Chem. Mater.

1996, 8, 2109-2113

2109

Preparation of Ultrafine Carbon Tubes in Nanochannels

of an Anodic Aluminum Oxide Film
Takashi Kyotani,* Li-fu Tsai, and Akira Tomita
Institute for Chemical Reaction Science, Tohoku University, Katahira, Sendai 980-77, Japan
Received January 30, 1996. Revised Manuscript Received March 21, 1996X

A template carbonization technique was applied to prepare carbon nanotubes and

submicron-tubes in one dimensional channels. An anodic aluminum oxide film that has
uniform and straight channels with a diameter at the nanometer level was used as a onedimensional template. The pyrolytic carbon deposition on the channel wall was carried out
by exposing the anodic oxide film to propylene at 800 C. The carbon was then liberated
from the anodic oxide film by HF washing. It was found that the resultant carbon is
comprised only of uniform hollow tubes with open ends. The infiltration of poly(furfuryl
alcohol) into the channels of the film followed by heat treatment led to the formation of
bamboolike carbon tubes. For all the carbon tubes obtained here, the outer diameter (30 or
230 nm) and the length (60 or 75 m) precisely reflect the channel diameter and the thickness
of the template used, respectively. Furthermore, in the case of the carbon tubes from
propylene, the wall thickness of the tubes was easily controllable by changing the carbon
deposition period.

Introduction
We developed a template carbonization technique for
synthesizing novel carbon materials. This technique
requires using the opening or the pores whose size and
shape can be controlled at the nanometer level. First,
we succeeded in preparing a new type of thin graphite
film from the carbonization of organic polymer in the
two-dimensional opening between the lamellae of a
layered clay such as montmorillonite and taeniolite.1-5
Then, we applied this template technique to the preparation of new types of porous carbon by using the threedimensionally controlled pores of a zeolite as a space
for carbonization of poly(furfuryl alcohol).6
Our next approach in the template carbonization
technique is the preparation of one-dimensional carbon
such as carbon tubes by using the one-dimensional
channel. Carbon tubes of nanometer level are now
being expected as a new carbon material. Such tubes
have been prepared thus far mainly using an arcdischarge evaporation technique7,8 or by thermal decomposition of benzene vapor.9,10 We have attempted
to prepare carbon nanotubes and submicron-tubes by
using an anodic aluminum oxide film as a one-dimensional template. This template was originally used by
Martin et al. to synthesize nanocylinders or tubes of
polymers, metals, and semiconductors.11-14 Very recently, they have attempted to prepare carbon tubes
Abstract published in Advance ACS Abstracts, July 15, 1996.
(1) Kyotani, T.; Sonobe, N.; Tomita, A. Nature 1988, 331, 331.
(2) Sonobe, N.; Kyotani, T.; Hishiyama, Y.; Shiraishi, M.; Tomita,
A. J. Phys. Chem. 1988, 92, 7029.
(3) Sonobe, N.; Kyotani, T.; Tomita, A. Carbon 1991, 29, 61.
(4) Kyotani, T.; Yamada, H.; Sonobe, N.; Tomita, A. Carbon 1994,
32, 627.
(5) Kyotani, T.; Mori, T.; Tomita, A. Chem. Mater. 1994, 6, 2138.
(6) Kyotani, T.; Nagai, T.; Tomita, A. Extended Abstracts of Carbon
92; Essen, 1992; 437.
(7) Iijima, S. Nature 1991, 354, 56.
(8) Ebbesen, T. W.; Ajayan, P. M. Nature 1992, 358, 220.
(9) Endo, M. Chemtech 1988, 18, 568.
(10) Endo, M.; Takeuchi, K.; Igarashi, S.; Kobori, K.; Shiraishi, M.;
Kroto, H. W. J. Phys. Chem. Solids 1993, 54, 1841.
X

S0897-4756(96)00063-4 CCC: $12.00

from polyacrylonitrile using the template.15 Our attempt for the carbon tube preparation was briefly
reported in a short communication.16 In the present
work, we report the more details of the method and the
characteristics of the resultant carbon tubes. Furthermore, we would like to emphasize here how effectively
uniform carbon tubes can be prepared and how easily
their size can be controlled by using this method.
Experimental Section
Preparation of Anodic Aluminum Oxide Films. The
following two types of circular anodic oxide films were used
as a template. One was a commercially available membrane
filter of 25 mm o.d. wide and 60 m thick (Whatman Ltd.,
Anodisc 25), whose porosity consisted of an array of parallel
and straight channels with a diameter of about 230 nm.
Another one was prepared by anodic oxidation of aluminum
plate (purity 99.99%) at a cell voltage of 20 V in 20 wt %
sulfuric acid at 0 C for 2 h. Following the electrooxidation,
an anodic oxide film was separated from the aluminum
substrate by reversing the polarity of the cell voltage. Then
an impervious layer (referred to as the barrier layer) was
etched by immersing the film in 20 wt % sulfuric acid for 1 h.
The diameter and the thickness of this film were 15 mm and
75 m, respectively, and the diameter of its straight channels
was about 30 nm.
Carbonization of Poly(furfuryl alcohol). The commercial anodic films were dried at 150 C for 3 h under
vacuum. After cooling to room temperature, the films (5 or 6
sheets) were impregnated under vacuum with a mixture of
furfuryl alcohol (100 cm3) and oxalic acid (0.25 g) as an acid
catalyst. The mixture was stirred with the films for 3 days.
The polymerization of furfuryl alcohol and its subsequent
(11) Liang, W.; Martin, C. R. J. Am. Chem. Soc. 1990, 112, 9666.
(12) Brumlik, C. J.; Martin, C. R. J. Am. Chem. Soc. 1991, 113,
3174.
(13) Klein, J. D.; Herrick, II, R. D.; Palmer, D.; Sailor, M. J.;
Brumlik, C. J.; Martin, C. R. Chem. Mater. 1993, 5, 902.
(14) Martin, C. R. Science 1994, 266, 1961.
(15) Parthasarathy, R. V.; Phani, K. L. N.; Martin, C. R., Adv.
Mater. 1995, 7, 896.
(16) Kyotani, T.; Tsai, L.; Tomita, A. Chem. Mater. 1995, 7, 1427.

1996 American Chemical Society

2110 Chem. Mater., Vol. 8, No. 8, 1996

Kyotani et al.

Figure 1. Schematic drawing of the formation process of

carbon tubes.
carbonization in the channels of the films were carried out by
heating the impregnated films under N2 flow up to 900 C at
a rate of 5 C/min and holding it for 3 h. The resultant
carbon-anodic oxide composite films were washed with an
excess amount of 46% HF solution at room temperature to
dissolve the anodic aluminum oxide template. As a result,
carbon was obtained as an insoluble fraction.
Pyrolytic Carbon Deposition from Propylene. In addition to the above impregnation method, we attempted to
deposit pyrolytic carbon on the inside of the straight channels
of anodic oxide film in the following way. Each anodic oxide
film was cut into two semicircular films. About four pieces of
the semicircular films were placed on quartz wool in a vertical
quartz reactor (i.d. 20 mm). The reactor temperature was
increased to 800 C under N2 flow and then propylene gas
(2.5% in N2) was passed through the reactor at a total flow
rate of 200 cm3 (STP)/min. The thermal decomposition of
propylene in the uniform straight channels of the anodic oxide
films results in carbon deposition on the channel walls. After
the desired period of time, the reactor was cooled to room
temperature and the films were removed. Then they were
washed with HF as described above, and only carbon was left
as an insoluble fraction. The formation process of carbon tubes
is illustrated in Figure 1.
Some of the carbon samples from the two methods were
heat-treated at 2800 C for 1 h under Ar flow, using an electric
furnace with a graphite heater.
Characterization. The microscopic features of the samples
were observed with a scanning electron microscope (SEM,
Hitachi, S900) and transmission electron microscopes (TEM;
JEOL, JEM-2000EXII, and JEM-ARM1250). Electron diffraction patterns for selected areas (SAD) were also taken. The
interplanar spacing, d002, and the average crystallite size, Lc,
of the carbon samples were determined with an X-ray diffractometer (XRD, Shimadzu, XD-D1 with Cu KR radiation) by
referring to a silicon standard. The surface area and the pore
structure were investigated with an automatic volumetric
sorption analyzer (Quantachrome, Autosorb-1) using N2 as
adsorbent at -196 C. The surface area and the pore size
distribution were determined using the BET equation and the
BJH method, respectively.

Results and Discussion

Structure of Anodic Aluminum Oxide Films.
Figure 2 shows the SEM photographs of the surface and
the cross section for the two types of the anodic oxide
films. Many openings with an uniform diameter (about
230 and 30 nm for photographs a and c, respectively)
are observed on the surface. The cross-section views
(b and d) indicate the presence of parallel and straight
channels perpendicular to the film surface. The BET
surface area for the anodic oxide film with the channels
of 30 nm in diameter was indicated in Table 1, where
the area calculated by the geometrical dimension esti-

Figure 2. SEM photographs of the aluminum anodic oxide

film with 230 nm channels (surface a, cross section b) and the
film with 30 nm channels (surface c, cross section d).

Figure 3. Pore distribution curves for the anodic oxide films

with 30 nm channels and the carbon tubes prepared by carbon
deposition of propylene for 1 h on the above film.
Table 1. BET Surface Area and the Area Calculated by
the Geometrical Dimension from the SEM Observation
surface area (m2/g)
sample

BET

calc

anodic oxide film with 30 nm channels

carbon tubes from the above film

23
225

19
270

mated from the SEM observation was put in the third

column. The BET surface area is in fair agreement with
the calculated value. The pore distribution curve of this
film (Figure 3) has a sharp peak at about 30 nm of pore
diameter, which is consistent with the channel diameter
from the SEM observation (Figure 2c). These findings
imply the size and shape of the channels to be very
uniform and non porous nature of the aluminum oxide
substrate.
Carbon from Poly(furfuryl alcohol). Figure 4a,b
shows the SEM photographs of the resultant carbon and
its 2800 C treated sample. Figure 4a indicates the
formation of tubular carbon, whose diameter is almost
equal to the channel diameter (230 nm) of the corresponding anodic oxide film as a template. Each tube

Preparation of Ultrafine Carbon Tubes

Chem. Mater., Vol. 8, No. 8, 1996 2111

Figure 4. SEM and TEM photographs of the original carbon

prepared from PFA (SEM image a, TEM image c) and the
sample following heat treatment at 2800 C (SEM image b,
TEM image d). In the TEM images a microscope grid is seen
under the sample.

Figure 5. SEM photographs of the carbon tubes prepared by

carbon deposition of propylene: (a and b) a carbon deposition
period of 1 h on the anodic oxide film with 30 nm channels; (c
and d) a period of 12 h on the film with 230 nm channels.

ramifies into several thin tubes near the end. This

branchlike structure originates from a similar branching
in pore structure of the commercial anodic oxide film
near its surface. The heat treatment at 2800 C
changed the microscopic feature of the tubes, although
the tubular structure still remained. The surface of the
walls became rough, and there are several holes on some
parts on the surface (Figure 4b). The following TEM
photographs (Figure 4c,d) give a clearer view of the
structure of these carbon samples. The photograph of
the carbon without the heat treatment shows voids and
knots in the tubes (Figure 4c). This is just like bamboo.
After the heat treatment, some carbon tubes kept the
bamboolike structure, but the others show the structure
as in Figure 4d, where they look as if many bubbles
appeared in the tubes. Although we do not have now a
clear explanation for the formation of such structure as
bamboo and bubble, it was confirmed that the carbonization in the one-dimensional channels gives the
formation of one-dimensional carbon.
Formation of Carbon Tubes from Propylene.
After the pyrolytic carbon deposition from propylene,
the anodic films were treated with HF to obtain carbon
as a residue. Figure 5 shows the SEM photographs of
the carbon samples from the two types of the films (the
film with 30 nm channels was subjected to the carbon
deposition for 1 h and the film with 230 nm channels
for 12 h). These figures reveal that in both cases the
samples consist of only cylindrical tubes and their outer
diameter is almost the same as the channel diameter
of the corresponding anodic oxide film. No other form
of carbon was found in the SEM and TEM observation.
In the SEM photographs with low magnification (Figure
5c), many bundles of the tubes can be observed, and the
length of a whole tube in a bundle was almost the same
as the thickness of the corresponding template film. The
presence of so many bundles implies that most of the
tubes are connected at both ends of each tube, because

the carbon deposition also took place on the outer

surface of the anodic film. But some of the tubes were
separated from the others during the stirring in the HF
solution, as observed in the other pictures. It is
noteworthy that the tubes from the anodic oxide film
with the smaller channels (Figure 5b) look transparent
under the SEM observation with an acceleration voltage
of 15 kV, indicating that the wall thickness of these
tubes is very thin.
The BET surface area and the pore distribution curve
of the tubes from the film with 30 nm channels are
shown in Table 1 and Figure 3. As in the case of the
anodic oxide film, the surface area of the carbon tubes
was calculated from their geometrical dimension with
assuming carbon density to be about 2 g/cm3. A good
agreement between the BET and the calculated surface
areas was obtained. On the other hand, their pore
distribution curve is found to be broader especially to
the large diameter side than that of the anodic oxide
film used as a template. This can be explained from
the fact that there is a long space among the carbon
tubes in a bundle, and it was also counted as pore.
Figure 6 shows the SEM and TEM photographs for
the carbon tubes under different carbon deposition
periods. The SEM photographs (Figure 6a-c), which
were taken from the end of the tubes, show the openend structure of these carbon tubes, and the TEM
photographs (Figure 6d-f) confirm that the shape of the
carbon samples here are tubular. These were also the
case for the carbon tubes from the anodic oxide film with
the 30 nm channels (see Figure 7a). These images
clearly demonstrate that the wall thickness of the tubes
increases with an increase in the deposition period. The
wall thickness can be roughly estimated from the TEM
photographs, where the thicknesses are in the ranges
3-5, 40-45, and 60-80 nm for 1, 6, and 12 h deposition,
respectively. Furthermore, the TEM observation exhibits that the morphology of the inside of the tubes
looks rougher with increasing the deposition period.

2112 Chem. Mater., Vol. 8, No. 8, 1996

Figure 6. SEM and TEM photographs of the carbon tubes

prepared in the anodic oxide film (channel diameter 230 nm)
under different carbon deposition periods: (a-c) SEM images;
(d-f) TEM images.

Figure 7. TEM images of the carbon tubes prepared by

carbon deposition for 1 h: (a) bright-field image of the tubes
prepared from the anodic oxide film with 30 nm channels; (b)
SAD pattern from the area indicated by a circle in a; (c) brightfield image of the tubes from the film with 230 nm channels;
(d and e) SAD patterns from the areas indicated by circles in
c.

Structural Analysis for Carbon Tubes from Propylene. The structural parameters such as the interplanar spacing (d002) and the average crystallite size (Lc)

Kyotani et al.

Figure 8. High-resolution TEM image of the carbon tubes

from the anodic oxide film with 30 nm channels under a carbon
deposition period of 6 h.

were, respectively, determined from the position and the

half-width of a broad (002) X-ray diffraction peak for
the carbon tubes with a 30 nm-diameter (carbon deposition period 1 h) and the ones with a 230 nm diameter
(carbon deposition period 3 h). The values of d002 and
Lc were 0.352 and 2.0 nm for the former carbon tubes
and 0.354 and 2.7 nm for the latter tubes, respectively.
Their d002 values are much larger than that of ideal
graphite (d002 ) 0.3354 nm) and Lc are very small in
comparison with these of graphitized carbon materials,
indicating low crystallinity of the carbon tubes obtained
here.
Figure 7 presents the TEM bright field images of the
carbon tubes and their corresponding SAD patterns. The
selected area for each SAD pattern is indicated by a
circle in the images. The SAD patterns (Figure 7b,d)
exhibit a pair of small but strong arcs for 002, together
with weak 10 and 11 diffraction rings. The appearance
of the 002 diffraction neither as a ring nor as clear spots,
but as a pair of small arcs, indicates some orientation
of the 002 planes in the carbon tubes and its poor
crystallinity. The latter accords with the results from
the XRD measurement. For the carbon tubes with a
230 nm diameter, the SAD pattern (Figure 7e) was
taken from a different area. It is noted that, in the area
d of Figure 7c, part of electron beam passed through
the tube wall in parallel, but no beam paralleled the
wall in the area e. The 002 arcs as observed in Figure
7d are not seen in the case of Figure 7e, where only 10
and 11 rings are observed. This finding suggests that
the tube wall consists of cylindrically stacked 002
planes. This would also be the case for the tubes with
a diameter of 30 nm, which are, however, too thin to
take such a SAD pattern as in Figure 7d,e. Accordingly,
the observation of 002 lattice image was attempted for
these carbon tubes.
The lattice image for the carbon tubes with a diameter
of 30 nm is shown in Figure 8, where at least four tubes
cross each other. The thickness of the walls is about
10 nm, and consequently the carbon has a hollow with
a diameter as small as 10 nm. Many small lines, which
correspond to 002 lattice planes, are observed in the
cross section of the walls for each tube. This image
demonstrates that the size of most 002 planes is less

Preparation of Ultrafine Carbon Tubes

Chem. Mater., Vol. 8, No. 8, 1996 2113

observed in the lattice image (Figure 8) would cause

such high graphatizabilty.
Conclusions

Figure 9. TEM images of the carbon tubes (diameter; 230

nm, carbon deposition period; 3 h) heat-treated at 2800 C:
(a) bright-field image; (b) SAD pattern from the area indicated
by circle in a.

than 10 nm, and they wrinkle to a great extent. This

structure is far from a graphitic one. It should be noted
that all 002 planes are orientated toward the direction
of carbon tube axis. This lattice image confirms that
the thinner tubes are also comprised of cylindrically
stacked 002 planes as well as the case of the thick tubes.
The carbon tubes prepared from the anodic oxide film
with 230 nm channels were subjected to the heat
treatment at 2800 C. Figure 9 shows the TEM
photograph and the SAD pattern for the heat-treated
carbon tubes. Unlike the case of the carbon tubes from
PFA, no bubble pattern appeared even after the heat
treatment. Some of the tube walls look broken or
collapsed. The SAD pattern was taken from one of the
carbon tubes, which is indicated by a circle in the brightfield image. In the pattern, 002 diffraction was observed
as a pair of spots, not as arcs. Furthermore, 100 (and
probably 101) and 110 diffractions appeared as many
spots along circle lines, not as continuous circles. These
findings clearly indicate the significant increase in the
degree of ordering of 002 planes with the heat treatment. The carbon tubes from the pyrolytic carbon
deposition can be, thereby, considered as a graphitizable
carbon. The orientated structure in 002 planes as

We can conclude that one-dimensional carbon was

prepared by the template carbonization method using
an anodic oxide film. The infiltration of furfuryl alcohol
into the channels of the film followed by heat treatment
led to the formation of bamboolike carbon tubes and the
pyrolytic carbon deposition from propylene on the channels allowed us to synthesize uniform carbon nanotubes
and submicron-tubes. The nanotubes prepared from the
other techniques such as arc-discharge evaporation are
much more graphitic and finer than the present carbon
tubes, and their structure is close-ended. These techniques do not, however, allow the preparation of uniform
carbon tubes and the accurate control of the tube length,
diameter, and thickness. In addition, the arc-discharge
evaporation technique produces a large amount of
amorphous carbon as well as carbon nanotubes. On the
other hand, one can prepare monodisperse carbon tubes
by the present method. Since the length and the inner
diameter of the channels in an anodic oxide film can
easily be controlled by changing the anodic oxidation
period and the current density during the oxidation, it
is possible to control the length and the diameter of the
carbon tubes. Furthermore, by changing the carbon
deposition period, the wall thickness of the carbon tubes
is controllable. The present method makes it possible
to produce no other form of carbon, but only carbon
tubes that are not capped at both ends. Thus, the
encapsulation of other material into the present tubes
would be very easy. This opens up the possibility for
the formation of a novel one-dimensional composite with
carbon.
Acknowledgment. We thank the High Voltage
Electron Microscope Laboratory of Tohoku University
for microscopical analysis. This study was partly supported by Special Coordination Funds for Promising
Science and Technology from Science and Technology
Agency, Japan, the Ministry of Education, Culture
and Science of Japan (07650777) and the Mitsubishi
Foundation.
CM960063+