Size analysis of single fullerene molecules by electron microscopy

Abstract

Individual fullerene molecules were observed and sized using high resolution transmission electron microscopy. Fullerene C 60 molecules were tethered by chemical bonding to carbon black particles to facilitate HRTEM imaging and sizing of known fullerenes. HRTEM analysis of soot samples from a fullerene-forming flame revealed the presence of a range of fullerenes from C 36 to C 176 and larger fullerene-like structures. The observation of fullerenes smaller than C 60 is noteworthy in that such structures necessarily contain adjacent pentagons and hence are strained and expected to have interesting reactivity that may be useful in certain applications. HRTEM can be used to detect and size fullerenes in samples containing fullerenes but not in sufficient quantity, or not sufficiently removable, to be detectable by chemical analysis.

Full text

Page 1

Size analysis of single fullerene molecules by electron microscopy
Anish Goela,*, Jack B. Howarda,*, John B. Vander Sandeb
aDepartment of Chemical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139-4307, USA
bDepartment of Materials Science and Engineering, Massachusetts Institute of Technology 77 Massachusetts Avenue, Cambridge,
MA 02139-4307, USA
Received 8 August 2003; accepted 10 March 2004
Available online 24 April 2004
Abstract
Individual fullerene molecules were observed and sized using high resolution transmission electron microscopy. Fullerene C60
molecules were tethered by chemical bonding to carbon black particles to facilitate HRTEM imaging and sizing of known fullerenes.
HRTEM analysis of soot samples from a fullerene-forming flame revealed the presence of a range of fullerenes from C36to C176and
larger fullerene-like structures. The observation of fullerenes smaller than C60 is noteworthy in that such structures necessarily
contain adjacent pentagons and hence are strained and expected to have interesting reactivity that may be useful in certain
applications. HRTEM can be used to detect and size fullerenes in samples containing fullerenes but not in sufficient quantity, or not
sufficiently removable, to be detectable by chemical analysis.
? 2004 Elsevier Ltd. All rights reserved.
Keywords: A. Carbon black, Fullerene, Soot; C. Electron microscopy; D. Microstructure
1. Introduction
An ability to observe and measure the size of an
individual fullerene molecule by high resolution trans-
mission electron microscopy (HRTEM) is of interest for
several reasons. Such a technique would offer a possible
means for extending the detection and analysis of
fullerenes to lower limits of detection than can be at-
tained by conventional chemical analysis. For example,
circular objects approximately the size of C60 and C70
fullerenes can be seen in HRTEM images of soot not
only from certain low-pressure benzene/oxygen flames
well known to contain fullerenes but also from atmo-
spheric-pressure ethylene/air flames in which fullerenes
could not be detected by state-of-the-art chemical
analysis [1].
The HRTEM technique would be useful also for the
detection and characterization of fullerenes which are
too strongly bound to, or within, the material with
which they are condensed in the synthesis process to
permit easy removal for chemical analysis. Such species
include fullerenes smaller than C60, all of which neces-
sarily contain adjacent pentagons in their structure and
are strongly curved and strained, and hence interactive,
and much larger fullerenes whose size facilitates exten-
sive contact and increases the opportunity for bonding
interactions. Curved carbon structures seen in HRTEM
images of soot from fullerene-forming flames exhibit
radii of curvature which include not only values com-
parable to those of the prevalent fullerenes C60and C70
but also smaller and larger values [1–3]. Carbon struc-
tures with radii of curvature less than that of C60may be
indicative of adjacent pentagons, the presence of which
in carbon layers would be of sufficient practical and
scientific interest to merit confirmation and further
study by more quantitative HRTEM analysis.
Mass spectra which show the presence of C60, C70and
higher fullerenes in the vapor from graphite vaporiza-
tion [4] and in low-pressure flames of certain hydrocar-
bons [5] also exhibit mass peaks corresponding to C50
and other carbon species lighter than C60. Under some
flame conditions the apparent abundance of C50is even
comparable to that of C60and C70[5,6]. However, C50
fullerene has never been found in chemical analysis of
condensed material from fullerene-containing flames or
carbon vaporization systems, possibly reflecting the
*Corresponding authors. Tel.: +1-202-736-4256; fax: +1-202-736-
4259 (A. Goel), Tel.: +1-617-253-4574; fax: +1-617-324-0066 (J.B.
Howard).
E-mail addresses: anish@mit.edu (A. Goel), jbhoward@mit.edu
(J.B. Howard).
0008-6223/$ - see front matter ? 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2004.03.022
Carbon 42 (2004) 1907–1915
www.elsevier.com/locate/carbon

Page 2

difficulty of extracting this compound from the soot and
other components of the condensate. The structure of
C50is expected to contain five pairs of adjacent penta-
gons and hence to be highly strained and strongly at-
tracted to the other molecules, including C50, and
surfaces in the condensed material. Fullerene C36which,
even more than C50, is highly strained and prone to
forming strong bonds with the condensed material, has
been identified in material condensed from a fullerene-
forming graphite vaporization process using fraction-
ation and analysis methods especially designed to deal
with the strong bonding [7]. The smallest possible ful-
lerene, C20, consisting solely of 12 pentagons, has been
generated from a brominated hydrocarbon, dodeca-
hedrane, by gas-phase debromination [8]. The lifetime of
the C20cage under the experimental conditions used was
at least 0.4 ms. Owing to the extreme curvature and
reactivity of this structure, it would not be expected to
attain substantial gas-phase concentrations in most
methods of fullerene synthesis. Nevertheless, it is con-
ceivable that the C20cage might be stabilized for longer
lifetimes through strong bonding with soot or other
condensed species. A HRTEM capability to observe
directly in the condensed phase C50, C36, C20and other
highly interactive fullerenes, both smaller and much
larger then C60, and to contribute to their character-
ization would clearly be useful.
The observation of single fullerene molecules by
HRTEM has been addressed previously. Cox et al. [9]
deposited C60 on MgO crystals supported on holey
carbon films. Circular contrast patterns with about 0.8
nm diameter, consistent with that of C60, were observed
on the MgO crystals and could be seen most clearly on
the edges of crystals hanging over holes in the support
film. The circular images were not seen on MgO crystals
without C60deposition. Cox et al. [9] regarded it to be
improbable that the circular patterns could represent
two or more C60molecules aligned parallel to the elec-
tron beam axis. The strong inference is that the observed
circular patterns were images of individual C60 cages.
Ajayan and Iijima [10] observed with HRTEM a 0.8 nm
diameter circular image at the outside edge of the image
of a multilayer truncated conical carbon nanostructure.
The authors point out that the image could correspond
to a single C60structure but they argue against such an
interpretation on the grounds that the contrast exhibited
by the small circular image is comparable to that of the
lattice fringes of the much larger supported multilayered
structure. Instead, they interpret the circular image to be
an end-on view of a nanotube having the same diameter
as a C60molecule and a length of a few such units in the
direction perpendicular to the plane of the image. Saito
et al. [11] present a HRTEM image of a round, hollow
single-layered particle of 1.3 nm diameter, which they
suggest may be a fullerene having approximately 200
carbon atoms, held by a single-walled carbon nanotube
of approximately the same diameter. Based on this
observation the authors suggest it should be possible,
although they had not been able yet, to observe by TEM
more prevalent fullerenes such as C60, C70, and C84
suspended on the tip or side of a carbon nanotube.
Fuller and Banhart [12] using HRTEM observed not
only fullerene cages but also the formation of the cages
during electron irradiation of carbon specimens in the
electron microscope. They observed tube-like structures
as well as spherical cages, and saw that tube-like struc-
tures viewed end-on resembled, but gave much more
contrast than, spherical cages. In a HRTEM study of
fullerenic nanostructures from fullerene producing
flames, Das Chowdhury et al. [13] observed on a mul-
tiwalled carbon nanotube an ellipsoidal structure having
a single closed-shell whose major and minor diameters
were about 1.1 and 0.7 nm, corresponding approxi-
mately to a C94 fullerene molecule. C94 fullerene has
been isolated since and identified by solvent extraction
and HPLC analysis of condensable material collected
from flames [14]. Burden and Hutchison [15] using
HRTEM observed features consistent with fullerenes
C60 and C70 on the surface of carbon black particles
irradiated with an electron beam in the presence of he-
lium. Smith et al. [16] and Sloan et al. [17] obtained
HRTEM images of round, closed shells of carbon,
apparently fullerenes, encapsulated within the hollow
central region of individual single-wall carbon nanotu-
bes produced in carbon vaporization systems. The
diameters of the apparent fullerenes correspond to C60
[16,17] and various other fullerenes both smaller and
larger than C60, ranging from 0.4 to 1.6 nm [17]. The
estimated precision in the diameter measurements of
Sloan et al. [17] was ±0.05 nm.
Although previous studies (mentioned above) provide
strong evidence that individual fullerene molecules can
be observed using HRTEM, such capability has been
questioned [10]. There is need for unequivocal demon-
stration of the capability of HRTEM for observing and
sizing fullerene molecules on or within condensed car-
bon material in order to establish the viability of the
technique for use in the detection and characterization
of fullerenes which are below the detection limit, or are
difficult to transfer to the gas or liquid phase, for
chemical analysis. Furthermore, there is need to apply
the capability, once it is established, to the study of
bound fullerenes in combustion-generated material,
which could be produced in whatever large quantities
might eventually be demanded by practical applications.
To these ends, the present study (1) tests the capability
of HRTEM for observing individual C60fullerene mol-
ecules added in a controlled manner to a sample of
carbon material, (2) calibrates the HRTEM measure-
ment of fullerene diameters against a known standard,
and (3) uses the established HRTEM technique to
identify and size interesting flame-generated fullerenes
1908
A. Goel et al. / Carbon 42 (2004) 1907–1915

Page 3

that cannot be removed easily from the other condensed
material for conventional chemical analysis.
2. Experimental
Three different samples were prepared for investiga-
tion by HRTEM. The first sample was pure carbon
black (Regal 330; Cabot Corp.) suspended in a toluene
solution. The other two samples were prepared from a
portion of the carbon black–toluene solution to which
wasaddedaspecified
hano[60]fullerene (obtained from Professor Mark S.
Meier, the University of Kentucky). In this functional-
ized fullerene, the carbon atom of the functional group
is bridged to two carbon atoms of the fullerene mole-
cule. It was expected, based on known behavior of the
similar dibromomethano[60]fullerene [18], that the two
chlorine atoms could be substituted readily to provide a
chemical tether to a compound of interest. After a uni-
form dispersion was ensured with vigorous mixing, the
toluene was allowed to evaporate and the resulting dry
powder mixture was sealed inside an argon-filled glass
tube. The entire unit was then heat treated at approxi-
mately 400 ?C for 4.5 h in a tubular furnace (Lindberg
Model 55036) and then cooled. The material was re-
moved from the tube and divided into two parts. One of
these two samples was not treated further and hence
consisted of carbon black with tethered fullerenes and
any fullerenes that remained untethered. This sample is
referred to as pre-extraction. The other of these two
samples was extracted by sonication in toluene for 13
min followed by vacuum filtration with a 0.45-lm nylon
filter to remove any untethered fullerenes. Thus, this
sample, referred to as post-extraction, consisted of car-
bon black with only tethered fullerenes.
A diluted suspension of each of the three samples in
toluene was deposited onto a lacey carbon grid and the
toluene was allowed to evaporate. The samples were
analyzed in a JEOL 2010 electron microscope operating
at 200 kV. The images obtained were analyzed for the
presence of fullerene-type structures, i.e., structures that
appear to be completely closed cages. In each image, the
number of fullerene-type structures per length of peri-
meter, referred to as linear concentration, was deter-
amountof dichloromet-
mined and the diameter of each of those structures was
measured. The data then were aggregated across all the
images of a particular sample to provide fullerene linear
concentration data and fullerene size distribution data.
Additional HRTEM analyses were performed on
soot material collected from a pre-mixed benzene/oxy-
gen/argon flame that has been extensively characterized
and studied previously [1,19]. The conditions of this
flame are: pressure, 40 Torr; gas velocity at burner, 25
cm/s (25 ?C); fuel equivalence ratio, 2.4 (atomic C/O
ratio, 0.96); and percentage diluent in feed gas, 10%
argon. Samples of soot and all other condensables from
this flame were collected in the manner described pre-
viously [1] and HRTEM analysis was done using the
same JEOL 2010 operating at 200 kV. Gold islands were
deposited on the surface of several of these samples to
provide a magnification calibration for the HRTEM
images. Gold has a stable planar structure with a con-
stant interplanar spacing of 2.039?A for the {111}
atomic planes. By observing and measuring this known
spacing in an image, the image length scale thus is cal-
ibrated allowing the dimensions of other structures to be
accurately measured. This calibration was used to
measure the sizes of several of the closed-cage structures
that were observed in the images and the data were
compiled across all of the samples into a size distribu-
tion histogram.
Three influences on the accuracy of the electron
microscopy observations have been considered in this
work. First, the influence of the degree of objective lens
defocus (measured in nm with negative numbers being
underfocus and positive numbers being overfocus) has
been investigated. Calculations [20] were performed on
C60 using 200 kV electrons, a spherical aberration
coefficient of 0.5 mm, an incident beam spread of 4 nm
over a defocus range of )35 to +35 nm in 5 nm intervals.
One result from such calculations is shown in Fig. 1.
Note that the only interpretable images are at an un-
derfocus of ?)30 nm yielding an image consisting of a
very dark ‘‘doughnut’’ surrounded by a ‘‘halo’’. The
variation in measured size (using the interface between
the ‘‘doughnut’’ and the ‘‘halo’’) is ±0.02 nm in the
underfocus range of )30±5 nm.
The second source of inaccuracy pursued was that
associated with the practical obtaining of images, i.e.,
Fig. 1. A series of simulated electron microscope images of C60viewed in the [100] direction. This sequence consists of images over an objective lens
defocus of )35 nm (left-most image) to )15 nm (right-most image), bracketing the defocus where interpretable images are produced. The C60image
consists of a dark ‘‘doughnut’’ surrounded by a bright ‘‘halo’’. See text for additional information.
A. Goel et al. / Carbon 42 (2004) 1907–1915
1909

Page 4

the experimental work. Fig. 2 shows a series of images
taken from the same area. The black dashes in Fig. 2
indicate a structure indicative of those measured in this
analysis. The variation of diameters measured is a con-
volution of objective lens defocus and changes in the
object over the time associated with sample radiation
and heating. Observation of images such as those in Fig.
2 lead to the establishment of the accuracy of observa-
tions of ±0.05 nm.
Third, the use of the gold islands as a magnification
calibration leads to a measurement precision of ±0.01
nm. It is clear that the practical aspects of performing
the electron microscopy, as embodied in Fig. 2, limit the
accuracy of observations in this work to ±0.05 nm.
Finally, a sample of C60molecules (99.5% pure; SES
Corporation) was examined under HRTEM. The fulle-
renes were dissolved into toluene and drops of the
solution were placed on TEM grids. The toluene was
allowed to evaporate before the HRTEM analysis.
3. Results
Figs. 3 and 4 show two images that are representative
of the images analyzed from the different carbon black
samples. Fig. 3 is an image of a particle taken from pure
carbon black while Fig. 4 shows a particle from the post-
extraction sample.
The black dashes in Fig. 4 are observer-added indi-
cations of structures that were deemed to be fullerenic
and included in the concentration and size data. Note
that the contrast from the fullerenes in Fig. 4 are qual-
itatively similar to the calculated images shown in Fig. 1.
The absence of black dashes in Fig. 3 highlights the lack
of fullerene-type structures in the carbon black sample.
Only carbon structures along the periphery of the par-
ticles were analyzed as only the periphery was thin en-
ough toallow forobservation
measurements of the structures [2]. The hand-drawn
black line in the inset to Fig. 4 shows the boundary
between the area that was analyzed and the particle
interior, whose thickness presents too many stacked
andaccurate
Fig. 2. A series of three electron microscope images, taken of the same area of a combustion generated fullerenic sample, over a short time span. The
degree of objective lens defocus (in the range of )30 nm) is also varying amongst the images. A C60-like molecule can be seen in each image,
the measured size of which varies.
Fig. 3. HRTEM image of a particle from a pure carbon black sample.
Fig. 4. HRTEM image of a particle from post-extraction tethered
fullerene sample.
1910
A. Goel et al. / Carbon 42 (2004) 1907–1915

Page 5

carbon layers to allow for accurate structural identifi-
cation. It is unclear whether perceived structures in the
particle interior inside the boundary are in fact single
structures or the result of superpositioning of two or
more different structures. Only the material outside the
boundary was sufficiently thin to ensure interpretable
observations. Qualitatively, the images show quite
clearly that the carbon black doped with tethered
fullerenes has many more fullerene-type structures than
the pure carbon black particles.
Quantitative analyses of the same images reinforces
the qualitative observation. Fig. 5 shows the method
used to perform the quantitative analyses. It can be seen
from this cartoon, corresponding to the five condensed
structures in the inset to Fig. 4, that both vertical and
horizontal height (diameter) were measured and then
averaged. This averaged diameter was then used for size
distribution purposes. Table 1 gives a summary of the
fullerene concentration data.
From Table 1, it is seen that both samples containing
tethered fullerenes have a fullerene concentration almost
an order of magnitude greater than the concentration of
what appears to be fullerenes in the pure carbon black
sample. It should be noted that the post-extraction
sample does have a slightly lower concentration than the
pre-extraction samples. This is not surprising as it is
expected that less than 100% of the functionalized
fullerenes would react with the carbon black, leaving
some untethered fullerenes to be separated during
extraction.
It should be noted also that both the pre- and post-
extraction samples exhibit concentrations less than what
would correspond to the total amount of functionalized
fullerenes added in the experiment. Considering the
relative amounts of carbon black and functionalized
fullerenes utilized, and assuming a uniform distribution
of fullerenes over the superficial surface of the carbon
black, the calculated area concentration of fullerene
molecules would be 0.25 molecules/nm2. The corre-
sponding linear concentration of fullerenes would be
0.50 molecules/nm. Both tethered samples yield a linear
concentration approximately 20% of this theoretical
value indicating that many of the fullerenes are not
observed. This result is not surprising given the difficulty
of finding and observing fullerenes on the carbon black
particles (see next section).
Figs. 6–8 show representative images taken from the
analysis of samples of flame-generated soot. The striped
patterns in Fig. 6 are the lattice fringe images of the
{111} planes from the deposits of gold that were used
to calibrate the microscope. Figs. 7 and 8 show other
areas of the soot and several key structures are indicated
by the arrows. The numbers associated with the high-
lighted structures are the observed diameters using the
gold calibration as identified in Fig. 6. It can be seen
a
b
c
d
e
Horizontal
(A)
7.0
6.5
6.5
6.0
6.5
Vertical
(A)
6.5
7.0
7.0
7.5
7.5
Average
(A)
6.8
6.8
6.8
6.8
7.0
a
b
c
d
e
Structure

Fig. 5. Measurement method used for structure diameter size distri-
bution.
Fig. 6. HRTEM image of flame soot with gold island deposits and
showing structures smaller than C60.
Table 1
Linear concentration analysis of C60fullerene-like structures in HRTEM images of carbon black with and without tethered C60fullerene molecules
Sample number and descriptionNo. of fullerenic structures Perimeter length (nm)Fullerenic structures per 1000
nm of perimeter
1. Without tethered C60
2. With tethered C60; pre-extraction
3. With tethered C60; post-extraction
21
209
172
1775
2220
1970
12
94
87
A. Goel et al. / Carbon 42 (2004) 1907–1915
1911