Can recoil distribution models account for end-of-range damage?

Abstract

A quantitative evaluation of two existing models for end-of-range damage in Ge + implanted silicon is presented. The models propose that the amount of end-of-range damage is related to either (1) the number of ions or (2) the number of recoiling Si atoms that come to rest beyond the amorphous/crystalline (α/c) interface. TRIM (Monte Carlo) simulations of the implants studied were used to obtain the predictions of each model. These predictions were compared to experimental results obtained from implants at 30, 100 and 150 keV and doses of 5 × 10 14, 1 × 10 15 and 2 × 10 15. Plan view and cross-sectional TEM observations were used to determine the concentration of atoms bound by the end-of-range dislocation loops and the location of the α/c interface. Increasing the energy or dose resulted in significant increases in the amount of end-of-range damage. Neither model was able to consistently predict these trends at the lower doses and energies where ion beam induced epitaxial crystallization was suppressed. However, the recoil model was found to be superior overall to the ion model when compared to data from a wider range of implant conditions.