308
The European Physical Journal D
When the coverage is increased more big clusters will
be formed during the deposition. Some of them may still
diffuse on the surface and aggregate with others. But for
large clusters, it is rather difficult to be absorbed by an-
other one and completely coalesce to a larger cluster. In-
stead, they can eliminate the interface by neck forma-
tion between them and form some short one-dimensional
cluster assemblies under the high-temperature ambiance.
What’s more they can grow further into long wire by neck-
link with other clusters or cluster assemblies. Such mor-
phologies can be observed from specimen S2 with coverage
of about half cluster monolayer. The structure evolution
process can be deduced by a serial of SEM images taken
from it locally, namely Figures 2a–2d.
Fig. 1. Isolated clusters are observed in sample S1 with a
coverage of 0.1 cluster monolayer.
Isolated large clusters are still remained in Figure 2a.
They are randomly distributed and piled up and keep the
spherical appearance. As a contrast clusters in Figure 2b
are no more distributed in random. Instead, they form
short one-dimensional assemblies, within which adjacent
clusters are connected with each other by neck formation.
Figure 2c shows similar structure, but the length of the
cluster assemblies is much longer with less breakpoints.
The morphologies of these one-dimensional cluster assem-
blies are obviously undulated, owing to the size mismatch
between the sintering necks and the clusters. A relatively
smooth nanowire is shown in Figure 2d. The appearance of
the smooth morphology may be explained by considering
the sintering mechanism. Atomic diffusion across the neck
during the sintering reduces the size difference between
the necks and the clusters. If the sintering time is long
enough, the necks are eliminated and the whole nanowire
develops into a smooth one in the end.
rate occurs around 970 ◦C evaporation temperature when
the carrier gas pressure is kept at 300 Pa. The deposition
rate increased from 0.15 A/s at 950 C to about 30 A/s at
970 ◦C in a few seconds. Therefore, the inference from the
deposition under the evaporation temperature lower than
◦
˚
˚
◦
970 C is neglectable, if only the structure evolution at a
◦
temperature of 970 C is concerned. Four specimens were
deposited at 970 ◦C with different coverage and named S1,
S2, S3, and S4 with the coverage from low to high. After
deposition, the electric current of the graphite heater was
turned off immediately and the oven temperature dropped
to 950 ◦C within 10 seconds. The morphology of each spec-
imen is observed by scanning electron microscopy (Leo-
1530).
3 Result and discussion
The series of SEM micrographs of Figures 2a−2d il-
lustrate the transitional stage when the cluster assem-
blies are formed by cluster aggregation. At high cover-
age, large clusters are formed from the growth of smaller
clusters. Under the high temperature ambiance clusters
are high mobile and can rearrange on the substrate sur-
face by themselves [10]. For large clusters, coarseness is
energy-unfavorable. Instead, necks are formed due to the
atomic diffusion on the interfaces while these clusters con-
tact and connect them to a one-dimensional structure.
Given enough annealing time, nanowire can be synthe-
sized by cluster sintering. An interesting phenomenon here
is the ordered arrangement of cluster assemblies. Since sin-
tering itself can not result in ordered growth of nanowires,
this phenomenon is discussed based on the steering of car-
rier gas that lead to the preferred arrangement under the
direction of gas flow. The detail will be published else-
where [9].
On the common process of gas condensation of SiO va-
por, SimOn clusters (zero-dimensional) are formed by free
nucleation in the vapor at the first step shortly after the
evaporation of the SiO [8]. At the early stage of deposition,
the cluster number density is low both in the gas phase as
well as on the substrate surface. Figure 1 is a SEM micro-
graph of sample S1 with a coverage of less than 0.1 cluster
monolayer deposited within this stage. Only individual
clusters with the diameter about hundreds of nanome-
ters are randomly distributed without any aggregating ev-
idence. It should be noted that the lack of smaller clusters
on the specimen surface at such a low coverage may re-
veal that the clusters are highly mobile on the substrate,
considering the substrate is close to the oven and the de-
◦
position temperature is about 900 C. Coalescence takes
place between smaller clusters on their diffusion path. The
clusters may also be pinned at some sites on the substrate,
such as the surface defect, where smaller clusters on diffu-
According to our suggested mechanism, more and more
sion can be absorbed and the growth process conduce to smooth nanowires should be synthesized if the coverage is
the formation of larger clusters. As a result, if the density further increased and annealing time is further prolonged.
of the pinning sites is not so high, only some big clus- Figures 3 and 4 show the micrographs of sample S3 with
ters can be remained individually on the substrate at low a growth time of 5 minutes and S4 with a growth time
coverage. These big clusters can either serve as seeds for of 10 minutes respectively. Smooth nanowires that can be
one-dimensional growth or be building blocks of the one- identified in Figure 3 are much more than those in Fig-
dimensional structure under high temperature [7] for fur- ure 2. In fact, smooth nanowires are the main portion of
ther growth.
specimen S3. The emergence of smooth nanowire is the
Zhou
