Modeling and simulation of arsenic activation and diffusion in silicon

Abstract

One of the critical problems that exist in a Very Large Scale Integrated circuit (VLSI) process modeling today is the prediction and simulation of arsenic behavior in silicon during the integrated circuit fabrication. This work is focused on understanding and modeling of the physical processes which occur during arsenic diffusion and deactivation.Arsenic deactivation is a complicated process due to fast kinetics and strong interstitial ejection which accompanies deactivation. We have used ab-initio calculations in order to gain insight into the fundamental processes involved in arsenic activation/deactivation. It has been proposed that several second nearest neighbor As atoms (two or more) may kickout an adjacent Si atom forming an electrically inactive arsenic-vacancy cluster and a self-interstitial. A physical model based on this mechanism was derived and used successfully to match a variety of electrical data as well as interstitial supersaturation data measured during deactivation. This model was applied to understanding and prediction of the ultra shallow junction formation. Several effects such as rapid As diffusion at high concentrations and the existence of grown-in vacancies after amorphous/crystalline regrowth must be included to account for the discrepancies between what seems to be a mobile fraction of the profile and the electrically active fraction.An integral part of this work involves the simulation of self-interstitial cluster formation. Total energy calculations based on empirical Molecular Dynamics (MD) simulations were performed in order to study the energetics of interstitial aggregates as a function of size and configuration. These results provide insight into ion implant annealing processes. A Kinetic Precipitation Model (KPM) was used to analyze the evolution of {311} defects. We discuss small interstitial clusters and their role in the initial stages of annealing after ion implantation and during As deactivation.