Ligand Engineering for the Efficient Dye-Sensitized Solar Cells with Ruthenium Sensitizers and Cobalt Electrolytes

Article abstract of DOI:10.1021/acs.inorgchem.6b00842

Over the past 20 years, ruthenium(II)-based dyes have played a pivotal role in turning dye-sensitized solar cells (DSCs) into a mature technology for the third generation of photovoltaics. However, the classic I₃^-/I^- redox couple limits the performance and application of this technique. Simply replacing the iodine-based redox couple by new types like cobalt(3+/2+) complexes was not successful because of the poor compatibility between the ruthenium(II) sensitizer and the cobalt redox species. To address this problem and achieve higher power conversion efficiencies (PCEs), we introduce here six new cyclometalated ruthenium(II)-based dyes developed through ligand engineering. We tested DSCs employing these ruthenium(II) complexes and achieved PCEs of up to 9.4% using cobalt(3+/2+)-based electrolytes, which is the record efficiency to date featuring a ruthenium-based dye. In view of the complicated liquid DSC system, the disagreement found between different characterizations enlightens us about the importance of the sensitizer loading on TiO₂, which is a subtle but equally important factor in the electronic properties of the sensitizers.

Full text of DOI:10.1021/acs.inorgchem.6b00842

Article
pubs.acs.org/IC
Ligand Engineering for the Efficient Dye-Sensitized Solar Cells with
Ruthenium Sensitizers and Cobalt Electrolytes
Sadig Aghazada,† Peng Gao,† Aswani Yella,‡ Gabriele Marotta,§,⊥ Thomas Moehl,‡ Joel Teuscher,∥
̈
Jacques-E. Moser,∥ Filippo De Angelis,§,# Michael Gratzel,‡ and Mohammad Khaja Nazeeruddin*,†
̈
†
́
́ ́
Group for Molecular Engineering of Functional Materials, Ecole Polytechnique Federale de Lausanne (EPFL), CH-1950 Sion,
Switzerland
‡Laboratory of Photonics and Interfaces and Group for Photochemical Dynamics (GDP), EPFL, CH-1015 Lausanne, Switzerland
∥
§Computational Laboratory for Hybrid Organic Photovoltaics, Istituto CNR di Scienze e Tecnologie Molecolari, via Elce di Sotto 8,
I-06123 Perugia, Italy
⊥Dipartimento di Chimica, Biologia e Biotecnologie, Universita
̀
di Perugia, via Elce di Sotto 8, I-06123 Perugia, Italy
#CompuNet, Instituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy
S
* Supporting Information
ABSTRACT: Over the past 20 years, ruthenium(II)-based dyes have
played a pivotal role in turning dye-sensitized solar cells (DSCs) into a
mature technology for the third generation of photovoltaics. However,
the classic I3 /I− redox couple limits the performance and application
−
of this technique. Simply replacing the iodine-based redox couple by
new types like cobalt(3+/2+) complexes was not successful because of
the poor compatibility between the ruthenium(II) sensitizer and the
cobalt redox species. To address this problem and achieve higher
power conversion efficiencies (PCEs), we introduce here six new
cyclometalated ruthenium(II)-based dyes developed through ligand
engineering. We tested DSCs employing these ruthenium(II)
complexes and achieved PCEs of up to 9.4% using cobalt(3+/2+)-
based electrolytes, which is the record efficiency to date featuring a
ruthenium-based dye. In view of the complicated liquid DSC system, the disagreement found between different characterizations
enlightens us about the importance of the sensitizer loading on TiO2, which is a subtle but equally important factor in the
electronic properties of the sensitizers.
complex/iodide system is around 750 mV, which limits the
maximum obtainable conversion efficiency to 13.4%.9,11
INTRODUCTION
■
The global challenge to develop carbon-neutral renewable
energy sources can be addressed by harnessing solar power
using photovoltaics.1 As an alternative to conventional solar
cells, third-generation photovoltaic devices with dye-sensitized
solar cells (DSCs) at the forefront have been extensively
studied.2 In standard DSCs, mesoporous TiO2 is sensitized by a
To boost the efficiencies further, redox couples with a smaller
loss in potential were introduced. Among the one-electron
redox pairs,12,13 cobalt(3+/2+) is the most promising for the
following two reasons: (i) DSCs with organic dyes and cobalt
electrolytes are more stable in comparison to the cells with
other one-electron shuttles;14 (ii) it is possible to obtain cobalt
complexes with various redox potentials just by ligand
modification.15,16 It is worth noting that the recent advances
with copper(2+/1+) phenantroline-based electrolytes may
result in an improved performance.17,18 Contrary to the
ruthenium(II) complex or an organic dye, and I3 /I− is widely
−
employed as the most effective redox couple.3,4 Congruence
between the dye molecule chemisorbed on a mesoporous oxide
and redox pair in the electrolyte in DSCs should be fine-tuned
to obtain fast dye regeneration and ideally slow charge
recombination. The present certified record power conversion
−
reduction of I3 , the unwanted recapture of the conduction-
efficiencies (PCEs) of up to 11.1% were achieved with I3 /I−
−
band electrons by the cobalt(3+) complex is a simple one-
electron outer-sphere redox reaction, which can attain fast rates
depending on the driving force, even though the cobalt(3+/2+)
self-exchange reaction is slow due to spin change. This is
prominent, in particular, for conventional ruthenium dyes such
and heteroleptic ruthenium dyes.5−7 However, regeneration of
a ruthenium(II) complex by electron donation from the I3 /I−
−
redox couple entails a loss of around 500 mV, with over 300
mV of that being directly related to the complicated sequence
of reactions associated with the two-electron oxidation of
iodide that do not involve the sensitizer molecule.8−10 The
estimated lowest potential loss for the ruthenium metal
Received: April 7, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.6b00842
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
as N3, N719, and N749.19−23 This caveat was ascribed to (i)
the net negative charge on NCS-containing complexes causing
columbic attraction between the positively charged cobalt
complex and sensitizer, which results in adsorption of the cobalt
complexes on the semiconductor surface24 and (ii) the
quenching of the triplet metal-to-lignad charge-transfer
(3MLCT) state of ruthenium dyes by cobalt(3+) species,
giving rise to a lower photocurrent.25 To alleviate this
problems, new ruthenium-free dye structures were accom-
modated with long alkyl or alkoxy chains to keep the redox
pairs away from the surface.14,26,27 In this manner, PCEs of over
14% and open-circuit voltage higher than 1 V were achieved by
employing bulky D−π−A dyes with a cobalt-based electro-
lyte.28−31 A similar strategy was introduced for ruthenium dyes
to improve their compatibility with the cobalt-based electro-
lyte.32−34
On the other hand, the bulky nature of the cobalt-based
complexes compels one to use thinner mesoporous films in
order to avoid mass-transport limitations of the photocurrent.
This, in turn, requires the use of dyes with higher extinction
coefficients. Ruthenium(II)-based sensitizers generally suffer
from relatively low extinction coefficients, which motivated us
to design and synthesize new cyclometalated tris-heteroleptic
ruthenium(II) complexes. Through modification of the
auxiliary ligands by attaching polyaromatic moieties, we can
fine-tune the photophysical properties of the complexes,
especially the extinction coefficient, and consequently the
incident photon-to-current conversion efficiency (IPCE) in a
device. It is known that the highest occupied molecular orbital
(HOMO) in a cyclometalated ruthenium(II) complex involves
the auxiliary ligand to a lesser degree than the cyclometalated
ligand.23,32 This provides a wide window for modification of the
auxiliary ligand without destabilization of the HOMO’s energy.
Here, we introduce six new tris-heteroleptic cyclometalated
ruthenium(II) dyes (SA22, SA25, SA246, SA282, SA284, and
SA285; Figure 1). All of these complexes possess the same
anchoring ligand, 4,4′-dicarboxy-2,2′-bipyridine, and the same
cyclometalated ligand as previously optimized.32 2,2′-Bipyr-
idines substituted in the 4 and 4′ positions with polyaromatic
rings were used as auxiliary ligands. With these sensitizers, we
systematically investigated the influence of the auxiliary ligand
by optical and electrochemical measurement, theoretical
calculation, transient absorbance spectroscopy (TAS), and
electrochemical impedance spectroscopy (EIS).
In the presence of cobalt(3+/2+) tris-phenantroline-based
redox electrolytes, the highest PCE came from SA246, which
possesses an insignificant absorption spectrum and the lowest
electron lifetime. These contradictory results motivated us to
unravel the critical parameters among the intricate effect of the
kinds of factors. Through desorption experiments, we show that
a proper substitution on the auxiliary ligands can dramatically
increase the dye loading, which will play a decisive role in
boosting the current density and quasi-Fermi level of TiO2 and
lead to an overtaking efficiency.
RESULTS AND DISCUSSION
■
Synthesis. We have previously reported32 2,6-didodecyloxy-
3,2′-bipyridine as a superior cyclometalating ligand over 2,6-
dimethoxy-3,2′-bipyridine because of its ability to keep the
redox shuttle away from the semiconductor surface.32 The
synthesis of the auxiliary ligands was carried out through
palladium(0)-catalyzed cross-coupling of 4,4′-dibromo-2,2′-
bipyridine with corresponding adducts of polyaromatic
substituents. For the synthesis of tris-heteroleptic cyclo-
metalated ruthenium(II) complexes, we adopted the proce-
dures developed by Bomben et al., which starts from reacting
the cyclometalating ligand with [Ru(C6H6)Cl2]2 or [Ru(p-
cymene)Cl2]2 (Scheme 1).36 After that, the intermediates were
Scheme 1. Synthesis of the Tris-heteroleptic Cyclometalated
Ruthenium(II) Complexes
coordinated with the auxiliary and anchoring ligands in a one-
pot reaction, resulting in three products: two cyclometalated
bis-heteroleptic complexes and one tris-heteroleptic complex.23
It is worth mentioning that from the two possible isomers,
where the cyclometalated ligand is in the trans position to the
auxiliary or anchoring ligand pyridines, the former one is
usually formed, as was shown from the single-crystal X-ray
diffraction structures.23,32 Thus, the new ruthenium(II)
complexes were separated and hydrolyzed to yield the final
Figure 1. Molecular structures of cyclometalated ruthenium(II)
complexes.
1
dye. All intermediates were characterized by H NMR and the
B
DOI: 10.1021/acs.inorgchem.6b00842
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
1
final complexes by H NMR, 13C 135 DEPT NMR, and high-
having more red-shifted bands in the near-UV region. SA25
and SA285 exhibit similar absorption spectra because of the
same cyclopentadithiophenes. However, SA285, with longer
alkyl chains on the cyclopentadithiophene moieties, has a
higher extinction coefficient for the MLCT band than SA25.
Figure S3 presents the cyclic voltammograms of the SA dyes.
From the intersection of normalized absorbance and emittance
(Figure S1), E0−0 was calculated, and the excited-state oxidation
potentials were calculated by subtracting E0−0 from the ground-
state oxidation potentials. Table 1 and Figure 3 summarize the
obtained optical and electrochemical data.
resolution mass spectrometry. The results are provided in the
Supporting Information.
Optical and Electrochemical Properties. Figure 2 shows
the absorption spectra of the SA dyes in dichloromethane
Figure 2. UV−vis absorption spectra of SA22, SA25, SA246, SA282,
SA284, and SA285 in DCM solution.
Figure 3. Energy diagram representing sensitizers’ ground-state (GS)
and excited-state (ES) oxidation potentials and redox couples’ Nernst
potentials with respect to the conduction band of TiO2. The
sensitizer’s GS oxidation potential was determined from cyclic
voltammetry measurements, and the ES oxidation potential was
obtained by subtracting E0−0 from the GS oxidation potential. Redox
couple Nernst potentials are shown considering concentrations in a
Z96040 iodine-based electrolyte and in an optimized cobalt-based
electrolyte.
(DCM). All sensitizers have a set of π−π* transitions in the UV
and near-UV regions. In the visible region, all dyes have
characteristic metal-to-ligand charge-transfer (MLCT) bands.
The extinction coefficient for the MLCT bands around 600 nm
varies from 15 × 103 M−1·cm−1 (for SA22) to 24 × 103 M−1·
cm−1 (for SA285). Except SA22 and SA284, all other dyes have
more intense absorption bands than the sensitizer 3b reported
in ref 28, which has hexylthiophene substitution on the auxiliary
ligand. Moreover, except SA282, in the presented sensitizers,
the low-energy MLCT band is red-shifted by 5−15 nm. SA22
and SA282 have similar shapes of absorption spectra. However,
in the case of SA22 with the indenothiophene-substituted
auxiliary ligand, the MLCT bands are red-shifted compared to
that of SA282 with fluorene moieties on the auxiliary ligands.
Despite the substantial MLCT band shift between SA22 and
SA282 (≈21 nm), the steeper drop of the MLCT band in long
wavelengths in SA22 compared to SA282 results in a similar
E0−0 value (1.77 eV; Table 1). SA246 and SA284 with
thienothiophene and bithiophene moieties, respectively, also
have shapes of absorption spectra similar with that of SA284,
All dyes have oxidation potentials between 0.88 and 0.92 V,
3+/2+
which are higher than the Co(phen)3
standard oxidation
potential (0.62 V) by at least 260 mV. This difference should
create enough driving force for efficient dye regeneration. SA22
and SA282 have similar E0−0 values, which ideally should lead
to identical MLCT band positions. The ground-state oxidation
potential of SA285 (0.89 V) is surprisingly lower than that of
SA25 (0.92 V) by 30 mV. Because of the same E0−0 values, the
difference is maintained in the excited state. However, the error
in cyclic voltammetry measurements up to 0.1 V has to be
considered. All presented sensitizers have the excited-state
oxidation potentials in the range of −0.79 to −0.89 V versus
Table 1. Photophysical and Electrochemical Properties of Dyes
c
λabs/max, nm (ε × 10−3 L·mol−1·cm−1
E0(S+/S), V (vs
λem/max, nm (in solution/on
τ1/2
,
E0−0, eV (in solution/on
E0(S+/S*), V (vs
a
b
a
d
e
dye
)
NHE)
titania)
ns
titania)
NHE)
SA22
592 (15.2), 511 (12.5), 399 (39.2)
593 (19.3), 448 (36.7)
0.97
0.92
0.88
0.88
0.89
0.89
0.86
771/776
808/811
798/814
775/798
803/810
804/822
789/−
46
1.77/1.77
1.71/1.71
1.72/1.71
1.77/1.74
1.71/1.70
1.71/1.67
1.76/−
−0.80
−0.79
−0.84
−0.89
−0.82
−0.82
−0.90
SA25
SA246
SA282
SA284
SA285
3b32
585 (21.5), 431sh (26.4), 382 (36)
571 (22), 508sh (14.2), 411 (28.2)
592 (16.3), 398 (30.4)
38
48
44
82
595 (24.1), 456 (44)
580 (19.6), 507 (14.1), 418 (23.4)
a
b
Absorption and emission spectra were measured in DCM at room temperature. Oxidation potentials were determined from cyclic voltammetry in
0.1 M NBu4PF6 in DCM. The working electrode was glassy carbon, the counter and reference electrodes were platinum wires and ferrocene (Fc)
c
was used as an internal standard. To calculate the potentials versus NHE, 0.7 V was used as the Fc oxidation potential. Excited-state lifetimes in a
d
DCM solution were determined from fluorescence time-correlated single-photon-counting measurements. E0−0 was determined from the
e
intersection of normalized absorption and emission spectra. Excited-state dye oxidation potentials were calculated by subtracting E0−0/q (where q is
a charge and is equal to one electron) from the ground-state oxidation potential.
C
DOI: 10.1021/acs.inorgchem.6b00842
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
NHE, which should guarantee an efficient charge injection into
the titania conduction band.
between the substituents and pyridine rings of the auxiliary
ligand may also play a role (Table S1). Because the dihedral
angles between the indenothiophene or fluorene moieties and
pyridine planes are close values according to the density
functional theory (DFT) calculations (36 and 38°, respectively;
Table S1), the red-shifted spectrum of SA22 (or SA285) with
respect to SA282 is more reasonably due to the higher
donating power of indenothiophene than fluorene. Although
the shapes of theoretical spectra do not perfectly follow the
experimental results (Figure S8), the absorption maxima and
the main calculated transitions (Table S4) are in perfect
agreement with the experimental optical data.
TAS. To evaluate the influence of different substituents on
the lifetimes of the photooxidized dyes in the environment of
cobalt-based electrolytes, we performed TAS measurements on
unbiased devices. Measurements were carried out on two
different types of cells containing (a) redox-inactive pure
acetonitrile and (b) a cobalt-based electrolyte. Dye molecules
were excited with a low-intensity pulsed laser at 510 nm to
ensure an average of less than one injected electron per
nanoparticle, i.e., typically 40 μJ·cm−2. The probe was
monitored at 900 nm, following the oxidized dye signature,
and transient absorbance decay was fitted with a mono-
exponential function (Figure S9).
We find a regeneration yield of above 94% for all dyes, except
SA282 and SA285. A low regeneration yield for SA282 and
SA285 is reasonable considering the steric hindrance of their
hexyl chains, which drastically affects the regeneration lifetime.
We also observe that, for most of the dyes, regeneration does
not seem complete and the dynamics reach a plateau,
suggesting remaining oxidized species in the system. It is
worth noting that, to estimate the regeneration efficiency, we
consider that electron recombination with photooxidized dye
and dye regeneration with the electrolyte rates follow first-order
dynamics on the reductant concentration ηreg = τrec/(τrec + τreg),
where τrec is the electron-oxidized dye recombination lifetime
and τreg the oxidized dye regeneration lifetime in the presence
of an electrolyte; Table S6]. This procedure may not exactly
represent the situation in a device under full sun illumination
considering the following facts: (a) dye regeneration is not
necessarily a first-order reaction42 and (b) the electron density
in TiO2 created by the laser pulse is not comparable to the
electron density in the performing device at maximum power
point.43,44 The second fact may bring an overestimated lifetime
of photooxidized dyes in the devices with a redox-inactive
electrolyte at a low-light regime. Although not exactly picturing
devices in working conditions, we use these data as an
approximation of the charge-transfer dynamics.45
One of the desired properties of a DSC sensitizer is its
capability of sustaining numerous oxidation−reduction cycles
under long-term operation, e.g., turnover numbers reaching 100
million for an outdoor lifetime of 20 years. Isothiocyanate
ligands employed often for ruthenium(II) complexes cannot
survive because of ligand exchange by electrolyte components
such as tert-butylpyridine.41
In this regard, cyclometalated ruthenium(II) complexes were
introduced as being potentially more stable, by ruling out the
possibility of ligand exchange. However, on the basis of cyclic
voltammetry measurements, our ruthenium(II) complexes are
quasi-reversible because the obtained shifts between the peaks
of the oxidative and reductive waves are usually around 120
mV.5,23,28,35−39
To understand the origin of the electrochemical irrever-
sibility in cyclometalated ruthenium dyes, we conducted
spectroelectrochemical measurements in solution and in the
complete device without a redox pair. We focused on the bands
above 300 nm. Typical examples are represented in Figure S4.
In solution (Figure S4A), none of these six complexes showed
complete reversibility of both the π−π* and MLCT bands.
Among them, SA22 and SA246 showed the best reversibility.
However, these results should not be discouraging because dye
features on titania may strongly change. In complete devices,
SA22, SA282, and SA284 showed reversible behavior only for
the π−π* transitions. Conversely, SA25, SA246, and SA285
have reversible MLCT bands in a device, but the π−π*
transitions around 450 nm become extinct, indicating that the
coordination core is more stable than the polyaromatic
substituents (Figure S4C).
Computational Analysis. In Figure S5, the optimized
geometries of studied SA dyes are shown. All of them have the
typical geometry of ruthenium tris-bipyridine complexes. These
optimizations were carried out for the isomers, which have a
cyclometalated pyridine ring in the position trans to one of the
anchoring carboxypyridine rings. It is worth noting that
calculated molecular volumes meet expectations, and SA246
and SA284 with the least number of substituents have the
lowest molecular volumes (Table S1).
Among the entire series of SA dyes, the lowest unoccupied
molecular orbitals (LUMOs) have very close energies (Table
S2) and are localized on the anchoring ligand, as can be
observed from the isodensity plots in Figures S6 and S7.
However, the HOMOs present variation in the energies mainly
because of localization on the different substituents. In
particular, SA25 and SA285 exhibit the highest HOMO
energies with a consequent decrease of the HOMO−LUMO
energy gap. As expected, the HOMOs not only are localized on
the metal t2g orbitals but also are partially delocalized through
the auxiliary ligand, the degree of which depends on the
different substituents. In SA25 and SA285, both cyclo-
pentadithiophene moieties on the auxiliary ligands contribute
to a major fraction of the HOMO. In SA22, SA246, and
SA284, only one donating moiety on the auxiliary ligands,
which is in the trans position to the anchoring carboxypyridine
ring, owns a fraction of the HOMO. Fluorene moieties on the
auxiliary ligand in SA282 do not possess any substantial
fraction of the HOMO (Figure S7). These observations imply
that, among all of these substituents on the auxiliary ligands,
cyclopentadithiophene and fluorene own the strongest and
weakest donating power, considering that dihedral angles
EIS. The VOC trends of the devices with various dyes can be
predicted by EIS analysis on complete devices in the dark. The
Nyquist plots were fitted according to the transmission-line
model developed by Bisquert et al., and the main parameters
were extracted.46,47 They comprise the charge-transfer resist-
ance, Rct, representing the charge recombination resistance for
the electron in the TiO2 conduction band with the oxidized
form of the redox couple, the chemical capacitance, Cμ,
representing the density of states (DOS) accessible to electrons
in the TiO2 nanocrystals, and the transport resistance, Rtrans
,
representing the resistance for the transport of the electrons
through the mesoporous TiO2 network. Using the obtained
charge recombination (Rct) and transport (Rtrans) resistances
along with the chemical capacitance (Cchem) of the titania, one
D
DOI: 10.1021/acs.inorgchem.6b00842
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 4. Key parameters extracted from EIS analyses of the cobalt-based devices: recombination resistance (solid lines); transport resistance
(dotted lines) and capacitance (dashed lines) over potential (left); electron recombination lifetime (solid lines) and transport lifetime (dotted lines)
as a function of the capacitance (right). In all figures, lines with red, blue, green, black, yellow, and cyan colors refer to the devices with SA22, SA25,
SA246, SA282, SA284, and SA285, respectively.
can calculate the electron lifetime (τn) and transport time
(τtrans) as τn = RctCchem and τtrans = RtransCchem
sensitizer called SA246 showed the best IPCE, which is in
agreement with the measured JSC value. Inconsistencies are
found in both JSC and VOC, especially with SA246, which has
relatively blue-shifted absorption spectrum and the lowest
electron lifetime. An unexpected trend in the VOC and JSC
values indicating some more vital parameters is controlling the
photovoltaic performance.
.
The main parameters from the EIS fitting of cobalt-based
devices are shown in Figures 4 and S10. The cobalt-based
devices showed pronounced changes in the conduction-band
edge position. It is interesting to note that the highest chemical
capacitance, and so the lowest conduction band, and the
highest electron lifetimes were observed for SA22, followed by
SA285, while the lowest values of the electron lifetime were
obtained for SA246 and SA282.
Despite the indistinctive molar extinction coefficient and
moderate absorption range with SA246, the high JSC over other
sensitizers is a baffling feature. To explain the highest JSC
provided by SA246, we analyzed the amount of dye adsorbed
on the surface. One way to evaluate the absolute dye loading is
the desorption experiment of sensitized titania films (Figure 6).
The amount of SA246 adsorbed on titania is drastically higher
compared to other dyes. It is worth noting that SA285 also
provides high dye loading; however, because of inefficient dye
regeneration (vide supra), SA285 fails to yield high PCE. The
high dye loading of SA246 increased the absorbance (optical
density) of the sensitized titania, which, in turn, boosted JSC
more than the other sensitizers with a higher molecular molar
extinction coefficient and a wider absorption range. This effect
has already been indicated by the IPCE spectra (Figure 5B).
At the same time, it is quite unexpected that the device with
SA246 gave the highest VOC even though it possesses the
lowest electron lifetime and therefore the highest recombina-
tion. It is known that the long electron lifetime is a necessary
but not sufficient condition to lead to high VOC. We have to
take the change in JSC under light into consideration, which has
been completely neglected in ESI analyses; e.g., the inversed
correlation of the electron lifetime and VOC in the case of SA22
and SA246 can be explained by the higher JSC in the case of
SA246. Higher JSC will induce a higher steady-state electron
density in titania in the performing device, which yields an
upwardly shifted quasi-Fermi level E*F,n and, hence, VOC. In
this case, the rise in VOC compensates for the loss due to high
electron recombination.
The differences in VOC due to the conduction-band shift and
electron lifetime were estimated in reference to the values
obtained with SA22 and are presented in Table 2.
Table 2. Comparison of the VOC Differences Estimated from
EIS Analyses and Obtained from the J−V Measurements for
the Cobalt-Based Electrolytes
b
τn, s (ΔV due to a
estimated ΔVOC =
a
d
ΔECB
,
change in τn with
respect to SA22, mV)
ΔECB + ΔV (due ΔVOC
,
c
dye
mV
to Δτn)
mV
SA22
0
24
40
22
23
15
0.180
0
SA25
0.052 (−32)
0.024 (−52)
0.025 (−51)
0.035 (−43)
0.120 (−11)
−8
−12
−27
−20
4
−17
18
SA246
SA282
SA284
SA285
−33
−33
−20
a
The shift in the conduction-band edge was estimated from EIS
b
analyses. The change in the voltage due to a change in the electron
lifetime was calculated using the diode equation. Total estimated
c
change in VOC due to the conduction-band edge difference and
d
electron lifetime. The difference in VOC obtained from the J−V
measurements. All values are brought in reference to the case with
SA22. Positive values indicate the rise in VOC
.
Photovoltaic Performance. Finally, to evaluate the
influence of ligands on the photovoltaic performance, we put
the complete devices with the cobalt-based electrolytes under
photovoltaic characterization and present the results in Table 3.
The iodine-based devices were also tested, and the results are
shown in Figure S11B and Table S7. In the case of the DSCs
with cobalt-based electrolytes, both JSC and VOC vary
significantly between 9 and 14 mA·cm−2 and 794 and 845
mV with the following trends: for JSC, SA282 < SA25 < SA284
< SA285 < SA22 < SA246; for VOC, SA282 = SA284 < SA285
< SA25 < SA22 < SA246 (Figure 5 and Table 3). The
CONCLUSION
■
Six new cyclometalated tris-heteroleptic ruthenium complexes
were synthesized and characterized in the course of ligand
engineering compatible with cobalt-based electrolytes in DSCs.
The substituents on the auxiliary ligand were the focus of this
study. Various substituents result in different photophysical
properties and different performances in the DSCs. With the
cobalt-based devices, the record efficiency to date featuring
E
DOI: 10.1021/acs.inorgchem.6b00842
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Table 3. Photovoltaic Performance of Iodine- and Cobalt-Based DSCs with SA Dyes
a
dye
electrolyte
JSC, mA/cm2
VOC, mV
FF, %
PCE(10%)
PCE(100%)
b
b
b
b
b
c
SA22
[Co(phen)3]3+/2+
[Co(phen)3]3+/2+
[Co(phen)3]3+/2+
[Co(phen)3]3+/2+
[Co(phen)3]3+/2+
12.25
10.68
14.55
9.89
827
810
845
794
794
807
75.5
77.9
74.7
78.5
76.9
73.6
8.4
7.4
9.4
6.7
7.2
6.1
7.9
6.9
9.4
6.3
7.0
7.2
SA25
SA246
SA282
SA284
SA285
11.28
[Co(phen)3]3+/2+
11.85
a
b
All cells were measured under AM 1.5 simulated solar light irradiation with a power of 100 mW·cm−2 at room temperature; Cobalt-based
c
electrolyte: 0.25 M [CoII(phen)3](TFSI)2, 0.05 M [CoIII(phen)3](TFSI)3, 0.25 M 4-(5-nonyl)pyridine, and 0.1 M LiTFSI; Cobalt-based
electrolyte: 0.25 M [CoII(phen)3](TFSI)2, 0.05 M [CoIII(phen)3](TFSI)3, 0.5 M 4-(5-nonyl)pyridine, and 0.1 M LiTFSI.
Figure 5. (A) J−V curves and (B) IPCE curves for DSCs based on cobalt electrolytes and cyclometalated ruthenium dyes SA22, SA25, SA246,
SA282, SA284, and SA285.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
chem.6b00842.
Synthetic procedures and characterization of all products,
descriptive preparation of solar cells, photoluminescence
spectra and excited-state decay for the final complexes,
cyclic voltammograms for the final complexes, and
DFT−time-dependent DFT calculation parameters and
results (PDF)
Figure 6. Dye-loading values obtained by the desorption of SA dyes
from the titania films.
AUTHOR INFORMATION
Corresponding Author
*E-mail: mdkhaja.nazeeruddin@epfl.ch.
■
ruthenium-based dyes was obtained for SA246 as 9.4%. The
highest efficiency is explained by the higher dye loading, a
factor that is normally neglected by the designers. The high
absolute dye loading increases the optical density, which can
lead to a higher photocurrent and voltage. This effect can even
overwhelm the influence of the recombination rate. Provided in
this work, different analyses indicate the complexity of the
processes taking place and underline the compromise between
various parameters to be considered from the sensitizer-design
viewpoint. For the cobalt-based devices, the sensitizer molecule
should have small enough size to maintain (i) efficient dye
regeneration and (ii) high dye loading and, at the same time,
(iii) a bulky periphery to keep the redox mediator away from
the semiconductor surface. These results could be very
instructive for further dye engineering to reach even higher
PCEs.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
S.A. thanks the European Community’s Seventh Framework
Programme under Grants 308997 (“NanoMatCell”) and
309194 (“GLOBASOL”).
REFERENCES
■
(1) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U. S. A. 2006,
103, 15729−15735.
(2) O’Regan, B.; Gratzel, M. Nature 1991, 353, 737−740.
̈
(3) Nazeeruddin, M. K.; Kay, A.; Rodicio, I.; Humphry-Baker, R.;
Muller, E.; Liska, P.; Vlachopoulos, N.; Gratzel, M. J. Am. Chem. Soc.
̈
̈
1993, 115, 6382−6390.
(4) Mishra, A.; Fischer, M. K. R.; Bauerle, P. Angew. Chem., Int. Ed.
̈
2009, 48, 2474−2499.
F
DOI: 10.1021/acs.inorgchem.6b00842
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(5) Nazeeruddin, M. K.; Pec
Humphry-Baker, R.; Comte, P.; Liska, P.; Cevey, L.; Costa, E.;
Shklover, V.; Spiccia, L.; Deacon, G. B.; Bignozzi, C. A.; Gratzel, M. J.
́
hy, P.; Renouard, T.; Zakeeruddin, S. M.;
(34) Wu, K. L.; Hu, Y.; Chao, C. T.; Yang, Y. W.; Hsiao, T. Y.;
Robertson, N.; Chi, Y. J. Mater. Chem. A 2014, 2, 19556−19565.
(35) Bessho, T.; Yoneda, E.; Yum, J.-H.; Guglielmi, M.; Tavernelli, I.;
̈
Imai, H.; Rothlisberger, U.; Nazeeruddin, M. K.; Gratzel, M. J. Am.
̈
Am. Chem. Soc. 2001, 123, 1613−1624.
Chem. Soc. 2009, 131, 5930−5934.
(6) Yu, Q.; Wang, Y.; Yi, Z.; Zu, N.; Zhang, J.; Zhang, M.; Wang, P.
ACS Nano 2010, 4, 6032−6038.
(36) Bomben, P. G.; Koivisto, B. D.; Berlinguette, C. P. Inorg. Chem.
2010, 49, 4960−4971.
(7) Chen, C. Y.; Wang, M.; Li, J. Y.; Pootrakulchote, N.; Alibabaei,
(37) Bomben, P. G.; Borau-Garcia, J.; Berlinguette, C. P. Chem.
Commun. 2012, 48, 5599−5601.
L.; Ngoc-le, C.; Decoppet, J. D.; Tsai, J. H.; Gratzel, C.; Wu, C. G.;
̈
Zakeeruddin, S. M.; Gratzel, M. ACS Nano 2009, 3, 3103−3109.
̈
(38) Martiniani, S.; Anderson, A. Y.; Law, C.; O’Regan, B.; Barolo, C.
Chem. Commun. 2012, 48, 2406−2408.
(8) Ardo, S.; Meyer, G. J. Chem. Soc. Rev. 2009, 38, 115−164.
(9) Boschloo, G.; Hagfeldt, A. Acc. Chem. Res. 2009, 42, 1819−1826.
(10) Peter, L. M. Phys. Chem. Chem. Phys. 2007, 9, 2630−2642.
(11) Snaith, H. J. Adv. Funct. Mater. 2010, 20, 13−19.
(39) Anderson, A.; Barnes, P.; Durrant, J.; O’Regan, B. J. Phys. Chem.
C 2011, 115, 2439−2447.
(40) Barnes, P.; Anderson, A.; Durrant, J.; O’Regan, B. Phys. Chem.
Chem. Phys. 2011, 13 (13), 5798−5816.
(12) Zhang, Z.; Chen, P.; Murakami, T. N.; Zakeeruddin, S. M.;
Gratzel, M. Adv. Funct. Mater. 2008, 18, 341−346.
̈
(41) Jennings, J. R.; Liu, Y.; Wang, Q. J. Phys. Chem. C 2011, 115,
15109−15120.
(13) Daeneke, T.; Kwon, T. H.; Holmes, A. B.; Duffy, N. W.; Bach,
U.; Spiccia, L. Nat. Chem. 2011, 3, 213−217.
(42) Martiniani, S.; Anderson, A. Y.; Law, C.; O’Regan, B.; Barolo, C.
Chem. Commun. 2012, 48, 2406−2408.
(14) Feldt, S. M.; Gibson, E. A.; Gabrielsson, E.; Sun, L.; Boschloo,
G.; Hagfeldt, A. J. Am. Chem. Soc. 2010, 132, 16714−16724.
(15) Feldt, S. M.; Wang, G.; Boschloo, G.; Hagfeldt, A. J. Phys. Chem.
C 2011, 115, 21500−21507.
(43) Anderson, A.; Barnes, P.; Durrant, J.; O’Regan, B. J. Phys. Chem.
C 2011, 115, 2439−2447.
(44) Jennings, J. R.; Liu, Y.; Wang, Q. J. Phys. Chem. C 2011, 115,
15109−15120.
(16) Giribabu, L.; Bolligarla, R.; Panigrahi, M. Chem. Rec. 2015, 15,
760−788.
(17) Colombo, A.; Dragonetti, C.; Magni, M.; Roberto, D.;
Demartin, F.; Caramori, S.; Bignozzi, C. A. ACS Appl. Mater. Interfaces
2014, 6, 13945−13955.
(18) Magni, M.; Giannuzi, R.; Colombo, A.; Cipolla, M. P.;
Dragonetti, C.; caramori, S.; Carli, S.; Grisorio, R.; Suranna, G. P.;
Bignozzi, C. A.; Roberto, D.; Manca, M. Inorg. Chem. 2016, 55, 5245−
5253, DOI: 10.1021/acs.inorgchem.6b00204.
(45) Li, F.; Jennings, J. R.; Wang, Q. ACS Nano 2013, 7, 8233−8242.
(46) Fabregat-Santiago, F.; Bisquert, J.; Garcia-Belmonte, G.;
Boschloo, G.; Hagfeldt, A. Sol. Energy Mater. Sol. Cells 2005, 87,
117−131.
(47) Fabregat-Santiago, F.; Garcia-Belmonte, G.; Mora-Sero, I.;
Bisquert, J. Phys. Chem. Chem. Phys. 2011, 13, 9083−9118.
(19) Nusbaumer, H.; Moser, J.-E.; Zakeeruddin, S. M.; Nazeeruddin,
M. K.; Gratzel, M. J. Phys. Chem. B 2001, 105, 10461−10464.
̈
(20) Sapp, S. A.; Elliott, C. M.; Contado, C.; Caramori, S.; Bignozzi,
C. A. J. Am. Chem. Soc. 2002, 124, 11215−11222.
(21) Klahr, B. M.; Hamann, T. W. J. Phys. Chem. C 2009, 113,
14040−14045.
(22) Liu, Y.; Jennings, J. R.; Huang, Y.; Wang, Q.; Zakeeruddin, S.
M.; Gratzel, M. J. Phys. Chem. C 2011, 115, 18847−18855.
̈
(23) Bomben, P. G.; Gordon, T. J.; Schott, E.; Berlinguette, C. P.
Angew. Chem., Int. Ed. 2011, 50, 10682−10685.
(24) Mosconi, E.; Yum, J. H.; Kessler, F.; Gom
́
ez García, C. J.;
Zuccaccia, C.; Cinti, A.; Nazeeruddin, M. K.; Gratzel, M.; De Angelis,
̈
F. J. Am. Chem. Soc. 2012, 134, 19438−19453.
(25) Omata, K.; Kuwahara, S.; Katayama, K.; Qing, S.; Toyoda, T.;
Lee, K.-M.; Wu, C.-G. Phys. Chem. Chem. Phys. 2015, 17, 10170−
10175.
(26) Tsao, H. N.; Yi, C.; Moehl, T.; Yum, J.-H. H.; Zakeeruddin, S.
M.; Nazeeruddin, M. K.; Gratzel, M. ChemSusChem 2011, 4, 591−594.
̈
(27) 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.
̈
(28) Mathew, S.; Yella, A.; Gao, P.; Humphry-Baker, R.; Curchod, B.
F.; Ashari-Astani, N.; Tavernelli, I.; Rothlisberger, U.; Nazeeruddin, M.
K.; Gratzel, M. Nat. Chem. 2014, 6, 242−247.
̈
(29) Yella, A.; Mai, C.-L. L.; Zakeeruddin, S. M.; Chang, S.-N. N.;
Hsieh, C.-H. H.; Yeh, C.-Y. Y.; Gratzel, M. Angew. Chem., Int. Ed. 2014,
̈
53, 2973−2977.
(30) Yum, J.-H. H.; Baranoff, E.; Kessler, F.; Moehl, T.; Ahmad, S.;
Bessho, T.; Marchioro, A.; Ghadiri, E.; Moser, J.-E. E.; Yi, C.;
Nazeeruddin, M. K.; Gratzel, M. Nat. Commun. 2012, 3, 631.
̈
(31) Kakiage, k.; Aoyama, Y.; Yano, T.; Oya, K.; Fujisawa, J.-I.;
Hanaya, M. Chem. Commun. 2015, 51, 15894−15897.
(32) Polander, L. E.; Yella, A.; Curchod, B. F.; Ashari-Astani, N.;
Teuscher, J.; Scopelliti, R.; Gao, P.; Mathew, S.; Moser, J.-E. E.;
Tavernelli, I.; Rothlisberger, U.; Gratzel, M.; Nazeeruddin, M. K.; Frey,
̈
J. Angew. Chem., Int. Ed. 2013, 52, 8731−8735.
(33) Wu, K. L.; Clifford, J. N.; Wang, S. W.; Aswani, Y.; Palomares,
E.; Lobello, M. G.; Mosconi, E.; De Angelis, F.; Ku, W. P.; Chi, Y.;
Nazeeruddin, M. K.; Gratzel, M. ChemSusChem 2014, 7, 2930−2938.
̈
G
DOI: 10.1021/acs.inorgchem.6b00842
Inorg. Chem. XXXX, XXX, XXX−XXX