Larger-scale growth studies

Using a high-temperature STM, we have been able to image the Si(001) surface while it is under a flux of disilane. This gas 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 above the hydrogen desorption temperature, around 670 K, this gas can be used to grow layers of silicon. Below this temperature, surface hydrogen desorbs very slowly, so that growth can only be achieved at very slow rates. The hydrogen also appears to saturate step edges, blocking step-flow growth, and promoting island growth.
The adsorbing dimers coagulate into islands, one atomic layer high, in the middle of the terraces. We have been investigating how the shape and size of the islands varies with growth temperature and disilane pressure.
At around 650 K, islands are stable, even under zero flux conditions. No step-flow growth occurs. As the flux is increased, there is a transition from a few large islands, with an aspect ratio around 3:1, to smaller, thinner islands, with a much higher aspect ratio. Eventually, at very high fluxes, the surface becomes clogged with disilane fragments, and no growth can occur.

At around 750 K, the steps are no longer clogged with hydrogen, and islands will grow or shrink according to the flux. Because we are able to grow while scanning, we have been able to take movies of the growing surface, and image both island growth and step growth.
Single frames from these movies are shown here. To reduce download times, each actual movie is shown on a separate page.

The mature growth surface is shown below. This is a large-scale pair of images of the Si(001) surface, about 140 nm in width. These two frames were taken about 13 minutes apart. The shape and size of the islands have changed, and the step edges have become smoother over time.

Dimer string nucleation

Nucleation of islands is first as dimer strings, due to a high sticking coefficent only at the ends of the dimer rows. This interaction pins dimers into long chains. The dimers that make up these chains are able to move form one chain to another. Thus a chain may fill in to form a complete dimer string, or else it may break up again. Both cases are seen in The first movie. Thus island formation is not irreversible condensation, but a dynamic process.

The effect of the hydrogen flux

By imaging between successive doses, we were able to grow a whole monolayer. This situation is effectively the zero-flux limit, and the surface has time to reach equilibrium. The islands are much larger than those in the images above. Coarsening amongst islands has therefore taken place. However, there has been no step-edge adsorption, as shown by the fact that there are islands very close to a B-type step edge.

We have found that as the hydrogen coverage increases, it progressively blocks surface diffusion, and prevents the growth of new islands. The next two movies show growth sequences at 650 K, over the same timespan (about 9 minutes).

At a flux of 5 x 10^-7 Pa, we were able to achieve a complete monolayer of growth (Movie 2). The islands are long and thin, about 1-2 dimer rows wide. After most of the surface is covered, the second layer begins to nucleate.

At 3 x 10^-6 Pa, the surface begins clean, as shown by the long streaks, but after a few minutes, the surface became clogged with disilane fragments (Movie 3), and only small, short dimer strings were able to form. Further exposure beyond the end of the sequence did not result in any significant increase in silicon coverage.

Higher-temperature growth

At 750 K, we were able to achieve stable growth, by both island nucleation and growth, and also by step flow. At this temperature, however, the hydrogen is no longer blocking step-edge adsorption, and so coarsening between islands and steps may be seen.
Movie 4 shows what happens under a relatively high flux. Two islands have nucleated at the boundary between two islands in the lower layer, and over the course of the movie they grow and merge. Other islands and the step edges also grow.

Movie 5 shows what happens under a lower flux. The islands which are present at the beginning of the movie are evaporating dimers faster than the disilane flux is replacing them, and so over time they shrink, and their material diffuses to the step edges. This is known as coarsening.

Conclusions

High-temperature STM studies of growth phenomena give a uniques insight into surface growth processes. While an island as an entity may appear to be stable, in reality, atoms are attaching to and detaching from the step edges continuously, and in the case of the dimer chains, dimers can in and out of the middle of the chain.
At larger scales, the surface hydrogen coverage, controlled by the temperature and disilane flux, have a large effect on the growth morphology. A low hydrogen coverage allows stable growth of many monolayers of silicon, while a large hydrogen coverage blocks adsorption of disilane, and diffusion of silicon atoms, preventing stable growth.