eLearning

Atomic force microscopy was performed using tap-
ping mode on the surface of island covered samples
(Fig. 2). The density of islands on InAlAs templates is
high (about 1.5 × 1011 cm–2 for 3 MLs InAs coverage) as
compared to islands grown on GaAs (several 1010 cm–2).
The dependence of the density with the InAs coverage
is shown in Fig. 3. A first-order phase transition
model3 fits the present data with a critical InAs
coverage of 1.7 ML on the GaAs buffer, but 1.2 ML and
0.9 ML on the pseudomorphic and metamorphic buff-
ers, respectively. For InAs on GaAs the results agree
with RHEED observations. The indium segregation
during the growth of InAlAs [about 0.9 ML of InAs at
the surface as measured by x-ray photoelectron spec-
troscopy (XPS)] partially explains the early onset of
3D growth of InAs on InAlAs layers, but not the
higher density; this confirms that Al containing buff-
ers with a rougher surface generate high density of
islands.
6,7 The AFM image of a MM-InAlAs layer
covered with 1.7 ML of InAs (Fig. 2a) shows that the
first islands nucleate at step edges of the rough cross-
hatched surface, not on the topmost regions which are
known to be less strained. As a result, the electron
beam of the RHEED can not probe the nucleation of
these islands. The reduction of the mismatch strain
between InAs and the relaxed InAlAs layer and the
lack of surface steps in the topmost regions of the
cross-hatch delay the onset of the 3D growth in these
regions probed by the RHEED. This explains the dis-
crepancy between the RHEED evolutions and the is-
land densities on metamorphic InAlAs buffer layers.
However, at higher InAs coverage, islands also nucle-
ate on topmost regions of the cross-hatch (Fig. 2b);
this results in a uniform coverage of the surface with
islands. The lattice mismatch between InAs and the
GaAs (or the pseudomorphic InAlAs) is high (da/a =
7%). However, the strain relaxation in the metamor-
phic InAlAs buffer provides a larger in-plane lattice
parameter at the surface. Then the lower mismatch
between InAs and the metamorphic layer leads to
very few coalescence of the islands as compared with
InAs grown on the pseudomorphic InAlAs buffer
(Fig. 2c) which presents the same mismatch as InAs
on GaAs (Fig. 2d).
Figure 4 shows the height and the radius of the
islands measured by AFM. The height of InAs islands
grown on InAlAs layers is small (less than 5 nm) as
compared with InAs on GaAs (8 nm for 3 MLs InAs
coverage). Furthermore, the islands grown on the
metamorphic buffer are smaller than those grown on
the pseudomorphic buffer and their size is similar to
that of the first islands nucleated on the GaAs buffer
layer (1.7 ML InAs coverage). A general trend “high
density-small size’’ is observed and the ratio of the
total volume in islands over the amount of InAs con-
firms that less surface mass transport is involved on
the InAlAs buffer layers.