Channeling characterization of defects in silicon: an atomistic approach

Abstract We review here the possibilities opened by a recent development of the Monte Carlo binary collision approximation (MC-BCA) simulation of Rutherford backscattering spectrometry-channeling (RBS-C) spectra for the study of radiation damage in monocrystalline materials. The ion implantation of silicon has been chosen as a case study. Atomic-scale modeling of defect structures was used to determine the location of interstitial atoms in the host lattice. Among possible candidate defects, we have considered the elementary hexagonal, tetrahedral, 〈1 1 0〉-split interstitials, the Bond-defect and one type of tetra-interstitial cluster. For each defect model a large Si supercell was populated with a proper defect depth distribution and then it was structurally relaxed by the application of the classical EDIP potential. This model system was then given as an input to the MC-BCA simulation code and the spectra corresponding to nine different axial and planar alignments were calculated. For low defect concentration (a few atomic percent), the scattering yields are strongly dependent on the orientation and a distinct signature characteristic of the limited number of allowed interstitial positions in Si could be found. The comparison of simulations and experiments in the case of 180 keV self ion implantation allowed the identification of the dominant interstitial defect whose structural properties are represented by the split-〈1 1 0〉 interstitial. By increasing the concentration of defects (and their mutual interaction) the technique looses sensitivity and, at the same time, the contribution of lattice relaxation becomes important. Under these conditions, although the RBS-C response becomes similar to the one obtained from a random distribution of displaced atoms, the major structural features of a heavily damaged sample could be still observed.