Optical properties of zinc-blende semiconductor alloys: Effects of epitaxial strain and atomic ordering

Abstract

Spontaneous ordering of III-V alloys is known to cause a band-gap reduction \ensuremath{\Vert}\ensuremath{\Delta}${\mathit{E}}_{\mathit{g}}$\ensuremath{\Vert} and a splitting \ensuremath{\Delta}${\mathit{E}}_{12}$ of the valence-band maximum. Strain also leads to a valence-band splitting and, depending on the sign of the strain \ensuremath{\epsilon}, to an increase (for \ensuremath{\epsilon}0) or decrease (for \ensuremath{\epsilon}>0) in the band gap. We present a general theory explaining how the strain produced by lattice mismatch with the substrate interacts with ordering effects. We find for (001) strain and (111) ``CuPt'' ordering that (i) atomic ordering removes the cusp in the band gap vs strain curve of random alloys; (ii) epitaxial strain always leads to an increase in the ordering-induced valence-band splitting \ensuremath{\Delta}${\mathit{E}}_{12}$; (iii) atomic ordering reduces the slope of \ensuremath{\Delta}${\mathit{E}}_{12}$ vs strain; (iv) while (a) strain, (b) ordering, and (c) clustering can all lead to a band-gap reduction, we show here that the three effects can be partially distinguished on the basis of a \ensuremath{\Delta}${\mathit{E}}_{12}$ vs \ensuremath{\Vert}\ensuremath{\Delta}${\mathit{E}}_{\mathit{g}}$\ensuremath{\Vert} plot; (v) the wave-function type at the valence-band maximum (VBM) (and, hence, the cause of the splitting) can be further determined by measuring the polarization dependence of the intensities of the transitions between the VBM split components and the conduction-band minimum; (vi) we predict that ordering can significantly enhance the degree of spin polarization of photoelectrons emitted from the VBM. Ordered III-V alloys can thus be used as a good polarized electron source. These general results open avenues of band-gap engineering by combining epitaxial strain with atomic ordering. Specific experimentally testable predictions are presented.