Influence of dislocations and annealing cap on the electrical activation of silicon implanted in semi-insulating GaAs: Implications for field-effect transistors

Abstract

Numerous studies have shown that the threshold voltage (Vth) of field-effect transistors fabricated on Czochralski-grown semi-insulating GaAs substrates using silicon ion implantation is sometimes dependent and sometimes independent of proximity to dislocations, according to the processing method used. We present a comprehensive model that reconciles these diverse observations and semiquantitatively describes the dependence of Vth on EL2 concentration, boron concentration, silicon fluence, proximity to a dislocation, and post-implant annealing method. Under conditions holding in most of the cited studies, silicon fluence between 2×1012 cm−2 and 5×1012 cm−2 at 60 keV followed by an anneal with a SiN cap, we find that Vth depends on the substrate concentration of EL2 and on the concentration of the boron-silicon acceptor complex BGaSiAs. We suggest that a nonuniform stress field across a wafer reduces the binding energy of BGaSiAs complexes within 50 μm of a dislocation resulting in a local increase in silicon activation and a more negative Vth. When the post-implant anneal is done under a SiO2 cap we suggest that the infusion of VGa due to the out-diffusion of Ga eliminates the stress field nonuniformities and results in Vth showing no dislocation-proximity effect. Under these latter conditions we find that any variation in local substrate EL2 concentration across the wafer is reflected in a variation of Vth.