First-principles-based predictive simulations of B diffusion and activation in ion implanted Si

We present a kinetic Monte Carlo model for boron diffusion, clustering and activation in ion implanted silicon. The input to the model is based on a combination of experimental data and ab initio calculations. The model shows that boron diffusion and activation are low while vacancy clusters are present in the system. As the vacancy clusters dissociate, boron becomes substitutional and the active fraction increases rapidly. At the same time, the total boron diffusion length also increases rapidly while interstitial clusters ripen. The final burst of boron diffusion occurs as the large interstitial clusters dissolve, but most of the transient diffusion of the implanted boron has already taken place by this time. We show that these results are in excellent agreement with experimental data on annealed dopant profiles and dopant activation as function of annealing time.