Indirect tunneling in a short GaAs-AlAs superlattice detected by photoluminescence under hydrostatic pressure.

Abstract

We have measured, as a function of hydrostatic pressure, the photoluminescence (PL) associated with the quasiconfined states in a short GaAs-AlAs superlattice (SL) embedded in a p-i-n diode. In order to be able to observe a measurable PL signal, we find that it is necessary to short the SL diode to allow tunneling of electrons and holes from the depletion region on either side into the SL. We observe quenching of this PL at approximately 15 kbar---the pressure at which we expect the changeover from a direct (or type-I) to an indirect (or type-II) SL. This indicates a nonresonant tunneling process through the AlAs barriers mediated by the lowest potential profile in the system (i.e., \ensuremath{\Gamma}-X). In previous work we found that our results were best modeled by using a \ensuremath{\Gamma}-X barrier and light holes in an effective-mass approximation. However, by using an empirical pseudopotential model, which takes account of the full band structure of the semiconductors, and comparing with our previous calculations, we see that the resonance energies of the electrons themselves are determined predominantly by the \ensuremath{\Gamma}-\ensuremath{\Gamma} potential profile. By also allowing the well widths to vary by up to a monolayer, we see that our experimental results are reproduced accurately by assuming recombination between n=1 electrons and n=1 heavy holes, where the states they occupy are determined by a \ensuremath{\Gamma}-\ensuremath{\Gamma} potential profile. We have also modeled the energy dependence of the pressure coefficient, taking into account well-width variations, effective-mass variations, and the different pressure coefficients for the two materials, and find good agreement with the experimental results for the electron--to--heavy-hole recombination process. We deduce that the value for the pressure coefficient of the \ensuremath{\Gamma} gap in GaAs most consistent with our data is 11.0\ifmmode\pm\else\textpm\fi{}0.1 meV ${\mathrm{kbar}}^{\mathrm{\ensuremath{-}}1}$.