The growth of carbon nanotubes at predefined locations using whole nickel nanowires as templates

Article abstract of DOI:10.1016/j.cplett.2004.06.075

In this process carbon nanotubes were prepared by the pyrolysis of a gaseous hydrocarbon in contacting with Ni nanowires, which were typically a few microns long, electrochemically produced from an anodic aluminium oxide template. Under controlled conditions, carbon nanotubes were formed at the location of the Ni-nanowire template. They inherited the shapes of the Ni nanowires, with their inner diameter being same as the outer diameter of the Ni nanowires. A novel growth mechanism which involves the diffusion of Ni atoms and the reshape of Ni nanowires is discussed.

Full text of DOI:10.1016/j.cplett.2004.06.075

Chemical Physics Letters 393 (2004) 511–516
www.elsevier.com/locate/cplett
The growth of carbon nanotubes at predefined locations
using whole nickel nanowires as templates
*
Han Gao, Jianyi Lin , Hui Pan, Guotao Wu, Yuanping Feng
Department of Physics, National University of Singapore, 10 Kent Ridge Cresent, Singapore 119260, Singapore
Received 16 December 2003; in final form 24 May 2004
Abstract
In this process carbon nanotubes were prepared by the pyrolysis of a gaseous hydrocarbon in contacting with Ni nanowires,
which were typically a few microns long, electrochemically produced from an anodic aluminium oxide template. Under controlled
conditions, carbon nanotubes were formed at the location of the Ni-nanowire template. They inherited the shapes of the Ni nano-
wires, with their inner diameter being same as the outer diameter of the Ni nanowires. A novel growth mechanism which involves the
diffusion of Ni atoms and the reshape of Ni nanowires is discussed.
Ó 2004 Elsevier B.V. All rights reserved.
1. Introduction
bution of CNTs can be controlled to some extent [5]. In
particular by using porous anodic aluminium oxide
(AAO) as a template, highly ordered, size-controllable
CNTs array can be grown from the pyrolysis of a gase-
ous carbon-containing material [9–13]. Unfortunately,
this kind of CNTs is normally ill-graphitized, consisting
of numerous stacked carbon flakes [14] since the AAO
itself can catalyze the pyrolysis of the carbon-containing
gas into amorphous carbon film on the walls, blocking
the formation of well-graphitized CNTs.
In this Letter, we report a novel process to grow
CNTs, using whole Ni nanowires as templates. The Ni
nanowires were prepared by AAO-templated electro-
chemical deposition. CNTs were then synthesized from
the decomposition of C2H2 on the Ni nanowires. The
outer diameter and length of CNTs were controllable,
correlated with the AAO channels. More exactly, the
CNTs were grown at the initial location of Ni nanowire
templates. Their size and shape were tallied with the Ni
nanowire. Moreover the grown CNTs were shown to be
well graphitized. This novel method therefore exhibits a
potential for the structure-predefined, size-controllable
growth of CNTs. For instance it is possible to prepare
a Ni nanostructure of designed patterns, using a routing
Since the first discovery in 1991 [1], carbon nanotubes
(CNTs) have stimulated tremendous development of na-
nomaterials and nanotechnology. Large amount of re-
searches have demonstrated the potential application
of CNTs in many fields including nanoscaled electronic
and optoelectronic devices, sensors, composite materi-
als, and energy storage [2]. But from the viewpoint of
practical applications there remain many issues to be ad-
dressed, for instance, bulk synthesis of CNTs with pre-
scribed size and predefined structure, manipulating
CNT to the desired locations, and assembling them to
form special structure among others.
Among four main approaches: arc-discharge [3], laser
ablation [4], chemical vapour deposition (CVD) [5–7]
and high pressure carbon monoxide process (HiPco)
[8], CVD is a popular method for mass and site-control-
lable production of CNTs. By CVD the diameter distri-
*
Corresponding author. Fax: +65-6777-6126.
E-mail address: phylinjy@nus.edu.sg (J. Lin).
0009-2614/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2004.06.075

512
H. Gao et al. / Chemical Physics Letters 393 (2004) 511–516
lithography technology. Alternatively, with the assist-
ance of magnetic-field [15] or electric-field techniques
[16], it is possible to manipulate an individual Ni nano-
wire into a specific location. The subsequent CVD
growth then enables to fabricate CNTs at the predefined
location and with predefined shape.
sample is prepared by dropping the solution on a copper
grid.
The silicon wafer with Ni nanowires was placed in a
horizontal quartz tube reactor for growing CNTs. The
sample was heated by a rate of 10 °C minÀ1 in a 5%
H2/Ar flow of 100 mLminÀ1. After the reactor tempera-
ture reached 700 °C, an ethylene flow of 5 mLminÀ1 was
introduced for 10 min, and meanwhile the flow rate of
the carrier gas was also increased to 300 mLminÀ1. Sub-
sequently, the reactor was cooled down to room temper-
ature in the 5% H2/Ar gas. The as-grown CNTs were
then burned in the air at 500 °C to remove amorphous
carbon, and immersed in a HCl (v:v 1:1) solution for
2 h to dissolve Ni (or nickel oxide). The as-grown and
post-growth-treated CNTs on the silicon wafer were di-
rectly observed under SEM. A Renishaw Raman Spec-
trometer 1000 with an excitation wavelength at 514
nm was employed to study the prepared CNTs. For
TEM and HRTEM (Philip CM300) observation, the
CNT/silicon wafer was sonicated in ethanol for a few
minutes to make the CNTs dispersed in the solution
and then dropped it on a copper grid.
As discussed later in this Letter the growth mecha-
nism of CNTs during the above process may be different
from the generally accepted vapour–liquid–solid (VLS)
process [6,7,17–20]. In VLS, growth occurs via the de-
composition of carbon-containing molecules on the sur-
face of metal catalyst nanoparticles, the bulk diffusion
and dissolution of carbon species in metal particles,
and the precipitation of saturated dissolved carbons
from the metal particle surface. This growth mechanism
was recently challenged by a time-resolved, in situ high-
resolution TEM study of carbon nanofibre growth on
supported Ni nanoparticles [21]. In the latter growth
scenario Ni nanoclusters tended to transfer their initial
equilibrium shape into a highly elongated shape. Ni step
edges at the surface played a key role in the nucleation
and growth of graphene sheet. Carbon atoms generated
at the step edge sites diffused to and finally bonded to
the graphene, which were developed through a reac-
tion-induced elongation/contraction reshaping of the
Ni nanocrystals. Our novel synthesis process appears
to follow the new growth mechanism, since the reshape
of Ni nanowires, i.e., the transfer from the elongated
wire to round particles, were evidently observed by
our SEM image.
3. Results and discussion
Fig. 1a is a SEM image of Ni nanowires dispersed on a
silicon wafer, and Fig. 1b is a TEM image of an individual
Ni nanowire. All the Ni nanowires in our sample are very
straight, having uniform sizes with the average length and
diameter around 20 lm and 40 nm, respectively. Never-
theless by varying the preparation conditions for the
AAO template the diameter and length of Ni nanowires
can be easily controlled [9–13]. The polycrystalline nature
of the Ni nanowires was confirmed by selected-area elec-
tron diffraction (SAD). Fig. 1c is a SEM image of as-pre-
pared CNTs on the silicon wafer. It is interesting to note
that after the CVD process CNTs were grown at the place
of the Ni nanowires. There were no Ni nanowires to sur-
vive, but attached many round-shaped nanoparticles,
which are graphite-encapsulated nickel (see Fig. 4b). Af-
ter the post-growth acid treatment, the round-shaped na-
noparticles were dissolved in HCl solution, and the
hollow core of the CNTs can be clearly observed in Fig.
1d. The grown CNTs inherit the shape of Ni nanowires.
Unlike the conventional CNTs, they look very straight
and uniform. As measured by SEM and shown in Fig.
1e most (>80%) of the CNTs have outer diameter in the
range of 55–60 nm. Considering that the mean wall thick-
ness of the CNTs is about 10 nm (see Fig. 2b,c), the inner
diameter of the CNTs ranges mostly between 35 and 40
nm, basically consistent with the outer diameter of the
Ni nanowire templates. (Note that the inner diameter of
the CNT in Fig. 2b is almost of the same size as the outer
diameter of the Ni nanowire in Fig. 1b.) To further dem-
onstrate the diameter controllability of our CNT growth,
2. Experimental
Ni nanowires, here serving as templates for preparing
CNTs, were made by the electrochemical deposition of
nickel into nanochannels of the AAO template [22].
Both-side-opened AAO were prepared following the
previously reported method [9,10]. In order to improve
the quality and uniformity of Ni nanowires, we first
coated one side of the AAO template with a Pt film of
less than 10 nm thickness by ion sputtering (JFC-1600
Coater). Then the coated side was subject to electroplat-
ing in a solution containing 300 gLÀ1 NiSO4 Æ6H2O,
50 gLÀ1 HBO3 at 60 °C for 5–10 min till a thick nickel
film was formed. This nickel film served as a working
electrode to further electrodeposit Ni into the pores of
AAO at the same solution and conditions as in the first
step. The Ni-embedded AAO sample was then treated in
a 3 molLÀ1 NaOH solution to fully remove the AAO
template, followed by repeat rinse with distilled water
till the water pH 7. They were then dispersed on a silicon
wafer for the use in growing CNTs and the observation
of their morphology under a scanning electron micro-
scope (SEM, JSM-6700F). TEM (Philips-CM100)

H. Gao et al. / Chemical Physics Letters 393 (2004) 511–516
513
Fig. 1. (a) SEM image of Ni nanowires liberated from AAO templates and dispersed on the silicon wafer for growing CNTs. (b) TEM image of a Ni
nanowire with a diameter of about 40 nm. (c) SEM image of the as-prepared CNTs on the silicon wafer. (d) SEM image of the post-grown treated
CNTs on the silicon wafer. (e) Outer diameter distribution of CNTS.
we have been able to prepare the CNTs with different in-
ner diameters in the range of 20–60 nm by varying the di-
ameter of AAO nanopores (and hence that of Ni
nanowires). Adjusting the concentration of Ni nano-
wire/ethanol solution, we can disperse only a few pieces
of Ni nanowires on a silicon wafer, and hence grow only
a few pieces of CNTs on the silicon wafer (see Fig. 3a).
Though the CNTs still keep the location of the Ni-nano-
wires template, the tube is broken at several places along
the tubes, with small round particles attaching at the bro-
kenness. In the case that the Ni nanowires are only par-
tially liberated from the AAO template, they would be
standing in vertically alignment on the AAO template
base. Fig. 3b is a SEM image of the as-prepared aligned
CNTs arrays from standing Ni nanowires array. This re-
sult implies that we can produce individual CNTs in the
designed locations where individual Ni nanowires are
placed prior to the CVD process.
strong peak at 1580 cmÀ1 is attributed to the crystalline
graphite structure while the D lines being regarding as
disorder-induced features due to the finite particle size
effect or lattice distortion [23–25]. The TEM images in
Figs. 2b,c reveals the tubular structure of the as-pre-
pared carbon nanotubes, with a hollow core of ꢀ40
nm in diameter and a multiple-layered wall of ꢀ10 nm
in thickness. Only one side of the tube wall was shown
in Fig. 2c, as the diameter of the CNT is large. The wall
of the CNT consists of many crystalline domains with
average domain size of 30 nm. It seems that our CNTs
have more defects than conventional CNTs but less de-
fects than those CNTs directly grown by the pyrolysis of
C2H4 on AAO template [9,10,14]. The graphitized struc-
ture can also be demonstrated by a typical SAD pattern
(inset in Fig. 2b), with an interlayer separation of about
˚
˚
3.4 A, similar to that of graphite (d002 =3.35 A).
The growth of CNTs using whole Ni nanowires as
templates appears to follow different mechanism from
the typical VLS process. We have carried out the follow-
ing 5 tests to probe the growing mechanism. In the first
test, Ni nanowires/silicon wafer was heated to 700 °C in
Fig. 2 displaces the results of the structural character-
ization of the CNTs. The Raman spectrum in Fig. 2a
shows two sharp peaks at 1580 (G line) and 1345 (D
line) cmÀ1 with a relative intensity of 1.44 (IG/ID). The

514
H. Gao et al. / Chemical Physics Letters 393 (2004) 511–516
Fig. 3. SEM imges of (a) sparse as-prepared CNTs lying on silicon
wafer and (b) aligned standing CNTs array fabricated from aligned Ni
nanowires array.
and would later break out from defects of the CNT
shell, as shown in Fig. 3a. In the third test, longer reac-
tion time (20 min for example) was applied, the break-
ing-out Ni particles then became the centers for
subsequent CNT growth. The second-generation CNTs
have smaller tube diameter (commonly 20 nm) as com-
pared to that of the CNTs grown from Ni-nanowires
template (40 nm in this case). The catalytic decomposi-
tion of ethylene on Ni nanowires was found to start at
as low temperature as 400 °C in our experiments. In
the fourth test, the CVD was carried out on Ni nano-
wires/silicon wafer at 500 °C for 20 min. The SEM im-
age in Fig. 4c show that the Ni nanowires have
transferred into Ni particles and squeezed out from
the carbon nanotubes. The carbon product in this figure
actually is an amorphous carbon with low mechanical
strength, which was easily broken into pieces when sub-
jected to a sonicate treatment for preparing TEM sam-
ple. We can also find in the figure many small-sized
CNTs (or carbon filament) grown from the squeezing-
Fig. 2. (a) Raman spectrum of the post-grown treated CNTs. (b) TEM
image of CNT with an inner diameter of 40 nm, inset is a typical
electron diffraction pattern of as-prepared CNT. (c) HRTEM image of
one side of CNT walls with a mean thickness of 10 nm.
a 5% H2/Ar flow of 100 mLminÀ1 without introducing
any hydrocarbon reactant, and we found that Ni nano-
wires were stable at 700 °C, without any round Ni par-
ticles appearing. In the second test, after the reactor
temperature reached 700 °C, an ethylene flow of
5 mLminÀ1 was introduced for 2 min, which was shorter
than the 10 min for producing the sample shown in Fig.
2. The SEM image of the product in the second test (see
Fig. 4a) show that some Ni nanowires are wrapped by
CNTs while some have been reshaped and aggregated
into Ni particles, leaving a hollow core under the CNT
shell (see the arrows in Fig. 4a). If the supply of ethylene
was long (10 min), the Ni particles continuously grew

H. Gao et al. / Chemical Physics Letters 393 (2004) 511–516
515
Fig. 4. (a) SEM image of CNTs prepared by a very short time of ethylene pyrolysis. (b) TEM image of graphite-encapsulated Ni particles, inset is a
HRTEM image of graphene layers; SEM images of (c) the sample prepared at 500 °C. (d) The CNTs prepared with high volume ratio of ethylene vs.
carrier gas.
out Ni nanoparticles due to long reaction time. In the
fifth test, after the reactor temperature reached 700 °C,
an ethylene flow of 15 mLminÀ1 (vs 5 mLminÀ1 in the
above tests) was introduced for 5 min with a 5% H2/
Ar flow of 100 mLminÀ1 to grow CNTs on silicon
wafer. The SEM image in Fig. 4d shows no Ni-nano-
wire-templated CNTs (which are straight and 40 nm in
diameter). However, curly and randomly oriented CNTs
with diameter far less than 40 nm are found along the
location of Ni nanowires. These CNTs were apparently
grown according to conventional VLS mechanism on
Fig. 5. Schematic illustration of growth mechanism for the CNTs
prepared by using whole Ni nanowires as templates.
the metal nanoparticles instead on the whole nanowires.
On the basis of the above five tests, we proposed a
growth mechanism for the CNTs prepared by using Ni
nanowires as templates. In this scenario ethylene is first
pyrolyzed on the surface of Ni nanowires. The precipita-
tion of carbon starts with the development of oversatu-
rated Ni–C alloys on the surface of the Ni nanowires
[17,18]. The growth of graphene layers takes place first
on the surface of Ni nanowires while the bulk of Ni na-
nowires remains intact (see the first step in Fig. 5). The
grown CNT covers the Ni nanowire forming CNT shells
which inherit the shape of Ni nanowires, and their inner
diameter is determined by the outer diameter of Ni na-
nowires. This growth process is driven by the energy
gain when binding the grapheme sheet to the Ni surface.
The growth ceases when the graphene layers eventually
encapsulate the Ni nanowire. In the case that the gra-
phene sheets are not completely closured while Ni nano-
wires are kept in the contact with ethylene at 700 °C for
longer time (such as those in Test 2), the continuous in-
crease of dissolved carbon may result in the develop-
ment of bulk Ni–C alloys. Density of functional
theory (DFT) predicts that the carbon–Ni bonding ener-
gy is larger than the energy cost for Ni–carbon interface
step formation [21]. Hence the diffusion of Ni atoms
along interface towards the free surface is energetically
favourable, and may finally cause the reshape of Ni into

516
H. Gao et al. / Chemical Physics Letters 393 (2004) 511–516
particles appearing at the brokenness along the carbon
tube (see the second step in Fig. 5), and leaving behind
the CNTs with a hollow core of the same diameter. This
process is very similar to that in [21], in which Ni parti-
cles were elongated to align graphene layers into a tubu-
lar form, and then contract back towards particles when
the increasing Ni surface energy during the elongation
cannot be compensated by the energy gained from bid-
ing the graphitic fibre to the nickel surface. In our at-
tempts, the low ratio of ethylene vs. carrier gas at the
beginning of the growth was necessary to ensure the gra-
phene layers occurring only on the surface of Ni nano-
wires. Alternatively, with a high ratio of ethylene vs.
carrier gas at the beginning, the speeds of the ethylene
decomposition and the liquid Ni–C alloy formation
could be so fast that Ni nanowires reshaped to Ni parti-
cles immediately along the their original locations with-
out forming nanotube shell, and curly, small-sized
CNTs will be grown from these particles, as seen in
the fifth test. If the growth temperature is relatively
low (such as 500 °C in the fourth test), carbon atoms
are either difficult to be dissolved into Ni nanowires or
do not diffuse fast inside the Ni nanowires, which results
in an amorphous carbon coating [26].
References
[1] S. Iijima, Nature 354 (1991) 58.
[2] R.H. Baughman, A.A. Zakhidov, W.A. de Heer, Science 297
(2002) 787.
[3] T.W. Ebbesen, P.M. Ajayan, Nature 258 (1992) 16.
[4] A. Thess, R. Lee, P. Nikdaev, H.J. Dai, P. Petit, J. Robert, C.H.
Xu, Y.H. Lee, S.G. Kim, A.G. Rinzler, D.T. Colbert, G.E.
Scuseria, D. Tomanek, J.E. Fischer, R.E. Smalley, Science 273
(1996) 483.
[5] P. Nikolaev, M.J. Bronikowski, R.K. Bradley, F. Rohmund,
D.T. Colbert, K.A. Smith, R.E. Smalley, Chem. Phys. Lett. 313
(1999) 91.
[6] H.J. Dai, Acc. Chem. Res. 35 (2002) 1035.
[7] R. Andrews, D. Jacques, D.L. Qian, T. Rantell, Acc. Chem. Res.
35 (2002) 1008.
[8] M.J. Bronikowski, P.A. Willis, D.T. Colbert, K.A. Smith, R.E.
Smalley, J. Vac. Sci. Technol. A 19 (2001) 1800.
[9] H. Gao, C. Mu, F. Wang, D.S. Xu, K. Wu, Y.C. Xie, S. Liu,
E.G. Wang, J. Xu, D.P. Yu, J. Appl. Phys. 93 (2003) 5602.
[10] H. Gao, X.B. Wu, J.T. Li, G.T. Wu, J.Y. Lin, K. Wu, D.S. Xu,
Appl. Phys. Lett. 83 (2003) 3391.
[11] G. Che, B.B. Lakshmi, R.E. Fisher, C.R. Martin, Nature 393
(1998) 346.
[12] T. Kyotani, L. Tsai, A. Tomita, Chem. Mater. 8 (1996) 2109.
[13] J. Li, C. Papadopoulos, J.M. Xu, Appl. Phys. Lett. 75 (1999) 367.
`
`
´
[14] Y.C. Sui, D.R. Acosta, J.A. Gonzalez-Leon, A. Bermudez, J.
Feuchtwanger, B.Z. Cui, J.O. Flores, J.M. Saniger, J. Phys.
Chem. B 105 (2001) 1523.
[15] M. Tanase, L.A. Bauer, A. Hultgren, D.M. Silevitch, L. Sun,
D.H. Reich, P.C. Searson, G.J. Meyer, Nano Lett. 1 (2001) 155.
[16] P.A. Smith, C.D. Nordquist, T.N. Jackson, T.S. Mayer, B.R.
Martin, J. Mbindyo, T.E. Mallouk, Appl. Phys. Lett. 77 (2000)
1399.
4. Conclusions
In summary, we have demonstrated a novel process
to prepare CNTs by using whole Ni nanowires as a tem-
plate. This process shows the potential of controllable
growth of CNTs at predefined locations and with prede-
fined size. High temperature and low volume ratio of
ethylene vs. carrier gas are the appropriate conditions
for growing higher quality CNTs. The growth of carbon
nanotubes may follow a new mechanism which involves
the diffusion of nickel atoms and the reshape of nickel.
[17] R.T.K. Baker, Carbon 27 (1989) 315.
[18] Y.H. Mo, A.K.M.F. Kibria, K.S. Nahm, Synth. Met. 122 (2001)
443.
[19] Y.N. Xia, P.D. Yang, Y.G. Sun, Y.Y. Wu, B. Mayers, B. Gates,
Y.D. Yin, F. Kim, H.Q. Yan, Adv. Mater. 15 (2003) 353.
[20] J.S. Lee, G.H. Gu, H. Kim, J.S. Suh, I. Han, N.S. Lee, J.M.
Kim, G.S. Park, Synth. Met. 124 (2001) 307.
´
[21] S. Helveg, C. Lopez-Cartes, J. Sehested, P.L. Hansen, B.S.
Clausen, J.R. Rostrup-Nielsen, F. Abild-pedersen, J.K. Nøkskov,
Nature 427 (2004) 426.
[22] D. Routkevitch, A.A. Tager, J. Haruyama, D. Almawlawi, M.
Moskovits, J.M. Xu, IEEE Trans. Electron. Dev. 43 (1996) 1646.
[23] W.Z. Li, H. Zhang, C.Y. Wang, Y. Zhang, L.W. Xu, K. Zhu,
S.S. Xie, Appl. Phys. Lett. 70 (1997) 2684.
Acknowledgements
The work is financially supported by Agency for
Science, Technology and Research of Singapore. The as-
sistance of Dr. Wang SJ and Ms. Chow XY of Institute
of Materials Research and Engineering in the HRTEM
study is highly appreciated.
[24] D.G. McCulloch, S. Prawer, A. Hoffman, Phys. Rev. B 50 (1994)
5905.
[25] V. Barbarossa, F. Galluzzi, R. Tomaciello, A. Zanobi, Chem.
Phys. Lett. 185 (1991) 53.
[26] Y.M. Shyu, F.C.N. Hong, Diam. Relat. Mater. 10 (2001) 1241.