Real-space real-time atomic studies of growing
surfaces
The growth of semiconductors by MBE (Molecular Beam Epitaxy), so that
the thickness of the layers is controlled to atomic precision, is
growing in importance for the development of new devices that take
advantage of quantum effects to achieve higher speeds or better
performance.
Traditionally, studies of growth have been done using indirect
techniques, usually diffrection techniques, such as RHEED (Reflectance
High Energy Electron Diffraction). Here the surface roughness, and the
growth mode are inferred from their effect on the intensity of the
RHEED pattern.
In order to gain local, real-space information, the
surfaces may be studied using STM (Scanning Tunnelling Microscopy),
but in order to do this in a conventional microscope, the growth is
interrupted, and the sample is quenched to room temperature, so as to
freeze-in the growth morphology. However, these experiments suffer
from the problem of relating the quenched surface to the growing
surface, and dynamic effects are impossible to study.
A new
technique, high-temperature STM, with in-situ dosing of
reactants, allows us to look at the growing surface at atomic
resolution, and therefore to observe the effect on surface morphology,
diffusion rates etc. much closer to real conditions.
A pair of STM images, showing the growth of islands are shown below. In the left-hand image, there is an atomic step running roughly vertically down the picture, 0.13 nm high. The bright features are islands of newly-grown silicon. The direction of the long axis of the islands alternates from one atomic step to the next. As more silicon is deposited, these thin islands grow and merge to form a new layer of silicon. A more-developed stage of growth is shown in the right-hand picture. Here over half of the first epitaxial layer has been completed, and some second-layer islands are beginning to nucleate on top of the larger first-layer islands.
Several experimental groups are doing hot STM growth studies, using beams of atomic silicon to study silicon MBE. The first two of these (Ganz and Swartzentruber) usually dose with silicon with the STM tip retracted and then observe the reactions between atoms, dimers and dimer complexes as a function of time. Measurements of atom and dimer diffusion have been made using atom-tracking STM, and the reaction pathway from silicon atoms to small islands is being studied.
Voigtlaender's group, by contrast study growth under a continuous flux, at large scales, and study the transition from island growth to step-flow growth as a function of flux and temperature. There is a strong line-of-sight shadowing of the silicon flux by the tip, which is partially alleviated by having the atomic beam at a low angle.
In our work, we are studying CVD (Chemical Vapour Deposition), otherwise
known as GS-MBE (Gas-Source MBE). GS-MBE is an alternative method for
the growth of silicon and germanium alloys, which is thought to reduce
problems of segregation of the germanium to the growth surface, due to
the presence of hydrogen. For this purpose, we use gas precursors,
disilane (Si 2H 6) and germane (Ge H 4).
One experimental advantage of using gas sources
is that tip shadowing is not so geometric as an atomic beam of silicon.
There is a local reduction in flux underneath the tip, however.
The disilane breaks up to form silicon dimers, and atomic hydrogen. The
hydrogen desorbs from the surface slowly, so that the dynamic surface
hydrogen coverage depends upon the disilane flux and the temperature of
the substrate. At temperatures where the hydrogen desorption
rate is higher than the silicond deposition rate, around 670 K for the
range of fluxes we have used, this gas can be used to grow layers of silicon,
at reasonable rates. Below this temperature, the hydrogen
saturates the step edges, blocking step adsorption, and promoting
island growth.
Two aspects of GS-MBE growth have been studied: the
nucleation pathway of silicon dimer strings from disilane fragments;
and the growth of larger islands and the transition from island growth
to step-flow growth as a function of flux and temperature. In this way,
our work bridges the work of the three MBE groups.
The details of the reaction pathway from SiH 2 groups to dimer strings, may be found here [Still under construction]
Larger-scale growth issues, such as the poisoning of growth by surface hydrogen, and the transition from island growth to step-flow growth may be found here.
See my publications page for the three papers that were written.