The optimization of InxGa1-xAs and InP growth conditions by CBE

Abstract

Minimization of the number of experiments needed to fully characterize and optimize the growth of epitaxial material is the first important step in realizing state of the art device structures. While widely used in some fields such as chemical engineering, response surface modeling (RSM) has been little used in crystal growth applications. Using RSM, input parameters such as substrate temperature hydride injector temperature and V/III ratio, were simultaneously adjusted to characterize the crystal growth process. This technique identified interactions among parameters, minimized the number of experiments necessary to understand and optimize the process, and minimized the variability of the growth process. RSM has been applied to the CBE growth of InGaAs and InP with the purpose of generating an operating point at which both good surface morphology and high mobility material can be produced. Although the best 77 K InP mobility was 70,000 cm2/V...s, in order to improve the surface quality the input parameters were changed so that the final mobility was 37,000 cm2/V...s. Although the quality of the InGaAs layers showed a dependence on the reactor history, there did not appear to be any sensitivity to variations made in the operating conditions. The best 77 K InGaAs mobility was 62,500 cm2/V...s.