Modeling co-implanted silicon and beryllium in gallium arsenide

The implantation and diffusion of Si and Be are modeled using SUPREM 3.5, a GaAs process simulator. According to the new model used for Be diffusion with a hole-dependent effective diffusivity, the presence of the Si should decrease the Be diffusion in the overlap region due to a Fermi-level effect, and the experimental results confirm this. The damage created by the Si implantation is also seen to affect the Be diffusion, as confirmed by Ar co-implantation, but the relatively simple diffusion models in SUPREM 3.5 do not currently model this effect. The Be segregation coefficient is seen to be a function of implant energy. The diffusion of Si, which is negligible to begin with, is not affected by the presence of Be. The implantation profiles of both Be and Si are not appreciably affected by the implantation damage caused by the other dopant and are well modeled by Pearson IV parameters. (author)