dottorato di ricerca in nanoscience and …modification of ruthenium bipyridyl complex with a...
TRANSCRIPT
DOTTORATO DI RICERCA IN
NANOSCIENCE AND NANOTECHNOLOGY
XXII CICLO
Sede Amministrativa
Universita degli Studi di MODENA e REGGIO EMILIA
TESI PER IL CONSEGUIMENTO DEL TITOLO DI
DOTTORE DI RICERCA
Time Dependent DFT Investigation of
Optical Properties and Charge Dynamics
in Light-Harvesting Assemblies
Candidato: Nicola Spallanzani
Relatori: Prof.ssa Franca Manghi
Dr. Carlo Andrea Rozzi
a Chiara e Greta
i
ii
As far as the laws of mathematics refer to reality, they are not certain;
and as far as they are certain, they do not refer to reality.
A. Einstein, “Geometry and Experience”, January 27, 1921.
Contents
1 Introduction 1
1.1 Third Generation Photovoltaics . . . . . . . . . . . . . . . . . 3
2 Time-Dependent Density Functional Theory 11
2.1 An introduction to DFT . . . . . . . . . . . . . . . . . . . . . 11
2.1.1 The Hohenberg-Kohn Theorem . . . . . . . . . . . . . 12
2.1.2 Degenerate Ground States and v-Representability . . . 14
2.1.3 The Kohn-Sham Equations . . . . . . . . . . . . . . . 15
2.2 Excited States . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.1 The Runge-Gross Theorem . . . . . . . . . . . . . . . . 18
2.2.2 Time Dependent Kohn-Sham Equations . . . . . . . . 21
2.3 Photo-absorption Spectra . . . . . . . . . . . . . . . . . . . . 22
3 Carotenoid-Porphyrin-C60 Molecular Triad 25
3.1 Photochemistry of Carotenoporphyrin-Fullerene Triads . . . . 30
3.2 Why TDDFT? . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4 Ground State 35
4.1 Geometry Optimization . . . . . . . . . . . . . . . . . . . . . 35
4.2 Convergence Tests . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3 Ground State Electronic Structure . . . . . . . . . . . . . . . 39
4.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5 Excited States 45
5.1 Propagation Scheme . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Photoabsorption cross-section . . . . . . . . . . . . . . . . . . 46
iii
iv CONTENTS
5.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6 Charge Transfer Dynamics 53
6.1 Photo-induced Charge SeparationMechanism . . . . . . . . . . 54
6.2 Simulation of the Charge Transfer Process . . . . . . . . . . . 56
6.3 Charge-density evolution . . . . . . . . . . . . . . . . . . . . . 57
6.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Bibliography 68
Chapter 1
Introduction
The quest for efficient ways to exploit renewable energy sources has become
one of the key problems of our times. The idea of a solid-state device capable
of transforming solar light energy into electric potential energy is quite old,
and enormous progress has been made in optimizing traditional junction-
based devices with respect to their thermodynamical efficiency and produc-
tion cost [1].
Figure 1.1 shows the energy band diagram of a standard p–n junction
Figure 1.1: Energy band diagram of a p–n junction with energy conversion
loss processes.
1
2 Chapter 1. Introduction
Figure 1.2: Solar electricity costs as function of module efficiency and cost.
solar cell. The right-hand side of the figure represents the n-type region of
the device where a semiconductor is doped by adding some amount of an
element with more valence electrons than the semiconductor. The left-hand
side represents a p-type region obtained by doping the same semiconductor
with an element with less valence electrons. When these two types of doped
semiconductors are in contact, once the equilibrium is reached the junction
is to behave as an elementary photovoltaic unit. As shown in Figure 1.1, a
photoexcited electron in the p side of the junction is under the influence of
an electrochemical potential gradient that moves the electron to the n side.
The n-type region is well predisposed to accept electrons due to the presence
of the dopant that yields a conduction band where electrons can flow easily
to and from cells contacts. The mobility of the charge carriers produces the
current flow.
Figure 1.2 shows the fully amortized cost of electricity as a function of
the efficiency and cost of an installed photovoltaic module for several classes
of devices. The single-band gap solar cell modules, like the one described
above, lie in zone I, whereas zones II and III are related to second and third
1.1. Third Generation Photovoltaics 3
generation devices respectively. The theoretical efficiency limit for even an
optimal single-band gap solar conversion device is 31%, because photons with
an energy below the band gap of the absorber material cannot generate a
hole-electron pair, and so their energy is not converted to useful output and
the absorbed energy is dissipated into hat. Photons with an energy larger
than the band gap energy rapidly release heat to the ions lattice and only a
fraction of the energy can be converted into useful electric potential. In addi-
tion a carrier able to reach the p-n junction could lead to the recombination
process, since one electron and one hole recombine and thereby annihilate
the associated free charge. The carriers that recombine do not contribute to
the generation of electrical current.
Recently with the up-coming of nanosciences and nanotechnologies, the
study and the control of matter at an atomic and molecular scale (generally
structures of the size 100 nanometers or smaller), entirely new classes of
materials have been proposed as building blocks for next-generation solar
cells. Among these nanostructured semiconductors, organic-inorganic hybrid
and supra-molecular assemblies appear to be the most promising [2].
The problems concerning the manipulation of matter at nano size were
first illustrated by the physicist Richard Feynman in the famous lecture
“There’s Plenty of Room at the Bottom” at an American Physical Society
meeting at Caltech on December 29, 1959. He pointed that as the sizes get
smaller, we would have to reconsider the physics of the system because the
relative strength of various forces would change. Gravity would become less
important, surface tension would become more important as well as Van der
Waals attraction, etc. This opened the doors to a new science dedicated at
the study of the “nano-world”, the nanoscience.
1.1 Third Generation Photovoltaics
In green plants photosynthesis light-harvesting complexes, systems that ab-
sorb photons with very high efficiency, are able to capture solar energy and
to store it in a charge-separated state. The solar energy is first used to create
an excited state of the molecule, and it is later transferred in a reaction cen-
ter. This reaction center is composed by several electron acceptor molecules
4 Chapter 1. Introduction
(a) (b)
Figure 1.3: (a) Gold nanoparticle functionalized with pyrene (from Ref. [2]).
(b) Binding of fluorophore-linked gold nanoparticles to nanostructured film
using a bifunctional surface modifier (from Ref. [3]).
disposed in a chain. The first molecule transfers its excited electron to the
next creating an electron transport chain. The solar energy is thus converted
into potential energy in terms of charge-separation.
Organic-Inorganic Hybrid Assemblies
Organic-inorganic nanohybrids can be developed by mimicking the photoin-
duced electron-transfer process of natural photosynthesis [3], tailoring the
optoelectronic properties of metal nanoparticles. This nanohybrids are made
by assembling monolayers of organic molecules containing functional groups,
such as amines, thiols, isothiocynate and silanes, on three-dimensional surface
of metal nanoparticles, such as gold nanoparticles. Such monolayer protects
the metal cluster and can be functionalized with a chromophore monolayer.
This type of spherical shape hybrid clusters exhibit efficient light-harvesting
capability and suppresses undesirable energy transfer quenching of singlet-
excited-state of the chromophores due to the fact that they are densely packed
on metal nanoparticle surfaces (Figure 1.3(a)).
The nonmetallic property of metal nanoparticles can also be utilized to
accept and then to transfer the electrons from an excited sensitizer to another
acceptor molecule bound to the surface [3]. For example, a gold-nanoparticle
1.1. Third Generation Photovoltaics 5
(a) (b)
Figure 1.4: Design of semiconductor-metal nanocomposites using (a) coupled
(from Ref. [4]) and (b) core–shell (from Ref. [5]) geometry.
could behave like a bridge for a chromophore and a TiO2 surface, in which
a bifunctional surface-linking molecule such as mercaptopropionic acid was
used to link the gold nanoparticle to the TiO2 surface (thiol group to gold
and carboxylic group to TiO2) (Figure 1.3(b)).
Semiconductor/Metal Nanocomposites
Of particular interest is the use of semiconductor nanostructures for solar
hydrogen production by the photocatalytic splitting of water. An example
are the TiO2–Gold composite nanoparticles [4], where the gold nanoparticles
capped with organic molecules exhibit unusual redox activity by readily ac-
cepting electrons from the charged semiconductor TiO2 nanostructure. The
Au-mediated electron transfer to an electron acceptor, such as C60, can be
followed by introducing Au nanoparticles to UV-irradiated TiO2 suspension
(Figure 1.4(a)).
This type of coupled metal–semiconductor structure is obtained by expos-
ing both metal and oxide surfaces to reactants, products, and the medium.
Corrosion or dissolution of the noble metal particles during the operation of
photocatalytic reaction is problematic. A better synthetic design would be
to employ the metal as a core and a semiconductor photocatalyst as a shell
[5]. For example, the Ag@TiO2 core–shell clusters are able to store electrons
under UV irradiation and discharge them on demand in the dark (Figure
6 Chapter 1. Introduction
(a) (b)
Figure 1.5: (a) Schematic of operation of a dye-sensitized electrochemical
photovoltaic cell (from Ref. [6]). (b) Ru-complex adsorbed on TiO2 nanopar-
ticle (from Ref. [7]).
1.4(b)).
Photoinduced charge separation in a semiconductor nanoparticle can be
greatly improved by coupling it with another semiconductor particle having
favorable energetics. By selecting a small band gap semiconductor, one can
also harvest photon in the visible, and by controlling the particle size of quan-
tum dots, one can vary the energetics of the particles. Opportunities exist to
develop ordered assemblies of small and large band gap semiconductors and
harvest photons over a wide spectral range of the visible and infrared light
with better charge separation.
Dye-Sensitized Solar Cells
The first example of solar cells in which the units responsible of the photo-
excitation and of the charge collection are separated, are the so-called dye-
sensitized (or Gratzel) solar cells [8, 6] (Figure 1.5(a)). The dye-modified
semiconductor films provide an efficient method to mimic the photosynthetic
process. The charge separation in this case is facilitated by a semiconduc-
tor particle. In this devices the nanostructured TiO2 films modified with
a ruthenium complex (Figure 1.5(b)) exhibit photoconversion efficiencies in
1.1. Third Generation Photovoltaics 7
the range of 11%, which is comparable to that of amorphous silicon-based
photovoltaic cells. When the electrode is illuminated with visible light, the
sensitizer molecules absorb light and inject electrons into the semiconductor
particles. These electrons are then collected at the conducting glass surface
to generate anodic photocurrent. The charge injection from excited sensitizer
into semiconductor nanoparticles is an ultrafast process occurring on the time
scale of femtoseconds to nanoseconds. The redox couple (e.g., I−3 /I−) present
in the electrolyte quickly regenerates the sensitizer. The iodide is regenerated
in turn by the reduction of triiodide at the counter-electrode, the circuit be-
ing completed via electron migration through the external load. The voltage
generated under illumination corresponds to the difference between the Fermi
level of the electron in the solid and the redox potential of the electrolyte.
Overall the device generates electric power from light without suffering any
permanent chemical transformation.
Recently this type of cells have been subject of several studies toward fine-
tuning the performance of electrode structure, pretreatment of TiO2 surface,
modification of ruthenium bipyridyl complex with a variety of functional
groups, regenerative redox couples, and electrolyte medium. Despite this
large body of work, the maximum attainable efficiency has remained in the
range of 10-11%.
Donor-Acceptor Assemblies
In the class of supra-molecular assemblies the donor-acceptor dyads and tri-
ads are object of careful experimental study. In fact they are considered an
ideal mimic for the primary photo-synthetic process, which basically consists
of a photo-excitation followed by a charge-transfer between the component
units [9]. The research for ideal components of light-harvesting dyads has
led to porphyrin-C60 assemblies [10, 11] (Figure 1.6(a)).
Porphyrin molecules show extensive absorption of visible light, and an
optical gap of approximately 2 eV and are therefore ideal photo-reaction
centers and electron donors. C60, on its side, is the ideal electron acceptor.
It has a particularly deep triply degenerate LUMO at approximately -4.3 eV,
capable of accommodating six electrons from neighboring molecules. More-
8 Chapter 1. Introduction
(a) (b)
Figure 1.6: (a) Porphyrin-C60 dyad (from Ref. [9]) and (b) oligo(p-
phenylene vinylene)-C60 dyad (from Ref. [14]).
over the highly conjugated nature of its bonds, together with a cage radius
of 3.45 A, stabilizes the extra charges in a very efficient way, so that little
repulsive forces are felt by the excess electrons.
It is interesting the use of a porphyrin-C60 dyad as a light receptor in
the spectral sensitization of nanostructured tin oxide semiconductor (SnO2)
electrode [12, 13]. In the sensitization of wide band gap semiconductors,
the energy difference between the conduction band edge of an n-type semi-
conductor film and the oxidation potential of an excited dye present as an
adsorbate provides a driving force for interfacial state charge injection.
From a study of the photo-induced electron transfer process in poly-
mer/fullerene blends [15] derives the idea of oligo(p-phenylene vinylene)-
fullerene dyads [14] (Figure 1.6(b)). The most relevant difference is that,
for covalently bound dyads, intramolecular energy and electron transfer pro-
cesses compete on short timescales after initial excitation. In fact, intramolec-
ular photoinduced charge transfer in these dyads in solution occurs via a two-
step mechanism that involves an initial photoinduced intramolecular singlet-
energy transfer from the oligomer to the fullerene moiety, followed by an
intramolecular charge transfer reaction. But, a second important difference
with polymer/fullerene blends is that the lifetime of the chargeseparated
state in a dyad in solution is much shorter.
A further step toward solar cells based on these components consisted in
transforming the donor-acceptor unit into a donor-bridge-acceptor unit. This
switching from dyads to triads has the main purpose of obtaining a better
1.1. Third Generation Photovoltaics 9
charge separation and a longer life time of the final (excited) state, that will
be the subject of the present study.
Outline
The thesis is organized as follows: in the first part of chapter 2 is introduced
the basic formalism of the Density Functional Theory (DFT) illustrating
the scheme proposed by Kohn and Sham that is based on the Hohenberg-
Kohn theorem. Then, in the second part of the chapter, we show how to
extend the theory on time dependent fenomena. In fact Time-Dependent
Density Functional Theory (TDDFT), based on the Runge-Gross theorem,
allow us to describe time dependent fenomena like photon-absorption and
quantum dynamics. In chapter 3 is introduced the system studied with a
large overview of the photochemistry of Carotenoporphyrin-Fullerene Triads
from an experimental point of view. From chapter 4 it starts the description
of our calculations with the discussion of the results. In particular the chapter
4 is focused on the ground state properties of the system while chapter 5
introduce the study of the excited states with the analysis of the photo-
absorption spectra. The thesis concludes in the chapter 6 with the most
relevant novelty, the study of the charge transfer dynamics that is the crucial
step in the realization of third-generation photovoltaics.
10 Chapter 1. Introduction
Chapter 2
Time-Dependent Density
Functional Theory
The aim of this chapter is to introduce the basic formalism of the Density
Functional Theory (DFT) and its time-dependent extension (TDDFT). We
start section 2.1 with the statement of two important theorems that are the
foundation of the DFT. In section 2.2 we illustrate how to extend the basic
idea of DFT to time-dependent phenomena.
2.1 An introduction to DFT
The problem of the determination of ground state properties of a quan-
tum many-body system is formulated in terms of the solution of the static
Schrodinger equation. This procedure is applicable only for small systems.
As pointed by W. Kohn in his Nobel Lecture, in January 28 1999, for systems
with a number of interacting electrons N > O(10), we have a so called “ex-
ponential wall” that makes the traditional wavefunction method unfeasible.
The aim of DFT is to replace the many-body wave-function, a very com-
plex mathematical object that depends on 3N variable, with a simple func-
tion, the charge density, that depends solely on the three spatial coordinates.
The advantage is clear in particular from a computational point of view. In
this section we illustrate the scheme proposed by Kohn and Sham that is
based on the Hohenberg-Kohn theorem.
11
12 Chapter 2. Time-Dependent Density Functional Theory
2.1.1 The Hohenberg-Kohn Theorem
The Hohenberg-Kohn (HK) theorem [16] states that the ground state expec-
tation value of any observable O is a unique functional of the exact ground
state density n(r)
〈Ψ[n]|O|Ψ[n]〉 = O[n] .
To prove the HK theorem we start considering a many-fermion system
characterized by a non-relativistic time-independent Hamiltonian H, com-
posed by a kinetic contribution T , a Coulomb (two-particle interaction) con-
tribution W , and an external potential V that belongs to a set V of local
one-particle potentials
H|Ψ〉 = (T + W + V )|Ψ〉 = Egs|Ψ〉 , V ∈ V . (2.1)
The states |Ψ〉 are collected in a set Ψ of antisymmetric ground state
wavefunctions, that solves the fermions eigenvalue problem, and defines the
surjective map
C : V → Ψ .
Calculating the ground state density
n(r) = N∑
σ2...σN
∫
dr2 . . .∫
drN |Ψ(rσ, r2σ2, . . . , rNσN )|2 ,
where N is the number of fermions
N =∑
σ
∫
dr3nσ(r) ,
another surjective map is generated collecting the ground state densities in
the set N
D : Ψ → N .
Now we demonstrate that the maps C and D are also injective. For the
map C we consider the eigenvalue problem of a ground state |Ψ′〉
H ′|Ψ′〉 = (T + W + V ′)|Ψ′〉 = E ′gs|Ψ
′〉 , V ′ ∈ V , (2.2)
2.1. An introduction to DFT 13
where the Hamiltonian differs from the one in (2.1) only by the external po-
tential. With the assumption |Ψ〉 = |Ψ′〉 and subtracting the two Schrodinger
equations (2.1) and (2.2) we obtain
(V − V ′)|Ψ〉 = (Egs − E ′gs)|Ψ
′〉 ,
and
V = V ′ + (Egs − E ′gs) . (2.3)
The last equation shows that the potentials differ only by an additive
constant, thus they are equivalent. For the map D we start to calculate the
ground state energy as the expectation value of the Hamiltonian, and for
|Ψ〉 6= |Ψ′〉 we get
Egs = 〈Ψ|H|Ψ〉 < 〈Ψ′|H|Ψ′〉 = 〈Ψ′|H ′ + V + V ′|Ψ′〉
then
Egs < E′gs +
∫
n′(r)[v(r) − v′(r)]d3r .
A similar procedure for E ′gs leads to
E ′gs < Egs +
∫
n(r)[v′(r) − v(r)]d3r ,
and additioning the last two inequalities with the assumption n(r) = n′(r)
we obtain
Egs + E ′gs < Egs + E ′
gs . (2.4)
The contradiction (2.4) proves that the maps C and D are also injective
and thus bijective. In particular the unique inversion of the map D
D−1 : n(r) → |Ψ[n]〉
proofs the Hohenberg-Kohn (HK) theorem.
In particular, if our observables are the kinetic energy and the two-particle
interaction, we can define the functional
FHK [n] = 〈Ψ[n]|T + W |Ψ[n]〉
14 Chapter 2. Time-Dependent Density Functional Theory
that is universal, i.e., valid for any number of fermions and does not depend
on the one-particle external potential. The energy functional now can be
written as
Ev0[n] = 〈Ψ[n]|T + W + V0|Ψ[n]〉 = FHK [n] +
∫
d3rv0(r)n(r) (2.5)
and another important statement, of the HK theorem, shows its variational
character: the exact ground state density can be determined by minimization
of the energy functional
E0 = minn∈N
Ev0[n] . (2.6)
2.1.2 Degenerate Ground States and v-Representability
In the case of a system with degenerate ground states we can define a subset
ΨV as a linear combination of the states that solve the eigenvalue problem
(2.1), related to a potential V ∈ V , where the union of all subset gives the
set Ψ. It is also possible to define a subset NV that groups all the ground
state densities related to a potential V ∈ V , where the union of all subset
gives the set N .
In this case the relation C cannot be considered a map, because one
potential now is associated with more then one ground state. Moreover the
map D is now not invertible.
We can overcome the problem proving that (CD) is a proper map and it
is invertible. Kohn [17] showed that the ground state energy corresponding
to the unique external potential V [n] of a given density n(r) is the same for
all the degenerate ground states |Ψi〉 ∈ ΨV
〈Ψi|T + W + V [n]|Ψi〉 = 〈Ψj |T + W + V [n]|Ψj〉 ≡ E .
Thus
FHK [n] = E −∫
n(r)v([n]; r)d3r
is an unique functional of the density and the relations (2.5) and (2.6) are
conserved.
2.1. An introduction to DFT 15
A function n(r) is termed pure-state v-representable if it is the density
of a (possibly degenerate) ground state of the Hamiltonian (2.1) with some
suitably chosen local external potential v(r) (the number of particles, N , and
their mutual interaction w(r, r′) being specified). But not all “reasonably
well behaved” non-negative functions are pure-state v-representable. A first
extension of the domain of the functional FHK [n] was proposed by Levy and
further investigated by Lieb [18] and Levy [19]. The functional is defined as
FLL[n] = infΨ→n
〈Ψ|T + W |Ψ〉 ,
where the notation Ψ → n indicates that the minimum must be searched
over all antisymmetric, normalized N -particle functions Ψ(r1σ1, . . . , rNσN )
whose density equals the prescribed function n(r). As a consequence of the
Rayleigh-Ritz principle FLL[n] = FHK [n] for all pure-state v-representable
functions n(r), thus
Ev0[n] = FLL[n] +
∫
n(r)v0(r)d3r .
2.1.3 The Kohn-Sham Equations
The HK theorem establishes a one-to-one correspondence between a given
local external potential v(r) and the ground state density n(r) of a N -fermion
system. Now we need a scheme to make the theorem suitable for solving
actual quantum-mechanical problems.
To achieve this Kohn and Sham introduce an auxiliary system of N non-
interacting particles denoted with the letter “s” and described by the same
density as the interacting one n(r) = ns(r).
Following the HK theorem there exists a potential vs(r) generated by a
given density n(r), thus the single-particle orbitals obtained by solving the
Schrodinger equation(
−h2
2m∇2 + vs(r)
)
ϕi(r) = εiϕi(r) (2.7)
with
n(r) =N∑
i=1
|ϕi(r)|2 , (2.8)
16 Chapter 2. Time-Dependent Density Functional Theory
are unique functional of the density, ϕi(r) = ϕi([n]; r) as well as the non-
interacting kinetic energy
Ts[n] =N∑
i=1
∫
ϕ∗i (r)
(
−h2
2m∇2
)
ϕi(r)d3r .
Now considering an interacting system, with external potential v0(r) and
ground state density n0(r), we can rewrite its total energy functional in terms
of the non-interacting kinetic energy
Ev0[n] = Ts[n] +
∫
d3rv0(r)n(r)
+1
2
∫ ∫
d3rd3r′n(r)w(r, r′)n(r′) + Exc[n] ,
where the exchange-correlation (xc) functional is defined as
Exc[n] = FHK [n] −1
2
∫ ∫
d3rd3r′n(r)w(r, r′)n(r′) − Ts[n] .
Defining the xc potential as
vxc([n0]; r) =δExc[n]
δn(r)
∣
∣
∣
∣
∣
n0
we can obtain the auxiliary potential, the so called Kohn-Sham potential,
vs,0(r) = v0(r) +∫
d3r′w(r, r′)n0(r′) + vxc([n0]; r) . (2.9)
Equations (2.7) (2.8) and (2.9) allow us to apply the Kohn-Sham theorem:
the exact ground state density n(r) of an arbitrary interacting system can
be obtained by the self-consistent solution of the following set of equations
(
−h2
2m∇2 + v(r) +
∫
d3r′w(r, r′)n(r′) + vxc([n]; r)
)
ϕi(r) = εiϕi(r)(2.10)
with
n(r) =N∑
i=1
|ϕi(r)|2 .
2.1. An introduction to DFT 17
Finally the ground state total energy of the interacting system is given
by
E0 =N∑
i=1
εi −1
2
∫ ∫
d3rd3r′n(r)w(r, r′)n(r′)
+Exc[n] −∫
vxc([n]; r)n(r)d3r .
This expression contains no approximations and it is able, in principle,
to predict all the ground state properties of a system provided we can write
the exact exchange and correlation functional expression.
Unfortunately we don’t know the exact formulation of the functional
Exc[n], in all real-world applications approximations are therefore always
required. In the present thesis we will use the oldest and most widely used
approximation, the local density approximation (LDA). LDA is based on the
concept that the electron density can be considered locally constant, thus
the xc energy is written as
Exc[n]LDA =∫
d3r εHEG(n)|n=n(r) ,
where εHEG(n) is the xc energy per unit volume of the homogeneous electron
gas. From this expression the xc potential is readily found as
vLDAxc ([n]; r) =
δ
δnεHEG(n)|n=n(r) .
For systems with a truly slowly varying density one can expect reasonable
results from the LDA. On the other hand one finds that rather favorable
results are also obtained for strongly inhomogeneous systems when the KS
scheme is employed with an xc potential derived from ELDAxc [n]. The method
works surprisingly well in many cases in spite of some deficiencies such as
the inadequate cancellation of self-interaction contributions in the long-range.
As a consequence of this drawback, for instance, the local xc potential does
not exhibit the correct asymptotic behavior proportional to 1/r for localized
systems.
18 Chapter 2. Time-Dependent Density Functional Theory
One reason for the success of the LDA seems to be the fact that the xc
energy, if expressed in terms of an xc hole xc(r, r′) as
Exc[n] =1
2
∫
d3r∫
d3r′n(r)w(r, r′)xc(r, r′) ,
is determined by the spherical average of xc(r, r′) alone. The xc hole in LDA
is spherical and appears to be a good representation of the spherical average
of the true xc hole, in part due to a rather subtle cancellation of errors from
the exchange and the correlation parts.
2.2 Excited States
This section is dedicated to the study of excited states. The simplest approx-
imation in the calculation of the excited energies is to take the differences
between the ground state KS eigenvalues. However is important to empha-
size that the KS eigenvalues, obtained from (2.10), do not have a physical
meaning, thus this procedure is not entirely justifiable. In particular, as
mentioned in section 2.1, with DFT we have a scheme to obtain the expec-
tation value of a ground state observable as a functional of the ground state
density. Thus we are not able to predict correctly the expectation value of
a time dependent observable only with the stationary density. Several tech-
niques have been proposed to overcome this problem, and one of the most
successful for the calculation of excitations in finite system is the TDDFT.
In this section we state the fundamental theorem of TDDFT, and outline a
procedure to extract absorption spectra from the solution of TDDFT equa-
tions. Application to the real-world system of interest here is presented in
Chapter 5.
2.2.1 The Runge-Gross Theorem
The Runge-Gross (RG) theorem [20] is the time dependent extension of the
HK theorem. It states that, given the initial state of a many-fermion system,
there is a one-to-one correspondence between the time dependent external
potential and the time dependent density
v(r, t) ↔ n(r, t) .
2.2. Excited States 19
The proof of the RG theorem, requires the demonstration that if two
potentials, v(r, t) and v′(r, t), differ by more than a purely time dependent
function c(t), they cannot produce the same time dependent density n(r, t).
For simplicity we consider external potentials that are Taylor expandable
with respect to the time coordinate around the initial time t0
v(r, t) =∞∑
k=0
ck(r)(t − t0)k ,
where
ck(r) =1
k!
∂k
∂tkv(r, t)
∣
∣
∣
∣
∣
t=t0
.
Now we define the function
uk(r) =∂k
∂tk[v(r, t) − v′(r, t)]
∣
∣
∣
∣
∣
t=to
, (2.11)
where
∃k≥0 : uk(r) 6= constant
is true for the initial assumption v(r, t) 6= v(r, t) + c(t). We start to demon-
strate this first showing that the current densities, j and j′, generated by v
and v′, are also different. In second step we prove that j 6= j′ implies n 6= n′.
The equations of motion for the current density in the primed and un-
primed systems are
id
dtj(r, t) = 〈Ψ(t)|
[
j(r), H(t)]
|Ψ(t)〉
id
dtj′(r, t) = 〈Ψ′(t)|
[
j(r), H ′(t)]
|Ψ′(t)〉 .
Using the same initial state, at t0 the wave-functions, the densities, and the
current densities are equal in the primed and unprimed systems. Then taking
the difference between the lasts equations for t = t0 we obtain
20 Chapter 2. Time-Dependent Density Functional Theory
id
dt[j(r, t) − j′(r, t)]t=t0 = 〈Ψ0)|
[
j(r), H(t0) − H ′(t0)]
|Ψ(t)〉
= 〈Ψ0)|[
j(r), v(r, t0) − v′(r, t0)]
|Ψ(t)〉
= in0(r)∇[v(r, t0) − v′(r, t0)] .
The right-hand side of the equation differs from zero by assumption, and the
two current densities will consequently deviate for t > t0, as well as, using
the function (2.11) for k > 0,
dk+1
dtk+1[j(r, t) − j′(r, t)]t=t0 = n0(r)∇uk(r) .
This again implies that j(r, t) 6= j′(r, t) for t > t0.
Now we will make use of the continuity equation
∂
∂tn(r, t) = −∇ · j(r, t)
for the primed and unprimed systems and taking the difference, we arrive
at
∂
∂t[n(r, t) − n′(r, t)] = −∇ · [j(r, t) − j′(r, t)] .
As before, using the function (2.11) for k > 0, we obtain at t = t0
∂k+2
∂tk+2[n(r, t) − n′(r, t)]t=t0 = −∇ ·
∂k+1
∂tk+1[j(r, t) − j′(r, t)]t=t0
= −∇ · [n0(r)∇uk(r)] .
We conclude the demonstration noting that the last step of the above
equation could not be zero by the assumption uk(r) 6= const, then n 6= n′
that proof the RG theorem.
As well as in the static case we need now a scheme to make the theorem
suitable for solving actual time dependent quantum-mechanical problems.
2.2. Excited States 21
2.2.2 Time Dependent Kohn-Sham Equations
The time dependent KS equations are an exact reformulation of the time
dependent Schrodinger equation
i∂
∂tϕi(r, t) =
[
−∇2
2+ vKS(r, t)
]
ϕi(r, t) , (2.12)
where vKS(r, t) is the potential operating on an auxiliary system of non-
interacting fermions, and the density of the interacting system is obtained
from the KS orbitals
n(r, t) =occ∑
i
|ϕi(r, t)|2 . (2.13)
We can rewrite the KS potential as the sum of three terms
vKS = vext(r, t) + vH(r, t) + vxc(r, t) ,
where, vext(r, t) is the external potential, vH(r, t) is the classical electrostatic
interaction, the so called Hartree potential, and vxc(r, t) is the Exchange and
Correlation potential.
This last term in DFT is formally obtained as functional derivative of
the Exchange and Correlation energy. But since we don’t know its exact
expression, we are obliged to introduce approximations.
Time-dependent exchange-correlation functionals
The exact xc potential has a functional dependence on n(r, t) that is nonlocal
in space and time, i.e., contains information about the previous history of
the system, including its initial state. However, almost all present applica-
tions of TDDFT employ an adiabatic approximation, ignoring all functional
dependence of the xc potential on prior time-dependent densities.
In fact we can obtain the time dependent xc potential using the so called
adiabatic LDA (ALDA) that considers the xc potential as a functional local
in time
vALDAxc (r, t) = vLDA
xc [n](r)|n=n(r,t) .
22 Chapter 2. Time-Dependent Density Functional Theory
For most applications, especially regarding finite systems with no memory
effects the use of adiabatic functionals is well justified. Major failures are
on the other hand evident when ALDA is applied to infinite systems. For
systems where the electron density is not slowly varying it is necessary to go
beyond ALDA; for a discussion of the role of non adiabatic approximations
see for example Refs. [21, 22, 23].
2.3 Photo-absorption Spectra
We conclude this chapter showing how the TDDFT allows us to obtain photo-
absorption spectra by simplifying the perturbation theory or by propagating
in real-time the time dependent KS equations.
Linear Response Theory
From perturbation theory the quantity that allows us to calculate the optical
absorption spectrum is the linear change of the density, i.e. the response
of the system when perturbed by a weak time dependent external electric
potential δvext(r, t). At the first order of a perturbative expansion the time
dependent perturbation of the density is related to the perturbing potential
by the following expression
δn(r, ω) =∫
dr′χ(r, r′, ω)δvext(r′, ω) ,
where χ(r, r′, ω) is the linear density-density response function of the system.
Unfortunately, the evaluation of χ through perturbation theory is a very
demanding task. TDDFT is able to simplify this process calculating the
induced density using the KS auxiliary system
δn(r, ω) =∫
dr′χKS(r, r′, ω)δvKS(r′, ω) ,
where χKS(r, r′, ω) is the density-density response function of a system of
non-interacting electrons. Thus applying first order perturbation theory to
KS equations (2.10), χKS can be expressed in term of the ground state KS
eigenvalues εi and eigenfunctions ϕi
χKS(r, r′, ω) =∑
ij
(fi − fj)ϕi(r)ϕ
∗j(r)ϕj(r
′)ϕ∗i (r
′)
ω − (εi − εj) + iη,
2.3. Photo-absorption Spectra 23
where fi is the occupation number of the KS orbitals.
We can now define the dynamical polarizability in terms of the linear
response function as
αij(ω) = −∫ ∫
drdr′xiχKS(r, r′, ω)xj .
Finally the photo-absorption cross-section is proportional to the imaginary
part of the dynamical polarizability averaged over the three spatial directions
σ(ω) =4πω
c
1
3ℑ[Trαij(ω)] , (2.14)
where c is the velocity of light.
Full solution of the KS equations
If we don’t want to restrict the perturbation to a weak field, or if we want to
study both linear and non-linear response functions, a very efficient method
consists in the direct solution of the time dependent KS equations (2.12).
In Quantum Mechanics a time dependent problem is mathematically defined
as an initial value problem. Using as starting point the KS ground state,
obtained from a DFT calculation, we can perturb it with an instantaneous
electric field δυext = −k0xνδ(t), where xν = x, y, z. All frequencies of the
system are excited with equal weight, and at time t = 0+ we have
ϕi(r, t0+) = eik0xν ϕi(r) ,
where ϕi(r) are the KS ground state wave-functions. Now the orbitals are
propagated in time using the KS time-evolution operator U
ϕi(r, tf) = U(tf , t0+)ϕi(r, t0+) = T exp
[
−i∫ tf
t0+
dτHKS(τ)
]
ϕi(r, t0+) .(2.15)
Once the desired propagation time is reached we can obtain the change
of the density through equation (2.13) as
δn(r, t) = n(r, t) − n(r) , (2.16)
24 Chapter 2. Time-Dependent Density Functional Theory
where n(r) is the ground state density of the system. With a Fourier transfor-
mation in frequency space of the induced density the dynamical polarizability
can be obtained directly from its definition:
ανν(ω) = −1
k0
∫
d3rxνδn(r, ω) ,
Finally the photo-absorption cross-section can be calculated by equation
(2.14).
Approximation of the time-evolution operator
In order to reduce the error in propagation from t0+ to tf , the time interval
is splitted into smaller sub-intervals of duration ∆t. The orbitals are then
propagated from tn to tn + ∆t, step by step, until to reach tf . The simplest
approximation to the time-evolution operator U is a direct expansion of the
exponential in a power series of ∆t
U(t + ∆t, t) ≈k∑
l=0
[−iH(t + ∆t/2)∆t]l
l!+ O(∆tk+1) . (2.17)
Unfortunately, this expression does not retain the unitarity of the operator,
and it makes propagation unstable. To remedy the defect it is possible to
use the enforced time-reversal symmetry method [24]. This procedure relies
on the fact that we can obtain the state at time t + ∆t/2 either by forward
propagating the state a t by ∆t/2, or by backward propagating the state at
t + ∆t. In fact we can write
ϕ(t + ∆t/2) = U(t + ∆t/2, t)ϕ(t)
= U(t − ∆t/2, t + ∆t)ϕ(t + ∆t) ,
that imposing the inverse relation U−1(t + ∆t, t) = U(t − ∆t, t) becomes
ϕ(t + ∆t) = U(t + ∆t/2, t + ∆t)U(t + ∆t/2, t)ϕ(t) . (2.18)
The whole procedure amounts to the following steps: i) obtain an estimate of
KS wave-functions at time t+∆t using a “low quality” formula (for example
the expression (2.17) expanded to third or forth order); ii) construct an
approximation to H(t + ∆t) and U(t + ∆t/2, t + ∆t); iii) apply the equation
(2.18). This procedure leads to a very stable propagation.
Chapter 3
Carotenoid-Porphyrin-C60
Molecular Triad
In this thesis we study the electronic structure, the absorption spectrum, and
the charge dynamics of the Carotenoid-Porphyrin-C60 supra-molecular triad
(abbreviated hereafter as triad or C-P-C60), depicted in Figure 3.1.
Figure 3.1: Structure formula for C-P-C60.
As mentioned in Chapter 1, this type of system is considered an ideal
mimic for the primary photosynthetic process, that occurs in green plant and
bacterial photosynthetic membrane. In these biological systems, the mem-
brane is composed by a lipid bilayer containing several protein components,
each of which has a specific role in the photo-induced process. An antenna
component captures the solar photon energy. The photo-excitation in the an-
tenna migrates to a reaction center where it is converted to chemical energy
in the form charge separation across the bilayer in a series of steps involv-
25
26 Chapter 3. Carotenoid-Porphyrin-C60 Molecular Triad
ing several donor-acceptor cofactors. The photosynthesis culminates in the
production of ATP, the natural energy storage. Thus the question is: what
are the design requirement for an artificial antenna plus reactor center? A
minimal design includes an electron donor chromophore that absorbs visible
light, and an electron acceptor. In this molecular assembly triad a covalent
bond joins a diaryl-porphyrin, the donor chromophore, with a pyrrole-C60,
the acceptor, and provides electronic coupling that allows photoinduced elec-
tron transfer. The β-carotenoid is an additional electron donor moiety and
it is joined to the diaryl-porphyrin by an amide connector. The role of the
β-carotenoid is to prolong the lifetime of the charge separated final state by
removing the hole from the porphyrin.
Carotenoids
The carotenoids are compounds made by eight isoprenic units and contain-
ing rings and some conjugated double bonds. They are insoluble in water,
but soluble in fats and in hydrocarbon solvents; all the carotenoids, which
generally have colors from yellow to red, are widespread in nature, both in
plants and animals.
Lycopene is a precursor of carotenoids, from which all those known by
various changes such as reduction of part of the double bonds, the cycliza-
tion, the isomerization of double bonds position and the introduction of oxy-
genated functional groups. Note that in the lycopene all the double bonds are
in trans configuration and that, except at the two extremities of the molecule,
they are all conjugated. This extended system of conjugation is responsible
for the color of these compounds, which are the pigments of many plants.
Figure 3.2: Structure formula for lycopene.
27
The carotene, the pigment in carrots, is particularly interesting because
it is closely related to vitamin A, a diterpenoid essential for growth, health
of the membranes and the view. In carrots are present in abundance two
isomers of carotene, which differ from each other for the position of one of
the double bonds, and both, in biological assays, show strong activities as
vitamin A.
Figure 3.3: Structure formula for β-carotene and α-carotene.
Figure 3.4: Structure formula for vitamin A.
The way in which carotenes derived from lycopene by cyclization of both
ends it is evident. β-carotene, in which all the double bonds are conjugated, it
is more abundant that α-carotene in which one double bond is not conjugated
with the others. The molecule of vitamin A is half of the carotene molecule
and it has an alcoholic function, and, unlike the crystalline carotene that has
an intense red-orange color, it is pale yellow.
28 Chapter 3. Carotenoid-Porphyrin-C60 Molecular Triad
Nitrogen Heterocycles
The pyrrole, which has a pentagonal ring structure with nitrogen in one of
the corners, plays an important role in the chemistry of living organisms.
The conversion of light energy supplied by the sun in the energy of chem-
ical bonds of carbohydrates, synthesized by green plants, occurs by means
of compounds known as chlorophylls. The fundamental structural charac-
teristics of chlorophylls is a system of four pyrroles rings held together by
bridges, each of which contains one carbon atom. This ring system, which is
called porphin, is also present in heme, associated with proteins hemoglobin
and myoglobin, which are responsible for the transport and accumulation of
oxygen in body tissues of warm-blooded animals.
Figure 3.5: Structure formula for porphin.
The porphin, with four nitrogen atoms of the rings pyrrole, directed to-
ward the center of its extensive ring system, complexes effectively with metal
ions. In heme the ion is iron (II), in the chlorophylls it is the magnesium
(II), and these metal ions can also complex with additional bonds above and
below the heterocyclic plane. Cyclic porphin systems can have different sub-
stituents on peripheral atoms; at substituted porphins is given, however, the
general name of porphyrins.
The ring system of porphyrins is very stable and has aromatic character;
the extended conjugation present in these compounds is responsible for their
intense blue color. Porphin and heme have 22 π electrons, but only 18 of these
are part of a ring system for which is possible to write resonance structures.
29
Figure 3.6: Resonance structure formula for porphyrins.
Buckminsterfullerene
In 1985, Harold Kroto of the University of Sussex and Richard Smalley of
the Rice University have announced the discovery of stable structures con-
sisting of only carbon atoms [25], different from graphite and diamond. The
more abundant and more stable of the new structures was a C60 molecule
(720 amu), and since then has been the subject of intense research. The C60,
when it is isolated from soot, is a mustard-colored solid, soluble in common
organic solvents, especially in solvents such as benzene and toluene, which
form magenta solutions. C60 has a very high melting points and, although
they are soluble in organic solvents, dissolve very slowly, which shows that
its molecules are packed very well in its respective crystal lattices.
The spectrum of Nuclear Magnetic Resonance (NMR) of carbon-13 of C60
shows a single peak at 142.68 ppm in the aromatic region of the spectrum,
indicating that all the C60 carbon atoms are equivalent to each other. This is
only possible with a highly symmetrical, confirmed by the high melting point
and by compact crystal structure packaging of C60. The proposed structure
is the family of football, whose technical name is truncated icosahedron; in
this structure each carbon atom is at the junction of two esatomic rings and
one pentatomic ring, and each esatomic ring contain three double bonds.
To C60 species has been given the name Buckminsterfullerene in honor of
Buckminster Fuller, inventor of the architectural structures called geometri-
cal domes, but it is more familiarly called “buckyball”. The term fullerenes
has been chosen to indicate the class of structures consisting of penta- and
esatomic rings, for example, the C70 is a fullerene.
30 Chapter 3. Carotenoid-Porphyrin-C60 Molecular Triad
Figure 3.7: Buckminsterfullerene (C60).
Chemists know how to easily get the C60, by evaporating graphite with
lasers, and how to purify it, and this has allowed to study the C60 from a
chemical point of view. The studies made so far indicate that the C60 is
not exactly aromatic such as benzene, for example, the NMR peak is more
typical for a tensioned aromatic structure.
Electrophilic compounds shall be added easily to double bonds. Further-
more, from chlorinated C60 it is possible, for nucleophilic substitution, add
aliphatic chains.
3.1 Photochemistry of Carotenoporphyrin-Fullerene
Triads
The observed absorption spectrum of a similar triad in a toluene solution is
shown in Figure 3.8. Superimposed on the bands at 629, 576, and 411 nm
that are caracteristic of the porphyrin is the broad carotene absorption with
maxima at 509, 477, and 448 nm. The weak fullerene absorption throughout
the visible with a long-wavelength maximum at 705 nm is also present, but
the amplitude is too low to be seen in the figure. The absorption spectrum is
essentially a linear combination of the spectra of porphyrin, carotenoid, and
3.1. Photochemistry of Carotenoporphyrin-Fullerene Triads 31
Figure 3.8: Absorption spectra in toluene solution at 292 K of a triad [26]
(—), and a model porphyrin (- - -), carotenoid (− · − · −), and fullerene
(− · ·− · · −). Also shown is a linear combination of the spectra of the model
compounds that approximates the spectrum of the triad (· · ·).
fullerene compounds. Thus, electronic interaction among the chromophores
is too weak to cause significant spectral perturbations. D. Kuciauskas et
al. [26] have observed in benzonitrile and 2-methyltetrahydrofuran small
(≤3 nm) spectral shifts to longer wavelengths.
Steady-state fluorescence emission studies of a triad in polar solvents,
with excitation at wavelengths where the porphyrin and fullerene absorb,
demonstrated very strong quenching of the emission from both the porphyrin
and fullerene chromophores. This is ascribed to rapid photoinduced electron
transfer involving either excited state. Excitation of the carotenoid moiety
did not lead to the observation of any carotenoid fluorescence, as the fluo-
rescence quantum yields of carotenoids of this type are ≤ 10−4. Studies of
carotenoporphyrin molecules have shown that singlet-singlet energy transfer
from the carotenoid to the porphyrin is negligible, and that charge-separated
species that might be generated by carotenoid excitation are not observed.
32 Chapter 3. Carotenoid-Porphyrin-C60 Molecular Triad
Figure 3.9: Relevant transient species created after excitation of triad 2 and
interconversion pathways. The energies of the charge-separated states are
estimated from the cyclic voltammetric data obtained in polar solvents, as
reported in Ref. [26]
Thus, the carotenoid is photochemically inert in the singlet manifold of the
triad.
Transient absorption experiments were performed to learn more about
electron-transfer processes in a P-C60 dyad and a C-P-C60 triad by Kuci-
auskas et al. [26]. The samples were excited at 590 nm with 200-fs laser
pulses. Their conclusion were that the first excited singlet state of the por-
phyrin moiety in the dyad, 1P-C60, decays at least in part by photoinduced
electron transfer to yield the P+-C−60 charge-separated state. The same result
was obtained for the triad and, after the formation of C-P+-C−60, the tran-
sient absorption experiments revealed a strong new absorption assigned to
the carotenoid radical cation of the C+-P-C−60 charge-separated state.
The electron-transfer events leading to the final state in the triad are
shown in Figure 3.9. Excitation at 590 nm yields mostly C-1P-C60, which
decays by step 3 to yield C-P+-C−60. The fullerene excited singlet state also
undergoes photoinduced electron transfer to give the same species. Electron
3.2. Why TDDFT? 33
transfer from the carotenoid (step 4) gives the final C+-P-C−60 state. Kuci-
auskas et al. have observed the formation of the carotenoid radical cation as
a function of time measured at 970 nm in 2-methyltetrahydrofuran at 292,
200, 120, and 77 K. They relate formation time constants of 59 ps (292 K),
90 ps (200 K), 247 ps (120 K), and 615 ps (77 K).
The time-scale of the transient absorption experiments discussed above
is too short to allow observation of the decay of the C+-P-C−60 state. There-
fore, transient absorption spectroscopy on the nanosecond time-scale was
employed by Kodis et al. [27] to investigate the long-lived state. The decay
of the absorption maximum at ∼980 nm is fitted by an exponential function
with a time constant of 57 ns. The decay is accompanied by a complemen-
tary exponential rise (fitted by an exponential function with a 57 nm time
constant) of absorbance at 550 nm. Absorbance at this wavelength is due to
the carotenoid triplet state, and its rise indicates the formation of 3C-P-C60
by charge-recombination of C+-P-C−60. The decay of this absorbance can be
observed on an even longer time-scale, and has a time costant of 4.9 µs in
absence of oxygen.
3.2 Why TDDFT?
While many experimental properties of porphyrin-C60 based dyads and tri-
ads are known, less is known from an ab-initio theoretical point of view. The
main reason is that the large size of the systems yields traditional accurate
quantum-chemistry methods such Configuration Interaction (CI) impractical
or infeasible. Less accurate but computationally efficient methods such as
DFT have successfully been employed to describe the ground state of such
compounds [28], but unfortunately, the prediction of excited state properties,
like the optical absorption spectrum, from the ground state Kohn-Sham or-
bitals and eigenvalues usually leads to results that are in disagreement with
the observed spectra. In particular the role of the interfaces between different
parts of the triad and between the molecule and the solvent are not under-
stood in spite of the fact that they are of paramount relevance to the problem
of improving the energy conversion efficiency of next generation solar cells.
TDDFT is a rigorous and in principle exact method that allows the calcu-
34 Chapter 3. Carotenoid-Porphyrin-C60 Molecular Triad
lation of excited-states through an extension of DFT to the domain of time-
dependent external potentials [20], and therefore allows us to get a much
better insight on all those properties that are most relevant for the photoab-
sorption process, and the subsequent transfer of charge, that are crucial steps
in any photovoltaic device.
Chapter 4
Ground State
Our first goal is to study the ground state properties of the molecule, and to
compare its electronic structure with the electronic structure of the separated
components. For all our calculations we used an optimized geometry obtained
by Baruah et al. [28]. In this structure the molecule can be entirely embedded
in a box approximately 55 × 16 × 20 A wide, the longest side being along
the x axis. As it is also clear following a steric argument, the porphyrin
ring is perpendicular to the aryl rings, which, in turn, are coplanar with
the carotenoid. The optimized geometry shows a slight bent in the carotene
chain (see Figure 4.1). In the calculations of the separated moieties the same
geometries used in the triad were employed.
4.1 Geometry Optimization
The C-P-C60 system is composed by 207 atoms with 632 valence electrons.
Its structure corresponds to the formula C132H68N6O. The geometry opti-
mization was carried out using the LBFGS scheme [29, 30], by Baruah et
al. [28]. Initially the geometries of the separate units were optimized be-
fore constructing the triad. After forming the composite molecular triad,
its geometry was optimized again. Since the molecule has a quite floppy
structure, two different geometries were optimized. In the first geometry, the
three subunits were attached in such a way that the resultant triad has an
elbow shaped structure. Its energy after optimization turned out to be higher
35
36 Chapter 4. Ground State
Figure 4.1: Orthogonal projections of the optimized geometry of C-P-C60.
From top to bottom: xy, xz and yz projections.
by about 0.6 eV and also the dipole moment for the final charge separated
structure is much smaller. The linear geometry as shown in Figure 4.1 was
found to be lower in energy and all subsequent calculations were carried out
on this geometry. In the optimization stage, the process was carried out until
the smallest force was 0.003 eV/A and RMS (root mean square) force was
0.076 eV/A. It can be pointed out here that NMR (nuclear magnetic reso-
nance) studies on carotenoid-porphyrin dyad shows a structure similar to the
linear one [31]. Similar structure was found by Smirnov et al. as the lowest
energy structure based on a molecular mechanics level calculation [32]. We
4.2. Convergence Tests 37
would like to point out that the molecule has a long chainlike structure and
can exhibit substantial temperature induced structural deformation. Since
our ground-state study is aimed at understanding the zero-temperature state,
such calculations are beyond the scope of present work. On the other hand
the temperature will be significant in the charge transfer calculations (see
Chapter 6).
4.2 Convergence Tests
In a real-space based approach the choice of the grid mesh is a critical param-
eter, and the consistency of the calculations must be checked with a careful
set of convergence tests. A cartesian grid was chosen with uniform spacing of
the mesh. We analyze the convergence of the eigenvalues, of the last 6 occu-
pied states and of the first 4 unoccupied states as a function of the distance
between grid points.
The first test was performed to determine the volume necessary to enclose
the states of the molecules. In the open source code octopus [33, 34] it is
possible to choose a particular box shape that is composed putting a sphere
of radius R over all the atoms. The union of all these spheres makes the
box shape that enclose the molecule. As boundary conditions we imposed
the wave-functions to be zero at the surface of the box. This approximation
is adequate only if the surface of the box is far enough from the system,
because if the box is too small the wave-functions will be forced to go to
zero unnaturally. Moreover if the box is too large, will increase the memory
requirement and the computational time because of the large number of grid
points. The test analyzes the convergence of the states changing the values
of radius sphere, and for all the moieties a good convergence was obtained
for value of R ≥ 4.0 A.
The second test was performed analyzing the energy of the states changing
the value of the mesh spacing. This test allow us to find out the largest value
of spacing for which ground state quantities are expect to converge. Figure
4.2 shows that the convergence of the systems was maintained until a spacing
of 0.26 A.
38 Chapter 4. Ground State
−7−6−5−4−3−2−1
0.18 0.2 0.22 0.24 0.26 0.28 0.3 0.32
ener
gy [e
V]
spacing [Å]
β−carotenoid−7−6−5−4−3−2−1
ener
gy [e
V]
Diaryl−porphyrin−8
−7
−6
−5
−4
−3
ener
gy [e
V]
Fullerene (C60)
Figure 4.2: Plots of the last six occupied orbitals (red lines) and first four
unoccupied orbitals (green-dashed lines) for the three moieties of the triad
at different values of spacing.
4.3. Ground State Electronic Structure 39
4.3 Ground State Electronic Structure
We have thus performed a plain DFT calculation at the LDA level, using
the Perdew-Zunger correlation functional [35]. The core electrons were rep-
resented via Troullier-Martins pseudo-potentials [36], and the Interpolating
Scaling Functions method [37] was used to efficiently solve the Poisson’s equa-
tion in the calculation of the Hartree potential. All the simulations in the
present work, a part from the geometry optimization, were performed using
the open source code octopus on a Linux cluster (IBM BCX/5120) com-
posed on Opteron Dual Core 2.6 GHz processors, maintained by CINECA
Bologna. All the quantities were discretized in real-space using a uniform
mesh with 0.2 A spacing that guarantees the convergence of the total energy
of the system [34]. This value of spacing, coupled with a radius R = 4.0 A
of the atomic spheres that compose the simulation box, gives 862,017 points
on the inner mesh, and 1,287,768 points adding the boundary mesh.
In Table 4.1 we report the ground state energies of the highest occupied
molecular orbitals (HOMOs) and the lowest unoccupied molecular orbitals
(LUMOs) for the isolated moieties and for the full triad. The detailed energy
levels for other orbitals are depicted in Figure 4.3.
Comparing the values of the HOMO and LUMO energies of the triad
with those of the isolated moieties we notice that the energy of the HOMO
in the triad and in the isolated β-carotenoid differ by only 40 meV, as well
as the energy of the LUMO in the triad and in the isolated C60. Moreover,
looking at Figure 4.4 and 4.5 we observe that it is possible to find a close
correspondence between many of the orbitals localized on each molecule,
molecules ǫHOMO (eV) ǫLUMO (eV)
β-carotenoid -4.47 -3.49
Diaryl-porphyrin -5.15 -3.31
pyrrole-C60 -5.89 -4.42
Triad -4.51 -4.38
Table 4.1: Energy of HOMO and LUMO respectively for β-carotenoid,
Diaryl-porphyrin, pyrrole-C60 and the whole triad.
40 Chapter 4. Ground State
-6.5
-6
-5.5
-5
-4.5
-4
-3.5
-3E
nerg
y [e
V]
βcar Dpor pyr-C60 Triad
Figure 4.3: Ground state energy levels of the triad and of the separated moi-
eties: β-carotenoid (βcar), Diaryl-porphyrin (Dpor), and Pyrrole-C60 (pyr-
C60).
and orbitals of the full triad. In particular the HOMO wave-function of the
triad is localized on the carotenoid, and it corresponds to the HOMO of the
isolated β-carotenoid, while the LUMO of the triad is fully localized on the
C60 and it corresponds to the LUMO of the C60 alone. The same is for the
HOMO-2 and LUMO+4 of triad that correspond respectively to the HOMO
and LUMO of the isolated Diaryl-porphyrin.
In some other cases the correspondence between the localization of or-
bitals remains, but some degeneracies are lifted, for example the triply degen-
erate C60 LUMO is split into three separate levels in the triad. Finally some
orbitals present a clear mixed character (Figure 4.5), such as the LUMO+5,
LUMO+6, LUMO+7 with mixed P-C60 character, or HOMO-6, with mixed
C-P character. These orbitals however do not appear to have a major role
4.4. Conclusions 41
Figure 4.4: Ground state energy levels of Diaryl-porphyrin and Fullerene.
Some orbitals of the two moieties are depicted.
in the photo-excitation process (see below in Chapter 6).
4.4 Conclusions
From the discussion above we conclude that the electronic structure of the
ground state of the triad is well approximated by the overlap of the electronic
structures of the isolate molecules (Figure 4.3). It is noteworthy that this
property is not an artefact of the local density approximation adopted here,
but it is also an experimentally observed property of the system [27]. This
point will become more clear with the analysis of the excited-state properties
in Chapter 5.
In Chapter 1 we illustrated the basic working mechanism of a single-
42 Chapter 4. Ground State
Figure 4.5: Ground state energy levels of the triad. Some orbitals of the
triad are depicted.
junction cell, and we mentioned that the necessary condition for the electrons
migration is that the valence band of a side of the junction falls into the band
gap of the other side. All the examples of third-generation solar devices
presented in the Chapter 1 are based on the same idea, where the sides
of the junction are replaced by electron donor/acceptor moieties creating
assemblies. In the triad presented here the porphyrin is the donor moiety
and the fullerene is the acceptor moiety.
There are two reason because this type of assembly is considered a good
candidate for third-generation photovoltaics. The first could be detected only
looking the electronic structures of the electron donor, the dyaril-porphyrin,
and of the electron acceptor, the Fullerene. In Figure 4.4 we can notice
4.4. Conclusions 43
the same type of structure of the p–n junction, where the HOMO of the
porphyrin, at -5.15 eV, falls into the HOMO-LUMO gap of the fullerene, gap
of about 1.81 eV, from the HOMO at -6.14 eV to the LUMO at -4.33 eV. Then
the above mentioned necessary condition is fulfilled because the photoexcited
electron can migrate to the almost degenerate LUMO of the C60.
The second important consideration is the presence of mixed P-C60 char-
acter orbitals in the triad at an energy close to the LUMO+4 at -3.39 eV
(see Figure 4.5). This is the state supposed to be the end point of the pho-
toexcitation, and it corresponds to the LUMO of the porphyrin moiety. For
example, the shape of the LUMO+5 and LUMO+6, respectively at only
28 meV and 64 meV above the LUMO+4, suggests the possibility that the
electron migration could be made easier by the presence of these mixed or-
bitals.
As mentioned above, the β-carotenoid is an additional electron donor
moiety. His role will be better explained in Chapter 6.
44 Chapter 4. Ground State
Chapter 5
Excited States
In this Chapter we show the application of TDDFT to the calculation of the
optical absorption spectra of the triad and of its components. The method
used, illustrated in Section 2.3, is recalled and deepened before the results
are analyzed.
5.1 Propagation Scheme
Starting from the ground state Kohn-Sham orbitals ϕj , the system is in-
stantaneously perturbed by a weak external electric potential of magnitude
k0 along the principal cartesian direction (i.e. by applying the external po-
tential δυext(r, t) = −k0xνδ(t)). The magnitude of the perturbation is kept
small in order to keep the dipolar response linear [34]. In this way all the
frequencies of the system are excited with the same weight. The initially
perturbed states therefore become
ϕi(r, t0+) = eik0xν ϕi(r) .
Now the orbitals are propagated in time using the time-evolution operator
U ,
ϕi(r, tf) = U(tf , t0+)ϕi(r, t0+) = T exp
[
−i∫ tf
t0+
dτHKS(τ)
]
ϕi(r, t0+) ,
45
46 Chapter 5. Excited States
and the new wave-functions are used to solve the time dependent KS equa-
tions
i∂
∂tϕi(r, t) =
[
−∇2
2+ vKS(r, t)
]
ϕi(r, t) .
In order to reduce the error in propagation from t0+ to tf , the time inter-
val is splitted into smaller sub-intervals of duration ∆t. The orbitals are then
propagated from tn to tn + ∆t, step by step, until to reach tf . For our cal-
culations we use, to approximate the operator U , the enforced time-reversal
symmetry method (see Section 2.3). This procedure leads to a very stable
propagation.
The numerical discretization of the propagation equations requires that
a finite difference equation is defined on a discretized grid in both space
and time. This lead to a dependence of the time step on the size of space
mesh. In addition the time-step depends on the method used to numerically
calculate the exponential of the Hamiltonian in the propagation operator.
For the triad we used a fourth order Taylor expansion that forces us to use
a small time-step (1.69 as). For the isolated molecules we use the Lanczos
method [38] that is compatible with a larger time-step (7.89 as), resulting
in a window of approximately 8 eV for the described spectral range. The
KS orbitals were propagated for 15 fs in total, that corresponds to a spectral
resolution of about 0.15 eV.
Finally the dynamical polarizability is obtained from
ανν(ω) = −1
k0
∫
drxνδn(r, ω),
where δn(r, ω) is the Fourier transform of n(r, t) − n(r, t = 0). The photo-
absorption cross section is related to the imaginary part of the polarizability
through the expression
σ(ω) =4πω
c
1
3Im[Tr (α(ω))].
5.2 Photoabsorption cross-section
We start our description of the optical properties of the system analyzing the
absorption spectra of the separated parts of the triad. Figure 5.1 reports the
5.2. Photoabsorption cross-section 47
2 2.5 3 3.5 4 4.5 5 5.5 6
Abs
orpt
ion
cros
s se
ctio
n
Photon energy [eV]
EXPTDDFT
2 2.1 2.2 2.3 2.4
a
b
Figure 5.1: Comparison between theoretical TDDFT optical absorption in
diaryl-porphyrin and observed free-base porphyrin absorption as in [39]. In
the inset the region between 2.0 to 2.4 eV is magnified, and the vertical lines
are the transition energies obtained by linear-response calculations.
comparison between theoretical optical absorption in diaryl-porphyrin and
the observed optical absorption in a free-base porphyrin [39]. The TDDFT
spectra is obtained with a real time propagation. In order to get a plausible
attribution of the single-excitation states involved in the process of photo-
excitation, we have also performed a linear response calculation [40, 41] of
the absorption spectrum on the diaryl-porphyrin moiety.
The spectra obtained by the two methods are quite similar, even if they
are not expected to be identical. In particular the analysis of the absorption
of the diaryl-porphyrin allows us to clarify that, even if an intense shoulder
only appears in the UV region around 3.3 eV, the so called B-band of the por-
phyrin, the actual very weak onset of absorption, the Soret Q-band, is in the
orange-yellow visible region at approximately 2.2 eV. In the inset of Figure
48 Chapter 5. Excited States
3 3.5 4 4.5 5 5.5 6 6.5
Abs
orpt
ion
cros
s se
ctio
n
Photon energy [eV]
EXPTDDFT
0.5 1 1.5 2 2.5
Abs
orpt
ion
cros
s se
ctio
n
Photon energy [eV]
Figure 5.2: Left: comparison between theoretical TDDFT optical and ob-
served absorption for fullerene as in [42]. Right: comparison between the
β-carotenoid main absorbance peak as observed in [43] and the calculated
one.
5.1, that magnifies the region between 2.0 to 2.4 eV, are reported the transi-
tion energies obtained by linear response calculation. The excitation named
“a” at 2.14 eV is composed at 60% by the transition HOMO→LUMO+1 and
at 38% by HOMO-1→LUMO. The excitation “b” at 2.32 eV is composed at
56% by the transition HOMO→LUMO and at 42% by HOMO-1→LUMO+1.
The remaining 2% for both the excitation is due to other transition with a
not relevant weight. These two transitions were observed also in the spectra
obtained by time propagation (thick line). The non-symmetric shape of the
large peak at about 2.2 eV suggest that a longer propagation could resolve
two peaks.
In Figure 5.2 left, it is shown the comparison between theoretical TDDFT
optical and observed absorption for fullerene as in [42]. In these spectra we
5.2. Photoabsorption cross-section 49
1 2 3 4 5 6 7 8
Abs
orpt
ion
cros
s se
ctio
n
Photon energy [eV]
Triadβcar
Dporpyr-C60
Figure 5.3: Calculated absorption spectrum of the triad, and the isolated
pyrrole-C60, Diaryl-porphyrin and β-carotenoid.
can notice a good correspondence of the peaks, provided a mean red-shift
of about 0.28 eV is taken into account. The right plot shows the compar-
ison between the β-carotenoid main absorbance peak as observed in [43],
at approximately 1.24 eV and the calculated one. Here a blue-shift is visi-
ble of about 0.47 eV. For an in depth discussion about the effect of exciton
confinement in molecular chains see also Ref. [44].
The TDDFT photo-absorption cross-section of the triad and of the sep-
arated molecules is shown in Figure 5.3. We can easily distinguish one dom-
inant peak at 1.7 eV, several small features between 2.0 and 3.0 eV, another
peak at 3.3 with a shoulder at 3.6 eV, and a group of peaks between 4.0 and
8.0 eV.
The decomposition of the spectrum into optical densities of the isolated
moieties clearly shows again that the total spectrum is very well approxi-
mated by the sum of the spectra of the parts. In particular the main feature
50 Chapter 5. Excited States
0 1 2 3 4 5
Abs
orpt
ion
cros
s se
ctio
n
Photon energy [eV]
EXPTDDFT
DFT
Figure 5.4: Comparison between theoretical and observed optical absorption
for the full C-P-C60. DFT is the calculated spectrum as in [28]. TDDFT
present work. EXP depicts the observed spectrum as reported in [27].
at 1.7 eV appears to be entirely originated by the β-carotenoid, which has
a weak contribution from the rest of the spectral range, apart from a small
feature at 2.3 eV that partially obscures the weak Soret Q-band of the diaryl-
porphyrin. The B-band of the diaryl-porphyrin is the main contribution at
3.2 eV, and the diaryl-porphyrin has another prominent peak in the ultra-
violet region at 6.6 eV. The contribution of pyrrole-C60 counts many peaks,
notably two isolated at 5.3 and 5.9 eV, and several more from 4.0 to 5.0 eV.
The comparison of our calculations with the experimentally observed ab-
sorption in similar triads clearly shows that TDDFT improves on DFT in
describing the main features of the spectra. In Figure 5.4 we compare our
data with the absorption reported in Ref. [27], that refers to a triad in 2-
methyltetrahydrofuran solution. We observe a good one-to-one correspon-
dence between the calculated and the observed peaks, whose positions ap-
5.3. Conclusions 51
pear overestimated in the calculation by an overall rigid shift of about 0.3
eV. We do not investigate here the possible role of the environment, that
might well be responsible for the rigid shift, but is unlikely to qualitatively
change the shape of the spectrum. In fact triads made of slightly different
components, or immersed in different solvents, often have very similar spec-
tra (see Ref. [45]). In particular the prominent porphyrin peak around 3 eV
and the wide band between 2.0 and 3.0 eV seem to be common features in all
the experimental data. The transitions approximately at 2.0 eV corresponds
to the calculated transitions in the diaryl-porphyrin alone, and they are also
responsible for the optical excitation of the triad as a whole.
Also notice that, despite the apparent superposition of the main exper-
imental peak to a peak in the DFT calculation, the nature of the latter is
predicted to be of C60 character, while the corresponding TDDFT feature at
3.5 eV correctly is attributed to the porphyrin (see decomposition in Figure
5.3). The experimental data range in Refs. [27, 45] are limited above approx-
imately 2.0 eV, but as mentioned above, the β-carotenoid absorption does
not appear to be relevant for the photo-conversion in the triad.
5.3 Conclusions
In summary, we have addressed the fundamental problem of describing the
creation of electron-hole pairs and the efficient charge separation in a supra-
molecular assembly of interest for third generation solar devices. Since not
all the effects leading to the final-state charge separation are fully known the
accurate description of the excited states of such an object is a crucial step in
order to understand the details of the photo-absorption and charge-transfer
processes occurring in the system.
We have calculated the optical absorption cross section of such a large
system using TDDFT as a rigorous ground for the excited states dynamics.
In addition to the computational challenge, our calculations demonstrate that
the simple ground-state DFT description of the system is not able to capture
the correct shape of the photo-absorption spectrum. In contrast we have
shown that the main features of the TDDFT spectrum are in good agreement
with the experimental data, and the analysis of the total absorption in terms
52 Chapter 5. Excited States
of the absorption of the isolated moieties indicates that even at the TDDFT
level (at least in the weak field limit) the component molecules in the triad
do not appear to strongly interact.
Finally, in Figure 4.5 we have shown the energy levels for the ground
state of the full triad, and we have emphasized the two orbitals, HOMO-2
and LUMO+4, localized on the porphyrin moiety. From the analysis of the
electronic structures we supposed that these two orbitals are responsible for
the optical excitation. From the analysis of the spectra of the porphyrin we
find that the orbitals that could be involved in the photo-excitation process
are the HOMO-1, HOMO, LUMO and LUMO+1 of the moiety. This allow us
to think that also in the triad there are other orbitals involved in the photo-
excitation, in particular considering the presence of the orbitals with mixed
P-C60 character. As concluding remarks we must recall that the 15 fs time
span of our simulation is unlikely to account for the complete charge transfer
process, that occurs on a different time scale, but it is able to accurately
describe the photo-excitation of the molecule. Nevertheless we have been
able to get a valuable insight about the charge transfer from the careful
examination of the energy levels and of the calculated absorption spectrum
of the system.
Chapter 6
Charge Transfer Dynamics
In Chapters 4 and 5 we have analyzed respectively the ground state and
optical properties of the triad and of its parts. In this chapter we address
the charge transfer dynamics in the whole triad.
The charge transfer dynamics is a crucial step in the realization of third-
generation photovoltaics devices such as Gratzel or donor-acceptor based
systems, and it happens in a subsequent step respect to the photo-excitation
process. Also the time scales are different.
We stress again the needed features of a good candidate for third-generation
photovoltaics. It should be composed at minimum by two units each with
a different role in the photo-induced charge separation mechanism. The
first unit is a chromophore that absorbs visible light and the second unit is
an electron acceptor capable to provide electronic coupling that allows the
photo-induced electron transfer. The central idea is to separate the light
absorption process from the charge collection process, mimicking natural
light harvesting procedures in photosynthesis. In order to make solar cells
efficient, intense visible absorption in addition to rapid charge transfer are
required. The efficiency is determined by two main factors: i) the maximum
photocurrent density, related to the charge transfered rate from the chro-
mophore (donor) HOMO to the acceptor unoccupied orbitals (probably the
LUMO), and ii) the open circuit potential, related to the energy difference
between the HOMO of the electron donor to the LUMO of the acceptor. In
addition to those, other two features make a device a good candidate, namely
53
54 Chapter 6. Charge Transfer Dynamics
a long mean lifetime of the charge separated state and good photochemistry
stability.
All the properties described above are present in the Carotenoid-Porphyrin-
C60 triad, where an additional unit takes the role of a second electron donor.
The function of the carotene is described in the next sections.
6.1 Photo-induced Charge Separation
Mechanism
Resting on the careful examination of the structure, levels, and orbital local-
ization we concluded in Chapter 4 that the triad has the perfect electronic
configuration for a photo-induced charge separation. This occurs thanks to
the fact that the HOMOs and LUMOs of the moieties are localized in the
respective sub-molecule, and right positioned. In particular the HOMO of
the porphyrin falls into the energy range of the HOMO-LUMO gap of the
fullerene, and the presence of mixed P-C60 character orbitals in the triad at
an energy close to the LUMO of the porphyrin makes the electron migration
easier.
We are now able to achieve a theoretically motivated view of the photo-
excitation process, and of the subsequent charge-transfer that can be sum-
marized in the following chain of steps:
C-P-C60 + hν → C-P∗-C∗60
C-P∗-C∗60 → C-P+-C∗−
60 → C+-P-C−60
The first step corresponds to the optical transition from HOMO-2 to
an orbital between LUMO+4 and LUMO+7. We call this excited state
C-P∗-C∗60. The electron in a previously unoccupied orbital on the porphyrin
is now under the influence of the high affinity of the C60, and it is easily
accommodated on the almost-degenerate orbitals LUMO+1, LUMO+2 at
0.14 eV above LUMO localized on the C60. In the assembled triad the three-
fold degeneracy of the C60 T1u LUMO is broken due to the presence of the
Pyrrole and the LUMO of the triad is a not degenered state. We have
6.1. Photo-induced Charge Separation
Mechanism 55
Figure 6.1: Photo-induced charge separation mechanism scheme, ground
state levels (red occupied, green unoccupied) for the triad and its moieties.
The scheme shows the photo-excitation step and the two steps representing
the electron and hole migration.
56 Chapter 6. Charge Transfer Dynamics
then reached a configuration C-P+-C∗−60 . The last step consists in a hole
delocalization from the Porphyrin HOMO to the Carotenoid HOMO to give
the final state C+-P-C−60. Obviously only the study of the full dynamics of
the whole process can support the latter two steps interpretation, which is
inferred here on the basis of the analysis of the energy levels in Figure 6.1,
and on the electrochemical properties of the moieties. Despite the fact that
the lifetime of the final state is estimated to be in the order of tens of ns, it is
possible to get an accurate description of the photo-excitation, and possibly
charge-transfer processes that occur on a much faster time-scale.
As we have already stressed, it is the almost non-interacting character
of the moieties, demonstrated by the shape of the absorption spectrum, and
the nature of the molecular orbitals, that makes it possible to describe this
process in terms of orbitals localized on each component. In fact the charge-
transfer states associated with this molecule are not dipole allowed transitions
due to the fact that the hole on the (donor) carotene and the particle-state
on the C60 (acceptor) molecule are essentially non overlapping.
6.2 Simulation of the Charge Transfer Pro-
cess
In order to focus on the charge transfer process we have chosen as the initial
state for the propagation a single-particle state, presumably close to the
real excited state after a photon absorption. Namely the state obtained by
occupating the LUMO+4 of the triad with one electron, taken from the state
HOMO-2. As mentioned above these two orbitals are almost fully localized
on the porphyrin. This excitation corresponds to an energy increase of about
1.85 eV.
To reduce the computational time for this demanding task, we decided
to use a mesh spacing of 0.24 A, an acceptable value that guarantees the
convergence of the total energy of the system (see Section 4.2 about the tests
performed to obtain a good simulation grid). In addition we increase the
radii, of the spheres that compose the volume of the simulation box, from
4.0 A to 5.0 A. This to be sure to contain the orbitals also for long time
6.3. Charge-density evolution 57
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
0 10 20 30 40 50 60 70
Ele
ctro
nic
Cha
rge
[ele
ctro
ns]
time [fs]
a b
c d
pyrrole-C60Carotenoid-Porphyrin
Figure 6.2: Fraction of electrons transfered to the pyrrole-C60 part (black
line) from the Carotenoid-Porphyrin part (cyan line) of the triad as a function
of time after excitation.
propagations.
We have performed calculations both with clamped and mobile ions, in
order to investigate the role of ion-electron dynamics coupling in the charge-
transfer process. In the mobile case we have assigned random velocities to
the atoms following a Boltzmann distribution with a temperature of 300 K.
6.3 Charge-density evolution
In order to obtain evidence of the charge-transfer process we have partitioned
the simulation box into three regions matching the moieties boundaries along
the longitudinal axis of the molecule. Figure 6.2 shows the evolution of
the excess electronic charge contained in the volume including the pyrrole-
C60 and in the remaining part of the molecule. The black line represents
58 Chapter 6. Charge Transfer Dynamics
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 2 4 6 8 10 12 14 16
Ele
ctro
nic
Cha
rge
[ele
ctro
ns]
time [fs]
dynamical ionsclamped ions
Figure 6.3: Comparison between charge transfer dynamics obtained through
calculations with dynamical ions (solid line) or clamped ions (dashed line)
for the pyrrole-C60 part.
the fraction of electronic charge transferred into a volume that contain the
pyrrole-C60 moiety as a function of time. From this plot it is easy to see
that there is an increase of the electronic charge into this volume, which
corresponds to a decrease of electronic charge into the volume containing the
carotenoid and the porphyrin. Less than 70 fs are needed to transfer 0.9 e to
the fullerene.
The fact that the charge transfer is tied to the dynamics of the ions is
visible in the Figure 6.3. The solid line shows the charge transferred to
fullerene when the ions are allowed to move freely, while the dashed line
shows the charge transfer if the triad’s ions are hold fixed. In the last case
the fraction of electron transferred to the fullerene volume after 17 fs is too
small to be considered a trend rather than a fluctuation. This plot clearly
shows that the energy recombination due to a rearrangement of the ions plays
6.3. Charge-density evolution 59
(a) time = 9.31 fs
(b) time = 17.69 fs
(c) time = 41.74 fs
(d) time = 68.15 fs
Figure 6.4: Plots of the time dependent perturbation of the density δn(r, t)
(see equation (2.16)) at different values of time. The letters “a”, “b”, “c”,
and “d” are referred at the points of the Figure 6.2.
60 Chapter 6. Charge Transfer Dynamics
a major role in the process.
In Figure 6.4 we can see the time dependent perturbation of the density
δn(r, t) at different values of the time propagation. The pictures are snap-
shots of an animation visible following the link:
http://www.quantumsolids.unimore.it/animations/triad_density.avi.
It is obtained by computing the difference between the charge density at time
t and the density of the KS ground state (see equation (2.16)). The magenta
isosurface corresponds to a positive difference and the green to a negative
difference. In the figure it is evident an increase of the magenta color during
the time evolution on the outer fullerene region. It means an increase of
charge density in this region. Form Figure 6.4(a) to Figure 6.4(b) it is also
possible to notice a decrease of the charge density on the porphyrin region
in terms of increase of the green isosurface. Form Figure 6.4(c) to Figure
6.4(d) is visible a charge polarization, that starts at about 40 fs, where in
the region between the fullerene and the porphyrin an emptying of charge is
observable. Since we know from Figure 6.2 that a net charge is transfered
from the porphyrin to the fullerene, we interpreted these changes of signs as
a local polarization of the moieties.
6.4 Conclusions
In order to clarify the spectral weight of different excitations in the charge
fluctuations, we have performed a Fourier analysis of the charge variation on
the fixed volumes. We have decomposed the charge density ρ(t) into a linear
term and an oscillating term
ρ(t) −→ At + B + f(t) .
Then we have performed a Discrete Fourier Transform of the oscillating term
alone
f(t) −→ dft[f(t)] = F (ωn) .
The functions in the frequency space are plotted in Figure 6.5. In all
the three regions it is possible to see frequencies of relevant amplitude in the
6.4. Conclusions 61
0
50
100
150
200
0 0.5 1 1.5 2 2.5 3 3.5 4
Am
pliti
de
energy [eV]
pyrrole-C60×50
0
50
100
150
200
Am
pliti
de
Porphyrin×50
0
50
100
150
200
Am
pliti
de
Carotene×50
0 0.05 0.1
0.15 0.2
3 3.2 3.4 3.6
0 0.05 0.1
0.15 0.2
3 3.2 3.4 3.6
0 0.05 0.1
0.15 0.2
3 3.2 3.4 3.6
Figure 6.5: Discrete Fourier transform of the excess of electronic charge con-
tained in the volumes matching the moieties. In the vertical axis is reported
the amplitude of the Fourier coefficients.
energy region up to 1 eV. These small energetic frequencies could be associ-
ated to vibrational modes of the molecule. In addition, in the fullerene and
carotene regions relevant frequencies from 3.0 to 3.5 eV are also present that
could be assigned to electronic modes. As an assessment of our procedure in
Figure 6.6 we have reconstructed the signal back to the time domain. Using
62 Chapter 6. Charge Transfer Dynamics
-0.2
0
0.2
0.4
0.6
0.8
1
0 10 20 30 40 50 60 70
Ele
ctr.
Cha
rge
[e]
time [fs]
pyrrole-C6050 harmonic components10 harmonic components
-2
-1.6
-1.2
-0.8
-0.4
0
0.4
Ele
ctr.
Cha
rge
[e]
Porphyrin50 harmonic components10 harmonic components
-0.4-0.2
0 0.2 0.4 0.6 0.8
1E
lect
r. C
harg
e [e
] Carotenoid50 harmonic components10 harmonic components
0 2.5 5
Figure 6.6: Harmonic reconstruction of the excess of electronic charge con-
tained in the volumes matching the moieties.
only the first 10 frequencies we are able to obtain a good matching with the
main features of the original function. Moreover we needed the first 50 fre-
quencies to reproduce properly also the small fluctuations visible in the inset
of the fullerene plot in the Figure 6.6.
6.5. Summary 63
The next step in the study of the charge transfer dynamics is the attribu-
tion of the frequencies used for the reconstruction to the mode (rotational,
vibrational, electronic, etc. . . ) of the molecule. This could help us to bet-
ter understand the role of the ions dynamic in the charge transfer process.
Some of these modes are visibly indispensable for the charge transfer (see
Figure 6.3) and non influential in the photo-excitation process, probably due
to the different time scale in which these process are brought to completion.
In addition, with a single-particle excitation in a single state (the LUMO+4
of the triad), probably we have provided the amount of energy necessary for
the excitation of other transitions. In the Figure 6.6, the plot concerning the
carotene shows an increase of charge comparable with that on the fullerene.
This could be due to the excitation of some electronic transitions that involve
the orbitals localized on the carotene. For this reason further calculations
could be done performing an excitation with partial occupation of the states
with mixed P-C60 character.
6.5 Summary
In Chapter 4 we showed the theoretical calculations concerning the electronic
structure of the triad and its parts. In particular, it is seen that the energy
levels of the moieties have the necessary requirements for the construction of
a photovoltaic device.
In Chapter 5 we compared the experimental absorption spectra with the-
oretical spectra, showing a good description for the parts. In addition, for the
whole triad we compare our results with the spectra of a previous theoretical
work based solely on the DFT. In this way is possible to see not only the
good agreement of our calculations with the experimental spectrum, as well
as the TDDFT, as compared to DFT, best describes the optical property of
a system like this one.
After the study of the ground state properties and optical properties, of
the triad and its parts, it is possible to state that the system as a whole acts
as the sum of its parts, each of which has a specific role in the photo-induced
charge separation mechanism that makes this molecule a good candidate
for third generation solar energy conversion devices. This fact has allowed
64 Chapter 6. Charge Transfer Dynamics
us to sketch a diagram of all the states involved in the photo-absorption
and charge transfer processes, and indicates that the design of solar devices
based on similar assemblies can be performed by individually addressing the
active components, and then chemically joining them into a unit capable of
transforming the energy of an incoming photon into electric potential energy
stored into a 50 A long dipole.
Finally, the most relevant novelty of this thesis work is described in Chap-
ter 6, where has been made the study of the dynamics of charge transfer
through the moieties of the triad. So we have shown that TDDFT is able
to produce a good dynamic even using local potentials, and that the time-
domain analysis of this process reveals the tight strong interaction between
the dynamics of the ions and the evolution of the electronic cloud in the
molecule.
Bibliography
[1] M. Green. Third Generation Photovoltaics: advanced solar energy con-
version. Springer-Verlag, Berlin, 2004.
[2] P. V. Kamat. Meeting the clean energy demand: Nanostructure archi-
tectures for solar energy conversion. J. Phys. Chem. C, 111:2834–2860,
2007.
[3] K. George Thomas and P. V. Kamat. Acc. Chem. Res., 36:888–898,
2003.
[4] V. Subramanian, E. E. Wolf, and P. V. Kamat. J. Am. Chem. Soc.,
126:4943–4950, 2004.
[5] T. Hirakawa and P. V. Kamat. Langmuir, 20:5645–5647, 2004.
[6] M. Gratzel. Nature, 414:338, 2001.
[7] F. De Angelis, S. Fantacci, A. Selloni, M. Gratzel, and M. K. Nazeerud-
din. Nano Letters, 7:3189–3195, 2007.
[8] B. O’Regan and M. Gratzel. Nature, 335:737, 1991.
[9] D. Gust, T. A. Moore, and A. L. Moore. Mimicking photosynthetic solar
energy transduction. Acc. Chem. Res., 34:40–48, 2001.
[10] D. M. Guldi. Fullerene–porphyrin architectures; photosynthetic antenna
and reaction center models. Chem. Soc. Rev., 31:22–36, 2002.
[11] S. Gadde, D. M. Shafiqul Islam, C. A. Wijesinghe, N. K. Subbaiyan,
M. E. Zandler, Y. Araki, O. Ito, and F. D’Souza. J. Phys. Chem. C,
111:12500–12503, 2007.
65
66 BIBLIOGRAPHY
[12] F. Fungo, L. Otero, C. D. Borsarelli, E. N. Durantini, J. J. Silber, and
L. Sereno. J. Phys. Chem. B, 106:4070–4078, 2002.
[13] T. Hasobe, H. Imahori, S. Fukuzumi, and P. V. Kamat. J. Phys. Chem.
B, 107:12105–12112, 2003.
[14] P. A. van Hal, S. C. J. Meskers, and R. A. J. Janssen. Appl. Phys. A,
79:41–46, 2004.
[15] C. J. Brabec, G. Zerza, G. Cerullo, S. De Silvestri, S. Luzzati, J. C.
Hummelen, and S. Sariciftci. Chem. Phys. Lett., 340:232–236, 2001.
[16] P. Hohenberg and W. Kohn. Inhomogeneous electron gas. Phys. Rev.
B, 136:864–871, 1964.
[17] W. Kohn. Highlights of Condensed Matter Theory. North-Holland, Am-
sterdam, 1985.
[18] E. H. Lieb. Physics as Natural Philosophy. MIT Press, Cambridge,
1982.
[19] M. Levy. Phys. Rev. A, 26:1200, 1982.
[20] E. Runge and E. K. U. Gross. Phys. Rev. Lett., 52:997, 1984.
[21] C. A. Ullrich and I. V. Tokatly. Phys. Rev. B, 73:235102, 2006.
[22] G. Vignale and C. A. Ullrich. Phys. Rev. Lett., 79:4878–4881, 1997.
[23] D. J. Tozer. J. Chem. Phys., 119:12697, 2003.
[24] A. Castro, M. A. L. Marques, and A. Rubio. Propagators for the time-
dependent kohn-sham equations. J. Chem. Phys, 121:3425, 2004.
[25] H. W. Kroto, J. R. Heath, S. C. O’Brien, R. F. Curl, and R. E. Smalley.
Nature, 318:162–163, 1985.
[26] D. Kuciauskas, P. A. Liddell, A. L. Moore, T. A. Moore, and D. Gust.
Photoinduced electron transfer in carotenoporphyrin-fullerene triads:
Temperature and solvent effects. J. Phys. Chem. B, 104:4307–4321,
2000.
BIBLIOGRAPHY 67
[27] G. Kodis, P. A. Liddell, A. L. Moore, T. A. Moore, and D. Gust. Syn-
thesis and photochemistry of a carotene–porphyrin–fullerene model pho-
tosynthetic reaction center. J. Phys. Org. Chem., 17:724–734, 2004.
[28] T. Baruah and M. R. Pederson. Density functional study on a light-
harvesting carotenoid-porphyrin-c60 molecular triad. J. Chem. Phys.,
125:164706, 2006.
[29] D. C. Liu and J. Nocedal. Math. Programming, 45:503, 1989.
[30] J. Nocedal. Math. Comp., 24:773, 1980.
[31] D. Gust, T. A. Moore, A. L. Moore, C. Devadoss, P. A. Liddell, R. M.
Hermant, R. M. Nieman, L. J. Demanche, J. M. DeGranziano, and
I. Gouni. J. Am. Chem. Soc., 114:3590, 1992.
[32] S. N. Smirnov, P. A. Liddell, V. Vlassiouk, A. Teslja, D. Kaciauskas,
C. L. Braun, A. L. Moore, T. A. Moore, and D. Gust. J. Phys. Chem.
A, 107:7567, 2003.
[33] M. A. L. Marques, A. Castro, G. F. Bertsch, and A. Rubio. octopus:
a first-principles tool for exited electron-ion dynamics. Comp. Phys.
Comm., 151:60–78, 2003.
[34] A. Castro, H. Appel, M. Oliveira, C. A. Rozzi, X. Andrade, F. Lorenzen,
M. A. L. Marques, E. K. U. Gross, and A. Rubio. octopus: a tool for
the application of time-dependent density functional theory. Phys. Stat.
Sol. B, 243:2465–2488, 2006.
[35] P. Perdew and A. Zunger. Phys. Rev. B, 23:5048–5079, 1981.
[36] N. Troullier and J. L. Martins. Phys. Rev. B, 43:8861–8869, 1991.
[37] L. Genovese, T. Deutsch, A. Neelov, S. Goedecker, and G. Beylkin.
Efficient solution of poisson’s equation with free boundary conditions.
J. Chem. Phys., 125:74105, 2006.
[38] M. Hochbruck and C. Lubich. SIAM J. Numer. Anal., 34:1911, 1997.
68 BIBLIOGRAPHY
[39] L. Edwards, D. H. Dolphin, M. Gouterman, and A. D. Adler. Vapor ab-
sorption spectra and redox reactions: Tetraphenylporphins and porphin.
J. Mol. Spect., 38:16, 1971.
[40] M. E. Casida. In D. P. Chong, editor, Recent Advances in Density Func-
tional Methods, volume I, page 155. World Scientific Press, Singapore,
1995.
[41] C. Jamorski, M. E. Casida, and D. R. Salahub. J. Chem. Phys.,
104:5134, 1996.
[42] R. Bauernschmitt, R. Ahlrichs, F. H. Hennrich, and M. M. Kappes. J.
Am. Chem. Soc., 120:5052–5059, 1998.
[43] A. S. Jeevarajan, L. D. Kispert, and X. Wu. Spectroelectrochemistry of
carotenoids in solutions. Chem. Phys. Lett., 219:427, 1994.
[44] D. Varsano, A. Marini, and A. Rubio. Optical saturation driven by
exciton confinement in molecular chains: A time-dependent density-
functional theory approach. Phys. Rev. Lett., 101:133002, 2008.
[45] A. C. Rizzi, M. van Gastel, P. A. Liddell, R. E. Palacios, G. F. Moore,
G. Kodis, A. L. Moore, T. A. Moore, D. Gust, and S. E. Braslavsky. En-
tropic changes control the charge separation process in triads mimicking
photosynthetic charge separation. J. Phys. Chem. A, 112:421, 2008.