature behavior are also presented and discussed based on physical mechanisms. Electron concentration obtained from bulk vacancy concentrations of Ga and N decreases linearly with ECR power, unlike the experimental observation of exponential decrease. The discrepancy is due to the fact that the electron concentration is strongly influenced by the incorporation of unintentional impurities from the plasma chamber such as Si, C, and O, which are not modeled in our study.