RICE UNIVERSITY
Investigating the Electronic Properties of Carbon
Nanostructures with Scanning Tunneling
Microscopy
by
Kevin F. Kelly
A Thesis Submitted
in Partial Fulfillment of the
Requirements for the Degree
Doctor of Philosophy
Approved, Thesis Committee:
Naomi J. Halas, Director
Associate Professor in Electrical and
Computer Engineering
Alexander J. Rimberg
Assistant Professor of Physics
Richard E. Smalley
Professor of Chemistry and Physics
Houston, Texas
March, 1999
Investigating the Electronic Properties of Carbon
Nanostructures with Scanning Tunneling
Microscopy
Kevin F. Kelly
Abstract
Investigating the Electronic Properties of Carbon
Nanostructures with Scanning Tunneling
Microscopy
Kevin F. Kelly
Abstract
Utilizing the scanning tunneling microscope, the electronic and structural proper-
ties of a variety of carbon based nanostructures are explored. Among these, threefold
electron scattering on graphite generated by vacancies and adsorbates is imaged using
C60-functionalized STM tips. By comparing a simple theoretical model to the experi-
mental results, it is possible to identify the exact location of the defect in the graphite
lattice. It is also possible to qualitatively identify the bonding strength of various ad-
sorbates. In addition to graphite, monolayers of disul�de derivatized fullerenes are
probed with the STM. The self-assembling and electronic properties of these molecules
are investigated. Lastly, comparisons are drawn from these experiments to similar
investigations of functionalized carbon nanotubes.
Contents
Abstract ii
Abstract iii
List of Illustrations v
1 Introduction 1
2 Tight-Binding Theory of Graphite Defects 32.1 Tight-Binding Theory . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 STM Imaging of Graphite . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Imaging Graphite Point Defects . . . . . . . . . . . . . . . . . . . . . 10
3 Comparison of Theory and Experiment 213.1 Alpha Versus Beta Site Defects . . . . . . . . . . . . . . . . . . . . . 22
3.2 Fluorinated Graphite . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Low-temperature STM of Graphite . . . . . . . . . . . . . . . . . . . 37
4 Self-Assembling Fullerene Monolayers 464.1 Alkanethiols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 Pure Fullerene SAMs . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3 Insertion into Alkanethiol SAMs . . . . . . . . . . . . . . . . . . . . . 57
5 Future Directions 67
Bibliography 70
A Curricula Vitae 75
Illustrations
2.1 Top view of the graphite crystal structure. . . . . . . . . . . . . . . . 8
2.2 Theoretical STM image for an isolated graphite monolayer. . . . . . . 10
2.3 Theoretical STM Images of the top layer of the graphite cluster at
di�erent bias voltages. . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Theoretical images for �-site (upper) and �-site (lower) vacancy defects. 13
2.5 Bias dependence for a �-site vacancy. . . . . . . . . . . . . . . . . . . 15
2.6 Bias dependence for a �-site vacancy. . . . . . . . . . . . . . . . . . . 16
2.7 Bias dependence for a carbon adatom located on a �-site. . . . . . . . 17
2.8 Dependence of electron scattering intensity on the coupling energy of
a carbon adatom bonded to a �-site. . . . . . . . . . . . . . . . . . . 19
3.1 Examples of �- and �-site scattering. . . . . . . . . . . . . . . . . . . 23
3.2 (a) A theoretical image of a �-site vacancy. (b) A theoretical image
calculated for an �-site vacancy. (c) An experimental image of a
graphite defect displaying superlattice features similar to a �-site
defect. (d) An image of a defect with superlattice features analogous
to the theoretical image of an �-site defect. Both experimental images
were obtained at 100 mV tip bias and 1 nA tunneling current. Inset
in each the four images is the center region of their Fourier transforms. 26
3.3 The same images from the previous �gure after cropping and
bandpass �ltering. The black circles represent the positions of carbon
atoms in the � site sublattice of each image. . . . . . . . . . . . . . . 28
3.4 An image of graphite uorinated at 300 �C obtained with a
C60-functionalized STM tip. The prominent threefold defect in the
lower right is attributed to a covalent C-F bond, while the other
defects represent uorine atoms adsorbed with an ionic bond. The
spacing between the atomic sites in the unperturbed areas of the
images is 2.46 �A. The inset is the Fourier transform of the
topographic image. The tunneling parameters were 1 nA and +100
mV sample bias voltage. . . . . . . . . . . . . . . . . . . . . . . . . . 31
vi
3.5 A constant current topographic image of graphite uorinated at 250�C imaged with a bare metal STM tip. The di�erences in defect size
correspond to di�erences in the C-F bonds. Inset in this image is the
corresponding Fourier transform. The six outer peaks are due to the
graphite lattice, while the less intense six inner peaks correspond to
the electronic superlattice. The tunneling parameters were 1 nA
tunneling current and +100 mV sample bias voltage. . . . . . . . . . 32
3.6 Theorectial images re ecting changes in the adsorbate interaction
with the graphite surface. In the upper image, the interaction of the
adsorbate with the surface is set equal to that of the neighboring
atoms. While in the lower one, the strength of that interaction is
decreased by a factor of three. . . . . . . . . . . . . . . . . . . . . . . 34
3.7 Atomic resolution STM image of anisotropic electron scattering from
a point defect on highly oriented pyrolytic graphite ovserved by a
metal STM tip at 77 K (75 �A x 75 �A image area, sample bias =
50mV, It = 500pA). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.8 STM image showing one each of �-site and �-site point defects on
highly oriented pyrolytic graphite exhibiting anisotropic electron
scattering (75 �A x 75 �A image area, bias = 50mV, tunneling It =
750pA, temp. = 77K). . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.9 Two images of a single point defect on highly oriented pyrolytic
graphite acquired at sample biases of -50mV (upper) and -100mV
(lower). With the larger magnitude bias voltage the anisotropic
nature of the perturbation is reduced (for both images 40 �A x 40�A image area, It = 1 nA, temperature = 77 K). . . . . . . . . . . . . 43
3.10 Fermi distribution of electron occupation in a metal at 300 K (solid
line) and 77 K (dotted line). Inset is a plot of df/dE vs. E showing
the e�ect of temperature on the energy broadening of the electron
distribution. The energy scales for the graphs are relative to the
Fermi level of the metal. The occupancy of the density of states
about the Fermi level is sharpened at 77 K and the FWHM of the
energy broadening is reduced to 23 meV from 91 meV at 300K. . . . 44
4.1 An illustration of the (p3�
p3)R30� crystal structure of alkanethiols
formed on Au(111). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 An STM image of a dodecanethiolate monolayer on Au(111) . . . . . 50
4.3 The Fourier transform of Fig. 4.2 highlighting the di�erent
reconstructions of the SAM. . . . . . . . . . . . . . . . . . . . . . . . 51
4.4 A large area STM image of a hexanethiol SAM demonstrating the
long range-order achievable with alkanethiols. . . . . . . . . . . . . . 52
4.5 A high resolution STM of a hexanethionl SAM. . . . . . . . . . . . . 52
4.6 An illustration of the structure of the short-chain fullerene disul�de. . 53
vii
4.7 A large area STM image (76 X 76 nm) of SAMs generated from the
short-chain C60 on Au(111). It was acquired at +1.0 V sample bias
and 50 pA tunneling current. . . . . . . . . . . . . . . . . . . . . . . 55
4.8 A high resolution STM image (18 X 18 nm) of the same SAM. The
scan parameters for this image were +700 mV sample bias and 80 pA
tunneling current. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.9 An image of short-chain C60 disul�de inserted into a hexanethiol
SAM. The bright areas are the fullerene moeity protruding above the
thiols, while the dark regions are etch pits in the gold surface due to
the self-assembly process. . . . . . . . . . . . . . . . . . . . . . . . . 59
4.10 Image of long-chain C60 inserted into undecanethiol SAM. . . . . . . 60
4.11 Image of long-chain C60 inserted into dodecanethiol SAM. This image
has been processed by unsharp masking to ehance the visibility of the
inserted molecules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.12 Image of long-chain C60 inserted into decanethiol SAM. . . . . . . . . 62
4.13 Scanning tunneling spectroscopy curves obtained in various locations
on the thiol �lms. The gap impedance was 2 x 1010 ohms (set at +1
V bias voltage and 50 pA tunneling current). . . . . . . . . . . . . . . 64
5.1 Front and back theoretical images of a vacancy in a (10,10) nanotube. 68
5.2 STM image of uorinated nanotubes on Au(111). . . . . . . . . . . . 69
1
Chapter 1
Introduction
Since its invention, the scanning tunneling microscope (STM) has become a powerful
tool in many �elds of science and engineering. It was the �rst in a series of scanning
probe microscopies to investigate the surface properties of materials on a nanometer
scale. The STM can consistently achieve atomic resolution on a wide range of conduc-
tors semiconductors and organic thin �lms. The atomic resolution of the STM makes
it a valuable surface probe for physicists, but its ability to work in air and solution
has also made it a useful tool for chemists, materials scientists, and biologists. In
recent years, it has evolved from a local probe to also become a tool for nanometer
lithography[1] and manipulation.[2, 3] This thesis is a summary of research conducted
with the STM on various carbon-based nanostructures.
Applications of STM to such areas as nanometer lithography, memory, and surface
chemistry, necessitate that the nature of STM images of point defects on the surface be
well understood. Recently, many researchers have used the atomic-scale imaging ca-
pability of the scanning tunneling microscope (STM) to observe long-range electronic
e�ects caused by impurities and steps on metal and semiconductor surfaces.[4, 5, 6, 7]
It has also been observed that electron scattering plays an important role in the ad-
sorption of molecules on surfaces.[8] STM imaging of electron scattering patterns
from point defects was �rst predicted and observed on graphite; however, the images
initially obtained did not resemble the theoretically predicted images.[9, 10, 11] In
2
contrast, when localized graphite defects are imaged with a C60 molecule adsorbed
onto the STM tip they yield highly reproducible electron scattering patterns which
correspond quite closely to the original theory.[12, 13]
The �rst part of this thesis is devoted to expanding the understanding of electron
scattering on graphite previously discussed. Chapter 2 discusses the theory behind
this phenomenon and presents a model for predicting the scattering patterns for a
variety of defect structures. Comparisons between these predictions and experimental
observations are discussed in Chapter 3. It will be demonstrated that information
such as bonding con�guration and bond strength can be determined from the STM
images. This is, to our knowledge, the �rst time that electron scattering modulations
have been used to obtain information about the local surface structure that gives rise
to this e�ect. Since the key to observing the threefold scattering is the C60 adsorbed
tip, investigations on funcationalizing fullerene molecules is presented in Chapter 4.
Using the STM, we have investigated the structural and electronic properities of
variations of disul�de C60 derivatives. Chapter 5 concludes with a discussion about
the future directions and applications of this research.
3
Chapter 2
Tight-Binding Theory of Graphite Defects
The theoretical work for STM imaging of defect structures on a graphite surface was
pioneered by Howard Mizes[9] and Maribel Soto.[14] In separate calculations, they
described the theoretical basis for calculating STM images from the local density of
states (LDOS) of a graphite surface. Mizes performed tight-binding calculations of
point defects using the Green function for the honeycomb lattice. The framework
for his predictions regarding graphite defects was that of a single graphene sheet,
and certain assumptions were necessary to translate these results to bulk graphite.
This technique yielded expressions for the LDOS at the Fermi energy (Ef) and thus
theoretical STM images of these defects. On the other hand, Soto's calculations on a
vacancy defect in a carbon cluster contradicted some of the predictions resulting from
Mizes' theory. Her technique used a graphite bilayer to represent the bulk graphite
and was implemented using an sp3 basis set. Constraints on available computer
time restricted the total number of carbon atoms to 32. The long-range scattering
predicted by Mizes' model, would then indicate a strong interference from nearby
vacancies in Soto's calculation due to the periodic boundary conditions imposed.
Since it is now possible to experimentally observe this threefold electron scattering
on graphite,[12, 13, 15] this chapter will re-examine and expand on the previous
theoretical descriptions of this phenomenon. Using the previous work as a starting
point, a more elaborate cluster calculation is used to model the e�ect of various defect
4
structures on the electron scattering patterns observed by the STM. The following
chapter will then utilize these predictions to identify some of the properties of the
experimentally observed defects.
2.1 Tight-Binding Theory
In tight-binding theory, the basis set for the wave functions of the solid is the atomic
orbitals of the free atoms. The electronic wave functions of the crystal are described
in terms of a linear superposition of these atomic eigenfunctions. This procedure is
also known as the LCAO method (Linear Combination of Atomic Orbitals). Since
only valence electrons may contribute to the tunneling current in an STM image,
only the valence orbitals are of interest in this calculation. The wavefunctions of the
valence electrons k of the crystal are the solutions of the Schr�odinger equation
H� E = 0 (2.1)
where =P
k k. We then substitute in a linear superposition of wave functions �i
of the atomic electron states for k
k =Xi
cik�i: (2.2)
Substituting Eq. 2.2 into Eq. 2.1 gives
Xk
Xi
cik(H � E)�ik = 0: (2.3)
Multiplying Eq. 2.3 by �j and integrating over all spatial coordinates of the wave
functions leaves the following three integrals to be evaluated:
5
Hii =
Z�iH�id� (2.4)
and
Hij =
Z�iH�jd� (2.5)
and
Sij =
Z�i�jd�: (2.6)
Once the values of these integrals are determined, Eq. 2.3 can be solved for the
energies of the electron orbitals of the solid. Once the energies are known, it is
possible to determine the coe�cients cik and thus obtain explicit expressions for the
wave functions of the valence electrons.
To simplify the calculation, the H�uckel approximation is used. This assumes that
all Sij = 0 and that all Hij = 0 unless the ith and jth orbitals are on adjacent
atoms. When the H�uckel approximation is applied to the � orbitals of hydrocarbons,
then Hii and Hij take on the following interpretation. The Hii integral represents
the energy of an electron in a carbon p� orbital. The Hij integral is the energy
of interaction between the orbitals of adjacent atoms and is also referred to as the
hopping parameter.
For a crystal of N atoms with m valence electron orbitals, the Hamiltonian in this
basis is an Nm � Nm matrix,
H =
0BB@
a1;1 � � � a1;Nm
.... . .
...
aNm;1 � � � aNm;Nm
1CCA (2.7)
6
with eigenvalues corresponding to the energies of the Nm one-electron states and
each eigenvector corresponding to the wave function for that particular energy. The
ith element of the eigenvector corresponds to the contribution of the ith atom to
the wave function. To go from this basis to an STM image, which is measured in
real space, we calculate the wave function at a point by summing the contribution
from each orbital at that point, keeping track of the correct phases. The local charge
density associated with that wave function is proportional to the square of the wave
function. The sum over all wave functions near Ef will be proportional to the local
density of states (LDOS). In order to simulate the observation of the surface LDOS
by the STM, we apply the same assumptions as Terso� and Hamann.[16] The main
assumption is to describe the STM tip by a single atom with only an s orbital. This
results in the STM image essentially re ecting the local density of states of the surface
only. Terso� and Lang later revised this theory for the explicit case of imaging the
graphite surface.[17] The validity of their assumptions will be discussed in Chapter
3 by comparing the results of metal and fullerene STM tips when imaging di�erent
defect structures and over di�erent temperatures.
2.2 STM Imaging of Graphite
The ideal crystal structure of graphite consists of stacked hexagonal planes of carbon
with two inequivalent atomic sites as shown in Fig. 2.1. The �-site atoms on the sur-
face are directly over another carbon atom, and the �-site atoms are located directly
over the center of a hexagonal ring, or hollow, in the next layer. This leads to an
ABABAB... stacking sequence and is commonly referred to as hexagonal or Bernal
graphite. The nearest-neighbor spacing in the plane is 1.42 �A, while the inter-planar
7
distance is 3.35 �A. This means that an �-site atom would couple strongly to its three
neighbors in the plane and weakly to its to neighbors out of the plane. Because each
�-site atom is positioned above and below the center of a hexagon, the coupling a
�-site atom to the atoms in the neighboring planes is much weaker than that of the �-
site. Accordingly, the interplane coupling is assumed to be zero for the �-site atoms.
This physical di�erence between the �- and �-site atoms leads directly to an electronic
di�erence as well. The weak interaction of the �-site atoms with the adjacent layers
causes a decoupling between the �- and the �-site atoms. This leads to a dispersion in
the density of states of graphite so that only the �-site atoms signi�cantly contribute
to the electron density at the Fermi energy in crystalline graphite.[18] This e�ect has
been observed experimentally. At low bias voltages, the STM images only the �-site
atoms and higher bias voltages are necessary to image the �-site atoms.[19]
The range of voltages accessible for imaging and spectroscopy in scanning tun-
neling microscopy is limited by the sample work function. Experimentally, images
can be taken for states approximately 5 eV above and below Ef . For graphite, the �
bands lie within a range of 3 eV of Ef , therefore, the � bands will make the dominant
contribution to the STM images. However, comparing our valence electron per atom
model to the experimental images in the next chapter will demonstrate the necessity
of including other electron orbitals for more accurate theoretical descriptions.
In the cluster method, the graphite lattice is approximated by a �nite number of
atoms with periodic boundary conditions and the full matrix is used. The advantage
of this method is that it can be easily extended to accommodate multiple impurities,
chemisorbed atoms, substitutional atoms, step edges, and the e�ects of the lower
layers. For the following calculations, the initial graphite cluster is composed of 2916
8
1st layer β-site atom1st layer α-site atom2nd layer β-site atom
Figure 2.1: Top view of the graphite crystal structure.
9
carbon atoms equally separated into two graphite sheets. The on-site p-orbital energy
of the carbon atoms equals -8.97 eV. The in-plane coupling of neighboring p-orbitals
is set to 3 eV and the interplane coupling of the �-site atom p-orbitals equals 0.3 eV.
When the cluster is only composed of carbon atoms, it is typical to set the on-site
energy equal to zero, which then becomes Ef for the cluster. The Hamiltonian then
becomes a very large, sparse matrix whose eigenvalues are the energy levels of the
cluster and whose corresponding eigenvectors are the wave functions of the electrons
at that energy. The STM image is constructed by multiplying the eigenvector by the
2pz orbital for that atom, summing over all the atoms, squaring this result and then
summing the wave functions over the desired bias voltage range. In these calculations,
the 2pz orbital is approximated using the atomic charge superposition model, which
has previously been proven to accurately simulate experimental STM images.[20]
When calculating the various defect structures in the following section, it is assumed
that no relaxation or reconstruction occurs in the surrounding bonds of either the
�rst or second layer. However, such a change in bond length could be incorporated
by appropriately adjusting the coupling energies.
For a graphite monolayer, we expect the theoretical STM image to show every
atom of the hexagon, which is indeed the case as shown in Fig. 2.2. This image is
generated from a sum of the discrete states at Ef . Following the method developed by
Tomanek et al:, we can calculate a continuous LDOS to better represent bulk graphite.
This is done by broadening the discrete energy levels (eigenvalues) of the cluster into
Gaussian distributions with a half-width at half-maximum of 0.2 eV. Then at a voltage
away from Ef equal to the simulated bias voltage, the LDOS stipulates how much to
weight each of the nearby energy levels and the corresponding wavefunctions. In this
10
way, the local charge density at a given bias voltage can be calculated. Figure 2.3
displays the results for simulations of +100 mV and -500 mV bias voltage. The two
sublattices can be distinguished by looking at the number of atoms around the edge
of each image.
2.3 Imaging Graphite Point Defects
Threefold electron scattering was �rst observed in the vicinity of graphite defects gen-
erated by low energy argon ion bombardment.[12] Molecular dynamics simulations of
this phenomenon indicate that vacancies are the primary type of defect formed.[21]
In order to better understand the experimental images, the above mentioned model
was used to simulate vacancy defects in graphite. Contributions from both �- and
�-site wave functions to the LDOS were included in calculating the electronic per-
Figure 2.2: Theoretical STM image for an isolated graphite monolayer.
11
Figure 2.3: Theoretical STM Images of the top layer of the graphite cluster
at di�erent bias voltages.
12
turbation on the STM image due to a vacancy defect at either site. The vacancy
was simulated by setting the hopping parameters to zero and the on-site energy at
the defect site equal to a large value, outside the range of the unperturbed graphite
density of states.[22]
Figure 2.4 displays the STM images for �-site and �-site vacancies predicted by
this model. The upper image shown in Fig. 2.4 for a �-site vacancy is in good
agreement with Mizes' original prediction. However, the lower image shows that an
�-site vacancy also gives rise to long-range threefold scattering near EF , which was
not originally predicted by Mizes and Foster.[9] Relative to the �-site vacancy, the
�-site vacancy scattering pattern is rotated by 60�. For an �-site vacancy, each of
the three neighboring �-sites has a dangling bond. This leads to a change in the
electronic structure of these sites and other nearby �-sites, but leaves the nearby
�-sites una�ected. For a �-site vacancy, the three neighboring �-sites have dangling
bonds, perturbing the LDOS of the surrounding �-sites. Therefore the direction of the
superlattice modulation of the �-site defect follows the orientation of the (perturbed)
�-site bonds, and for an �-site defect, the scattering modulation is oriented along the
�-site bonds. Thus one can clearly distinguish that these two modulations originate
from di�erent defect sites.
Figures 2.5 and 2.6 illustrate the bias dependence of the �-site and �-site vacancies,
respectively. For the �-site vacancy, an extra electronic state is generated at Ef . The
model of the �-site vacancy shows that two scattered waves are generated, one above
Ef and one below. This results from the fact that by removing a �-site atom, we are
perturbing the �-site energy levels of the cluster. The �-sites have no energy levels
at Ef in the unperturbed graphite. The nearest perturbation on the original states
13
Figure 2.4: Theoretical images for �-site (upper) and �-site (lower) vacancy
defects.
14
occur around �0.6 eV above and below Ef . In the case of the �-site vacancy, the
occupied state scattered wave matches the Green's function for the honeycomb lattice
very well. For the unoccupied state scattered wave, a slight di�erent threefold image
is generated. This is analogous to the hydrogen molecule where a bonding and an
antibonding state are created based on the relative phase of one atom's wavefunction
with its neighbor. The di�erence between the positive and the negative bias voltage
images is the phase shift between the scattered wave and that of the background
lattice. For the �-site vacancy, this e�ect occurs away from Ef .
This model can also be used to probe the e�ects that an adatom would have upon
the surrounding electron charge densities and the related STM image. The e�ect of
the adatom depends on the strength of its adsorption and the energy of its orbitals
relative to the graphite � bands. We would expect physisorbed atoms/molecules to
cause very little perturbation to the surrounding atoms and their e�ects to be short
range. On the other hand, chemisorbed atoms should have a pronounced e�ect on
the electronic states of the surface. As Fig. 2.7 demonstrates, chemisorbed carbon
adatoms generate a strong electron scattering e�ect near Ef . In this case, the coupling
energy of the adatom to the surface was set equal to the in-plane coupling energy.
The threefold scattering in Fig. 2.7 is very similar to that of a vacancy defect.
Likewise, the theoretical image for a carbon adatom strongly bonded to an �-site is
nearly equivalent to that for an �-site vacancy. This is due to the fact that energy
levels of the scattered wave for an adatom at Ef depends on the di�erence in the
pz orbital energy between the surface and the adatom. Thus, when both atoms
are carbon, the electrons at Ef are greatly perturbed. While the covalent bonding
of an isolated carbon adatom to graphite is not a realistic phenomenon, the covalent
15
Figure 2.5: Bias dependence for a �-site vacancy.
16
Figure 2.6: Bias dependence for a �-site vacancy.
17
Figure 2.7: Bias dependence for a carbon adatom located on a �-site.
18
bonding of carbon functional group such as a methyl or a butyl would likely generate a
similar electron scattering pattern. In addition, the interaction of the carbon adatom
with the surface can be changed to investigate changes in the electron scattering
patterns as the adatom's interaction with the surface increases from physisorption
to chemisorption. This calculation is performed by gradually increasing the coupling
energy, the results of which are shown in Fig. 2.8.
For the case of adsorption of non-carbon adatoms on the graphite surface, the
energy levels associated with these scattered states will occur away from Ef . In this
simple model, the electron scattering states are shifted by the di�erence of the on-
site pz orbital energies between that of the adatom and the underlying carbon atom.
However, in real experimental situations, non-carbon adatoms may still create reso-
nant states near Ef . The following chapter demonstrates robust threefold scattering
from vacancy defects scanned in air. Under ambient conditions, it is quite likely that
the dangling bonds react with oxygen or other elements. Therefore additional reso-
nance states must be created near Ef to account for the experimental observation.
Similarly, threefold scattering near Ef is observed for chemisorbed uorine adatoms,
even though it is not predicted by the one valence electron per atom model.
In summary, a bilayer cluster model of the graphite surface has been developed
to predict the electron scattering patterns from various defects on graphite. Unlike
Mizes' earlier theory, this model quantitatively accounts for the energy states as-
sociated with the �-site atoms. In addition, the cluster size is large enough that
perturbations from nearby defects do not interfere. The predictions of the model in-
clude threefold electron scattering from both �- and �-sites and an increase in electron
scattering with adsorbate bonding strength. The comparison between these results
19
Figure 2.8: Dependence of electron scattering intensity on the coupling
energy of a carbon adatom bonded to a �-site.
20
and experimental STM images of threefold electron scattering are discussed in the
following chapter.
21
Chapter 3
Comparison of Theory and Experiment
Recently, many researchers have used the atomic-scale imaging capability of the scan-
ning tunneling microscope (STM) to observe long-range electronic e�ects (Friedel os-
cillations) caused by impurities and steps on metal and semiconductor surfaces.[4,
5, 6, 7] The observation of these scattered electron states has proved bene�cial in
exploring adsorption on surfaces,[8] the depth of impurities in solids,[23, 7, 24] and
the mixing of surface and bulk electronic states.[25] In this chapter, comparisons of
the preceding theoretical descriptions of electron scattering on graphite are made to
experimental observations of this e�ect in the hopes of gaining greater insight into var-
ious graphite defect structures. In the �rst section, comparisons are made between
theory and defects produced by low energy ion bombardment. It will be demon-
strated that it is possible to identify the crystallographic location of the defects in
the graphite lattice from their scattered electron waves. This is, to our knowledge, the
�rst time that electron scattering modulations have been used to obtain information
about the local surface structures that give rise to this e�ect. The following section
will then describe the e�ects of bond strength and chemical identity of adsorbates on
the threefold electron scattering patterns. Speci�cally, electron scattering from dilute
coverages of uorine adatoms will be analyzed. In the �nal section of this chapter, we
wil brie y discuss the imaging of the threefold patterns by a metal STM at cryogenic
temperatures. These experiments demonstrate that a strong reason for the observa-
22
tion of these patterns with a fullerene-coated STM tip at room-temperature is due to
the narrow density of states of the C60 molecule.
3.1 Alpha Versus Beta Site Defects
Using a three dimensional tight-binding model of graphite, the previous chapter pre-
dicted that electron scattering would originate from defects on both graphite lattice
s. As mentioned earlier, this e�ect was not predicted by Mizes calculations.[9] This
poses the question as to whether or not we can observe scattering from both defects
experimentally. As displayed in Fig. 3.1, the answer to this question is yes. In the
upper image, two defects on graphite are imaged by an STM tip with 2 C60 molecules
at the tip apex. This gives the appearance of four defects when there are actually only
two. In the lower image of Fig. 3.1, we see two threefold defects displaying rotated
scattered patterns along with other defects that do not display strong electron scat-
tering. However, it is not readily obvious from these images at which lattice position
each defect resides.
From the theory described in the previous chapter, it might appear possible to
determine the defect location from changes in the strength of the electron scattering
patterns over a range of bias voltages. While this may be possible in UHV, where
the vacancies remain unreacted, these samples are imaged under ambient conditions.
Given this, it is likely that at least one, if not all, of the dangling bonds around the
vacancy sites have reacted. As the preceding chapter discussed, this alters the energies
of the electronic states of the defect. Moreover, the one-electron per atom model
only accounts for such changes qualitatively. It is quite likely that a many orbital
calculation would be necessary for a quantitative determination of the various defect
23
Figure 3.1: Examples of �- and �-site scattering.
24
structures. Accordingly, a new methodolgy is necessary to determine the location of
the defects expediently and independent of chemical composition. One possibility is
to take advantage of the di�erent wavelength of the scattered electron waves relative
to that of the unperturbed lattice. This section focuses on a procedure to identify the
defect location by �ltering the STM images in Fourier space to isolate the scattered
wave and compare its position to the background lattice.
Figures 3.2(a) and (b) demonstrate the STM images for �-site and �-site vacan-
cies predicted by the model in the preceeding chapter. As previously discussed, the
�-site scattering pattern is rotated by 60� relative to the �-site one. Figures 3.2(c)
and (d) are experimental STM images of graphite defects obtained by scanning with
a C60-adsorbed STM tip. The threefold scattering pattern was typical of about half
of the defects imaged. The point defects were generated by bombarding the graphite
with 60 eV argon ions in ultra-high vacuum (base pressure = 5 � 10�10 Torr) at a
uence of approximately 1011 ions/cm2. Molecular dynamic simulations have shown
that this process generates primarily vacancy defects on the graphite surface.[21] The
graphite is then scanned under ambient conditions with C60-adsorbed STM tips, the
preparation and characterization of which have been previously described.[26, 27] A
slight distortion of the scattered waves relative to the highly symmetric theoretical
calculations was typically seen in our experimental images. This may be a conse-
quence of localized relaxation at the defect site, or a reaction of the vacancy with
an unknown adsorbate. It was recently shown that graphite vacancies readily react
with molecular oxygen; leaving one oxygen atom satisfying two of three dangling
bonds, while the other oxygen atom bonds to the remaining carbon atom.[28] This
structure may very well explain the asymmetry seen in some of the threefold defects.
25
Nevertheless, both of these images display very good agreement with the theoretical
images.
In the four images of Fig. 3.2 (insets), the centers of the corresponding Fourier
transform of each STM image (FT-STM) are displayed. The Fourier space transforms
of STM images have proven useful for analyzing periodic features appearing in the
corresponding real space images.[6, 29] The six inner spots of the FT-STM images
correspond to the (p3�
p3)R30� superlattice generated by the electron scattering,
where R is the distance between �-site atoms (2.46 �A). The six outer spots along
the same direction are due to the second harmonic of the superlattice waves, while
the six brighter and rotated outer spots are due to the regular graphite lattice. By
designing a �lter based on the FT-STM image one can selectively remove or enhance
the contributions of the lattice and superlattice in the real space image.
Careful application of image processing techniques enables one to examine mod-
i�cations in the charge densities at the perturbed lattice sites in the vicinity of
each defect relative to the unperturbed lattice sites. This permits a determination
of which defect site gives rise to which electronic modulation pattern. Figure 3.3
shows the same four images of Fig. 3.2 after �ltering with a two-dimensional band-
pass �lter to enhance the superlattice in each image. The �lter was created in two
steps. First, a one-dimensional equiripple bandpass �lter centered on the spatial fre-
quency of the superlattice peaks is generated using the Parks-McClellan algorithm.[30]
This 1-D �lter is then transformed into a two-dimensional �lter using the McClellan
transformation.[31] This e�ectively isolates the scattered wave from the rest of the
image. We can now directly compare the registry of the scattered wave with the
underlying �-site sublattice, represented by the black open circles. Since the �-site
26
a) b)
c) d)
Figure 3.2: (a) A theoretical image of a �-site vacancy. (b) A theoretical
image calculated for an �-site vacancy. (c) An experimental image of a
graphite defect displaying superlattice features similar to a �-site defect.
(d) An image of a defect with superlattice features analogous to the theo-
retical image of an �-site defect. Both experimental images were obtained
at 100 mV tip bias and 1 nA tunneling current. Inset in each the four
images is the center region of their Fourier transforms.
27
defect gives rise to an electronic modulation that suppresses the � sublattice con-
tribution, the electron density of the scattered wave is observed to be strongest at
the �-sites of the lattice. For the �-site defect, the increased charge density at the
� sites results in a shift of the scattered wave away from the �-sites. This is the
result one would expect from the previous discussion of the e�ects of the dangling
bonds adjacent to each respective defect. For the �-site defect the � sublattice is still
suppressed, while the scattered wave from the �-site defect is spatially shifted by the
increased charge density from atoms located on �-sites. By performing a similar anal-
ysis on the scattered waves displayed in the experimental images, we can determine
that Fig. 3.3(c) is a �-site defect and Fig. 3.3(d) is an �-site defect. Applying this
procedure to observed images of electronic modulations permits the identi�cation of
the defect site independent of a strong real space correlation between the observed
superlattice and theory.
In conclusion, by imaging point defects on graphite with C60-adsorbed STM tips,
we are able to observe electron scattering phenomena that cannot be observed with
metal STM tips. We have also demonstrated that a bilayer model of the graphite sur-
face is necessary to accurately reproduce some of the features of the electron scattering
patterns observed. By isolating the scattered electron waves in frequency space, we
are able to identify the speci�c location of the original vacancy sites on the graphite
surface. The determination of which can only be made when the graphite defects are
imaged with a fullerene nanotip.
28
a) b)
c) d)
Figure 3.3: The same images from the previous �gure after cropping and
bandpass �ltering. The black circles represent the positions of carbon
atoms in the � site sublattice of each image.
29
3.2 Fluorinated Graphite
Fluorinated carbon materials are an area of intense research due to their important
practical applications such as lubricants, water repellants, and battery cathodes. The
chemical composition of uorinated carbons depends on both the structure of the
starting material and the uorination conditions.[32, 33] One of the most intriguing
aspects of uorinated graphite is the variability of its electrical conductivity. This is
due to the ability of the uorine atoms to form both ionic and covalent bonds with the
carbon atoms. Electrical conduction in uorinated graphite can be enhanced by the
ionic bond between the uorine and carbon atoms due to an increase in hole carriers,
while covalent bonding produces a decrease in the overall carrier concentration.[34]
In this letter, we use the STM to identify the ionic and covalent C-F bonds on
graphite by the di�erences induced by each bonding con�guration in the surrounding
local density of states of the graphite substrate. In this letter, we demonstrate the
ability to identify two di�erent bonding con�gurations of uorine atoms adsorbed
onto graphite via the easily distinguishable di�erences in their surrounding electron
scattering patterns.
The uorinated graphite samples were prepared by placing freshly cleaved highly
oriented pyrolytic graphite (ZYA grade) in a Monel ow reactor and exposing it to
uorine gas diluted with helium at elevated temperatures. The HF impurity of the
uorine gas was removed by passing it through a HF trap containing sodium uoride
pellets. The F2 and He ow rates for these reactions were 1 and 100 sccm, respectively.
The reaction time in each case was 2 minutes unless otherwise noted. The uorination
reactions were carried out at 200, 250, and 300 �C. Scanning was performed under
30
ambient conditions with both bare, mechanically cut Pt/Rh (87/13) tips and the
same tips functionalized with C60 molecules as previously described.[26]
Imaging the uorinated graphite surfaces with a C60-functionalized STM tip yields
highly reproducible images of the local uorination sites and allows for clear distinc-
tion between the two types of bonding. Figure 3.4 is an STM image of a graphite
sample uorinated at 300 �C using an STM tip that has been functionalized by the
adsorption of a C60 molecule at its apex. This procedure facilitates the imaging of
the predicted threefold electron scattering around graphite defects at room tempera-
ture. Inset in this image is the center of its Fourier transform. Besides the six outer
peaks due to the graphite lattice, a set of six very strong inner peaks corresponding
to ap3 �
p3 R30� superlattice (where R=2.46 �A) This superlattice represents a
perturbation to the graphite electronic structure due to the presence of the uorine
adsorbate. The feature in the lower right corner of the topographic image shows a
strong threefold electron scattering pattern, while the other features in this image are
surrounded by much weaker electronic perturbations. Shown in Fig. 3.4 is an STM
image of graphite uorinated at 250 �C obtained using a bare metal STM tip. In this
case, a fairly blunt STM tip is scanned over the adsorbates yielding images of the
STM tip at each bonding site. In this case, the appearance of the superlattice in the
topographic images is highly dependent on the metal tip structure, and is typically
much weaker than the superlattice detected using a C60 tip. In Fig. 3.5, virtually no
superlattice modulation is detected in the topographic image, although weak super-
lattice points are visible in the Fourier transform of this image (inset). Observation
of the superlattice by metal tips has also been reported in the vicinity of adsorbed
metal clusters[11] and defects induced by ion damage.[35] Correlating the number of
31
features in each image observed with C60-functionalized STM tips to those imaged by
bare metal tips, the strong threefold scattering features shown in Fig. 3.4 correspond
to the larger sized features imaged in Fig. 3.5. This is in agreement with previous ex-
periments that ascribed the large size of graphite defects in STM images to electronic,
not topographic, e�ects.[36] For either metal or C60 STM tips, two distinct types of
features were imaged on the uorinated surfaces. Only the overall density, not the
relative proportion, of these features varied with increasing uorination temperature.
Figure 3.4: An image of graphite uorinated at 300 �C obtained with a C60-
functionalized STM tip. The prominent threefold defect in the lower right
is attributed to a covalent C-F bond, while the other defects represent
uorine atoms adsorbed with an ionic bond. The spacing between the
atomic sites in the unperturbed areas of the images is 2.46 �A. The inset is
the Fourier transform of the topographic image. The tunneling parameters
were 1 nA and +100 mV sample bias voltage.
32
Figure 3.5: A constant current topographic image of graphite uorinated
at 250 �C imaged with a bare metal STM tip. The di�erences in defect
size correspond to di�erences in the C-F bonds. Inset in this image is
the corresponding Fourier transform. The six outer peaks are due to the
graphite lattice, while the less intense six inner peaks correspond to the
electronic superlattice. The tunneling parameters were 1 nA tunneling
current and +100 mV sample bias voltage.
33
Comparing the C60-tip obtained experimental images to theory allows the identi�-
cation of the C-F bonding con�guration. The theoretical images in Fig. 3.6 illustrate
the variations in the electron scattering due to di�erences in the adsorbate bond
strength. The upper image in Fig. 3.6 was calculated assuming a strong bonding of
the adsorbate to a single carbon by setting its coupling strength equal to that of the
in-plane neighbors. For the lower image, Vpp� was reduced by a factor of three to
re ect a decrease in the adsorbate bond strength. These images are in very good
agreement with the experimental results displayed in Fig. 3.4. Based on this agree-
ment, we assign the defects demonstrating long-range electron scattering to uorine
atoms having covalent bonds to the graphite surface. For weakly coupled adsorbates
which we attribute to ionic bonding, the intensity of the scattered wave is reduced
relative to the background lattice. As a result, the electronic perturbation surround-
ing the uorine appears more isotropic. However, the structure of the scattered wave
is strongly correlated with the Fermi surface of graphite and remains trigonally sym-
metric. An analogous situation is seen for the scattered electron waves on the (111)
surface of noble metals[25] and the (10�10) surface of beryllium[37] which remain spher-
ically and elliptically symmetric, respectively. While these images qualitatively re ect
the experimental results, a more thorough calculation is necessary for a quantitative
determination of the electronic structure and relative bonding energies.
We also investigated the HF catalyzed uorination of a graphite surface at room
temperature. In this case, the uorine would be expected to bond to the graphite
surface in a predominantly ionic fashion. This sample was exposed for ten minutes
to an atmosphere consisting of 100 He sccm, 1 HF sccm, and 1 F2 sccm at room
temperature. The HF catalyzes the uorination process by forming HF�2 species in
34
Figure 3.6: Theorectial images re ecting changes in the adsorbate interac-
tion with the graphite surface. In the upper image, the interaction of the
adsorbate with the surface is set equal to that of the neighboring atoms.
While in the lower one, the strength of that interaction is decreased by a
factor of three.
35
the presence of molecular uorine which subsequently reacts with the graphite.[38] On
this sample, we observed a decrease in the strongly threefold scattering defects and
an increase in the weakly scattering defects. This is consistent with previous results
which indicate that uorination in the presence of HF yields primarily ionic C-F
bonds.[38] Additionally, a control sample of graphite was uorinated with 1 sccm F2
and 100 sccm He at room temperature and showed neither type of defect to be present
on the graphite surface. This con�rms that uncatalyzed uorination of graphite does
not occur under these conditions. We attribute neither of the defects to molecular
uorine since it is generally assumed that uorine only adsorbs to the graphite surface
in an atomic state.[39] These results appear to con�rm our assignment of the scattered
states to the di�erent bonding con�gurations.
Table 3.1 summarizes the defect density and the ratios of covalently bonded to
ionically bonded uorine for the di�erent uorination conditions as determined from
the STM images. These results can be compared to previous investigations which
probed the nature of the C-F bond through nonlocal techniques such as x-ray pho-
toelectron spectroscopy[40] and nuclear magnetic resonance.[41] The defect density
demonstrates that the amount of uorine reacted with the graphite is highly tempera-
ture dependent which is in good agreement with previous studies. However, a striking
feature for the non-catalyzed uorination is the fairly constant ratios of ionic versus
covalent bonding over the temperature range studied. This is a very unexpected
result considering previous studies on uorinated graphite, although, this is a very
dilute coverage of uorine on the graphite, and it has been previously suggested that
the transformation to a covalent solid does depend on the presence of large amounts
of uorine.[39]
36
Ratios of ReactionGases (in sccm)
SampleTemp. (ºC)
Defect Density(per nm2)
% of F atoms withCovalent Bonds
100 He : 1 F2 200 9.4 x 10-7 18.7 ± 1
He : 1 F2 250 5.0 x 10-5 19.2 ± 1
He : 1 F2 300 1.4 x 10-4 17.3 ± 1
He : 1 HF : 1 F2 23 4.2 x 10-5 11.5 ± 1
100
100
100
Table 3.1: A summary of the results for the di�erent uorination condi-
tions. The reaction times were 2 minutes for the uncatalyzed uorination
and 10 minutes for the sample treated with HF. While there is a noticeable
increase in uorine adatoms with temperature for the uncatalyzed sam-
ples, the ratio of ionic to covalent C-F bonds remains relatively constant.
Whereas uorination with HF shows a reduced percentage of covalent
bonds.
37
In summary, the uorination of graphite has been investigated using bare metal
and C60-adsorbed STM tips. We have found that with C60 molecularly functionalized
tips the chemical nature of the bonding sites can be clearly and consistently distin-
guished at room temperature. In addition, we have found that the percentage of
covalent C-F bonds is independent of the uorination temperature up to 300 �C, in
contrast with results obtained with HF-catalyzed uorination. This, combined with
the ability to determine the exact adsorbate binding site on the substrate lattice,
should prove a highly detailed technique for imaging other adsorbates and for study-
ing other important chemical interactions on graphite surfaces, such as oxidation.[28]
It may also prove useful in analyzing other uorinated graphene structures such as
single-walled nanotubes.[42]
3.3 Low-temperature STM of Graphite
In this section, the observation of anisotropic scattering from point defects in a
graphite surface with the metal probe tip of a low temperature STM is demonstrated.
This anisotropic scattering is consistent with theoretical predictions made by Mizes
and Foster[9] and extended by Kelly and Halas[43] for scattering from point defects on
graphite. The images are also in agreement with previously reported observations of
threefold scattering using a C60-functionalized probe tip at room temperature.[12] The
ability to observe this anisotropic scattering with a metal tip only at low temperatures
is consistent with the proposed resolution enhancement found for C60-functionalized
tips. A narrow distribution of tip electronic states, whether molecular orbitals of a
C60-functionalized tip or the cryogenically sharpened Fermi distribution of a metal
tip, appears to be necessary to image the anisotropic electronic perturbations.
38
All experiments were performed in a low temperature ultra-high vacuum (UHV)
STM operated at 77 K, which has been described previously.[44] Brie y, three room
temperature UHV chambers contain the surface preparation and analysis tools. A
connected low temperature stage contains a beetle-style STM, with multiple layers
of vibration isolation separating the microscope from the surrounding environment.
The STM tips were prepared by mechanically cutting polycrystalline W wire. Tips
were examined using an optical microscope and any tips that were shattered or that
appeared to have unfavorable geometries were discarded. Tips so prepared readily
yielded atomic resolution of the HOPG surface and threefold scattering from point
defects. Once inserted into the UHV system we have no means of replacing the tip
thus in situ tip regeneration is paramount. Upon accidental tip-sample contact the
tunnel junction often became unstable. By scanning at large gap impedances (e.g.
50pA at 3V) a stable tip could be created, which once again was capable of imaging
the lattice and defect scattering. Alternatively, as sample exchange is possible in our
STM, the probe tip could be brought into contact with a metal substrate, or could
be cleaned and sharpened via �eld emission.[45] Although the tips are nominally W,
the chemical identity of the outermost atomic asperity is unknown.
The highly ordered pyrolytic graphite (HOPG) samples were prepared by cleaving
in air with adhesive tape followed by gentle resistive heating in the UHV system to
remove surface contaminants. Single atom surface vacancies (point defects) can be
formed by sputtering a graphite surface with argon ions,[27, 21] however, a small
number are present naturally on unsputtered surfaces and these are the focus of this
study. Point defects were located by imaging relatively large areas of the HOPG
surface (e.g. 500 x 500 nm), followed by higher resolution scans of suspected defects.
39
Defects exhibiting anisotropic scattering had an average density of greater than 1
per 200 x 200 nm area, and were randomly distributed (total point defect densities
were typically 1 per 100 x 100 nm area). No evidence of clustering or of defect-
defect repulsion could be inferred. All images were acquired in constant current mode
with the bias voltage applied to the sample relative to the tip, and are shown here
un�ltered. Images of defect-free areas of the graphite surface exhibited the trigonal
symmetry and lattice spacing of the �-site sublattice, which is exclusively imaged at
the low bias voltages used in this study.[18, 19]
Figure 3.7 is a representative atomic-resolution STM image of anisotropic electron
scattering from a point defect on a graphite surface obtained with aWmetal probe tip.
The scattering pattern is in excellent agreement with both the theoretical predictions
and the previously reported threefold scattering observed with C60-modi�ed tips. The
�-site carbon lattice as well as the threefold symmetric scattering "arms" of the point
defect are resolved. In the lower left corner of the image a second defect or adsorbate
not exhibiting threefold scattering is also found. Anisotropic scattering from point
defects was observed with both positive and negative sample biases over a small bias
window, which is discussed in greater detail below.
As previously mentioned, the most prominent di�erence between the two types of
point defects is that they scatter electrons along di�erent crystallographic directions.
Defects at �-sites (�-sites) scatter electrons along the rows of the �-site (�-site)
sublattice, which was originally predicted by Soto.[14] Figure 3.8 is an STM image of
two point defects exhibiting anisotropic scattering. The perturbations due to the two
defects are rotated 60o with respect to one another and thus are taken to be examples
of �-site and�-site defects. The assignment of the defect position can be obtained by
40
Figure 3.7: Atomic resolution STM image of anisotropic electron scatter-
ing from a point defect on highly oriented pyrolytic graphite ovserved by
a metal STM tip at 77 K (75 �A x 75 �A image area, sample bias = 50mV,
It = 500pA).
41
comparing the theoretical model with careful Fourier space lattice-sublattice �ltering
analysis of the experimental STM image.[43] The results of this analysis indicate that
the defect on the left is located at an �-site and the right defect is located at a �-site.
The question of why these defects are observable with a metal tip at cryogenic
temperature, but not at room temperature, remains. Two atomic-resolution STM
images of a single defect acquired at two di�erent bias voltages are shown in Fig. 3.9.
The image acquired at -50 mV (upper image) exhibits anisotropic electron scattering.
Changing the applied bias voltage to -100 mV (lower image) dramatically changes
the image characteristics. The scattering "arms" are no longer resolved although
some of the threefold symmetry remains. As the bias voltage is further increased,
Figure 3.8: STM image showing one each of �-site and �-site point defects
on highly oriented pyrolytic graphite exhibiting anisotropic electron scat-
tering (75 �A x 75 �A image area, bias = 50mV, tunneling It = 750pA,
temp. = 77K).
42
the defect loses even more of its long-range scattering and becomes more isotropic.
The anisotropic threefold character of the defect is observable only over a very small
voltage (energy) window. The bias dependent characteristics of the STM images
depend on the electronic states of the defect which for some types of defects, such
as vacancies, lie very close to the Fermi energy of the graphite.[14, 46] As the bias
voltage is shifted away from resonance with the defect states, the intensity of the
scattered wave is reduced relative to the underlying lattice. As a result, the electronic
perturbation surrounding the defect appears more isotropic.
To understand the inability to observe such anisotropic scattering at room tem-
perature with a metal tip it is useful to consider the electron distribution of the metal
probe tip. The electron distribution as a function of temperature and energy is given
by the Fermi distribution function where Ef is the Fermi energy and k is Boltzmann's
constant. Figure 3.10 is a plot of the Fermi distribution function for electrons in a
metal at 300 K and 77 K. The electron distribution about the Fermi level is sharpened
at 77 K relative to that at room temperature (300 K). As can be seen in the inset
to Fig. 3.10 the full width half maximum (FWHM) of the energy broadening due to
thermally excited electrons scales as � 3:5kT yielding a FWHM of 23 meV at 77 K
and 91 meV at 300 K. It is the increased energy broadening and associated decreased
energy resolution at room temperature that precludes observation of the anisotropic
scattering with a metal tip. Thermally excited tip electrons tunnel into higher energy
sample states which due to their isotropic character mask the anisotropic nature of
the defect sites.[47] Thus the loss of resolution at higher bias voltages and the in-
ability to observe anisotropic scattering at room temperature with a metal tip are
both the result of higher-energy isotropic states associated with the graphite surface.
43
Figure 3.9: Two images of a single point defect on highly oriented py-
rolytic graphite acquired at sample biases of -50mV (upper) and -100mV
(lower). With the larger magnitude bias voltage the anisotropic nature of
the perturbation is reduced (for both images 40 �A x 40 �A image area, It= 1 nA, temperature = 77 K).
44
Certainly, a more thorough quantitative understanding of this phenomenon would be
obtained by including the tip-sample interaction at low bias voltages and di�erences
in the tunneling matrix element (orbital overlap) between the tip and the graphite
surface.
In summary, the observation of long-range, anisotropic electron scattering from
point defects on a graphite surface with a metal tip at 77K was reported. The observed
scattering is in close agreement with both theoretical predictions and previous obser-
vations with C60-functionalized tips. Cryogenic temperatures narrow the occupancy
of the probe tip density of electronic states about the Fermi level, thus enabling imag-
Figure 3.10: Fermi distribution of electron occupation in a metal at 300
K (solid line) and 77 K (dotted line). Inset is a plot of df/dE vs. E
showing the e�ect of temperature on the energy broadening of the electron
distribution. The energy scales for the graphs are relative to the Fermi
level of the metal. The occupancy of the density of states about the Fermi
level is sharpened at 77 K and the FWHM of the energy broadening is
reduced to 23 meV from 91 meV at 300K.
45
ing of the anisotropic perturbations of the local graphite surface electronic structure.
It is concluded that the image characteristics critical dependence on the tip density of
states is the reason why earlier studies of defects and adsorbates on graphite surfaces
with metallic probe tips were unable to observe anisotropic scattering.[11, 35, 48]
The previous studies were performed at room temperature, where there is signi�cant
thermal broadening of the density of states about the Fermi level of the STM tip.
These results appear to con�rm that the same narrowing of the STM tip occurs with
C60-functionalized tips.[12] In both cases, a narrow energy distribution of accessible
tip electronic states enables the resolution of long-range electronic scattering e�ects
on this surface.
46
Chapter 4
Self-Assembling Fullerene Monolayers
Since their discovery in 1985, buckminsterfullerenes and their related compounds
have been widely studied due to their unique electronic, spectroscopic, and structural
properties.[49] Thin �lms of fullerenes have been shown to exhibit a wide range of
properties such as charge transport, photochemical activity, superconductivity upon
reduction, biological activity, and photophysical behavior strongly in uenced by in-
termolecular interactions. As previously discussed, the modi�cation of STM tips via
the adsorption of a C60 molecule onto the tip apex results in improved atomic res-
olution on graphite[26] and the detection of threefold electron scattering from point
defects on graphite at room temperature under ambient conditions.[12] Along these
lines, this chapter describes the characterization of the development of fullerene thin
�lms and fullerene-coated STM tips, where we have targeted the covalent attachment
of C60 to well-de�ned substrates. The goal of this research to provide a versatile
and well-de�ned fullerene terminated coating generated by the self-assembled mono-
layer strategy. Furthermore, using well established SAM techniques,[50] we expect to
be able to generate and manipulate the structure of fullerene-terminated �lms with-
out the cumbersome necessity of ultra-high vacuum (UHV) conditions or deposition
techniques.
Self-assembled monolayers (SAMs) have become a key technology in fabricating
devices on the micrometer- and nanometer-scale. SAMs provide a unique link between
47
the science of organic surfaces and technologies that seek to exploit their adaptable
character. The initial development of SAMs focused on their usefulness in patterning
surfaces, such as forming masks for beams of metastable atoms[51] and microcontact
printing.[52] However, they have evolved to become an essential component in the de-
vices themselves. The insulating characteristics of some of these molecules has proven
bene�cial in generating well-de�ned tunnel junctions in quantum dot arrays[53] and
�eld-e�ect transistors.[54] By attaching a C60 molecule onto a disul�de, we have com-
bined the favorable characteristics of SAMs with the unique electronic and structural
properties of fullerenes. The dynamics of self assembly have been traditionally inves-
tigated using nonlocal techniques such as ellipsometry, contact angle measurements,
infrared spectroscopy, and x-ray photoelectron spectroscopy. Scanning tunneling mi-
croscopy, on the other hand, o�ers molecular-scale resolution of the surface structure
as well as the ability to perform electrical characterization on individual molecules.
The �rst section of this chapter will brie y discuss scanning tunneling microscopy
investigations of SAMs formed from alkanethiols and disul�des. A thorough un-
derstanding of these SAMs is necessary due to their similarity with the disul�de
fullerenes. The second section is concerned with the characterization of pure disul�de
fullerene monolayers. The �nal section discusses experiments concerning the phys-
ical and electronic interactions of the derivatized fullerenes with various preformed
alkanethiolate monolayers.
4.1 Alkanethiols
Self-Assembled Monolayers (SAMs) of alkanethiols and disul�des on gold surfaces
form organic interfaces with properties largely controlled by the end groups of the
48
molecules comprising the �lm. Nuzzo and Allara published the �rst paper in this area,
showing that dialkyldisul�des (RS-SR) form oriented monolayers on gold surfaces.[55]
Later it was found that sulfur compounds coordinate very strongly to gold, silver,
copper, and platinum.[56] It was also found that similar compounds form ordered
monolayers on gallium arsenide surfaces.[57] The most thoroughly studied and best
characterized SAMs are those formed by thiols adsorbed on Au(111) because of the
nonoxidizing character of the gold surface under ambient conditions. Recently, it was
found that the adsorption kinetics for alkanethiols is faster for longer alkyl chains,
probably due to the increased van der Waals interactions.[58]
The electron transfer from the gold surface to the sulfur atom of the thiol group
weakens the S-H bond and results in its dissociation. On Au(111) the S-H bond is
believed to be cleaved at the on-top site, and the thiolate then moves to its �nal
equilibrium position at the theefold hollow site. A similar cleaving of the disul�de
bond allows for these molecules to form ordered monolayers similar to the thiols. The
bonding of the thiolate group to the gold surface is very strong (approximately 44
kcal/mol).[59] Electron di�raction studies of monolyaers of alkanethiolates on Au(111)
show that the S-S spacing is 4.97 �A and a calculated area per molecule of 21.4 �A2.[60,
61, 62] Figure 4.1 displays the crystal structure of the alkanethiols when self-assembled
on the Au(111) surface.
The discovery in 1983 of the self-assembly of disul�des on gold and, soon there-
after, of alkanethiols coincided with the maturation of STM following its invention in
the early 80's. SAMs on gold have turned out to be a valuable type of sample for in-
vestigation by STM: STM reveals the molecular structure of SAMs and SAMs in turn
reveal information about the imaging mechanism of organic molecules on conductors.
49
Au
S(CH2)n-CH3
S
CH3
S
CH3
S
CH3
S
CH3
S
CH3
S
CH3
Figure 4.1: An illustration of the (p3�
p3)R30� crystal structure of alka-
nethiols formed on Au(111).
Due to the highly insulating nature of the alkane chain, large gap impedances must
be maintained in order to image the monolayers with the STM.[63] Typical scanning
parameters are approximately a volt bias and a set point current of a few tens of
picoamps. Besides the basic (p3 �
p3)R30� overlayer, alkanethiol SAMs can also
form various reconstructions of this structure on the gold surface. This is depicted in
the STM image of a dodecenathiolate monolayer shown in Fig. 4.2. Comparison of
the assorted superlattices is displayed in the corresponding Fourier transform of this
image.
Recent X-ray and helium di�raction studies[64] and STM[65] imaging have also
shown evidence of a rectangular primitive unit mesh with the dimension of 8.68 x 10.02
�A containing four chains, equivalent to a c(4x2) superlattice. Domains di�er by their
phase, their chain tilt or the origin of sulfur adsorption. Molecular arrangements
50
Figure 4.2: An STM image of a dodecanethiolate monolayer on Au(111)
are hexagonal with an addtional rectangular superlattice. The phases originate from
the particular distribution in the unit cell of molecules with di�erent twist angles
that move end groups at di�erent height levels in and out of the average plane of
the monolayer by 0.03 nm. The height di�erence corresponds to a chain rotation
of approximately 90o.[66] Additionally, alkanethiolates have two binding modes at
the Au(111) hollow site, one with a bond angle around the sulfur of 180o (sp) and
the other of 104o (sp3), the later being more stable by 0.41 kcal/mol.[67] This may
also contribute to the various crystallographic reconstructions. The energy barrier
between the two chemisorption modes is very small (2.5 kcal/mol),[67] suggesting
that the thiolate may easily cross from one of these minima to the other.
51
Figure 4.3: The Fourier transform of Fig. 4.2 highlighting the di�erent
reconstructions of the SAM.
Besides dodecanethiol, SAMs were also formed from hexanethiol, decanethiol, and
undecanethiol. These were used to investigate the in uence of the alkyl length has
on the insertion of the derivatized fullerenes described in the third section of this
chapter. Figure 4.4 and Fig. 4.5 are typical images of a hexanethiol SAM observed by
STM. Figure 4.4 demonstrates the long range order that is possible due to the strong
steric interaction of the alkanethiols. In the higher resolution image of Fig. 4.5, we
see similar c(4x2) reconstructions to that of the dodecanthiolate monolayer.
4.2 Pure Fullerene SAMs
In order to combine the favorable self-assembly properties of alkanethiols and the ro-
bust physical and electronic structure of buckminsterfullerene, CH3(CH2)nSS(CH2)n+1NC60
52
Figure 4.4: A large area STM image of a hexanethiol SAM demonstrating
the long range-order achievable with alkanethiols.
Figure 4.5: A high resolution STM of a hexanethionl SAM.
53
was synthesized in Professor T. Randall Lee's lab at the University of Houston. Two
varieties of this molecule were created, one where the C60 molecule is attached by an
aziridine linkage through a short-chain alkane (n=5) to the disul�de and the other
using a longer alkane chain (n=10). Figure 4.6 shows the structure of the shorter
of the two varieties of fullerene-terminated unsymmetrical disul�des. Y-S. Shon et
al: provide a detailed description of the synthesis along with the results of the char-
acterization by contact angle measurements, ellipsometry, and STM of the SAMs
formed from this molecule.[68] The synthesis of the long-chain version is performed
in a similar manner. This synthesis provides the �rst example of aziridine formation
on C60 in the presence of a disul�de moiety. An aziridine linkage was used to connect
the fullerene to the organosulfur since the chemistry of functionalizing C60 with an
aziridine group is well established.[69] The disul�de moeity was chosen over that of
a thiol because aziridines and thiols are uncompatible.[70] Additionally, the use of
an unsymmetrical disul�de provides for a "tunability" of the interfacial properties by
making possible the replacement of the methyl group with another substituent.
N
S
S
CH3
Figure 4.6: An illustration of the structure of the short-chain fullerene
disul�de.
54
Scanning tunneling microscopy has previously been applied to characterize fullerene
thin �lms on a wide variety of substrates formed by evaporation of C60 under UHV
conditions. To our knowledge, however, there have been no reports of STM studies
of fullerene-terminated SAMs on gold. Figure 4.7 shows a large area image of a SAM
derived from the short-chain fullerenes on atomically at Au(111). (This and the
following image have been median �ltered along the slow scan direction to remove
noise.) Films were prepared by a 24 hour immersion of gold on mica substrates in
0.01 mM benzene solutions of the derivatized fullerene molecules. This image demon-
strates a high coverage of surface-con�ned moieties, but no evidence of ordering or
any discernible overlayer structure.
A more detailed image of the SAM is shown in Fig. 4.8. The fullerene moieties
can be readily distinguished with a nearest-neighbor distance of �11 �A in the most
densely packed regions. This corresponds well to the measured van der Waals di-
ameter of underivatized C60 adsorbates obtained in previous STM measurements.
Through cross-sectional analysis of these images, it appears that the fullerene moiety
is extended above the gold substrate by the alkanethiolate tail. This is supported by
the fact that stable images of these molecules can only be obtained at very low tun-
neling currents, consistent with imaging of alkanethiol SAMs and inconsistent with
the imaging of unfunctionalized C60 molecules adsorbed on metal surfaces.
In this section, it has been shown that C60 can be functionalized via an aziridine
linkage in the presence of a disul�de group. The resultant unsymmetrical disul�des
of variable alkane chain length can be used to generate self-assembled �lms on gold.
Imaging of the �lms by STM shows a high coverage of discrete fullerene moieties.
However, the disordered nature of the �lm prevents an exact determination of their
55
Figure 4.7: A large area STM image (76 X 76 nm) of SAMs generated
from the short-chain C60 on Au(111). It was acquired at +1.0 V sample
bias and 50 pA tunneling current.
56
Figure 4.8: A high resolution STM image (18 X 18 nm) of the same SAM.
The scan parameters for this image were +700 mV sample bias and 80 pA
tunneling current.
57
molecular orientation. In fact, it is quite likely that the molecules are adsorbed in a
variety of con�gurations due to the strong interaction of C60 with gold surface and
each other, and the highly asymmetric nature of the molecule. Pure �lms formed from
the long-chain fullerenes displayed a similar structure. Therefore, the following section
will describe the insertion of these molecules in various preassembled alkanethiolate
SAMs in order to create more structured monolayers.
4.3 Insertion into Alkanethiol SAMs
Instead of pure derivatized fullerene SAMs, this section describes the investigation
of the physical and electronic properties of the disul�de fullerenes when they are
incorporated into premade alkanethiolate monolayers. The generation of heteroge-
neous SAMs is in uenced by a number of factors including the terminal functional
groups, the chain substituents, and the molecular length. Pure solutions of the disul-
�de fullerene generate heterogeneous SAMs upon adsorption due to their asymmetry.
However, in order to form a more uniform monolayer, SAMs were investigated which
were generated by adsorbing alkanethiols onto the gold surface and then inserting
the derivatized fullerenes into this matrix. In the previous insertion experiments,
conjugated organic molecules were inserted into boundaries between structural do-
mains and at substrate step edges of preassembled alkanethiolate monolayers.[71, 72]
They demonstrated that this technique preserves the crystalline lattice of the alka-
nethiol SAM as opposed to co-deposition which does not. In conjunction with this,
the electronic e�ects on the states of the C60 molecule due its proximity with the gold
substrate were studied with scanning tunneling spectroscopy. Furthermore, we believe
58
that this represents the �rst case of an insertion of a disul�de into an alkanethiolate
monolayer.
The alkanethiol matrices were prepared by immersing the gold substrates for 24
hours in 1 mM ethanol solutions. The samples were then removed and rinsed repeat-
edly with ethanol before a second, one hour immersion in 0.01 mM disul�de fullerenes
dissolved in benzene. After removal, the samples went through multiple benzene, then
ethanol rinses and were then dried with nitrogen before scanning in the STM.
Figure 4.9 is an STM image of a hexanethiol SAM after immersion in a solution of
the short-chain C60 molecules. The white protrusions in this image are the tethered
fullerenes, while the black holes are due to etch pits formed during the self-assembly
process of one gold monolayer depth.[65, 73] The atomic lattice of the underlying
hexanethiol monolayer is clearly observed, demonstrating that the lattice of the SAM
is preserved after the insertion process. As well as insertion near step edges, the
fullerene molecules can also be found incorporated into the hexanethiol matrix itself.
Presumably, these molecules are inserted at grain boundaries in the hexanethiol lat-
tice. However, in some cases the size of the C60 prevents a clear identi�cation of the
surrounding structure. In addition, we observe that the molecules can be inserted in
small clusters as well as individually. It should be noted that the asymmetric shape
of the fullerene moieties in this and the following images is due to convolution with
the STM tip structure.
In Fig. 4.10, long-chain C60 molecules are imaged by the STM in an undecanethiol
matrix. In this case, we see a similar density (same order of magnitude) of inserted
molecules as compared to the previous case. This result is somewhat surprising given
that the steric interaction between the undecanethiols is almost twice that of the
59
Figure 4.9: An image of short-chain C60 disul�de inserted into a hex-
anethiol SAM. The bright areas are the fullerene moeity protruding above
the thiols, while the dark regions are etch pits in the gold surface due to
the self-assembly process.
60
hexanethiols and might therefore greatly decrease the number of inserted molecules.
This observation will be discussed in greater detail shortly. It appears that SAMs
consisting of these molecules have a decreased tendency towards clustering of the
inserted molecules as compared to the hexanethiol case.
We also explored the insertion of the disul�de fullerenes in alkanethiol SAM with
di�ering numbers of methylene spacers as compared to those of the disul�des. The in-
sertion of the long-chain C60 molecules into a dodecanethiol SAM is shown in Fig. 4.11.
Here we observe an order of magnitude decrease in the number of inserted molecules
as compared to the previous case. In addition, we observe that insertion occurs al-
most exclusively at the terrace steps or at the edges of the etch pits. This di�erence
cannot be explained due to the increase in steric interaction going from undecane to
Figure 4.10: Image of long-chain C60 inserted into undecanethiol SAM.
61
dodecane because of the similar inserted molecule densities reported for cases in the
hexane and undecane matrices. Therefore, the two other possible explanations are
that the shorter chain length of the long-chain C60 is hindering contact with the gold
surface or else the odd/even di�erences of the methyl group orientation is hindering
the incorporation of the methyl half of the disul�de into the dodecanethiol matrix.[74]
To try and distinguish between these two e�ects, we also performed the insertion of
the long-chain C60 molecules in a decanethiol SAM.
The results of inserting the long-chain C60 molecule into a decanethiol SAM are
shown in Fig. 4.12. In this way, the di�erences in the orientation of the methyl
moiety are maintained, except now the tether on the C60 is longer than the underlying
Figure 4.11: Image of long-chain C60 inserted into dodecanethiol SAM.
This image has been processed by unsharp masking to ehance the visibility
of the inserted molecules.
62
alkane matrix. As compared to the previous results, we �nd that the density of
inserted molecules is approximately equal to the cases of the derivatized fullerenes
inserted into like alkanethiol matrices and much greater than the insertion of the
long-chain fullerenes into dodecanethiol. This result tends to support the argument
that the relative chain lengths are the dominant mechanism in the insertion process.
However, it is interesting to point out that with the aziridine linkage, the tether on
the C60 half of the molecule is nearly equal in length to an alkanethiol with an extra
methylene spacer. Yet, the associated reduction of the density of inserted molecules
seems to indicate that the other half of the disul�de also plays an important role in
the incorporation of these molecules into the alkanethiol SAM.
Figure 4.12: Image of long-chain C60 inserted into decanethiol SAM.
63
Besides the insertion dynamics, we also investigated the electronic properties of the
inserted molecules through scanning tunneling spectroscopy. The insertion method
has been previously shown to be ideal for studying the electronic properties of iso-
lated molecules.[71] Displayed in Fig. 4.13 are the current versus voltage (I-V) curves
taken both on and away from the inserted disul�de molecules for both the insertion
of short-chain and long-chain molecules in alkanethiol matrices formed from the same
molecular constituents. The nearly linear characteristic of the alkanethiols spectra are
in good agreement with previously published I-V data acquired under the same condi-
tions for pure alkanethiol SAM.[75, 76] The similar curves obtained from the di�erent
alkanethiols can be explained by the fact that tunneling through these molecules oc-
curs nonresonantly with a decay constant similar to that of free space and that the
I-V curves were acquired under similar gap impedances.[77]
The �2 eV bandgap measured for the functionalized C60 is similar to the spectra
of a second monolayer of C60 molecules on a gold surface and is drastically di�erent
than that for a single C60 molecule on gold.[78] In that case, the electronic di�erence
between the �rst and second fullerene monolayers was due to their charge transfer
interaction with the gold surface. In contrast, our results indicate that there is very
little coupling between the molecular levels of the derivatized fullerenes and the un-
derlying gold surface. This indicates that unlike previous fullerene monolayers formed
on metal surfaces, the disul�de fullerenes retain the semiconducting properties of bulk
C60. The symmetric character of the functionalized C60 can be explained in terms
of the chemical potential and applied bias voltage, and is similar to that of other
molecular SAM.[79] The symmetry of the states is a re ection of how e�ectively the
molecular energy levels are a�ected by the metal reservoir of electrons. These results
64
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-30
-20
-10
0
10
20
30
HexanethiolShort-chain C60Long-chain C60Undecanethiol
Sample Bias (V)
Tun
nelin
g C
urre
nt (
pA)
Figure 4.13: Scanning tunneling spectroscopy curves obtained in various
locations on the thiol �lms. The gap impedance was 2 x 1010 ohms (set at
+1 V bias voltage and 50 pA tunneling current).
65
indicate that the alkane chain functions as a very e�ective insulating layer indepen-
dent of the chain length. This e�ect has been seen in a similar alkane molecule SAM
on Si,[80] where they discovered that the leakage current was also independent of
monolayer thickness. However, the shortest alkane chain investigated in that study
was dodecane. We demonstrate that the same insulating characteristics still apply
for alkane chains of half that length.
In summary, we have investigated the adsorption dynamics of C60 molecules
derivatized by disul�des in pure and mixed SAM on Au(111) with the STM. We have
demonstrated that the disul�de fullerene molecules can be inserted into alkanethiol
matrices of varying chain length. From these experiments we observe di�erences in the
adsorption dynamics that depend on the length of the disul�de chain relative to the
underlying alkanethiol matrix. The insertion technique also provides a method for the
electrical characterization of isolated tethered fullerenes. The tunneling spectroscopy
of these molecules indicate that the C60-derivatized molecules maintain their semicon-
ducting nature even when the alkane spacing between the C60 functional group and
the metal surface is reduced to six methylene units. By using a molecular semicon-
ductor instead of a bulk one, this work represents an important step in the creation of
an all organic electronic device. Utilizing well-developed annealing and replacement
techniques associated with other SAMs, this molecular system opens the possibility
of generating well ordered uniform monolayers of functionalized C60 molecules. In
particular, a uniform monolayer of the disful�de fullerenes could serve as the molecu-
lar layer in a Mott transition �eld e�ect transistor.[81] The advantages of this include
the high dielectric constant of the alkane chains, the good ohmic contact obtained at
the C60-metal interface, and the fact that both properties are contained in a single,
66
self-assembling molecule. This technique should also prove useful for probing the
electronics e�ects of other chemical modi�cations made to fullerene molecules.
67
Chapter 5
Future Directions
The preceeding chapters have demonstrated a robust theoretical and experimental
understanding of graphite and fullerene structures on the nanometer scale. One area
where this research can �nd immediate application is the analysis of bare and function-
alized single-walled carbon nanotubes. Due to the unique structural and electronic
characteristics, carbon nanotubes will surely become one of the essential building
blocks of nanotechnology. However, electronic and structural modi�cations will prob-
ably be necessary to assemble them into functional devices. Unfortunately, the size of
the nanotubes limits the feasibility of nonlocal measurements such as x-ray di�action
and Auger electron spectroscopy to characterize these structures. Scanning tunneling
microscopy, on the other hand, has already proven very useful in characterizing bare,
single and multi-walled nanotubes and should do so for functionalized nanotubes as
well. Lastly, the nanotubes similarity to graphite and C60 makes research in this
thesis very relevant to these investigations.
As Fig. 5.1 demonstrates, the theoretical model in Chapter 2 can be easily ex-
tended to accommodate defect structures in carbon nanotubes. This is a theoretical
image of a vacancy defect in a (10,10) nanotube. One extra step as compared to the
graphite calculations is performed. In a simple approximation, the two dimensional
charge density plot is warped around a cylinder to account for the curvature of the
nanotube. The only other change from the graphite cluster is a modi�cation of the
68
boundary conditions. For the case of Fig. 5.1 the boundary conditions were chosen to
simulate a (10,10) nanotube, however, it is a straightforward matter to modify them
for other tube chiralities.
To demonstrate the ability of the STM to characterize functionalized tubes on the
nanometer scale, preliminary investigations have already been performed on uori-
nated nanotubes. Figure 5.2 shows an STM image of SWNT uorinated at 250 oC.
It is believed that the bright areas along the nanotube are regions of high uorine
coverage, while the dark regions are sections of bare nanotube. Assuming a (10,10)
nanotube, the perpendicular banding of the uorine coverage suggests that uori-
nation proceeds via 1,2 cycloaddition. This demonstrates that even without atomic
resolution, the STM o�ers insight that is not possible with any other technique.
Figure 5.1: Front and back theoretical images of a vacancy in a (10,10)
nanotube.
69
Figure 5.2: STM image of uorinated nanotubes on Au(111).
70
Bibliography
[1] E. S. Snow, P. M. Campbell, and P. J. McMarr, Appl. Phys. Lett. 63, 749 (1993).
[2] D. M. Eigler and E. K. Schweizer, Nature 344, 524 (1990).
[3] I.-W. Lyo and P. Avouris, Science 253, 173 (1991).
[4] Y. Hasegawa and P. Avouris, Phys. Rev. Lett. 71, 1071 (1993).
[5] M. F. Crommie, C. P. Lutz, and D. M. Eigler, Nature 363, 524 (1993).
[6] P. T. Sprunger, L. Peterson, E. W. Plummer, E. Laegsgaard, and F. Besenbacher,Science 275, 1764 (1997).
[7] M. C. M. M. van der Wielen, A. J. A. van Roij, and H. van Kempen, Phys. Rev.Lett. 76, 1075 (1996).
[8] M. M. Kamna, S. J. Stranick, and P. S. Weiss, Science 274, 118 (1996).
[9] H. A. Mizes and J. S. Foster, Science 244, 559 (1989).
[10] J. P. Rabe, M. Sano, D. Batchelder, and A. A. Kalatchev, J. Microsc. 152, 573(1988).
[11] J. Xhie, K. Sattler, U. M�uller, N. Venkateswaran, and G. Raina, Phys. Rev. B43, 8917 (1991).
[12] K. F. Kelly, D. Sarkar, G. D. Hale, S. J. Oldenburg, and N. J. Halas, Science273, 1371 (1996).
[13] K. F. Kelly, Functionalizing scanning tunneling microscope probes with buck-minsterfullerenes, Masters, Rice University, 1996.
[14] M. R. Soto, J. Microsc. 152, 779 (1988).
[15] J. Kushmerick, K. F. Kelly, H.-P. Rust, N. J. Halas, and P. S. Weiss, J. Phys.Chem. B 103, 1619 (1999), in press.
[16] J. Terso� and D. R. Hamann, Phys. Rev. Lett. 50, 1998 (1983).
[17] J. Terso� and N. D. Lang, Phys. Rev. Lett. 65, 1132 (1990).
[18] D. Tom�anek, S. G. Louie, H. J. Mamin, D. W. Abraham, R. E. Thomson,E. Ganz, and J. Clarke, Phys. Rev. B 35, 7790 (1987).
71
[19] S. Gwo and C. K. Shih, Phys. Rev. B 47, 13059 (1993).
[20] R. J. Hamers, R. M. Tromp, and J. E. Demuth, Phys. Rev. B 34, 5343 (1986).
[21] K. Nordlund, J. Keinonen, and T. Mattila, Phys. Rev. Lett. 77, 699 (1996).
[22] L. Chico, L. X. Benedict, S. G. Louie, and M. L. Cohen, Phys. Rev. B 54, 2600(1996).
[23] M. Schmid, W. Hebenstreit, P. Varga, and S. Crampin, Phys. Rev. Lett. 76,2298 (1996).
[24] C. Wittneven, R. Dombrowski, M. Morgenstern, and R. Wiesendanger, Phys.Rev. Lett. 81, 5616 (1998).
[25] L. Petersen, P. Laitenberger, E. L�gsgaard, and F. Besenbacher, Phys. Rev. B58, 7361 (1998).
[26] J. Resh, D. Sarkar, J. Kulik, J. Brueck, A. Ignatiev, and N. J. Halas, Surf. Sci.316, L1601 (1994).
[27] K. F. Kelly, D. Sarkar, S. Prato, J. S. Resh, G. D. Hale, and N. J. Halas, J. Vac.Sci. Technol. B 14, 593 (1996).
[28] S. M. Lee, Y. H. Lee, Y. G. Hwang, J. R. Hahn, and H. Kang, Phys. Rev. Lett.82, 217 (1999).
[29] B. Burk, R. E. Thomson, J. Clarke, and A. Zettl, Science 257, 362 (1992).
[30] T. W. Parks and J. H. McClellan, IEEE Trans. Circuit Theory CT-19, 189(1972).
[31] J. S. Lim, Two-dimensional signal and image processing, Prentice Hall, 1990.
[32] R. J. Lagow, R. B. Badachape, J. L. Wood, and J. L. Margrave, J. Chem. Soc.,Dalton Trans. 12, 1268 (1974).
[33] P. Kamarchik and J. L. Margrave, Acc. Chem. Res. 11, 296 (1978).
[34] M. S. Dresselhaus, M. Endo, and J.-P. Issi, Physical properties of uorine- and uoride-graphite intercalation compounds, in Fluorine-Carbon and Fluoride-Carbon Materials, edited by T. Nakajima, page 95, Marcel-Dekker, New York,1995.
[35] G. M. Shedd and P. E. Russell, J. Vac. Sci. Technol. A 9, 1261 (1991).
[36] J. R. Hahn, H. Kang, S. Song, and I. C. Jeon, Phys. Rev. B 53, R1725 (1996).
[37] B. G. Briner, P. Ho�man, M. Doering, H.-P. Rust, E. W. Plummer, and A. M.Bradshaw, Phys. Rev. B 58, 13931 (1998).
[38] T. Mallouk and N. Bartlett, Chem. Commun. , 103 (1983).
72
[39] S. L. di Vittorio, M. D. Dresselhaus, and G. Dressehaus, J. Mater. Res. 8, 1578(1993).
[40] A. Tressaud, F. Moguet, S. Flandrois, M. Chambon, C. Guimon, G. Nanse,E. Papirer, V. Gupta, and O. P. Bahl, J. Phys. Chem. Solids 57, 745 (1996).
[41] A. M. Panich, T. Nakajima, and S. D. Goren, Chem. Phys. Lett. 271, 381 (1997).
[42] E. T. Mickelson, C. B. Hu�man, A. G. Rinzler, R. E. Smalley, R. H. Hauge, andJ. L. Margrave, Chem. Phys. Lett. 296, 188 (1998).
[43] K. F. Kelly and N. J. Halas, Surf. Sci. 416, L1085 (1998).
[44] J. H. Ferris, J. G. Kushmerick, J. A. Johnson, M. G. Y. Youngquist, R. B.Kessinger, H. W. Kingsbury, and P. S. Weiss, Rev. Sci. Instrum. 69, 2691(1998).
[45] J. A. Meyer, S. J. Stranick, J. B. Wang, and P. S. Weiss, Ultramicroscopy 42-44,1538 (1992).
[46] H. A. Mizes, Interpretation of Scanning Tunneling Microscopy Images ofGraphite, Phd, Stanford University, 1988.
[47] An analogous argument in terms of hole tunneling can explain the observationof anisotropic scattering at low negative bias voltages.
[48] G. M. Shedd and P. E. Russell, Surf. Sci. 266, 259 (1992).
[49] H. Kroto, J. R. Heath, S. C. O`Brien, R. Curl, and R. Smalley, Nature 318, 162(1985).
[50] A. Ulman, A. Chem. Rev. 96, 1533 (1996).
[51] K. K. Berggren, A. Bard, J. L. Wilbur, J. D. Gillaspy, A. G. Helg, J. J. McClel-land, S. L. Rolston, W. D. Phillips, M. Prentiss, and G. M. Whitesides, Science269, 1255 (1995).
[52] R. J. Jackman, J. L. Wilbur, and G. M. Whitesides, Science 269, 664 (1995).
[53] R. P. Andres, J. D. Bielefeld, J. I. Henderson, D. B. Janes, V. R. Kolagunta,C. P. Kubiak, W. J. Mahoney, and R. G. Osifchin, Science 273, 1690 (1996).
[54] J. Collet and D. Vuillaume, Appl. Phys. Lett. 73, 2681 (1998).
[55] R. G. Nuzzo and D. L. Allara, J. Am. Chem. Soc. 105, 4481 (1993).
[56] A. Ulman, Introduction to Thin Organic Films: From Langmuir-Blodgett toSelf-Assembly, Academic Press, Boston, 1991.
[57] C. W. Sheen, J. X. Shi, J. Martensson, A. N. Parikh, and D. L. Allara, J. Am.Chem. Soc. 114, 1514 (1992).
73
[58] C. D. Bain, E. B. Troughton, Y.-T. Tao, J. Evall, G. M. Whitesides, and R. G.Nuzzo, J. Am. Chem. Soc. 111, 321 (1989).
[59] L. H. Dubois and R. G. Nuzzo, Annu. Rev. Phys. Chem. 43, 437 (1992).
[60] L. Strong and G. M. Whitesides, Langmuir 4, 546 (1988).
[61] C. E. D. Chidsey and D. N. Loiacono, Langmuir 6, 709 (1990).
[62] L. H. Dubois, B. Zegarski, and R. G. Nuzzo, J. Chem. Phys. 98, 678 (1993).
[63] P. S. Weiss, H. Yokota, R. Aebersold, G. van den Engh, L. A. Bumm, J. J.Arnold, T. D. Dunbar, and D. L. Allara, J. Phys. Condens. Matter 10, 7703(1998).
[64] N. Camillone, C. E. D. Chidsey, G.-Y. Liu, and G. Scoles, J. Chem. Phys. 98,3503 (1993).
[65] G. E. Poirier and M. J. Tarlov, Langmuir 10, 2853 (1994).
[66] E. Delamarche, B. Michel, C. Gerber, D. Anselmetti, G�untherodt, H. Wolf, andH. Ringsdorf, Langmuir 10, 2869 (1994).
[67] H. Sellers, A. Ulman, Y. Shnidman, and J. E. Eilers, J. Am. Chem. Soc. 115,9389 (1993).
[68] Y.-S. Shon, K. F. Kelly, N. J. Halas, and T. R. Lee, Langmuir , submitted.
[69] M. Prato, Q. C. Li, and F. Wudl, J. Am. Chem. Soc. 115, 1148 (1993).
[70] V. R. Gaertner, J. Heterocycl. Chem. 8, 519 (1971).
[71] L. A. Bumm, J. J. Arnold, M. T. Cygan, T. D. Dunbar, T. P. Burgin, L. Jones II,D. L. Allara, J. M. Tour, and P. S. Weiss, Science 271, 1705 (1996).
[72] M. T. Cygan, T. D. Dunbar, J. J. Arnold, L. A. Bumm, N. F. Shedlock, T. P.Burgin, L. Jones, D. L. Allara, J. M. Tour, and P. S. Weiss, J. Am. Chem. Soc.120, 2721 (1998).
[73] C. Sch�onenberger, J. A. M. Sondag-Huethorst, J. Jorritsma, and L. G. J.Fokkink, Langmuir 10, 612 (1994).
[74] P. E. Laibinis, G. M. Whitesides, D. L. Allara, Y.-T. Tao, A. N. Parikh, andR. G. Nuzzo, J. Am. Chem. Soc. 113, 7152 (1991).
[75] M. Dorogi, J. Gomez, R. Osifchin, R. P. Andres, and R. Reifenberger, Phys.Rev. B 52, 9071 (1995).
[76] W. Mizutani, T. Ishida, and H. Tokumoto, Langmuir 14, 7197 (1998).
[77] P. S. Weiss, L. A. Bumm, T. D. Dunbar, J. M. Tour, and D. L. Allara, AnnalsNew York Acad. Sciences 852, 145 (1998).
74
[78] J. K. Gimzewski, S. Modesti, and R. R. Schlittler, Phys. Rev. Lett. 72, 1036(1994).
[79] S. Datta, W. Tian, S. Hong, R. Reifenberger, J. I. Henderson, and C. P. Kubiak,Phys. Rev. Lett. 79, 2530 (1997).
[80] C. Boulas, J. V. Davidovits, F. Rondelez, and D. Vuillaume, Phys. Rev. Lett.76, 4797 (1996).
[81] C. Zhou, D. M. Newns, J. A. Misewich, and P. C. Pattnaik, Appl. Phys. Lett.70, 598 (1997).
75
Appendix A
Curricula Vitae
Kevin F. KellyRice University
Dept. of Electrical & Computer
Engineering
Houston, TX 77251-1892
(713) 527-8750, ext. 2329
6516 Travis
Houston, TX 77030
(713) 520-1621
INTERNET: [email protected]
Education
� Rice UniversityPhD research in progressThesis Research: Scanning tunneling microscopy of graphite point defectswith metal and molecular STM probes
� Rice UniversityMS in Applied Physics, May 1996Thesis Title: Functionalization of Scanning Tunneling Microsocpe Probes withBuckminsterfullerenesThesis Advisor: Dr. N.J. Halas
� Colorado School of MinesBS in Engineering Physics, June 1991
� Sterling High SchoolSterling, Colorado, May 1989
Professional Memberships
1. American Physical Society
2. American Association for the Advancement of Science
3. American Vacuum Society
4. American Chemical Society
Career Goals
� To continue research and development in the area of nanotechnology and molec-ular electronics
Research Experience
� Ion Implantation
� Scanning Probe Microscopy
� Thin Film Growth
� One and Two Dimensional Signal Processing
Honors and Awards
� Tau Beta Pi Engineering Honor Society
� Sigma Pi Sigma Physics Honor Society
� Robert A. Welch Foundation Predoctoral Fellow, 1994 - present
Relevant Job Experience
� Rice UniversityResearch Assistant in Electrical Engineering Department, 1993 - present
� Colorado School of MinesResearch Assistant, August 1992 - August 1993
Publications
1. K. F. Kelly, S. Prato, J. Resh, D. Sarkar, J. Hovis, and N. J. Halas, \ImagingFullerene Nanotips", Technical Report 95-06, Department of Electrical andComputer Engineering.
2. K. F. Kelly, D. Sarkar, S. Prato, J. Resh, G. Hale, and N. J. Halas, \DirectObservation of Fullerene-Adsorbed Tips by STM", J. Vac. Sci. Tech. B 14,593 (1996).
3. K. F. Kelly, G. Hale, D. Sarkar, and N. J. Halas, \Threefold Electron Scatteringon Graphite Observed with C60-Adsorbed STM Tips", Science 273, 1371 (1996).
4. K. F. Kelly, D. Sarkar. S. J. Oldenburg, G. D. Hale, and N. J. Halas, \FullereneFunctionalized Scanning Tunneling Microscope Tips- Preparation, Characterizationand Applications", Synthetic Metals 86, 2407-2410 (1997).
5. K. F. Kelly, D. Sarkar, S. J. Oldenburg, G. D. Hale, and N. J. Halas, \FullereneTips for Scanning Probe Microscopy", SPIE Proceedings, 2854, 114-121 (1996).(ref-ereed)
6. K. F. Kelly and N. J. Halas, \Di�erentiation of alpha-site and beta-site defectson a graphite surface", Surface Science 416, L1085 (1998).
7. J. G. Kushmerick, K. F. Kelly, H.-P. Rust, N. J. Halas, and P. S. Weiss,"Observations of Anisotropic Electron Scattering on Graphite with a Low-Temperature Scanning Tunneling Microscope", Journal of Physical ChemistryB bf103, 1619 .
8. Y.-S. Shon, K. F. Kelly, N. J. Halas, and T. R. Lee, "Fullerene-TerminatedAlkanethiolate SAMs on Gold Generated fromUnsymmetrical Disul�des", Langmuir,submited.
9. K. F. Kelly, E. T. Mickelson, R. H. Hauge, J. L. Margrave, and N. J. Halas,"Nanoscale Imaging of Chemical Interactions: Fluorine on Graphite", PhysicalReview Letters, submitted.
10. K. F. Kelly, Y. S. Shon, T. R. Lee, and N. J. Halas, "Scanning TunnelingMicroscopy and Spectroscopy of Dialkyl Disul�de Fullerenes Inserted into AlkanethiolateSAMs", J. Phys. Chem. B, submitted.
11. K. F. Kelly, E. T. Mickelson, R. H. Hauge, J. L. Margrave, R. E. Smalley,and N. J. Halas, "Scanning Tunneling Microscopy of Functionalized CarbonNanotubes, manuscript in preparation.
Conference Presentations
1. \Imaging Fullerene Nanotips" K. F. Kelly*, J. S. Hovis, D. Sarkar, S. Prato, J.S. Resh, and N. J. Halas, poster, Rice Summer Research Colloquium, Houston,TX, August 1994.
2. \Imaging Fullerene Nanotips", K. F. Kelly, S. Prato, D. Sarkar, J. Resh, andN. J. Halas, American Physical Society Meeting, Division of Condensed MatterPhysics, postdeadline poster, San Jose, CA, March 1995.
3. \Fullerene STM Tips: Characterization and Application" K. F. Kelly*, D.Sarkar, G. D. Hale, S. J. Oldenburg, and N. J. Halas, contributed talk, RiceSummer Research Colloquium, Houston, TX, August 1995.
4. \Preparation, Characterization and Imaging Applications of Fullerene STMTips", K. F. Kelly*, D. Sarkar, G. D. Hale, S. J. Oldenburg, and N. J. Halas,contributed talk at the American Physical Society March Meeting, St. Louis,MO, March 1996.
5. \Fullerene STM Tips", K. F. Kelly and N. J. Halas, invited talk at the ElectrochemicalSociety Meeting, Los Angeles, CA, May 1996.
6. \Fullerene Functionalized Scanning Tunneling Microscope Tips: Preparation,Characterization, and Applications", K. F. Kelly and N. J. Halas, invited talkat the International Conference on Synthetic Metals, Salt Lake City, July 1996.
7. \Preparation, Characterization and Imaging Applications of C60 STM Tips",K. F. Kelly and N. J. Halas, invited talk at the SPIE Meeting, August 1996.
8. \Preparation of Submonlayer Nanoparticles for SPM Imaging" S. J. Oldenburg,K. F. Kelly, and N. J. Halas, poster, Rice Summer Research Colloquium,Houston, TX, August 1996.
9. \Imaging Electron Scattering at Graphite Point Defects using Fullerene STMTips", K. F. Kelly* and N. J. Halas, invited talk at the Texas Section of theAmerican Physical Society Meeting, Arlington, TX, October 1996.
10. \Functionalization of STM Tips with C60", K. F. Kelly and N. J. Halas, con-tributed talk at the American Vacuum Society 43rd National Symposium, Philedelphia,PA, October 1996.
11. \Theory of Electron Scattering at Point Defects on Graphite Surfaces", K. F.Kelly* and N. J. Halas, contributed talk at the American Physical Society MarchMeeting, Kansas City, MO, March 1997.
12. \Imaging of Point Defects on Graphite Surfaces Using C60-adsorbed STM tips",K. F. Kelly* and N. J. Halas, contributed poster at STM 97, Hamburg, Germany,July 1997.
13. \Imaging of Alpha-site and Beta-site Defects on Graphite Surfaces", K. F.Kelly* and N. J. Halas, contributed talk, Rice Summer Research Colloquium,Houston, TX, August 1997.
14. "Characterization of Localized Defect Sites on Graphite Surfaces", K. F. Kelly*and N. J. Halas, contributed talk at the American Physical Society MarchMeeting, Los Angeles, CA, March 1998.
15. "Anisotropic Electron Scattering from Point Defects on Graphite at Low Temperature",K. F. Kelly, J. G. Kushmerick, H.-P. Rust, N. J. Halas, and P. S. Weiss, con-tributed poster at the American Vacuum Society Meeting, Baltimore, MD,November 1998.
16. "Functionalized Fullerene Self-Assembled Monolayers Studied by Scanning TunnelingMicroscopy", K. F. Kelly*, Y.-S. Shon, T. R. Lee, and N. J. Halas, contributedtalk, Rice Summer Research Colloquium, Houston, TX, August 1998.
17. \Functionalized Fullerene SAMs Investigated by STM", K. F. Kelly, Y.-S. Shon,T. R. Lee, and N. J. Halas, contributed talk at the American Physical SocietyMarch Meeting, Atlanta, GA, March 1999.
* Presenter