A scalable CVD method for the synthesis of single-walled carbon nanotubes with high catalyst productivity

Article abstract of DOI:10.1016/S0009-2614(00)00422-X

Single-walled carbon nanotubes (SWNT) are known to possess superior mechanical and electronic properties. However, the lack of methods for large-scale preparation limits fundamental research and application development of this unique material. Among all methods currently used for SWNT preparation, chemical vapor deposition (CVD) method represents the best hope for large-scale production. However, current CVD method has limited catalyst productivity (~40% the weight of the starting catalyst). Here we report an improved CVD method for preparation of SWNTs with high productivity using a novel aerogel supported Fe/Mo catalyst. The total amount of high-quality SWNTs produced is greater than 200% the weight of the catalysts.

Full text of DOI:10.1016/S0009-2614(00)00422-X

26 May 2000
Ž
.
Chemical Physics Letters 322 2000 321–326
www.elsevier.nlrlocatercplett
A scalable CVD method for the synthesis of single-walled carbon
nanotubes with high catalyst productivity
Ming Su, Bo Zheng, Jie Liu)
Chemistry Department, Duke UniÕersity, Durham, NC 27708, USA
Received 6 January 2000; in final form 20 March 2000
Abstract
Ž
.
Single-walled carbon nanotubes SWNT are known to possess superior mechanical and electronic properties. However,
the lack of methods for large-scale preparation limits fundamental research and application development of this unique
Ž
.
material. Among all methods currently used for SWNT preparation, chemical vapor deposition CVD method represents the
Ž
best hope for large-scale production. However, current CVD method has limited catalyst productivity ;40% the weight of
.
the starting catalyst . Here we report an improved CVD method for preparation of SWNTs with high productivity using a
novel aerogel supported FerMo catalyst. The total amount of high-quality SWNTs produced is greater than 200% the
weight of the catalysts. q 2000 Elsevier Science B.V. All rights reserved.
1. Introduction
otubes possess superior mechanical properties and
chemical stability. Experimental measurements of
w x
Ž
.
Ever since its discovery by Iijima in 1991 1 ,
carbon nanotube has been one of the most actively
studied materials in today’s research. This is not very
surprising given the outstanding chemical and physi-
cal properties that this material possess and its poten-
tial applications in many different fields. For exam-
ple, depending on the number of the concentric walls
and the ways that a graphene sheets being rolled into
a cylinder, carbon nanotubes can be either metallic
Young’s moduli by atomic force microscope AFM
w x
w x
5 and thermal vibrations 6 afforded values of 1.3
and 1.8 Tpa, which are higher than any other known
materials. Consequently, the chemical stability, supe-
rior mechanical properties, together with the ballistic
transport property of metallic nanotubes and richness
in electronic properties of tubes with different helic-
ity, make carbon nanotubes ideal candidates for high
strength composite materials, and for interconnec-
tions and functional devices in molecular electronics.
Although carbon nanotube materials possess many
unique and technically important properties, lack of
sufficient amount of materials limited the study of
the fundamental properties and development of more
practical applications. The discovery of a low cost,
high productivity method for preparation of high-qu-
ality SWNT material will certainly solve one of the
w x
or semiconductive 2 . Experiments showed that indi-
vidual carbon nanotubes can behave as quantum
wires 3 and can even be made into room tempera-
ture transistors 4 . In addition, these carbon nan-
w x
w x
)
Corresponding author. Fax: q1-919-660-1605; e-mail:
jliu@chem.duke.edu
0009-2614r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
Ž
.
PII: S0009-2614 00 00422-X

(
)
322
M. Su et al.rChemical Physics Letters 322 2000 321–326
biggest problems facing this field in the past and
open new opportunities for a wide variety of applica-
tions.
2.2. Catalyst preparation
Catalysts used in the experiment were prepared
Currently, carbon nanotubes are synthesized by
w
x
using sol–gel technique 30 followed by supercriti-
cal drying. In a typical experiment, 23 g ASB was
dissolved in 200 ml Ethanol in a round bottom flash
under reflux condition. Then 0.1 ml concentrated
HNO3 diluted with 1 ml water and 50 ml ethanol
was added into the mixture. It was refluxed for 2 h
or until a clear solution was formed before adding
Ž .
three different techniques: 1 arc discharge between
w
x Ž .
two graphite electrodes 7–11 ; 2 chemical vapor
Ž
.
deposition CVD through catalytic decomposition of
Ž .
w
x
hydrocarbon 12–21 ; and 3 laser evaporation of
w
x
carbon target 22–27 . Both the laser method and arc
method yield high-quality SWNTs. However, both
techniques suffer from the problem that it is hard to
scale up the production of the nanotube materials to
industrial scale. The CVD method presently repre-
sents the best hope for large scale production of
nanotube materials. This method had been reported
for preparation of various carbon materials such as
carbon fibers and multi-walled carbon nanotubes with
Ž
.
Ž
.
1.38 g Fe2 SO4 3.4H2O and 0.38 g MoO2 acac
2
Ž
the amounts of Fe and Mo were chosen so that the
.
molar ratio was Mo:Fe:Als0.16:1:16 into the mix-
ture. After refluxing for two more hours, the mixture
was cooled to room temperature and 5 ml concen-
trated NH4OH diluted with 5 ml of water was added
into the mixture under vigorous stirring. Within a
few minutes, a gel was formed. It was left to age for
about 10 h before the supercritical drying step was
performed.
Drying of the catalysts was carried under the
supercritical condition of carbon dioxide. First, the
catalyst gel was sealed in a 300 ml high-pressure
container, which was then cooled to 08C and pressur-
ized with CO2 to above 850 psi. This filled the
container with liquid CO2. A solvent exchange step
was followed to exchange the ethanol in the gel with
liquid CO2 by slowly releasing CO2 while keeping
the pressure above 850 psi for 2 to 3 h. Then, the
container was pressured up to above the critical
w
x
high yield and at large scale 12,28 . More recently,
there are reports of SWNT preparation by CVD of
w
x
w x
carbon monoxide 19,29 , benzene 20 , ethylene
w
x
w
x
16,17 , methane 15,21 etc. Among these, methane
CVD is reported to produce high-purity and high-qu-
ality SWNT materials. However, the catalyst produc-
tivity of the method is low and the best results
reported so far give a total output ratio of ;40%
w
x
16,21 defined as the weight gain ratio of the pro-
duced nanotubes to the starting catalyst. In this letter,
we report a methane CVD method that afford high-
Ž
.
quality SWNTs with higher output ratio )200%
using a FerMo catalyst supported on Al2O3 aerogel,
a high surface area, high porosity and ultra-low-den-
sity material made by the sol–gel synthesis and
subsequent removal of the liquid solvent by super-
critical drying.
Ž
.
pressure 1050 psi . and warmed up to above the
Ž
.
critical temperature of CO2 318C . The system was
held at this condition for a short time before the
pressure was slowly reduced while the temperature
was kept at the same. Finally, the temperature was
reduced to room temperature. The catalyst prepared
this way is in the form of very fine free-flowing
powder with surface areas ;600 m2rg. The powder
was calcined at 5008C for 30 min before used for
SWNT growth.
2. Experimental
2.1. Materials
All materials used in the experiment are research
grade materials purchased from different suppliers.
Ž
.
Ž
.
Aluminum tri-sec-butoxide ASB , Fe2 SO4 3 P
2.3. SWNT growth
Ž
.
4H 2 O, bis actylacetonato dioxomolybdenum
. .
Ž
Ž
MoO2 acac were purchased from SigmarAldrich
2
Chemicals. Reagent grade nitric acid, ammonium
hydroxide and ethanol were purchased from VWR.
High-purity methane, carbon dioxide and hydrogen
were supplied by National Welders.
SWNT were prepared in a simple CVD setup
made of a tube furnace and gas flow control units. In
a typical growth experiment, ;50 mg catalyst was
put into alumina boat inside a quartz tube and was

(
)
M. Su et al.rChemical Physics Letters 322 2000 321–326
323
heated to reaction temperature, 8508C to 10008C,
under Ar flow at a flow rate of 100 sccm. The Ar
Ž
.
was switched to H2 100 sccm flowrate for 30 min
Ž
.
before switching to methane flow ;1000 sccm .
The reaction was carried out for desired time before
methane flow was turned off and Ar flow turned on
and temperature reduced to room temperature. The
material was then weighted and characterized.
2.4. Characterization
SWNT samples were fully characterized using
TEM and SEM. TEM was performed on a Philip
CM-12 operating at 100 kV. The samples for TEM
were prepared by sonicating ;1 mg material in 10
ml methanol for 10 min and drying a few drop of the
suspension on holy-carbon grid. SEM was performed
on a Hitachi S-4700 with a beam energy of 4 kV by
placing the as-grown materials on conductive carbon
tape. The amount of the prepared SWNT material
with respect to the catalyst was measured on a
Ž .
Ž .
b purified
Fig. 1. Typical TGA curves for
SWNT materials in air.
a
as prepared and
nanometers. TEM image of the materials shows that
these fibers observed in SEM are actually bundles of
SWNTs. The diameters of the tubes measured from
high-resolution TEM images are between 0.9 and 2.7
nm. Both SEM and TEM images shows the charac-
teristics of high-quality SWNT materials similar to
Ž
w
x
Thermal Gravimetric Analyzer SDT 2960, TA in-
the materials prepares in laser method 22,23 and
.
w
x
strument under flowing air with a heating rate of
arc method 10,11 . It is worth to mention that the
SEM image shown here is of as-grown materials, no
purification was performed before the imaging. The
fact that the image showed only nanotubes indicates
that the catalyst surface is fully covered with nan-
otube materials. However, for samples with weight
gain higher than 300%, amorphous carbon overcoat
was observed. We are currently working on further
improving the methods to eliminate the amorphous
carbon deposit and to increase the productivity fur-
ther. Nevertheless, even the current method shows
catalyst productivity significantly better than the pre-
58Crmin.
3. Result and discussion
The key result we report in this Letter is the
dramatic increase in the amount of nanotubes pre-
pared per unit weight of catalyst using the new
catalyst. As shown in Fig. 1a, the weight gain of the
catalyst is measured by heating up the prepared
material under flowing air in a TGA. The total
weight gain of the SWNT material is calculated by
the weight loss between 3008C and 7008C, where
SWNT material burns in air, divided by the weight
left at 7008C, which is presumably the weight of the
catalyst and support materials. For a typical 60 min
growth experiment at 9008C, the average weight gain
using this catalyst is about 200%. The maximum
weight gain we had is about 600% for a 6.5 h growth
as shown in Fig. 2.
w
x
viously reported values 15,21 .
The increase in the productivity of catalyst in
current method can be mostly attributed to the effect
of the catalyst support. Aerogels are known to have
high surface area, high porosity and ultra-low-den-
sity 31 . Because of the high surface area associated
with these materials, they have found increasing
interest as catalysts and catalyst supports 32,33 . It
has been shown that Fe2O3 catalyst supported on
SiO2 or Al2O3 aerogels exhibit catalytic activity two
to three orders of magnitude higher than that of
catalyst supported on conventional supports in Fis-
w
x
w
x
The quality of the prepared nanotube material was
characterized by TEM and SEM. As shown in Fig. 3,
The SEM image of the as-prepared material shows
tangled web-like network of very clean fibers. The
diameters of the fibers are in the range of 10–20
w
x
cher–Tropsch synthesis 34 . Moreover, the catalyst
does not deactivate with time. The authors attributed

(
)
324
M. Su et al.rChemical Physics Letters 322 2000 321–326
tions, 9008C under methane flow for 60 minutes, is
also -10%. This strongly suggests that the interac-
tion of the catalyst particles and the support materi-
als is a key factor in the growth of SWNT materials.
It is known that the surface groups on Al2O3 have
stronger Lewis acidity than those on SiO2 , thus the
interactions of the metal catalysts with Al2O3 are
w
x
stronger than with SiO2 21 . The observation that
SWNT growth favor strong metal-support interaction
indicates that the main growth mechanism in current
method is ‘base growth’ mechanism, in which the
catalyst particles are attached to the support surfaces
Fig. 2. Weight gain versus reaction time plot at 9008C with
methane flow at 1158 sccm.
these improvements to the strong interactions be-
tween the aerogel support and the catalyst as well as
the high surface area of the support. We believe this
is also true in our discovery. The high catalytic
activity of the catalyst on this new support and the
high surface area are the key factors that greatly
enhanced the performance of the catalyst in prepar-
ing nanotube materials.
We have found that supercritical drying process is
a crucial step in preparing high performance cata-
lysts. It had been shown that drying of a wet gel is a
w
x
very complex process 31,35 . Simply evaporating
the liquid would cause the gel to shrink due to the
collapse of the porous structures by the strong forces
from surface tensions at the liquidrgas interfaces
within the pores in the gel. This shrinkage would
significantly reduce the total surface areas and pore
volume of the dried material, which is normally
called xerogel. In supercritical drying process, the
liquid in the wet gel is put into the supercritical state
and therefore there are no liquidrgas interfaces in
the pores during drying. The original porous struc-
ture in the wet gels is thus reserved in the dried
aerogels. We have compared the catalysts made from
the same wet gel but dried differently. The aerogel
supported catalyst showed a weight gain of ;200%
of high purity SWNT under methane flow at 9008C
for 60 min while the xerogel supported catalyst
showed a weight gain of -5% under the same
conditions. We have also compared the same FerMo
catalysts supported on Al2O3 aerogel and on SiO2
aerogels prepared with a similar method, the weight
gain of the catalyst on SiO2 under the same condi-
Ž .
Ž .
b TEM image of SWNT sample prepared
Fig. 3.
a
SEM and
on Al2O3 aerogel supported FerMo catalyst. Sample was pre-
pared at 9008C under CH4 flow. Flow rate is 1158 sccm; reaction
time is 30 min; yield measured by TGA is 100.2%.

(
)
M. Su et al.rChemical Physics Letters 322 2000 321–326
325
during the growth process. As more and more nan-
otubes are grown on the surface, the diffusion of the
methane gas to the catalyst particles becomes more
difficult. Furthermore, we have observed amorphous
carbon deposition on the nanotubes at longer growth
time, which further reduces the diffusion rate of
methane gas to the catalyst. This would explain why
the growth rate slowed down versus time as shown
in Fig. 2. We believe if we can eliminate the deposit
of amorphous carbon by optimizing the growth con-
ditions, we could increase the amount of SWNTs
produced per unit weight of catalyst further.
Another factor that may have contributed to in-
crease the productivity of the catalyst is the high
dispersity of the metal catalysts. Since the catalysts
were prepared in a sol–gel process by dissolving Fe
and Mo salts and subsequent hydrolysis at high pH
under vigorous stirring, well dispersed nanoparticles
composed of Fe hydroxide and Mo oxide were gen-
erated, which was then converted to metal oxides in
the calcination step. We did not observe aggregates
of metal particles and large metal particles, which
would be inactive for nanotube growth, in our TEM
images. These well-dispersed catalysts also result in
a narrower size distribution of the prepared nanotube
materials compared with previous CVD experiments
results also suggested that the growth mechanism of
the SWNT in current method is ‘base growth’ mech-
anism, and the possible growth-limiting factor is the
diffusion of methane gas through the nanotube net-
works. More works need to be done to further im-
prove the productivity of the catalyst and to control
their size distributions.
Acknowledgements
This work is in part supported by Office of Naval
Research grant a00014-98-1-0597 through UNC-
CH. We are grateful for Mr. Bo Gao and Professor
Otto Zhou for help on TEM imaging, and Mr. Neal
Snider and Professor Rich Superfine for help on
SEM works. We also would like to thank Professor
Hongjie Dai for measurement of the surface area of
our catalyst and helpful discussions.
References
w x
Ž
.
1
w x
2
S. Iijima, Nature 354 1991 56.
M.S. Dresselhaus, G. Dresselhaus, P.C. Eklund, Science of
Fullerenes and Carbon Nanotubes, Academic Press, San
Diego, CA, 1996.
w
x
15,17 . More works is currently underway on con-
w x
Ž
.
3
w x
4
S.J. Tans et al., Nature 386 1997 474.
trolling the diameter of the nanotubes using current
method of catalyst preparation.
S.J. Tans, A.R.M. Verschueren, C. Dekker, Nature 393
Ž
.
1998 49.
w x
5
Ž
.
E.W. Wong, P.E. Sheehan, C.M. Lieber, Science 277 1997
1971.
Purification of the material prepared in our cur-
rent CVD method was also studied. Because of the
highly amorphous nature of the aerogel support pre-
pared in the current method, it is possible to remove
the catalyst support by refluxing the materials in
either strong acid or strong base. TGA results have
shown that, reflux the material in 3 M HNO3 for 4 h
followed by filtration eliminates all the support ma-
w x
6
M.M.J. Treacy, T.W. Ebbesen, J.M. Gibson, Nature 381
Ž
.
1996 678.
w x
Ž .
7
w x
8
S. Iijima, T. Ichihashi, Nature 363 1993 603.
Ž .
S. Iijima, MRS Bull. 19 1994 43.
w x
Ž .
9
S. Zujin et al., Solid State Commun. 97 1996 371.
w
x
10 M. Wang, X.L. Zhao, M. Ohkohchi, Y. Ando, Fullerene Sci.
Ž
.
Technol. 4 1996 1027.
w
w
x
Ž
.
11 C. Journet et al., Nature 388 1997 756.
Ž
.
x
terials Fig. 1b , leaving nanotube materials with
high purity.
12 H.G. Tennent, Hyperion Catalysis International, USA Patent,
a4663230, 1987.
w
w
w
w
w
w
w
w
w
x
Ž
.
13 M. Terrones et al., Nature 388 1997 52.
In summary, we have discovered a new form of
catalyst that can be used to prepare high-quality
SWNT with high catalyst productivity using chemi-
cal vapor deposition method. The amount of nan-
otubes prepared per unit weight of catalyst was
improved by at least a factor of 5 compared with a
similar catalyst supported on Al2O3 powder. This
discovery provided a way to prepare high-quality
SWNT materials in large scale with low cost. The
x
Ž
.
14 Z.F. Ren et al., Science 282 1998 1105.
x
Ž
.
15 J. Kong, A. Cassell, H. Dai, Chem. Phys. Lett. 292 1998 4.
x
Ž
.
16 J.H. Hafner et al., Chem. Phys. Lett. 296 1998 195.
x
Ž
.
17 E. Flahaut et al., Chem. Phys. Lett. 300 1999 236.
x
Ž
.
18 S.S. Fan et al., Science 283 1999 512.
x
Ž
.
19 H.J. Dai et al., Chem. Phys. Lett. 260 1996 471.
x
Ž
.
20 H.M. Cheng et al., Appl. Phys. Lett. 72 1998 3282.
x
21 A.M. Cassell, J.A. Raymakers, J. Kong, H.J. Dai, J. Phys.
Ž
.
Chem. B 103 1999 6484.
w
x
Ž
.
22 A. Thess et al., Science 273 1996 483.

(
)
326
M. Su et al.rChemical Physics Letters 322 2000 321–326
w
w
x
w
w
w
x
Ž
.
23 T. Guo, P. Nikolaev, A. Thess, D.T. Colbert, R.E. Smalley,
29 P. Nikolaev et al., Chem. Phys. Lett. 313 1999 91.
x Ž .
30 D.J. Suh, T.-J. Park, Chem. Mater. 9 1997 1903.
Ž
.
Chem. Phys. Lett. 243 1995 49.
x
Ž
.
x
Ž
.
24 H. Kataura et al., Jpn. J. Appl. Phys. Part 2-Lett. 37 1998
31 N. Husing, U. Schubert, Angew. Chem. Int. Ed. 37 1998
L616.
22.
w
w
x
Ž
.
w
x Ž .
32 G.M. Pajonk, Appl. Catal. 71 1991 217.
25 W.K. Maser et al., Chem. Phys. Lett. 292 1998 587.
x
w
x
33 A.J. Fanelli, S. Verma, T. Engelmann, J.V. Butlew, Ind. Eng.
26 M. Yudasaka, T. Komatsu, T. Ichihashi, S. Iijima, Chem.
Ž
.
Ž .
Chem. Res. 30 1991 126.
Phys. Lett. 278 1997 102.
w
w
x
w
x
27 M. Yudasaka, T. Ichihashi, T. Komatsu, S. Iijima, Chem.
34 F. Blanchard, J.P. Reymond, B. Pommier, S.J. Teivhner, J.
Ž
.
Ž .
Mol. Catal. 17 1982 171.
Phys. Lett. 299 1999 91.
x
w
x
28 M.S. Dresselhaus, G. Dresselhaus, K. Sugihara, I.L. Spain,
H.A. Goldberg, Graphite Fibers and Filaments, Springer,
New York, 1988.
35 J. Fricke, Aerogels, Springer, Berlin, 1986.