The interdiffusion of Si, P, and In at polysilicon/GaAs interfaces

Abstract

Gallium arsenide (GaAs) encapsulated at 450 °C with thin films of amorphous silicon has been annealed at temperatures up to 1050 °C and the resulting polysilicon (poly-Si)/GaAs interfaces investigated with secondary ion mass spectroscopy, Rutherford backscattering, and transmission electron microscopy. Little or no interdiffusion is detected at undoped Si/GaAs interfaces whereas Si diffuses from P- or As-doped Si to depths as great as 550 nm in the GaAs after 10 s at 1050 °C. The flux of Si into the GaAs is correlated with the flux of Ga and As into the Si and both increase with increases in the dopant concentration of the Si. The diffusion of other Si dopants into the GaAs, including P and In, is also detected. This enhanced diffusivity of Si, P, and In in GaAs results from the diffusion of point defects into the GaAs created by the diffusion of the Ga and As into the encapsulant. Numerical simulations using position-dependent impurity diffusion coefficients predict that Si, P, and In diffusivities in GaAs at doped poly-Si interfaces are enhanced by factors of 104 above their respective intrinsic bulk equilibrium diffusivities, where known.

Dopant diffusion and segregation in semiconductor heterostructures: Part III, diffusion of Si into GaAs

“…The source material for diffusing Si into GaAs is a polycrystalline Si layer deposited on the surface of the GaAs crystal [1,2]. The Si-source/GaAs system constitutes a heterostructure.…”

“…The third is a self-consistent calculation of the solubilities of Si on the two GaAs sublattice sites [10]. In this paper we present an analysis of Si indiffusion results [1,2] by incorporating these advances in the Fermi-level effect model suggested by Yu et al [4]. Satisfactory fits to these Si indiffusion results were obtained, in particular the anomalous portions of these Si profiles close to the GaAs surface.…”

“…To illustrate the principles involved, Deppe and Holonyak [5] considered the singly-negatively-charged Ga vacancies. On the other hand, Yu et al [4] have quantitatively fitted some of the available experimental data [1,2] by considering the diffusion of both Si + Ga and Si − As species. For Si + Ga , they used mainly the triply-negatively-charged Ga vacancies V 3− Ga , and to a lesser extent also the neutral Ga vacancy species V 0 Ga , as the responsible point defects.…”

“…Greiner and Gibbons [1] assumed that Si diffusion is predominantly carried by SiAs−SiGa pairs. Kavanagh et al [2] suggested that the concentration dependence is due to a depth-dependent vacancy concentration generated by the Si source material containing As or P. Tan and Gösele [3], Yu et al [4], and Deppe and Holonyak [5] proposed that the concentration dependence of the Si diffusivity is an effect of the Fermi level, because the point defect species governing Si diffusion in GaAs are the negatively-charged Ga vacancies. To illustrate the principles involved, Deppe and Holonyak [5] considered the singly-negatively-charged Ga vacancies.…”

“…Yu et al [4] also showed that experimental results on Si diffusion into n-type (Sn-doped) GaAs are in contradiction with the Greiner-Gibbons pair-diffusion model [1], and hence it needs not to be further considered. On the other hand, the vacancy injection idea of Kavanagh et al [2] and the Fermi-level effect are not mutually exclusive on a qualitative basis.…”