Synthesis of Cu Nanoparticles and Microsized Fibers by Using Carbon Nanotubes as a Template

Article abstract of DOI:10.1021/jp983983j

Copper in the form of nanoparticles and microsized fibers was synthesized by using carbon nanotubes as a template. The size of the copper nanoparticles was controllable in the range between a few and tens of nanometers, depending on the diameter of carbon nanotubes and the preparation conditions. They could be detached from carbon nanotubes by ultrasonic treatment. Cu fibers with diameter ranging from 100 nm to several μm were obtained when increasing the amount of Cu coated onto carbon nanotubes, which in this case became undetectable under SEM and XRD observations. The electronic structure of these two forms of Cu materials was studied by XPS and UPS and compared with that of polycrystalline Cu metal. A negative core-level binding energy shift for Cu nanoparticles and an increase in work function for Cu fibers as compared with those of the polycrystalline Cu were observed and discussed in terms of size effect.

Full text of DOI:10.1021/jp983983j

© Copyright 1999 by the American Chemical Society
VOLUME 103, NUMBER 22, JUNE 3, 1999
LETTERS
Synthesis of Cu Nanoparticles and Microsized Fibers by Using Carbon Nanotubes as a
Template
P. Chen, X. Wu, J. Lin,* and K. L. Tan
Physics Department, National UniVersity of Singapore, 10 Kent Ridge Crescent, Singapore, 119260
ReceiVed: October 6, 1998; In Final Form: April 12, 1999
Copper in the form of nanoparticles and microsized fibers was synthesized by using carbon nanotubes as a
template. The size of the copper nanoparticles was controllable in the range between a few and tens of
nanometers, depending on the diameter of carbon nanotubes and the preparation conditions. They could be
detached from carbon nanotubes by ultrasonic treatment. Cu fibers with diameter ranging from 100 nm to
several µm were obtained when increasing the amount of Cu coated onto carbon nanotubes, which in this
case became undetectable under SEM and XRD observations. The electronic structure of these two forms of
Cu materials was studied by XPS and UPS and compared with that of polycrystalline Cu metal. A negative
core-level binding energy shift for Cu nanoparticles and an increase in work function for Cu fibers as compared
with those of the polycrystalline Cu were observed and discussed in terms of size effect.
Since the discovery of carbon nanotubes, there have been
intensive studies concerning the preparation, characterization,
and application of this fascinating material.1-4 Great interest in
finding a way to use this material arises worldwide because of
its superior properties. This includes using carbon nanotubes
as nanodevices, gas absorbents, one-dimensional quantum wire,
and nonlinear optical limiter among others.5-8 In this study,
carbon nanotubes were employed as templates to synthesize
small-sized Cu which has been found to exhibit unusual
properties as compared with the bulk material.9,10
Normally, Cu nanoparticles are prepared by wet chemistry
methods which can provide Cu particles as small as several
nanometers but may suffer from the uncontrollable size distribu-
tion.11 An alternative preparation route is pulsed laser ablation,
in which a Cu target is evaporated in an inert gas atmosphere,
but for effectiveness several parameters need to be carefully
controlled.12 The method reported in this paper provides an easy
and controllable way to synthesize nanoparticles. It also has an
advantage in separating Cu nanoparticles from the template.
Furthermore, a new form of Cu fibers with diameter ranging
from 100 nm to several µm can be synthesized using the same
method by increasing the amount of Cu added.
It has been found in our previous study that a size effect may
manifest itself in the variation in the core level binding energy
and in the work function.13 In this study, XPS and UPS were
applied to investigate and compare the electronic structure of
Cu nanoparticles, Cu fibers, and polycrystalline Cu metal. As
will be reported below, the results reveal that a similar size effect
does exist among these three kinds of Cu.
Experimental Section
The carbon nanotubes used in this study were prepared by
the decomposition of CH4 on a NixMg1-xO catalyst as described
previously.13,14 By changing catalyst composition, this method
enabled us to produce carbon nanotubes with narrow and
specific size distributions. When x ) 0.3, approximately 90%
(as measured from careful TEM inspection) of the tubes have
diameters between 5 and 10 nm, while 80% of the tubes are
* To whom correspondence should be addressed: telephone, 0065-
8742616; fax, 0065-7776216; e-mail, phylinjy@nus.edu.sg.
10.1021/jp983983j CCC: $18.00 © 1999 American Chemical Society
Published on Web 05/13/1999

4560 J. Phys. Chem. B, Vol. 103, No. 22, 1999
Letters
Figure 2. TEM image of Cu nanoparticles prepared under optimum
Figure 1. TEM image of Cu nanoparticles.
conditions.
between 10 and 20 nm when x ) 0.4, and 80% are between 25
and 35 nm when x ) 0.5. Little (<10 wt %) nonnanotube
material such as metal (catalyst) particles, amorphous carbon,
disordered carbon etc. was observed from TEM and thermo-
gravimetry (TG) study of high temperature hydrogen etching.
The preparation procedure for Cu nanoparticles and fibers
involved fully dispersing a certain amount of carbon nanotubes
in water, followed by adding a desired amount of a Cu salt
(normally 2 molar % for Cu nanoparticles, and 100 molar %,
i.e., 1:1 Cu to carbon nanotubes molar ratio, for Cu fibers) to
the suspension. The mixture was then strongly stirred at 373 K
until the solvent was vaporized completely, and finally it was
subjected to decomposition and reduction with H2 at tempera-
tures below 773 K.
Cu samples for TEM observation were dispersed in water
and subjected to ultrasonic treatment. A little drop of the
suspension was then deposited onto carbon TEM grids. TEM
and SEM observations were conducted on a JEM-100CX
electron microscope; XRD measurement was performed on a
Philips PW 1710 diffractometer. Both XPS and UPS measure-
ments were performed on a VG ESCALAB Mk II machine at
a vacuum better than 3 × 10-9 Torr. To avoid charging, samples
for XPS/UPS measurements were first dispersed in acetone,
dropped to the top of sample stubs, and then dried to secure a
good contact with the sample stub. Samples were pretreated in
purified H2 at 573 K for 4 h to remove oxygen followed by Ar
ion sputtering to ensure that the sample surface was free of
oxygen and other impurities. The analyzer pass energy used in
the XPS measurement was 5.0 eV, and the scan step was
selected at 0.025 eV to improve resolution. The work function
was measured by UPS using He(I) (21.2 eV) as an ionization
source and a 10 eV bias was applied to make a clear cutoff of
the secondary electron tail.
Figure 3. SEM image of Cu fibers.
diameter. Furthermore, the preparation conditions including the
amount of Cu salt added and the decomposition/reduction
temperature can also play important roles in controlling the size
of Cu particles. By reducing the amount of the Cu salt from 2
molar % to <1 molar % while keeping other conditions
identical, Cu nanoparticles synthesized have smaller size, 70%
in the range of 5-10 nm. Also, if the decomposition/reduction
temperature is high, close to the melting point of Cu, the size
of Cu particles will be large, since Cu atoms at such high
temperatures are prone to merge into large particles to reduce
surface energy. Figure 2 shows Cu particles prepared under
optimum conditions with diameters less than 5 nm. The TEM
image also shows that the sticking of the Cu particles to the
carbon nanotubes is not strong, with a large number of particles
being dropped off from the tube surface and spread on the TEM
grids after ultrasonic treatment. This provides an easy way for
separating Cu particles from carbon nanotubes.
When the amount of the Cu salt added to carbon nanotubes
was increased (from 2 molar % as mentioned above for
nanoparticles) to 1:1 molar ratio, a spongelike yellow-brown
product was obtained following the procedure described in the
Experimental Section. The SEM image in Figure 3 reveals that
this spongelike Cu is actually the piling up of microsized Cu
fibers with diameters ranging from 100 nm to 5 µm, and lengths
up to hundreds of µm. XRD measurement can hardly detect
the graphitelike diffraction peaks, which are regularly detected
from carbon nanotubes prior to Cu coating.13,15 Only diffraction
Results
Figure 1 is the TEM image of the Cu nanoparticles attached
on the carbon nanotubes, which was prepared from 2 molar %
Cu on the carbon nanotubes with diameters of 25-35 nm. It
can be seen that the Cu particles are in the form of round dots
stuck on the surface of carbon nanotubes, with diameter less
than that of carbon nanotubes, i.e., less than 35 nm. It has been
found that the size of Cu particles can be controlled by the
templates: the smaller the diameter of carbon nanotubes
template, the smaller the size of Cu particles prepared. When
carbon nanotubes with diameter of 5∼10 nm were used, all of
the Cu particles synthesized were found to be under 10 nm in

Letters
J. Phys. Chem. B, Vol. 103, No. 22, 1999 4561
Figure 5. UPS (He(I)) spectra of (a) polycrystalline Cu and (b) Cu
Figure 4. XPS core-level spectra of: (a) polycrystalline Cu, (b) Cu
fibers.
fibers, and (c) Cu nanoparticles.
that which has been previously reported, but 4.4 eV is obtained
for the Cu fibers giving an increase of 0.5 eV. The increase in
work function means the energy barrier for an electron in the
solid to escape to the vacuum is enhanced. Since low-index
surfaces normally possess higher work functions than do high-
index surfaces of the same identity, it may be inferred that
microsized Cu fibers expose mainly low-index surfaces and have
therefore fewer defects as compared with polycrystalline Cu
metal.
of Cu metal is observable (spectrum not shown), indicating a
thick coating of Cu on carbon nanotubes. The diameter of the
Cu fibers can be controlled by changing the amount of Cu added.
Nanosized Cu fibers with diameters around 100 nm can be
obtained if the Cu salt added is less than 40 molar %. In this
case Cu nanoparticles would coexist.
Figure 4 displays the XPS core-level spectra in the Cu 2p3/2
energy region obtained from Cu nanoparticles, Cu fibers, and
polycrystalline Cu. It can be found that the binding energy of
Cu (2p3/2) for the copper powder stays at the normal value of
932.6 eV, but it shifts to 932.3 eV for the Cu nanoparticles. A
slight negative shift for that of Cu fiber can also be observed.
It is interesting to note that our previous study concerning the
difference in electronic structure between carbon nanotubes and
graphite also came to the same conclusion, i.e., as the particle
size reduces to nanoscale, the core-level binding energy shifts
to smaller values as compared with that of bulk one.
The electronic structure of the valence band of these three
Cu materials was measured by UPS. As shown in Figure 1, Cu
nanoparticles coexist with carbon nanotubes, so that their
valence band spectrum shows characteristics both of Cu and
the dominant carbon. The work function measured is also close
to that of carbon nanotubes. As for Cu fibers, the carbon
nanotubes have been shown by SEM and XRD observations to
be fully coated by Cu, consequently the UPS spectrum obtained
from the Cu fibers is characteristic of metallic Cu, different from
that of carbon nanotubes. It can be found (see Figure 5) that
there is a notable change in the energy position of secondary
electron tails for Cu fibers and polycrystalline Cu. As the Fermi
edges of the Cu fibers and polycrystalline Cu coincide at the
same point, the change in the position of the secondary electron
tail reflects the variation of the work function. The work function
for polycrystalline Cu is measured to be 3.9 eV, the same as
References and Notes
(1) Iijima, S. Nature 1991, 354, 56-58.
(2) Hamada, N.; Sawada, S.; Oshiyama, A. Phys. ReV. Lett. 1992, 68,
1579-1581.
(3) Ajiayan, P. M.; Iijima, S.; Ichihashi, T. Phys. ReV. B 1993, 47,
6859-6862.
(4) Jishi, R. A.; Dresselhaus, M. S.; Dresselhaus, G. Phys. ReV. B 1993,
47, 16671-16674.
(5) Collins, P. G.; Zettl, A.; Bando, H.; Thess, A.; Smalley, R. E.
Science 1997, 278, 100-103.
(6) Tans, S. J.; Devoret, M. H.; Dai, H.; Thess, A.; Smalley, R. E.;
Geerligs, L. J.; Dekker, C. Nature (London) 1997, 386, 474-477.
(7) Dai, H.; Wong, E. W.; Lu, Y. Z.; Fan, S.; Lieber, C. M. Nature
1995, 375, 769-772.
(8) Dillion, A. C.; Jones, K. M.; Bekkedahl, T. A.; Kiang, C. H.;
Bethune, D. S.; Heben, M. J. Nature 1997, 387, 377-379.
(9) Wolf, D.; Yip, S. Materials Interfaces; Chapman & Hall: London,
1992; Part 2.
(10) Jang, J. S. C.; Koch, C. C. Scr. Metall. Mater. 1990, 24, 1675-
1678.
(11) Suryanarayanan, R.; Frey, C. A.; Sastry, M. L.; Waller, B. E.;
Buhro, W. E. J. Mater. Res. 1996, 11, 449-457.
(12) Pa´szti, Z.; Horva´th, Z. E.; Peto˜, G.; Karacs, A.; Guczi, L. Appl.
Surf. Sci. 1997, 109/110, 67-73.
(13) Chen, P.; Wu, X.; Lin, J.; Tan, K. L. Phys. ReV. Lett. 1999, 82,
2548-2551.
(14) Chen, P.; Zhang, H. B.; Lin, G. D.; Hong, Q.; Tsai, K. R. Carbon
1997, 35, 1495-1501.
(15) Chen, P.; Wu, X.; Lin, J.; Li, H.; Tan, K. L. Carbon, in press.