Chemical vapor deposition of methane for single-walled carbon nanotubes

Article abstract of DOI:10.1016/S0009-2614(98)00745-3

We report the synthesis of high-quality single-walled carbon nanotubes (SWNT) by chemical vapor deposition (CVD) of methane at 1000°C on supported Fe₂O₃ catalysts. The type of catalyst support is found to control the formation of ind

Full text of DOI:10.1016/S0009-2614(98)00745-3

14 August 1998
Ž
.
Chemical Physics Letters 292 1998 567–574
Chemical vapor deposition of methane for single-walled carbon
nanotubes
a
a,b
a,)
Jing Kong , Alan M. Cassell , Hongjie Dai
a
Department of Chemistry, Stanford UniÕersity, Stanford, CA 94305, USA
b
NASA Ames Research Center, Moffett Field, CA 94035, USA
Received 1 May 1998; in final form 12 June 1998
Abstract
Ž
.
Ž
.
We report the synthesis of high-quality single-walled carbon nanotubes SWNT by chemical vapor deposition CVD of
methane at 10008C on supported Fe2O3 catalysts. The type of catalyst support is found to control the formation of individual
or bundled SWNTs. Catalysts supported on crystalline alumina nanoparticles produce abundant individual SWNTs and small
bundles. Catalysts supported by amorphous silica particles produce only SWNT bundles. Studies of the ends of SWNTs lead
to an understanding of their growth mechanism. Also, we present the results of methane CVD on supported NiO, CoO and
NiOrCoO catalysts. q 1998 Published by Elsevier Science B.V. All rights reserved.
1. Introduction
met by developing CVD methods that utilize
carbon-rich gases as carbon feedstock. CVD of vari-
ous hydrocarbon gases on transition-metal catalysts
has been successful in obtaining carbon fibers and
Significant progress was made recently in synthe-
sizing gram quantities of single-walled carbon nan-
Ž
. w x
w x
otubes SWNT 1 using laser-ablation 2 and arc-
w
x
multi-walled nanotubes 6–13 . The CVD methods
are attractive because of their efficiency in generat-
ing carbon feedstock and the straightforward scale-
up. Furthermore, it could be possible to gain control
over the types of nanotubes grown by controlling
various parameters that are involved in CVD experi-
ments. Nevertheless, CVD approaches have not been
fully explored to synthesize single-walled carbon
w
x
discharge methods 3,4 . Both methods involve the
evaporation of carbon at ;30008C, with the synthe-
sized SWNTs predominantly in the form of rope-like
w
x
bundles 2–4 . Currently, several topics in SWNT
synthesis still remain challenging. First, samples con-
taining mostly defect-free individual SWNTs have
yet to be obtained. Such samples are desired for
w x
fundamental studies of the physics in nanotubes 5 .
w
x
nanotubes. So far, there are only two reported 14,15
CVD syntheses of SWNTs including the method that
Secondly, practical applications of nanotubes will
require the synthesis of high-quality SWNT materi-
als at a very large scale. These challenges could be
w
x
utilizes CO and supported MoOx catalyst 14 .
In this Letter, we report high-quality SWNT ma-
terials obtained by chemical-vapor deposition of
methane on supported transition-metal oxide cata-
)
Corresponding author. E-mail: hdai@chem.stanford.edu
0009-2614r98r$19.00 q 1998 Published by Elsevier Science B.V. All rights reserved.
Ž
.
PII: S0009-2614 98 00745-3

(
)
568
J. Kong et al.rChemical Physics Letters 292 1998 567–574
Table 1
Summary of results of methane CVD experiments using supported metal-oxide catalysts
Catalyst
composition
Support
material
SWNTs?
Description of synthesized material
Fe2O3
alumina
yes
abundant individual SWNTs; some bundles;
occasional double-walled tubes
abundant SWNT bundles
some SWNT bundles and individual SWNTs
no tubular materials synthesized
mainly defective multi-walled structures
with partial metal filling
Fe2O3
CoO
CoO
NiO
silica
alumina
silica
yes
yes
no
alumina
no
NiO
NiOrCoO
NiOrCoO
silica
alumina
silica
no
no
yes
no tubular materials synthesized
no tubular materials synthesized
some SWNT bundles
lysts. Our general approach is similar to previous
work in producing carbon fibers and multi-walled
nanotubes. However, we find several factors that are
key to producing single-walled carbon nanotubes.
These factors include the choice of methane, catalyst
composition and type of support.
Ž .
Ž .
Fig. 1. a TEM image of individual and bundled SWNTs produced on the Fe2O3ralumina catalyst. Scale bar: 100 nm. b High-resolution
TEM image of a ds5 nm individual SWNT. Scale bar: 50 nm. Inset shows the dome-closed end of a ds3 nm SWNT. Scale bar: 10 nm.

(
)
J. Kong et al.rChemical Physics Letters 292 1998 567–574
569
2. Experimental
nification TEM image is shown in Fig. 1b. By
imaging several samples obtained under the same
conditions, we measured that the individual SWNTs
have diameters in ranges of 1.3–5.4 nm, and their
lengths are up to 10 mm. In Fig. 2a, we present an
AFM image of the nanotubes in this sample. The
diameters of the nanotubes are determined from their
topographic heights. Fig. 2b shows a statistical diam-
eter histogram obtained from AFM data by measur-
ing ;300 nanotubes. The histogram shows that the
nanotube diameters are dispersed from 0.7 to 6 nm
with a broad peak at 1.3 nm. Besides individual
SWNTs, we do observe other forms of nanotubes
synthesized with the Fe2O3ralumina catalyst. First,
bundles of SWNTs Fig. 1a are present, with each
bundle containing a few tubes. The diameters of
SWNTs within these bundles are also dispersed and
range from ;1 to 6 nm. Secondly, we do occasion-
ally observe double-walled nanotubes in this mate-
rial. From high-resolution TEM images, we have
measured the diameters of the SWNT bundles and
the observed double-walled nanotubes. Statistically,
the smallest diameter of the bundles is ;4 nm, and
all of the double-walled nanotubes exhibit diameters
larger than 2.5 nm. These results lead to the conclu-
sion that in the AFM data, nanotubes that exhibit
The first step for our CVD synthesis is to prepare
a supported catalyst. As an example, a Fe2O3 cata-
lyst is prepared by impregnating 1 g of fumed-
2
Ž
.
alumina Degussa, 100 m rg surface area nanopar-
ticles with 30 ml of a methanol solution that contains
Ž
.
0.245 g of Fe NO3 3 P9H2O. The impregnation typi-
cally lasts for 1 h at room temperature. The methanol
solution was then removed via rotary evaporation at
808C. The material is then heated at 1508C overnight
followed by grinding into a fine powder. This result-
ing catalyst is denoted as Fe2O3ralumina. In this
Ž
. Ž
.
work, the metal-mol r alumina weight ratio is 0.6
mmolrg and is fixed for all of the alumina supported
Ž
.
Ž
metal-oxide catalysts NiO, CoO and NiOrCoO
.
Ž
mixture with 1:1 mol ratio . For all of the catalysts
supported by fumed-silica Degussa AEROSIL, 300
2
.
Ž
. Ž
.
m rg , the metal-mol r silica weight ratio is kept
at 0.9 mmolrg.
For a methane CVD experiment, ;10 mg of the
catalyst was placed in a quartz tube mounted in a
tube furnace. An argon flow was passed through the
quartz tube as the furnace was heated to reach
10008C. The Ar flow was then replaced by methane
3
Ž
.
99% purity at a flow rate of 6150 cm rmin under
1.25 atm. head pressure. The methane flow lasted for
10 min and was replaced by argon and the furnace
was cooled to room temperature. Samples for trans-
mission electron microscopy were prepared by soni-
cating ;2 mg of the synthesized material in 10 ml
of methanol or 1,2-dichloroethane for half an hour.
A few drops of the resulting suspension were then
put onto a holey-carbon TEM grid. The TEM was
operated at beam energies between 120 and 200 kV
using a Philips CM20 instrument. For atomic force
Ž
.
diameters less than 2.5 nm are mostly )90%
individual SWNTs. Also, we notice that the nan-
otubes examined by TEM appear free of amorphous
coating and free of topological defects at least on the
1 mm length scale.
When using silica supported catalysts in methane
CVD, we find that only bundles of SWNTs are
synthesized. The SWNT bundles have diameters be-
tween 4 and 47 nm, and lengths up to tens of
micrometers. In Fig. 3a,b respectively, we show
typical low- and high-magnification TEM images of
SWNT bundles produced by a Fe2O3rsilica catalyst.
The well-resolved and aligned fringes Fig. 3b sug-
gest that the nanotubes are well packed in each
bundle. High-magnification TEM images have al-
lowed statistical measurements of the spacing be-
tween fringes in several bundles. We find that the
spacing is constant within each bundle. By tilting our
sample in the TEM, we do observe variation in the
spacing between nanotube fringes in a given bundle.
These results suggest that the individual SWNTs in a
same bundle are nearly monodispersed in diameter.
Ž
.
microscopy AFM studies, samples were prepared
by allowing a drop of the suspension to evaporate on
a freshly cleaved mica surface.
Ž
.
3. Results
We summarize our CVD syntheses in Table 1,
and the overall result is that SWNTs are obtainable
using methane CVD. In particular, with the
Fe2O3ralumina catalyst, we obtained abundant indi-
vidual SWNTs. Fig. 1a shows a TEM image of
SWNTs synthesized with this catalyst. A high-mag-

(
)
570
J. Kong et al.rChemical Physics Letters 292 1998 567–574

(
)
J. Kong et al.rChemical Physics Letters 292 1998 567–574
571
NiO, CoO and NiOrCoO catalysts are prepared
ing up of carbon over-coating to an appreciable
Ž
.
Ž
.
Ž
.
by starting with Ni NO3 2 P 6H2O, Co NO3 2 P
thickness sub-micrometer over a time period of
Ž .
a mixture of Ni NO3 2 P 6H2Or
w
x
6H2O, and
several hours near 10008C temperatures 6,16,17 .
Thirdly, so far, we find that supported Fe2O3 cata-
lysts produces SWNTs much more efficiently than
NiO, Co, and the NiOrCo mixture. Note that using
Ž
.
Co NO3 2 P6H2O, respectively. These catalysts in
methane CVD have led to interesting but complex
results. First, none of the NiOralumina and
NiOrsilica catalysts produced SWNTs. Secondly,
CoO and NiOrCoO catalysts produced SWNTs only
Ž
.
X-ray photoelectron spectroscopy XPS on the cata-
Ž
.
lyst made from Fe NO3 3 P9H2O and heated to
10008C for 5 min in argon, we have confirmed that
the active catalyst composition is Fe2O3.
Ž
.
on one of the alumina or silica supports Table 1 .
Fig. 4a,b show TEM images of SWNTs synthesized
on CoOralumina, NiOrCoOrsilica catalysts, re-
spectively. In the material synthesized on the
NiOrCoOrsilica catalyst, we have observed only
SWNT bundles. In contrast, small bundles and the
occasional individual SWNTs are observed on the
CoOralumina catalyst. Thirdly, the yield of SWNTs
appears much lower than the Fe2O3 catalysts. Cur-
rently, a quantitative measure of SWNT yield in our
materials is lacking. Yield is only estimated by
counting the number of nanotubes around a given
volume of particles from TEM data.
To understand the growth mechanism of SWNTs
in the methane CVD process, we have carried out
systematic TEM imaging of nanotube ends that are
Ž
.
easily observable in our materials. In Fig. 1b inset ,
we show a typical high-resolution TEM image of the
end of an individual SWNT synthesized on the
Fe2O3ralumina catalyst. With all of the materials
listed in Table 1, we can only observe one end of a
individual SWNT or a bundle. The other end is
always found buried in particles. Furthermore, all of
the ends that we have imaged so far appear closed
and contain no metal particles.
Based on the states of the nanotube ends, we
propose that SWNTs in our methane CVD processes
grow via the ‘base-growth’ mechanism. Previously
work on CVD synthesis of carbon fibers and nan-
otubes have found that the initial growth step in-
volves absorption, decomposition of hydrocarbon
molecules on a catalyst particle, and diffusion of
carbon atoms into the catalyst bulk from a supersatu-
4. Discussion
We believe that a number of variables in our
CVD approach are key to the success of obtaining
high-quality SWNTs. First, we chose methane as
carbon feedstock at temperatures on the order of
10008C. Methane is known to be the most kinetically
stable hydrocarbon that undergoes the least pyrolytic
decomposition at high temperatures. Therefore, the
carbon atoms needed for nanotube growth are sup-
plied by the catalytic decomposition of methane on
transition metal surfaces. This is one of the main
reasons that the synthesized nanotubes are nearly
free of amorphous carbon coatings caused by self-
pyrolysis of methane. Secondly, we limit CVD reac-
tion times to ;10 min at a high methane flow rate
to prevent amorphous carbon accumulation. Previous
methane CVD syntheses of carbon fibers and multi-
walled nanotubes have found and utilized the build-
w
x
rated catalyst surface 10,18,19 . We believe that this
first step applies to our current process. However,
the ‘base-growth’ model proposes that a nanotube
lengthens with a particle-free closed-end, and carbon
feedstock is supplied from the base where the other
end of the nanotube interfaces with the catalyst
w
x
material 10,18,20 . Thus, the ‘base-growth’ is the
dominating growth mechanism for our SWNT mate-
rials. In contrast, the ‘tip-growth’ model proposes
that a nanotube lengthens while carrying away a
w
x
metal catalyst particle at its end 10,18,20 . The
carried-along particle supplies the carbon feedstock
Ž .
Ž
.
Fig. 2. a Tapping mode AFM image of carbon nanotubes produced on the Fe2O3ralumina catalyst. Z range dark to bright : 4 nm. Scale
w
x Ž .
bar: 0.5 mm. Sonication may have caused some of the short nanotubes 25 .
b Diameter histogram of nanotubes determined from AFM
data.

(
)
572
J. Kong et al.rChemical Physics Letters 292 1998 567–574
Ž .
Ž .
Fig. 3. a TEM image of SWNT bundles synthesized with the Fe2O3rsilica catalyst. Scale bar: 100 nm. b High-resolution TEM image of
SWNT bundles exhibiting fringes of individual SWNTs in the bundles. Scale bar: 50 nm.

(
)
J. Kong et al.rChemical Physics Letters 292 1998 567–574
573
Ž .
Ž .
c TEM image of a SWNT bundle
Fig. 4.
a
TEM image of SWNTs synthesized with the CoOralumina catalyst. Scale bar: 25 nm.
synthesized on the NiOrCoOrsilica catalyst. Scale bar: 50 nm.
for growth. This ‘tip-growth’ mechanism does not
apply to the current SWNT materials, since all the
observed ends appear closed and particle-free. Nev-
ertheless, we point out that, SWNTs synthesized in
catalyst yielded abundant individual SWNTs, while
the Fe2O3rsilica catalyst yielded abundant SWNT
bundles. We rationalize these results by considering
the structures of the catalyst support materials. The
fumed-alumina material consists of crystalline d-form
w
x
the CO–MoOx CVD approach 14 are results of
‘tip-growth’, since most of the observed nanotube
w
x
Al2O3 nanocrystals. 21 These nanocrystals are
anisotropic since they contain crystal edges and cor-
ners. Therefore, different types of metal catalytic
particles may form on the alumina particles. This
leads to catalytic particles of varying sizes and distri-
w
x
ends do contain a metal nanoparticle 14 .
The distribution of active catalytic particles on a
support should dictate the type of SWNTs that are
produced. As described earlier, the Fe2O3ralumina

(
)
574
J. Kong et al.rChemical Physics Letters 292 1998 567–574
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,
butions, thus producing a mixture of tube types. In
contrast, the fumed silica has an amorphous structure
Ž
.
Science 273 1996 483.
w
x
22 and an isotropic surface. We believe that active
w x
3
D.S. Bethune, C.H. Kiang, M.S. deVries, G. Gorman, R.
catalyst particles on this isotropic support are uni-
formly and closely distributed. Therefore, the SWNTs
nucleated on the close-by catalyst sites grow into
bundles to maximize the van der Waals interactions
between the walls of the nanotubes.
Ž
.
Savoy, J. Vazquez, R. Beyers, Nature 363 1993 605.
C. Journet, W.K. Maser, P. Bernier, A. Loiseau, M.L.
Delachapelle, S. Lefrant, P. Deniard, R. Lee, J.E. Fischer,
w x
4
Ž
.
Nature 388 1997 756.
w x
Ž .
5
S. Iijima, MRS Bull. 19 1994 43.
w x
Ž .
G.G. Tibbetts, Appl. Phys. Lett. 42 1983 666.
6
A detailed understanding of the different transi-
tion metal-oxide catalysts is currently lacking and
requires future studies. We note that previously, Fe,
Ni, Co, and mixed NirCo catalysts have all success-
fully produced SWNTs in laser-ablation and arc-dis-
w x
Ž
.
7
G.G. Tibbetts, Carbon 27 1989 745.
w x
8
G.G. Tibbetts, in Carbon Fibers, Filaments and Composites,
Kluwer Academic, Amsterdam, 1990, p. 73.
w x
9
Ž
.
R.T.K. Baker, in: P. Walker, P. Thrower Eds. , Physics and
Chemistry of Carbon, Vol. 14, 1978, pp. 83.
w
w
x
x
Ž
.
10 R.T.K. Baker, Carbon 27 1989 315.
w
x
charge approaches 1–4,23,24 . Interestingly, in arc-
discharge, SWNTs have been synthesized using Fe
catalyst when methane is present in the discharge
11 R.T.K. Baker, N.M. Rodriguez, Catalytic Growth of Carbon
Nanofibers and Nanotubes, Symposium of the Materials
Research Society, Materials Research Society, 1994.
12 H.G. Tennent, Hyperion Catalysis International, USA, 1987.
13 C.E. Snyder, W.H. Mandeville, H.G. Tennent, L.K. Trues-
dale, Int. Pat. WO 89r07163, 1989.
14 H.J. Dai, A.G. Rinzler, P. Nikolaev, A. Thess, D.T. Colbert,
w
w
x
x
w x
chamber 1 . Future work is also required to further
Ž .
explore 1 detailed pictures of nanotube growth as
Ž .
well as nucleation, 2 various catalyst compositions
and concentrations, 3 new types of catalyst support,
and 4 optimum growth conditions. The methane
w
x
Ž .
Ž
.
R.E. Smalley, Chem. Phys. Lett. 260 1996 471.
w
x
15 A. Fonseca, K. Hernadi, P. Piedigrosso, L.P. Biro, S.D.
Lazarescu, P. Lambin, P.A. Thiry, D. Bernaerts, J.B. Nagy,
Electrochem. Sci. Proc. 97–14, 1997, 884.
Ž .
CVD approach could bring new possibilities to
nanoscale science and technology.
w
x
Ž
.
16 H. Jaeger, T. Behrsing, Composites Sci. Tech. 51 1994
231.
w
w
x
Ž
.
17 L.C. Qin, S. Iijima, Mater. Lett. 30 1997 311.
Acknowledgements
x
Ž
.
18 G.G. Tibbetts, M.G. Devour, E.J. Rodda, Carbon 25 1987
367.
w
w
x
Ž
.
19 G.G. Tibbetts, J. Cryst. Growth 66 1984 632.
This work was in part supported by a Camille and
Henry Dreyfus New Faculty Award. We are grateful
to J. Brauman and J. Han for helpful discussions.
x
20 S. Amelinckx, X.B. Zhang, D. Bernaerts, X.F. Zhang, V.
Ž
.
Ivanov, J.B. Nagy, Science 265 1994 635.
w
w
w
x
21 Degusssa, Technical Bulletin N56.
22 Degussa, AEROSIL: Fumed Silica.
23 C.-H. Kiang, G. Wa III, R. Beyers, J.R. Salem, D.S. Bethune,
x
x
References
Ž
.
J. Phys. Chem. 98 1994 6612.
w
w
x
x
Ž
.
24 S. Seraphin, J. Electrochem. Soc. 142 1995 290.
25 K.L. Lu, R.M. Lago, Y.K. Chen, M.L.H. Green, P.J.F.
w x
Ž
.
1
w x
2
S. Iijima, T. Ichihashi, Nature 363 1993 603.
A. Thess, R. Lee, P. Nikolaev, H.J. Dai, P. Petit, J. Robert,
Ž
.
Harris, S.C. Tsang, Carbon 34 1996 814.