Compositional Ordering in Semiconductor Alloys

Compositional ordering has been observed in a wide variety of III/V semiconductor alloys as well as in SiGe alloys. The thermodynamic driving force is now understood in terms of minimization of the microscopic strain energy of the bonds in the solid. However, the mechanism leading to the specific ordered structures formed is only now beginning to be understood. It appears to be intimately related to the physical processes occurring on the surface during epitaxial growth, specifically surface reconstruction and the attachment of atoms at steps and kinks. Thus, an improved understanding the ordering process may lead to a better understanding of the surface processes occurring during epitaxial growth from the vapor.This paper will review the current understanding of the ordering process, including discussions of the arrangement of atoms on the surface and the nature of surface steps. The emphasis will be on the use of patterned surfaces to investigate and control the ordered structures formed during organometallic vapor phase epitaxial growth of GaInP. Using photolithography and chemical etching, [110]-oriented steps are formed on the (001) GaAs substrate. The direction of motion of these steps determines the specific variant of the Cu-Pt ordered structure (with ordering on (111) planes) formed. The step density at the edge of the groove apparently determines the degree of order. Highly stepped surfaces suppress ordering or lead to small domains of a single variant. When the steps are very shallow, the large domain of the predominant variant is filled with “inclusions” of the second variant. Step edges that are oriented at nearly 160 from (001) form a {511} variant during growth. This facet is observed to grow at the expense of adjacent (001) surfaces and to produce material that is completely disordered.Growing on intentionally misoriented substrates leads to interesting structures consisting of both large, highly-ordered domains and disordered material. This allows, using cathodoluminescence(CL) imaging, a direct determination of the effect of ordering on the energy band gap. In the GaInP samples studied, the CL images show that the disordered material has a distinct emission pattern consisting of a single, sharp peak at an energy more than 100 meV higher than that observed in the adjacent ordered region.