Atomic level resolution of dye regeneration in the dye-sensitized solar cell

Article abstract of DOI:10.1021/ja311640f

Two donor-acceptor organic dyes have been synthesized that differ only by a two-heteroatom change from oxygen to sulfur within the donor unit. The two dyes, (E)-3-(5-(4-(bis(4-(hexyloxy)phenyl)amino)phenyl)thiophen-2-yl)-2- cyanoprop-2-enoic acid (Dye-O) and (E)-3-(5-(4-(bis(4-(hexylthio)phenyl)amino) phenyl)thiophen-2-yl)-2-cyanoprop-2-enoic acid) (Dye-S), were tested in solar cell devices employing both I₃^-/I^--based and [Co(bpy)₃]^3+/2+ redox mediators. Power conversion efficiencies over 6% under simulated AM 1.5 illumination (1 Sun) were achieved in both electrolytes. Despite similar optical and redox properties for the two dyes, a consistently higher open-circuit voltage (V_oc) was measured for Dye-S relative to Dye-O. The improved efficiency observed with Dye-S in an iodide redox mediator is against the commonly held view that sulfur atoms promote charge recombination attributed to inner-sphere interactions. Detailed mechanistic studies revealed that this is a consequence of a 25-fold enhancement of the regeneration rate constant that enhances the regeneration yield under open circuit conditions. The data show that a high short circuit photocurrent does not imply optimal regeneration efficiency as is often assumed.

Full text of DOI:10.1021/ja311640f

Article
pubs.acs.org/JACS
Atomic Level Resolution of Dye Regeneration in the Dye-Sensitized
Solar Cell
Kiyoshi C. D. Robson,† Ke Hu,‡ Gerald J. Meyer,*,‡ and Curtis P. Berlinguette*,†
†Department of Chemistry and Centre for Advanced Solar Materials, University of Calgary, 2500 University Drive Northwest,
Calgary T2N 1N4, Canada
‡Departments of Chemistry and Materials Science & Engineering, Johns Hopkins University, 3400 North Charles Street, Baltimore,
Maryland 21218, United States
S
* Supporting Information
ABSTRACT: Two donor−acceptor organic dyes have been synthesized that differ only by a
two-heteroatom change from oxygen to sulfur within the donor unit. The two dyes, (E)-3-
(5-(4-(bis(4-(hexyloxy)phenyl)amino)phenyl)thiophen-2-yl)-2-cyanoprop-2-enoic acid (Dye-O)
and (E)-3-(5-(4-(bis(4-(hexylthio)phenyl)amino)phenyl)thiophen-2-yl)-2-cyanoprop-2-enoic acid)
(Dye-S), were tested in solar cell devices employing both I3 /I−-based and [Co(bpy)3]3+/2+
−
redox mediators. Power conversion efficiencies over 6% under simulated AM 1.5 illumina-
tion (1 Sun) were achieved in both electrolytes. Despite similar optical and redox properties
for the two dyes, a consistently higher open-circuit voltage (Voc) was measured for Dye-S
relative to Dye-O. The improved efficiency observed with Dye-S in an iodide redox mediator
is against the commonly held view that sulfur atoms promote charge recombination
attributed to inner-sphere interactions. Detailed mechanistic studies revealed that this is a consequence of a 25-fold enhancement
of the regeneration rate constant that enhances the regeneration yield under open circuit conditions. The data show that a high
short circuit photocurrent does not imply optimal regeneration efficiency as is often assumed.
the oxidized form of the redox mediator present in the elec-
trolyte, either molecular iodine, I2, or tri-iodide, I3 . The high
INTRODUCTION
■
−
Dye-sensitized solar cells (DSSC) continue to be actively pur-
sued as low cost alternatives to traditional photovoltaics.1
The successful utilization of CoIII/II diimine compounds as redox
mediators2−5 removes some of the inherent thermodynamic
surface concentration of these iodine-based electron acceptors at
the TiO2 interface results in conditions that favor unwanted
charge recombination and hence lower the quasi-Fermi level
of the illuminated TiO2 thin film.13 Based largely on this novel
work, it is now widely accepted that the presence of sulfur
heteroatom(s) in a dye molecule has a deleterious influence on
energy conversion when employed in DSSCs. This conclusion
is unfortunate as large classes of organic dye, such as poly-
thiophenes and phenothiazines, possess S atoms.
To resolve the nature of these interactions, two oxygen atoms
in the donor−acceptor organic dye (E)-3-(5-(4-(bis(4-(hexyloxy)-
phenyl)amino)phenyl)thiophen-2-yl)-2-cyanoprop-2-enoic acid
(Dye-O)14 were replaced with two sulfur atoms, (E)-3-(5-(4-
(bis(4-(hexylthio)phenyl)amino)phenyl)thiophen-2-yl)-2-cyano-
prop-2-enoic acid) (Dye-S) shown in Figure 1. The dye molecules are
“donor−acceptor” or “push−pull” type organic dyes wherein
charge transfer occurs from the triphenylamine group (TPA) to
the cyanoacetic acid moiety that can bind to the TiO2. This
orbital arrangement allows for efficient charge separation and
vectorial electron transfer toward the TiO2 surface subsequent to
light absorption. Moreover, these donor and acceptor units are
common modalities in high performance organic and/or
inorganic DSSC sensitizers.15−19
6
−
limitations of the I3 /I− mediators and could soon enable a
leapfrog jump in efficiency. Also impressive is the molecular
detail in which DSSCs can now be optimized.7,8 Isotopic substi-
tution has also been shown to enhance excited-state electron
injection yields.9 Indeed, several recent reports demonstrate that
exchange of a single atom, or a few atoms, in a dye molecule
markedly changes the efficiencies of DSSCs comprised of them.
While the magnitude of the efficiency change can be quite
dramatic, the origin(s) generally remains speculative. Adducts
comprised of one of the exchanged atoms with redox mediators
present in the electrolyte are generally invoked to rationalize
photoelectrochemical behavior.10,11 In all cases investigated thus
far, adduct formation lowers DSSC efficiencies. The findings dis-
closed herein suggest that adduct formation, most likely involv-
ing iodide activation, can result in enhanced power conversion
efficiencies when the position of the exchanged atom within the
dye structure is carefully taken into account.
One would indeed expect changes in DSSC efficiency when
heteroatoms in a dye molecule are exchanged. Adduct formation
between the dye and the semiconductor surface and/or species
present in the electrolyte alters redox potentials and hence the
interfacial electron-transfer kinetics that govern electrical power
generation. O’Regan and Palomares have reported evidence of
N−I and S−I binding.10,12 The iodine species was thought to be
Received: December 6, 2012
Published: January 10, 2013
© 2013 American Chemical Society
1961
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
open-circuit voltage (Voc), as the S atoms enhance the rate of the
reaction between the oxidized dye and the redox mediator by
3-fold with Co-based mediators, and 25-fold with iodide-based
mediators.
EXPERIMENTAL SECTION
■
Preparation of Compounds. All reactions and manipulations
were performed using solvents passed through an MBraun solvent
purification system prior to use. All reagents were purchased from
Aldrich. Purification by column chromatography was carried out using
silica (Silicycle: Ultrapure Flash Silica). Analytical thin-layer chromatog-
raphy (TLC) was performed on aluminum-backed sheets precoated
with silica 60 F254 adsorbent (0.25 mm thick; Merck, Germany) and
1
visualized under UV light. Routine H and 13C NMR spectra were re-
corded at 400 and 100 MHz, respectively, on a Bruker AV 400 instru-
ment at ambient temperature. Chemical shifts (δ) are reported in parts
per million (ppm) from low- to high-field and referenced to residual
nondeuterated solvent in the Supporting Information. Standard abbre-
viations indicating multiplicity are used as follows: s = singlet; d =
1
doublet; t = triplet; m = multiplet. All H chemical shifts follow the
general labeling scheme in Figure 2. Organic and inorganic compounds
Dye-O14 and P1-S,20 P2-S,20 P3-S,20 4,4,5,5-tetramethyl-2-(thiophen-
2-yl)-1,3,2-dioxaborolane,21 1,3-dimethylimidazolium iodide (DMII),22
Figure 1. Molecular structures of the organic dyes Dye-O and Dye-S.
The heteroatoms that have an effect on the regeneration reaction with
redox mediators are indicated. The β-LUSO (lowest-unoccupied spin
orbital) for the oxidized forms of each dye, Dye-O+ and Dye-S+, are
provided below each structure to highlight the delocalization of the wave
function onto the heteroatoms of interest (the terminal hexyl
substituents were replaced with methyl substituents for the
calculations).
5
[Co(bpy)3](PF6)2,5 and [Co(bpy)3](PF6)3 were synthesized accord-
ing to literature reports.
Substrate Preparations. Mesoporous nanocrystalline TiO2 thin
films used in the spectroscopy studies were prepared as previously
described.23 Film thickness was determined to be about 5 μm by
profilometry. The films were briefly immersed in acetonitrile solution
containing dye (∼75 μM) and coadsorbent chenodeoxycholic acid
(∼3.7 mM). Films were then washed with neat CH3CN and diagonally
positioned in a standard 1 cm2 quartz cuvette containing 0.5 M LiClO4
acetonitrile solution. In some experiments, the sensitized thin films were
sandwiched against a glass microscope with a vinyl film (Warps, 8 mil
In this study, the location of the substitution of the two-atom
changes was carefully taken into consideration. As shown in
Figure 1, there is a sizable orbital coefficient on the heteroatom
(either O or S, for Dye-O and Dye-S, respectively) in the lowest-
unoccupied spin orbital (LUSO) of the oxidized forms of the
dyes (i.e., Dye-O+ and Dye-S+). Consequently, the delocaliza-
tion of the LUSO onto the heteroatoms offers the opportunity to
interrogate how S and O atoms can affect DSSC performance,
particularly in view of the fact that it is this same portion of the
oxidized sensitizer that directly interacts with the redox mediator.
Comparative interfacial electron transfer studies were therefore
conducted with Dye-O and Dye-S anchored to mesoporous
2+
Vinyl-Pane) spacer that contained desired concentrations of Co(bpy)3
or LiI in 0.5 M LiClO4 acetonitrile solutions. The electrolyte solutions
were purged with Ar gas for at least 30 min prior to experimentation.
Spectroelectrochemistry. Steady-state UV−visible absorption
spectra were obtained on a Varian Cary 50 spectrophotometer at
room temperature. A potentiostat (BAS model CV-50W) was employed
for measurements in a standard three-electrode arrangement with a
sensitized TiO2 thin film deposited on a FTO substrate working
electrode, a platinum disk counter electrode, and a Ag/AgCl reference
electrode (Bioanalytical Scientific Instruments, Inc.) in 0.5 M LiClO4
acetonitrile solution. All potentials are reported versus the normal
hydrogen electrode (NHE). The ferrocenium/ferrocene half-wave
potential was measured at room temperature before and after each
−
TiO2 thin films in the presence and absence of CoIII/II and I3 /I−
redox mediators. It is shown that the identity of the heteroatom
can have a profound effect on dye regeneration, and in turn the
Figure 2. UV−vis absorption spectra of Dye-O (red) and Dye-S (blue) in (a) fluid CH3CN solution and (b) anchored to TiO2 immersed in CH3CN
solution at 298 K.
1962
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
experiment and was used as an external standard to calibrate the
reference electrode. Conversion constant of −640 mV from NHE to
Fc+/Fc was used in acetonitrile at 25 °C.
meter (model 2400). All cells were measured with a mask size of
0.13 cm2. IPCE measurements were performed on a QEX7 Solar Cell
Spectral Response Measurement System from PV Instruments, Inc. The
system was calibrated with a photodiode that was calibrated against
NIST standard I755 with a transfer uncertainty of less than 0.5%
between 400 and 1000 nm, and less than 1% at all other wavelengths. All
measurements were made in AC mode at a 10 Hz chopping frequency
under a bias light. The system was calibrated and operated in Beam
Power mode.
Transient Absorption Spectroscopy. Nanosecond transient
absorption measurements were obtained with an apparatus similar to
that which has been previously described.23 Briefly, samples were excited
by a frequency doubled Q-switched, pulsed Nd:YAG laser (Quantel
USA (BigSky) Brilliant B; 532 nm, 5−6 ns full width at half-maximum
(fwhm), 1 Hz, ∼10 mm in diameter) directed 45° to the film surface.
A Glan-Taylor polarizer was employed in the laser path to attenuate
the pulse fluence. A 150 W xenon arc lamp coupled to a 1/4 m
monochromator (Spectral Energy, Corp. GM 252) served as the probe
beam (Applied Photophysics) that was aligned orthogonally to the
excitation light. For detection at sub-100 μs time scales, the lamp
was pulsed with 80 V. Detection was achieved with a monochromator
(Spex 1702/04) optically coupled to an R928 photomultiplier tube
(Hamamatsu). Transient data were acquired on a computer-interfaced
digital oscilloscope (LeCroy 9450, Dual 350 MHz) with 2.5 ns resolu-
tion terminated at 50 Ω for sub-100 μs; for longer time scales, the signal
was terminated with a 10 kΩ resistor and bandwidth limited at 80 MHz.
Typically, 30−50 laser pulses were averaged at each observation wave-
length over the range 380−710 nm for full spectra generation. 300−500
laser pulses were typical averages for single wavelength measurement
to achieve satisfactory S/N. Kinetic data fitting and spectral modeling
was performed in Origin 8, and least-squares error minimization was
accomplished using the Levenberg−Marquardt iteration method.
DFT Calculations. Density functional theory (DFT) calculations
were carried out using B3LYP and the 6-31G(d) basis set. All geometries
were fully optimized in the ground states (closed shell) and the +1
oxidized states (open shell). Time-dependent density functional theory
(TD-DFT) calculations were performed with an IEFPCM solvation
model (CH3CN) using a spin-restricted formalism to examine low-
energy excitations at the ground-state geometry. All calculations were
carried out with the Gaussian 03W software package.24
RESULTS
■
Synthesis and Characterization. The preparation of Dye-O
followed a previously reported literature procedure.14 A similar
protocol was followed for Dye-S with the exception of the
generation of the P2-S precursor, which was isolated following
an electrophilic aromatic substitution reaction between P1-S and
C6H13S− (Scheme S1, Supporting Information). A subsequent
electrophilic bromination step with NBS furnished the
bromoderivative P3-S, which could undergo reaction with
4,4,5,5-tetramethyl-2-(thiophen-2-yl)-1,3,2-dioxaborolane
under Suzuki cross-coupling conditions to yield P4-S. A Vislmeier−
Haack formylation was employed to install the −CHO group
(P5-S), which was condensed with cyanoacetic acid to yield the
target Dye-S.
The optical spectra of Dye-O and Dye-S recorded in CH3CN
are very similar, Figure 2a. Dye-O is characterized by an intense
charge-transfer absorption band centered at λmax = 478 nm
(ε = 2.84 × 104 M−1 cm−1), while that of Dye-S is slightly
hypsochromically shifted to λmax = 470 nm (ε = 3.16 × 104 M−1
cm−1). The slight differences in absorption are attributed to less
orbital overlap of the sulfur lone pairs with the π system of the
aryl rings of the TPA groups relative to those of the oxygen
atoms, which was corroborated by the 1H NMR chemical shifts
and redox chemistry (vide infra) of the two complexes.20 Time-
dependent DFT indicates that, for both Dye-O and Dye-S, the
HOMO, HOMO−1, and HOMO−2 orbitals are localized to the
electron-rich TPA moiety, and the LUMO and LUMO+1 levels
are confined to the cyanoacetic acid portion of the molecule;
the lowest energy transition (Figures S2 and S3) is a LUMO ←
HOMO transition. Light excitation into the charge transfer absorp-
tion bands resulted in room temperature fluorescence (Figure S4).
The dyes were found to bind to thin films of anatase, mesoporous
TiO2 with high surface coverages over 1 × 10−7 mol/cm2, and are
herein abbreviated Dye-O/TiO2 and Dye-S/TiO2. When the
sensitized thin films were measured in neat CH3CN, the
absorption maxima were slightly blue-shifted from that observed
in fluid solution, Figure 2. The presence of 0.5 M LiClO4 induced
a significant bathochromic (red) shift in both cases, measured to
be 870 cm−1 for Dye-O/TiO2 and 750 cm−1 for Dye-S/TiO2
(Figure S5). Lewis acidic cations, such as Li+ and Na+, bind to
anatase TiO2, and the local field generated is known to induce a
bathochromic shift of dye molecules anchored to the same
surface.25
Cell Fabrication. Photoanodes were prefabricated by Dyesol Inc.
(Australia) with screen-printable TiO2 pastes (18-NRT and WER4-O,
Dyesol). The active area of the TiO2 electrodes was 0.28 cm2 with a
thickness of 6 μm (18-NRT) and 3 μm (WER4-O) on fluorine-doped
tin-oxide [FTO; TEC8 (8 Ω cm−2)]. The TiO2 substrates were treated
with TiCl4(aq) (0.05 M) at 70 °C for 30 min and subsequently rinsed
with H2O and then dried prior to heating. The electrodes were heated to
450 °C for 20 min in an ambient atmosphere and left to cool to 80 °C
prior to immersing into an acetonitrile solution containing the dye
(∼0.25 mM) and chenodeoxycholic acid (∼2.5 mM) for 16 h. The
stained films were then rinsed with copious amounts of CH3CN and
dried. The cells were fabricated using Pt-coated counter-electrode [FTO
TEC-15 (15 Ω cm−2)] and sealed with a 30 μm Surlyn (Dupont) gasket
−
by resistive heating. Two electrolytes were used for this study: (i) I3 /I−
[1.0 M 1,3-dimethylimidazolium iodide (DMII), 60 mM I2, 0.5 M tert-
butylpyridine, 0.05 M NaI, and 0.1 M GuNCS in a mixed solvent sys-
tem of acetonitrile and valeronitrile (85:15, v/v)]; (ii) CoIII/II [0.21 M
[Co(bpy)3](PF6)2,0.033 M [Co(bpy)3](PF6)3, 0.1 M NaClO4, and
0.2 M tert-butylpyridine in acetonitrile]. Each electrolyte was introduced
into the two sandwiched electrodes via vacuum backfilling through a
−
hole in the counter electrode. In the cases where the I3 /I− electrolyte
was used, the hole was sealed with an aluminum-backed Bynel foil
(Dyesol). In the cases where the CoIII/II electrolyte was used, the hole
was covered with a thin Surlyn (25 μm) microscope slide; aluminum foil
was then taped to the back of the cell. Silver bus bars were added to all
cells after sealing.
Cell Characterization. Photovoltaic measurements were recorded
with a Newport Oriel solar simulator (model 9225A1) equipped with a
class A 150 W xenon light source powered by a Newport power supply
(model 69907). The light output (area = 5 cm × 5 cm) was calibrated to
AM 1.5 using a Newport Oriel correction filter to reduce the spectral
mismatch in the region of 350−700 nm to less than 1.5%. The power
output of the lamp was measured to 1 Sun (100 mW cm−2) using a
certified Si reference cell. The current−voltage (I−V) characteristic of
each cell was obtained by applying an external potential bias to the cell
and measuring the generated photocurrent with a Keithley digital source
The cyclic voltammograms of Dye-O or Dye-S in 0.1 M
NBu4BF4/CH3CN electrolyte displayed quasi-reversible TPA•+/
TPA0 redox waves at +0.96 and +1.06 V vs NHE, respectively
(Figure S6). Cyclic voltammetry of the sensitized thin films
showed broad waves with large peak-to-peak splittings (Figure S7).
Spectroelectrochemistry was therefore used to estimate the reduc-
tion potentials of Dye-O/TiO2 and Dye-S/TiO2. Oxidation of
Dye-O/TiO2 or Dye-S/TiO2 resulted in characteristic absorp-
tion changes attributed to the oxidized triphenylamine, Figure 3.
Isosbestic points were maintained, and the spectral changes were
reversible upon stepping the potential back to less positive values.
1963
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
Figure 3. Visible absorption spectra of (a) Dye-O/TiO2 (dye surface coverage Γ = 1.0 × 10−8 mol cm−2) and (b) Dye-S/TiO2 (Γ = 5.3 × 10−8 mol cm−2)
■
measured at the indicated applied potentials in 0.5 M LiClO4 CH3CN solution. The insets show the fraction (x) of dye molecules present in the reduced ( )
●
and the oxidized state (red ). Overlaid on this data is a sigmoidal fit, x = 1/(1 + 10 exp((Eapp − E°1/2)/a × 59 mV)), where a is the ideality factor determined
to be 1.27 for Dye-O/TiO2 and 1.05 for Dye-S/TiO2.
The insets of Figure 3 show the fraction of molecules in each
oxidation state as a function of applied potential, Eapp. This frac-
tion of (x) was calculated by x = 1/(1 + 10 exp((Eapp − E°)/a ×
59 mV)). The formal reduction potential, E°, was taken as the
equilibrium potential where the fraction present was 0.5, and was
measured to be +1090 mV vs NHE for Dye-S/TiO2 while +1015 mV
for Dye-O/TiO2. Ideality factors of a = 1.27 for Dye-O/TiO2
and a = 1.05 for Dye-S/TiO2 were required to fit the data.13
In some cases, the fractional surface coverage was converted to
a chemical capacitance in units of mF/cm2 by (−dx/dEapp) ×
Table 1. Summary of Spectroscopic and Electrochemical
Properties for Dye-O and Dye-S in Solution and Anchored
to TiO2
λ
E1/2 (V vs E0−0 E(S+/S*)
c d e
a
PL b
(nm) NHE) (eV) (V vs NHE)
compound
λmax (nm)
478 (2.84)
468
Dye-O
598
0.96
1.02
1.06
1.09
2.31
2.03
2.32
2.12
−1.35
−1.01
−1.26
−1.03
Dye-O/TiO2
Dye-S
469 (3.16), 318 (2.96) 623
453
Dye-S/TiO2
a
Corresponds to maximum of lowest-energy absorption band re-
Γ ×
-
× 106, where Γ is the surface coverage in mol/cm2 and
corded in CH3CN; ε values indicated in parentheses with units of
is Faraday’s constant (96 485 C mol−1).
The E0−0 energy was calculated to be 2.31 and 2.32 eV for
b
×104 M−1 cm−1. Fluorescence maxima determined in CH3CN after
-
c
excitation at a wavelength corresponding to λmax
.
Corresponds to
Dye-O and Dye-S, respectively, from the intersection of the
absorption and fluorescence spectra (Table 1 and Figure S4).
The excited-state reduction potential, E(S+/S*), was then cal-
culated by E(S+/S*) = E(S+/S) − E0−0 to be −1.35 and −1.26 V
vs NHE for Dye-O and Dye-S, respectively.
TPA•+/TPA0 redox couple. Data collected in 0.1 M NBu4BF4 CH3CN
solutions at 100 mV/s and referenced to [Fc]+/[Fc] (+640 mV vs
NHE). Data for sensitized TiO2 thin films were collected in 0.5 M
LiClO4 CH3CN solutions by density of states measurements (see
d
Figure 11). In fluid solution, E0−0 was estimated from the intersection
of the normalized absorption and fluorescence spectra (see Figure S4).
For the sensitized TiO2 thin films, E0−0 was estimated from the inter-
The application of negative applied potentials to the sensitized
thin films immersed in 0.5 M LiClO4/CH3CN resulted in reduc-
tion of TiO2 and the characteristic absorption of the trapped
TiO2 electron, TiO2(e−). The potential dependence of this re-
duction was found to be the same for Dye-O/TiO2 and Dye-S/
TiO2, and was well described by a single exponential fit (Figure
S8). The measured absorbance was related to the number of
TiO2(e−)’s per 15 nm diameter spherical particle through Beer’s
law and the extinction coefficient ε = 1000 M−1 cm−1 with an
effective optical path length for a 5 μm thick film of 50% porosity
positioned 45° to the probe light. For example, the absorp-
tion due to 20 electrons per particle was calculated as Abs =
ε·b·c(TiO2(e−)) = 1000 (M−1 cm−1) × 5 (μm) × 1.414 × 10−4
(cm μm−1) × 50% × 20/(4/3 × π × (7.5 × 10−9(m))3 × NA/
103(L m−3)) = 0.007.
Interfacial Charge Recombination. Nanosecond transient
absorption spectroscopy was used to quantify interfacial electron
transfer from TiO2 to the oxidized sensitizers. Absorption dif-
ference spectra of Dye-O/TiO2 and Dye-S/TiO2 displayed in
Figure 4 were acquired in 0.5 M LiClO4 CH3CN after pulsed
532 nm excitation of the sensitized thin films. Simulations based
on a 1:1 stoichiometry of oxidized dyes and injected electrons,
cept of a tangent line drawn on the red edge of ground-state absorp-
e
tion. Estimated using E(S+/S*) = E1/2 − E0−0
.
Dye+/TiO2(e−), plus the first derivative of the ground-state
absorption were in good agreement with the experimental data.
Normalization of the measured spectra at different observation
times revealed that the bleach in the initial spectra was red-shifted
relative to those observed at longer delay times. The first
derivative absorption feature is consistent with previous studies
and the presence of an underlying electric field that influences the
sensitizer absorption spectra in a manner similar to that observed
in Stark spectroscopy. The time-dependent spectral changes are
due to screening of the electric field by solvents and ions.25,26 The
immediate appearance of the absorption spectrum of the
oxidized dye molecule indicated that electron injection occurred
with a rate constant kinj > 108 s−1.
Shown in Figure 5 are absorption changes measured after
pulsed light excitation of Dye-O/TiO2 and Dye-S/TiO2 in 0.5 M
LiClO4/CH3CN that correspond to charge recombination
between the injected electron and the oxidized sensitizer. Both
sensitized thin films had a ground-state absorption of 0.43 at the
1964
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
Figure 4. Absorption difference spectra measured at the indicated delay times after pulsed 532 nm excitation of (a) Dye-O/TiO2 and (b) Dye-S/TiO2
immersed in 0.5 M LiClO4/CH3CN solution. Overlaid in purple are simulated spectra.
Figure 5. Absorption change monitored at 635 nm after pulsed light excitation (532 nm; laser fluence: 0.3 mJ/cm2) of Dye-O/TiO2 (left) and
Dye-S/TiO2 (right) immersed in 0.5 M LiClO4/CH3CN at the indicated applied potentials. Overlaid in yellow are best fits to the KWW kinetic model.
532 nm excitation wavelength. The kinetics were nonexponential
but were satisfactorily modeled by Kohlrausch−Williams−Watts
(KWW) function, eq 1, with a β value of 0.20 0.01 for both sen-
sitizers on mesoporous nanocrystalline TiO2 surface.27 Average
charge recombination rate constants, kcr, were calculated as
Sensitizer Regeneration. The regeneration of Dye-O+/
TiO2(e−) and Dye-S+/TiO2(e−) by iodide and [Co(bpy)3]2+
donors was investigated. A laser pulse was used to generate the
interfacial charge separated state, Dye-O+/TiO2(e−) or Dye-S+/
TiO2(e−), in an electrolyte that contained known amounts of
I− or [Co(bpy)3]2+. The data were recorded at an observation
wavelength of 635 nm where the oxidized dye predominantly
absorbed light, Figure 6. The observed kinetic data could not be
fit to a single exponential function, but were adequately described
by a sum of two exponential functions, from which an average
rate constant was calculated, eqs 3 and 4.
the first moment28 of the underlying Lev
́
y distribution of rate
constants, eq 2.
ΔAbs = A exp[−(kt)β]
(1)
−1
⎡
⎢
⎣
⎤
⎥
⎦
⎛ ⎞
1
kβ
1
⎝ β ⎠
kcr =
× Γ
⎜ ⎟
ΔAbs = A0 + A1 exp(−k1t) + A2 exp(−k2t)
(2)
(3)
The rate constants were found to be within experimental error
the same, kcr = 1.3 0.1 × 102 s−1 for Dye-O/TiO2 and kcr =
1.4 0.1 × 102 s−1 for Dye-S/TiO2 (Figure S9). The reduction
of TiO2 with forward bias had a marked influence on the average
charge recombination rate constants as is shown in Figure 5. The
kcr values were found to increase exponentially with the applied
potential from +100 to −300 mV vs NHE (Figure S10).
The Voc was measured in the absence of redox mediator under
the same condition as the charge recombination measurement
described previously. Over 2 decades of irradiances, the Voc values
for Dye-O/TiO2 and Dye-S/TiO2 were within experimental error
the same (Figure S11).
A1/k1 + A2/k2
kobs
̅
=
A1/k12 + A2/k22
(4)
2+
The kobs
̅
values were linear in the Co(bpy)3 and I− as is
shown in Figure 7. The slopes from the linear fits provided the
second-order rate constants for dye regeneration. The values are
summarized in Table 2.
The regeneration efficiency, ϕreg, was quantified on the basis of
the charge recombination and regeneration rate constants, eq 5:
ϕ
= kreg[D]/(kcr + kreg[D])
reg
(5)
1965
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
Figure 6. Absorption changes measured after pulsed laser excitation of the sensitized film in the presence of the indicated concentration of
[Co(bpy)3](ClO4)2, (a) Dye-O/TiO2 and (b) Dye-S/TiO2, or LiI, (c) Dye-O/TiO2 and (d) Dye-S/TiO2 in 0.5 M LiClO4/CH3CN. Absorbance
changes were monitored at a probe wavelength of 635 nm (excitation wavelength, 532 nm; laser fluence, 0.1 mJ/cm2). Overlaid in yellow are the best fits
to biexponential kinetic model.
■
●
Figure 7. Plots of observed regeneration rate constants (kobs
̅
) of Dye-O/TiO2 (red ) and Dye-S/TiO2 (blue ) obtained from the kinetics fits versus
titrated (a) [Co(bpy)3]2+ and (b) [I−] concentrations.
−
where D is the donor concentration. The ϕreg values reflect the
facts that charge recombination rate constants were highly
sensitive to the extent of forward bias and that regeneration rate
constants were larger for Dye-S+/TiO2(e−), Table 3.
times (>10 μs), the injected electron and I3 were observed as
well as a derivative feature attributed to a shift in the ground-state
absorption of the dye molecules due to the injected electrons
(an electroabsorption or Stark effect)25,26 (Figure S12). Recom-
−
Transient absorption spectra were also recorded after pulsed
bination of the injected electron with I3 was found to be well
laser excitation in the presence of iodide. At long observation
modeled by KWW function with a β value of 0.8 (Figure S13).
1966
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
and experimentally measured values were within experimental
Table 2. Second-Order Regeneration Rate Constants (kreg)
Abstracted for Dye-O/TiO2 and Dye-S/TiO2 in 0.5 M
LiClO4/CH3CN Containing Cobalt- and Iodide-Based Redox
Mediators
•−
error the same for Dye-S/TiO2, consistent with quantitative I2•−
disproportionation (Figure S14). In other words, all of the I2
that was photogenerated underwent disproportionation with no
evidence for a TiO2(e−) + I2•− → reaction. This same conclusion
could not be reached for Dye-O/TiO2 due to the more sluggish
regeneration rates that made kinetic analysis difficult.
[Co(bpy)3]2+
iodide
a
b
c
kreg
(M−1 s−1
ΔGo
(V)
kreg
ΔGo
ΔGo
)
(M−1 s−1
)
(V)
(V)
Light-to-Electrical Energy Conversion. Photocurrent
density−voltage plots are shown for DSSCs with Dye-O and
Dye-S using an iodide- and cobalt-based electrolytes, Figure 8.
The DSSCs fabricated with Dye-S display power conversion
Dye-O/TiO2
Dye-S/TiO2
1.1 × 106
3.5 × 106
−0.46
−0.53
1.5 × 106
0.21
0.14
−0.09
−0.16
3.7 × 107
a
b
E°([Co(bpy)3]3+/2+) = 0.56 V vs NHE. Calculated free energy
change for outer-sphere reactions based on E°(I·/I−) = 1.23 V vs
NHE. Calculated free energy change for outer-sphere reactions based
−
efficiencies (η) similar to those fabricated with 1 with the I3 /I−-
c
based electrolytes, Table 4. The DSSCs fabricated with Dye-O
on E°(I2·−/2I−) = 0.93 V vs NHE.
−
and Dye-S and an I3 /I−-based electrolyte displayed PCEs of
6.1 0.2% and 6.2 0.3%, respectively. The DSSCs fabricated
with Dye-O or Dye-S and the CoIII/II redox mediator displayed η
of 4.6 0.4% and 6.3 0.7%, respectively.
Table 3. Regeneration Efficiencies Calculated by Equation 5
at the Indicated Forward Bias Conditions
bias (mV vs NHE) ϕreg 0.5 M I− ϕreg 0.2 M [CoII(bpy)3]2+
Photocurrent action spectra, measured as the incident
photon-to-current efficiency (IPCE) versus wavelength shown
in Figure 9, were in good agreement with the absorptance spectra.
The slight decrease in photocurrent efficiency that was measured
below 420 nm with the iodide containing electrolyte is due to
competitive light absorption by tri-iodide. The Dye-O absorbed
light to about 650 nm. When corrections were made for light
scattering and absorption by the FTO substrate, the efficiency
was near 100%.
Dye-O/TiO2
Dye-S/TiO2
80
0
1.0
1.0
1.0
1.0
−70
−140
−200
−270
100
0.99
0.97
0.83
0.57
1.0
0.97
0.91
0.60
0.28
1.0
30
1.0
1.0
−40
−110
−180
−240
1.0
1.0
DISCUSSION
0.99
0.99
0.99
0.99
0.95
0.85
■
The two dyes investigated in this study, which differ only by a
two-heteroatom change from oxygen to sulfur within the donor
unit, revealed nearly quantitative short-circuit photocurrent
efficiencies when employed in DCCS with iodide or cobalt redox
mediators. However, the open circuit photovoltage, Voc, was
found to be largest for Dye-S under all conditions investigated.
It is not the absolute values of the conversion efficiencies that are
novel in this study as Wang et al. have reported more optimal
data,14 and even higher values for structurally related organic
dyes have been realized by Bai and co-workers.30 Instead, this
comparative study revealed an improvement with Dye-S that
stands in stark contrast to the commonly accepted view that
sulfur atoms incorporated within the organic framework of a dye
molecule enhance recombination and hence lower the overall
light-to-electrical energy conversion efficiency. In fact, recombi-
nation of the injected electrons with the oxidized dye was within
experimental error the same for the two sensitized materials.
An average rate constant was calculated to be 2.3 0.4 s−1 with
eq 2.
At short observation times, there was clear kinetic evidence for
the presence of I2•−. It was of interest to test whether injected
electrons recombined with I2•−. This was accomplished by a
method that was previously described.29 The absorption change
at 396 nm, where both I3− and I2•− absorbed light, was measured
at a sufficiently long delay time (10 μs) to ensure that all of the
•−
photogenerated I2 had disproportionated. The initial ampli-
•−
tude of the absorption change expected if quantitative I2 dis-
proportionation occurred was then calculated, based on the
known relative extinction coefficients of I3 and I2
compared to the experimentally measured value. After a small
correction for contributions from the oxidized dye, the calculated
29
•−
−
,
and
−
Figure 8. Current−voltage curves for DSSCs fabricated with Dye-O and Dye-S at AM1.5 illumination using (A) I3 /I− and (B) CoIII/II redox mediators.
1967
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
Table 4. Solar Cell Performance Parameters Obtained under Simulated AM 1.5 Illumination (1 Sun) for Dye-O and Dye-S
a
b
compound
electrolyte
Jsc (mA/cm2)
Voc (mV)
fill factor
η (%)
I−/I3
11.7 0.2
10.3 0.6
10.1 0.5
10.8 0.4
734 21
798 24
714 31
828 38
70.5 1.5
74.4 2.2
63.7 4.8
70.0 3.6
6.1 0.2
6.2 0.3
4.6 0.4
6.3 0.7
−
Dye-O
Dye-S
Dye-O
I−/I3
−
CoIII/II
CoIII/II
Dye-S
a
−
The I3 /I− electrolyte solution is composed of 1.0 M 1,3-dimethylimidazolium iodide (DMII), 60 mM I2, 0.5 M tert-butylpyridine, 0.05 M NaI, and
0.1 M GuNCS in a mixed solvent system of acetonitrile and valeronitrile (85:15, v/v), while the CoIII/II electrolyte is composed of 0.21 M
b
[Co(bpy)3](PF6)2, 0.033 M [Co(bpy)3](PF6)3, 0.1 M NaClO4, and 0.2 M tert-butylpyridine in acetonitrile. A 0.13 cm2 mask was used; averaged
parameters, with standard deviations, from no fewer than three different cells are listed.
−
Figure 9. Incident-photon-to-current efficiency (IPCE) for DSSCs fabricated using (a) I3 /I− and (b) CoIII/II redox mediators.
Likewise, recombination of injected electrons with oxidized iodide
ions was also sensitizer independent. Surprisingly, the origin of the
disparate Voc values was instead traced to an enhanced regeneration
rate and efficiency for Dye-S. Detailed mechanistic studies revealed
that regeneration was optimal at the short circuit condition for both
dye molecules, but was not rapid enough for Dye-O to compete
with Dye-O+/TiO2(e−) → Dye-O/TiO2 recombination as the
number of electrons trapped in TiO2 was raised toward the values
expected at the open circuit condition. Below we discuss these two
key findings in regard to the interfacial energetics, kinetics, and
conversion efficiency in dye-sensitized solar cells.
Interfacial Energetics. Spectroelectrochemistry of the
sensitized thin films revealed that the electrochemical reduction
of TiO2 was insensitive to the identity of the surface anchored
dye molecules. Reduction of TiO2 results in a blue-black colora-
tion of the film that has been attributed to trapped TiO2(e−)’s,
most probably as TiIII sites.31 Employing the approach of Bisquert
and co-workers,32 the potential-dependent redox chemistry was
Figure 10. Interfacial energetics of Dye-O/TiO2 and Dye-S/TiO2 in 0.5
converted to a chemical capacitance. A key assumption was that
M LiClO4/CH3CN solution.
electrons could be independently added and removed from the
TiO2; that is, the electrons are screened from each other. With
this assumption, molecular excited states with reduction
potentials as low as 0 V vs NHE would be able to inject electrons
into TiO2, Figure 10. Dye-O and Dye-S are much more potent
potential and was found to be 75 mV more positive for Dye-S/
TiO2 than for Dye-O/TiO2. This difference was attributed to
the poorer orbital overlap of the lone pair of the sulfur atom with
the phenyl ring.20 Nevertheless, measurements of the photo-
reductants, and quantitative injection was expected. The knowl-
edge that the E(S+/S*) potential is ∼600 mV higher than needed
for efficient electron injection allows for futureworkon developing
these sensitizers by extending the π-bridge to lower the E(S+/S*)
and consequently covering a larger portion of the solar spectrum.
The dye molecules could be reversibly oxidized, and the poten-
tial dependence is displayed in the form of chemical capacitance.
The peak of the distributions shown corresponds to the poten-
tial where equal concentrations of the oxidized and reduced
forms were present. This value was taken as the formal reduction
voltage generated in the absence of redox mediators showed that
the free energy stored in the Dye-O+/TiO2(e−) and Dye-S+/
TiO2(e−) interfacial charge-separated states were within
experimental error the same. Taken together, these data indicate
that the change of heteroatoms did not appreciably influence
interfacial thermodynamics for Dye-O/TiO2 and Dye-S/TiO2.
Interfacial Kinetics. Charge recombination between the
injected TiO2(e−) and the oxidized dye molecules was within
experimental error the same for both Dye-O/TiO2 and Dye-S/
TiO2. This observation was expected as the driving force for the
1968
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
Figure 11. Molecular structures of two ruthenium(II)-based dyes studies by O’Regan and co-workers10 indicating the site where the heteroatoms have
been changed. The β-LUSO (lowest-unoccupied spin orbital) for the oxidized form of the ruthenium dye where X = S is shown to the left of the β-LUSO
for Dye-S+ (also shown in Figure 1) to highlight that there is no orbital character over the S atoms in the metal dye and significant orbital character over
the S atoms in the organic dye. The portions of the respective dyes that ostensibly interact with the redox mediator are illustrated. The equivalent spin-
density plots are found in Figure S15.
reactions differed by merely 75 mV.33,34 The average rate con-
stants, kcr, abstracted from a distributional analysis of the kinetic
data were highly sensitive to the presence of excess TiO2(e−)’s.
This behavior was first recognized by Durrant and co-workers for
endergonic by +140 mV (E°(I•/I−) =1.23 V vs NHE), while it is
exergonic by −160 mV for I2•− formation (E°(I2•−/2I−) = 0.93 V
vs NHE).40 The corresponding driving forces for regeneration
of Dye-S/TiO2 studies here are essentially identical, suggesting
specific interactions between the oxidized dye and iodide. While
these arguments are not definitive as inner-sphere dye−iodide
adducts may activate iodide and significantly alter the thermo-
dynamics,41 ion-pairing with the oxidized dye is generally invoked
to explain regeneration mechanisms at sensitized TiO2 interfaces,
reaction 7.42
35
cis-Ru(4,4′-dicarboxy-2,2′-bipyridine)2(NCS)2/TiO2 and was
immediately apparent for the organic dyes studied herein. The kcr
values increased exponentially with the applied forward bias that,
given the exponential density of states known for TiO2, implies a
reaction that is first-order in TiO2(e−) concentration.
The kinetics for sensitizer regeneration were quantified spec-
troscopically. With [Co(bpy)3]2+ as the electron donor, second-
order rate constants of the order of 106 M−1 ·s−1 were measured
for reaction 6.
[I−, Dye‐X+/TiO2] + I− → Dye‐X/TiO2 + I2
(X = O or S)
•−
(7)
In the present case, the rate constant for regeneration differs
by a factor of 25 with only a 75 mV change in driving force for
Dye-O+ and Dye-S+. This finding is striking in that previous
kinetic data indicate that a 75 mV change in driving force would
not have such a dramatic influence on the electron transfer rate
constant.11,30 For example, Palomares and co-workers observed
a 4-fold enhancement of regeneration with an additional
60 mV driving force,11 while O’Regan and co-workers observed
a 20-fold enhancement of regeneration with over 300 mV driving
force.43 Clearly the dependence of the measured electron
transfer rate constants on −ΔG will be a function of where they
sit on the Marcus parabola. Reactions that occur near the top of
the parabola, that is, where −ΔG = λ, may show a very shallow
ΔG dependence, whereas those that are highly unfavored show a
stronger dependence. Nevertheless, a 25-fold change in rate
constant cannot reasonably be attributed to only a 75 mV change
in driving force.
Inspection of the molecular orbitals of the two dyes reveals a
higher orbital coefficient on the S atom relative to the O atom.
This description is more predominant in the oxidized forms of
the dyes, which is the more relevant form to be considered here.
Dye-O and Dye-S offer the first examination of how different
heteroatoms affect the regeneration step, where the heteroatoms
constitute the portion of the molecule that will be interacting
with the electrolyte. Previous studies on the heteroatom effect
in coordination compounds placed the heteroatom on a remote
portion of a bipyridine ligand, Figure 11. Consequently, this
resulted in differences in behavior that were manifest in the
recombination step, but not necessarily the regeneration step.
The reaction between the oxidized forms of Dye-O and Dye-S
and the redox mediator will almost certainly involve the portion
2+
Dye‐X+/TiO2 + CoII(bpy)3 → Dye‐X/TiO2(e−)
3+
+ CoIII(bpy)3
(X = O or S)
(6)
At high [Co(bpy)3]2+ concentrations (i.e., above 0.13 M) and
low concentrations of trapped TiO2(e−)’s, the regeneration step
was quantitative, consistent with the high photocurrent efficiencies
measured at the short circuit condition. The high photocurrents also
indicated that the unwanted charge recombination of the injected
electron with the oxidized cobalt compound, [Co(bpy)3]3+, was
inefficient. This observation likely emanates from the well-
known spin change that accompanies CoIII/II redox chemistry
that results in anomalously small self-exchange rate constants.36,37
The [Co(bpy)3]2+ compound has a d7 high spin electron configura-
tion (t2g eg*2), and [Co(bpy)3]3+ has a d6 low-spin configuration
5
6
(t2g ). Consistent with other sensitizers that have displayed high
solar conversion efficiencies with a CoIII/II redox mediator, the long
hexyl hydrocarbon chains of the dye molecules are postulated to
effectively shield the TiO2 surface from the [Co(bpy)3]3+.2
Iodide oxidation has been the subject of many studies38,39 and
occurs by two different mechanisms: one that is first-order in
iodide to generate the iodine atom; and the other that is second-
order in iodide to yield an iodine radical anion called di-iodide,
I2•−. It remains unclear which mechanism(s) are operative at
sensitized TiO2 interfaces. This ambiguity is due in part to the
rapid reactivity of the iodine atom with iodide to yield the same
I2•− product. However, the latter mechanism is generally thought
to be operative in solar cells as the thermodynamics for this
outer-sphere electron transfer pathway are more favorable.6 For
example, the free energy change for regeneration of the classical
N3 sensitizer, cis-Ru(dcb)2(NCS)2, to yield the iodine atom is
1969
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
of the molecule bearing the triaryl fragment containing the
heteroatoms of interest. Consequently, the much faster rates of
dye regeneration provide incontrovertible evidence that the S
atoms mediate an interaction with the redox mediator that is
favorable relative to the O atoms. Adduct formation between
iodide and the soft more polarizable sulfur atom as the Lewis base
at the regeneration site would provide an inner-sphere electron
transfer pathway that could enhance electronic coupling and
hence alter the electronic transmission coefficient. Alternatively,
the higher rate of the reaction could simply be a result of the
greater orbital coefficient on the S atom on the O atom.
quantitative regeneration and the low Voc value for Dye-O/TiO2
results from poor regeneration when the TiO2(e−) concentration
is high. Hence, the detailed mechanistic studies reported herein
show that the observation of quantitative short circuit photo-
currents is not necessarily an indication that regeneration is
optimal. Quantitative regeneration of Dye-S+/TiO2 as the
TiO2(e−) concentration was raised toward the values expected
at the open circuit condition underlies its superior performance
and Voc values relative to Dye-O/TiO2.
CONCLUSIONS
■
In either event, transient absorption data clearly show that the
unwanted charge recombination reaction with oxidized iodide
species was not influenced by the presence of the sulfur atoms
positioned near the donor site of these organic dyes as was
invoked previously for the ruthenium-based sensitizers.10 Re-
To our knowledge, this study provides the first direct comparison
of molecules where a simple two-atom change, at the site of
regeneration, systematically probes the rates of regeneration by
I− (and Co(II)). The results of this study contrast the broadly
accepted view that S atoms enhance the rate of recombination
between the TiO2(e−) and the oxidized iodide mediator, thus
lowering the solar conversion efficiency of the DSSC. These
disparities in reactivities are manifest in the specific positioning of
the heteroatom change within the portion of the dye molecule
that is most likely to react with the incoming reductant I−. These
results highlight that the atomic composition of the dye can have
a profound effect on the dye regeneration process, and that
the careful design of molecular dyes can lead to an optimization
of interactions with the redox mediator that can lead to further
enhancements in dye regeneration. Furthermore, the data
provide compelling evidence that sluggish regeneration can
lower the open circuit photovoltage even when quantitative short
circuit photocurrents are measured.
−
combination of the injected electrons with I3 or I2 was within
experimental error the same for the two dye-sensitized TiO2
materials, 2.3
0.4 s−1. Di-iodide, an intermediate in the
−
regeneration reaction and a I3 reduction product, is also a
potential electron acceptor,44,45 yet there was no evidence for a
reaction between Dye-S/TiO2(e−) and I2•−. Instead, the I2
•−
formed during regeneration underwent disproportionation with
a quantum yield and reaction rate constant that was unchanged
from that measured in fluid acetonitrile solution.28
The question remains whether more rapid regeneration by
Dye-S+/TiO2(e−) could account for the larger open circuit
photovoltage, Voc, measured in the DSSCs. Generally speaking,
the observation of quantitative short circuit photocurrents and
IPCE values has led previous authors to conclude that regenera-
tion is optimal.7 However, at short circuit the injected electrons
are rapidly removed from TiO2 thin film, whereas at the open
circuit condition many electrons are trapped within each nano-
crystallite, behavior that is known to enhance charge recombina-
tion.35 Indeed, as electrons were thermally transferred to TiO2
with an external potentiostat, recombination to these oxidized
organic dyes increased significantly and the corresponding re-
generation efficiency decreased. At −200 mV vs NHE, the re-
generation efficiency for Dye-S+/TiO2(e−) was near unity for
both donors, while that for Dye-O+/TiO2(e−) had dropped by
about 20% for iodide and 40% for Co(bpy)32+. The more
significant decrease for the cobalt mediator reflects the fact that it
did not regenerate as quickly as did iodide. These regeneration
efficiencies explain the smaller Voc values measured for Dye-O
as well as it is greater sensitivity to the cobalt mediator as is
described more fully below.
Correlations of measurements made under potentiostatic con-
ditions with an operational solar cell are nontrivial. Ideally, one
would like to employ transient absorption spectroscopy directly
as an in situ characterization tool for DSSCs as has been done in
pioneering studies by O’Regan and co-workers.46,47 However,
transmission losses associated with the complete DSSC coupled
with the necessary corrections for redox chemistry that occurs at
the counter electrode have thus far precluded data of the quality
presented herein. Nevertheless, about 20 electrons have been
calculated to reside in each TiO2 nanocrystallite at the power
point, albeit a value noted to be quite sensitive to the distribution
of TiO2 particle sizes.46 With reasonable assumptions of extinc-
tion coefficient and path length, the measured TiO2(e−) absorp-
tion at −200 mV is in good agreement with this value, and the
number would be even higher at the open circuit condition.
Thus, the potentiostatic kinetic data provide compelling evidence
that the high short circuit photocurrents for both dyes are due to
ASSOCIATED CONTENT
* Supporting Information
■
S
Synthetic details, physical methods, and spectroscopic, electro-
chemical, and computational data. This material is available free
of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
meyer@jhu.edu; cberling@ucalgary.ca
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
K.H. and G.J.M. acknowledge support by a grant from the
Division of Chemical Sciences, Office of Basic Energy Sciences,
Office of Energy Research, U.S. Department of Energy (DE-
FG02-96ER14662). C.P.B. and K.C.D.R. are grateful to the
Canadian Natural Science and Engineering Research Council,
Canadian Foundation for Innovation, Alberta Ingenuity, Canada
School of Energy and Environment (CSEE); and the Alfred P.
Sloan Foundation for support.
REFERENCES
■
(1) O’Regan, B.; Gratzel, M. Nature 1991, 353, 737−740.
(2) Yella, A.; Lee, H. W.; Tsao, H. N.; Yi, C.; Chandiran, A. K.;
Nazeeruddin, M. K.; Diau, E. W. G.; Yeh, C. Y.; Zakeeruddin, S. M.;
̈
Gratzel, M. Science 2011, 334, 629−634.
̈
(3) Yum, J.-H.; Baranoff, E.; Kessler, F.; Moehl, T.; Ahmad, S.; Bessho,
T.; Marchioro, A.; Ghadiri, E.; Moser, J.-E.; Yi, C.; Nazeeruddin, M. K.;
Gratzel, M. Nat. Commun. 2012, 3, 631−633.
̈
(4) Bomben, P. G.; Gordon, T. J.; Schott, E.; Berlinguette, C. P. Angew.
Chem., Int. Ed. 2011, 50, 10682−10685.
1970
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971

Journal of the American Chemical Society
Article
(5) Feldt, S. M.; Gibson, E. A.; Gabrielsson, E.; Sun, L.; Boschloo, G.;
Hagfeldt, A. J. Am. Chem. Soc. 2010, 132, 16714−16724.
(6) Boschloo, G.; Hagfeldt, A. Acc. Chem. Res. 2009, 42, 1819−1826.
(7) Ardo, S.; Meyer, G. J. Chem. Soc. Rev. 2009, 38, 115−164.
(8) Clifford, J. N.; Martinez-Ferrero, E.; Viterisi, A.; Palomares, E.
Chem. Soc. Rev. 2011, 40, 1635−1646.
(9) Liu, F.; Meyer, G. J. J. Am. Chem. Soc. 2004, 127, 824−825.
(10) O’Regan, B. C.; Walley, K.; Juozapavicius, M.; Anderson, A.;
Matar, F.; Ghaddar, T.; Zakeeruddin, S. M.; Klein, C. d.; Durrant, J. R. J.
Am. Chem. Soc. 2009, 131, 3541−3548.
(11) Planells, M.; Pelleja, L.; Clifford, J. N.; Pastore, M.; De Angelis, F.;
Lopez, N.; Marder, S. R.; Palomares, E. Energy Environ. Sci. 2011, 4,
(40) Rowley, J. G.; Farnum, B. H.; Ardo, S.; Meyer, G. J. J. Phys. Chem.
Lett. 2010, 1, 3132−3140.
(41) Farnum, B. H.; Jou, J. J.; Meyer, G. J. Proc. Natl. Acad. Sci. U.S.A.
2012, 109, 15629−15633.
(42) Clifford, J. N.; Palomares, E.; Nazeeruddin, M. K.; Gratzel, M.;
̈
Durrant, J. R. J. Phys. Chem. C 2007, 111, 6561−6567.
(43) Martiniani, S.; Anderson, A. Y.; Law, C.; O’Regan, B. C.; Barolo,
C. Chem. Commun. 2012, 48, 2406−2408.
(44) Farnum, B. H.; Gardner, J. M.; Meyer, G. J. Inorg. Chem. 2010, 49,
10223−10225.
(45) Bauer, C.; Boschloo, G.; Mukhtar, E.; Hagfeldt, A. J. Phys. Chem. B
2002, 106, 12693−12704.
(46) O’Regan, B. C.; Durrant, J. R. Acc. Chem. Res. 2009, 42, 1799−
1808.
1820−1829.
́
(12) Reynal, A.; Forneli, A.; Martinez-Ferrero, E.; San
́
chez-Dlaz, A.;
(47) Anderson, A. Y.; Barnes, P. R. F.; Durrant, J. R.; O’Regan, B. C. J.
Phys. Chem. C 2010, 114, 1953−1958.
Vidal-Ferran, A. N.; O’Regan, B. C.; Palomares, E. J. Am. Chem. Soc.
2008, 130, 13558−13567.
(13) Hu, K.; Robson, K. C. D.; Johansson, P. G.; Berlinguette, C. P.;
Meyer, G. J. J. Am. Chem. Soc. 2012, 134, 8352−8355.
(14) Li, R.; Lv, X.; Shi, D.; Zhou, D.; Cheng, Y.; Zhang, G.; Wang, P. J.
Phys. Chem. C 2009, 113, 7469−7479.
(15) Mishra, A.; Fischer, M. K. R.; Bauerle, P. Angew. Chem., Int. Ed.
̈
2009, 48, 2474−2499.
(16) Zeng, W.; Cao, Y.; Bai, Y.; Wang, Y.; Shi, Y.; Zhang, M.; Wang, F.;
Pan, C.; Wang, P. Chem. Mater. 2010, 22, 1915−1925.
(17) Hagberg, D. P.; Edvinsson, T.; Marinado, T.; Boschloo, G.;
Hagfeldt, A.; Sun, L. Chem. Commum. 2006, 2245−2247.
(18) Robson, K. C. D.; Koivisto, B. D.; Yella, A.; Sporinova, B.;
Nazeeruddin, M. K.; Baumgartner, T.; Gratzel, M.; Berlinguette, C. P.
Inorg. Chem. 2011, 50, 5494−5508.
(19) Karthikeyan, C. S.; Wietasch, H.; Thelakkat, M. Adv. Mater. 2007,
19, 1091−1095.
(20) Robson, K. C. D.; Koivisto, B. D.; Berlinguette, C. P. Inorg. Chem.
2012, 51, 1501−1507.
(21) Dienes, Y.; Durben, S.; Karpati, T.; Neumann, T.; Englert, U.;
Nyulaszi, L.; Baumgartner, T. Chem.-Eur. J. 2007, 13, 7487−7500.
(22) Bonhote, P.; Dias, A.-P.; Papageorgiou, N.; Kalyanasundaram, K.;
Gratzel, M. Inorg. Chem. 1996, 35, 1168−1178.
(23) Argazzi, R.; Bignozzi, C. A.; Heimer, T. A.; Castellano, F. N.;
Meyer, G. J. Inorg. Chem. 1994, 33, 5741−5749.
(24) Frisch, M. J.; et al. Gaussian 03, revision C.02; Gaussian, Inc.:
Wallingford, CT, 2004.
(25) Ardo, S.; Sun, Y.; Staniszewski, A.; Castellano, F. N.; Meyer, G. J. J.
Am. Chem. Soc. 2010, 132, 6696−6709.
(26) Cappel, U. B.; Feldt, S. M.; Schoneboom, J.; Hagfeldt, A.;
Boschloo, G. J. Am. Chem. Soc. 2010, 132, 9096−9101.
(27) Williams, G.; Watts, D. C. Trans. Faraday Soc. 1970, 66, 3348−
3358.
(28) Lindsey, C. P.; Patterson, G. D. J. Chem. Phys. 1980, 73, 3348−
3357.
(29) Rowley, J. G.; Ardo, S.; Sun, Y.; Castellano, F. N.; Meyer, G. J. J.
Phys. Chem. C 2011, 115, 20316−20325.
(30) Bai, Y.; Zhang, J.; Zhou, D.; Wang, Y.; Zhang, M.; Wang, P. J. Am.
Chem. Soc. 2011, 133, 11442−11445.
(31) Boschloo, G.; Fitzmaurice, D. J. Electrochem. Soc. 2000, 147,
1117−1123.
́
(32) Bisquert, J.; Fabregat-Santiago, F.; Mora-Sero, I.; Garcia-
Belmonte, G.; Barea, E. M.; Palomares, E. Inorg. Chim. Acta 2008,
361, 684−698.
(33) Clifford, J. N.; Palomares, E.; Nazeeruddin, M. K.; Gratzel, M.;
̈
Nelson, J.; Li, X.; Long, N. J.; Durrant, J. R. J. Am. Chem. Soc. 2004, 126,
5225−5233.
(34) Hasselmann, G. M.; Meyer, G. J. J. Phys. Chem. B 1999, 103,
7671−7675.
(35) Haque, S. A.; Tachibana, Y.; Klug, D. R.; Durrant, J. R. J. Phys.
Chem. B 1998, 102, 1745−1749.
(36) Rillema, D. P.; Endicott, J. F. Inorg. Chem. 1972, 11, 2361−2366.
(37) Sutin, N.; Creutz, C. J. Chem. Educ. 1983, 60, 809.
(38) Nord, G. Comments Inorg. Chem. 1992, 13, 221−239.
(39) Wang, X.; Stanbury, D. M. Inorg. Chem. 2006, 45, 3415−3423.
1971
dx.doi.org/10.1021/ja311640f | J. Am. Chem. Soc. 2013, 135, 1961−1971