High Resolution TEM Observations
of Single-Walled Carbon Nanotubes
Tara Spires and R. Malcolm Brown, Jr.
Department of Botany, The University of Texas at Austin, Austin,Tx., 78713
Single-walled carbon nanotubes (SWNTs) kindly donated by Dr. Richard Smalley (Rice University, Houston) were imaged in high resolution mode (.335nm) using a Philips 420 transmission electron microscope (TEM). Digital images were captured at the atomic and molecular levels, demonstrating structural characteristics of SWNTs never before observed. As shown by the Smalley lab, the lengths of the SWNTs is so great that it is rare to find true ends; however, in several instances, views of individual nanotube tips reveal a tripartate and tetrapartate segmentation. A growing bundle of single-walled tubes was visualized. Linear chains of acetylenic carbon appear to be directly attached to the tips of nanotubes. These are apparently coated with metal atoms which could be the nickel/cobalt catalyst. An example of nanotube curvature of approximately 90 degrees was shown in which no visible tube distortion was revealed. This opens to suggestion that SWNT are even more flexible than previously suggested. New ideas about how SWNT's are assembled from linear acetylenic carbon chains are presented.
Since their discovery in 1991,1 single and multi-walled nanotubes of carbon have been extensively investigated.2,3 Based on scanning-tunneling microscopy (STM) images and electron diffraction studies,4,5 single-walled nanotubes (SWNTs) are proposed to consist of a seamless cylinder of a graphitic sheet capped by hemispherical ends composed of pentagons and hexagons. Curves observed in high-resolution transmission electron microscope (HRTEM) images of SWNTs indicate the single-walled tubes are more pliable than their multi-walled counterparts.6 New growth mechanisms of SWNTs have been proposed which differ from the present concepts. One school of thought states that the tube formation occurs from capped ends in which C2 is absorbed near pentagonal defects in the cap causing elongation. 2 The other school proposes that tubes grow only by addition of carbon clusters to the structures at open tips which are prevented from collapsing into the more stable hemisphere structure by metal catalyst chemisorption or electric field stabilization of the dangling bonds at the open ends. 3,7
Using the novel instrumentation with our Philips 420 TEM we have been investigating at the atomic
and molecular level of resolution (less than 0.335nm) the structure of single-walled carbon nanotubes. In particular, we have
studied the atomic co-ordinates of carbon within the walls of SWNTs. In addition, we report on the ability of the tubes to curve
at sharp angles. We also have observed heretofore never described internal structure within the SWNT tips.
Finally, we discuss the implications of our findings with respect to a more complete understanding of the mechanism of formation
of SWNTs.
Grid Preparation: A sample of single-walled nanotubes of carbon provided by the Smalley lab at Rice
University was dissolved in methanol and briefly sonicated. One drop of the suspension was placed on a 300 mesh
copper TEM grid coated with Formvar (=polyvinyl formal) from a 0.3% solution in dichloroethane. After approximately one minute,
excess liquid was removed by touching one edge of the grid to Whatman 1 filter paper.
TEM Observations: Operating voltages of 100kV and 120kV were used on the Philips 420 transmission
electron microscope to observe the specimen. Observations were made with a spot size setting of "1" for C1 and a maximum
intensity at crossover for C2. We used a condenser aperture of 100mm, an objective aperture of 70mm, no diffraction aperture,
and magnifications ranging from 160,000X to 750,000X on the TEM. Average beam current was 1-2 microamps. The TEM was
routinely calibrated to better than 3.35Å
resolution by focusing and stigmating on a graphite lattice.
Image Analysis: Images were sent from a GATAN camera to an IBAS digital image processing system where
they were digitized and stored for analysis. One unique feature of our system is the YAG crystal upon which the electron
image falls. This grainless detector converts the electron image into a photon image which is captured on a low light level
Newvicon camera using a special fiber optic interface. Final image magnifications of the digital images ranged from 6,912,000X
to more than 32,700,000X. These high magnifications have been employed to overcome any possible instrumental limitations
inherent in the television camera scanning. Furthermore, atomic distances less than 0.2nm can easily be measured on the
high resolution monitor and recorded. Fast Fourier Transforms (fft's) were used on indicated images to enhance traces of periodicity
detected in structures. An fft was run on several different images of nanotubes and one background image of the formvar film
support captured at the same magnification to ensure that an artifact was not mistaken for periodic structure.
Length Calculations: Measurements of nanotube widths were determined based on calibrations using images
of a graphite lattice with 3.35Å spacing between layers. The length (in centimeters) of 3.35Å was measured on the GATAN
monitor and IBAS imaging system monitor at each applicable magnification. Measurements of the widths of nanotubes
(in centimeters) were obtained on the same monitors and the actual width in Angstroms was calculated using simple
ratios from the 3.35Å calibrations. Measurements of individual units of substructure were obtained by measuring
the width of nanotubes and the diameter lengths of the units (in centimeters) on the IBAS imaging screen then calculating
the measurements of the substructures in Angstroms using the relationship of calculated tube width in Angstroms
to measured tube width on the screen.
Tube Curvature:
Tube curvature is viewed in figure 1 without visible distortion of the nanotube, and
although SWNTs are pliable,6 an angle of this magnitude is expected to
cause buckling in the tube.
Graphitic Lattice: Averages of ten measurements of the width of
nanotubes from three different images provided an average diameter of SWNTs of 1.2nm. Average diameter
of the periodic subdivisions enhanced by fft was 2.3Å.(fig. 2) The average diameter of the
subunits is within the range of 1.4Å to 2.8Å which are the minimum and maximum lengths observed in a hexagon
of sp2 bound carbon atoms in a graphitic lattice. (see fig. 3)
Tube Growth: In figure 4, a growing bundle of SWNTs is observed.
The curved tube on the left is coated with an electron diffracting substance making it darker. The close packing
arrangement of the SWNTs is also observed. A dark chain of carbon, seen more clearly in figure 5, is
attached to the tip of one of the tubes in the bundle. Figure 6 is another possible demonstration
of a chain of carbon adding to growing nanotubes.
Structure of SWNTs: The ability of single-walled carbon nanotubes to bend at the extreme angle observed in figure 1 opened the possibility that the proposed structure of SWNTs was not accurate since it is suspicious that a cylinder composed of a closed graphitic sheet could bend that far without visible damage to the tube.
Opposing the theory of the inaccuracy of the current structural model was the periodic structure found in images we captured at high magnification. This substructure resembles the hexagonal lattice predicted for SWNT structure.2,3,4 Measurements of the subunits of the periodic structure show that it is within the 1.4Å to 2.8Å range of the length of hexagons in a graphitic lattice. Deviation from the exact values for lengths of a hexagon as described in figure 2 is due to distortion of the observed periodic structure by viewing the sides of SWNTs where the hexagons are curving away from the electron beam. Electron interactions with the lattice on the opposite side of the tube may also affect the observed periodic structure.
Nanotube Interactions: The close packing arrangement of single-walled nanotubes in bundles due to van der Waals interactions is also demonstrated in figure 4. The packing can be observed as a diamond lattice or hexagonal close packing.
An Alternate Hypothesis to Nanotube Formation: The image captured of the growing bundle of SWNTs
(fig. 4) reveals several interesting properties of the nanotubes. First, the curved tube on the
edge of the bundle is more electron-dense than the remainder of the bundle indicating the presence of an electron
diffracting substance coating the tube. Clusters of single-walled tubes are observed to terminate at the same
time,3, and it is proposed that this substance coating the darker tube either is
the metal catalyst or acetylenic carbon chains that are associated with this tube. Shown in figure 5
which is a zoomup of figure 4, a linear acetylenic chain of carbon as described
by Lagow et al.9 is observed confirming the presence of chains that have been
used to explain nanotubes as field emitters.7 The carbon chain appears to be
coated in paracrystallized atoms of the metal catalyst making it dark and wider than expected. Chains such as this
may be a source of carbon for the growing tips of nanotubes as they "zipper" to become part of the tube structure.
We propose the possibility of generation of SWNTs not just as a single layer at a time, but a number of linear
acetylenic carbon chains integrating simultaneously into the growing end as suggested in figures 4 and 5.
Figure 7 shows a possible mechanism by which linear acetelynic carbon chains may
add to a growing SWNT. This mechanism assumes the existance of triple bonds between carbon atoms on the
top layer of the chair confirmation in a growing SWNT.8
A honeycomb graphitic lattice structure of single-walled carbon nanotubes has been obtained for the first time using HRTEM, and a chain of linear acetylinic carbon9 was observed attached to the end of a growing nanotube. HRTEM offers a convenient and useful tool in order to more fully understand the structure and growth mechanisms of SWNTs.
We would like to thank Dr. Richard E. Smalley and Dr. Andrew Thess for their generous donation of the sample of single-walled carbon nanotubes used in this study. This work is supported in part by Welch Foundation Grant F1217 to R.M.B. and the Office of Naval Research grant N00014-95-1-0933 (Dr. Randall Alberte, Program Director).
If you have any questions, please email them to us at rmbrown
@mail.utexas.edu.
Up to Posters Page