Formation of carbon nanotubes in silicon-coatedalumina nanoreactor

Mei Lu a, Zhe Wang a, Hu-Lin Li a,*, Xin-Yong Guo b, Kin-Tak Lau c

a Chemistry Department of Lanzhou University, TianShui Road 298, Lanzhou 730000, Chinab Laboratory of Special Functional Materials, Henan University, Kaifeng 475001, China

c Department of Mechanical Engineering, The Hong Kong Polytechnic University, Kowloon, Hong Kong

Received 28 March 2003; accepted 15 January 2004

Available online 17 March 2004

Keywords: A. Carbon nanotubes; B. Chemical vapor deposition; C. Electron microscopy

Over recent years, ordered carbon nanotubes (CNTs)

arrays have generally been prepared using the template

method, i.e., by pyrolyzing hydrocarbon into nano-

channels of alumina film [1–8]. The most notable feature

of the CNTs arrays thus prepared is that the arrays

perfectly copy the three-dimensional structures of the

nanochannels in the template. The resulting outer

diameter of the obtained CNTs matches the innerdiameter of the channels. Construction of CNTs

with diameter smaller than the lowest attainable diam-

eter of the template is not possible with this method. The

work presented in this paper describes a new technique

of CNTs production where the alumina nanochannel is

not used as a traditional template, but instead as a na-

noreactor to carry out submicroscopic chemical reac-

tions. CNTs of interesting formation and high density ina single cavity of silicon-coated alumina nanochannel

were observed. With this approach, higher surface area

of CNTs in a confined space could be obtained which

would be beneficial to applications in the areas of high

capacity hydrogen storage, solar thermal collectors and

super-capacitor batteries.

The method of preparation of alumina film with

channel-diameters of 250–300 nm has been described inearlier publications [9–11]. Prior to chemical vapor

deposition, the bottoms of the pores on the alumina film

were opened up via chemical etching in 0.3 mol/L

H3PO4 aqueous solution. The alumina film was then

placed in a quartz tube furnace with atmosphere in the

reactor pumped down to 50 Pa. A mixture of N2 (10 ml/

min) and Ar (30 ml/min) was purged for 0.5 h and fol-

lowed by heating the reactor to 800 �C. Then, a flow ofSiH4 (10 ml/min) was introduced for 1 h under a net

pressure of 150 Pa, which resulted in a uniform silicon

coating on the channel walls of the alumina film. The

silicon-coated film was then immersed into 0.1 mol/L

FeSO4 Æ 7H2O solution for 5 min and then subjected to

Fe ions reduction under the atmosphere of H2 at 800 �Cfor 0.5 h. Subsequently, the heating temperature was

lowered to 700 �C and a flow of acetylene (15 ml/min)

was allowed for 1 h. After cooling down, the composite

film was broken into fine pieces and treated with 10 M

NaOH solution to dissolve the alumina. The depositedcomposites were sonicated in ethanol and a drop of the

resulting sample was evaporated on a copper grid cov-

ered with a carbon film. TEM analysis was carried out

with JEOL-2010 microscope at 200 kV equipped with

Link-ISIS energy dispersive spectroscopy elemental

composition analyzer. Elementary X-ray mapping was

made by scanning the sample with an electron beam to

generate characteristic X-radiation from the elementsexcited in the sample with a dwell time of 50 ms. Digital

data of X-ray mapping were memorized in the computer

processing for imaging during 1 h.

Fig. 1a shows the TEM image of the composite dis-

solved from the alumina film at low magnification. The

outer diameter of silicon tube coating corresponds to the

inner diameter of the alumina nanochannels. Numerous

small wire-like or tube-like inserts can be seen within thecavity of the tube. The appearances of the inserts can be

further examined in Fig. 1b, where the composite is

placed under high magnification. The image shows

many tangled CNTs with diameters of 10�12 nm within

the hollow interior. These ‘‘as-grown’’ CNTs are quite

clean with few side products such as amorphous carbon

and carbon nanoparticles. The tops of the CNTs are

open and a few metal particles can be observed at thebase of the CNTs. Furthermore, the inset of Fig. 1a

shows multiple composite structures, evidence that the

formation of the CNTs/silicon coating is not accidental.

The elemental constitution of the composite is con-

firmed with elementary X-ray mapping (two-dimen-

sional energy dispersive X-ray analysis). By scanning the

composite structure indicated in the first frame of Fig. 2

*Corresponding author. Tel.: +86-931-891-2517; fax: +86-931-891-

2582.

E-mail address: lihl@lzu.edu.cn (H.-L. Li).

0008-6223/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.carbon.2004.01.052

1846 Letters to the Editor / Carbon 42 (2004) 1846–1849

for 1 h, Si-mapping presents a clear view of tube-struc-ture and suggests the sheath of the composite is indeed

silicon after the alumina film was dissolved. C-mapping

test shows that carbon is dispersed in the channel of the

silicon tube. In addition, oxygen can be observed to be

distributed over most of the carbon film, which could

have been caused by the contamination of the TEM

sample preparation. X-ray mapping results strongly

suggest that a large quantity of CNTs were actuallyproduced in a single silicon-coated alumina channel.

Based on these observations, the alumina channel trea-

ted with this method may be viewed as a nanometer-

sized test tube or nanoreactor, in which a variety of

nanoscale chemical reactions can be carried out.

Referring to HRTEM image of the silicon sheath(Fig. 3) one can see a clear lattice fringes paralleling to

the silicon tube growth axis. The regular distance of the

observed lattice planes is 0.315 nm, which is consistent

with the spacing of (1 1 1) plane of silicon. Here the fast

growth direction is along the ()2 1 1) axis of silicon tube.

The finding of free CNTs in TEM observation was

unexpected. These CNTs may have been released from

the destroyed silicon tube during the TEM samplepreparation and are shown in Fig. 4a. Other than hollow

CNTs, bamboo-like structured CNTs were also ob-

served in which the curvature of the compartment layer

is directed toward the top of the tube. HRTEM taken

from the open end of one of these tubes reveals an

Fig. 1. (a) Low magnification TEM image of a composite structure dissolved from alumina film. Inset: A TEM image of several composite

structures; (b) high magnification TEM image of a composite structure.

Fig. 2. TEM-EDS X-ray mapping of C, Si, O in the composite

structure.

Fig. 3. HRTEM image of the silicon coating in the composite struc-

ture.

1847Letters to the Editor / Carbon 42 (2004) 1846–1849

ordered graphitic structure of multi-walled tube (Fig.4b). The interlayer separation is approximately 0.333

nm, a figure that is close to the interplanar distance in

graphite (d0 0 2 ¼ 0:335 nm). The tube wall thickness is

about 4 nm, suggesting that the multi-walled carbon

nanotube consists of approximately 12 graphitic layers.

It is clear that the mechanism of the CNTs growth in

silicon-coated alumina nanoreactor is different from the

conventional CNTs arrays produced in alumina tem-plate. Since alumina itself is a catalyst in the hydrocar-

bon decomposition process due to the Lewis acid nature

of its surface sites [12], albeit a far less effective catalyst

than metal, its presence complicates the mechanism of

CNTs growth in pure alumina template. The general

belief is that the conventional CNTs are formed by the

joint catalytic activities of metal nanoparticles and alu-

mina walls [12,13]. The metal catalyst might merelyinitiate the growth of the tube through the initial cata-

lytic decomposition of the hydrocarbon reagent, then

carbon, possibly as a carbide transport along the surface

of the metal nanoparticle to the alumina channel wall,

where subsequent tube growth proceeds. In this case, the

inside of the channel wall will be covered by carbon

flakes that will form a tube-like structure.

With the silicon-coated alumina method presentedhere, the silicon coating on the channel wall may con-

tribute to the formation of CNTs by eliminating the

catalytic behavior of the alumina, in which case the

CNTs are only catalyzed by the small metal nanoparti-

cle independently throughout the entire growth process.

The function of the alumina channel is only as a reactor

or a tube for the reaction between the metal nanopar-

ticles and the carbon atoms. TEM observation hasshown that the metal nanoparticles are located at the

bases of the nanotubes, indicating that a base-growth

mechanism could mainly be responsible for the CNTs

growth in the silicon-coated alumina channel.

It should be emphasized that the pressure and flow

rate of the hydrocarbon inside the channels is different

than that on the outside. For this reason, the aluminachannels with larger diameter and the thinner silicon

coating may be favored for its function as nanoreactors.

In summary, high density CNTs have been fabricated

in a silicon-coated alumina nanochannel by chemical

vapor deposition. With this approach, the nanochannel

with silicon coating is akin to a nanoreactor, in which

CNTs of various small diameters can be obtained. It is

reasonable to believe that the described nanoreactorcould have a variety of industrial applications in the

production of silicon encapsulated metal, metal com-

pounds and semiconductor nanowires. Thus, the CNTs

formation described here is quite a unique phenomenon

and can provide a novel methodology for future nano-

fabrication technology as well as of significance in the

production of nanotube composites with alumina and

CNTs.

Acknowledgements

This work is supported by the National Natural

Science Foundation of China (grant no. 60171004) and

The Hong Kong Polytechnic University Research Grant

(G-T 684).

References

[1] Parthasarathy RV, Phani KLN, Martin CR. Template synthesis

of graphitic nanotubules. Adv Mater 1995;7(11):896–8.

[2] Kyotani T, Tsai L, Tomita A. Preparation of ultrafine carbon

tubes in nanochannels of an anodic aluminum oxide film. Chem

Mater 1996;8:2109–13.

[3] Lakshmi B, Fisher ER, Martin CR. Carbon nanotubule mem-

branes for electrochemical energy storage and production. Nature

1998;393:346–9.

[4] Che G, Lakshmi BB, Martin CR, Fisher ER. Chemical vapor

deposition based synthesis of carbon nanotubes and nanofi-

bers using a template method. Chem Mater 1998;10:260–

7.

Fig. 4. (a) TEM image of CNTs released from silicon coating; (b) HRTEM image of an open end of the carbon nanotube shown in (a).

1848 Letters to the Editor / Carbon 42 (2004) 1846–1849

[5] Sung SL, Tsai SH, Tseng CH, Chiang FK, Liu XW, Shih HC.

Well-aligned carbon nitride nanotubes synthesized in anodic

alumina by electron cyclotron resonance chemical vapor deposi-

tion. Appl Phys Lett 1999;74(2):197–9.

[6] Haruyama J, Takesue I, Sata Y. Coulomb blockade in a single

tunnel junction directly connected to a multiwalled carbon

nanotube. Appl Phys Lett 2000;77(18):2891–3.

[7] Han CC, Lee JT, Chang H. Thermal annealing effects on structure

and morphology of micrometer-sized carbon tubes. Chem Mater

2001;13:4180–6.

[8] Sui YC, Acosta DR, Gonzalez-leon JA, Bermudez A, Feuchtw-

anger J, Cui BZ, et al. Structure, thermal stability, and deforma-

tion of multibranched carbon nanotubes synthesized by CVD

in the AAO template. J Phys Chem B 2001;105:1523–7.

[9] Kyotani TB, Pradhan K, Tomita A. Synthesis of carbon nanotube

composites in nanochannels of an anodic aluminum oxide film.

Bull Chem Soc Jpn 1999;72(9):1957–70.

[10] Qin DH, Lu M, Li HL. Magnetic force microscopy of magnetic

domain structure in highly ordered Co nanowire arrays. Chem

Phys Lett 2001;350(1–2):51–6.

[11] Sha J, Niu J, Ma X, Xu J, Zhang X, Yang Q, et al. Silicon

nanotubes. Adv Mater 2002;14(17):1219–21.

[12] Scokart PO, Rouxhet PG. Comparison of the acid-basenproper-

ties of various of oxides and chemically treated oxides. J Colloid

Interface Sci 1982;86(1):96–104.

[13] Jing L, Martin M, Tom HL. Nanoscale electroless metal

deposition in aligned carbon nanotubes. Chem Mater 1998;

10:1963–7.

Effects of heat treatment conditions on the thermal propertiesof mesophase pitch-derived graphitic foams

James W. Klett *, April D. McMillan, Nidia C. Gallego,Timothy D. Burchell, Claudia A. Walls

Metals and Ceramics Division, Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, TN 37831, USA

Received 2 April 2003; accepted 20 January 2004

Available online 17 March 2004

Keywords: A. Mesophase pitch, porous carbon; B. Heat treatment; D. Thermal conductivity

Carbon foam was first developed by researchers in

the late 1960s as reticulated vitreous (glassy) carbon

foams [1,2]. More recently, Klett and coworkers [3–5] atthe Oak Ridge National Laboratory reported the first

graphitic foams with bulk thermal conductivities greater

than 58 W/mK; thermal conductivities up to 180 W/mK

have recently been reported [6–8]. By combining an

open cellular structure with a thermal conductivity to

weight ratio (k=q) greater than 200 (as compared to 45

for copper), the development of high conductivity gra-

phitic foam presents a unique opportunity to radicallychange the approach to many heat transfer problems.

For many graphitic materials, the heating rate during

graphitization is critical to optimizing the material

properties. Therefore, in this project we examine the

influence of heating rate during graphitization on key

properties of the foam, such as density, thermal diffu-

sivity, and crystallinity.

Several billets of foam (~15 cm� ~10 cm� ~2.5 cm)were produced from Mitsubishi AR mesophase pitch

powder with a softening point of 235 �C. The mesophase

pitch was foamed under 1000 psi pressure in aluminum

pans utilizing two heating rates during (3.5 and 10 �C/min foaming rate) [9]. The foaming step consisted of

heating under vacuum to 250 �C and soaking for 1 h,

applying the foaming pressure, and then heating at the

specified heating rate to 600 �C, soaking for 1 h, andthen cooling to room temperature (while simultaneously

reducing pressure) at approximately 1.25 �C/min. All

billets were marked for their position in the furnace (i.e.

bottom or top of furnace) and then carbonized to 1000

�C under an atmospheric nitrogen purge at a heating

rate of 0.2 �C/min. The carbonized billets were then cut

into 1.59 cm cubes in a regular pattern throughout the

foam billets and the Euclidian density was measured foreach cube (ASTM C559). The foam cubes were sepa-

rated into four different groups for each initial foaming

rate (3.5 and 10 �C/min) and subsequently heated to a

graphitization temperature of 2800 �C in four separate

runs with different heating rates (1, 5, 10, and 15 �C/min)

under an argon purge. Key material properties such as

crystallographic structure, thermal diffusivity (measured

by ASTM C714) and density were determined.Table 1 reports the density uniformity for the billets

processed under different foaming rates and in different

positions within the furnace. As can be seen, there is

little effect on uniformity (smaller deviations between

the maximum and minimum values) due to the heating

rate during foaming. However, a significant effect on

density uniformity with the position in the furnace was

observed. While there appeared to be some effect ondensity uniformity due to the foaming rate, this effect

*Corresponding author. Tel.: +1-8655-745220; fax: +1-8655-

768424.

E-mail address: klettjw@ornl.gov (J.W. Klett).

0008-6223/$ - see front matter Published by Elsevier Ltd.

doi:10.1016/j.carbon.2004.01.057

1849Letters to the Editor / Carbon 42 (2004) 1846–1849