Gas-Source Molecular Beam Epitaxy

Abstract

The replacement of the elemental sources of conventional MBE with simple compounds, first reported in 1980 [1], was initiated in order to bring the advantages of molecular beam epitaxy to the growth of GaxIn1−xAs1−yPy/InP heterostructures. These advantages center about precision in layer thickness and abruptness in doping and heterojunction interfaces. This replacement of elemental sources was necessary because III–V semiconductors containing P, and particularly As and P simultaneously, are very difficult to grow by conventional MBE. A well controlled and useful beam flux from an effusion cell containing elemental phosphorus is difficult to achieve because of the presence of allotropic forms of solid P, each having a different vapor pressure, and because condensed P vaporizes to yield P4 molecules. The morphological observations of Asahi et al [2] of InP grown with P4, and the studies of the relative incorporation of As and P during MBE of GaAs1−yPy and InAsyP1−y by Foxon et al [3], suggest that P4 has a small accommodation coefficient on the III–V surface. It is possible, of course to thermally crack P4 to P2, and P2 can readily be used for epitaxy of P containing III–V compounds. Its accommodation coefficient is approximately unity [4]. However, the generation of P2 by adding a thermal cracker to a conventional effusion oven does not eliminate the underlying stability problem and has added control problems.