Controlled growth of silicon nanowires synthesized via solid–liquid–solid mechanism

Abstract

The growth of silicon nanowires using solid–liquid–solid method is described. In this method, silicon substrates coated with a thin layer of gold were heat treated in nitrogen ambient. Gold particles started to diffuse into the silicon substrate and Au–Si alloy formed at the interface. The alloy would have molten to form liquid droplets on the substrate when temperature increases above their eutectic point, and more Si atoms diffused into these alloy droplets when heating continues. Rapid cooling of the droplet surface due to nitrogen flow into the chamber would eventually lead to the phase separation of silicon atoms from the surface of the alloy, created the nucleation and thus the growth of silicon nanowires. Controlled growth of the nanowire could be achieved by annealing the sample at 1000°C with nitrogen flow rate set to around 1.5l/min. The synthesized nanowires with diameter varied from 30 to 70nm, were straight and grew along the N2 flow. Larger amount and longer nanowires were grown when longer period of heating was applied. Nanowires with lengths more than several hundreds of micrometers were achieved by annealing the sample for 4h.

Full text

Controlled growth of silicon nanowires synthesized via solid–liquid–solid mechanism
This content has been downloaded from IOPscience. Please scroll down to see the full text.
Download details:
IP Address: 84.22.32.222
This content was downloaded on 28/09/2013 at 19:13
Please note that terms and conditions apply.
2005 Sci. Technol. Adv. Mater. 6 330
(http://iopscience.iop.org/1468-6996/6/3-4/A22)
View the table of contents for this issue, or go to the journal homepage for more
Home Search Collections Journals About Contact us My IOPscience

Controlled growth of silicon nanowires synthesized via
solid–liquid–solid mechanism
Y.Y. Wong
*
, M. Yahaya, M. Mat Salleh, B. Yeop Majlis
Institute of Microengineering and Nanoelectronics (IMEN), 43600 UKM, Bangi, Selangor, Malaysia
Received 12 January 2005; revised 16 February 2005; accepted 16 February 2005
Available online 20 June 2005
Abstract
The growth of silicon nanowires using solid–liquid–solid method is described. In this method, silicon substrates coated with a thin layer of
gold were heat treated in nitrogen ambient. Gold particles started to diffuse into the silicon substrate and Au–Si alloy formed at the interface.
The alloy would have molten to form liquid droplets on the substrate when temperature increases above their eutectic point, and more Si
atoms diffused into these alloy droplets when heating continues. Rapid cooling of the droplet surface due to nitrogen flow into the chamber
would eventually lead to the phase separation of silicon atoms from the surface of the alloy, created the nucleation and thus the growth of
silicon nanowires. Controlled growth of the nanowire could be achieved by annealing the sample at 1000 8C with nitrogen flow rate set to
around 1.5 l/min. The synthesized nanowires with diameter varied from 30 to 70 nm, were straight and grew along the N
2
flow. Larger
amount and longer nanowires were grown when longer period of heating was applied. Nanowires with lengths more than several hundreds of
micrometers were achieved by annealing the sample for 4 h.
q 2005 Elsevier Ltd. All rights reserved.
Keywords: Silicon nanowire; Solid–liquid–solid; Gold; Thermal anneal; N
2
1. Introduction
Successful synthesis of carbon nanotubes (CNTs) by
Iijima in 1991 [1] and the growth of some wire-like
nanowhiskers [2,3] around that time have stimulated new
momentum in the study of nanoscience. Much interest has
been drawn towards these nanomaterials because of their
special characteristics in one-diamensionality [4]. Mean-
while, fabrication of these one-dimensional nanomaterials
can be achieved by using either ‘Top-down’ or ‘Bottom-up’
approaches [5]. Whilst top-down approach usually requires
precision engineering and lithography for patterning and
removing of bulk material, bottom-up approach on the other
hand involves the building of nanostructure by self-
arrangement of atoms or molecules through the natural
interaction amount themselves. However, the later process
receives more research interest as it provides a feasible and
inexpensive alternative for nanomaterials fabrication.
In this paper, the synthesis of silicon nanowires using the
approach of bottom-up fabrication is described. Silicon
nanowires have been grown in a furnace under the flow of
nitrogen gas. The growth mechanism can be explained by
solid–liquid–solid method. In general, silicon from the
substrate which was in solid phase would first be transferred
into liquid phase upon heating with the helped of gold as
catalyst. The silicon atoms were then self-assembled into
wire shape (solid rod) when rapid cooling occurred at the
liquid surface which was induced by the nitrogen gas flow.
The effect of the nitrogen flow rate, the heating temperature
and its duration onto the growth of silicon nanowires were
also discussed.
2. Experimental details
The p-type silicon substrates with (100) orientation were
cleaned with acetone and diluted hydrofluoric acid followed
by rinsing under running deionized water. The cleaned
samples were dried and transferred into a metal sputter
coater, where a thin layer (10 nm) of gold was deposited.
The samples were then ready to be annealed. A quartz tube
Science and Technology of Advanced Materials 6 (2005) 330–334
www.elsevier.com/locate/stam
1468-6996/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.stam.2005.02.011
*
Corresponding author. Tel.: C603 8925 9080; fax: C603 8925 9080.
E-mail address: yywong@vlsi.eng.ukm.my (Y.Y. Wong).
The STAM archive is now available from the IOP Publishing website http://www.iop.org/journals/STAM

4 cm in diameter and 60 cm in length was used to place the
samples and was inserted into a furnace. Purified nitrogen
gas (99.999%) was flowed into the quartz tube throughout
the annealing process. Typical synthetic conditions for the
nanowires were 2 l/m of N
2
flow rate and 1000 8C of heating
temperature for 2 h duration.
The as-growth nanowires were observed by the scanning
electron microscope (SEM, XL 30 Philips) as well as the
tunneling electron microscope (TEM, JEM2010 JEOL).
Meanwhile, chemical compounds presented in the nano-
wires were investigated using the energy dispersive X-ray
spectroscopy (EDS, Oxford Instrument, model 7353).
3. Results and discussions
3.1. Characterization of silicon nanowires
The grown of silicon nanowires can be indicated by
observing a thin layer in gray (or whitely) formed on the
silicon substrate which was initially covered with gold. At
the typical condition of growth, some fiber-like wires can be
seen (with SEM) to have covered the treated silicon
substrate (Fig. 1). Diameters of wires varied from 50 to
150 nm and their lengths were usually longer than several
tens of micrometers though they were hard to be determined
due to the highly compact and entangle growth.
To examine the chemical components of the nanowires,
the EDS scans (Fig. 2) had been conducted by focusing onto
the nanowires located at the edge of the silicon substrate. In
the experiment, growth directions of the nanowires were
rather irregular and some of them were grown extended out
from the substrate (and suspend in air). This had provided an
idea location to obtain related information from nanowires
under the EDS scans without interference signals from
silicon substrate underneath. As suggested by the exper-
imental setup, Si peak was originated from the body of the
wires, and not from the silicon substrate. O peak came from
the silicon oxide layer. Although nitrogen was flowed
throughout the experiment, small amounts of oxygen were
believed still presented in the chamber and caused the
formation of silicon oxide. Growth of a native oxide layer on
the silicon nanowres surface after the experiment was also
contributing to this signal. This argument is supported by
further testing using the TEM. Fig. 3 has clearly shown a thin
layer of oxide covering the surface of the silicon core wire.
The strong Al signal came from beneath the nanowires, i.e.
aluminum holder for the sample meanwhile the weak C peak
is caused by the used of carbon tape to attach the sample.
3.2. Growth mechanism of silicon nanowires
The growth of the silicon nanowires in the experiments
can be understood from the concept of eutectic. In our
works, silicon and gold were the main elements in concern.
By referring to the phase diagram for Au–Si eutectic alloy
Fig. 4, lowest melting temperature (363 8C) can be achieved
for the formation of alloy with about 20% of silicon atoms.
Fig. 1. SEM image of nanowires on silicon substrate.
Fig. 2. EDX spectrum of the nanowires taken from the rectangle box as
shown in inset. (Inset: Gray surface on the left is the silicon substrate
covered by nanowires, black area on the right is outside of the substrate).
Fig. 3. TEM image of the nanowires.
Y.Y. Wong et al. / Science and Technology of Advanced Materials 6 (2005) 330–334 331

When the gold-coated sample being heated in the furnace,
gold may diffuse into the silicon substrate to form alloy at
their interface. This layer started to melt when temperature
in the furnace increased above eutectic point and the process
continue until all the gold would have mixed with some
silicon and molten. Inter-atomic interaction and strong
surface tension caused the formation of round Au–Si liquid
droplets on the silicon substrate. Temperature dependent
mixing as indicated by the phase diagram suggested that
more silicon would diffuse in the alloy droplets when
temperature continues to increase. (It is therefore super
saturation of silicon in the droplets may have occurred when
temperature reached 1000 8C as in the experiment).
The nitrogen flow had played the critical role for silicon
nanowires growth. The cooler nitrogen molecules that
flowed into the quartz tube caused cooling at the surface of
the hot liquid droplets. Rapid cooling at these Au–Si
droplets surface tended to separate the elements in the
mixture. The nucleation of silicon may have occurred at the
droplet surface where more silicon atoms could be
precipitated from the alloy droplet to grow into nanowires.
The growth can therefore be regarded as solid–liquid–
solid (SLS) method as suggested by Yu and coworkers [6,7].
In this mechanism, the silicon source for nanowires growth
originated from the substrate, which was a solid. Forming
and melting of the alloy would transform the silicon into
liquid. These silicon atoms then self-assembled into
nanowire which was in solid again upon cooling. It has to
be noted that the nanowires grown using the SLS method are
differed from those grown using vapor–liquid–solid (VLS)
method [8–10] which is a more popular growth mechanism
for nanowires. In this later mechanism, source materials for
the nanowire introduced as vaporized form, before it is
transformed into liquid and finally into solid nanowire.
The main difference in nanowires produced is the presence
of gold cluster on the tip of nanowires grown by the later
process. No Au had been found on our nanowires as
indicated by EDS result. Au had played a role as catalyst for
the growth, not forming any part on the nanowire except at
the growth location and would stay on the silicon substrate.
In order to achieve better control on the silicon nanowires
synthesized using SLS method, further studies have also
been carried out to investigate the effect of annealing
parameters to the growth. The synthetic conditions as well
as their effects on nanowires growth were summarized in
Table 1.
0 20 40 60 80 100
Au Si
Atomic
p
ecenta
g
e of Si
C
o
1400
1200
1000
800
600
400
200
0
363
o
C
19.5%
Fig. 4. Phase diagram of Au–Si eutectic alloy. The eutectic alloy forms at
363 8C.
Table 1
Studies of the parameter effects on nanowire grown by SLS method
Annealing condition Nanowires growth
Temperature (8C) N
2
flow rate (l/min) Duration (min)
800 2.0 120 None, only small clusters formed.
900 2.0 120 Very few, lot of small clusters remain.
950 2.0 120 Few, but can easily be found.
1000 2.0 120 Dense, entangled growth.
1050 2.0 120 Very dense at certain area, but the diameters usually large (O200 nm).
Lot of big clusters with various shapes and sizes but no nanowire formed.
1100 2.0 120 None, only big clusters formed (w1 mm).
1000 0.6 30 None, only small clusters seen.
1000 0.8 30 Started to form, but not straight, much entangled.
1000 1.2 30 More directional than at 0.8 l/m N
2
, diameter 100–150 nm.
1000 1.5 30 Straight and follow the N
2
flow, diameter 30–70 nm.
1000 2.0 30 Quite straight and follow the N
2
flow, diameter: 50–100 nm.
1000 2.5 30 Started to become more entangle again.
1000 R3.0 30 Direction become irregular again, diameter: 40–70 nm.
1000 1.5 15–240 Nanowire can be formed by annealed the sample for15 min under N
2
ambient. The growth was less compact and their direction can be
controlled by the N
2
flow. The longer the heating duration, the longer
the nanowires can be grown (few hundreds mm for 4 h), but too compact
and entangled.
Y.Y. Wong et al. / Science and Technology of Advanced Materials 6 (2005) 330–334332

3.3. Temperature effect to the growth of nanowires
The heating of the eutectic alloy had supplied the source
for growth of the nanowires in SLS method. While a too low
temperature may not create saturation of silicon atoms in the
alloy to be nucleated and grown into nanowires upon
cooling; a too high temperature may consume excessive
energy and will not favor the cost for fabrication. In our
works, the growth of nanowires could only be achieved
when temperature increased above 900 8C. Below this
critical point, the percentage of silicon atoms present in the
alloy may not be sufficient to form the nucleation and
precipitation for the nanowires growth. Only small cluster
of the Au–Si alloy were found on the substrate (Fig. 5)in
these cases.
Significant growth of nanowires had been achieved by
setting the temperature to 1000 8C. The growth was
relatively dense as shown in Fig. 1. When the temperature
increased above this point (1050 and 1100 8C), the wires
formed were generally larger in diameter (w200 nm). This
may be due to too much of silicon in the alloy creating larger
nucleation seeds for the growth of the wires. It is also noted
that only small amounts of nanowires were able to form at
certain areas on the substrate. Most of the areas remain
covered with a thick layer of alloy clusters (Fig. 6). Almost
none of the fiber-like feature could be found on the sample
heat treated at 1100 8C. The reason why only relatively big
alloy clusters with irregular sharps and sizes formed on the
substrate as compared to those seen in Fig. 5 has yet to be
understood.
3.4. Nitrogen flow rate effect to the growth of nanowires
Nitrogen flowed into the chamber was not only to
mitigate the present of unwanted gases, but also to induce
rapid cooling at the surface of the hot droplets. Again, a too
low the flow rate may not be sufficient to cool down the
surface effectively, and a too high flow will increase the
production cost. In the experiments, there were no
nanowires formed for the flow rates below 0.8 l/m. This
may be because of too few N
2
molecules to bring away the
thermal energy from the droplet surface, thus no nucleation
occurred. At 0.8 l/m flow, the nanowires started to form but
entangled with each other, their orientations also irregular.
The growth of nanowires became more directional
(along the flow of N
2
) as the flow rate increased to 1.5 l/m
(Fig. 7). The nanowires were relatively straight at this flow
rate and their diameters were also the smallest (30–70 nm)
among others. The growth direction became irregular again as
the flow rate increased above 2 l/m. Supply of N
2
into the
quartz at this rate may have led to the formation of turbulence
flow which disturbed the directional growth of nanowires.
3.5. Heating duration effect to the growth of nanowires
The amount of silicon nanowires grown can be
controlled by the annealing duration. Generally, if heating
time had been kept short the growth of nanowires was less
compact and their direction could be controlled by N
2
flow.
Good example is shown in Fig. 7 in which the sample had
Fig. 7. SEM image of directional growth of nanowires along the N
2
flow.
Fig. 6. SEM image shown thick layer of Au–Si alloy clusters. No nanowire
formed.
Fig. 5. SEM image of the clusters of Au–Si alloy on the substrate. Their
sizes vary from 30 to 100 nm.
Y.Y. Wong et al. / Science and Technology of Advanced Materials 6 (2005) 330–334 333

been annealed for about 30 min. Annealing time of 15 min
had also been used, where much less compact of nanowires
growth had been achieved. Experiment had also been
carried for a duration of 4 h. The growth was highly
compact and entangled. However, the length of nanowires
grew more than several hundreds of micrometers in this
case.
4. Conclusion
The synthesis of nanowires using solid–liquid–solid
mechanism has been demonstrated in this paper. The
as-growth nanowires were composed of Si and its oxide at
the surface. SLS provides a relatively simple and
straightforward method to grow large amount of
nanowires.
Major factors that affect the growth including the
heating temperature, the N
2
flow rate and the duration of
annealing have been investigated. In our experiment,
relatively straight and directional growth of nanowires
along the gas flow could be achieved by choosing heating
temperature of 1000 8C and N
2
flow at 1.5 l/min. The
as-growth nanowires have diameters varied from 30 to
70 nm. It is also believed that a laminar flow of nitrogen
gas with sufficient number of molecules to cool the droplet
surface is essential to produce straight and directional
nanowires. Heating duration on the other hand was used to
control the amount of the nanowires formed. Less compact
growth of nanowires was a result of short annealing
duration, nanowires. In contrast, nanowires with lengths as
long as several hundreds of micrometers were grown by
heating the substrate for about 4 h.
Acknowledgement
The study is financially supported by Malaysian Ministry
of Education, through the fundamental research grant
RRR1-001-2004.
References
[1] S. Iijima, Nature 354 (1991) 56.
[2] M. Yazawa, M. Koguchi, K. Hirutn, Appl. Phys. Lett. 58 (1991) 1080.
[3] K. Haraguchi, K. Haraguchi, T. Katsuyama, K. Hiruma, K. Ogawa,
Appl. Phys. Lett. 60 (1992) 745.
[4] M.S. Dresselhaus, Y.M. Lin, O. Rabin, A.G. Souza Filho,
M.A. Pimenta, R. Saito, Ge.G. Samsonidze, A.G. Dresselhaus,
Mater. Sci. Eng. C 23 (2003) 129.
[5] M.J. Madou, Fundamentals of Microfabrication, The Science of
Miniaturization, 2nd ed., CRC Press, 2002.
[6] Y.J. Xing, D.P. Yu, Z.H. Xi, Z.Q. Xue, Appl Phys. A 76 (2003) 551.
[7] H.F. Yan, Y.J. Xing, Q.L. Hang, D.P. Yu, Y.P. Wang, J. Xu, Z.H. Xi,
S.Q. Feng, Chem. Phys. Lett. 323 (2000) 224.
[8] Y. Wu, P. Yang, J. Am. Chem. Soc. 123 (2001) 3165.
[9] C.Y. Lee, T.Y. Tseng, S.L. Li, P. Lin, Tamkang, J. Sci Tech. 6 (2003)
127.
[10] T. Ma
˚
rtensson, M. Borgstrom, W. Seifert, B.J. Ohlsson,
L. Samuelson, Nanotechnology 14 (2003) 1255.
Y.Y. Wong et al. / Science and Technology of Advanced Materials 6 (2005) 330–334334