We report on the use of electrochemical atomic-layer epitaxy (EC-ALE) to gr
ow thin-films of the III-V compounds InAs, InSb, and an InAsxSb1-x, superla
ttice. EC-ALE is a method for forming compound semiconductors with improved
control, compared to other electrodeposition methodologies. It involves th
e use of surface limited reactions to form deposits an atomic layer at a ti
me, in a cycle. The EC-ALE cycle uses underpotential deposition (upd) to fo
rm atomic layers of each of the component elements. One cycle ideally produ
ces one monolayer (ML) of the desired compound. Studies to optimize the InA
s cycle are reported, specifically the dependence on the In and As depositi
on potentials. These studies show that the potentials must be adjusted for
each of the first 25 or more cycles, as a contact potential between the Au
substrate and the growing semiconductor develops. After deposition of this
initial 'buffer layer', steady state conditions are reached, and the same p
otentials can be used without change, for the remaining cycles. The formati
on of InSb has also been investigated, and the EC-ALE growth of InSb deposi
ts is reported for the first time. Due to a 6% lattice mismatch, and a less
than fully optimized cycle, the InSb deposits on Au appear composed of 70
nm particles. By combining the InAs and InSb programs, a superlattice was f
ormed with 41 periods, where each period involved ten cycles of InAs follow
ed by ten cycles of InSb. X-ray diffraction (XRD) indicated a period of 5.5
nm, whereas a 7.4 nm period was expected, based on 1 ML/cycle and the (111
) interplanar spacing, derived from the lattice constants for InAs and InSb
. Given the stoichiometry of the resulting deposit, and the shorter periodi
city observed, it appears that 1 ML/cycle of InAs was formed, while only a
1/2 ML/cycle of InSb was obtained. IR absorption measurements indicate that
the deposit was red shifted relative to the lower bandgap compound, InSb (
0.17 eV), which is consistent with a type II superlattice. If an alloy had
been formed, the bandgap should have been a linear function of the bandgaps
and relative mole fractions of InAs and InSb, or about 0.31 eV, twice the
observed bandgap. (C) 2001 Elsevier Science B.V. All rights reserved.