Theoretical Characterization of
Optical Processes in Molecular
Complexes
Kai Liu
Department of Theoretical Chemistry
School of Biotechnology
Royal Institute of Technology
Stockholm, Sweden 2008
c© Kai Liu, 2008
ISBN 978-91-7178-857-3
Printed by Universitetsservice US AB, Stockholm, Sweden, 2008
3
Abstract
The main theme of this thesis is to study effects of different environments on geometric and
electronic structures, as well as optical responses, of molecules using time-(in)dependent
density functional theory. Theoretical calculations have been carried out for properties that
can be measured by conventional and advanced experimental techniques, including one-
photon absorption (OPA), two-photon absorption (TPA), surface-enhanced Raman scatter-
ing (SERS) and second order nonlinear optical (NLO) response. The obtained good agree-
ment between the theory and the experiment allows to further extract useful information
about inter- and intra-molecular interactions that are not accessible experimentally.
By comparing calculated one-photon absorption spectra of aluminum phthalocyanine chlo-
ride (AlPcCl) and AlPcCl -water complexes with the corresponding experiments, detailed
information about the interaction between water molecules and AlPcCl, and geometric
changes of AlPcCl molecule has been obtained. Effects of aggregation on two-photon ab-
sorption spectra of octupolar molecules have been examined. It is shown that the formation
of clusters through inter-molecular hydrogen bonding can drastically change profiles of TPA
spectra. It has also demonstrated that a well designed molecular aggregate/cluster, den-
drimer, can enhance the second order nonlinear optical response of the molecules. In collabo-
ration with experimentalists, a series of end-capped triply branched dendritic chromophores
have been characterized, which can lead to large enhancement of the second order NLO
property when the dipoles of the three branches in the dendrimers are highly parallelized.
Surface-enhanced Raman scattering has made the detection of single molecules on metal
surface become possible. Chemically bonded molecule-metal systems have been extensively
studied. We have shown in a joint experimental and theoretical work that stable Raman
spectra of a non-bonding molecule, perylene, physically adsorbed on Ag nano-particles can
also be observed at low temperature. It is found that the local enhanced field has a tendency
to drive molecule toward a gap of two closely lying nano-particles. The trapped molecule
can thus provide a stable Raman spectrum with high resolution when its thermal motion is
reduced at low temperature.
For the ever growing size of molecular complexes, there is always the need to develop
new computational methods. A conceptually simple but computationally efficient method,
named as central insertion scheme (CIS), is proposed that allows to calculate electronic
structure of quasi-periodic system containing more than 100,000 electrons at density func-
tional theory levels. It enables to monitor the evolution of electronic structure with respect
to the size of the system.
4
Preface
The work presented in this thesis has been carried out at the Department of Theoretical
Chemistry, School of Biotechnology, Royal Institute of Technology, Stockholm, Sweden.
List of papers included in the thesis
Paper I Origin of the Q-band splitting in the absorption spectra of aluminum phthalo-
cyanine chloride, K. Liu, Y.-H. Wang, J.N. Yao, and Y. Luo, Chem. Phys. Lett., 438
(2007) 36.
Paper II Aggregation effects on two-photon absorption spectra of octupolar molecules,
K. Liu, Y.-H. Wang, Y.Q. Tu, H. Agren, and Y. Luo, J. Chem. Phys., 127 (2007) 026101.
Paper III Two-photon absorption of hydrogen bonded octupolar molecular clusters, K.
Liu, Y.-H. Wang, Y.Q. Tu, H. Agren, and Y. Luo, J. Phys. Chem. A, 00 (2008) 000.
Paper IV Hyperpolarizabilities of end-capped triply branched dendrimers: a theoretical
study, Y.H. Wang, K. Liu, J. Heck, H.G. Kuball, H. Agren, and Y. Luo, J. Chem. Phys.,
submitted.
Paper V Controlling single molecular Raman behavior of a non-bonding molecule on Ag
nanoparticles, Z.X. Luo, Y. Luo, J. Li, K. Liu, H.B. Fu, Y. Ma, and J.N. Yao, submitted.
Paper VI Raman spectra of a single perylene molecule on silver clusters, K. Liu, Z.X.
Luo, J.N. Yao, and Y. Luo, in manuscript.
Paper VII An efficient first-principle approach for electronic structure calculations of
nanomaterials, B. Gao, J. Jiang, K. Liu, Z. Wu, W. Lu, and Y. Luo, J. Comput. Chem.,
29 (2008) 434.
5
List of papers not included in the thesis
Paper VIII An elongation method for first principle simulations of electronic structures
and transportation properties of finite nanostructures, J. Jiang, K. Liu, W. Lu, and Y.
Luo, J. Chem. Phys., 124 (2006) 214711.
Paper IX Hierarchical chiral supra-molecular structures by the self-assembly of achiral
azobenzene derivative with Water, X. Sheng, G. Zhang, J. Wang, Y. Wang, H. Fu, X. Gao,
K. Liu, Y.H. Wang, Y. Luo, W. Yang, A. Peng, and J.N. Yao, J. Am. Chem. Soc.,
Submitted.
Paper X Structure-property relationship in organometallic compounds regarding SHG, J.
Heck, M. H. Prosenc, T. Meyer-Friedrichsen, J. Holtmann, E. Walczuk, M. Dede, T. Farrell,
A. R. Manning, H.-G. Kuball, G. Archetti, Y.-H. Wang, K. Liu, and Y. Luo, Proc. SPIE,
6653 (2007) 66530R
Paper XI Three-branched dendritic NLOphores, more than three times a single-strand
chromophore? J. Holtmann, E. Walczuk, M. Dede, C. Wittenburg, J. Heck, G. Archetti, R.
Wortmann, H.G. Kuball, Y.-H. Wang, K. Liu, and Y. Luo, Chem. Euro. J., submitted.
6
Comments on my contribution to the papers included
• I was responsible for calculations and writing of manuscripts for papers I, II, III and
VI.
• I was responsible for part of the calculations for papers IV and VII.
• I was responsible for most of the calculations for paper V.
7
Acknowledgments
First of all, I would like to give my greatest thanks to my supervisor Prof. Yi Luo for giving
me the opportunity to study in Sweden and for teaching me a lot for my research work.
And I should thank Prof. Luo and his wife Dr. Xing together for their kindly help in my
daily life.
I am heartily grateful to Prof. Hans Agren for creating such a warm atmosphere in the
Department of Theoretical Chemistry. I have really enjoyed my time studying here.
I would like to give my special thanks to Prof. Margareta Blomberg, Prof. Boris Minaev,
Prof. Faris Gel’mukhanov, Docent Fahmi Himo and Docent Pawel Salek for their excellent
courses. And I am deeply thankful to Dr. Jun Jiang, Bin Gao and Dr. Barbara Brena for
helping me in my work. I would like to thank all my friends for your sincere friendship.
Many thanks to the Swedish National Infrastructure for Computing (SNIC) for computer
time at national supercomputer facilities.
I would like to express my warmest thanks to my parents, my brother, my wife Dr. Yanhua
Wang and all the other close relatives for their love and support.
Kai Liu
2008-01
8
Contents
1 Introduction 11
2 Density Functional Theory and Response Functions 15
2.1 Density Functional Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.1 Hohenberg-Kohn Theorems . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.2 Kohn-Sham equations . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1.3 Exchange and Correlation Functionals . . . . . . . . . . . . . . . . . 17
2.2 Response Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2.1 Response Theory for Exact States . . . . . . . . . . . . . . . . . . . . 19
2.2.2 One-Photon Absorption . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.3 Two-Photon Absorption . . . . . . . . . . . . . . . . . . . . . . . . . 21
3 Linear and Nonlinear Optical Spectroscopy 23
3.1 One-photon Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 Two-photon Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 Selection Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.4 Hyperpolarizabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.4.1 Finite Field Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.4.2 Analytical Derivative Method . . . . . . . . . . . . . . . . . . . . . . 30
3.4.3 Response Theory Method . . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.5.1 Q-band Splitting of Aluminum Phthalocyanine Chloride . . . . . . . 31
9
10 CONTENTS
3.5.2 Two-Photon Absorption of Octupolar Molecules . . . . . . . . . . . . 32
3.5.3 One- or Two-end-capped Dendritic Structures . . . . . . . . . . . . . 34
4 Raman Spectroscopy 37
4.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.3 Surface Enhanced Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . 40
4.3.1 Electromagnetic Enhancement . . . . . . . . . . . . . . . . . . . . . . 41
4.3.2 Chemical Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5 Central Insertion Scheme 47
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.2 Central Insertion Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2.1 Basic Idea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2.2 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2.3 Central Insertion Process . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.2.4 Approximations and Error Control . . . . . . . . . . . . . . . . . . . 51
5.2.5 BioNano Lego . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Chapter 1
Introduction
If we knew what we were doing, it wouldn’t be called research, would it?
Albert Einstein
Since there were human being, people kept trying to understand and to change everything
around them. It is no doubt that many common phenomena can be studied by watching,
listening, touching, smelling and tasting directly. However, there are still enormous details
and ultimate truths behind the phenomena can not be obtained by these normal means.
A wide variety of experimental tools and techniques have been developed and applied for
different purposes. Molecular optical spectroscopy is one of such tools that is designed for
studying molecular structures and properties.
Optics is an old science started several centuries ago. It is a branch of physics that ex-
plains the behavior and properties of light and the interaction between light and matter.
The research area of optics covers a wide spectrum of light, including infrared, visible and
ultraviolet. The beginning of optical spectroscopy is marked by the first spectrum that
was obtained by Newton in 1666 with the help of a triangular prism. The spectrum was a
rainbow of colors generated by dispersing sunlight. By analysis of the emitted light, one can
immediately get information about temperature, mass and luminosity of an object. Nowa-
days, many modern spectrometers have been developed that allow to effectively collect high
resolution spectra of different systems. In 1859, Kirchhoff showed that elements or com-
pounds have their own unique spectra, which means that one can determine the chemical
composition of an unknown system by studying its spectrum. Those information usually
come from the positions and the intensities of spectral lines. Optical spectroscopy has been
heavily used in physics, analytical chemistry, astronomy and remote sensing since it reveals
11
12 CHAPTER 1. INTRODUCTION
microscopic information of the matter without imposing physical damage. It can also pro-
vide information about chemical surrounding and intra/inter-molecular interactions of the
studied systems by comparing spectra gathered under different conditions.
The interpretation of various spectra requires good knowledge of electronic structures and
properties of the given systems. Physical and chemical intuitions might be useful for under-
standing the spectra of simple and small systems, but they become quickly inefficient for
large and complicated systems. Theoretical modeling has played a significant role in the de-
velopment and application of various spectroscopic techniques. With the fast development
of computer technology and computational methods, nowadays it is possible to calculate
systems of very large size with good accuracy in a reasonable time. In comparison with
experiments, theoretical study usually costs much less and can provide massive information
that are normally not accessible experimentally. For instance, geometric structures of single
polyatomic molecules can nowadays be accurately determined by highly correlated theoreti-
cal methods, which can largely replace experimental measurements. However, for molecular
complexes or clusters the most powerful approach would be a combination of theoretical
modelings and experimental measurements. A good agreement between theoretical and ex-
perimental results for certain molecular properties provides not only good interpretation
and understanding of observed phenomena, but also useful inputs that enable to carry out
further theoretical modelings to obtain a more complete picture of the whole system under
investigation.
In a technology-driven society, the need for new materials is becoming severer. Molecular
materials have become increasingly important owing to their exceptionally optical and elec-
tronic properties. One of the examples is the molecular nonlinear optical (NLO) materials
which can be used to change the color of the laser light and to storage information. Theo-
retical investigations have made significant contribution to the understanding of nonlinear
optical processes, and can give a mechanistic insight to the connections between molecular
structure and nonlinear optical properties. They have helped chemists to design and synthe-
size organic molecules with optimized performance. The technological applications require
to use a large number of NLO active molecules and the interaction between molecules could
thus affect the performance of the material. An extreme example is the so-called push-
pull molecule attached with strong electron donor and acceptor groups. A single push-pull
molecule often possesses very large second order nonlinear optical response. When a large
number of molecules get together (in the case of liquid or solution phase), the strong dipole
interaction between molecules can quench the second order NLO of the complexes com-
pletely. Therefore, a good understanding of single molecular behavior might not be always
relevant to molecular complexes. The inter-molecular interactions under different conditions
should thus be taken into account.
13
The main theme of this thesis is to study effects of different environments on geometric and
electronic structures, as well as optical responses, of molecules. Theoretical calculations
have been carried out for experimental observeables, such as one-photon absorption (OPA),
two-photon absorption (TPA), Raman spectra and second order NLO response, of various
systems to extract useful information about molecular interactions.
Experimentally it was observed that the well defined so-called Q-band of aluminum phthalo-
cyanine chloride (AlPcCl) molecule displays a splitting in water solution. By computing
OPA of different possible water-AlPcCl complexes with time-dependent density functional
theory (TD-DFT), one can reveal the origin of the splitting and the geometric change of
the molecule in water, something that can not be directly observed experimentally. By
constructing different water-AlPcCl complexes, it is found that the distortion of AlPcCl
molecule could be the major cause of the Q-band splitting.
To avoid the possible cancelation of dipoles, a molecule, named as octupolar molecule that
possesses zero permeant dipolar moment but large second order NLO response was proposed
by Zyss.1,2 We have used a model system to examine the possible aggregation effects on
TPA spectra of octupolar molecules. With the help of molecular dynamics simulations and
TD-DFT calculations, we have found that the thermal motion can induced intra- and inter-
molecular hydrogen bonds among molecules to form clusters of different sizes. This study
demonstrates that even for molecule without dipole moment, the aggregation effect can still
take place and result in significant changes in TPA spectra in comparison with the case of
a single molecule.
Molecular aggregates can sometime enhance certain nonlinear optical properties if they
are arranged in certain ways. One of these examples is the so-called dendrimer, in which
the molecules are organized through very connecting groups. A well defined architecture
could avoid the cancelation of dipoles and result in larger second order NLO properties.
In collaboration with experimentalists, a series of end-capped triply branched dendritic
chromophores have been designed and characterized. It is shown that large enhancement of
the second order NLO property can be expected, when the dipoles of the three branches in
the dendrimers are highly parallelized.
Raman scattering is a general process that can provide more useful information on molecu-
lar structures, orientations and interactions with neighboring species than one-photon and
two-photon absorptions/emissions. However, the ordinary Raman effect is often very weak
and can not be employed in the analysis of small number of molecules. Situation becomes
quite different when a molecule is adsorbed on a metal surface. In this case, Raman scat-
tering can be dramatically enhanced due to the presence of surface plasmon coupled with
chemical bonding between the molecule and the metal surface. A single molecule can thus
14 CHAPTER 1. INTRODUCTION
be effectively determined by the so-called surface enhanced Raman scattering (SERS). We
have used a system that consists of a non-dipolar molecule, perylene, physically adsorbed on
Ag clusters to examine the possibility of extending the applicability of SERS to non-bonding
systems.
For very large and complex systems, there are always needs to increase the capacity of
computational programs. We have recently proposed and implemented a conceptually simple
but computationally efficient method, named as central insertion scheme (CIS), to calculate
electronic structures of quasi-periodic systems containing more than 100,000 electrons at
density functional theory levels. With this method, it thus becomes possible to monitor the
evolution of electronic structure with respect to the size of the system and the formation of
band from discrete molecular orbitals.
Chapter 2
Density Functional Theory and
Response Functions
2.1 Density Functional Theory
Based on the fact that the masses of nuclei are much heavier than those of electrons, molec-
ular systems could be viewed as interacting electrons moving in the static potential of nuclei
according to Born-Oppenheimer approximation. With fixed nuclear geometry ~R, one can
solve the electronic Schrodinger equation,
He | Ψe(~r; ~R)〉 = Ee | Ψe(~r; ~R)〉, (2.1)
and get the corresponding electronic energy eigenvalue Ee and eigenfunctions (or orbitals)
| Ψe(~r; ~R)〉, ~r is coordinates of electrons. The potential energy surface (PES): Ee(~R) on
which the nuclei move can be obtained by changing the nuclear positions ~R and repeatedly
solving the electronic Schrodinger equations. By solving the nuclear Schrodinger equation,(−
M∑
A=1
1
2MA
∇2A + Ee(~R)
)| Ψnuc(~R)〉 = E | Ψnuc(~R)〉. (2.2)
the total energy E of the molecule can be obtained. And the total molecular wavefunction
can be written as,
| Ψtot(~r, ~R)〉 =| Ψe(~r; ~R)〉 | Ψnuc(~R)〉. (2.3)
It is relatively easy to define the Hamiltonian operator H. However, the wave function Ψ
for large system is difficult to determine because it depends on 3N spatial coordinates and
N spin coordinates, where N is the number of electrons.
15
16 CHAPTER 2. DENSITY FUNCTIONAL THEORY AND RESPONSE FUNCTIONS
Moreover, electrons repel each other and their motions are strongly correlated in reality.
In the mean field approximation, like Hartree-Fock,3 this correlation effect of electrons is
completely neglected. To improve the quality of the wavefunctions, many so-called post-
Hartree-Fock methods, such as Configuration Interaction (CI),4 Coupled Cluster (CC),5 and
Møller-Plesset perturbation theory6 (MP2, MP4, etc), have been developed to take into ac-
count the electron correlation at different levels. In general, highly correlated computational
methods require much more computational power and time.
Since the late eighties and early nineties, Density Functional Theory (DFT) has become
more popular in molecular physics and chemistry. Compare with wavefunction based ab
initio methods, the main advantages of DFT methods are the sufficiently high accuracy
and short computational time, in particular for large systems, since it reduces the spatial
dimension of the problem from 3N to three. This simplification is done by assuming the
energy as a function of the electron density ρ(~r) which is only related to 3 space variables.
2.1.1 Hohenberg-Kohn Theorems
Before the introduction of Hartree-Fock theory, L. H. Thomas and E. Fermi proposed the
first DFT model independently.7–9 But only from 1964, with the help of two important
theorems proved by Hohenberg and Kohn,10 the Thomas-Fermi theory started to give good
results. These two theorems are usually considered as the base of modern DFT.
The first theorem states that the external potential V (~r) of a system of N electrons is
uniquely determined by its density ρ(~r). And then the Hamiltonian
H = −1
2
∑
i
∇2i +
∑
i
V (~ri) +∑
i>j
1
~iij(2.4)
is determined by ρ(~r). ρ(~r) can be got with the equation,
ρ(~r) =N∑
i=1
|ψi(~r)|2 . (2.5)
The energy can be written as a functional of the density
E = E[ρ]. (2.6)
The second one proves that if ρ(~r) is the exact ground state density, the total electronic
energy of the system becomes a minimum.
E[ρ] ≤ E[ρt], (2.7)
where ρt is any trial density.
2.1. DENSITY FUNCTIONAL THEORY 17
2.1.2 Kohn-Sham equations
Since the kinetic energy functional T [ρ] is easily calculated from the wave function, we can
express the energy functional of an isolated molecular system as,
E[ρ] = T [ρ] + VNe[ρ] + Vee[ρ], (2.8)
where VNe[ρ] and Vee[ρ] account for the nuclear-electron attraction and the electron-electron
interaction functional, respectively.
It has been assumed by the Kohn-Sham method that a fictitious non-interacting system has
the same electron density distribution ρs(~r) as that of the real molecular system ρ(~r). The
energy functional of this non-interacting system can then be written as,
Es[ρ] = Veff [ρ] + Ts[ρ], (2.9)
where Veff is an effective potential functional and Ts[ρ] is the non-interacting kinetic energy
functional. From the relationship ρs(~r) = ρ(~r) and Es[ρ] = E[ρ], we can easily get
Veff [ρ] = VNe[ρ] + (T [ρ] − Ts[ρ]) + Vee[ρ]. (2.10)
In detail,
Veff (~r) = VNe(~r) +
∫ρ(~r′)
|~r − ~r′|d ~r′ + Vxc[ρ(~r)], (2.11)
where the second term is the electron-electron Coulomb repulsion functional, and Vxc[ρ(~r)] is
the so-called exchange correlation potential functional. By solving the Schrodinger equations
of this non-interacting system,
[−1
2∇2
i + Veff(~r)]ψi(~r) = ǫi ψi(~r), (2.12)
the orbitals ψi(~r) that reproduce the electron density of the real molecular system will be
obtained.
2.1.3 Exchange and Correlation Functionals
The key problem of DFT is to find efficient exchange correlation functional since there is
no straightforward way to systematically improve the exact form of Exc[ρ]. Exc[ρ] has often
been separated in two parts, a pure exchange part Ex[ρ] and a pure correlation part Ec[ρ].
18 CHAPTER 2. DENSITY FUNCTIONAL THEORY AND RESPONSE FUNCTIONS
Local Density Methods
Local Density Approximation (LDA) assumes that the density is local and can be treated
as a homogeneous electron gas. For a system that an infinite number of electrons N in an
infinite volume V, the density ρ can be remaining finite and written as,
ρ = N/V. (2.13)
Then we can get the exact exchange energy Ex[ρ]
Ex[ρ] =4
3(3π−1)
2
3
∫ρ(~r)
4
3d~r. (2.14)
The correlation energy functional of the electron gas has been developed by Vosko, Wilk
and Nusair (VWN).11 When considering α and β spins separately for an open shell system,
LDA is replaced by the Local Spin Density Approximation (LSDA).
Gradient Corrected Approximation
For a non-uniform electron gas, LDA approach has been extended to functionals depending
on the gradient of the density ▽ρ(~r) based on the Generalized Gradient Approximation
(GGA). The general form of the GGA can be written as
EGGAxc [ρα, ρβ] =
∫f(ρα, ρβ,▽ρα,▽ρβ)d~r. (2.15)
By considering boundary conditions and fitting to accurate numerical or experimental data,
several gradient corrected functionals have been proposed for the exchange energy such as
Perdew and Wang (PW86),12 Becke (B88),13 and correlation energy such as Perdew (P86).14
And Lee, Yang and Parr (LYP)15 developed new correlation functionals that are not based
on corrections of the LSDA.
Hybrid Functionals
Another way of constructing functionals is to combine the LSDA, exact exchange and the
gradient correction terms together. Among many approximations that have been developed
over the years, one of the most successful functionals is the hybrid B3LYP method which
combine the Becke-3 (B3) parameter exchange functional16 and LYP correlation functional.
The B3LYP exchange-correlation functional can be written as,
EB3LY Pxc = ELDA
xc + a(EHFx −ELDA
x ) + b(EGGAx −ELDA
x ) + c(EGGAc − ELDA
c ), (2.16)
2.2. RESPONSE THEORY 19
where a = 0.20, b = 0.72, and c = 0.81 are three empirical parameters; EGGAx and EGGA
c
are the generalized gradient approximation formulated with the B88 exchange functional
and the LYP correlation functional, and ELDAc is the VWN correlation functional. B3LYP
method can often give results in good agreement with experiments and thus be widely used
in many theoretical calculations.
2.2 Response Theory
Among many fundamentally different ways of describing the response of the electronic de-
grees of freedom in the presence of a perturbing field, response theory17 has enjoyed a great
success in providing accurate results for a wide variety of molecular properties. With re-
sponse theory, the exact information about the excited states is no longer needed. A set
of response functions and their residues are used to express molecular properties, such as
(hyper)polarizabilities, excitation energies, spin-orbit coupling matrix elements, phospho-
rescence lifetimes, nuclear shielding constants and (hyper)magnetizabilities, just to name a
few. Moreover, results of response theory can be compared with experimental measurements
directly, since calculations are performed for external fields of any frequencies. Response
theory deals with the time dependent and time independent properties on the same footing.
2.2.1 Response Theory for Exact States
In detail, we can write Hamiltonian H of a molecular system which is perturbed by a time
dependent perturbation field as the summation of the unperturbed Hamiltonian H0 and the
time dependent perturbation term V (t).
H = H0 + V (t), (2.17)
V (t) =
∫ ∞
−∞
V ω e(−iω+ǫ) t dω, (2.18)
where ǫ is a positive infinitesimal. When t = −∞, the system is in the unperturbed state
V (t) = 0. The unperturbed hamiltonian H0 can be used to solve
H0|0〉 = E0|0〉 (2.19)
and
H0|n〉 = E0|n〉 (2.20)
20 CHAPTER 2. DENSITY FUNCTIONAL THEORY AND RESPONSE FUNCTIONS
Let A be a Hermitian operator describing an observable molecular property. When the
perturbation V (t) caused by the external field is sufficiently small, one can write the time-
evolution of the expectation value of A as,
⟨A⟩
(t) =⟨A⟩
0
+
∫ ∞
−∞
⟨⟨A; V ω1
⟩⟩e−iω1t dω
+1
2!
∫ ∞
−∞
∫ ∞
−∞
⟨⟨A; V ω1, V ω2
⟩⟩e−i(ω1+ω2)t dω1 dω2
+1
3!
∫ ∞
−∞
∫ ∞
−∞
∫ ∞
−∞
⟨⟨A; V ω1 , V ω2 , V ω3
⟩⟩e−i(ω1+ω2+ω3)t dω1 dω2 dω3
+ · · · ,
(2.21)
⟨⟨A; V ω1, · · · , V ωn
⟩⟩is the response function, which describes the response of the molecular
properties A to the external perturbation field.
For example, if A represents the dipole moment operator µ of the molecular system and the
perturbing field is electric, the corresponding response functions are the molecular linear
polarizability and nonlinear hyperpolarizabilities α, β, γ, etc.
αij(−ωσ;ω1) = −〈〈µi; µj〉〉ω1, (2.22)
βijk(−ωσ;ω1, ω2) = −〈〈µi; µj, µk〉〉ω1,ω2, (2.23)
γijkl(−ωσ;ω1, ω2, ω3) = −〈〈µi; µj, µk, µl〉〉ω1,ω2,ω3. (2.24)
By choosing other operators, one can obtain other properties through the corresponding
response functions. And for time-independent perturbations, Eq. 2.21 naturally reduces
to the ordinary Taylor expansion of the static expectation value of A, where the response
function⟨⟨A; V ω1, · · · , V ωn
⟩⟩ω1=0,··· ,ωn=0
becomes the expansion coefficient.
When a system is subject to one or several fields, the effects of the fields can be also deter-
mined by response functions. Each molecular property A can be described within a frame-
work by suitable choice of fields. For instance, the linear response function⟨⟨A; V ω1
⟩⟩
ω1
contains all terms that are linear in V ω1 . And the quadratic and cubic response functions
functions⟨⟨A; V ω1 , V ω2
⟩⟩
ω1,ω2
and⟨⟨A; V ω1, V ω2, V ω3
⟩⟩
ω1,ω2,ω3
contain all contributions
to the expectation values that are linear in V ω1 , V ω2 and V ω1 , V ω2 , V ω3 , respectively.
2.2. RESPONSE THEORY 21
2.2.2 One-Photon Absorption
Electron transitions between different states of a molecule can also be obtained with the
response functions. For instance, the linear response function⟨⟨A; V ω
⟩⟩ω
can be written
as,
⟨⟨A; V ω
⟩⟩
ω=∑
n 6=0
⟨0∣∣∣A∣∣∣n⟩⟨
n∣∣∣V ω
∣∣∣ 0⟩
ω − (En − E0)−∑
n 6=0
⟨0∣∣∣V ω
∣∣∣n⟩⟨
n∣∣∣A∣∣∣ 0⟩
ω + (En − E0), (2.25)
where En is the energy corresponding to eigenfunction |n〉. For frequencies equal to plus or
minus the excitation energies of the unperturbed system, the linear response function has
poles shown as,
limω−ωk
⟨⟨A; V ω
⟩⟩ω
=⟨0∣∣∣A∣∣∣ k⟩⟨
k∣∣∣V ω
∣∣∣ 0⟩, (2.26)
limω+ωk
⟨⟨A; V ω
⟩⟩ω
= −⟨0∣∣∣V ω
∣∣∣ k⟩⟨
k∣∣∣A∣∣∣ 0⟩, (2.27)
where ωk = Ek −E0. Clearly one-photon absorption process can be described by these two
functions that contains information about excitation energies from the reference state |0〉
and the corresponding transition matrix elements. The transition dipole moments between
the ground state |0〉 to an excited state |f〉 can be written as
limω1→−ωf
(ω1 − ωf) 〈〈µi;µj〉〉ω1= 〈0 |µi| f〉 〈f |µj| 0〉 . (2.28)
2.2.3 Two-Photon Absorption
The two-photon cross section σTPA is related to the second hyperpolarizability γ. If one
chooses ω as the half of the energy ωf which is the excitation energy from the ground state
to the final two-photon state |f〉
ω =1
2ωf , (2.29)
22 CHAPTER 2. DENSITY FUNCTIONAL THEORY AND RESPONSE FUNCTIONS
The second hyperpolarizability, γαβγδ(−ω;ω,−ω, ω), can be written as,
γαβγδ(−ω;ω,−ω, ω) =h−3∑
P1,3×
∑
nm
[〈0 |µα|n〉 〈n |µγ| f〉 〈f |µβ|m〉 〈m |µδ| 0〉
(ωn − ω)(−iΓf/2)(ωm − ω)
+〈0 |µγ|n〉 〈n |µα| f〉 〈f |µβ|m〉 〈m |µδ| 0〉
(ωn − ω)(−iΓf/2)(ωm − ω)]
=i2h−3
Γf
∑P−δ,2
∑
n
〈0 |µα|n〉 〈n |µγ | f〉
ωn − ω
∑P1,3
∑
m
〈f |µβ|m〉 〈m |µδ| 0〉
ωm − ω
=i2h−1
Γf
SαγS∗δβ .
(2.30)
And two-photon transition matrix elements Sαβ can be identified as
Sαβ = h−1∑
n
[〈0 |µα|n〉 〈n |µβ| f〉
ωn − ω+
〈0 |µβ|n〉 〈n |µα| f〉
ωn − ω
]. (2.31)
Other than getting the TPA cross section from γ, the TPA transition matrix elements can
also be deduced from the single residue of the quadratic response function, as shown below,
limω2→−ωf
(ω2 − ωf) 〈〈µi;µj, µk〉〉−ω1,ω2
= −∑
n
[〈0 |µi|n〉 〈n |(µj − 〈0 |µj| 0〉)| f〉
ωn − ω2
+〈0 |µj|n〉 〈n |(µi − 〈0 |µi| 0〉)| f〉
ωn − ω1
]〈f |µk| 0〉 .
(2.32)
And the transition dipole moments between excited states can be obtained from the double
residue of the same response function, which can be written as
limω1→ωf
(ω1 − ωf)
[lim
ω2→−ωm
(ω2 − ωi) 〈〈µi;µj, µk〉〉−ω1,ω2
]
= −〈0 |µi| f〉 〈f |(µj − 〈0 |µi| 0〉)| i〉 〈i |µk| 0〉 ,
(2.33)
where ωm = ω1 + ω2.
Chapter 3
Linear and Nonlinear Optical
Spectroscopy
This chapter gives a brief introduction about the linear and nonlinear optical processes
that have been studied in this thesis with special attention to one-photon and two-photon
absorptions.
Absorption spectroscopy is based on absorption of photons by pure substances or mixtures
in solid, liquid, or gas phases. An absorption spectrum of a particular substance is obtained
by relating the amount of absorbed photon with its wavelength. Every compound has
absorption lines at particular wavelengths. And if the structure of the compound is stable,
the normalized absorption spectrum does not change much at different concentrations and
thus can be taken as a kind of chemical “fingerprint”. For instance, C2H4 has its absorption
band of longest wavelength at bout 180 nm, while for COH2 it is at about 280 nm.
The absorbed wavelength of every compound is determined by the energy difference between
the initial and final states. Atomic states are defined by arranging electrons in different
atomic orbitals, while molecular states are defined by quantized molecular states mixed
with vibrational and rotational modes. By absorbing one or several photons whose total
energy is exactly equal to the energy difference between the initial and final states, an
electron will be excited to the final state of higher energy. After some time, the excited
electron will be relaxed to the initial state of the system, which is a process that can be
accompanied by light emitting. One of the combined absorption and emission processes
gives Raman spectroscopy which will be discussed in next chapter.
Absorption spectroscopy covers a wide range of wavelengths, going from infrared, optical to
ultraviolet. With the conventional light source, only one photon absorption is detectable.
With the help of powerful laser, multi-photon absorptions can be routinely achieved.
23
24 CHAPTER 3. LINEAR AND NONLINEAR OPTICAL SPECTROSCOPY
3.1 One-photon Absorption
When a molecule absorbs one photon whose energy matches the energy difference between
the ground state and an excited state, electrons can be promoted from the ground state to the
excited state. This process is called one photon absorption (OPA) process as schematically
shown in Fig. 3.1. By tuning the frequency of the incoming light, electrons can be excited to
Figure 3.1: Scheme of one photon absorption process.
different excited states at where a strong absorption of the incoming light can be observed.
For a given final state f , the one-photon absorption cross section is associated with the
oscillator strength
δopa =2ω0f
3
∑
α
〈0 |µα| f〉2 (3.1)
ω0f is the energy deference between the ground |0〉 and the final states |f〉, 〈0 |µα| f〉 is
the electronic transition dipole moment between two states and the summation goes over
x, y and z axes of the molecule. These are the peak values of the absorptions. In reality,
each excited state has its own lifetime which give rise to the lifetime broadening of each
absorption peak. In the optical region, the lifetime of the charge transfer state, which often
possesses very large transition dipole moment, of a gas phase molecule is often in the range
of nanosecond. When the molecule is put into the solution, the lifetime of the state can
be drastically reduced, to the order of femto-second, because of the collision between the
molecule and the solvent. Furthermore, the involvement of vibronic structures can further
complicate the absorption spectrum. The experimental OPA spectra can be very broad and
3.2. TWO-PHOTON ABSORPTION 25
ill-defined, which make it difficult to extrapolate useful information about the molecular
system under investigation. Theory can in many cases provide much needed assistance.
3.2 Two-photon Absorption
Quantum mechanically, the possibility of absorbing several photons simultaneously by an
atom or a molecule is always existed, although it could be too small to be observed under
the excitation of conventional light. Two-photon absorption (TPA) was first predicted by
Goppert-Mayer in 193118 but had to wait for 30 years to be observed experimentally with
the help of newly invented laser sources. The TPA processes are schematically drawn in Fig.
3.2 which can be utilized by absorption of two photons of the same or different energies.
Two-photon absorption has been primarily used as a spectroscopic tool that complements
Figure 3.2: Scheme of TPA process.
one-photon absorption spectroscopy. It enables to study states that are inaccessible for
ordinary one-photon excitations due to symmetry or parity selection rules. It furthermore
stretches the accessible range of conventional lasers by using a frequency half of the actual
energy gap. In recent years, TPA has been found many technological applications, such
as two-photon excited fluorescence microscopy,19,20 optical limiting,21–23 three-dimensional
optical data storage,24 two-photon induced biological caging studies,25 and nano- or mi-
crofabrication.26 All these applications take the advantage of two unique features of TPA,
namely the ability to create excited states with photons of half the nominal excitation en-
26 CHAPTER 3. LINEAR AND NONLINEAR OPTICAL SPECTROSCOPY
ergy, which can provide improved penetration in absorbing or scattering media, and a high
degree of spatial selectivity in three dimensions because of its quadratic dependence on the
intensity. Moreover, because the wavelength used for TPA is much longer than that of OPA,
the influence of scattering on beam intensity is greatly reduced. These are clear advantages
for applications in imaging of absorbing or scattering media, like biological tissues.
As we shown in previous chapter, TPA matrix element for the resonant absorption of two
photons with identical energy can be expressed as
Sαβ = h−1∑
i
[〈0|µα|i〉〈i|µβ|f〉
ωi − ω0f/2+
〈0|µβ|i〉〈i|µα|f〉
ωi − ω0f/2
], (3.2)
in a sum-over-states form, where ωi is the excitation energy for the intermediate state |i〉,
ω0f is the energy difference between the initial and final states, and thus ω0f/2 means the
fundamental frequency of the laser beam. TPA cross section can be given as,
δtpa =∑
αβ
[F × SααS
∗ββ +G× SαβS
∗αβ +H × SαβS
∗βα
], (3.3)
by orientational averaging over the TPA probability for molecules in gas phase and solu-
tion,27 where the coefficients F , H , and G are related to the polarization of the radiation
source and the summation goes over the molecular x, y, and z axes. For linearly polarized,
the values of F , H , and G are 2, 2 and 2, for the circular case, F , H , and G are −2, 3 and
3. In this thesis, only linearly polarized case has been considered so that the microscopic
TPA cross section becomes,
δtpa = 6(S2xx + S2
yy + S2zz) + 8(S2
xy + S2yz + S2
zx) + 4(SxxSyy + SyySzz + SzzSxx) (3.4)
To be comparable with the experiment directly, the TPA cross section needs to be re-defined
as,
σtpa =4π3a5
0α
15c
ω2
Γf
δtpa, (3.5)
where a0 is the Bohr radius, c is the speed of light, α is the fine structure constant, hω is
the photon energy, and Γf is the lifetime broadening of the final state. The unit of TPA
cross section is GM named after Goppert-Mayer ( 1 GM= 10−50 cm4s/photon).
3.3 Selection Rules
In an atomic or molecular system, there are huge numbers of excited states. However, not all
states can be detected by one particular absorption process within dipole approximation.
3.3. SELECTION RULES 27
Only the states that obey dipole selection rules are allowed. For instance, for a center-
symmetric molecule, if the initial state has a parity of grade (g), only the final state with
parity of ungrade (u) can be reached by OPA since the dipole operator has always an ungrade
parity (u). In general, if the product of symmetry groups of the initial state, the dipole
operator, and the final state contains the symmetry group of the initial state, the transition
between the initial and final states is allowed. TPA is a product of two OPA processes,
involving the transition element of 〈0 |r| i〉 〈i |r| f〉, which should obey a rule different from
that of OPA. For instance, starting from a ground state with parity g, the final states with
parity of g become accessible by TPA.
Here we take benzene molecule, shown in Fig. 3.3, which has very high symmetry of D6h as
an example. The calculated OPA and TPA spectra using DFT at B3LYP level with 6-31G*
Figure 3.3: Optimized structure of benzene (C6H6) molecule.
basis set are shown in Fig. 3.4. In a D6h group, there are twelve symmetry elements, namely
A1g, A1u, A2g, A2u, B1g, B1u, B2g, B2u, E1g, E1u, E2g and E2u. The dipole operators, x, y, z,
belong to symmetries of E1u and A2u, respectively. The ground state of benzene molecule
is of symmetry A1g, therefore, states with symmetries E1u and A2u can be reached by OPA,
but are not accessible with TPA. As shown in the calculated spectrum, the degenerate state,
1E1u, at 7.4 eV, has large OPA cross section and zero TPA strength. At the same time,
degenerate states, 1E2g, at 9.1 eV, and 2E1g, 10.0 eV, show considerable TPA cross section
with zero OPA strength. It is noticed that for the high symmetry molecule, like benzene,
there are many states with right parity can be symmetry forbidden. For instance, the lowest
28 CHAPTER 3. LINEAR AND NONLINEAR OPTICAL SPECTROSCOPY
single state 1B2u is a both OPA and TPA forbidden state.
2 3 4 5 6Fundamental energy (eV)
0
1
2
3
4
5
6
7
OPA
(R
el. I
nt)
TPA
cro
ss-s
ectio
n (G
M)OPA TPA
Figure 3.4: Calculated OPA (solid lines) and TPA (dashed lines) spectra of optimized
benzene molecule. Lifetime broadening of 0.1 eV is used. Energy scale in TPA is normalized
to coincident with that of OPA.
3.4 Hyperpolarizabilities
Nonlinear optics is a way of manipulation light through the interaction between the light
and the matter. The frequency, the polarization or the phase of an incoming laser light
can be altered after interacting with materials possessing strong nonlinear optical proper-
ties. Nonlinear optical (NLO) materials have wide potential applications in various photonic
technologies such as optical switching, telecommunications and optical computing, just name
a few.28–32 Therefore, finding materials with unique NLO properties has attracted consid-
erable attention in last decades.
3.4. HYPERPOLARIZABILITIES 29
From a microscopic view, the induced dipole moment µ of the molecular system and the
external electric field strength F have the relationship
µ = µ0 + αF (3.6)
in linear optics, where µ0 is the permanent dipole moment of the molecular system, and α
is the polarizability. While in nonlinear optics, the influence of the high power terms of the
electric field strength should be considered on the induced dipole moment. And then the
induced dipole moment can be written as,
µ = µ0 + αF +1
2!βF2 +
1
3!γF3 + · · · . (3.7)
i.e. by a Taylor series expansion. β and γ are the molecular first hyperpolarizability and
second hyperpolarizability that control the second and third order NLO processes.
If we write the external electric in a Fourier expansion form of the sum of electric fields with
discrete frequencies,
F(t) =∑
p
F(ωp) e−i ωp t. (3.8)
µi for component i can be got
µi =µ0i +
∑
p
αij(−ωσ1;ωp)Fj(ωp) e
−i ωσ1t
+1
2!
∑
pq
βijk(−ωσ2;ωq, ωp)Fj(ωq)Fk(ωp) e
−i ωσ2t
+1
3!
∑
pqr
γijkl(−ωσ3;ωr, ωq, ωp)Fj(ωr)Fk(ωq)Fl(ωp) e
−i ωσ3t
(3.9)
by using the Einstein summation convention. ωσ1= ωp, ωσ2
= ωq + ωp and ωσ3= ωr + ωq +
ωp showing different frequencies of the output light. This expression represents different
frequency dependent NLO processes.
3.4.1 Finite Field Method
Finite Field (FF) method is used to calculate linear and nonlinear polarizabilities at the
static limit. It is easily to be understood and to be realized in programming.
The total molecular energy E of a molecular system can be expanded into Taylor series over
the electric field strength F,
E = E0 − µ0iFi −
1
2!αijFiFj −
1
3!βijkFiFjFk −
1
4!γijklFiFjFkFl − · · · , (3.10)
30 CHAPTER 3. LINEAR AND NONLINEAR OPTICAL SPECTROSCOPY
in uniform electrostatic fields, where E0 is the total energy of the molecular system without
the external electric field, µi is the ith component of the molecular dipole moment vector
and Fi is the component in the ith direction of the external uniform electrostatic field with
each i goes over all Cartesian coordinates x, y and z.
In FF method, a series of field dependent total energy of the molecular system can be
obtained by applying electrostatic fields with different strengths at certain directions. And
then the corresponding coefficients in Eq. 3.10 can be determined by polynomial fitting.
For example, the dipole moment of the excited charge transfer state can be obtained with
hωeg(Fz) = hωeg(0) − (µzee − µz
gg)Fz −1
2!(αzz
ee − αzzgg)F
2z − · · · . (3.11)
3.4.2 Analytical Derivative Method
Analytical Derivative (AD) method is a generalized approach that can avoid the numerical
uncertainty problem of the FF method. In this case, µ, α, β, and γ are expressed in
derivative forms of the total molecular energy E over the electric field strength F.
µi = −∂E
∂Fi
,
αij = −∂2E
∂Fi∂Fj
,
βijk = −∂3E
∂Fi∂Fj∂Fk
,
γijkl = −∂4E
∂Fi∂Fj∂Fk∂Fl
,
(3.12)
...
Still, AD method can only obtain static properties.
3.4.3 Response Theory Method
This method is a very powerful tool to study dynamic NLO properties and has been well
described in Chapter 2.
3.5. APPLICATIONS 31
3.5 Applications
3.5.1 Q-band Splitting of Aluminum Phthalocyanine Chloride
Molecular phthalocyanines have been the subject of theoretical and experimental stud-
ies over the past decades owing to a variety of applications in areas such as xerography,
chemical sensors, and photodynamic therapy.33–41 The properties of phthalocyanine can be
effectively tuned by inserting different metal atom in the center of the molecule. Most met-
allophthalocyanines are nontoxic, water, air and thermally stable, which made them very
attractive for biotechnology. For instance, the second generation photosensors are mainly
based on the phthalocyanines, because metallophthalocyanine molecules absorb intensely in
the long-wavelength side of the visible region, which is particularly useful for diagnostics
and therapy of malignant neoplasms.42
One of important absorption bands in metallophthalocyanines is the so-called Q-band, lo-
cated at around 600-700 nm.43 Many studies have been devoted to understand the properties
of this band. One of the most interesting experimental observations is the sensitivity of this
band to the changes of the surrounding. A separation as large as 10.14 nm for the Q-band
splitting has been observed in fluorescence excitation spectra of aluminum phthalocyanine
tetrasulfonate (APT) in hyper-quenched glassy water. The aluminum phthalocyanine (APc)
molecule has been found to show a similar behavior in hyperquenched glassy ethanol with
APT.44 A study of Q-band splitting should provide useful information about the interaction
between the molecule and the surrounding.
We have carried out time-dependent density functional theory (TD-DFT) calculations for
AlPcCl molecule and AlPcCl-water complexes. Furthermore, the polarizable continuum
model (PCM) is employed for complex of AlPcCl and three water molecules, to explore the
effect of the long range inter-molecular interactions. It has been found that the interaction
with water molecules can only result in relatively smaller splitting for the Q-band, less than
3 nm in most cases. The dimerization of molecules also fails to induce larger splitting.
The most possible mechanism to explain the large splitting of the Q-band in the AlPcCl
(eventually the APT) molecule is found to be associated with the bond changes between the
metal atom and neighboring nitrogen atoms. Symmetrically changing the N−Al bonds by
reducing the α angles as shown in Fig. 3.5, we have found that the value of Q-band splitting
can be as large as 24 nm when a small geometry distortion that costs extra energy as less
as 2.8 kcal/mol occurs. This molecular geometry distortion is expected when the AlPcCl
molecule is mixed with the glasses. This result also shows that the interaction strength
between the surrounding and the AlPcCl molecule might be expressed by the magnitude of
the Q-band splitting.
32 CHAPTER 3. LINEAR AND NONLINEAR OPTICAL SPECTROSCOPY
Figure 3.5: Structure of AlPcCl monomer. α is a tuneable angle that represents the geometry
distortion.
3.5.2 Two-Photon Absorption of Octupolar Molecules
Strong conjugated molecules that are asymmetrically substituted with electron donor and
acceptor groups, such as para-nitroaniline (PNA) derivatives, substituted stilbenes, and
push-pull polyenes have been found to exhibit extremely large second order nonlinear optical
properties. However, they also possess some major drawbacks, such as a high tendency
toward unfavorable aggregation, difficult in non-centrosymmetric crystallization, and small
off-diagonal tensor components associated with their anisotropic dipolar character.45 To
avoid these weaknesses, octopolar molecules with zero permanent dipole moment and large
first hyperpolarizability have been designed and synthesized. They have been considered to
be the most promising compounds for second-order NLO applications.
With the help of molecular dynamics at room temperature, it is, however, found that the
charge transfer octupolar molecules can also form aggregates through intermolecular hy-
drogen bonds activated by thermal motion of the molecules.46 We have studied TPA cross
section of different clusters taken from snapshots of molecular dynamics simulations for a
model molecule 1,3,5-triamino-2,4,6-trinitrobenzene (TATB), including dimers, trimers and
tetramers. Aggregation effect on TPA cross section has been clearly demonstrated. Taken
the results of TATB dimers as an example, five representative dimer structures and their
TPA spectra are shown in Fig. 3.6. It can be seen that for all dimers, maxima of their TPA
3.5. APPLICATIONS 33
Figure 3.6: Calculated TPA spectra of five representative dimers, together with those of the
optimized monomer. A lifetime broadening of 0.1 eV is used.
34 CHAPTER 3. LINEAR AND NONLINEAR OPTICAL SPECTROSCOPY
spectra are red shifted by as much as 0.6 eV to 0.9 eV comparing with that of the monomer.
Enhancement of TPA cross section for certain structure, like dimer E, could also be found.
3.5.3 One- or Two-end-capped Dendritic Structures
The concentration of nonlinear optical chromophores should be sufficiently high for realistic
applications. In that case, as we mentioned above the push-pull molecules turn to sponta-
neously form aggregates that then drastically reduce the nonlinear optical response of the
total system. A controlled aggregation of push-pull molecules could avoid this probelm. We
have constructed three-fold symmetric dendritic architectures of dipolar chromophores that
effectively enhance the local density of nonlinear optical molecules and their responses.
We have studied a series of end-capped triply branched dendritic chromophores including
one- and two-end-capped models as displayed in Fig. 3.7. It is found that the second order
nonlinear optical properties of one-end-cappedtype dendrimitic arrangements are strongly
dependent on the mutual orientations of the three chromophores and large enhancement of
the first hyperpolarizability can be obtained when the dipole moments of the three branches
in the dendrimers are highly parallelized. The increase of the conjugation length can also
improve the performance of the dendrimers. However, the use of two-end-caps is less useful
to improve the performance of these systems. The calculated structure-to-property relations
are successful in explaining various experimental observations.
3.5. APPLICATIONS 35
Figure 3.7: Optimized molecular structures of one- and two-end-capped dendritic structures.
36 CHAPTER 3. LINEAR AND NONLINEAR OPTICAL SPECTROSCOPY
Chapter 4
Raman Spectroscopy
4.1 History
Quantum mechanically, light has the dual character of waves and particles. When consid-
ering light as particles, inelastic scattering processes could be expected. Compton showed
inelastic scattering of x-rays from a graphite target in 1923.47 And Smekal predicted the
inelastic scattering of light theoretically in the same year.48 It was known that the inten-
sity of inelastic scattering scales to the fourth power of the energy, which means that the
observed scattering cross section at optical wavelength of 500 nm should be 10 orders of
magnitude larger than that at x-ray wavelength of 0.7 nm. Raman focused a large telescope
on the sun to obtain strong green light through a filter. After the beam of green light
passed through a solution of chloroform, a weak yellow light was observed, resulting from
the inelastic scattering. In 1928, C. V. Raman and K. S. Krishnan published a paper in
the journal Nature to report their experiments which proved Smekal’s prediction of inelastic
scattering of light.49 The inelastic scattering of visible light from molecular transitions was
named after Raman. In 1930, Raman won the Nobel Prize in Physics.
Nowadays, with the help of a small HeNe laser and a CCD detector, modern Raman spec-
troscopy can be easily carried out in a matter of few seconds. Raman spectroscopy has been
widely used to identify chemical bonds as “fingerprint” of molecular compounds. Raman
spectra complements with infrared spectra, because some excitations can only happen in
one of them. And for the both allowed excitations, the intensities are always different.
37
38 CHAPTER 4. RAMAN SPECTROSCOPY
4.2 Basic Theory
When photons are scattered by molecular systems, most photons are elastically scattered
and this process can be called Rayleigh scattering as shown in Fig. 4.1. During this process,
Figure 4.1: Processes of Rayleigh scattering, Stokes Raman scattering and anti-Stokes Ra-
man scattering.
the emitted photon has the same energy as the incoming photon. Raman spectroscopy is
based on the inelastic scattering of photons. When light shines upon molecules and interacts
with the electron cloud of the molecules, a molecule can be excited from the ground state to
a virtual state by absorbing one incident photon and relaxes to an excited state. A photon
will then be emitted with less energy,
Eout = Ein − ωj, (4.1)
where ωj is energy level of the excited state. This is called Stokes Raman scattering. If
the molecule is in an excited state at initial, and relaxes to the ground state, the emitted
photon will have energy
Eout = Ein + ωj. (4.2)
Obviously, the energy of the scattered photon is smaller than the incident photon for the
Stokes scattering and while it is larger for the anti-Stokes case. When the initial and final
electronic states belong to the same electronic state, the energy difference in each process
4.2. BASIC THEORY 39
is related to the energy of the vibrational excited state. Therefore the Stokes or anti-Stokes
lines are a direct measurement of the vibrational energy levels of the molecule.
Although the anti-Stokes and Stokes lines are equally displaced from the Rayleigh line,
anti-Stokes lines are much less intense than the Stokes lines because only molecules that
are vibrationally excited at the initial state can give rise to the anti-Stokes lines. Therefore,
more intense Stokes lines are normally measured in Raman spectroscopy.
When a molecule exhibits vibrational Raman effect, there should be deformation of the
electron cloud or change of molecular polarizability. Raman intensity is determined by the
amount of the change of polarizability. Raman scattering transition moment can be written
as,
R = 〈i |α| j〉 , (4.3)
where α is the polarizability of the molecule, |i〉 and |j〉 are the initial and final states. α
can be expressed as,
α = α0 + (~r − ~re)
(dα
d~r
)+ · · · , (4.4)
where α0 is the polarizability at the equilibrium bond length ~re, ~r is the distance between
atoms. Since α0 is a constant and 〈i|j〉 = 0, R can be simplified to,
R =
⟨i
∣∣∣∣(~r − ~re)dα
d~r
∣∣∣∣ j⟩. (4.5)
It shows that to have a nonzero Raman scattering transition moment, there must be a
change in polarizability.
However, not all vibrational modes can take part in the Raman scattering. Here we use
the asymmetric stretch and the symmetric stretch of CO2 molecule, shown in Fig. 4.2, as
examples. The polarizability changes during vibration because it depends on how tightly
the electrons bind to the nuclei. Obviously, in the symmetric stretch mode, the strength of
electron binding to the nuclei is different with the minimum and maximum internuclear dis-
tances. Therefore the symmetric vibrational mode is Raman active. And for the asymmetric
stretch, there is no overall changes in polarizability so that it is Raman inactive vibration. In
general, Raman scattering obeys the same selection rule as that for two-photon absorption.
The vibrational Raman-scattering cross section is proportional to
σFI,αβ = |〈F, 0 |ααβ| I, 0〉|2 , (4.6)
40 CHAPTER 4. RAMAN SPECTROSCOPY
Figure 4.2: Symmetric stretch and asymmetric stretch modes of CO2.
within the Born-Oppenheimer approximation, where |I, 0〉 is the initial vibrational state
and |F, 0〉 is the final state, and ααβ is the polarizability tensor. ααβ can be written as50,51
ααβ =∑
k
∑
J
〈0 |µα| k〉 |J, k 〉〈 J, k| 〈k |µβ| 0〉
EJk − EI
0 − ω − iΓ
+〈0 |µα| k〉 |J, k 〉〈 J, k| 〈k |µβ| 0〉
EJk − EI
0 + ωS + iΓ,
(4.7)
where EJk is the energy for the |J, k〉 state, ω is the frequency of the incident light, and ωS
is that of the scattered light. The summation of k and J is over all the electronic surfaces
|k〉 and all the vibrational levels of each surface |J, k〉. The above equation is only valid
for nonresonance scattering, because the polarizability is approximated by the electronic
polarizability tensor and expanded in a Taylor series about the equilibrium geometry in the
standard theory of Placzek.
4.3 Surface Enhanced Raman Spectroscopy
In 1974, Martin Fleischman and coworkers found large Raman signal from pyridine adsorbed
on electrochemically roughened silver surface.52 However, they did not notice that this was
a major enhancement effect. Three years latter, two groups realized that the large signal
was from Raman enhancement effect independently.53,54
It has been found that the Raman scattering from a compound adsorbed on a metal sur-
face can be 103-106 times greater than that in solution. Silver, gold and copper55 are the
most often used surface materials for surface enhanced Raman spectroscopy (SERS). It has
4.3. SURFACE ENHANCED RAMAN SPECTROSCOPY 41
also known that small particles or atomically rough surfaces can also produce large SERS
signals. Therefore, SERS can be used to study molecular mono-layer adsorbed on metals
or electrodes. Over the years, SERS technique has been applied to many metal systems,
such as metal films on dielectric substrates, colloids, and arrays of metal particles bound
to metal or dielectric colloids through short linkages. However, it has often been limited to
electron-rich molecules, like pyridine, aromatic amines or phenols, and carboxylic acids.
Two possible mechanisms for surface-enhanced Raman scattering that are substantially
different from each other have been proposed. One of them is an electromagnetic effect
proposed by Jeanmaire and Van Duyne,53 and the other is a charge-transfer effect or chemical
effect proposed by Albrecht and Creighton.54
4.3.1 Electromagnetic Enhancement
The electromagnetic enhancement relies upon the fact that the electric field provided by
particular surfaces where the molecules adsorbed on can be resonantly enhanced. When
light interacts with the metal surfaces, localized surface plasmons can be excited. When
the frequency of the plasmons ωp is in resonance with the incoming light, the greatest field
enhancement can be obtained. Moreover, the direction of the plasmon oscillation should be
perpendicular to the surface in order to realize the enhancement. Scattering will not occur if
the plasmon oscillations are in-plane with the surface.56 For this reason, roughened surfaces
or nano-particles are typically employed to get large SERS factor. Under these conditions,
surface areas in which the localized collective oscillations can take place are easily identified.
The wavelength of the incoming light is much longer than the features of the surfaces
or particles, only the dipolar contribution of the light needs to be considered. The field
enhancement can be divided into two steps. First, it magnifies the intensity of incident
light with a factor of E2. And then, it magnifies the intensity of the Raman signal by
the same mechanism with the same factor. For the total process, an enhancement factor
of E4 is obtained.57 Noticed, the maximum factor E4 is only for the Raman frequencies
that are slightly shifted from the incident light. In this case, the Raman signals are also
near resonance with the plasmon frequency. When there is larger Raman frequency shift,
the incident light and the Raman signal can not be both on resonance with ωp and thus
show smaller enhancement.58 Because most of SERS studies are in visible and near-infrared
radiation (NIR) ranges, the plasmon resonance frequency should be in these ranges to get
maximal enhancement. For this reason, silver, gold and copper55 are chosen to be typical
metals for SERS with plasmon resonance frequencies match the requirement.
42 CHAPTER 4. RAMAN SPECTROSCOPY
4.3.2 Chemical Enhancement
Figure 4.3: Optimized structure of pyrazine (left) and structure of pyrazine molecule bonded
to two Ag10 clusters (right).
Figure 4.4: Calculated Raman spectra of pyrazine (left) and pyrazine molecule bonded to
two Ag10 clusters (right).
Although the electromagnetic theory gives a general explanation for the enhancement of
SERS regardless of the molecular system involved, it has difficulty to explain experimental
observations for certain specific systems, for instance, molecules with a lone pair of electrons
bonded to the surface. It was found that the enhancement magnitude of such systems are
too large to be explained by the electromagnetic mechanism alone. A chemical enhancement
mechanism involves charge transfer between the adsorbate and the metal surface has then
been proposed. Because electronic transitions of many charge transfer complexes are in the
visible range, resonance enhancement can easily occur. Here we take pyrazine molecule, see
Fig. 4.3, bonded to two Ag10 clusters as an example to demonstrate the possible chemical
enhancement of SERS. From the calculated Raman spectra as shown in Fig. 4.4, a large
enhancement, a factor of 20000, due to the presence of chemical bonds between the molecule
and metal clusters can be easily observed. It is also shown that Raman active modes for
the molecule in metal junction are different from that of a single molecule.
4.4. APPLICATIONS 43
The chemical enhancement occurs mostly for chemically bonded molecule-metal complexes.
For physically adsorbed molecule-metal systems, the chemical enhancement can be occasion-
ally observed for certain specific configurations. It should also mentioned that the chemi-
cal enhancement mechanism is strongly associated with the electromagnetic enhancement
mechanism.59
4.4 Applications
The possibility of using SERS to probe single molecular behavior has attracted considerable
attentions in recent years.60–68 However, it is not a trivial task to determine the existence of
single molecule from SERS measurements due to many uncertainties.69–75 Up to now single
molecular SERS studies have been limited to a few dye molecules that often form strong
chemical bonds with the metal substrate. The formation of the chemical bond between
the molecule and the substrate can lead to relatively stable spectra with strong signal
as a result of the so-called chemical enhancement. Even with these systems, there is no
consensus on how to determine the structure of the single molecule, which has limited
the applicability of single molecular (SM-) SERS in general. In a joint experimental and
theoretical work, we have conducted SM-SERS measurements for a non-bonding molecule,
perylene, physically adsorbed on uniformly assembled colloidal Ag nanoparticles on glass
cover-slips. We have demonstrated that it is possible to control the thermal motion of the
single molecule by lowing the temperature of the sample, and to obtain very stable spectra
with high resolutions. With the help of first principles calculations, the position of perylene
molecule inside the Ag nanoparticles has been determined unambiguously. Our work has
not only significantly widened the scope of practical applications of SM-SERS technique,
but also been conceptually important since it shows for the first time that the chemical
enhancement is not a necessary condition for the generation of SM-SERS spectrum, hence
resolving the long standing debate in the field about the role of the chemical enhancement.
We have tested many model systems with DFT calculations. It is found that only when the
molecule is placed in between two Ag clusters, the calculated Raman spectrum can repro-
duce the most stable experimental spectrum. The agreement between the theory and the
experiment is remarkable, as nicely demonstrated in Fig. 4.5B. These results are consistent
with the observation of Xu et al63 that the molecule likes to sit in between two nanoparticles
with small gap in SM-SERS, which might due to the fact that in this situation the electric
field gradient force for a molecule could become zero and the molecule can safely remain
there.
Based on the evolutions of laser power dependent and temperature dependent experimental
44 CHAPTER 4. RAMAN SPECTROSCOPY
spectra, we could propose a model to describe the dynamic process of a single perylene
molecule on Ag nanoparticles under low temperature as given in Fig. 4.5. When a perylene
molecule is spread on the nanoparticules, the most stable position for the perylene is to lie
on the surface of a nanoparticle. It is energetically unfavorable for a molecule to drop in
between two nanopartciles with a gap of less than 1 nm. Calculations indicate that when the
distance between molecule and the nanoparticle is sufficiently small, around 0.6 nm, more
vibrational modes become Raman active because of the interaction between the molecule
and the metal. As shown in Fig. 4.5A, the calculated spectrum resembles reasonably well
the experimental spectrum taken at the very beginning of the measurement. Under the
guidance of the strong electric field gradient force, the molecule eventually moves into a
gap of two nanoparticles as described by Fig. 4.5C. It should be mentioned that Raman
intensity of the single molecule in the gap shown in Fig. 4.5 is quite similar to that of in the
gas phase, indicating clearly that the chemical enhancement is not the necessary condition
for the observation of single molecular Raman spectrum.
4.4. APPLICATIONS 45
Figure 4.5: (A) Calculated Raman activity of perylene based on the model by assuming a
single free molecule perylene adsorbed near Ag10 clusters, (B) and trapped into two pyra-
midal Ag10 clusters, in comparison with low-temperature experimental SM-SERS spectra.
(C) Schematic draws for the behavior of a perylene molecule near a pair of electronegative
Ag nano-particles: (K and E0 refer to wave vector and electric vector of incident light) the
molecule is initially adsorbed on top of one of the Ag particle, and then moved into the gap
of two particles driven by electric field gradient force.
46 CHAPTER 4. RAMAN SPECTROSCOPY
Chapter 5
Central Insertion Scheme
5.1 Introduction
Nanomaterials in many cases could be considered as molecular complexes. In these systems,
electrons are often confined in a finite region. For this reason, nanostructures have been
expected to possess unique physical and chemical properties different from the corresponding
bulk materials. Nanotechnology rapidly became a growing interdisciplinary research area
since it was first envisioned in the 1980’s. And novel quantum phenomena, such as single-
electron charging effects, quantization of conductance, just name a few, have been exploited
near room temperature.76–78 For instance, a single carbon nanotube can be used to make a
transistor that works at room temperature,79,80 and a single C60 molecule can be operated as
an amplifier.81 Generally, nanomaterials can lead to very good performance with decreased
size, weight, and cost.
It is important to understand the structures and properties of nano-sized systems for the
development of nanotechnology. However, when nanomaterials are lack of infinite period-
icity, it is difficult to get the electronic structures of these relatively large systems with ab
initio or first principle methods because of the O(N3) or worse scaling behavior.
Following the first proposal of Yang,82,83 many computational methods with linear scaling
behavior have been developed. These methods can be divided in two main categories. The
first category is to calculate the whole system at once, combining several efficient techniques:
the “order-N exchange”84 and “near-field-exchange”85 methods for exchange matrix,86 the
fast multipole methods for the Coulombic matrix,87–90 and density matrix search methods91
for replacing the diagonalization of the Fock matrix. The second category is to divide the
whole system into a series of fragments, and construct the properties of the whole system
47
48 CHAPTER 5. CENTRAL INSERTION SCHEME
with the results of quantum chemical calculations of all the fragments.
5.2 Central Insertion Scheme
5.2.1 Basic Idea
The central insertion scheme (CIS) is a method that belongs to the second category.92 It
is based on a simple fact that for a large enough finite periodic system, the interaction
between different units in the middle of the system should be converged. In another word,
the Hamiltonian of those units in the middle becomes identical. It is thus possible to
obtain the Hamiltonian of a larger system by continuously adding the identical units in the
middle of the initial system. Apparently the application of CIS requires to have an initial
Hamiltonian possessing identical central parts, which can only be achieved by computing a
fairly large initial system. Fortunately this condition can be fulfilled routinely by modern
quantum chemistry programs. It should be noted that this central insertion process can be
applied to three-dimensional systems.
5.2.2 Hamiltonian
In CIS method, the initial system can be considered with several parts, including two end
parts, namely L and M , and n central uniform units Ui with periodic structures as described
in Fig. 5.1 (a)I. The wavefunction | Ψη〉 for a certain state of the system obeys
H | Ψη〉 = εη | Ψη〉, (5.1)
where H is the Hamiltonian of the system. When describing a periodic system, the wave-
functions are more suitable to be expanded in terms of the site basis: | L〉 and | M〉 for two
end parts, and | Ki〉 (i runs over U1, U2, ... , Un) for the middle parts. Each of them can
be described as linear combination of atomic orbitals (LCAO) shown as,
| L〉 =∑
Lα
c(L)Lα | ψLα〉;
|M〉 =∑
Mα
c(M)Mα | ψMα〉;
| Ki〉 =∑
iα
c(i)iα | ψiα〉. (5.2)
By defining
Hi,j = 〈ψi | H | ψj〉 , Si,j = 〈ψi | ψj〉 (5.3)
5.2. CENTRAL INSERTION SCHEME 49
Figure 5.1: (a) I and II: Schematic draw of the geometry and Hamiltonian matrix of a n
unit periodic structure, respectively; (b) I and II: The geometry and Hamiltonian matrix
of two stretched periodic structures from (a), respectively; (c) I and II: The geometry and
Hamiltonian matrix of the elongated n+ 1 unit periodic structure, respectively.
50 CHAPTER 5. CENTRAL INSERTION SCHEME
where i and j run over units L, U1, U2, · · · , Un and M , Eq. 5.1 can be described as,
HL,L HL,1 ... HL,n HL,M
H1,L H1,1 ... H1,n H1,M
... ... ... ... ...
Hn,L Hn,1 ... Hn,n Hn,M
HM,L HM,1 ... HM,n HM,M
CηL
Cη1
...
CηN
CηM
= εη
SL,L SL,1 ... SL,n SL,M
S1,L S1,1 ... S1,n S1,M
... ... ... ... ...
Sn,L Sn,1 ... Sn,n Sn,M
SM,L SM,1 ... SM,n SM,M
CηL
Cη1
...
CηN
CηM
(5.4)
where Cηi = (cηi1 c
ηi2 ... c
ηiα)T is the coefficient submatrices of unit L, Ui (i = 1, 2, . . . , n) and
M in the molecular orbital εη, respectively.
5.2.3 Central Insertion Process
The Hamiltonian of the initial system can be represented by a (n+ 2)× (n+ 2) matrix (L,
Ui, M), as shown schematically in Fig. 5.1 (a)II. It should be noted that the using of one
unit to represent each end part is only to simplify the description. And the sizes of them
are not constrained. The structures of the matrices for those two mismatched systems are
illustrated in Fig. 5.1 (b)II and are denoted as (L′
, U′
i , M′
) and (L′′
, U′′
i , M′′
), respectively.
If we put these two mismatched systems together, a new (n+ 3)× (n+ 3) matrix (L′′′
, U′′′
i ,
M′′′
) with the structure of Fig. 5.1 (c)II will be obtained for the elongated system. And in
the constructing of the new matrix, the following relationship,
U′′′
i,j = αU′
i,j + (1 − α)U′′
i−1,j−1, (5.5)
is adopted, where α is a weighting factor depending on the system under investigation. In
most cases, α = 0.5 is considered.
As shown in Fig. 5.1(c)II, there are still many matrix elements can not be generated
by the combination. However, these elements normally represent the long-rang interaction
between two ends of the system, and thus can be set to zero without losing any accuracy. By
repeating this process, Hamiltonian of even larger systems can be constructed. With these
new Hamiltonian, molecular orbitals energies and wavefunctions of the elongated systems
with high accuracy can be calculated.
5.2. CENTRAL INSERTION SCHEME 51
5.2.4 Approximations and Error Control
CIS method relies on two key approximations.92 First, the central parts of the initial system
are assumed to have converged electronic structures. For instance, as shown in Fig. 5.1 (a)I,
the wavefunction, charge density and the Hamiltonian of unit Un2−1 is approximately the
same as that of unit Un2. Second, interaction between long range (e.g., > 20A) units can be
neglected without loss of accuracy. For example, the interaction between two ends (L′′ +U ′′1
and U ′′n+1 +M ′′ as shown in Fig. 5.1 (c)II).
If we define the possible error arises from the first approximation as δ′ and that of the second
approximation as δ′′, the error matrix ∆ can be written as,
∆ = H −H =
δL,L δL,1 ... δL,n δL,M
δ1,L δ1,1 ... δ1,n δ1,M
... ... ... ... ...
δn,L δn,1 ... δn,n δn,M
δM,L δM,1 ... δM,n δM,M
, (5.6)
where H is the true Hamiltonian of the new system, H is the Hamiltonian obtained from
the CIS method, δi,j = δ′i,j + δ′′i,j is the total error.
Let u and λ be the eigenvector and eigenvalue of H , and the Euclidean norm of vector u
satisfies ‖u‖2 = 1. And let r = Hu− λSu = Hu − Hu be the residual vector. Then there
exists an eigenvalue λ of H ,93
|λ− λ| ≤ ‖r‖2
= ‖Hu− Hu‖2
= ‖∆u‖2
≤ ‖∆‖2‖u‖2
= ‖∆‖2
= ρ(∆), (5.7)
where ρ(∆) is the spectral radius of ∆.
Using Gerschgorin’s theorem,93 an approximated estimation of ρ(∆) can be obtained,
ρ(∆) ≤ maxi
∑
j
|δi,j| = ‖∆‖∞, (5.8)
where ‖∆‖∞ is the infinite norm of matrix ∆.
From Eq. 5.8, we can see that, for a large enough system whose central parts have converged
interaction energies, the insertion of a new unit at the center of the system would hardly
affect the convergence and consequently reduce remarkable errors.
52 CHAPTER 5. CENTRAL INSERTION SCHEME
5.2.5 BioNano Lego
“BioNano Lego” is an efficient tool package to proceed CIS using effective parallelization
technique.94 It can be divided into four parts as shown bellow.
Step 1
1. Calculate Hamiltonian H0 and overlap matrix S0 of the initial system L+⋃n
i=1 Ui +M
with well-converged central parts. H0 and S0 are usually sparse Matrices.
2. Perform the screw operator i times and calculate the analytical form of rotation matrix
for coefficient matrices of molecular orbitals RiC .
3. Calculate the weighting factor α for each unit according to the distance from the center
of mass of the system.
4. Compute S−10 by using SuperLU package.95
5. Calculate S−10 H0 by using Sparse BLAS package.96 S0 and H0 are saved in compressed
sparse row (CSR) format, while S−10 H0 is in full matrix format.
6. Check whether the initial system is converged.
Step 2
1. Perform the screw operator N times successively on the initial system to get Hi (i =
1, · · · , N).
Step 3
1. Get the large CIS Hamiltonian HNCIS by combining Hi (i = 0, 1, 2, · · · , N) together.
2. Then calculate the overlap matrix of the large system with the same sparse pattern
of HNCIS.
Step 4
1. Solve the general eigenvalue problems HNCISC
NCIS = SNCN
CISΛNCIS.
5.3. APPLICATIONS 53
5.3 Applications
Figure 5.2: (a) Spatial distributions of wavefunction for LUMO of CNT31 calculated by
GAUSSIAN03 (I) and the CIS method (II). (b) Molecular orbitals of (5, 5) SWCNTs with
N = 9 + 12 × i (i=0, 1, ... ,134) units, corresponding to 1.1 nm to 200 nm in length.
Carbon nanotubes (CNT)are considered to be the most promising material for future molec-
ular and nono-electronics. Single-walled carbon nanotubes (SWCNTs) with finite lengths,
ranging from 10 to 50 nm, have found to possess unique electron transport properties.80
Since the electrons in the CNTs can only propagate in the direction of the tube axis,79,97
CNTs are therefore interesting systems for studying quantum behavior of electrons in one
dimension (1D). However, the finite-length CNTs present a great challenge for the first
principle modeling because of both the involvement of vast number of electrons and the
54 CHAPTER 5. CENTRAL INSERTION SCHEME
breakdown of periodic boundary conditions.
Here we show the result of a (5, 5) metallic SWCNT with 31 layers (CNT31, 3.7nm in length)
with CIS method using 19 layers SWCNT as the initial system. The spatial distribution
of the LUMO of the CNT31 by mapping the Cartesian(xyz)-coordinate into a cylindrical
one has been illustrated in Fig. 5.2(a), which it is almost identical as the one calculated
by GAUSSIAN03 program directly. Very good agreement has also found for other orbitals.
With the CIS, it is possible to calculate very large systems. The electronic structures of
sub-200-nm-long single-walled (5,5) carbon nanotubes (SWCNT(5,5)) are shown in Fig.
5.2(b). As expected, the density of states (DOS) gets higher for longer SWCNTs. It is also
interesting to see that the energy gaps (Eg) oscillate periodically with the increase of the
length of the tube.
Bibliography
[1] J. Zyss. Nonlinear Opt. 1991, 1: 3
[2] J. Zyss and I. Ledoux. Chem. Rev. 1994, 94: 77
[3] C. C. J. Roothaan. Rev. Mod. Phys. 1951, 23: 69
[4] B. O. Roos and P. E. M. Siegbahn. The direct Configuration Interaction Method from
Molecular Intergrals, in Methods of Electronic Structure Theory (H. F. Schaefer III,
ed), Plenum Press, New York, 1977
[5] J. Cızek. J. Chem. Phys. 1966, 45: 4256
[6] C. Møller, M. S. Plesset. Phys. Rev. 1934, 46: 618
[7] L. H. Thomas. Proc. Camb. Phil. Soc. 1927, 23: 542
[8] E. Fermi. Rend. Accad. Lincei. 1927, 6: 602
[9] E. Fermi. Z. Phys. 1928, 48: 73
[10] P. Hohenberg and W. Kohn. Phys. Rev. 1964, 136: B864
[11] S. H. Vosko, L. Wilk and M. Nusair. Can. J. Phys. 1980, 58: 1200
[12] J. P. Perdew and Y. Wang. Phys. Rev. B. 1986, 33: 8800
[13] A. D. Becke. Phys. Rev. A. 1988, 38: 3098
[14] J. P. Perdew and Y. Wang. Phys. Rev. B. 1986, 33: 8822
[15] C. Lee, W. Yang and R. G. Parr. Phys. Rev. B. 1988, 37: 785
[16] A. D. Becke. J. Chem. Phys. 1993, 98: 5649
[17] J. Olsen and P. Jørgensen. J. Chem. Phys. 1985, 82: 3235
55
56 BIBLIOGRAPHY
[18] Goppert-Mayer. Ann. Phys. 1931, 9: 273
[19] W. Denk, J. H. Strickler and W. W. Webb. Science. 1990, 248: 73
[20] D. A. Parthenopoulos and P. M. Rentzepis. Science. 1989, 245: 843
[21] J. E. Ehrlich, X. L. Wu, I.-Y. S. Lee, Z.-H. Hu, H. Rockel, S. R. Marder and J. W.
Perry. Opt. Lett. 1997, 22:1843
[22] A. A. Said, C. Wamsely, D. J. Hagan, E. W. Van stryland, B. A. Reinhardt, P. Roderer
and A. G. Dillard. Chem. Phys. Lett. 1994, 228: 646
[23] G. S. He, R. Gvishi, P. N. Prasad and B. Reinhardt, Opt. Commun. 1995, 117: 133
[24] J. H. Strickler and W. W. Webb. Opt. Lett. 1991, 16: 1780
[25] D. L. Pettit, S. S. H. Wang, K. R. Gee and G. J. Augustine. Neuron. 1977, 19: 465
[26] B. H. Cumpston, S. P. Anthavel, S. Barlow, D. L. Dyer, J. E. Ehrlich, L. L. Erskine,
A. A. Heikal, S. M. Kuebler, I.-Y. S. Lee, D. McCord-Maughon, J. Qin, H. Rockel, M.
Rumi, G. Subramaniam, W. W. Webb, X.-L. Wu and C. Xu. Science. 1999, 398:51
[27] P. R. Monson and W. M. McClain. J. Chem. Phys. 1970, 53: 29
[28] J. Jortner and M. Ratner, Molecular Electronics. Blackwell Science, Oxford, 1997
[29] P.N. Parasad, D.J. Williams, Introduction to Nonlinear Optical Effects in Molecules
and Polymers. Wiley, New York, 1991
[30] L.R. Dalton, W.H. Steier, B.H. Robinson, C. Zhang, A. Ren, S. Garner, A. Chen, T.
Londergan, L. Irwin, B. Carlson, L. Fifield, G. Phelan, C. Kincaid, J. Amend and A.
Jen. J. Mater. Chem. 1999, 9: 1905
[31] Y. Shi, C. Zhang, J. H. Bechtel, L. R. Dalton, B. H. Robinson, W. H. Steier. Science.
2000, 288: 119
[32] J. Zyss, Molecular Nonlinear Optics: Materials, Physics and Devices. Academic Press,
Boston, 1993
[33] A. B. P. Lever. Strem chem. 1986, 11: 1
[34] T. Hara, M. Toriyama and K. Tsukagoshi. Bull. Chem. Soc. Jpn. 1983, 56: 2267
BIBLIOGRAPHY 57
[35] T. H. Stanton, D. C. Schindele, G. E. Renzoni, B. V. Pepich, N. H. Anderson, J. A.
Clagett and K. E. Opheim. International Patent Cooperation Treaty. Publication No.
WO88/04777
[36] J. D. Spikes. Photochem. Photobiol. 1986, 43: 691
[37] W.S. Chan, R. Svenson, D. Phillips and I. R. Hart. Br. J. Cancer. 1986, 53: 225
[38] N. Brasseur, H. Ali, R. Langlois, J. R. Wagner, J. Rousseau and J. E. van Lier. Pho-
tochem. Photobiol. 1987, 45: 581
[39] J. R. Wagner, H. Ali, R. Langlios, N. Brasseur and J. E. van Lier. Photochem. Photobiol.
1987, 45: 587
[40] N. Brasseur, H. Ali, L. Rejean and J. E. van Lier. Photochem. Photobiol. 1988, 47: 705
[41] J. D. Spikes and J. C. Bommer. Radiat. Biol. 1986, 1: 41
[42] N. Sh. Lebedeva. Russian Chemical Bulletin. 2004, 53: 2674
[43] M. Gouterman, in: D. Dolphin (Ed.), The Porphyrins, vol. III. Academic Press, New
York, 1978, p. 1
[44] T. Reinota, J. M. Hayesb, G. J. Smalla and M. C. Zernerc. Chem. Phys. Lett. 1999,
299: 410
[45] T. Verbiest, K. Clays, C. Samyn, J. Wolff, D. Reinhoudtand A. Persoons. J. Am. Chem.
Soc. 1994, 116: 9320
[46] Y.Q. Tu, Y. Luo and H. Agren. J. Phys. Chem. B. 2005, 109: 16730
[47] A. H. Compton. Phys. Rev. 1923, 21: 483
[48] A. Smekal. Naturwiss. 1923, 11: 873
[49] C. V. Raman and K. S. Krishnan. Nature. 1928, 121: 501
[50] H. A. Kramers and W. Heisenberg. Z. Phys. 1925, 31: 681
[51] P. A. M. Dirac. Proc. Roy. Soc. 1927, 114: 710
[52] M. Fleischmann, P. J. Hendra and A. J. McQuillan. Chem. Phys. Lett. 1974, 26: 163
[53] D. L. Jeanmaire and R. P. Van Duyne. J. Electroanal. Chem. 1977, 84: 1
58 BIBLIOGRAPHY
[54] M. G. Albrecht and J. A. Creighton. J. Am. Chem. Soc. 1977, 99: 5215
[55] J. A. Creighton and D. G. Eadon. J. Chem. Soc. Faraday Trans. 1991, 87: 3881
[56] E. Smith and G. Dent. Modern Raman Spectroscopy: A Practical Approach. John Wiley
and Sons, 2005
[57] M. Moskovits. Surface-Enhanced Raman Spectroscopy: a Brief Perspective in Surface-
Enhanced Raman Scattering: Physics and Applications. Springer, 2006, p. 1-18
[58] A. Campion and P. Kambhampati. Chem. Soc. Rev. 1998, 27: 241
[59] J. R. Lombardi, R. L. Birke, T. Lu and J. Xu. J. Chem. Phys. 1986, 84: 4174
[60] S. Nie and S. R. Emory. Science. 1997, 275: 1102
[61] J. A. Dieringer, R. B. Lettan II, K. A. Scheidt and R. P. Van Duyne. J. Am. Chem.
Soc. 2007, 129: 16249
[62] K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. R. Dasari and M. S.
Feld. Phys. Rev. Lett. 1997, 78: 1667
[63] (a) H. Xu, J. Aizpurua, M. Kall and P. Apell. Phys. Rev. E. 2000, 62: 4318; (b) H. Xu,
E. J. Bjerneld, M. Kall and L. Borjesson. Phys. Rev. Lett. 1999, 83: 4357
[64] F. Svedberg, Z. Li, H. Xu and M. Kall. Nano Lett. 2006, 6: 2639
[65] J. Jiang, K. Bosnick, M. Maillard and L. Brus. J. Phys. Chem. B. 2003, 107: 9964
[66] P. J. Moyer, J. Schmidt, L. M. Eng and A. J. Meixner. J. Am. Chem. Soc. 2000, 122:
5409
[67] Y. Sawai, B. Takimoto, H. Nabika, K. Ajito and K. Murakoshi. J. Am. Chem. Soc.
2007, 129: 1658
[68] A. Otto. J. Raman Spec. 2002, 33: 593
[69] T. Lemma and R. F. Aroca. J. Raman Spec. 2002, 33: 197
[70] A. A. Moore, M. L. Jacobson, N. Belabas, K. L. Rowlen and D. M. Jonas. J. Am.
Chem. Soc. 2005, 127: 7292
[71] S. Habuchi, M. Cotlet, R. Gronheid, G. Dirix, J. Michiels, J. Vanderleyden, F. C. De
Schryver and J. Hofkens. J. Am. Chem. Soc. 2003, 125: 8446
BIBLIOGRAPHY 59
[72] E. J. Bjerneld, Z. Foldes-Papp, M. Kall and R. Rigler. J. Phys. Chem. B. 2002, 106:
1213
[73] T. O. Shegai and G. Haran. J. Phys. Chem. B, 2006, 110: 2459
[74] E. C. Le Ru, M. Meyer and P. G. Etchegoin. J. Phys. Chem. B. 2006, 110: 1944
[75] Z. Wang and L. J. Rothberg. J. Phys. Chem. B, 2005, 109: 3387
[76] L. L. Sohn, L. P. Kouwenhoven and G. Schon, (eds). Mesoscopic Electron Transport.
Kluwer, Boston, 1997
[77] J. R. Heath and M. A. Ratner. Phys. Today. 2003, 56: 43
[78] A. Nitzan, and M. A. Ratner. Science. 2003, 300: 1384
[79] S. J. Tans, M. H. Devoret, H. J. Dai, Andreas Thess, Richard E. Smalley, L. J. Geerligs
and Cees Dekker. Nature. 1997, 386: 474
[80] A. Javey, P. Qi, Q. Wang and H. Dai. Proc. Natl. Acad. Sci. USA. 2004, 101: 13408
[81] C. Joachim and J. M. Gimzewski. Chem. Phys. Lett. 1997, 265: 353
[82] W. Yang. Phys. Rev. Lett. 1991, 66: 1438
[83] W. Yang and T. Lee. J. Chem. Phys. 1995, 103: 5674
[84] E. Schwegler and M. Challacombe. J. Chem. Phys. 1996, 105: 2726
[85] J. C. Burant, G. E. Scuseria and M. J. Frisch. J. Chem. Phys. 1996, 105: 8969
[86] G. E. Scuseria. J. Phys. Chem. A. 1999, 103: 4782
[87] L. Greengard and V. Rokhlin. J. Comput. Phys. 1987, 73: 325
[88] M. C. Strain, G. E. Scuseria and M. J. Frisch. Science. 1996, 271: 51
[89] C. A. White, B. G. Johnson, P. M. W. Gill and M. Head-Gordon. Chem. Phys. Lett.
1996, 253: 268
[90] C. A. White and M. Head-Gordon. J. Chem. Phys. 1994, 101: 6593
[91] X. Li, J. M. Millam, G. E. Scuseria, M. J. Frisch and H. B. Schlegel. J. Chem. Phys.
2003, 119: 7651
60 BIBLIOGRAPHY
[92] J. Jiang, K. Liu, W. Lu and Y. Luo. J. Chem. Phys. 2006, 124: 214711. ibid, 2006,
125: 149902
[93] Y. Saad. Numerical Methods for Large Eigenvalue Problems (first edition). Halsted
Press, New York, 1992
[94] BioNano Lego V1.0 is a tool package for central insertion scheme approach written by
B. Gao, J. Jiang, K. Liu and Y. Luo. Royal Institute of Technology, Sweden, 2007
[95] J. W. Demmel, S. C. Eisenstat, J. R. Gilbert, X. Li and J. Liu. SIAM J. Matrix Anal.
Appl. 1999, 20: 720
[96] I. S. Duff and C. Vomel. ACM Trans. Math. Softw. 2002, 28: 268
[97] L. C. Venema, J. W. G. Wildoer, J. W. Janssen, S. J. Tans, H. L. J. T. Tuinstra, L. P.
Kouwenhoven and C. Dekker. Science. 1999, 283: 52