Theor Chem Acc (2016) 135:14

DOI 10.1007/s00214-015-1769-5

REGULAR ARTICLE

Coumarin‑based donor–π–acceptor organic dyes

for a dye‑sensitized solar cell: photophysical properties
and electron injection mechanism
Supawadee Namuangruk1 · Siriporn Jungsuttiwong2 · Nawee Kungwan3 ·
Vinich Promarak4 · Taweesak Sudyoadsuk4 · Bavornpon Jansang5 · Masahiro Ehara6

Received: 14 September 2015 / Accepted: 20 November 2015 / Published online: 28 December 2015
© Springer-Verlag Berlin Heidelberg 2015

Abstract The electronic structure and photophysi- frontier orbitals and charge separation were analyzed, and
cal properties of five coumarin-based donor–π–acceptor the thiophene linker was found to be effective for the elec-
(D–π–A)-type organic dyes for a dye-sensitized solar cell tron injection in DSSC. The adsorption of these dyes on the
(DSSC) which were recently developed have been investi- TiO2 anatase (101) surface and the electron injection mech-
gated using the time-dependent density functional theory anism were also investigated using a dye–(TiO2)38 cluster
and the symmetry-adapted cluster configuration interac- model employing PBE and TD-CAM-B3LYP calculations,
tion method. Theoretical calculations including the sol- respectively. The adsorption energies of these dyes were
vent effect in state-specific and linear-response scheme estimated to be ~14 kcal/mol, indicating strong adsorption
reproduced the experimental UV–Vis absorption spectra of dye to a TiO2 surface by carboxylate group. The possible
of these dyes satisfactorily. The π-spacers, thiophene and direct electron injection mechanism was suggested in the
thiophene–phenylene mixed units, affect the planarity of present coumarin-based D–π–A dyes in a dye–TiO2 inter-
the molecular structures which is relevant to the photophys- acting system.
ical properties and charge polarization. Energy levels of the
Keywords Coumarin · Dye-sensitized solar cells · DFT ·
SAC-CI · Donor–π-spacer–acceptor
Published as part of the special collection of articles “Health and
Energy from the Sun.”

This work is dedicated to the UNESCO International Year of 1 Introduction

Light and Light-based Technologies (IYL 2015). TCA Special
Issue on Health and Energy from the Sun: a Computational
Perspective.
Over the past decade, solar cells based on semiconductors
with small band gaps as well as dye-sensitized solar cells
Electronic supplementary material The online version of this (DSSCs) have attracted great attention as possible low-cost
article (doi:10.1007/s00214-015-1769-5) contains supplementary
material, which is available to authorized users.
3
* Siriporn Jungsuttiwong Department of Chemistry, Faculty of Science, Chiang Mai
[email protected] University, Chiang Mai 50200, Thailand
4
* Masahiro Ehara School of Molecular Science and Engineering,
[email protected] Vidyasirimedhi Institute of Science and Technology,
Wangchan, Rayong 21210, Thailand
1
National Nanotechnology Center, National Science 5
PTT Research and Technology Institute, PTT Public
and Technology Development Agency, Klong Luang,
Company Limited, Phahonyothin Rd., Sanubtub, Wangnoi,
Pathumthani 12120, Thailand
Ayutthaya 13170, Thailand
2
Department of Chemistry and Center of Excellence 6
Institute for Molecular Science and Research Center
for Innovation in Chemistry, Faculty of Science, Center
for Computational Science, 38 Nishigonaka, Myodaiji,
for Organic Electronic and Alternative Energy, Ubon
Okazaki 444‑8585, Japan
Ratchathani University, Ubon Ratchathani 34190, Thailand

13
14 Page 2 of 13 Theor Chem Acc (2016) 135:14

photovoltaic devices alternative to conventional silicon- Stability of dyes in its excited and ionized states is impor-
based solar cells. The first DSSC system was developed tant for efficiency and durability. These requirements often
in 1991 by Grätzel’s group [1]. They reported that DSSCs conflict to each other; therefore, flexible molecular design
based on ruthenium (Ru) complexes such as N3 and N719 is necessary. In this sense, organic dyes that are compris-
as dye sensitizer achieve considerable photoelectric conver- ing from donor (D), π-conjugation unit (π), and acceptor
sion efficiencies of >12 % at the standard air mass 1.5 sun- (A) have potential flexibility for molecular design as DSSC
light [2, 3]. On the other hand, Ru-based dyes (Ru dyes) dyes [6–8]. Recently, so-called type II mechanism, namely
have difficulties, like limited Ru resources, difficult purifi- direct injection of an electron from dye to conduction band
cation relative to organic dyes, generally moderate absorp- (CB) of TiO2 by one step, has attracted the interest because
tion intensity, and an undesirable environmental impact. this type of electron injection makes the sensitization sim-
Thus, research is also directed toward searching for cheaper pler and more efficient [15–17]. In the series of works, we
and safer dyes. Recently, the use of metal-free organic dyes have developed various types of D–π–A organic DSSC
has been the subject of an active aspect of DSSC develop- dyes without including metals under the background of
ment because of their desirable advantages, such as modest above-mentioned basic guidelines [18–23].
costs, high molar extinction coefficients, ease of synthesis Quantum chemical calculation has been established as a
and modification, long-term stability, and environmental powerful tool that can be used for the rationalization and
friendliness [4–8]. prediction of the electronic structures and various proper-
Among metal-free organic dyes, coumarin dyes have ties of molecules. In particular, for nanoscale systems like
been proposed as strong candidates because of their excel- DSSCs, density functional theory (DFT) has been utilized
lent photoelectric conversion efficiency [9]. The DSSC widely to calculate the electronic structure and geometry of
based on coumarin dyes was first developed by Hara et al. systems. The time-dependent DFT (TDDFT) has also been
[10]. They designed a π-conjugated system by introducing widely used to evaluate photophysical properties because
a methine unit in C343 [9], which achieved the efficiency of of its reasonable computational cost and reliability. The
5.6 % (NKX-2311). Introduction of bulky substituents that B3LYP functional [24, 25] has been extensively adopted
prevent aggregation increased the efficiency up to 6.7 % among many functionals in the case of the DSSC-related
(NKX-2753) [11]. Extending the π-conjugated system works [26–29]. However, with this functional, the excita-
using a thiophene moiety increased the efficiency to 7.7 % tion energies of charge-transfer transition are sometimes
(NKX-2677) [12]. Later on, inserting a thiophene-based severely underestimated for large π-conjugated systems
bridge between the donor–acceptor parts of the push–pull which causes problem for the organic dyes for DSSC [30–
structure, the stability under sun soaking and the efficiency 32]. In this case, the long-range corrected (LC) function-
were further improved, reaching 8.2 % with NKX-2700. als like LC-BLYP [33] and LC-BOP [34], or CAM-B3LYP
These works indicate that further chemical modifications [35], PBE0(PBE1PBE) [36] are candidates for obtaining
of these coumarin-based dyes may realize higher conver- reliable results. The photoelectronic processes of some
sion efficiency. Thus, theoretical analysis of the electronic DSSC systems have been intensively investigated using
structure, photophysical properties, and electron injection PBE0 [37, 38]. In the series of our works, CAM-B3LYP
mechanism of this series of dyes is important for their fur- has been successfully utilized for evaluating the photoab-
ther development. sorption of the DSSC dyes and anchored dye systems [18,
To improve their photoelectronic conversion efficiency, 19, 39].
dyes must be designed with the following guidelines in On the other hand, a theoretical study based on ab ini-
terms of electronic structure. (1) The dye should absorb tio wave function theory is also important to obtain reli-
light over the wide energy range of UV–Vis and near-IR able information. Various methods including electron cor-
region to achieve excellent light harvesting. (2) The energy relations are now applicable to large nanoscale systems
level of the highest occupied molecular orbital (HOMO) [40]. The symmetry-adapted cluster configuration inter-
must be established below the redox couple of the I− /I− 3 action (SAC-CI) method [41, 42] based on cluster expan-
electrolyte. (3) The energy level of the lowest unoccupied sion theory has been established as a useful approach for
molecular orbital (LUMO) must be situated above, and molecular excited states [43]. Recently, the direct algorithm
close to, the conduction band of TiO2 to enhance efficient of the SAC-CI method has been developed, and this algo-
charge injection. In view of molecular geometry, unfavora- rithm has enabled efficient and accurate calculations [44].
ble aggregation of dyes should be prohibited to achieve The SAC-CI method has been applied to various types of
high conversion efficiency. The close π–π aggregation molecular spectroscopy and photochemistry of molecules
leads to self-quenching and reduction of electron injec- [43]. Related to the present subject of photofunctional
tion to TiO2 as well as to instability due to the formation molecules, it has been successfully applied to the absorp-
of excited triplet states and unstable radicals [13, 14]. tion/emission spectra of organic light-emitting diodes [45,

13
Theor Chem Acc (2016) 135:14 Page 3 of 13 14

46], organic sensitizers for DSSCs [20], and UV-B block- 2 Computational detail
ing molecules [47]. The method has also been applied to
the absorption spectra of extended π-conjugated systems The model system in which a dodecyloxy unit is replaced by
like the phthalocyanine congeners [48] and fused phthalo- a methoxy group was used for calculations, as this replace-
cyanine dimer [49], aiming at the molecular design of dyes ment does not significantly affect the conjugation relevant to
utilizing the near-IR region. These works indicate that the excitation. The geometric structures of the five organic dyes
SAC-CI method is useful to investigate the excited states were fully optimized without any symmetry constraints by
of large π-conjugated systems, including charge-transfer DFT calculations using the B3LYP hybrid functional [24].
excitation [50–53]. Recently, we have performed bench- The 6-31G(d,p) basis set [57] was adopted in the geome-
mark calculations of heteroaromatic compounds regard- try optimization. The natural bond orbital (NBO) analysis
ing photoemissions for comparing TD-DFT and SAC-CI [58] was performed for evaluating charge density or charge
approaches [54, 55]. separation. The frontier orbital analysis and the calculation
In this study, the photophysical properties of recently of energy gaps were carried out for optimized structures at
synthesized coumarin-based donor–π–acceptor (D–π–A)- the same level of theory. The vertical absorption spectra of
type organic dyes [56] were theoretically investigated using all dyes were calculated by direct SAC-CI calculations with
TDDFT and SAC-CI calculations. These dyes consist of the double-zeta basis set of Huzinaga and Dunning (D95)
coumarin as electron donor, thiophene and/or phenylene [59] solving the lowest 10 excited states. They were also
as π-spacer, and cyanoacrylic acid as acceptor, as shown obtained using TDDFT with both B3LYP [24] and CAM-
in Fig. 1. The five dyes are denoted here as UB10, UB11, B3LYP [35] functionals with the 6-31G(d,p) basis set. In the
UB12, UB13, and UB14; their formal names are given in SAC-CI calculation, the perturbation selections of excita-
the Supporting Information (SI). For these molecules, the tion operators were performed with LevelThree accuracy to
effect of different π-spacer moieties was examined with reduce the computational cost. The solvent effect in dichlo-
respect to the geometries and electronic structures—more romethane (CH2Cl2) solution was implicitly accounted
specifically, the planarity, energy levels, and charge separa- for using a conductor-like polarizable continuum model
tion—for molecular design. The calculated spectra of these (C-PCM) [60, 61]. All electronic structure calculations of
five dyes were compared with our experimental spectra these dyes, and their photophysical properties, were carried
[56] and detailed assignments performed. To understand out using the Gaussian 09 suite of programs [62].
the adsorption and electron injection mechanism of the pro- To gain insight into the electron injection mechanism
totype system, the electronic structure and electronic transi- in this coumarin-based D–π–A-type dyes, calculations
tion in a model system, dye–(TiO2)38 complexes, were also of the prototype model systems that represent the dyes
investigated. and TiO2 anatase (101) surface were performed using the
DMol3 program [63, 64] in Materials Studio version 5.5. In
DMol3 calculations, the TiO2 cluster was constructed with
the (TiO2)38 cluster model, which was similar to what was
used in our previous study [65]. The relaxed structures of
isolated dyes, TiO2 surface, and dye–TiO2 complexes were
obtained using the generalized gradient-corrected approxi-
mation (GGA) method of the Perdew–Burke–Ernzerhof
(PBE) [66] functional with a double-numeric-polarized
(DNP) basis set. Note that, although the size of the DNP
basis set is comparable with the 6-31G(d,p) basis set, DNP
is usually more accurate than a Gaussian basis set of the
same size [67]. The core electron was treated with DFT
semicore pseudopotentials (DSPPs) [68].
The adsorption energies (Eads) were computed from the
total energies of the adsorption system by
Eads = E[dye] + E[(TiO2 )38 ] − E[dye + (TiO2 )38 ] (1)
where E[dye] is the energy of the isolated dye, E[(TiO2 )38 ]
is the energy of the (TiO2)38 cluster modeling a TiO2 clean
surface, and E[dye + (TiO2 )38 ] is the energy of the dye–
Fig. 1  Molecular structure of UB10, UB11, UB12, UB13, and UB14 (TiO2)38 model system. According to this expression, a
(real system) positive value of Eads means a stable adsorption of dye on

13
14 Page 4 of 13 Theor Chem Acc (2016) 135:14

the TiO2 surface. Finally, the electron injection mechanism Table 1  Selected bond lengths (r, in Å) and dihedral angles (ϕ, in
between dye and TiO2 surface was investigated using the degree) of all dyes optimized by B3LYP/6-31G(d,p) in CH2Cl2 with
C-PCM
TD-CAM-B3LYP/6-31G(d) level, in vacuum, at ground-
state DMol3 optimized geometry. Dyes Parameter C–T1 T1–T2 T2–T3/T1(T2)–P T–A/P–A

UB10 r 1.459 – – 1.425

ϕ 0.31 – – −0.15
3 Results and discussion UB11 r 1.458 1.439 – 1.422
ϕ 5.87 –3.40 – 0.09
3.1 Geometric structure of dyes
UB12 r 1.459 1.441 1.438 1.421
ϕ 6.25 –7.04 0.545 −0.35
The optimized structures of the model systems UB10–
UB13 r 1.460 – 1.458 1.445
UB14 determined by B3LYP/6-31G(d,p) are shown in
ϕ 11.79 – –12.09 0.47
Fig. 2. The selected dihedral angles and bond distances
UB14 r 1.459 1.443 1.457 1.445
characterizing the molecular structure are summarized
ϕ 6.50 –11.39 11.50 −0.12
in Table 1. The geometric structures of these analogues
show almost coplanar structure, but, not strictly planar, in C, T, P, and A denote coumarin, thiophene, phenylene, and cyanoacr-
C1 point group, and there is no significant difference in ylic acid, respectively
bond distance parameters. In all dyes, the zigzag structure

in π-spacers, namely alternative plus/minus dihedral

angles, was found at the local minimum because of keep-
ing π-conjugation over the molecule. The major difference
was found in the dihedral angle of these dyes, in which
extension of the π-conjugated system by introducing phe-
nylene rings in UB13 and UB14 increases the twisted con-
formation angle in the π-conjugated backbone compared
with those in UB10–UB12; namely, the dihedral angles of
π-spacer in UB14 are −11.4° and 11.5° for T1–T2 and T2–
P, while those in UB12 are −7.0° and 0.5° for T1–T2 and
T2–T3, respectively. Note that the distortion in T2–P affects
the T1–T2 angle in UB14 because of the π-conjugation.
This effect is ascribed to the steric repulsion between two
H atoms on the thiophene and phenylene rings in UB13
and UB14 (Fig. 2), while it does not exist in UB10–UB12.
Considering the polarization and π-conjugation, this struc-
tural feature implies that the electron injection from the
coumarin donor to the cyanoacrylic acceptor of UB13 and
UB14 might not be so effective compared with those of
UB10–UB12.

3.2 Electronic structure of dyes

It is important to examine the distribution of frontier orbit-

als because it provides a measure of the charge separa-
tion accompanied by the excitation, which relates signifi-
cantly to the electron injection. The HOMO and LUMO
of UB12 and UB14 obtained at the optimized structures
using B3LYP/6-31G(d,p) are shown in Fig. 3. Those of
all dyes are given in SI (Figure S1). To create an efficient
charge-separated state by photoabsorption, it is prefer-
able that the HOMO is localized on the donor moiety and
Fig. 2  Molecular structures of UB10-14 optimized by B3LYP/6- the LUMO on the acceptor moiety. In general, all frontier
31G(d,p) (model system) orbitals show π-character and similar orbital shape, where

13
Theor Chem Acc (2016) 135:14 Page 5 of 13 14

Fig. 3  HOMO, LUMO, and

charge density difference (Δρ)
between the excited and ground
states of UB12 and UB14. The
contour thresholds for MO and
densities are 0.02 and 0.004
a.u., respectively. The red and
blue colors indicate a decrease
and increase in charge density

the HOMO spreads over the entire π-conjugated backbone, Table 2  Natural charges of different groups in coumarin-based dyes
but the LUMO is dominantly localized on the π-spacer and calculated by B3LYP/6-31G(d,p) in CH2Cl2
cyanoacrylic acid acceptor, with partial distribution on the Dyes C T1 T2 T3/P A
coumarin ring. This orbital picture clearly indicates that
UB10 0.045 0.131 – – −0.176
the charge-separated state is generated by the transition
UB11 0.009 0.072 0.109 – −0.190
from HOMO to LUMO; in these dyes, the HOMO–LUMO
excitation can be attributed to intramolecular charge-trans- UB12 0.008 0.044 0.052 0.104 −0.192
fer (ICT) transition from donor to acceptor. An important UB13 0.000 0.068 – 0.070 −0.137
characteristic of electron density in the frontier MOs is the UB14 –0.013 0.037 0.050 0.065 −0.139
overlap between the HOMO and LUMO, the well-over- C, T, P, and A denote coumarin, thiophene, phenyl, and cyanoacrylic
lapped HOMO and LUMO suggesting good induction and acid, respectively
electron–withdrawing properties for the donor and accep-
tor, respectively. Thus, we anticipate that this strong-over-
lapping character in our dyes will facilitate ICT between a thiophene–phenyl (TP) bridge; for example, the nega-
the donor and acceptor, subsequently toward conduction tive charge on cyanoacrylic acid in TT dyes (UB12) was
band of TiO2. −0.192 e, which remarkably decreased to −0.139 e when
Furthermore, we perform the Δρ plots revealing elec- replacing the thiophene with a phenylene unit in TP dyes
tron density differences between the ground and excited (UB14). This indicates that the charge polarization in the
states to provide further evidence for ICT. Obviously, the ground state of UB10–UB12 is larger than that of UB13
decreasing electron density is primarily from the electron and UB14. Considering the electron injection from dye to
donor, whereas the increasing electron density is mainly TiO2 surface, this negative charge on an acceptor unit may
on the acceptor group, which is indicative of the ICT when also be a relevant factor. Furthermore, as shown later, in the
transition occurs upon photoirradiation. results of the solvent effect, the change in dipole moments
Next, NBO analysis based on the optimized structure by photoabsorption in UB10–UB12 is also more significant
in the ground state was performed to obtain further insight than that in UB13 and UB14, which means that the ICT in
into the charge distribution in these dyes. Table 2 shows UB10–UB12 is larger than in UB13 and UB14. These facts
the natural charges of coumarin, thiophene, phenylene, and indicate that the electron injection into the substrate TiO2
the cyanoacrylic acid group in dye molecules. In UB10, upon photoabsorption in UB10–UB12 dyes is more effec-
coumarin has a positive or nearly zero NBO charge, while tive than that in UB13 and UB14 dyes.
cyanoacrylic acid has a negative charge, as expected.
Comparing the calculated NBO charges of dyes UB10- 3.3 HOMO–LUMO gap and orbital energy of dyes
UB12, and UB13 and UB14, the NBO charges of the elec-
tron acceptor are –0.176 ~ –0.192 and –0.137 ~ –0.139 e, The HOMO–LUMO energy gaps were evaluated using
respectively. It shows that the electron acceptor of dyes B3LYP/6-31G(d,p) calculations, and the results are sum-
with a thiophene–thiophene (TT) bridge exhibits a strong marized in Table 3. The HOMO–LUMO gap in vacuum
electron-withdrawing ability compared with dyes with was calculated to decrease from 2.94 to 2.42 eV for the

13
14 Page 6 of 13 Theor Chem Acc (2016) 135:14

Table 3  The HOMO–LUMO gaps (in eV) of all dyes calculated by these molecules. However, replacing one thiophene unit
B3LYP/6-31G(d,p) with phenylene in UB13 and UB14 increases the energy
Dyes ΔHOMO–LUMO vacuum ΔHOMO–LUMO in CH2Cl2 Exp.a gap by ~0.19 and 0.15 eV, compared with the cases of the
thiophene ring in UB11 and UB12, respectively. This is
UB10 2.94 2.90 2.51
originated from the twisted structure of UB13 and UB14,
UB11 2.64 2.57 2.48
as discussed above. Because the dihedral angle in thio-
UB12 2.42 2.36 2.39
phene–phenylene affects the other thiophene–thiophene
UB13 2.82 2.76 2.56
dihedral angle, simple estimation or parameterization of
UB14 2.57 2.51 2.47 the band gap or excitation energy by the number/type of
a
Experimental ΔHOMO–LUMO estimated from the onset of the absorp- π-spacers is difficult for the dyes with mixed thiophene–
tion spectra (Eg (eV) = 1240/λonset) in dilute CH2Cl2 solution phenylene π-spacer.
Figure 4 compares the Kohn–Sham (KS) orbital energy
diagram of five dyes with the TiO2 cluster. The energy lev-
series UB10–UB12 and from 2.82 to 2.57 eV for UB13 and els of HOMO and LUMO, and the HOMO–LUMO gap of
UB14 along the effective conjugation length which corre- the (TiO2)38 cluster were calculated to be –7.74, –3.27, and
lates with planarity of structures. We also examined these 4.46 eV, respectively, while the excitation energy for the
HOMO–LUMO gaps in CH2Cl2 solution by C-PCM. The lowest transition was 3.73 eV, by TD-B3LYP/6-31G(d,p)
HOMO–LUMO gap decreases in CH2Cl2 as 2.90–2.36 eV calculation. In these calculations, the structure of (TiO2)38
for UB10–UB12 and 2.76–2.51 eV for UB13 and UB14 was relaxed. These values show reasonable agreement with
compared with in vacuum. The solvent effect amounted the experimental band gaps of typical TiO2 nanoparticles,
to –0.04 ~ –0.07 eV in orbital energy levels. Experimen- namely a few nanometers in energy (3.2–3.3 eV) [70, 71].
tal values were determined from the onset of the absorption The deviation from the experimental values is partly attrib-
spectra in dilute CH2Cl2 solution. Obviously, the gaps cal- uted to the confinement effect [38] which originates in the
culated with including the solvent effect using C-PCM are finite surface modeling. Although the present model of the
in better agreement with the experimental values (Table 3) dye-substrate system lacks in this effect, it does not signifi-
than those calculated in vacuum, which indicates that the cantly affect the qualitative interpretation of the electron
solvent effect is relevant to examine the energy levels of injection mechanism as discussed later.
orbitals. Deviation from the experimental values is due to As mentioned in introduction, the LUMO of dyes must
the insufficiency of the functional. Recently, it was shown be located above and close to the conduction band (CB) of
that the DFT with LC functional satisfies the Koopmans’ a TiO2 semiconductor, while the HOMO should be estab-
theorem and gives reasonable band gaps [69]. lished below the redox couple of the I− /I− 3 electrolyte.
The HOMO–LUMO gap decreases as the number of From Fig. 4, it is clear that the insertion of a π-spacer
π-spacers increases. The extension of the π-spacer linker group gradually decreases the LUMO energy level and
using a thiophene ring enhances the conjugation system of increases the HOMO level, and the resulting HOMO and

Fig. 4  Orbital energy diagram

of five coumarin analogues
and (TiO2)38 by B3LYP/6-
31G(d,p) with the lowest
excitation energy of (TiO2)38
by TD-B3LYP/6-31G(d,p) in
parenthesis

13
Theor Chem Acc (2016) 135:14 Page 7 of 13 14

LUMO energy levels of UB12 are closest to the redox ener- spectra. For all dyes, the solvent effect works to reduce
gies of I− /I−
3 and CB of TiO2, respectively. The extension the excitation energies. The red shift of this peak in polar
of the π-conjugated backbone by insertion of the phenylene solvent is anticipated because the first excited state is the
ring in UB13 and UB14, on the other hand, increases the ICT transition and the dipole moment of the excited state
LUMO energy level compared with UB11 and UB12, is much larger than that of the ground state. This red shift
respectively. This suggests that the phenylene ring is not was evaluated to be larger when using C-PCM with the SS
preferable for improving the conjugation system in view scheme than that when using C-PCM with the LR scheme,
of the energy levels; the electron injection efficiencies of which means that the SS scheme is important for evaluat-
UB13 and UB14 might not be better compared with those ing the solvent effect on the excitation energy in the case of
of UB11 and UB12, respectively. solute molecule with permanent dipole.
For calculating the solvent effect using the SAC-CI
3.4 UV–Vis absorption spectra of dyes method, the PCM-SAC-CI method has been developed
in the SS scheme [76]. However, in the present study, we
Another important factor of dyes in DSSC is the capability adopted the approximation with including the solvent effect
of light harvesting. To gain insight into the optical proper- based on the TDDFT results. The excitation energies and
ties, excitation energies, and oscillator strengths, SAC-CI oscillator strengths in solvent for i → a transition were
and TDDFT calculations were performed for the UV–Vis estimated with the help of the correction based on C-PCM-
absorption spectra of these dyes. Direct SAC-CI calcula- TD-CAM-B3LYP calculations as
tions were performed for solving the 10 lowest singlet tran-
SAC-CI SAC-CI
sitions. In TDDFT, two types of functional, B3LYP and �Ei→a,(solv) = �Ei→a,(vac) + ��Ei→a,(solv) (2)
CAM-B3LYP, were employed to see the functional depend-
ence. The 10 lowest singlet excited states were calculated and
in both vacuum and solvent phase.
SAC-CI
First, we will discuss the solvent effect that is not neg- SAC-CI SAC-CI
�Ei→a,(vac) CAM-B3LYP
ligible to these absorption spectra. To examine the solvent
fi→a,(solv) = fi→a,(vac) + CAM-B3LYP
�fi→a,(solv) (3)
�Ei→a,(vac)
effect on excitation energy, the TDDFT calculations with
C-PCM were performed in both state-specific (SS) [72, 73] SAC-CI
where �Ei→a,(solv) is the SAC-CI excitation energy in
and linear-response (LR) [74, 75] schemes. The results of vacuum, ��Ei→a,(solv) is the difference in excitation ener-
the solvent effect in CH2Cl2 for the first excited state which gies between solution and vacuum obtained by C-PCM-
is relevant for light harvesting are summarized in Table 4. SAC-CI
TD-CAM-B3LYP, fi→a,(solv) is the SAC-CI oscillator
CH2Cl2 was used in the experimental UV–Vis absorption CAM-B3LYP is the TD-CAM-B3LYP
strength in vacuum, �Ei→a,(vac)

Table 4  Excitation energies Dyes Phase B3LYP CAM-B3LYP

(ΔE, eV), absorption
wavelengths (λabs, nm), and ΔE λabs f ΔE λabs f
oscillator strength (f) for all
dyes in vacuum and in CH2Cl2 UB10 Vacuum 2.78 446 1.181 3.07 404 1.290
calculated by TDDFT with CH2Cl2 (LR) 2.59 478 1.296 2.91 426 1.437
B3LYP and CAM-B3LYP
CH2Cl2 (SS) 2.47 501 1.444 2.79 445 1.560
functionals
UB11 Vacuum 2.46 505 1.439 2.80 443 1.621
CH2Cl2 (LR) 2.28 543 1.540 2.65 467 1.746
CH2Cl2 (SS) 2.17 571 1.690 2.54 488 1.848
UB12 Vacuum 2.23 555 1.639 2.65 468 2.048
CH2Cl2 (LR) 2.08 597 1.755 2.52 492 2.158
CH2Cl2 (SS) 1.98 627 1.926 2.42 513 2.245
UB13 Vacuum 2.60 477 1.367 3.06 405 1.690
CH2Cl2 (LR) 2.44 508 1.476 2.94 422 1.849
CH2Cl2 (SS) 2.34 529 1.639 2.84 436 1.976
UB14 Vacuum 2.46 505 1.439 2.85 435 2.095
CH2Cl2 (LR) 2.20 563 1.701 2.75 451 2.253
CH2Cl2 (SS) 2.12 586 1.880 2.66 465 2.369

Solvent effect was included by C-PCM with LR and SS scheme

13
14 Page 8 of 13 Theor Chem Acc (2016) 135:14

Fig. 5  UV–Vis absorption spectra of UB10-UB12 compared to ab initio direct SAC-CI spectra for vacuum (black) and in CH2Cl2 (red) with
including solvent effect by TD-CAM-B3LYP (C-PCM-SS)

Fig. 6  UV–Vis absorption

spectra of UB13 and UB14
compared to ab initio direct
SAC-CI spectra for vacuum
(black) and in CH2Cl2 (red)
with including solvent effect by
TD-CAM-B3LYP (C-PCM-SS)

CAM-B3LYP is the dif-

excitation energy in vacuum, and �fi→a,(solv) experimental data, while the TD-CAM-B3LYP spectra
ference in oscillator strengths between vacuum and solu- were in better agreement with the experimental spectra.
tion calculated by TD-CAM-B3LYP. This result shows that the LC functional of DFT is also
The absorption spectra of UB10–UB12 and UB13– important in calculating the excited states in the present
UB14 calculated by ab initio direct SAC-CI/D95, in system with large π-conjugation.
vacuum (black solid line) and in CH2Cl2 (red solid line), Referring to the experimental spectra, UB12 and UB14
are compared with the experimental spectra observed have broad absorption band compared to other dyes, which
in CH2Cl2 solution in Figs. 5 and 6, respectively. The is preferable to light harvesting in wide energy region. This
absorption spectra by TD-B3LYP/6-31G(d,p) and TD- is attributed to the flexible structure of UB12 and UB14.
CAM-B3LYP/6-31G(d,p) for all the other dyes in the pre- The flexibility in the ground state with thermal distribu-
sent study are also given with the experimental UV–Vis tion and the Franck–Condon factor reflecting the geometry
spectra in Supporting Information (Figures S2 and S3). change between the ground and excited states are important
Figure 5 shows that the lowest strong peak with the high for the bandwidth. UB12 and UB14 have complex spectral
oscillator strength is responsible for the light harvesting. shape with shoulder in higher energy side which may be
The absorption spectra of UB10-UB14 were reasonably due to the non-planarity with thermal distribution.
reproduced by the SAC-CI calculations including the sol- Table 5 summarizes the SAC-CI results for the S1 state
vent effect, although the excitation energies in UB13 and which is responsible for the light harvesting. The excita-
UB14 were overestimated. As shown in Figures S2 and S3, tion energies and oscillator strengths are given for in vac-
the TD-B3LYP calculation significantly underestimated uum and in CH2Cl2, where the solvent effect was evalu-
the excitation energies of absorption peaks compared with ated in the above-mentioned approximation. The SAC-CI

13
Theor Chem Acc (2016) 135:14 Page 9 of 13 14

Table 5  Excitation energies Dyes SAC-CI/D95 Exp.

(ΔE, eV), absorption
wavelengths (λabs, nm), and Vacuum CH2Cl2
oscillator strength (f) for all
dyes in vacuum and in CH2Cl2 ΔE λabs f ΔE λabs f ΔE (λabs)
calculated by the SAC-CI
method with D95 basis set UB10 3.29 377 1.290 3.01 412 0.703 2.98 (416)
UB11 3.26 380 0.854 3.00 414 0.590 2.87 (432)
UB12 2.78 446 1.446 2.55 487 1.239 2.83 (439)
UB13 3.59 345 1.517 3.38 367 1.182 2.93 (423)
UB14 3.43 361 1.936 3.25 382 1.606 2.92 (425)

Solvent effect was evaluated by TD-CAM-B3LYP

results for 10 lowest singlet excited states in vacuum and in even-numbered thiophene spacer have vertical adsorbed
CH2Cl2 are summarized in Tables S1 and S2, respectively. structure. The Ti–O bond distances and adsorption ener-
The S1 state is characterized as HOMO–LUMO transi- gies calculated by PBE/DNP are given in Table 6. The
tion, and the transition energies decrease as the effective bond distances of Ti–O and Ti′–O′ of dyes were 2.15–2.13
π-conjugation length, which is in accordance with the trend and 2.03–2.04 Å for UB10–UB12 and 2.14–2.15 and
in the HOMO–LUMO gap. The solvent effect reduces 2.04–2.03 Å for UB13 and UB14. The calculated adsorp-
the excitation energies of these transitions by about 0.22– tion energies of dyes on the (TiO2)38 cluster were 13.6–
0.28 eV for UB10–UB12 and by 0.18–0.21 eV for UB13 14.8 kcal/mol for these five prototype systems. These val-
and UB14, compared with those in vacuum. The excita- ues indicate that these dyes strongly adsorb on the TiO2
tion energies in solution are in better agreement with the surface.
experimental data, and the solvent effect becomes smaller Finally, the electron injection mechanism from dyes to
as the effective π-conjugation length. More importantly, TiO2 substrate by photoabsorption was also investigated
the solvent effect in the UB10–UB12 series is significant for the dye–(TiO2)38 complexes. The 10 lowest singlet
than that in UB13 and UB14. This indicates that the charge transitions were solved for these systems by TD-CAM-
reorganization in the excited state of UB10–UB12 is larger B3LYP/6-31(d). The calculated excitation energies for the
than in UB13 and UB14, which is relevant for the electron peak having the highest oscillator strength are summarized
injection to substrate together with the charge separation in in Table 7 together with the transition character for all dyes.
the ground state, as described in the previous section. The results of excitation energies and oscillator strengths
for all the 10 low-lying states are presented in Table S3.
3.5 Dye–TiO2 adsorption and electron injection The results show that the strongest transition in all dyes is
mechanism characterized as the linear combination of some configura-
tions. The KS orbitals that are most relevant to these transi-
The adsorption of a dye with carboxylic acid is possible by tions are shown in Fig. 8 for UB12 and UB14. Other KS
either physisorption or chemisorption. In the case of chem- orbitals that also significantly contribute to this transition
isorption, there are some possibilities in terms of bond- are displayed in Figures S4–S9 for all the dye–(TiO2)38
ing to the surface such as via a monodentate ester, biden- systems. These orbitals together with TDDFT functions
tate chelating, or bidentate bridging [77, 78]. It has been show that the transition with large oscillator strength is
reported in many experimental studies, such as attenuated characterized as the transition from the orbitals localized in
total reflectance (ATR)-FTIR spectroscopy, that the termi- the donor–π-spacer (D–π) unit to the orbitals delocalized
nal carboxylic acid usually adsorbs on the TiO2 surface via over the acceptor (A) unit of dyes and the (TiO2)38 cluster.
bidentate bridging in the dye-substrate system [4, 78, 79]. The former orbitals are similar to HOMOs of dyes, and the
Therefore, chemisorption via bidentate bridging was inves- latter orbitals correspond to the interacting orbitals between
tigated in the present study. the LUMOs of dyes and the conduction band of TiO2. This
To obtain the adsorption energy, the relaxed structures means that the electron excitation of this system directly
of both free (TiO2)38 cluster and dye–(TiO2)38 complexes induces the electron injection from dye into TiO2 surface.
were calculated. The relaxed structures of the chemisorbed This implies the high photoelectric conversion efficiency of
dye–(TiO2)38 complexes are shown in Fig. 7. Because of this system. Note that the latter orbitals are embedded in
the molecular structure of dyes, dyes with odd number the conduction band of TiO2, and then the injected electron
of thiophene spacer (UB10, UB13) have relatively tilted is transferred to the conduction bands of TiO2. We should
adsorbed structure relative to the surface, while dyes with be careful for the artificial confinement effect [38] caused

13
14 Page 10 of 13 Theor Chem Acc (2016) 135:14

Fig. 7  Relaxed structures of dye–TiO2 complexes optimized by PBE/DNP calculation

Table 6  Ti–O bond distances and adsorption energies of dye– 4 Summary

(TiO2)38 cluster optimized by PBE/DNP calculation
Dye Ti–O (Å) Ti′–O′ (Å) Eads (kcal/mol) Theoretical investigations were performed to understand
the electronic structure and photophysical properties of
UB10 2.15 2.03 14.6
five coumarin-based D–π–A-type organic dyes for DSSC,
UB11 2.14 2.04 14.4
namely UB10, UB11, UB12, UB13, and UB14, using the
UB12 2.13 2.04 14.8
TDDFT and SAC-CI methods. The photophysical prop-
UB13 2.14 2.04 13.6
erties of these molecules are shown to be related to the
UB14 2.15 2.03 14.5
planarity of their molecular structure. The π-conjugated
backbones of UB10–UB12 are more planar than those
of UB13 and UB14. Increasing π-spacer moieties gradu-
by the substrate modeling: The finite cluster model esti- ally enhances the charge-separated state of these dyes,
mates the LUMO energy level higher and band gap larger. and therefore, injection of electrons from coumarin
Using the infinite substrate model, the LUMO of substrate donor to cyanoacrylic acceptor via ICT transition would
would be lower in energy than the present result, and there- be improved. Introducing the thiophene unit practically
fore, the interaction between the LUMO of dyes and TiO2 improves both the electron injection and regeneration pro-
conduction band exists even when we adopt the infinite cesses, which facilitates the conversion efficiency, while
substrate. As this transition is related to the ICT transition the phenylene unit significantly increases the LUMO level.
of dyes, the intramolecular charge separation discussed in Theoretical calculations, with inclusion of the solvent
the preceding sections is also important. effect, in SS and LR scheme, reproduced the experimen-
The better-overlapped HOMO and LUMO orbitals of tal UV–Vis absorption spectra of these dyes satisfactorily
UB12 with TT linker than UB14 with TP linker have a sig- and provided detailed assignments of the spectra. The sol-
nificant effect on facilitating electron injection from dyes vent effect on excitation energies calculated by C-PCM in
toward conduction band of TiO2, see Fig. 8. Therefore, the SS scheme is larger than that in the LR scheme. The
electron injection process could be more efficient in UB12 π-spacers, thiophene and/or phenylene units, affect the
than in UB14 according to Marcus theory. planarity of molecular structures. This is relevant to the

Table 7  Excitation energies Dye–TiO2 ΔE (λabs) f Transition character

(ΔE/eV, λabs/nm), oscillator
strength (f), and transition UB10 2.84 (434) 1.682 0.53 (H → L + 8) − 0.25 (H → L + 6) + 0.16 (H → L + 7)
character of the S1 state of
UB11 2.59 (478) 2.174 0.37 (H → L + 5) − 0.34 (H → L + 6) + 0.29 (H → L + 7)
dye–TiO2 complexes calculated
by TD-CAM-B3LYP/6-31G(d) UB12 2.42 (511) 2.556 0.33 (H → L + 6) − 0.32 (H → L + 5) + 0.25 (H → L + 7)
level of theory UB13 2.82 (439) 2.175 0.33 (H → L + 13) + 0.22 (H → L + 18) − 0.22 (H → L + 10)
UB14 2.61 (475) 2.633 0.33 (H → L + 10) + 0.22 (H → L + 13) − 0.22 (H → L + 17)

13
Theor Chem Acc (2016) 135:14 Page 11 of 13 14

Education Commission, and the Ministry of Education, Thailand.

National Nanotechnology Center (NANOTEC), Research Center for
Computational Science, and Institute for Molecular Science (IMS)
are acknowledged for financial support and research facilities. M.E.
acknowledges the financial support from a Grant-in-Aid for Scientific
Research from the Japan Society for the Promotion of Science (JSPS),
Nanotechnology Platform Program (Molecule and Material Synthe-
sis) of the Ministry of Education, Culture, Sports, Science, and Tech-
nology (MEXT) of Japan.

References

1. O’Regan B, Grätzel M (1991) A low-cost, high-efficiency

solar cell based on dye-sensitized colloidal TiO2 films. Nature
353(6346):737–740
2. Hagfeldt A, Boschloo G, Sun L, Kloo L, Pettersson H (2010)
Dye-sensitized solar cells. Chem Rev 110(11):6595–6663
3. Chung I, Lee B, He J, Chang RPH, Kanatzidis MG (2012) All-
solid-state dye-sensitized solar cells with high efficiency. Nature
485(7399):486–489
4. Hara K, Sato T, Katoh R, Furube A, Ohga Y, Shinpo A, Suga S,
Sayama K, Sugihara H, Arakawa H (2002) Molecular design
of coumarin dyes for efficient eye-sensitized solar cells. J Phys
Chem B 107(2):597–606
5. Hara K, Wang Z-S, Sato T, Furube A, Katoh R, Sugihara H, Dan-
oh Y, Kasada C, Shinpo A, Suga S (2005) Oligothiophene-con-
taining coumarin dyes for efficient eye-sensitized solar cells. J
Phys Chem B 109(32):15476–15482
6. Li G, Jiang KJ, Li YF, Li SL, Yang LM (2008) Efficient struc-
tural modification of triphenylamine-based organic dyes for dye-
sensitized solar cells. J Phys Chem C 112:11591–11599
7. Wu W, Yang J, Hua J, Tang J, Zhang L, Long Y, Tian H (2010)
Efficient and stable dye-sensitized solar cells based on phe-
nothiazine sensitizers with thiophene units. J Mater Chem
Fig. 8  MOs relevant during photoexcitation for TiO2-UB12 and 20:1772–1779
TiO2-UB14 complexes in vacuum calculated by TD-CAM-B3LYP/6- 8. Mishra A, Fischer MKR, Bauerle P (2009) Metal-free
31G(d) organic dyes for dye-sensitized solar cells: from structure:
property relationships to design rules. Angew Chem Int Ed
48:2474–2499
photophysical properties and charge polarization. Charge 9. Rehm JM, McLendon GL, Nagasawa Y, Yoshihara K, Moser J,
Grätzel M (1996) Femtosecond electron-transfer dynamics at a
separation was examined by NBO analysis and also found sensitizing dye-semiconductor (TiO2) interface. J Phys Chem
to be an important factor for the electron injection in DSSC. 100(23):9577–9578
The adsorption of these dyes on a TiO2 anatase (101) 10. Hara K, Sayama K, Ohga Y, Shinpo A, Suga S, Arakawa H
surface and the electron injection mechanism were also (2001) A coumarin-derivative dye sensitized nanocrystalline
TiO2 solar cell having a high solar-energy conversion efficiency
investigated with the dye–(TiO2)38 cluster model, employ- up to 5.6%. Chem Commun 6:569–570
ing PBE and TD-CAM-B3LYP calculations, respectively. 11. Wang ZS, Hara K, Dan-oh Y, Kasada C, Shinpo A, Suga S,
The adsorption energies of these dyes were estimated to be Arakawa H, Sugihara H (2005) Photophysical and (photo)
~14 kcal/mol, implying strong adsorption on the TiO2 sur- electrochemical properties of a coumarin dye. J Phys Chem B
109(9):3907–3914
face with a carboxylate group. It was found that the elec- 12. Hara K, Kurashige M, Dan-oh Y, Kasada C, Shinpo A, Suga S,
tron injection occurs by direct charge-transfer transition in Sayama K, Arakawa H (2003) Design of new coumarin dyes
a dye–(TiO2)38 interacting system, which is mainly charac- having thiophene moieties for highly efficient organic-dye-sensi-
terized as the transition from the orbitals localized on the tized solar cells. New J Chem 27(5):783–785
13. Fink RF, Seibt J, Engel V, Renz M, Kaupp M, Lochbrunner S,
donor and π-spacer part of the dyes to the orbitals distrib- Zhao HM, Pfister J, Würthner F, Engels B (2008) Exciton trap-
uted over the acceptor part of the dyes and (TiO2)38. ping in π-conjugated materials: a quantum-chemistry-based pro-
tocol applied to perylene bisimide dye aggregates. J Am Chem
Acknowledgments The authors acknowledge Department of Chem- Soc 130:12858–12859
istry, Faculty of Science, Ubon Ratchathani University, and Chiang 14. Tatay S, Haque SA, O’Regan B, Durrant JR, Verhees WJH,
Mai University. This research was supported by Thailand Research Kroon JM, Vidal-Ferran A, Gavina P, Palomares E (2007)
Fund (Grant Number RSA5780048) and the Center of Excellence Kinetic competition in liquid electrolyte and solid-state cyanine
for Innovation in Chemistry (PERCH-CIC), the office of the Higher dye sensitized solar cells. J Mater Chem 17:3037–3044

13
14 Page 12 of 13 Theor Chem Acc (2016) 135:14

15. Angelis FD, Tilocca A, Sellon A (2004) Time-dependent DFT 32. Zhao Y, Truhlar DG (2008) Density functionals with broad appli-
study of [Fe(CN)6]4− sensitization of TiO2 nanoparticles. J Am cability in chemistry. Acc Chem Res 41(2):157–167
Chem Soc 126:15024–15025 33. Iikuwa H, Tsuneda T, Yanai T, Hirao K (2001) A long-range cor-
16. Jono R, Fujisawa J, Segawa H, Yamashita K (2011) Theoretical rection scheme for generalized-gradient-approximation exchange
study of the surface complex between TiO2 and TCNQ show- functionals. J Chem Phys 115:3540–3544
ing interfacial charge-transfer transitions. J Phys Chem Lett 34. Tawada Y, Tsuneda T, Yanagisawa S, Yanai T, Hirao K (2004) A
2:1167–1170 long-range-corrected time-dependent density functional theory. J
17. Agrawal S, Dev P, English NJ, Thampi KR, MacElroy JMD Chem Phys 120:8425–8433
(2012) A TD-DFT study of the effects of structural variations on 35. Yanai T, Tew DP, Handy NC (2004) A new hybrid exchange-
the photochemistry of polyene dyes. Chem Sci 3:416–424 correlation functional using the Coulomb-attenuating method
18. Kungwan N, Khongpracha P, Namuangruk S, Meeprasert J, (CAM-B3LYP). Chem Phys Lett 393(1–3):51–57
Chitpakdee C, Jungsuttiwong S, Promarak V (2014) Theoreti- 36. Adamo C, Barone V (1999) Toward reliable density functional
cal study of linker-type effect in carbazole–carbazole-based dyes methods without adjustable parameters: the PBE0 model. J
on performances of dye-sensitized solar cells. Theor Chem Acc Chem Phys 110(13):6158–6170
133(8):1–14 37. Labat F, Ciofini I, Hratchian HP, Frisch M, Raghavachari K,
19. Namuangruk S, Meeprasert J, Jungsuttiwong S, Promarak Adamo C (2009) First principles modeling of eosin-loaded ZnO
V, Kungwan N (2014) Organic sensitizers with modified films: a step toward the understanding of dye-sensitized solar
di(thiophen-2-yl)phenylamine donor units for dye-sensitized cell performances. J Am Chem Soc 131:14290–14298
solar cells: a computational study. Theor Chem Acc 133(9):1–15 38. Le Bahers T, Labat F, Pauporté T, Lainé PP, Ciofini I (2011) The-
20. Namuangruk S, Fukuda R, Ehara M, Meeprasert J, Khanasa T, oretical procedure for optimizing dye-sensitized solar cells: from
Morada S, Kaewin T, Jungsuttiwong S, Sudyoadsuk T, Promarak electronic structure to photovoltaic efficiency. J Am Chem Soc
V (2012) D–D–π–A type organic dyes for dye-sensitized solar 133:8005–8013
cells with a potential for direct electron injection and a high 39. Namuangruk S, Sirithip K, Rattanatwan R, Keawin T, Kung-
extinction coefficient: synthesis, characterization, and theoretical wan N, Sudyodsuk T, Promarak V, Surakhot Y, Jungsuttiwong S
investigation. J Phys Chem C 116(49):25653–25663 (2014) Theoretical investigation of the charge-transfer properties
21. Jungsuttiwong S, Yakhanthip T, Surakhot Y, Khunchalee J, in different meso-linked zinc porphyrins for highly efficient dye-
Sudyoadsuk T, Promarak V, Kungwan N, Namuangruk S (2012) sensitized solar cells. Dalton Trans 43(24):9166–9176
The effect of conjugated spacer on novel carbazole derivatives 40. Akasaka T, Nagase S (2002) Endofullerenes: A new family of
for dye-sensitized solar cells: density functional theory/time- carbon clusters. Kluwer Academic Publishers, Kluwer Dordrecht
dependent density functional theory study. J Comput Chem 41. Nakatsuji H (1978) Cluster expansion of the wavefunction.
33(17):1517–1523 Excited states. Chem Phys Lett 59:362–364
22. Sudyoadsuk T, Pansay S, Morada S, Rattanawan R, Namuangruk 42. Nakatsuji H (1979) Cluster expansion of the wavefunction. Cal-
S, Kaewin T, Jungsuttiwong S, Promarak V (2013) Synthesis and culation of electron correlations in ground and excited states by
characterization of D–D–π–A type organic dyes bearing carba- SAC and SAC CI theories. Chem Phys Lett 67(329):334–342
zole–carbazole as a donor moiety (D–D) for efficient dye-sensi- 43. Ehara M, Hasegawa J, Nakatsuji H (2005) Theory and applica-
tized solar cells. Eur J Org Chem 23:5051–5063 tions of computational chemistry, SAC-CI method applied to
23. Sirithip K, Prachumrak N, Rattanawan R, Keawin T, Sudyoad- molecular spectroscopy: the first 40 years. Elsevier, Oxford
suk T, Namuangruk S, Jungsuttiwong S, Promarak V (2015) 44. Fukuda R, Nakatsuji H (2008) Formulation and implementation
Zinc–porphyrin dyes with different meso-aryl substituents for of direct algorithm for the symmetry adapted cluster and sym-
dye-sensitized solar cells: experimental and theoretical studies. metry adapted cluster-configuration interaction method. J Chem
Chem Asian J 10(4):882–893 Phys 128:094105
24. Becke AD (1993) Density-functional thermochemistry. III. The 45. Saha B, Ehara M, Nakatsuji H (2007) Investigation of the elec-
role of exact exchange. J Chem Phys 98(7):5648–5652 tronic spectra and excited-state geometries of poly(para-phe-
25. Lee C, Yang W, Parr RG (1988) Development of the Colle–Sal- nylene vinylene) (PPV) and poly(para-phenylene) (PP) by the
vetti correlation-energy formula into a functional of the electron symmetry-adapted cluster configuration interaction (SAC-CI)
density. Phys Rev B 37:785–789 method. J Phys Chem A 111:5473–5481
26. Liu Z (2008) Theoretical studies of natural pigments rel- 46. Poolmee P, Ehara M, Nakatsuji H (2011) Photophysical proper-
evant to dye-sensitized solar cells. J Mol Struct THEOCHEM ties and vibrational structure of ladder-type penta p-phenylene
862(1–3):44–48 and carbazole derivatives based on SAC-CI calculations. Theor
27. Zhang X, Zhang JJ, Xia YY (2008) Molecular design of cou- Chem Acc 130:161–173
marin dyes with high efficiency in dye-sensitized solar cells. J 47. Promkatkaew M, Suramitr S, Monhaphda TK, Namuangrukd S,
Photochem Photobiol A 194(2–3):167–172 Ehara M, Hannongbua S (2009) Absorption and emission spectra
28. Cai-Rong Z, Zi-Jiang L, Yu-Hong C, Hong-Shan C, You-Zhi W, of ultraviolet B blocking methoxy substituted cinnamates inves-
Li-Hua Y (2009) DFT and TDDFT study on organic dye sensitiz- tigated using the symmetry-adapted cluster configuration inter-
ers D5, DST and DSS for solar cells. J Mol Struct THEOCHEM action method. J Chem Phys 131:2243060(10)
899(1–3):86–93 48. Fukuda R, Ehara M, Nakatsuji H (2010) Excited states and
29. Ham HW, Kim YS (2010) Theoretical study of indoline dyes for electronic spectra of extended tetraazaporphyrins. J Chem Phys
dye-sensitized solar cells. Thin Solid Films 518(22):6558–6563 133(14):144316(16)
30. Cai Z-L, Sendt K, Reimers JR (2002) Failure of density-func- 49. Fukuda R, Ehara M (2012) Excited states and electronic spec-
tional theory and time-dependent density-functional theory for tra of annulated dinuclear free-base phthalocyanines: a theo-
large extended pi systems. J Chem Phys 117(12):5543–5549 retical study on near-infrared-absorbing dyes. J Chem Phys
31. Cai Z-L, Crossley MJ, Reimers JR, Kobayashi R, Amos RD 136:114304(15)
(2006) Density functional theory for charge transfer: the nature 50. Ehara M, Fukuda R, Adamo C, Ciofini I (2013) Chemically intu-
of the N-bands of porphyrins and chlorophylls revealed through itive indices for charge-transfer excitation based on SAC-CI and
CAM-B3LYP, CASPT2, and SAC-CI calculations. J Phys Chem TD-DFT calculations. J Comput Chem 34(29):2498–2501
B 110(31):15624–15632

13
Theor Chem Acc (2016) 135:14 Page 13 of 13 14

51. Fukuda R, Ehara M (2014) Efficiency of perturbation-selec- 64. Delley B (2000) From molecules to solids with the DMol3
tion and its orbital dependence in the SAC-CI calculations for approach. J Chem Phys 113(18):7756–7764
valence excitations of medium-size molecules. J Comput Chem 65. Yakhanthip T, Jungsuttiwong S, Namuangruk S, Kungwan N,
35(30):2163–2176 Promarak V, Sudyoadsuk T, Kochpradist P (2011) Theoretical
52. Fukuda R, Ehara M (2014) An efficient computational scheme investigation of novel carbazole-fluorene based D-π-A conju-
for electronic excitation spectra of molecules in solution using gated organic dyes as dye-sensitizer in dye-sensitized solar cells
the symmetry-adapted cluster-configuration interaction method: (DSCs). J Comput Chem 32(8):1568–1576
the accuracy of excitation energies and intuitive charge-transfer 66. Perdew JP, Burke K, Ernzerhof M (1996) Generalized gradient
indices. J Chem Phys 141(15):154104 approximation made simple. Phys Rev Lett 77(18):3865–3868
53. Adamo C, Le Bahers T, Savarese M, Wilbraham L, García G, 67. Kusama H, Orita H, Sugihara H (2008) TiO2 band shift by nitro-
Fukuda R, Ehara M, Rega N, Ciofini I (2015) Exploring excited gen-containing heterocycles in dye-sensitized solar cells: a peri-
states using time dependent density functional theory and den- odic density functional theory study. Langmuir 24(8):4411–4419
sity-based indexes. Coord Chem Rev 304–305:166–178 68. Delley B (2002) Hardness conserving semilocal pseudopoten-
54. Bousquet D, Fukuda R, Maitarad P, Jacquemin D, Ciofini I, tials. Phys Rev B 66(15):155125(8)
Adamo C, Ehara M (2013) Excited-state geometries of heteroar- 69. Tsuneda T, Song JW, Suzuki S, Hirao K (2010) On Koop-
omatic compounds: a comparative TD-DFT and SAC-CI study. J mans’ theorem in density functional theory. J Chem Phys
Chem Theor Comput 9(5):2368–2379 133:174101(9)
55. Bousquet D, Fukuda R, Jacquemin D, Ciofini I, Adamo C, Ehara 70. Weng YX, Wang YQ, Asbury JB, Ghosh HN, Lian T (1999)
M (2014) Benchmark study on the triplet excited-state geome- Back electron transfer from TiO2 nanoparticles to FeIII(CN)63-:
tries and phosphorescence energies of heterocyclic compounds: origin of non-single-exponential and particle size independent
comparison between TD-PBE0 and SAC-CI. J Chem Theor dynamics. J Phys Chem B 104(1):93–104
Comput 10(9):3969–3979 71. Yang M, Thompson DW, Meyer GJ (2000) Dual pathways
56. Potjanasopa S (2011) Synthesis and characterization of organic for TiO2 sensitization by Na2[Fe(bpy)(CN)4]. Inorg Chem
materials for optoelectronic devices. Ph.D. thesis, Faculty of Sci- 39(17):3738–3739
ence, Ubon Ratchathani University 72. Cossi M, Barone V (2000) Solvent effect on vertical electronic
57. Hehre WJ, Ditchfield R, Pople JA (1972) Self-consistent molec- transitions by the polarizable continuum model. J Chem Phys
ular orbital methods. XII. further extensions of Gaussian-type 112(5):2427–2435
basis sets for use in molecular orbital studies of organic mole- 73. Improta R (2006) A state-specific polarizable continuum model
cules. J Chem Phys 56:2257–2261 time dependent density functional theory method for excited
58. Glendening ED, Reed AE, Carpenter JE, Weinhold F. NBO ver- state calculations in solution. J Chem Phys 125(5):054103(9)
sion 3.1 74. Cammi R, Mennucci B, Tomasi J (2000) Fast evaluation of
59. Dunning TH Jr (1970) Gaussian basis functions for sse in molec- geometries and properties of excited molecules in solution: a
ular calculations. I. Contraction of (9s5p) atomic basis sets for tamm-dancoff model with application to 4-dimethylaminobenzo-
the first-row atoms. J Chem Phys 53:2823–2834 nitrile. J Phys Chem A 104(23):5631–5637
60. Cossi M, Rega N, Scalmani G, Barone V (2003) Energies, struc- 75. Cossi M, Barone V (2001) Time-dependent density func-
tures, and electronic properties of molecules in solution with the tional theory for molecules in liquid solutions. J Chem Phys
C-PCM solvation model. J Comput Chem 24(6):669–681 115(10):4708–4717
61. Takano Y, Houk KN (2004) Benchmarking the conductor-like 76. Cammi R, Fukuda R, Ehara M, Nakatsuji H (2010) Symmetry-
polarizable continuum model (CPCM) for aqueous solvation free adapted cluster and symmetry-adapted cluster-configuration
energies of neutral and ionic organic molecules. J Chem Theor interaction method in the polarizable continuum model: Theory
Comput 1(1):70–77 of the solvent effect on the electronic excitation of molecules in
62. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, solution. J Chem Phys 133(2):024104(24)
Cheeseman JR, Scalmani G, Barone V, Mennucci B, Petersson 77. Vittadini A, Selloni A, Rotzinger FP, Grätzel M (2000) Formic
GA, Nakatsuji H, Caricato M, Li X, Hratchian HP, Izmaylov AF, acid adsorption on dry and hydrated TiO2 anatase (101) surfaces
Bloino J, Zheng G, Sonnenberg JL, Hada M, Ehara M, Toyota K, by DFT calculations. J Phys Chem B 104(6):1300–1306
Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao 78. Nazeeruddin MK, Humphry-Baker R, Liska P, Grätzel M (2003)
O, Nakai H, Vreven T, Montgomery Jr JA, Peralta JE, Ogliaro Investigation of sensitizer adsorption and the influence of pro-
F, Bearpark M, Heyd JJ, Brothers E, Kudin KN, Staroverov tons on current and voltage of a dye-sensitized nanocrystalline
VN, Kobayashi R, Normand J, Raghavachari K, Rendell A, TiO2 solar cell. J Phys Chem B 107(34):8981–8987
Burant JC, Iyengar SS, Tomasi J, Cossi M, Rega N, Millam JM, 79. Srinivas K, Yesudas K, Bhanuprakash K, Rao VJ, Giribabu L
Klene M, Knox JE, Cross JB, Bakken V, Adamo C, Jaramillo (2009) A combined experimental and computational investi-
J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, gation of anthracene based sensitizers for DSSC: comparison
Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski of cyanoacrylic and malonic acid electron withdrawing groups
VG, Voth GA, Salvador P, Dannenberg JJ, Dapprich S, Daniels binding onto the TiO2 anatase (101) surface. J Phys Chem C
AD, Farkas Ö, Foresman JB, Ortiz JV, Cioslowski J, Fox DJ 113(46):20117–20126
(2009) Gaussian 09, Revision B.01. Gaussian, Inc., Wallingford,
CT
63. Delley B (1990) An all-electron numerical method for solving
the local density functional for polyatomic molecules. J Chem
Phys 92(1):508–517

13