N-Aryl stilbazolium dyes as sensitizers for solar cells

Article abstract of DOI:10.1016/j.dyepig.2011.06.017

Eight new N-arylstilbazolium chromophores with electron donating -NR ₂ (R = Me or Ph) substituents have been synthesized via Knoevenagel condensations and isolated as their PF₆^- salts. These compounds have been characteriz

Full text of DOI:10.1016/j.dyepig.2011.06.017

Dyes and Pigments 92 (2011) 766e777
Contents lists available at ScienceDirect
Dyes and Pigments
journal homepage: www.elsevier.com/locate/dyepig
N-Aryl stilbazolium dyes as sensitizers for solar cells
,
a
a
a
Octavia A. Blackburn a, Benjamin J. Coe a , Valentine Hahn , Madeleine Helliwell , James Raftery ,
*
Yien T. Ta a, Laurence M. Peter b, Hongxia Wang b, Juan A. Anta c, Elena Guillén c
a School of Chemistry, University of Manchester, Oxford Road, Manchester M13 9PL, UK
b Department of Chemistry, University of Bath, Bath BA2 5NB, UK
c Area de Quimica Fisica, Departamento de Sistemas Físicos, Químicos y Naturales, Universidad Pablo de Olavide, 41013 Sevilla, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Eight new N-arylstilbazolium chromophores with electron donating eNR2 (R ¼ Me or Ph) substituents
have been synthesized via Knoevenagel condensations and isolated as their PF6ꢀ salts. These compounds
have been characterized by using various techniques including 1H NMR and IR spectroscopies and
electrospray mass spectrometry. UVevis absorption spectra recorded in acetonitrile are dominated by
Received 18 May 2011
Accepted 12 June 2011
Available online xxx
intense, low energy
p / p* intramolecular charge-transfer (ICT) bands, and replacing Me with Ph
Keywords:
Stilbazolium salts
Dye sensitizers
Solar cells
UV-visible absorption spectroscopy
Electrochemistry
increases the ICT energies. Cyclic voltammetric studies show irreversible reduction processes, together
with oxidation waves that are irreversible for R ¼ Me, but reversible for R ¼ Ph. Single crystal X-ray
structures have been determined for three of the methyl ester-substituted stilbazolium salts and for the
Clꢀ salts of their picolinium precursors. Time-dependent density functional theory calculations afford
reasonable predictions of ICT energies, but greater rigour is necessary for eNPh2 derivatives. The four
new acid-functionalized dyes give moderate sensitization efficiencies (ca. 0.2%) when using TiO2-based
photoanodes, with relatively higher values for R ¼ Ph vs Me, while larger efficiencies (up to 0.8%) are
achieved with ZnO substrates.
Density functional theory
Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction
organic solvent containing the Iꢀ/I3ꢀ couple). Driven largely by the
desire to avoid the use of relatively expensive metals such as
Given the urgent need to develop clean and renewable energy
sources, there is much current interest in using molecular dyes,
especially transition metal complexes, as sensitizers in solar energy
harvesting [1]. Traditional silicon solar cells, which rely on
absorption of photons by a pen semiconductor junction, require
high purity materials. Newer inorganic thin film alternatives, such
as cadmium telluride or copper indium gallium selenide, are based
on toxic or rare elements. In a major breakthrough in 1991, O’Regan
and Grätzel reported efficient photosensitization of a wide band-
gap semiconductor by a trinuclear RuII-2,20-bipyridyl-based dye
[2]. The chromophoric properties of such complexes derive from
intense, low energy metal-to-ligand charge-transfer transitions.
A dye-sensitized solar cell (DSSC) contains a nanoparticulate
film of a wide band-gap semiconductor such as TiO2, coated with
a dye monolayer. Photoexcitation of the dye results in electron
injection into the conduction band of the semiconductor, and the
dye is then regenerated by a species in the electrolyte (usually an
ruthenium, many recent studies have focused on using purely
organic dyes in DSSCs [3]. Although such an approach initially
proved somewhat less effective when compared with using metal-
based dyes, state-of-the-art organic sensitizers are now becoming
competitive [4].
Motivated by potential applications in the field of nonlinear
optics, we have investigated previously a range of stilbazolium dyes
that contain N-arylpyridinium groups which act as strong electron
acceptors [5]. Although pyridinium groups are often exploited in
nonlinear optical chromophores [6], and also as electron acceptors
in other photo-active molecular assemblies [7], we are not aware of
any instances in which such substituents have been used in dyes for
solar cell applications. Since they are much stronger electron
acceptors than the carboxylate group which is typically found in
dye sensitizers, strongly red-shifted
p / p* intramolecular charge-
transfer (ICT) bands may be anticipated; increased absorption into
the red and NIR regions will improve matching with the AM 1.5
solar spectrum when compared with most existing sensitizers. In
addition, the ICT absorptions of previously studied N-aryl-
stilbazolium species are relatively intense (with molar extinction
* Corresponding author. Tel.: þ44 161 275 4601; fax: þ44 161 275 4598.
E-mail address: b.coe@manchester.ac.uk (B.J. Coe).
coefficients
3
of up to ca. 75 ꢁ 103 Mꢀ1 cmꢀ1), fulfilling another of
0143-7208/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.dyepig.2011.06.017

O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
767
the key requirements for potentially efficient sensitization. This
report concerns experimental and theoretical studies including
four such new chromophores functionalized with ester groups
which act as precursors to carboxylate units for anchoring to oxide
surfaces.
dichloromethane/methanol. The major colourless band (fluores-
cent under short-wave UV light) was collected and evaporated to
dryness to afford a peach-coloured solid. Yield: 0.194 g (35%, based
on [dnppicþ]Cl$H2O; 23% based on 4-picoline). dH (400 MHz,
CD3OD) 9.15 (2 H, d, J ¼ 6.8 Hz, C5H4N), 8.87 (1 H, t, J ¼ 1.5 Hz, C6H3),
8.67 (2 H, d, J ¼ 1.5 Hz, C6H3), 8.17 (2 H, d, J ¼ 6.3 Hz, C5H4N), 4.02
2. Experimental
(6 H, s, 2CO2Me), 2.83 (3 H, s, Me). n
(C]O) 1734 s cmꢀ1. Anal. Calcd
(%) for C16H16ClNO4$1.3H2O: C, 55.67; H, 5.43; N, 4.06. Found: C,
2.1. Materials and procedures
55.56; H, 5.10; N, 3.88. m/z: 286 ([M ꢀ Cl]þ).
We have reported previously the synthesis of the precursor
compound N-(2,4-dinitrophenyl)-4-picolinium chloride, [dnppicþ]
Cl; this material is unstable in air and quickꢀly forms a tar-like
substance which may be converted into its PF6 salt for improved
handling [5a]. Therefore, [dnppicþ]Cl was treated as a reactive
intermediate in this study and used without purification for reac-
tions with aniline derivatives. All other reagents were obtained
commercially and used as supplied. Products were dried at room
temperature in a vacuum desiccator (CaSO4).
2.3.2. N-(4-Methoxycarbonylphenyl)-4-picolinium chloride,
[(4-MCPh)picþ]Cl
This compound was prepared and purified in a manner similar
to [(3,5-MC2Ph)picþ]Cl$1.3H2O by using methyl-4-aminobenzoate
(2.56 g, 16.9 mmol) in place of dimethoxy-5-aminoisophthalate
(crude product ¼ 0.246 g). The product was finally crystallized by
diffusion of diethyl ether vapour into an ethanol solution. Yield:
63 mg (14%, based on [dnppicþ]Cl$H2O; 9% based on 4-picoline). dH
(400 MHz, CD3OD) 9.13 (2 H, d, J ¼ 6.8 Hz, C5H4N), 8.35 (2 H, d,
J ¼ 8.8 Hz, C6H4), 8.15 (2 H, d, J ¼ 6.3 Hz, C5H4N), 7.97 (2 H, d,
2.2. General physical measurements
J ¼ 9.1 Hz, C6H4), 3.99 (3 H, s, CO2Me), 2.81 (3 H, s, Me).
n(C]O)
1717 s cmꢀ1. Anal. Calcd (%) for C14H14ClNO2$0.8H2O: C, 60.46; H,
5.65; N, 5.04. Found: C, 60.31; H, 5.56; N, 5.12. m/z: 228 ([M ꢀ Cl]þ).
1H NMR spectra were recorded on a Bruker AV-400 spectrom-
eter and all shifts are referenced to TMS. The fine splitting of pyridyl
or phenyl ring AA0BB0 patterns is ignored and the signals are
reported as simple doublets, with J values referring to the two most
intense peaks. Elemental analyses were performed by the Micro-
analytical Laboratory, University of Manchester. IR spectroscopy
was performed on solid samples using an Excalibur BioRad FT-IR
spectrometer, and UVevis spectra were obtained by using a Shi-
madzu UV-2401 PC spectrophotometer. Mass spectra were
measured by using þ electrospray on a Micromass Platform II
spectrometer with acetonitrile as the solvent.
2.3.3. (E)-40-(Dimethylamino)-N-(3,5-bismethoxycarbonylphenyl)-
4-stilbazolium hexafluorophosphate, [1]PF6
[(3,5-MC2Ph)picþ]Cl$1.3H2O (100 mg, 0.290 mmol), 4-(dime-
thylamino)benzaldehyde (102 mg, 0.684 mmol) and piperidine
(4 drops) were added to methanol (20 mL) and the mixture was
heated at reflux for 4 h. Addition of diethyl ether (200 mL) to the
purple solution afforded a dark precipitate which was filtered off,
washed with diethyl ether and dried. Data for crude chloride salt,
[1]Cl: dH (400 MHz, CD3OD) 8.86e8.84 (3 H, C5H4N þ C6H3), 8.61
(2 H, d, J ¼ 1.5 Hz, C6H3), 8.12 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.02 (1H, d,
J ¼ 15.6 Hz, CH), 7.68 (2 H, d, J ¼ 9.1 Hz, C6H4), 7.20 (1 H, d,
J ¼ 15.9 Hz, CH), 6.81 (2 H, d, J ¼ 9.1 Hz, C6H4), 4.02 (6 H, s, 2CO2Me),
Cyclic voltammetric measurements were carried out with an
Ivium CompactStat. An EG&G PAR K0264 single-compartment
microcell was used with a silver/silver chloride reference elec-
trode (3 M NaCl, saturated AgCl) separated by a salt bridge from
a glassy carbon disk working electrode and Pt wire auxiliary elec-
trode. Acetonitrile was freshly distilled (from CaH2) and [NBu4n]PF6,
as supplied from Fluka, was used as the supporting electrolyte.
Solutions containing ca. 10ꢀ3 M analyte (0.1 M electrolyte) were
deaerated by purging with N2. All E1/2 values were calculated from
3.10 (6 H, s, NMe2). n
(C]O) 1721 s cmꢀ1. This material was dissolved
in 1:1 water/methanol and aqueous NH4PF6 (1 M) was added to
give a dark purple precipitate which was filtered off, washed with
water and dried. Yield: 120 mg (72%). dH (400 MHz, CD3COCD3) 9.12
(2 H, d, J ¼ 7.1 Hz, C5H4N), 8.79 (1 H, t, J ¼ 1.4 Hz, C6H3), 8.70 (2 H, d,
J ¼ 1.5 Hz, C6H3), 8.31 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.17 (1 H, d,
J ¼ 15.9 Hz, CH), 7.71 (2 H, d, J ¼ 9.1 Hz, C6H4), 7.36 (1 H, d,
J ¼ 16.1 Hz, CH), 6.85 (2 H, d, J ¼ 9.1 Hz, C6H4), 4.00 (6 H, s, 2CO2Me),
(Epa þ Epc)/2 at a scan rate of 200 mV sꢀ1
.
2.3. Synthesis
3.12 (6 H, s, NMe2). n
(C]O) 1726 s cmꢀ1. Anal. Calcd (%) for
C
25H25F6N2O4P$0.5H2O: C, 52.54; H, 4.59; N, 4.90. Found: C, 52.40;
2.3.1. N-(3,5-Bismethoxycarbonylphenyl)-4-picolinium chloride,
[(3,5-MC2Ph)picþ]Cl
H, 4.48; N, 4.68. m/z: 417 ([M ꢀ PF6]þ).
A
solution of 4-picoline (2.4 mL, 25 mmol) and 2,4-
2.3.4. (E)-40-(Diphenylamino)-N-(3,5-bismethoxycarbonylphenyl)-
4-stilbazolium hexafluorophosphate, [2]PF6
dinitrochlorobenzene (5.00 g, 25 mmol) in ethanol (25 mL) was
heated under reflux for 2 h. After cooling to room temperature,
diethyl ether was added and the black precipitate was filtered off,
washed with diethyl ether and dried. The crude product was
purified by precipitation from boiling ethanol/diethyl ether. The
resulting gray solid was filtered off, washed with diethyl ether and
dried to afford crude [dnppicþ]Cl (5.01 g). A portion of this material,
assumed to be monohydrated (0.500 g, 1.59 mmol), and dime-
thoxy-5-aminoisophthalate (3.54 g, 16.9 mmol) were added to
isopropanol (50 mL), and the suspension was heated at reflux for
72 h. After cooling to room temperature, a dark green/brown
residue was filtered off. The yellow filtrate was evaporated to
dryness, then dissolved in water (100 mL) and extracted with
chloroform. The pale brown aqueous layer was reduced to dryness,
giving the crude product (0.305 g). Purification was effected by
column chromatography on silica gel eluting with 80:20
This compound was prepared in a manner similar to [1]PF6 by
using 4-(diphenylamino)benzaldehyde (187 mg, 0.684 mmol) in
place of 4-(dimethylamino)benzaldehyde to give a dark red solid.
Yield: 177 mg (88%). dH (400 MHz, CD3COCD3) 9.25 (2 H, d,
J ¼ 7.1 Hz, C5H4N), 8.81 (1 H, t, J ¼ 1.5 Hz, C6H3), 8.73 (2 H, d,
J ¼ 1.5 Hz, C6H3), 8.45 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.21 (1 H, d,
J ¼ 16.1 Hz, CH), 7.72 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.53 (1 H, d,
J ¼ 16.1 Hz, CH), 7.43e7.39 (4 H, Ph), 7.24e7.18 (6 H, Ph), 7.01 (2 H, d,
J ¼ 8.8 Hz, C6H4), 4.01 (6 H, s, 2CO2Me).
n
(C]O) 1730 s cmꢀ1. Anal.
Calcd (%) for C35H29N2O4PF6$0.25H2O: C, 60.83; H, 4.30; N, 4.05.
Found: C, 60.85; H, 4.10; N, 4.00. m/z: 541 ([M ꢀ PF6]þ). Data for
crude chloride salt, [2]Cl: dH (400 MHz, CD3OD) 8.98 (2 H, d,
J ¼ 7.1 Hz, C5H4N), 8.87 (1 H, t, J ¼ 1.5 Hz, C6H3), 8.64 (2 H, d,
J ¼ 1.5 Hz, C6H3), 8.24 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.04 (1 H, d,
J ¼ 16.1 Hz, CH), 7.67 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.39e7.33 (5 H,

768
O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
Ph þ CH), 7.20e7.14 (6 H, Ph), 7.00 (2 H, d, J ¼ 8.8 Hz, C6H4), 4.02 (6
H, s, 2CO2Me).
(C]O) 1721 s cmꢀ1
J ¼ 7.1 Hz, C5H4N), 8.84 (1 H, t, J ¼ 1.5 Hz, C6H3), 8.69 (2 H, d,
J ¼ 1.5 Hz, C6H3), 8.42 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.19 (1 H, d,
J ¼ 16.1 Hz, CH), 7.68 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.50 (1 H, d,
J ¼ 15.9 Hz, CH), 7.40e7.36 (4 H, Ph), 7.20e7.15 (6 H, Ph), 6.98 (2 H,
n
.
2.3.5. (E)-40-(Dimethylamino)-N-(4-methoxycarbonylphenyl)-4-
stilbazolium hexafluorophosphate, [3]PF6
d, J ¼ 8.8 Hz, C6H4).
n
(C]O) 1713 s cmꢀ1. Anal. Calcd (%) for
This compound was prepared in a manner similar to [1]PF6 by
using 4-(dimethylamino)benzaldehyde (125 mg, 0.838 mmol) and
[(4-MCPh)pic]Cl$0.8H2O (100 mg, 0.360 mmol) in place of
[(3,5-MC2Ph)picþ]Cl$1.3H2O to give a dark purple solid. Yield:
109 mg (60%). dH (400 MHz, CD3COCD3) 9.05 (2 H, d, J ¼ 7.1 Hz,
C5H4N), 8.33 (2 H, d, J ¼ 8.6 Hz, C6H4), 8.29 (2 H, d, J ¼ 7.1 Hz,
C5H4N), 8.15 (1 H, d, J ¼ 15.9 Hz, CH), 8.07 (2 H, d, J ¼ 8.8 Hz, C6H4),
7.70 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.35 (1 H, d, J ¼ 16.1 Hz, CH), 6.85 (2 H,
C33H25F6N2O4P$2.3H2O: C, 56.62; H, 4.26; N, 4.00. Found: C, 56.62;
H, 3.96; N, 3.80. m/z: 557 ([M
ꢀ
PF6
þ
2Na]þ), 535
([M ꢀ PF6 þ Na]þ), 513 ([M ꢀ PF6]þ).
2.3.9. (E)-40-(Dimethylamino)-N-(4-carboxyphenyl)-4-stilbazolium
hexafluorophosphate, [7]PF6
The crude chloride salt [3]Cl was prepared exactly as described
above, then treated as for [5]PF6 to give a dark purple solid. Yield:
46 mg (26%). dH (400 MHz, CD3COCD3) 9.06 (2 H, d, J ¼ 7.1 Hz,
C5H4N), 8.35 (2 H, d, J ¼ 8.8 Hz, C6H4), 8.29 (2 H, d, J ¼ 7.3 Hz,
C5H4N), 8.16 (1 H, d, J ¼ 15.9 Hz, CH), 8.06 (2 H, d, J ¼ 8.8 Hz, C6H4),
7.71 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.35 (1 H, d, J ¼ 15.9 Hz, CH), 6.85 (2 H,
d, J ¼ 9.1 Hz, C6H4), 3.97 (3 H, s, CO2Me), 3.12 (6 H, s, NMe2).
n(C]O)
1714 s cmꢀ1. Anal. Calcd (%) for C23H23N2O2PF6: C, 54.77; H, 4.60; N,
5.55. Found: C, 54.84; H, 4.46; N, 5.37. m/z: 359 ([M ꢀ PF6]þ). Data
for crude chloride salt, [3]Cl: dH (400 MHz, CD3OD) 8.83 (2 H, d,
J ¼ 7.1 Hz, C5H4N), 8.33 (2 H, d, J ¼ 8.8 Hz, C6H4), 8.12 (2 H, d,
J ¼ 7.1 Hz, C5H4N), 8.02 (1 H, d, J ¼ 15.9 Hz, CH), 7.90 (2 H, d,
J ¼ 8.8 Hz, C6H4), 7.68 (2 H, d, J ¼ 9.1 Hz, C6H4), 7.20 (1 H, d,
J ¼ 15.9 Hz, CH), 6.82 (2 H, d, J ¼ 9.1 Hz, C6H4), 3.98 (3 H, s, CO2Me),
d, J ¼ 9.1 Hz, C6H4), 3.13 (6 H, s, NMe2).
n
(C]O) 1702 s cmꢀ1. Anal.
Calcd (%) for C22H21F6N2O2P$0.5H2O: C, 52.91; H, 4.44; N, 5.61.
Found: C, 52.84; H, 4.03; N, 5.48. m/z: 367 ([M ꢀ PF6 þ Na]þ), 345
([M ꢀ PF6]þ).
3.10 (6 H, s, NMe2). n .
(C]O) 1715 s cmꢀ1
2.3.10. (E)-40-(Diphenylamino)-N-(4-carboxyphenyl)-4-
stilbazolium hexafluorophosphate, [8]PF6
2.3.6. (E)-40-(Diphenylamino)-N-(4-methoxycarbonylphenyl)-4-
stilbazolium hexafluorophosphate, [4]PF6
The crude chloride salt [4]Cl was prepared exactly as described
above, then treated as for [5]PF6 to give a dark red/purple solid.
Purification was effected by reprecipitation from acetone/diethyl
ether. Yield: 22 mg (10%). dH (400 MHz, CD3COCD3) 9.19 (2 H, d,
J ¼ 7.1 Hz, C5H4N), 8.43 (2 H, d, J ¼ 7.2 Hz, C5H4N), 8.36 (2 H, d,
J ¼ 8.7 Hz, C6H4), 8.20 (1 H, d, J ¼ 16.1 Hz, CH), 8.10 (2 H, d, J ¼ 8.8 Hz,
C6H4), 7.71 (2 H, d, J ¼ 8.3 Hz, C6H4), 7.52 (1 H, d, J ¼ 16.1 Hz, CH),
7.43e7.39 (4 H, Ph), 7.23e7.19 (6 H, Ph), 7.01 (2 H, d, J ¼ 8.3 Hz,
This compound was prepared in a manner similar to [3]PF6 by
using 4-(diphenylamino)benzaldehyde (228 mg, 0.834 mmol) in
place of 4-(dimethylamino)benzaldehyde to give a dark red solid.
Yield: 140 mg (62%). dH (400 MHz, CD3COCD3) 9.16 (2 H, d,
J ¼ 7.1 Hz, C5H4N), 8.41 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.34 (2 H, d,
J ¼ 8.8 Hz, C6H4), 8.19 (1 H, d, J ¼ 16.1 Hz, CH), 8.09 (2 H, d, J ¼ 8.8 Hz,
C6H4), 7.71 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.51 (1 H, d, J ¼ 16.1 Hz, CH),
7.43e7.39 (4 H, Ph), 7.23e7.18 (6 H, Ph), 7.00 (2 H, d, J ¼ 8.8 Hz,
C6H4).
n
(C]O) 1707
s
cmꢀ1
.
Anal. Calcd (%) for
C6H4), 3.97 (3 H, s, CO2Me).
n
(C]O) 1715 s cmꢀ1. Anal. Calcd (%) for
C32H25F6N2O2P$0.7H2O: C, 61.29; H, 4.24; N, 4.47. Found: C, 61.34;
C
33H27N2O2PF6: C, 63.06; H, 4.33; N, 4.46. Found: C, 62.86; H, 4.18;
H, 4.13; N, 4.49. m/z: 483 ([M ꢀ PF6 þ Na]þ), 469 ([M ꢀ PF6]þ).
N, 4.27. m/z: 484 ([M ꢀ PF6]þ). Data for crude chloride salt, [4]Cl: dH
(400 MHz, CD3OD) 8.87 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.34 (2 H, d,
J ¼ 8.8 Hz, C6H4), 8.24 (2 H, d, J ¼ 7.1 Hz, C5H4N), 8.04 (1 H, d,
J ¼ 16.1 Hz, CH), 7.93 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.66 (2 H, d, J ¼ 8.8 Hz,
C6H4), 7.39e7.33 (5 H, CH þ Ph), 7.20e7.15 (6 H, Ph), 7.00 (2 H, d,
2.4. X-ray crystallographic studies
Crystals of the salts [(4-MCPh)picþ]Cl and [(3,5-MC2Ph)picþ]
Cl$EtOH were grown by vapour diffusion of diethyl ether into
ethanol solutions, while those of [1]PF6, [2]PF6$MeCN, and [4]
PF6$MeCN were obtained similarly with acetonitrile solutions.
Data were collected on a Bruker APEX CCD X-ray diffractometer by
J ¼ 8.8 Hz, C6H4), 3.99 (3 H, s, CO2Me).
n .
(C]O) 1713 s cmꢀ1
2.3.7. (E)-40-(Dimethylamino)-N-(3,5-biscarboxyphenyl)-4-
stilbazolium hexafluorophosphate, [5]PF6
using MoKa radiation (
by using the Bruker SAINT and SADABS [8] software packages. The
structures were solved by direct methods using SHELXS-97 [9],
l
¼ 0.71073 Å), and the data were processed
The crude chloride salt [1]Cl was prepared exactly as described
above, then dissolved in methanol (5 mL) and aqueous NaOH (2%,
1 mL) was added. The mixture was stirred at room temperature for
7 h and then acidified with concentrated H2SO4 (5 drops). Aqueous
NH4PF6 (1 M) was added and the flask was left in a refrigerator
overnight. The dark purple precipitate was filtered off, washed with
water and dried. Yield: 33 mg (21%). dH (400 MHz, CD3COCD3) 9.15
(2 H, d, J ¼ 7.1 Hz, C5H4N), 8.86 (1 H, t, J ¼ 1.4 Hz, C6H3), 8.69 (2 H, d,
J ¼ 1.3 Hz, C6H3), 8.32 (2 H, d, J ¼ 6.8 Hz, C5H4N), 8.18 (1 H, d,
J ¼ 15.9 Hz, CH), 7.71 (2 H, d, J ¼ 8.8 Hz, C6H4), 7.37 (1 H, d,
J ¼ 15.6 Hz, CH), 6.85 (2 H, d, J ¼ 9.1 Hz, C6H4), 3.13 (6 H, s, NMe2).
2
and refined by full-matrix least-squares on all F0 data using
SHELXL-97 [10]. All non-hydrogen atoms were refined anisotrop-
ically and hydrogen atoms were included in idealized positions
using the riding model, with thermal parameters of 1.2 times
those of aromatic parent carbon atoms, and 1.5 times those of
methyl parent carbons. The crystal of [(4-MCPh)picþ]Cl was
a weakly diffracting needle, so the data were cut at 0.9 Å resolu-
tion. The asymmetric unit of [1]PF6 contains two cations and two
ꢀ
PF6 anions with a disordered solvent fragment that could not be
n
(C]O) 1716 s cmꢀ1. Anal. Calcd (%) for C23H21F6N2O4P: C, 51.69; H,
identified, but is presumably diethyl ether; all atoms of this group
were defined as C at 0.5 occupancy and refined isotropically. For
this structure, the phenyl rings were constrained to be regular
hexagons, and restraints were also applied to the geometry of the
pyridyl group. The crystal was very weakly diffracting and so the
data were cut at 1.2 Å resolution. The crystals of [4]PF6$MeCN
were also extremely weakly diffracting, so the data were cut at 1 Å
resolutioꢀn. In this case, the asymmetric unit contains two cations,
two PF6 anions and two acetonitrile molecules. The atoms
C36eC53 are disordered over two sites, and the two components
3.96; N, 5.24. Found: C, 51.36; H, 4.11; N, 5.11. m/z: 411
([M ꢀ PF6 þ Na]þ), 389 ([M ꢀ PF6]þ).
2.3.8. (E)-40-(Diphenylamino)-N-(3,5-biscarboxyphenyl)-4-
stilbazolium hexafluorophosphate, [6]PF6
The crude chloride salt [2]Cl was prepared exactly as described
above, then treated as for [5]PF6 to give a dark red/purple solid.
Purification was effected by reprecipitation from acetone/diethyl
ether. Yield: 24 mg (12%). dH (400 MHz, CD3COCD3) 9.24 (2 H, d,

O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
769
of the disordered phenyl group were constrained to be regular
hexagons. All other calculations were carried out by using the
SHELXTL package [11]. Crystallographic data and refinement
details are presented in Table 1.
anatase structure of thickness around 13
m
m. After cooling to about
100 ꢂC, the films were immersed into 5 ꢁ 10ꢀ4 M solutions of the
test dyes in acetonitrile and left for 16 h. Chenodeoxycholic acid
(10ꢀ3 M) was added to the dye bath to reduce dye aggregation on
the TiO2 film. The dye-coated film was washed thoroughly with
HPLC grade acetonitrile and then dried under nitrogen. The cells
were assembled by sealing the dye-coated electrodes to thermally
The CCDC depositions 809 929 ([(4-MCPh)picþ]Cl), 809 930
([(3,5-MC2Ph)picþ]Cl$EtOH), 809 931 ([1]PF6), 809 932 ([2]
PF6$MeCN) and 809 933 ([4]PF6$MeCN) contain the supplementary
crystallographic data for this paper. These data can be obtained free
of charge via www.ccdc.cam.ac.uk/data_request/cif, data_request@
ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic
Data Centre, 12, Union Road, Cambridge CB2 1EZ, UK; fax: þ44 1223
336033.
platinized FTO (TEC8) counter electrodes using a 25 mm thermo-
plastic gasket (Surlyn) at 80 ꢂC under pressure. The narrow gap
between the two electrodes was vacuum filled with electrolyte via
holes predrilled in the counter electrode. The filling holes were
sealed with a microscope slip by using Surlyn. The active area of the
cells was 1.0 cm2.
2.5. Theoretical calculations
Two types of electrolyte were used in the TiO2-based cells.
Electrolyte A was composed of 0.06 M I2, 0.6 M 1-propyl-3-
methylimidiazoline iodide (PMII, Merck), 0.2 M NaI, 0.1 M guani-
dinium thiocyanate (GuSCN) and 0.5 M tert-butylpyridine (TBP) in
3-methoxypropionitrile (MPN). Electrolyte B (sodium-free) was
composed of 0.03 M I2, 0.6 M PMII, 0.1 M GuSCN and 0.5 M TBP in
acetonitrile/valeronitrile (85:15 v/v). All of the chemicals used in
the cell fabrication were obtained from SigmaeAldrich unless
stated otherwise.
Structure optimizations (B3LYP/6-311g*) [12,13] and time-
dependent density functional theory calculations were carried
out by using the Gaussian 03 software package [14]. The first 50
excited states were calculated in each case. The conductor-like
polarizable continuum model (CPCM) [15] was employed to take
account of the effect of acetonitrile on chromophore 6. When using
the crystal structure geometry, all counterions and solvent mole-
cules were removed and methyl groups (on the esters) were
replaced with protons (bond distance between O and H modified to
1.00 Å). The UVevis absorption spectra were simulated by using the
GaussSum program [16].
ZnO-based cells were fabricated as follows. The ZnO paste was
prepared by using a 1:1 mixture of two commercial ZnO powders,
Evonik VP AdNano@ZnO20 (particle size ca. 20 nm) and PI-KEM
(particle size ca. 50 nm). For thin film preparation the mixture
was dispersed in water and ethanol (30:70) and stirred overnight to
obtain a colloidal suspension of 30 wt%. This suspension was spread
onto previously cleaned FTO glass with a glass rod using Scotch
tape as spacer, and the film was then heated at 420 ꢂC for 30 min.
The ZnO substrates were coated with the test dyes in similar
manner to the TiO2 films, but using an immersion time of only
30 min. A photoanode coated with the reference dye D149 (Mit-
subishi Paper Mills Limited) was prepared similarly, but using
7 ꢁ 10ꢀ4 M chenodeoxycholic acid in tert-butyl alcohol/acetonitrile
2.6. Fabrication of dye-sensitized solar cells
TiO2-based cells were fabricated as follows. Fluorine-doped tin
oxide (FTO) glass (TEC15, Hartford Glass) was cleaned by successive
sonication for 15 min in aqueous detergent, acetone, isopropanol
and ethanol. A thin compact layer (60 nm) of TiO2 was then
deposited on the cleaned FTO by spray pyrolysis. The mesoporous
layers were prepared by doctor-blading a commercial TiO2 paste
(Dyesol, DSL-18-NR) onto the coated FTO substrates. The film was
dried at 80 ꢂC on a hotplate for 15 min and then sintered at 500 ꢂC
for 30 min to burn out the organic binder, leaving the mesoporous
(1:1). The counter electrode was prepared by spreading 15 ml of
platisol (Solaronix) on the conductive side of TEC8 electrodes and
subsequent annealing at 400 ꢂC for 5 min. Cells with an active area
Table 1
Crystallographic data and refinement details for the salts [(4-MCPh)picþ]Cl, [(3,5-MC2Ph)picþ]Cl$EtOH, [1]PF6, [2]PF6$MeCN and [4]PF6$MeCN.
[(4-MCPh)picþ]Cl
[(3,5-MC2Ph)picþ]Cl$EtOH
[1]PF6
[2]PF6$MeCN
[4]PF6$MeCN
formula
MW
C
14H14ClNO2
C18H22ClNO5
367.82
C56H50F12N4O8P2
1196.94
C37H32F6N3O4P
727.63
C35H30F6N3O2P
669.59
263.71
cryst syst
space group
a (Å)
b (Å)
c (Å)
orthorhombic
Fdd2
28.673(15)
42.40(2)
4.232(2)
90
90
90
5145(5)
16
monoclinic
P2/c
12.232(2)
6.7253(12)
23.092(4)
90
102.791(3)
90
1852.6(6)
4
triclinic
P1
triclinic
P1
triclinic
P1
9.832(7)
10.682(7)
29.87(2)
81.414(13)
86.504(13)
76.879(12)
3020(4)
6.914(9)
11.460(14)
17.32(2)
84.56(2)
84.58(2)
87.50(2)
1359(3)
8.0011(16)
10.175(2)
21.333(4)
99.113(4)
95.656(4)
96.570(3)
1691.4(6)
2
a
b
g
(deg)
(deg)
(deg)
U (Å3)
Z
1
4
Dcalcd (Mg mꢀ3
T (K)
)
1.362
100(2)
0.290
1.319
100(2)
0.233
1.463
100(2)
0.180
1.429
100(2)
0.160
1.473
100(2)
0.167
m
(mmꢀ1
)
crystal size (mm)
crystal appearance
no. of reflns collected
no. of independent reflns (Rint
qmax (deg) (completeness)
0.80 ꢁ 0.02 ꢁ 0.02
colourless needle
7750
1844 (0.1909)
23.25 (99.9%)
1127
0.33 ꢁ 0.30 ꢁ 0.25
orange block
13 878
3774 (0.0835)
25.00 (99.7%)
2387
0.60 ꢁ 0.10 ꢁ 0.05
red plate
3952
1650 (0.1961)
17.22 (99.9%)
820
0.75 ꢁ 0.15 ꢁ 0.05
red plate
10 448
4838 (0.0596)
23.26 (99.7%)
3850
0.60 ꢁ 0.10 ꢁ 0.04
red plate
10 926
6256 (0.1265)
20.81 (98.9%)
2927
)
reflns with I > 2
s(I)
GOF on F2
0.914
1.020
1.107
1.135
0.873
final R1, wR2 [I > 2
s
(I)]
0.0762, 0.1286
0.1278, 0.1468
0.340, ꢀ0.277
0.0695, 0.1313
0.1169, 0.1481
0.678, ꢀ0.880
0.1434, 0.3187
0.2254, 0.3658
0.675, ꢀ0.334
0.0713, 0.1443
0.0925, 0.1534
0.951, ꢀ0.412
0.0700, 0.0947
0.1586, 0.1158
0.294, ꢀ0.309
(all data)
peak and hole (eÅꢀ3
)

770
O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
of 0.81 cm2 were assembled exactly as those containing the TiO2
films.
Five different electrolyte solutions were tested in the ZnO-based
cells with the dye salt [7]PF6: (1) 0.5 M LiI, 0.05 M I2, 0.5 M TBP in
MPN; (2) 5 ꢁ 10ꢀ5 M I2, 0.5 M TBAI in acetonitrile/ethylene
carbonate (1:4); (3) 0.5 M LiI, 0.03 M I2, 0.5 M TBP, 0.1 M GuSCN in
acetonitrile; (4) 0.03 M I2, 0.6 M PMII, 0.1 M GuSCN, 0.5 M TBP in
acetonitrile; (5) 0.06 M I2, 0.6 M PMII, 0.2 M NaI, 0.1 M GuSCN, 0.5 M
TBP in MPN. Electrolyte 2 gave the best cell performance, so data
were obtained for [8]PF6 in only this electrolyte.
2.7. Currentevoltage measurements
The currentevoltage characteristics of the TiO2-based cells were
measured by using a solar simulator (Müller) equipped with 1 kW
xenon lamp. The intensity of the illumination was calibrated with
a standard silicon reference cell (Fraunhofer ISE) to provide 1 sun
(100 mW cmꢀ2). AM 1.5 and KG5 filters were used to minimize the
mismatch between the solar simulator and the AM 1.5 solar spec-
trum. The currentevoltage plots were recorded by using
a computer-controlled system (Whistonbrook).
The ZnO-based cells were characterized with a solar simulator
(ABET) combined with a AM 1.5G filter. A reference cell with
temperature output (Oriel, 91 150) was used to calibrate the illu-
mination output to 1 sun. Photocurrents, photovoltages and cur-
rentevoltage curves were measured by using a 2400 Keithley
SourceMeter.
2.8. Incident-photon-to-electron conversion efficiency (IPCE)
measurements
The IPCE of the TiO2-based cells was measured with a home-
made setup. Illumination was provided by a tungsten lamp in
combination with a monochromator (Bentham, England). A yellow
filter was employed at wavelengths above 550 nm to eliminate
second-order diffracted light, and no bias illumination was used.
The photon flux at each wavelength was measured by a calibrated
photodiode (SD-112 UV), and the photocurrent generated by the
cell at each wavelength was recorded by a current amplifier and
A/D converter (SR 850, Stanford Research System).
Fig. 1. Chemical structures of the stilbazolium cations investigated; all were isolated
and studied as their PF6 salts.
ꢀ
little effect; (ii) the disubstituted species always have higher n(C]
O) values (by 6e17 cmꢀ1) when compared with their mono-
substituted analogues; (ii) ester hydrolysis lowers v(C]O) by
8e17 cmꢀ1
.
3. Results and discussion
3.1. Synthesis and characterization
3.2. Electronic absorption spectroscopy
We have investigated previously salts of N-arylstilbazolium
cations as materials for nonlinear optics [5]. These studies confirm
the strong electron-accepting nature of the pyridinium groups in
such molecules, revealing intense ICT absorptions in the visible
region that may be exploitable in DSSCs. The new cations 1e4
(Fig. 1) were synthesized by using an established method that
involves Knoevenagel condensation of an aldehyde with a picoli-
nium derivative. The required new chloride salt precursors [(3,5-
MC2Ph)picþ]Cl and [(4-MCPh)picþ]Cl were prepared via Zincke-
type reactions with [dnppicþ]Cl [5a], and isolated in moderate
yields after column chromatography on silica gel. Base-catalyzed
ester hydrolysis of 1e4 affords the acid-functionalized dyes 5e8
(Fig. 1), albeit in low isolated yields (10e26%).
UVevis absorption data for [1e8]PF6 in acetonitrile are pre-
sented in Table 2, and representative spectra of [3]PF6 and [4]PF6
are shown in Fig. 2. These spectra are dominated by intense ICT
bands in the visible region, and also show one or more weaker
bands due to
at higher energies.
In each case, replacing R ¼ Me with Ph causes the ICT Emax value to
increase by ca. 0.1 eV (Fig. 2). This trend reflects the weaker -elec-
p / p* transitions having limited directional nature
p
tron donating strength of the diarylamino group inferred by Kwon
et al. by using density functional theory (DFT)-calculated bond
distances and 13C NMR chemical shifts [17]. The ICT band intensities
show little sensitivity to R. The spectra are not affected significantly
either by hydrolysis of the ester groups or by varying the substitution
pattern on the phenyl ring attached to the pyridyl group.
The IR spectra of the new compounds all show relatively intense
n
(C]O) stretching bands, together with various other absorptions,
ꢀ
most notably those due to the PF6 anions in [1e8]PF6 (a very
intense band due to a stretching mode at ca. 835e840 cmꢀ1
together with a sharper but somewhat less intense band at ca.
560 cmꢀ1 due to a bending mode). The energies of the
(C]O)
bands show several general trends: (i) replacing R ¼ Me with Ph has
,
3.3. Electrochemistry
n
Cyclic voltammetric data for [1e8]PF6 in acetonitrile are pre-
sented in Table 2, and representative voltammograms of [1]PF6 and

O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
771
Table 2
UVevis absorption and electrochemical data for salts [1e8]PF6 in acetonitrile.
Salt
lmax, nm
Emax
,
Assignment E, V vs AgeAgCl (D
Ep, mV)b
a
(
3
, 103 Mꢀ1 cmꢀ1
)
eVa
Epa Oxidation Reductiond
c
E1/2
[1]PF6
276 (11.7)
307 (10.7)
518 (51.2)
4.49
4.04
2.39 ICT
p
p
/
/
p
p
*
*
0.92
ꢀ0.79
[2]PF6 279sh (18.5)
303 (23.2)
4.44
4.09
p
p
/
/
p
p
*
*
1.09 1.04 (100) ꢀ0.74
502 (54.5)
2.47 ICT
[3]PF6
276 (12.7)
310 (13.0)
519 (54.9)
4.49
4.00
2.39 ICT
p
p
/
/
p
p
*
*
0.92
ꢀ0.79
[4]PF6 279sh (20.1)
305 (23.4)
4.44
4.07
p
p
/
/
p
p
*
*
1.09 1.04 (100) ꢀ0.73
502 (53.6)
2.47 ICT
[5]PF6
276 (8.63)
310 (9.21)
516 (41.5)
4.49
4.00
2.40 ICT
p
p
/
/
p
p
*
*
0.96
ꢀ0.80
Fig. 3. Cyclic voltammograms for [1]PF6 (black) and [2]PF6 (gray) recorded at
200 mV sꢀ1 in acetonitrile with a glassy carbon working electrode. The arrow indicates
the direction of the initial scans.
[6]PF6 278sh (16.5)
303 (22.0)
4.46
4.09
p
p
/
/
p
p
*
*
1.04 1.02 (60) ꢀ0.71
500 (44.7)
2.48 ICT
presumably located largely on the pyridinium moiety, with
Epc z e(0.7e0.8) V vs AgeAgCl. Similar behaviour is found for
related stilbazolium derivatives [5], and the irreversible nature of
these processes is attributable to the presence of the ethenylene
units which can engage readily in chemical reactions upon reduc-
tion. Small increases in Epc (50e90 mV) on replacing Me with Ph in
the new compounds show that the reduction occurs more readily
when the weaker electron donor group is present, i.e., the LUMO
becomes stabilized.
[7]PF6
276 (8.54)
310 (9.27)
516 (44.5)
4.49
4.00
p
p
/
/
p
p
*
*
0.93
ꢀ0.79
2.40 ICT
[8]PF6 278sh (16.8)
304 (22.2)
4.46
4.08
p
p
/
/
p
p
*
*
1.08 1.04 (80) ꢀ0.72
498 (41.9)
2.49 ICT
Solutions ca. 10e5 M.
a
Solutions ca. 10e3 M in analyte and 0.1 M in [NBun4]PF6; potentials quoted for
b
voltammograms recorded at a scan rate of 200 mV sꢀ1 by using a glassy carbon
working electrode. Ferrocene internal reference E1/2 ¼ 0.45 V,
D
Ep ¼ 80 mV.
c
For an irreversible process.
3.4. X-ray crystallography
d
Epc for an irreversible process.
Representations of the molecular structures of the salts
[(4-MCPh)picþ]Cl, [(3,5-MC2Ph)picþ]Cl$EtOH, [1]PF6, [2]PF6$MeCN
and [4]PF6$MeCN are shown in Figs. 4e8. These structures show
geometric parameters that are generally similar to those found in
related compounds [5], so detailed discussion is unnecessary. The
respective dihedral angles between the pyridyl and phenyl ring in
[(4-MCPh)picþ]Cl and [(3,5-MC2Ph)picþ]Cl$EtOH are ca. 39.5ꢂ and
48.5ꢂ, while the corresponding twists in [1]PF6, [2]PF6$MeCN and
[4]PF6$MeCN are ca. 31.2ꢂ, 44.8ꢂ and 34.2/38.6ꢂ (two independent
[2]PF6 are shown in Fig. 3. Each compound shows a single process
attributable to one-electron oxidation of the HOMO, presumably
located primarily on the phenylamino unit, at ca. 1 V vs AgeAgCl;
these processes are irreversible when R ¼ Me, but reversible when
R ¼ Ph (Fig. 3). This difference is likely attributable to delocaliza-
tion, and hence stabilization, of the positive charge over the phenyl
rings in the eNPh2 derivatives. The Epa values for the latter are
always somewhat higher than those for their eNMe2 analogues, in
keeping with the stronger relative electron donating ability of the
dimethylamino group (see above). In each case, an irreversible
wave is also observed due to one-electron reduction of the LUMO,
cations), respectively. As expected, the increased p-conjugation and
lack of steric hindrance leads to smaller dihedral angles between
the two aryl rings of the stilbazolium fragments; ca. 25.0ꢂ, 8.7ꢂ and
31.3/29.5ꢂ in [1]PF6, [2]PF6$MeCN and [4]PF6$MeCN, respectively.
3.5. Theoretical studies
In order to rationalize the electronic structures and optical
properties of the new chromophores, time-dependent DFT
(TD-DFT) calculations were carried out by using Gaussian 03 [14].
Fig. 2. UVevis absorption spectra of [3]PF6 (black) and [4]PF6 (gray) at 293 K in
Fig. 4. Representation of the molecular structure of [(4-MCPh)picþ]Cl (50% probability
acetonitrile.
ellipsoids).

772
O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
Fig. 5. Representation of the molecular structure of [(3,5-MC2Ph)picþ]Cl$EtOH (50%
probability ellipsoids).
Fig. 7. Representation of the molecular structure of [2]PF6$MeCN (50% probability
ellipsoids).
moment m12. Inspection of the frontier orbitals (Fig. 11) shows that
the HOMOs reside mainly on the aminophenyl and ethenylene
groups, with some contribution from the pyridyl rings. The LUMOs
are spread across the whole molecule, but with major contributions
from the pyridyl rings and ethenylene units. The higher energy
band is in all cases calculated to be made up of two separate
transitions, and the intensities of these relative to the main band
vary somewhat. The ICT energies predicted via TD-DFT when using
the crystal structure geometries are lower than the experimental
values (Tables 2 and 3). Such differences are commonly found when
using TD-DFT to model ICT transitions, as noted in other studies
relevant to DSSCs [3c]. This underestimation of the excitation
energy is reasonable for 5 (0.10 eV), but very large for 6 (0.81 eV)
and 8 (0.78 eV). While structure optimization and inclusion of the
The predicted transition energies and other details are presented in
Table 3. The UVevis spectra were simulated by using GaussSum
[16], and selected spectra are shown in Figs. 9 and 10.
For the eNMe2-substituted cation 5, using the crystal structure
geometry determined for the corresponding ester (in [1]PF6, see
above) gave an adequate fit with the observed UVevis spectrum
(Fig. 9). In the case of the related monosubstituted cation 7, no
directly relevant crystallographic data are available, so the structure
was optimized before the TD-DFT calculations were carried out. For
the eNPh2-substituted chromophores, the calculations produced
less consistent results. For 6, using the crystal structure geometry
determined for the corresponding ester (in [2]PF6$MeCN, see
above) gave a very poor fit with the observed UVevis spectrum
(Fig. 10a). Therefore, the structure was optimized and the TD-DFT
calculations repeated, giving a better match between the pre-
dicted and observed spectra. Additional inclusion of an acetonitrile
solvent continuum (CPCM) further improves the agreement with
the experimental spectrum (Fig. 10b). In the case of 8, using the
crystallographic coordinates for [4]PF6$MeCN also gave relatively
inaccurate results, but geometry optimization has not been per-
formed with this molecule. Optimization of 6 alters the geometry
when compared with the related crystal structure. The stilbazole
portion is flattened out slightly to become almost perfectly planar,
while the dihedral angles between this fragment and the acid-
substituted phenyl ring and between the N-phenyl and phenylene
rings increase. Similar differences are evident between the crys-
tallographic model and optimized structures of 7.
CPCM for 6 produces a closer match between the calculated DE and
the experimental Emax value, this energy is still underestimated by
0.19 eV. Studies by Liu et al. with arylfluorene derivatives having
two eNPh2 substituents underestimated
DE to a similar extent
(ca. 0.2e0.4 eV) at the B3LYP/6-31G(d) level of theory [18]. Such
discrepancies may be attributable to a neglect of solvation effects,
but other factors may also be important. In contrast, geometry
optimization for 7 gives an overestimation of the excitation energy,
with DE being larger than Emax by 0.16 eV.
3.6. Sensitized solar cell studies
The basic physics and chemistry of DSSCs have been reviewed
elsewhere [1d,19,20], including the derivations of key expressions
used here [1d].
In all cases, the lowest energy absorption band is confirmed to
have ICT HOMO / LUMO character with a large transition dipole
Fig. 8. Representation of the moleculaꢀr structure of [4]PF6$MeCN, excluding the
disordered cation and associated PF6 and MeCN molecules (50% probability
ellipsoids).
Fig. 6. Representation of the molecular structure of [1]PF6 (50% probability ellipsoids).

O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
773
Table 3
Data from TD-DFT (B3LYP/6-311g*) calculations on the cations 5e8.
Cation
DE (eV)
f os
Major contributions
m12 (D)
5a
2.30
3.52
1.14
0.23
HOMO / LUMO
11.44
4.17
HOMOe2 / LUMO
HOMO / LUMOþ3
HOMOe2 / LUMO
HOMO / LUMOþ3
HOMO / LUMOþ4
3.71
0.30
4.63
6a
1.67
3.07
0.84
0.23
HOMO / LUMO
11.53
4.42
HOMOe4 / LUMO
HOMO / LUMOþ3
HOMOe4 / LUMO
HOMO / LUMOþ3
3.14
0.40
5.76
6b
6c
7b
2.20
3.07
3.28
1.41
0.25
0.37
HOMO / LUMO
HOMOe1 / LUMO
HOMOe3 / LUMO
12.99
4.63
5.49
2.29
3.42
1.66
0.34
HOMO / LUMO
13.82
5.09
HOMOe1 / LUMO
HOMO / LUMOþ2
2.56
3.59
1.59
0.08
HOMO / LUMO
12.78
2.44
HOMO / LUMOþ1
HOMO / LUMOþ2
HOMOe1 / LUMO
HOMO / LUMOþ4
3.82
0.20
3.68
8a
1.71
3.06
0.85
0.30
HOMO / LUMO
11.47
5.09
HOMOe3 / LUMO
HOMO / LUMOþ2
HOMOe4 / LUMO
HOMOe3 / LUMO
HOMO / LUMOþ2
3.19
0.43
5.98
a
Obtained by using the atomic coordinates from the X-ray crystal structure of the
corresponding ester derivative.
b
Obtained by using the atomic coordinates from the B3LYP/6-311g*-optimized
structure.
c
Obtained by using both the atomic coordinates from the B3LYP/6-311g*-opti-
Fig. 10. TD-DFT-calculated (gray) and experimental (black) UVevis spectra of 6 and [6]
PF6, the latter recorded in acetonitrile. The calculations used (a) the crystal structure
geometry for the corresponding ester (in [2]PF6$MeCN), and (b) the B3LYP/6-311g*-
optimized structure with CPCM. All other details as for Fig. 9.
mized structure and the CPCM (acetonitrile).
The carboxylate-functionalized dye salts [5e8]PF6 were tested
initially in liquid electrolyte cells with mesoporous TiO2 layers.
Plots showing the IeV performance of these cells with electrolyte A
are depicted in Fig. 12, and the AM 1.5 characteristics of the cells
together with information on the LUMO energy levels of the dyes
are summarized in Table 4.
The cells containing [5]PF6, [7]PF6 and [8]PF6 give the same
open-circuit voltage (VOC ¼ 467 mV), while [6]PF6 gives a value
higher by 26 mV (Table 4). The excess electron density nlight ꢀ ndark
under illumination and hence the VOC of a DSSC is determined by
the balance between electron injection from the photoexcited
sensitizer and back elecꢀtron-transfer from the TiO2 to the oxidized
component of the Iꢀ/I3 redox couple. For the usual case where
nlight >> ndark, the electron density under open-circuit conditions
and first-order back electron-transfer is given by
hLHhinjI0
s
nlight
¼
(1)
d
Here hLH is the light harvesting efficiency (i.e., the fraction of inci-
dent photons absorbed), hinj is the electron-injection efficiency, I0 is
the incident photon flux,
d is the thickness of the TiO2 layer (
s
is the lifetime of injected electrons and
corresponds to the inverse of
s
the pseudo first-order rate constant for back electron-transfer to
I3ꢀ). The VOC value is determined by the ratio of electron concen-
trations under illumination and in the dark,
Fig. 9. TD-DFT-calculated (gray) and experimental (black) UVevis spectra of chro-
mophore 5 and [5]PF6, the latter recorded in acetonitrile. The calculations used the
crystal structure geometry determined for the corresponding ester (in [1]PF6). The
nlight
ndark
kBT
q
VOC
¼
ln
(2a)
3
-axes refer to the experimental data only and the vertical axes of the calculated data
where ndark is determined by the difference between the conduc-
tion band edge energy Ec and the redox Fermi level energy EFn
are scaled to match the main experimental absorptions. The oscillator strength axes
refer to the individual calculated transitions (vertical gray lines).
.

774
O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
Fig. 11. TD-DFT-derived depictions of the main orbitals involved in the ICT transitions in 5 (using the X-ray crystal structure of the corresponding ester) and 6 (using the B3LYP/6-
311g*-optimized structure) (isosurface value 0.03 au).
ꢀ
ꢁ
collection losses explain the differences in JSC in the present studies.
All of the dyes [5e8]PF6 absorb very strongly in the visible region,
Ec ꢀ EFn
e
kBT
ndark ¼ Nce
(2b)
with
3 values substantially higher than those of N719 and similar
species (ca. 1.5e2 ꢁ 104 Mꢀ1 cmꢀ1). The red shifts of the ICT bands
of the Me-substituted dyes with respect to their Ph analogues
(Fig. 2) could lead to slightly higher hLH values, so this factor cannot
explain the higher JSC values for [6]PF6 and [8]PF6. This leaves
differences in the injection efficiencies, hinj as the most likely
explanation.
The difference in performance of the dyes is also evident in the
IPCE plots (Fig. 13). The highest IPCE is observed for the cell fabri-
cated with [8]PF6, and the lowest for that with [5]PF6. The peak IPCE
observed for [8]PF6 is only ca. 9%, an order of magnitude lower than
that for the best Ru-based sensitizers. These low IPCE values are
consistent with the low JSC values recorded for the cells.
In order to test whether poor injection is the cause of the low
currents and voltages observed, the performance of cells sensitized
with [6]PF6 and filled with electrolytes A and B was compared. The
results are shown in Fig. 14.
The higher VOC for the cell sensitized with [6]PF6 could be due to
several different factors. Since hLH and are expected to remain
essentially constant for all four dyes, the most probable difference
between them is either hinj or the difference between Ec and EFn
The second important characteristic of a DSSC is the short-
circuit current density, JSC. For AM 1.5 illumination, JSC is deter-
mined by the product of the efficiencies for light harvesting, elec-
tron injection and electron collection (hcol). In principle, all three
efficiencies may depend on photon energy, so that JSC corresponds
to the integral
s
.
Z
JSC ¼ q hLH
ð
l
Þ
hinj
ð
Þ
l hcol
ð
l
ÞIð
l
Þd
l
(3)
where q is the elementary charge. The JSC values, which follow the
sequence [8]PF6 > [6]PF6 >> [7]PF6 ¼ [5]PF6 (Table 4, Fig. 12) are at
least an order of magnitude lower than those measured for iden-
tical cells sensitized with the standard ruthenium complex N719
(JSC ¼ 11.4 mA cmꢀ2), so clearly electron losses are much higher
with the new stilbazolium dyes. The electrolytes used here give hcol
close to 100% with N719-based cells [21], so it is unlikely that
Substitution of electrolyte A (which contains Naþ) by the
sodium-free electrolyte B lowers the JSC values by a factor of 3, but
increases VOC by ca. 60 mV. The lower current and higher voltage
in the sodium-free electrolyte can be attributed to upward
movement of the TiO2 conduction band relative to the I3ꢀ/Iꢀ Fermi
level. Alkali metal ions are known to lower the Ec of TiO2 as
a consequence of changing the surface dipole [22]. This sensitivity
of the cell performance to electrolyte composition suggests that
hinj is influenced by the alignment of the excited state energy and
Ec, but other factors may also be responsible for the low injection
efficiencies.
Table 4
Current/voltage characteristics of TiO2-based DSSCs containing [5e8]PF6 with
electrolyte A and the measured and calculated LUMO energies.
Salt
JSC (mA cme2
)
VOC (mV) FF (%)
h
(%)
LUMO (V)a LUMO (eV)b
[5]PF6 0.73
[6]PF6 0.92
[7]PF6 0.73
[8]PF6 0.95
467
493
467
467
57
59
57
56
0.193 ꢀ0.80
0.268 ꢀ0.71
0.195 ꢀ0.79
0.248 ꢀ0.72
ꢀ5.71
ꢀ5.80
ꢀ5.66
ꢀ5.74
a
Measured via cyclic voltammetry in acetonitrile (V vs AgeAgCl).
b
Obtained by using TD-DFT (B3LYP/6-311g*) calculations on the cation with the
Fig. 12. AM 1.5 Current density-voltage characteristics of TiO2-based DSSCs fabricated
by using salts [5e8]PF6 with electrolyte A.
atomic coordinates from the X-ray crystal structure of the corresponding ester
derivative, excepting 7 for which the B3LYP/6-311g*-optimized geometry was used.

O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
775
Table 5
Current/voltage characteristics of ZnO-based DSSCs containing [7]PF6, [8]PF6 or
D149.
Dye
Electrolyte
JSC (mA cmꢀ2
)
VOC (mV)
FF (%)
h (%)
[7]PF6
[7]PF6
[7]PF6
[8]PF6
D149
1
6
2
2
2
1.68
0.97
2.33
2.70
8.38
503
485
524
492
668
48
55
56
60
35
0.4
0.3
0.7
0.8
2.0
between the dyes and ZnO, which is known to be unstable under
both acidic and basic conditions [1f,25]. Five different electrolytes
were tested with [7]PF6, in order to find the most suitable. Dye
desorption and thus a significant decrease of the photocurrent with
time is observed for electrolytes 4 and 5. Desorption is also
observed with electrolyte 6, although not as strongly as with 4 and
5. The data obtained by using electrolytes 1, 2 and 6 are shown in
Table 5 and IeV curves are depicted in Fig. 15. Given that electrolyte
2 clearly gives the best performance, [8]PF6 was also tested with
this mixture. The IeV curves obtained under these conditions with
[7]PF6, [8]PF6 or the indoline reference dye D149 [26] are shown in
Fig. 16.
Comparison with the data obtained when using TiO2-based
photoanodes (Table 4) shows that the overall efficiencies of solar
cells sensitized with [7]PF6 or [8]PF6 are improved 3e4-fold by
switching to ZnO. Since the Ec value of ZnO is more negative by ca.
100 mV when compared with that of TiO2 (anatase) [27], this factor
cannot be the cause of the increased performance of the new dyes
with ZnO. However, electron injection depends also on other
factors such as the local electric dipole at the surface and the
dielectric constant of the oxide. It may be that the relatively higher
electron mobility of ZnO which facilitates electron transport is also
important, and/or recombination losses may be more significant in
TiO2. While the studies using both metal oxides show that the Ph-
substituted dyes are more efficient than their Me analogues, the
difference between [7]PF6 and [8]PF6 is less pronounced with ZnO.
The performance of the new dyes is still substantially below that
of a common benchmark (in this case D149, Table 5), with JSC values
about 30% as large and VOC values 143e176 mV lower. Nevertheless,
these results are reasonably encouraging in the context of a first
investigation with a new class of sensitizer molecule. In order to
improve the performance of stilbazolium dyes, further tuning of
both the energy levels and the electron donor-acceptor coupling
will be necessary. One possibility that may be expected to affect
Fig. 13. IPCE plots for TiO2-based DSSCs containing [5e8]PF6.
While the data presented here do not relate to excited state
properties, so provide only rough a guide, both the electrochemical
measurements and the TD-DFT calculations (Table 4) indicate that
the LUMO energies of the new dyes are too low with respect to the
conduction band of TiO2 to allow efficient electron injection.
Experiments in acetonitrile-based electrolytes have yielded an Ec
value of ca. ꢀ0.7 V vs AgeAgCl [23], similar to or only a little higher
than the potentials for reduction of the pyridinium groups in [5e8]
PF6. Furthermore, DFT calculations have been used to derive Ec
values of ca. ꢀ(4e5) eV for TiO2 nanoparticles [24], higher than our
predicted LUMO energies. Our TD-DFT calculations also indicate
that the LUMOs feature very little electron density on the acid-
functionalized phenyl rings (Fig. 11). By contrast, the direct
attachment of carboxylate groups to the electron-accepting 2,20-
bipyridyl ligands in RuII-based dyes gives a much more favourable
electron density distribution in the excited states. Time-resolved
studies of electron injection could give more information on this
aspect of the sensitization process.
Although TiO2 has been used most widely, various metal oxides
can be used in DSSCs [1]. Therefore, we have also carried out studies
with the representative dye salts [7]PF6 and [8]PF6 by using ZnO
instead of TiO2-based photoanodes, noting that the number of
carboxylate substituents does not greatly affect the dye perfor-
mance on the latter (Table 4). When compared with the TiO2-based
photoanodes, much shorter immersion times were used (30 min cf.
16 h). This change is intended to minimize chemical reactions
Fig. 14. AM 1.5 currentevoltage characteristics of TiO2-based DSSCs containing [6]PF6
Fig. 15. AM 1.5 currentevoltage characteristics of ZnO-based DSSCs containing [7]PF6
with electrolytes A and B.
with electrolytes 1, 2 and 6.

776
O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
Ingenio 2010), CTQ2009-10477 (TRANSLIGHT), and
a FPU
studentship, and Junta de Andalucia (Andalusian Regional
Government) under projects P07-FQM-02595, P07-FQM-02600,
and P09-FQM-04938.
References
[1] (a) Nazeeruddin MdK, Grätzel M. Conversion and storage of solar energy using
dye-sensitized nanocrystalline TiO2 cells. In: McCleverty JA, Meyer TJ, editors.
Comprehensive coordination chemistry II, vol. 9. Oxford, UK: Pergamon Press;
2004. p. 719e58;
(b) Robertson N. Optimizing dyes for dye-sensitized solar cells. Angew Chem
Int Ed 2006;45:2338e45;
(c) Durrant JR, Haque SA, Palomares E. Photochemical energy conversion:
from molecular dyads to solar cells. Chem Commun; 2006:3279e89;
(d) Peter LM. Characterization and modeling of dye-sensitized solar cells.
J Phys Chem C 2007;111:6601e12;
(e) Special issue on organic photovoltaics. Acc Chem Res 2009;42:1689e857;
(f) Hagfeldt A, Boschloo G, Sun L-C, Kloo L, Pettersson H. Dye-sensitized solar
cells. Chem Rev 2010;110:6595e663.
[2] O’Regan B, Grätzel M. A low-cost, high-efficiency solar cell based on dye-
sensitized colloidal TiO2 films. Nature 1991;353:737e40.
Fig. 16. AM 1.5 currentevoltage characteristics of ZnO-based DSSCs containing elec-
trolyte 2 and [7]PF6, [8]PF6 or D149.
[3] (a) Ning Z-J, Tian H. Triarylamine: a promising core unit for efficient photo-
voltaic materials. Chem Commun; 2009:5483e95;
(b) Mishra A, Fischer MKR, Bäuerle P. Metal-free organic dyes for dye-
sensitized solar cells: from structure:property relationships to design rules.
Angew Chem Int Ed 2009;48:2474e99;
(c) Pastore M, Mosconi E, De Angelis F, Grätzel M. A computational investi-
gation of organic dyes for dye-sensitized solar cells: benchmark, strategies,
and open issues. J Phys Chem C 2010;114:7205e12.
both aspects beneficially is to use N-(2-carboxyvinyl)- instead of
N-arylpyridinium groups.
4. Conclusions
[4] (a) Selected examples: Zhang G-L, Bala H, Cheng Y-M, Shi D, Lv X-J, Yu Q-J,
et al. High efficiency and stable dye-sensitized solar cells with an organic
ꢀ
We have synthesized and characterized a series of new PF6
chromophore featuring
a binary p-conjugated spacer Chem Commun;
salts of N-arylstilbazolium chromophores with eNR2 (R ¼ Me or
2009:2198e200;
(b) Li Q-Q, Lu L-L, Zhong C, Huang J, Huang Q, Shi J, et al. New pyrrole-based
organic dyes for dye-sensitized solar cells: convenient syntheses and high
efficiency. Chem Eur J 2009;15:9664e8;
Ph) substituents. Replacing Me with Ph causes blue shifts of the
intense ICT absorption bands, due to the relatively weaker p-elec-
tron donating ability of the eNPh2 group. In keeping with these
observations, cyclic voltammetry shows that the eNPh2 derivatives
are oxidised less readily than their eNMe2 analogues. These
oxidation processes are irreversible when R ¼ Me, but reversible
when R ¼ Ph. Irreversible waves at negative potentials are due to
pyridinium reductions, with small increases in Epc on replacing Me
with Ph showing that the eNPh2 derivatives are reduced more
readily. Single crystal X-ray structures have been determined for
three of the stilbazolium salts and also for the Clꢀ salts of their
picolinium precursors. TD-DFTcalculations confirm the character of
the visible absorptions, and predict the ICT energy reasonably
accurately for a eNMe2 derivative when using a crystallographi-
cally determined geometry. In contrast, for a eNPh2 derivative,
structure optimization and inclusion of an acetonitrile solvent
continuum is necessary to avoid severe underestimation with
respect to the measured ICT energy. DSSC studies with the new
acid-functionalized compounds on TiO2-based photoanodes afford
moderate efficiencies of ca. 0.2%, with relatively higher values for
R ¼ Ph vs Me. When using ZnO substrates, larger sensitization
efficiencies of up to 0.8% are achieved. The modest performance of
these compounds is attributable to inefficient electron injection
arising from thermodynamic and/or kinetic factors. However,
considerable scope exists for future improvements in the sensi-
tizing abilities of stilbazolium dyes via appropriate molecular
engineering.
(c) Choi H, Kang SO, Ko J, Gao G, Kang HS, Kang M-S, et al. An efficient dye-
sensitized solar cell with an organic sensitizer encapsulated in a cyclodex-
trin cavity. Angew Chem Int Ed 2009;48:5938e41;
(d) Im H, Kim S, Park C, Jang S-H, Kim C-J, Kim K, et al. High performance organic
photosensitizers for dye-sensitized solar cells. Chem Commun 2010;46:1335e7;
(e) Choi H, Raabe I, Kim D, Teocoli F, Kim C, Song K, et al. High molar extinction
coefficient organic sensitizers for efficient dye-sensitized solar cells. Chem Eur J
2010;16:1193e201;
(f) Chen C-H, Hsu Y-C, Chou H-H, Thomas KRJ, Lin JT, Hsu CP. Dipolar compounds
containingfluorene and a heteroaromatic ring as the conjugating bridge forhigh-
performance dye-sensitized solar cells. Chem Eur J 2010;16:3184e93;
(g) Liu X-Z, Zhang W, Uchida S, Cai L-P, Liu B, Ramakrishna S. An efficient organic-
dye-sensitized
solar
cell
with
in
situ
polymerized
poly(3,4-
ethylenedioxythiophene) as a hole-transporting material. Adv Mater 2010;22:
E150e5.
[5] (a) Coe BJ, Harris JA, Asselberghs I, Clays K, Olbrechts G, Persoons A, et al.
Quadratic nonlinear optical properties of N-aryl stilbazolium dyes. Adv Funct
Mater 2002;12:110e6;
(b) Coe BJ, Harris JA, Asselberghs I, Wostyn K, Clays K, Persoons A, et al.
Quadratic optical nonlinearities of N-methyl and N-aryl pyridinium salts. Adv
Funct Mater 2003;13:347e57;
(c) Coe BJ, Foxon SP, Harper EC, Harris JA, Helliwell M, Raftery J, et al. The
syntheses, structures and nonlinear optical and related properties of salts with
julolidinyl electron donor groups. Dyes Pigments 2009;82:171e86.
[6] (a) Other recent examples: Coe BJ. Switchable nonlinear optical metal-
lochromophores with pyridinium electron acceptor groups Acc Chem Res
2006;39:383e93;
(b) Kim HS, Sohn KW, Jeon Y, Min H, Kim D, Yoon KB. Aligned inclusion of n-
propionic acid tethering hemicyanine into silica zeolite film for second
harmonic generation. Adv Mater 2007;19:260e3;
ꢀ
(c) Figi H, Mutter L, Hunziker C, Jazbinsek M, Günter P, Coe BJ. Extremely large
nonresonant second-order nonlinear optical response in crystals of the stil-
bazolium salt DAPSH. J Opt Soc Am B 2008;25:1786e93;
(d) Compain J-D, Mialane P, Dolbecq A, Marrot J, Proust A, Nakatani K, et al.
Second-order nonlinear optical properties of polyoxometalate salts of a chiral
stilbazolium derivative. Inorg Chem 2009;48:6222e8;
Acknowledgements
(e) Coe BJ, Fielden J, Foxon SP, Brunschwig BS, Asselberghs I, Clays K, et al.
Combining very large quadratic and cubic nonlinear optical responses in
We thank the EPSRC for support in the form of a PhD student-
ship (YTT) and also DyStar UK Ltd and the University of Manchester
for funding a PhD studentship (OAB). We are grateful to Dr Joseph
J. W. McDouall (Manchester) for advice concerning TD-DFT calcu-
lations. The work in Bath was supported by the EPSRC (grant
EP/E035469/1). EG and JAA thank the Ministerio de Ciencia e
Innovación of Spain for project HOPE CSD2007-00007 (Consolider-
extended, tris-chelate metallochromophores with six
p-conjugated pyr-
idinium substituents. J Am Chem Soc 2010;132:3496e513.
[7] (a) Selected examples: Kim Y-H, Das A, Zhang H-Y, Dutta PK. Zeolite
membrane-based artificial photosynthetic assembly for long-lived charge
separation J Phys Chem B 2005;109:6929e32;
(b) Lainé PP, Bedioui F, Loiseau F, Chiorboli C, Campagna S. Conformationally
gated photoinduced processes within photosensitizereacceptor dyads based
on osmium(II) complexes with triarylpyridinio-functionalized terpyridyl

O.A. Blackburn et al. / Dyes and Pigments 92 (2011) 766e777
777
ligands: insights from experimental study.
7510e21;
J
Am Chem Soc 2006;128:
[19] Peter LM. Dye-sensitized nanocrystalline solar cells. Phys Chem Chem Phys
2007;9:2630e42.
(c) Weber JM, Rawls MT, MacKenzie VJ, Limoges BR, Elliott CM. High energy
and quantum efficiency in photoinduced charge separation. J Am Chem Soc
2007;129:313e20;
(d) Zhang H-Y, Rajesh CS, Dutta PK. Ruthenium polypyridyl complexes con-
taining a conjugated ligand LDQ (LDQ ¼ 1-[4-(40-methyl)-2,20-bipyridyl]-2-[4-
(40- N, N0-tetramethylene-2,20-bipyridinum)]ethene): synthesis, characteriza-
tion, and photoinduced electron transfer at solution-zeolite interfaces. J Phys
Chem C 2009;113:4623e33.
[20] Kalyanasundranam K, editor. Dye sensitized solar cells. Lausanne,
Switzerland: EPFL Press; 2010.
[21] Wang HX, Peter LM. A comparison of different methods to determine the
electron diffusion length in dye-sensitized solar cells. J Phys Chem C 2009;
113:18125e33.
[22] Park N-G, Chang S-H, van de Lagemaat J, Kim K-J, Frank AJ. Effect of cations on
the open-circuit photovoltage and the charge-injection efficiency of dye-
sensitized nanocrystalline rutile TiO2 films. Bull Korean Chem Soc 2000;21:
985e8.
[23] Lemon BI, Liu F, Hupp JT. Electrochemical, spectral, and quartz crystal
microgravimetric assessment of conduction band edge energies for nano-
crystalline zirconium dioxide/solution interfaces. Coord Chem Rev 2004;248:
1225e30.
[8] SAINT (Version 6.45) and SADABS (Version 2.10). Madison: Wisconsin, USA:
Bruker AXS Inc; 2003.
[9] Sheldrick GM. Phase annealing in SHELX-90: direct methods for larger struc-
tures. Acta Crystallogr Sect A 1990;46:467e73.
[10] Sheldrick GM. SHELXL 97, Program for crystal structure refinement. Göttin-
gen, Germany: University of Göttingen; 1997.
[11] SHELXTL (Version 6.10). Madison: Wisconsin, USA: Bruker AXS Inc; 2000.
[12] Becke AD. Density-functional thermochemistry. III. The role of exact
exchange. J Chem Phys 1993;98:5648e52.
[24] (a) Kusama H, Orita H, Sugihara H. DFT investigation of the TiO2 band
shift by nitrogen-containing heterocycle adsorption and implications on
dye-sensitized solar cell performance. Sol Energ Mat Sol Cells 2008;92:
84e7;
[13] Krishnan R, Binkley JS, Seeger R, Pople JA. Self-consistent molecular orbital
methods. XX. A basis set for correlated wave functions. J Chem Phys 1980;72:
650e4.
[14] Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman VG,
et al. Gaussian 03, Revision C.02. Wallingford CT: Gaussian, Inc.; 2004.
[15] (a) Barone V, Cossi M. Quantum calculation of molecular energies and energy
gradients in solution by a conductor solvent model. J Phys Chem A 1998;102:
1995e2001;
(b) De Angelis F, Fantacci S, Selloni A. Alignment of the dye’s molecular
levels with the TiO2 band edges in dye-sensitized solar cells:
a DFTeTDDFT study. Nanotechnology 2008;19:424002.
[25] Keis K, Lindgren J, Lindquist S-E, Hagfeldt A. Studies of the adsorption
process of Ru complexes in nanoporous ZnO electrodes. Langmuir 2000;16:
4688e94.
[26] (a) Onwona-Agyeman B, Kaneko S, Kumara A, Okuya M, Murakami K,
Konno A, et al. Sensitization of nanocrystalline SnO2 films with indoline dyes.
Jpn J Appl Phys, Part 2 2005;44:L731e3;
(b) Cossi M, Rega N, Scalmani G, Barone V. Energies, structures, and electronic
properties of molecules in solution with the C-PCM solvation model. J Comp
Chem 2003;24:669e81.
(b) Minoura H, Yoshida T. Electrodeposition of ZnO/dye hybrid thin films for
dye-sensitized solar cells. Electrochemistry 2008;76:109e17;
(c) Guillén E, Idígoras J, Berger T, Anta JA, Fernández-Lorenzo C, Alcántara R,
et al. ZnO-based dye solar cell with pure ionic-liquid electrolyte and organic
sensitizer: the relevance of the dye-oxide interaction in an ionic-liquid
medium. Phys Chem Chem Phys 2011;13:207e13.
[16] O’Boyle NM, TenderholtAL, Langner KM. cclib: a library for package-independent
computational chemistry algorithms. J Comp Chem 2008;29:839e45.
[17] Kwon O, Barlow S, Odom SA, Beverina L, Thompson NJ, Zojer E, et al. Aromatic
amines: a comparison of electron-donor strengths. J Phys Chem A 2005;109:
9346e52.
[18] Liu Y-L, Feng J-K, Ren A-M. Theoretical study of optical and electronic prop-
erties of the bis-dipolar diphenylamino-endcapped oligoarylfluorenes as
promising light emitting materials. J Phys Org Chem 2007;20:600e9.
[27] (a) Nozik A. Photochemistry: applications to solar energy conversion. Annu
Rev Phys Chem 1978;29:189e222;
(b) Serpone N, Pelizzetti E, editors. Photocatalysis: fundamentals and appli-
cations. New York: Wiley; 1989.