Monte-Carlo Simulations of the Clustering of Carbon in Silicon and Its Influence on Carbon Diffusion

Carbon in silicon appears in high concentrations, ranging from 10 sup (16) to 10 sup (18) cm sup (-3), well above its solubility at the usual annealing temperatures. Carbon diffusion in Si is therefore intimately linked to C clustering and precipitation. Only simulation tools that take both phenomena into account will be able to model experiments accurately. Such a tool is the Monte-Carlo atomistic simulator DADOS. Through the use of DADOS it has been shown earlier that the formation of carbon/Si self-interstitial clusters with a low content of Si interstitials accounts for the faster diffusion of C compared with that predicted by the kick-out mechanism alone. In this work, we show that, at higher temperatures, the Carbon/Si self-interstitial clusters are bigger and have a higher C content. This fact results in an immobilized fraction of C in the high concentration regions, in good agreement with the experimental results of C diffusion in doping-superlattice C structures grown by MBE. Furthermore, under conditions of thermal generations of intrinsic point defects, the low temperature dependence of Si interstitial under-saturation in the C rich-regions can be well reproduced. Finally, we will show that this model is also consistent with recent experiments of formation of the C sub 2 cluster detected by infrared absorption spectroscopy at room temperature.