DX center in Ga1-xAlxAs alloys.

Abstract

From the electron-emission and -capture variations with the alloy composition, we deduce that the DX center in ${\mathrm{Ga}}_{1\mathrm{\ensuremath{-}}\mathrm{x}}$${\mathrm{Al}}_{\mathrm{x}}$As alloys is linked to the L conduction band. This is verified by the study of the DX center in GaAs-${\mathrm{Ga}}_{1\mathrm{\ensuremath{-}}\mathrm{x}}$${\mathrm{Al}}_{\mathrm{x}}$As superlattices, where it is observed that the DX center does exist only in GaAs-AlAs superlattices such that the L miniband which replaces the L band of the AlAs barrier is at the same energy as the bottom of this L band, although the X conduction band has been replaced by a \ensuremath{\Gamma} miniband. This property can only find an explanation in the effective-mass approximation and cannot be justified in large--lattice-relaxation models. Within this model the DX level is the effective-mass state of the doping impurity associated with the L band which undergoes a shallow-deep instability due to intervalley mixing. This description implies a small, if any, electron-phonon interaction. From our electron-paramagnetic-resonance studies we conclude that the large threshold for optical ionization involves a transition to a higher state than the bottom of the conduction band and must probably be attributed to an internal transition. The barrier for electron capture is explained as originating from a change of the local force constants of the defect with its charge state. The variation of the capture barrier height with alloy composition is quantitatively accounted for by electron emission from the bottom of the conduction band into the L band, followed by a multiphonon-emission process over a constant barrier.