Provided that the shape, size, and composition profile of semiconductor-emb
edded quantum dots are given, theory is able to accurately calculate the ex
citonic transitions, including the effects of inhomogeneous strain, alloy f
luctuations, electron-hole binding, and multiband and intervalley coupling.
While experiment can accurately provide the spectroscopic signature of the
excitonic transitions, accurate determination of the size, shape, and comp
osition profile of such dots is still difficult. We show how one can arrive
at a consistent picture of both the material and the electronic structure
by interactive iteration between theory and experiment. Using high-resoluti
on transmission electron microscopy, electron-energy-loss spectroscopy, and
photoluminescence (PL) spectroscopy in conjunction with atomistic empirica
l pseudopotential calculations, we establish a model consistent with both t
he observed material structure and measured electronic/optical properties o
f a quantum dot sample. The structural model with best agreement between me
asured and predicted PL is a truncated cone with height 35 Angstrom, base d
iameter 200 Angstrom, and top diameter 160 Angstrom, having a nonuniform, p
eaked composition profile with average 60% In content. Next, we use our bes
t structure to study the effect of varying (i) the amount of In in the dots
, and (ii) the spatial distribution of In within the dots. We find that by
either increasing the amount of In within the dot or by concentrating a giv
en amount of In near the center of the dot, both electrons and holes become
more strongly bound to the dot. A small change of In content from 50 to 60
% causes an exciton redshift of about 70 meV. Changing the composition prof
ile from a uniform In distribution to a centrally peaked distribution can r
edshift the exciton by an additional 20-40 meV.