Nucleation and growth of Si nanowires from silicon oxide
N. Wang, Y. H. Tang, Y. F. Zhang, C. S. Lee, and S. T. Lee
*
Center of Super-Diamond and Advanced Films, Department of Physics and Materials Science, The City University of Hong Kong,
Hong Kong, China
~Received 18 August 1998; revised manuscript received 28 September 1998!
Nucleation and growth of Si nanowires by laser ablation and thermal evaporation of Si powder sources
mixed with SiO
2
have been investigated by means of transmission electron microscopy. At the initial nucle-
ation stage, Si oxide vapor condensed on the substrate and formed Si nanoparticles ~the nuclei of nanowires!.
Each Si nanowire nucleus consisted of a polycrystalline Si core containing a high density of defects and a Si
oxide shell. A growth mechanism was proposed based on the microstructure and different morphologies of the
Si nanowires observed. @S0163-1829~98!51348-3#
The synthesis of one-dimensional nanostructures is a
challenge in materials science. Since the successful growth
of Si whiskers by the vapor-liquid-solid ~VLS! reaction,
1,2
many efforts have been made to synthesize Si nanowires by
employing different methods, such as photolithography tech-
nique combined with etching
3–5
and scanning tunneling
microscopy.
6,7
Recently, Si nanowires have been success-
fully synthesized by a different method, namely, the laser
ablation of metal-containing Si targets.
8–11
Previous
investigations
8,9
showed that metal or metal-silicide nanopar-
ticles acted as the critical catalyst during the synthesis as-
sisted by laser ablation. Therefore, a growth mechanism of Si
nanowires has been extrapolated from the VLS reaction.
8,9
However, a different model has been proposed which is sup-
ported by the fact that metal catalysts are not necessary dur-
ing synthesis by laser ablation.
12
SiO
2
was discovered to be
the special and effective catalyst which largely enhances Si
nanowire growth ~see Table I!. Moreover, structural investi-
gations show that high-density defects and silicon oxide
outer layers may play an important role for Si nanowire
growth.
12
In this paper, we report the synthesis of Si nano-
wires by laser ablation and thermal evaporation of highly
pure Si powder mixed with SiO
2
. Included are observations
of the microstructure and growth morphology of Si nanowire
nuclei by transmission electron microscopy ~TEM! and a dis-
cussion of the growth mechanisms.
The Si nanowires investigated in this work have been syn-
thesized by laser ablation
10,11
and thermal evaporation in an
evacuated quartz tube containing Ar gas ~500 Torr!. The
growth rate of Si nanowires was higher with the assistance of
laser ablation than with thermal evaporation alone. Si
nanowires synthesized by both methods had the same mor-
phology. The solid source was highly pure Si powder mixed
with approximately 70 wt % SiO
2
~all materials from Good-
fellow, purity 99.99%!. The temperatures around the source
and the area of the quartz tube where the nanowires grew
were about 1200 °C and 930 °C, respectively. After 12 h of
thermal evaporation, a Si nanowire product ~spongelike, dark
red in color! formed on the inside wall of the quartz tube. To
collect the Si nanowire nuclei, a Mo grid was placed in the
region of the quartz tube where the nanowires grew. Some Si
nanowires nucleated and grew on the grid. Structural inves-
tigation was carried out using a Philips CM200FEG trans-
mission electron microscope working under 200 kV.
As demonstrated from the data in Table I the presence of
SiO
2
in the powder targets significantly enhances Si nanow-
ire growth. The Si nanowire product obtained by using a
powder target composed of 50% SiO
2
and 50% Si is 30 times
greater than the amount generated by using a metal contain-
ing target. In addition, very few Si nanowires form if Si
single-crystal wafers or pure Si powder are used as targets.
Since SiO
2
is the dominant impurity in the two kinds of
targets mentioned, it is believed that in this case, the limited
amount of Si nanowire formation that does occur is also due
to the presence of SiO
2
. Pure SiO
2
powder can not produce
Si nanowires.
A typical morphology of the as-grown Si nanowires is
shown in the TEM image in Fig. 1, where two major forms
can be seen. Si nanowires with uniform diameters and
smooth surfaces are one major component, while the other
kind, Si nanoparticles coexist with the nanowires. Most Si
nanowires are extremely long ~.10
m
m! and randomly ori-
ented. It is interesting to note that most Si nanoparticles ap-
pear in the form of chain ~as marked by the arrow!. Short
bars of amorphous silicon oxide connect these nanoparticles.
Figure 2~a! shows the initial nucleation stage of the Si
nanowires on the Mo grid. Si nanoparticles formed first, as
TABLE I. Yields of Si nanowires for different experimental
conditions.
Target
~wt. %!
Furnace
temperature ~°C!
Yield
~mg!
Si1 1–3% Fe 1200 ;0.1
Si water 1200 Very few ,0.01
Si powder 1200 Very few ,0.01
Si1 5% SiO
2
1200 ;0.3
Si1 10% SiO
2
1200 ;0.5
Si1 30% SiO
2
1200 ;2.5
Si1 50% SiO
2
1200 ;3.0
Si1 70% SiO
2
1200 ;2.5
Si1 90% SiO
2
1200 ;0.5
SiO
2
800–1200 0
RAPID COMMUNICATIONS
PHYSICAL REVIEW B 15 DECEMBER 1998-IIVOLUME 58, NUMBER 24
PRB 58
0163-1829/98/58~24!/16024~3!/$15.00 R16 024 ©1998 The American Physical Society
identified by electron diffraction and most piled up on the
substrate. However, some particles ~as marked by the arrow!
stood alone from the others and grew fast since their prefer-
able growth directions were normal to the surface of the
substrate @see also Fig. 2~b!#. These preferable particles were
indeed the nuclei of the nanowires. As revealed in Fig. 2~b!,
no obvious metal catalysts or impurities formed on the tip of
the nucleus. The nucleus simply consisted of a Si crystalline
core with a high density of defects and an amorphous outer
layer. Since only silicon and oxygen were detected by elec-
tron energy dispersive spectrometer ~EDS! in situ the micro-
scope, the amorphous outer layer should be silicon oxide.
Because tip areas determine the one-dimensional growth
phenomenon of Si nanowires, high-resolution transmission
electron microscopy ~HRTEM! was employed to investigate
the microstructure of the tips. A typical Si nanowire tip is
shown in the HRTEM image in Fig. 3~a!. The tip is generally
round and covered by a relatively thick amorphous layer.
Because of this feature, under low magnification, a tip
looked thicker than the nanowire itself. The contrast of the
amorphous layer was quite uniform and only Si crystalline
structure was observed within the tip. This is contrary to the
nanowires grown by the VLS where metal drops were al-
ways formed on the tips. Similar to the microstructure of the
nuclei, the Si crystal core in the nanowire tip contained a
high density of stacking faults and micro-twins, which were
along the axis of the nanowire in the
@
1
¯
12
#
direction. These
kinds of defects have been observed in most Si nanowires in
our previous work.
10
We believe that the amorphous silicon oxide was the key
factor which significantly enhanced the nucleation and one-
dimensional growth of the Si nanowires. The growth mecha-
nism was different from that of the classic VLS growth of Si
whiskers where Si vapor diffused into metal drops ~e.g., Au!
and accumulated causing one-dimensional growth of Si. In
the present case, however, little Si vapor was generated un-
der 1200 °C. This was evidenced by the fact that the material
deposited on the water-cooled Cu finger surface was Si
x
O
y
~x5 0.51, y50.49! as determined by EDS. Since the vapor
phase was quenched on the cooled finger, the chemical com-
position of the material did not change. This indicated that
the vapor phase generated by thermal evaporation of the
solid source (Si1SiO
2
) mainly consisted of SiO. The forma-
tion of SiO was due to the reaction of Si and SiO
2
since SiO
is known to be an amorphous semiconductor of high resis-
tivity that is easily generated from powder mixtures ~espe-
cially in equimolar amounts of mixtures! of Si and SiO
2
by
heating.
13,14
The nucleation of nanoparticles may involve dif-
ferent decompositions of the Si oxide vapor phase at a rela-
tively low temperature of 930 °C as shown below:
Si
x
O→Si
x2 1
1 SiO
~
x. 1
!
,
2SiO→Si1SiO
2
.
These decompositions resulted in the precipitation of silicon
nanoparticles ~the nuclei of Si nanowires! surrounded by
shells of silicon oxide as observed in Fig. 2~a!. Such precipi-
tation of Si nanoparticles from SiO by annealing has been
quite well known since the 1950s.
15
FIG. 1. TEM image showing two major morphologies of Si
nanowires.
FIG. 2. ~a! Si nanowire nuclei formed on the Mo grid and ~b!
initial growth stage of Si nanowire.
RAPID COMMUNICATIONS
PRB 58 R16 025NUCLEATION AND GROWTH OF Si NANOWIRES FROM...
The growth mechanism of Si nanowires is suggested to be
determined by the following factors: ~1! the catalytic effect
of the relatively thick Si
x
O layer on nanowire tips; ~2! the
SiO
2
~much more stable than SiO! component in the shell,
which is formed from the SiO decomposition and retards the
lateral growth of nanowire; ~3! the main defects in the Si
nanowires, which are stacking faults ~along the nanowire
growth direction of ^112&! and microtwins. The presence of
these kinds of defects at the tip areas should result in the fast
growth of Si nanowires; ~4! the $111% surface, which has the
lowest surface energy among the Si surfaces, plays an im-
portant role during nanowire growth. Since surface energy is
more important when the crystal size is reduced to nanom-
eter scale, the presence of the $111% surfaces parallel to the
axes of the nanowires reduces the system energy. These im-
portant factors determine the growth direction of Si nanow-
ires to be ^112&.
According to the mechanism discussed above, nucleation
and growth occurred at all times during evaporation since the
SiO vapor phase was continually generated. Nuclei formed
on nanowire tips with different crystalline orientations dur-
ing growth. These unfavorable nuclei could not grow fast
along the nanowire and would cause a change of growth
direction or renucleation. Such renucleations resulted in the
formation of nanoparticle chains ~see Fig. 1! which coexisted
with the nanowires. A HRTEM image taken from a Si nano-
particle chain supports this growth mechanism. As shown in
Fig. 3~b!, the Si particles in the chain have different orienta-
tions and most are not aligned with their ^112& directions
parallel to the growth direction.
In summary, TEM study of Si nanowire nucleation and
growth indicated that thermal evaporation of Si powder
mixed with SiO
2
generated a Si oxide vapor phase. At the
initial nucleation stage, Si oxide vapor condensed on the sub-
strate and formed Si nanoprecipitates. Si nanowire nuclei
consisted of a polycrystalline Si core with a high density of
defects. The Si oxide vapor phase and defect structure acted
as key factors which greatly enhanced the nucleation and
one-dimensional growth of Si nanowires.
This work was financially supported in part by the Re-
search Grants Council of Hong Kong and the Strategic Re-
search Grants of the City University of Hong Kong.
*
Author to whom correspondence should be addressed. Electronic
address: APANNALE@cityu.edu.hk
1
R. S. Wagner and W. C. Ellis, Appl. Phys. Lett. 4,89~1964!.
2
E. I. Givargizov, J. Cryst. Growth 32,20~1975!.
3
H. I. Liu, N. I. Maluf, and R. F. W. Pease, J. Vac. Sci. Technol.
B 10, 2846 ~1992!.
4
H. Namatsu, S. Horiguchi, M. Nagase, and K. Kurihara, J. Vac.
Sci. Technol. B 15, 1688 ~1997!.
5
Y. Wada, T. Kure, T. Yoshimura, Y. Sudou, T. Kobayashi, Y.
Gotou, and S. Kondo, J. Vac. Sci. Technol. B 12,48~1994!.
6
T. Ono, H. Saitoh, and M. Esashi, Appl. Phys. Lett. 70, 1852
~1997!.
7
R. Hasunuma, T. Komeda, H. Mukaida, and H. Tokumoto, J.
Vac. Sci. Technol. B 15, 1437 ~1997!.
8
A. M. Morales and C. M. Lieber ~unpublished!.
9
A. M. Morales and C. M. Lieber, Science 279, 208 ~1998!.
10
N. Wang, Y. H. Tang, Y. F. Zhang, D. P. Yu, C. S. Lee, I. Bello,
and S. T. Lee, Chem. Phys. Lett. 283, 368 ~1998!.
11
Y. F. Zhang, Y. H. Zhang, N. Wang, D. P. Yu, C. S. Lee, I. Bello,
and S. T. Lee, Appl. Phys. Lett. 72, 1835 ~1998!.
12
N. Wang, Y. H. Tang, Y. F. Zhang, C. S. Lee, and S. T. Lee
~unpublished!.
13
S. W. Roberts, G. J. Parker, and M. Hempstead, Opt. Mater. 6,99
~1996!.
14
U. Setiowati and S. Kimura, J. Am. Ceram. Soc. 80, 757 ~1997!.
15
G. Hass and C. D. Salzberg, J. Opt. Soc. Am. 44, 181 ~1954!.
FIG. 3. HRTEM images of ~a! Si nanowire tip and ~b! Si nano-
particle chain.
RAPID COMMUNICATIONS
R16 026 PRB 58WANG, TANG, ZHANG, LEE, AND LEE
