Atoms that are deposited on crystal surfaces, through a method known
as molecular beam epitaxy, often form surfaces covered with numerous
small mounds rather than smooth layers, if the substrate temperature
is sufficiently low. For higher temperatures, an atom near the top of
a mound can often move about and diffuse down toward the crystal surface.
Conventional wisdom holds that upward diffusion, on the other hand,
is essentially negligible. Recently, however, a collaboration of researchers
at the INFM-Università di Genova in Italy, the Chinese Academy
of Sciences, and Oak Ridge National Laboratory has found that deposited
atoms may sometimes diffuse upward spontaneously, forming faceted mountains
that tower over the surrounding crystal plane. Although the formation
of faceted nanocrystals has been observed before, these were generally
thought to be due to a mismatch between a crystal substrate and the
crystal structure of the deposited atoms (for example, when germanium
atoms are deposited on silicon, differences in the spacing of the two
types of crystals lead to a strain that encourages the growth of large,
hut-shaped crystals). The new research, by contrast, reveals for the
first time that the crystalline mountains (see
image) can form even when the deposited atoms and the substrate
crystal consist of the same element, and no strain energy is involved.
Specifically, aluminum atoms deposited on an aluminum crystal substrate
may diffuse upward into crystal structures that rise upward as much
as ten times higher than the thickness of the surrounding planes.
Computer simulations seem to indicate that the growth may be caused
by processes thought to be insignificant in previous deposition studies.
In particular, an atom sitting at the inner corner near the base of
a crystal protrusion may jump out of place and onto the crystal facet,
or a pair of atoms can conspire to exchange positions as they leapfrog
up a crystal slope. The counterintuitive formation of tall nanocrystals
via upward diffusion of aluminum atoms only occurs within a temperature
window of about 330 K to 500 K, when the total crystal surface coverage
exceeds critical values of about 10 or more deposited layers, depending
on the specific temperature. The researchers (Francesco
Buatier de Mongeot, +39-10-3536324, and Zhenyu
Zhang) predict that the often neglected processes leading to upward
atom diffusion are likely to be important for other crystals grown via
molecular beam epitaxy, leading to much richer dynamics in the growth
of thin films than previously suspected. (F. Buatier de Mongeot et
al., Physical Review Letters,
upcoming article, probably 4 July)