Chemistry Reference
In-Depth Information
terraces, which were almost free of steps. It is interesting to note that the rearrange-
ment of the regularly distributed steps into step bands is a reversible process in the
sense that one can transform the developed step bands into a regular system of steps
by reversing the direction of the electric current. This phenomenon (electric current-
induced bunching of steps) has a very complicated temperature dependence - there
are three temperatures (in the interval between 1170 K and the melting temperature)
where bunching-debunching transition takes place at a fixed direction of the electric
current [
1
,
2
]. It is instructive to mention again that the observations outlined above
refer to the step dynamics at vicinal (111) face of Si crystal.
The electric current flowing through the Si crystal has also an impact on the step
configuration of the (001) vicinal surface. The (001) surface of Si crystal is known
to manifest surface reconstruction and to contain (2
2) domains formed
by the dimerization of the surface atoms, which have two broken bonds each. Due to
the surface reconstruction the terraces at the (001) vicinal surface are not identical -
the direction of dimerization rotates by 90
◦
when one moves from one terrace to the
adjacent one. When the electric current through the Si crystal is perpendicular to the
steps it has a dramatic effect on the ratio of the areas of the neighbouring terraces.
One type of terraces (say (2
×
1) and (1
×
×
1)) increase in width on expense of the terraces of the
other type (terraces with reconstruction (1
2)). If the electric current direction is
reversed the narrow terraces increase their width whereas the wide terraces shrink
[
3
,
4
].
The REM observations of Latyshev et al. [
1
] and Kahata and Yagi [
4
]marked
the beginning of rather active research on electric current-induced bunching of steps
and conversion between 1
×
1 reconstruction domains at silicon surfaces. In
1990, Stoyanov [
5
] proposed a model explaining both phenomena step bunching and
conversion of reconstruction domains. A fundamental feature of this model is the
electromigration of the adatoms which are assumed to jump more frequently in the
direction of the electric current (heating the crystal) than in the opposite direction.
The proposed model, however, does not operate at the finest scale, where the crystal
growth is described by hopping of adatoms between different lattice sites on the
interface. Like the classic Burton-Cabrera-Frank (BCF) theory the model of crystal
growth and sublimation in the presence of electromigration of the adatoms operates
at the next scale where the basic elements are the terraces and the steps. Thus the
crystal growth is reduced to diffusion of the adatoms on the terraces (this process
is characterized by the coefficient
D
s
of surface diffusion) and their subsequent
attachment to the steps. The processes at the steps are not simple. The adatoms
find their “proper” sites in the crystal lattice after many “trials and errors.” The
atoms attach to the step edges, migrate along them, reach kinks, and attach into kink
position. The last event does not mean that the atom attaches to the crystal phase
“forever.” In fact, this atom could later detach from the kink position and even leave
the crystal surface by desorption. In the special case of crystal-vapour equilibrium
an atom attaching to a kink position has zero probability to stay in the crystal phase
“forever.” If, however, the vapour pressure is higher than its equilibrium value, an
atom attaching to a kink position has a chance to stay in the crystal phase “forever.”
As a result the crystal grows. In the scale where BCF operates the crystal growth
×
2 and 2
×
