The formation and evolution of nanoscopic voids during silicon crystal growth has received tremendous attention because of its technological importance in setting microelectronic device reliability and also because of its fundamental role as a well-defined example of solid-state homogenous nucleation and growth. Current continuum simulators now are able to accurately predict the density and size distribution of voids as a function of operating conditions during silicon crystal growth and are frequently used by industry in process optimization.

An outstanding problem however, is the prediction of void morphology, which can vary dramatically depending on the growth conditions. Morphology is an important parameter in setting the overall stability of voids, which in turn determines how easily the void can be dissolved by thermal annealing. Both the crystal growth operating conditions and the chemical environment can play an important role in setting void morphology. For example, recent experiments [1] show that voids formed in fast-cooled, oxygen-doped floating-zone crystals can dramatically change morphology depending on the oxygen concentration. Without oxygen, or at low cooling rates, voids can grow into single highly faceted octahedral structures with length scales of about 100 nm. As the oxygen concentration and cooling rate is increased, the void morphology becomes less regular and eventually transforms into a core-and-halo structure. The halo is comprised of very small vacancy clusters with characteristic size of 2-3 nm.

In this talk, we present a highly efficient and accurate lattice kinetic Monte Carlo (LKMC) model [2] to explain these observations. We show how the presence of oxygen, which serves to reversibly trap single vacancies, affects the aggregation process and leads to structures that are in agreement with experimental measurements. The LKMC model is based on a bond-counting framework and has been carefully verified by direct comparison to large-scale molecular dynamics simulations. We investigate the effect of trap density, trap binding strength, and trap capture radius on the predicted morphologies and make comparisons to experimental data.

[1] K. Nakai, A. Akari, A. Huber, W. Haeckl, and W. von Ammon, Void formation mechanism in floating zone silicon crystal. 4th International Symp. Adv. Sci. Tech. Silicon Mater, JSPS Silicon Symp. Kona, Hawaii (2004) 23.

[2] J. Dai, J. M. Kanter, S. S. Kapur, W. D. Seider and T. Sinno, On-lattice kinetic Monte Carlo simulations of point defect aggregation in entropically influenced crystalline systems. Phys. Rev. B, 72 (2005) 134102.