Interdiffusion experiments and results for InP/GaInAs(P) heterostructures were considered in terms of a thermodynamic model. Important factors which affected interdiffusion in the GaInAsP system were shown to include a miscibility gap, differing diffusivities on each of the sub-lattices of the 2 materials, Fermi level or impurity-induced changes in diffusivity or diffusion mechanism, and the type of experiment. When a miscibility gap was present, the activity coefficients and solubilities of all of the species varied near to a heterojunction and caused the interdiffusion to become strongly composition-dependent. At the usual growth and annealing temperatures, many superlattices were expected to equilibrate as 2 quaternary superlattices rather than as an homogeneous alloy. Differing diffusivities on the sub-lattices of a superlattice could lead to widening or narrowing of quantum wells. When this occurred, optical measurements of the band-gap energy were likely to be misleading, because of quantum size effects. The diffusivity on each sub-lattice could be altered by the presence of group-II, -IV, or -VI dopants. Diffusion on the group-III sub-lattice in p-type GaInAsP was found to be consistent with an interstitialcy mechanism. The mechanism remained unknown for n-type doping and for the group-V sub-lattice. Poorly designed and controlled experiments were found to be associated with large discrepancies in the observed diffusivities, with unreliable concentration profiles, and with the appearance of new condensed phases in the solid. Experiments indicated that the ordered Cu-Pt structure which was often found in GaIn(As)P epilayers was unstable, and was not strain-stabilized relative to the disordered structure at normally used growth and annealing temperatures.

R.M.Cohen: Journal of Applied Physics, 1993, 73[10], 4903-15