© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
p s s
basic solid state physics
b
st
at
us
so
lid
i
www.pss-b.comph
ys
ic
a
Wide-range optical spectra of
carbon nanotubes: a comparative
study
K. Kamara´s1,*, ´A. Pekker1, M. Bruckner1, F. Borondics1, A. G. Rinzler2, D. B. Tanner2, M. E. Itkis3,
R. C. Haddon3, Y. Tan4, and D. E. Resasco4
1 Research Institute for Solid State Physics and Optics, Hungarian Academy of Sciences, P.O. Box 49, 1525 Budapest, Hungary
2 Department of Physics, University of Florida, Gainesville, FL 32611, USA
3 Center for Nanoscale Science and Engineering, Departments of Chemistry and Chemical and Environmental Engineering, University
of California, Riverside, CA 92521-0403, USA
4 Department of Chemical, Biological, and Materials Engineering, University of Oklahoma, 100 East Boyd St, Norman, OK 73019, USA
Received 15 May 2008, accepted 27 June 2008
Published online 3 September 2008
PACS 78.30.Na, 78.40.Ri
∗ Corresponding author: e-mail kamaras@szfki.hu
We present optical spectra from the far infrared through
the ultraviolet spectral region of freestanding transparent
carbon nanotube films: single-, double- and multiwalled,
and containing varying amounts of tubes of different chi-
rality. By comparing the spectral features in the far in-
frared and the near infrared/visible, we can estimate the
metallic/semiconducting content. We show by spectro-
scopic methods that doping the tubes increases both the
conductivity and the transparency for visible light.We
also discuss the influence of sample preparation and sub-
sequent treatment on application possibilities.
1 Introduction Optical spectroscopy is one of the
most widely used methods in characterization of carbon
nanotubes. Recently, it was demonstrated that absorption
and especially fluorescence studies can detect individual
nanotubes and identify them by chirality index [1]. Most
studies concentrated on the NIR/VIS spectral range where
transitions between Van Hove singularities occur: based on
these observations, selectivity by semiconducting/metallic
character was reported both for ionic doping [2,3] and co-
valent functionalization [4]. Relatively less attention was
devoted to the low-frequency part of the spectrum. Here we
want to emphasize the importance of far-infrared measure-
ments as a sensitive indicator of intrinsic charge carriers in
metallic tubes and extrinsic carriers in doped materials.
2 Experimental Six types of tubes were used for this
study: four were single-walled, one double-walled and one
multiwalled. Single-walled laser-ablated nanotubes were
produced by Tubes@Rice, arc-derived ones by Carbon So-
lutions, Inc. HiPCO tubes were commercial samples from
CNI Nanotechnologies and two types of CoMoCat sam-
ples, commercial grade and research grade, from South-
west NanoTechnologies. Based on the average diameter,
we divide the single-walled samples into small- and large-
diameter groups (see Table 1). The width of the diameter
distribution, though, is also varying among the families:
HiPCO has the widest distribution, and the two CoMoCat
samples differ in the content of metallic tubes, SG grade
having a narrower distribution and consisting almost ex-
clusively of semiconducting tubes. Double- and multiwall
tubes were commercial products by Nanocyl SA.
All measurements were done on freestanding films,
prepared according to Ref. [5] from nanotube suspensions
in Triton-X, filtered on a mixed cellulose ester filter which
was subsequently dissolved in acetone and the nanotube
films captured on graphite frames. Films of thickness 300-
400 nm were produced in this way, still thin enough to be
transparent throughout the whole spectral range. The films
were annealed in vacuum at 200 ◦C for two hours before
measurement; this step is important in order to remove ac-
cidental doping by the chemicals used for purification [6]
© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
phys. stat. sol. (b) 245, No. 10, 2229–2232 (2008) / DOI10.1002/pssb.200879647

2230 K. Kamarás et al.: Wide-range spectra of nanotubes
p
h
y
s
i
c
a
p s s
s
t
a
t
u
s
s
o
l
i
d
i
b

© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com
Table 1 Summary of the samples investigated
Sample Origin Average
diameter (nm)
Arc Carbon Solutions 1.4
Laser Tubes@Rice 1.3
HiPCO CNI Nanotechnologies 0.7
CoMoCat CG Southwest Nanotechnologies 1.0± 0.3
CoMoCat SG Southwest Nanotechnologies 0.7
Double-Walled Nanocyl 2 -3 (outer)
Multi-Walled Nanocyl
and thereby allowing the comparison of pristine nanotubes
without doping effects.
Several spectrometers were used, covering the full op-
tical range from the far infrared through the ultraviolet:
two Fourier-transform interferometers (Bruker 66v/S in the
FIR/MIR, Bruker Tensor 28 in the MIR/NIR) and a JASCO
grating instrument as well as an Ocean Optics fiber-optic
spectrometer in the VIS/UV. Spectra were taken in trans-
mission mode. Doping studies were conducted in a 10 cm
gas cell with KBr windows. We converted the transmission
to optical density (OD) by simply taking−logT . This pro-
cedure neglects reflectance corrections and is therefore not
fully quantitative, especially in the far infrared [7], how-
ever, the resulting OD value is monotonouswith concentra-
tion, albeit not linear. This means that although we cannot
use this procedure for estimating the metallic to semicon-
ducting tube ratio, we can establish an order of the sam-
ples according to both dc conductivity and IR/VIS trans-
mittance. A more detailed analysis of the optical constants,
combining spectroscopy with thickness measurements by
atomic force microscopy, will be presented in a forthcom-
ing publication.
3 Results and discussion
3.1 Undoped films Figure 1 shows the optical den-
sity of all six samples investigated. In single-walled tubes,
three typical features can be distinguished: 1. a low-
frequency absorption, either peaked around 100 cm−1
or increasing smoothly towards zero; 2. distinct peaks in
the NIR/VIS range, corresponding to transitions between
Van Hove singularities in the density of states; 3. a broad
absorption peaking in the UV, caused by π → π∗ excita-
tions of the entire π-electron system. Between regions 1
and 2 the absorption is low and this region constitutes the
transparency window, important for applications.
It is apparent by comparing Figs. 1a and 1b that the
small-diameter tubes used in this study consist of fewer
types of nanotubes than the large-diameter ones, and that
probably the metal/semiconducting ratio is also different
between these samples. The structured appearance of the
S
11
transition in HiPCO as well as the extreme small width
of this peak in scientific grade CoMoCat indicates that the
distribution is not continous as in the laser and arc samples.
Figure 1 Optical density of three types of carbon nanotubes:
a) large diameter (arc and laser), b) small diameter (HiPCO and
CoMoCat) and c) double-walled and multiwalled. Spectra are
scaled at the π → π∗ peak.
Moreover, the difference between commercial and scien-
tific grade CoMoCat samples manifests itself in both the
narrowing of the S
11
transition and the decrease in intensity
of the low-frequency peak, indicating that the SG material
contains not only a narrower distribution of semiconduct-
ing tubes, but the ratio of semiconducting to metallic tubes
has also increased. Estimations of metal content in CoMo-
Cat tubes vary considerably in the literature from 10 per
cent [8] to 25 per cent [9]; according to the data presented
here, the metal content of the CG type is close to HiPCO,
whereas that of the SG type is less.
All three small-diameter samples show a distinct peak
around 100 cm−1, in contrast to the large-diameter types,