Reduction of Defects in Germanium-Silicon

Crystals grown without being in contact with a container have superior quality to otherwise similar crystals grown in direct contact with a container, especially with respect to impurity incorporation, formation of dislocations, and residual stress in the crystals. In addition to float-zone processing, detached Bridgman growth, although not a completely crucible-free method, is a promising tool to improve crystal quality. It does not suffer from the size limitations of float zoning and the impact of thermocapillary convection on heat and mass transport is expected to be negligible. Detached growth has been observed frequently during g experiments. Considerable improvements in crystalline quality have been reported for these cases. However, neither a thorough understanding of the process nor a quantitative assessment of the quality of these improvements exists. This project will determine the means to reproducibly grow GeSi alloys in a detached mode and seeks to compare processing-induced defects in Bridgman, detached-Bridgman, and floating-zone growth configurations in GeSi crystals (Si less than or equal to 10 at%) up to 20mm in diameter. Specific objectives include: measurement of the relevant material parameters such as contact angle, growth angle, surface tension, and wetting behavior of the GeSi-melt on potential crucible materials; determination of the mechanism of detached growth including the role of convection; quantitative determination of the differences in defects and impurities for crystals grown using normal Bridgman, detached Bridgman, and floating zone (FZ) methods; investigation of the influence of a defined flow imposed by a rotating magnetic field on the characteristics of detached growth; control of time-dependent Marangoni convection in the case of FZ growth by the use of a rotating magnetic field to examine the influence on the curvature of the solid-liquid interface and the heat and mass transport; and growth of benchmark quality GeSi-single crystals.