Change of dislocation density in silicon wafers during thermal processing

Dislocations are known to be a limiting parameter for higher efficiencies in crystalline silicon solar cells. They can increase during the crystallization process, if thermal stress is acting as the driving force for dislocation motion. In today's point of view, the dislocation density is assumed to be invariant during thermal processing of the wafers after crystallization. With this paper we want to motivate a discussion about this assumption. As a worst case study, numerical simulations are performed describing the thermal conditions in a wafer in the middle of a boat during the loading and unloading stage of a phosphorus diffusion process. Under extreme conditions a raise of dislocation density from the starting value of 1/spl middot/10/sup 4/ cm/sup -2/ up to 6.8/spl middot/10/sup 6/ cm/sup -2/ is calculated at the edge of the wafer for loading and again up to 7.2/spl middot/10/sup 6/ cm/sup -2/ during unloading. Nevertheless the expected effect is strongly coupled to the type of wafer materials and their specific dislocation density and distribution. This subject is investigated by heat treatment experiments focusing on dislocation clusters in multicrystalline silicon. Opposed to simulation results, no change of the dislocation density in cluster configurations is observed.