Correlating cobalt redox couples to photovoltage in the dye-sensitized solar cell

Article abstract of DOI:10.1039/c8dt01921e

We report a series of structurally analogous cobalt mediators related to [Co-bpy]^Z (bpy = 2,2′-bipyrimidine, Z = 2+ or 3+) to demonstrate a linear relationship between the redox potential of the Co(iii/ii)-based redox couple (E_med) and open-circuit voltage (V_OC) of the DSSC. The E_med values vary from 0.42 to 1.07 V vs. NHE depending on the number of nitrogen atoms and the presence of tert-butyl substituents on the ligand. A 64-mV gain in photovoltage was calculated for every +100 mV shift in E_med. Differences in the mediator sizes, diffusion coefficients, light absorption profiles, and spin state configurations for the complexes were not significant and therefore not expected to contribute to changes in the V_OC. A decrease in the photocurrent, downward shift in quasi-Fermi level (E_F,n) and shorter electron lifetime (T_n) with increasingly positive E_med were instead attributed to enhanced electron recombination from the TiO₂ film to oxidized mediator species in the electrolyte.

Full text of DOI:10.1039/c8dt01921e

Dalton
Transactions
View Article Online
View Journal
PAPER
Correlating cobalt redox couples to photovoltage
in the dye-sensitized solar cell†
Cite this: DOI: 10.1039/c8dt01921e
a,b,c
Kitty Y. Chen,a Phil A. Schauer,a Brian O. Patricka and Curtis P. Berlinguette
*
We report a series of structurally analogous cobalt mediators related to [Co-bpy]Z (bpy = 2,2’-bipyrimi-
dine, Z = 2+ or 3+) to demonstrate a linear relationship between the redox potential of the Co(III/II)-
based redox couple (Emed) and open-circuit voltage (VOC) of the DSSC. The Emed values vary from
0.42 to 1.07 V vs. NHE depending on the number of nitrogen atoms and the presence of tert-butyl
substituents on the ligand. A 64-mV gain in photovoltage was calculated for every +100 mV shift in
Emed. Differences in the mediator sizes, diffusion coefficients, light absorption profiles, and spin state
configurations for the complexes were not significant and therefore not expected to contribute to
changes in the VOC. A decrease in the photocurrent, downward shift in quasi-Fermi level (EF,n) and
shorter electron lifetime (Tn) with increasingly positive Emed were instead attributed to enhanced elec-
tron recombination from the TiO2 film to oxidized mediator species in the electrolyte.
Received 12th May 2018,
Accepted 25th June 2018
DOI: 10.1039/c8dt01921e
rsc.li/dalton
the TiO2 conduction band is typically enhanced with increas-
ingly positive Emed due to a greater thermodynamic driving
Introduction
The dye-sensitized solar cell (DSSC) converts light into electri- force for the reaction.8,9 Similarly, the improved overlap
cal energy through photo-excitation of a TiO2 anode film sensi- between the electronic density of states for TiO2 and oxidized
tized with molecular dye molecules.1 An electrolyte redox mediator species increases with rate of recombination from
mediator subsequently transports charge from the cathode to surface states at higher Emed values.10,11 A consequence of the
the photo-oxidized dye to complete the circuit. The open enhanced recombination is a decrease in the short-circuit
circuit voltage (VOC) of the cell is defined by the energy differ- current density (JSC) as more photo-injected electrons are
ence between the quasi-Fermi level of TiO2 film (EF,n) and the intercepted by oxidized mediators.12 As a result of these
relevant redox potential of the mediator (Emed).2 Because the recombination mechanisms VOC rarely scales linearly with
power conversion efficiency (PCE) of the photovoltaic cell is Emed (i.e., a 200 mV increase in Emed corresponds to a less than
proportional to VOC, it follows that affecting a positive shift of 200 mV increase in VOC). The overall decrease in VOC caused by
the Emed can increase overall cell performance.
interfacial recombination has been linked to a positive shift in
Redox shuttles with very positively shifted Emed values, the EF,n due to a concomitant increase in the dark current and
however, often demonstrate low power conversion decrease in TiO2 electron density under illumination.12,13
efficiencies.3–7 This effect can be attributed to deleterious elec-
A number of studies have applied Marcus theory to explain
tron recombination occurring from either the conduction how the driving force of recombination affects recombination
band or surface states of the TiO2 film and the rates of both kinetics and device performance.8,9,14 Polypyridyl cobalt(III/II)
these processes are a function of Emed. Recombination from redox mediators are extensively utilized in these comparative
studies because they are straightforward to synthesize with a
15,16
variety of ligand combinations that directly influence Emed
.
The challenge of these studies lies in comprehensively addres-
sing the additional influences of polypyridyl ligand variance,
specifically how the physical and magnetic properties of the
mediator impact the recombination reaction. The volume and
geometry of the mediator affect both mass-transport coeffi-
cients and the interfacial coupling factor which are critical to
the recombination rate.14,17–19 Mediator spin state also affects
the recombination rate because spin change during the
aDepartment of Chemistry, University of British Columbia, 2036 Main Mall,
Vancouver, British Columbia, V6T 1Z1, Canada. E-mail: cberling@chem.ubc.ca
bDepartment of Chemical & Biological Engineering, University of British Columbia,
2360 East Mall, Vancouver, British Columbia V6T, Canada
cStewart Blusson Quantum Matter Institute, The University of British Columbia,
2355 East Mall, Vancouver, BC V6T 1Z4, Canada
†Electronic supplementary information (ESI) available. CCDC 1842553. For ESI
and crystallographic data in CIF or other electronic format see DOI: 10.1039/
c8dt01921e
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Results
Synthesis and characterization
Cobalt complexes were synthesized by dissolving separately
CoCl2 and an excess of the relevant ligand in methanol, com-
bining the two solutions, and refluxing for two hours.
Dropwise addition of the CoCl2 solution to the ligand solution
was performed in the synthesis of [Co-bpm]2+ and
[Co-(bpm-DTB)]2+ to prevent potential bridging coordination
of the bipyrimidine ligands. Potassium hexafluorophosphate
or lithium bis(trifluoromethane)sulfonimide salt was added to
the solution in order to precipitate the target Co(II) complexes.
Excess ligand and salts remain dissolved in solution for facile
product separation by filtration, although [Co-(bpm-DTB)]
(PF6)2 required a modified purification procedure due to high
solubility in methanol. The Co(III) forms of the complexes were
generated upon reacting the Co(II) complexes with nitrosonium
hexafluorophosphate (NOPF6).
All Co(II) and Co(III) complexes were characterized by
1H NMR spectroscopy. Attempts to confirm the mass-to-charge
ratio of the cationic complexes using high-resolution mass
spectrometry were unsuccessful. Even in the case of electro-
spray ionization it was only possible to infer fragmented
ligand ions from the obtained spectra. The structure of the
newly-reported complexes [Co-(bpm-DTB)](PF6)2 and [Co-bpm]
(TFSI)2 were confirmed through single-crystal X-ray diffracto-
metry and elemental analysis respectively.
Fig. 1 Chemical structure of tris(2,2’-bipyridine)cobalt [Co-bpy]Z, tris
(2,2’-bipyrimidine)cobalt [Co-bpm]Z, tris(4,4’-tert-butyl-2,2’-bipyridine)
cobalt [Co-(bpy-DTB)]Z, and a novel compound tris(4,4’-tert-butyl-2,2’-
bipyrimidine)cobalt [Co-(bpm-DTB)]Z where Z = 2+ or 3+ charge.
reduction of low-spin Co(III) to high-spin Co(II) requires a
larger inner-sphere reorganization energy and acts as a barrier
for recombination compared to the reduction of low-spin
Co(III) to low-spin Co(II).20–22
X-ray crystallography
We report here the comparison of two sets of structurally
analogous tris(2,2′-diimine)cobalt complexes (Fig. 1) to
examine the direct effect of Emed on VOC and electron recombi-
nation in the DSSC while holding the sizes, absorption pro-
files, and spin states at near parity. The Emed of these cobalt
complexes range from 0.42 V to 1.07 V vs. NHE. The four com-
plexes only differ by the number of nitrogen atoms in the aro-
matic system and the presence of tert-butyl groups. [Co-bpy]Z,
[Co-(bpy-DTB)]Z, and [Co-(bpm-DTB)]Z were balanced by hexa-
fluorophosphate counterions (PF6−), whereas [Co-bpm]Z was
balanced by bis(trifluoromethane)sulfonimide counterions
(TFSI−) in consideration of solubility issues.4 A linear relation-
ship between Emed and VOC was observed when testing each of
the four complexes into functional DSSCs. The size difference
between complexes with and without tert-butyl groups did not
affect the overall trend in VOC, but did appear to influence the
accessibility of surface state-mediated recombination accord-
ing to open-circuit decay measurements.
The two pairs of analogous cobalt complexes exhibit near-iden-
tical structures according to single crystal X-ray diffraction
data. Crystal structure parameters for [Co-(bpy-DTB)](PF6)2,23
[Co-bpy](PF6)2,24 and [Co-bpm](PF6)2 were obtained from the
4
Cambridge Structural Database.25 The crystallographic and
refinement parameters for [Co-bpm-DTB](PF6)2 are shown in
Fig. S1, Tables S1 and S2.† The Co–N bond lengths and N–Co–
N angles are summarized in Table 1 (ϕ represents the N–Co–N
angle of the same ligand and θ is the N–Co–N angle between
adjacent ligands). All four cobalt complexes exhibit distorted
octahedral geometries with comparable metal–ligand bond
lengths (ca. 2.1 Å) and angles (ϕ = 76–77°; θ = 90–99°)
(Table 1). The estimated molecular radii of the cobalt com-
plexes with tert-butyl groups were 2.2 Å larger than those
without the group, but no significant difference in radii was
observed between the respective bipyridine and bipyrimidine
complexes (Table 2).
Table 1 Range of Co–N bond lengths and N–Co–N bond angles of [Co-(bpy-DTB)](PF6)2, [Co-bpy](PF6)2, [Co-(bpm-DTB)](PF6)2, and [Co-bpm]
(PF6)2 determined by X-ray crystallography
Parameters
[Co-(bpy-DTB)](PF6)2
[Co-bpy](PF6)2
[Co-(bpm-DTB)](PF6)2
[Co-bpm](PF6)2
Co–N length (Å)
ϕa (°)
2.095–2.133
77.22–77.43
93.16–99.37
2.117(4)–2.146(4)
77.02(15)–77.66(14)
90.63(15)–96.86(15)
2.126(3)–2.158(3)
76.46(10)–76.49(10)
92.00(10)–99.64(10)
2.1210(18)–2.1494(18)
76.46(7)–76.86(7)
90.42(7)–98.29(7)
θb (°)
a ϕ = N–Co–N angle of the same ligand. b θ = N–Co–N angle between adjacent ligands. Refer to ref. 4, 23, 24 and ESI for more details.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
Table 2 The molecular radii, effective magnetic moments (µeff) and diffusion coefficients (DCo(II)) of Co(II) complexes, and redox potentials of the
Co(III/II) couples (Emed) in MeCN
Molecular radiia
Å
µeff
Emed
DCo(II)
b
c
Compound
Bohr magneton
V vs. NHE
×10−5 cm2 s−1
[Co-(bpy-DTB)]Z
[Co-bpy]Z
8.0
5.8
8.0
5.8
—
4.42
4.72
4.62
4.86
—
0.42
0.57
0.89
1.07
0.64
0.83e
1.21d
—
[Co-(bpm-DTB)]Z
[Co-bpm]Z
FcH
1.20d
2.48e
a Estimated from crystallographic structures by averaging the distance between the Co to the center of the most peripheral H atoms on each
ligand. b Cyclic voltammograms measured in 0.1 M n-NBu4PF6 electrolyte at 100 mV s−1 and internally referenced against FcH/FcH+ (+0.64 V vs.
NHE). c Obtained from linear sweep voltammograms measured in 0.1 M n-NBu4PF6 electrolyte at 100 mV s−1 and referenced against Ag/AgCl in
saturated KCl solution or Pt wire. d GC working electrode. e Au working electrode.
Magnetic susceptibility
Magnetic susceptibilities were assessed using the Evans
method26 and results indicate that all four cobalt complexes
exist in a high-spin d7 Co(II) configuration (Table 2). The
effective magnetic moments (µeff) fell between 4.4–4.8 Bohr
magnetons which is the characteristic range for paramagnetic
high-spin Co(II).27 The spin-state assignment is further vali-
dated by the Co–N bond lengths characteristic of high-spin
complexes extracted from crystallographic analysis (∼2.1 Å)23
and the wide chemical shift ranges of 0–105 ppm observed in
1H NMR spectroscopy (Fig. S2†). All four complexes exist in a
low-spin d6 Co(III) state as indicated by proton resonances
located between 7–10 ppm with clearly identifiable JH,H coup-
1
ling constants in H NMR spectroscopy. These features corres-
pond to diamagnetic Co(III) species (Fig. S2†).
UV-Vis absorption spectroscopy
Fig. 2 UV-Vis absorption spectra of [Co-(bpy-DTB)](PF6)2 (blue), [Co-
bpy](PF6)2 (grey), [Co-(bpm-DTB)](PF6)2 (red), and [Co-bpm](PF6)2
(orange) in MeCN at ambient temperature. Inset shows enlarged spectra
between 400–1100 nm.
The four cobalt complexes are weak light absorbers in the
visible region with molar extinction coefficients (ε) less than
200 M−1 cm−1 (Fig. 2). The absorption features near 950 nm
and 450 nm are respectively consistent with the T1g(4F) →
4
4T2g(4F) and 4T1g(4F) → 4T1g(4P) transitions previously assigned
to this family of complexes (Fig. 2 inset, Table S3†).28 The
with literature values.29 The tert-butyl groups affect a 0.15–0.18
V decrease in Emed relative to the protio derivative whilst the
substitution of bipyrimdine ligands for bipyridine results in a
0.47–0.50 V increase of the measured Emed value. Similar
increases in Emed upon substituting bipyridine with bipyrimi-
dine have been reported for related ruthenium, iron, and
nickel complexes.28,30,31
4
4
expected T1g → A2g transition bands were not observed and
are likely concealed within the intense intraligand π → π* tran-
sition bands at 200 to 400 nm. The spectra of [Co-bpy](PF6)2
and [Co-bpy-DTB](PF6)2 show an additional absorption feature
associated with intraligand transitions at 295 nm and a hypso-
chromic shift of the near-infrared absorption band near
900 nm compared to the spectra of [Co-bpm](PF6)2 and
[Co-bpm-DTB](PF6)2. The wavelengths of local absorption
maxima (λmax) for [Co-bpy](PF6)2/[Co-(bpy-DTB)](PF6)2 and for
[Co-bpm](PF6)2/[Co-(bpm-DTB)](PF6)2 are nearly identical,
indicating that the tert-butyl groups do not appreciably affect
absorption properties.
Solution mass transport properties of the Co(II) complexes
were derived from diffusion coefficient (DCo(II)) measurements
obtained by rotating disk electrode measurements on glassy
carbon or gold electrodes in acetonitrile solution. The
diffusion coefficients were extracted from the slopes of
Koutecky–Levich plots (Fig. S3 and S4†) and are tabulated in
Table 2. Ferrocene (FcH) was utilized as a reference standard
for which the measured value of 2.48 × 10−5 cm2 s−1 is in close
agreement with the literature value of 2.43 × 10−5 cm2 s−1 32
.
Electrochemical measurements
Cyclic voltammetry performed in acetonitrile solution revealed The diffusion coefficient values decreased with increasing
a
single, quasi-reversible Co(III/II) redox process between molecular size FcH > [Co(bpy)]2+ = [Co(bpm)]2+ > [Co-(bpy-
0.42–1.07 V vs. NHE in all four complexes (Fig. 3; Table 2). DTB)]2+. Note that measurements for [Co-(bpy-DTB)]2+ and
Emed values for [Co-bpy]Z and [Co-(bpy-DTB)]Z are in agreement FcH were carried out on a gold working electrode due to
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Fig. 4 Redox potential levels of the four cobalt redox mediators,
ground-state and excited-state Dye-I, and conduction band level (EC)
and quasi-Fermi level (EF,n) of TiO2. The molecular structure of Dye-I
(5-(4-(bis-(4-iododiphenyl)amino)phenyl)thiophene-2-cyanoacrylic acid)
is shown at left.
Fig. 3 Cyclic voltammograms of Co-(bpy-DTB) (blue), Co-bpy (grey),
Co-(bpm-DTB) (red), and Co-bpm (orange) in 0.1 M n-NBu4PF6 in
MeCN. 100 mV s−1 scan rate with Au working, Pt wire counter and Ag/
AgCl reference electrodes. Arrow indicates the VOC and initial scan
direction with depicted trace from the 2nd scan.
irreversible behaviour at glassy carbon. The diffusion coefficient
for [Co-(bpm-DTB)]2+ could not be obtained using either Au or
GC working electrodes as there was no detectable oxidative
current. This absence of current is unusual given the reversible
redox behaviour observed by cyclic voltammetry (Fig. 3).
One possible explanation is that [Co-(bpm-DTB)]2+ may be
chemically oxidized by oxygen introduced into the electrolyte
by electrode rotation under ambient conditions.33
Photovoltaic device performance
The four cobalt complexes were evaluated as redox mediators
in DSSC devices containing a nanoparticulate TiO2 thin-film
sensitized with a triphenylamine dye (Dye-I) and platinized
counter electrode. The energetics of the dye, cobalt mediators,
and TiO2 in the DSSC are presented in Fig. 4. The excited-state
redox level (Dye-I*/Dye-I+) of −1.29 vs. NHE was estimated from
the intersection of normalized UV-Vis and emission spectra
(Fig. S5†) and is sufficiently negative for photo-injection into
the TiO2 conduction band (i.e. <−0.5 V vs. NHE).1 The ground
state redox potential (Dye-I+/Dye-I) was previously determined
by cyclic voltammetry34 and is sufficiently positive of the
Co(III)/Co(II) redox couples (Emed) to facilitate regeneration by
these mediators. As the solubility of [Co-bpm](PF6)2 in aceto-
nitrile is limited to ca. 0.027 M4 it was necessary to utilize
[Co-bpm](TFSI)2 with an ambient solubility limit of ca. 0.19 M.
The remaining Co(II) complexes were utilized as the hexa-
fluorophosphate salts at 0.2 M concentration. The best J–V
curves are plotted in Fig. S6,† and the average device para-
meters are summarized in Fig. 5 and Table 3.
Fig. 5 Average PCE, JSC, and VOC vs. Emed of [Co-(bpy-DTB)]Z (blue),
[Co-bpy]Z (grey), [Co-(bpm-DTB)]Z (red), and [Co-bpm]Z (orange). The
performance of two replicate devices for each type of mediator are
plotted.
gain in VOC for every +100 mV shift in Emed should be +100 mV
A near unity linear correlation between VOC and Emed was if the quasi-Fermi level of TiO2 (EF,n) is constant, but the
observed using the four mediators (Fig. 5). The theoretical experimentally observed gain in VOC was only 64 2 mV. Back
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
Table 3 Average device performance, quasi-Fermi level (EF,n), and maximum IPCE of DSSCs containing [Co-(bpy-DTB)]Z, [Co-bpy]Z, [Co-(bpm-
DTB)]Z, and [Co-bpm]Za
b
PCE
%
VOC
V
JSC
FF
EF,n
IPCE (420 nm)
%
Mediator
mA cm−2
V vs. NHE
[Co-(bpy-DTB)]Z
[Co-bpy]Z
2.44 (0.10)
3.39 (0.43)
1.94 (0.15)
0.88 (0.13)
0.67 (0.01)
0.77 (0.02)
1.00 (0.01)
1.08 (0.01)
9.16 (1.03)
8.73 (0.74)
5.45 (0.09)
2.27 (0.02)
0.41 (0.02)
0.45 (0.14)
0.37 (0.04)
0.37 (0.05)
−0.25
−0.20
−0.11
−0.01
74
69
56
28
[Co-(bpm-DTB)]Z
[Co-bpm]Z
a Two replicates were measured. The range is indicated in the parentheses. b EF,n = Emed − VOC
.
calculation of EF,n revealed a positive shift in EF,n from
−0.252 V to −0.008 V vs. NHE as Emed increases from 0.42V to
1.07 V vs. NHE (Table 3).
The increase in VOC with increasing Emed is mitigated by a
parallel decrease of both the JSC (Fig. 5) and fill factor (FF).
The trend of decreasing JSC is also observed in incident photon
to current efficiency (IPCE) spectra (Fig. 6) and photocurrent
transient measurements (Fig. 7). The maximum IPCE at
420 nm dropped from 74% for [Co-(bpy-DTB)]Z to only 28% for
[Co-bpm]Z (Table 3), indicating a significantly lower photo-
current generation at higher Emed. Photocurrent transient plots
show that the current densities at the initial peak response are
slightly higher than the values at steady-state, but the order in
which current densities decrease with increasing Emed remains
unchanged. The overall photo-conversion efficiency (PCE) thus
tracks the increasing Emed for the bipyridyl complexes, but the
effects of decreasing JSC and FF result in significantly lower
Fig. 7 Photocurrent transient of best performing cell devices with
cobalt mediator Co-(bpy-DTB) (blue), Co-bpy (grey), Co-(bpm-DTB)
(red), and Co-bpm (orange) under simulated AM 1.5 irradiation at
1000 mW cm−1. Dash lines indicate the current density at the initial
response upon illumination.
PCE values for the highest Emed bipyrimidine functionalized
mediators (Fig. 5).
Open-circuit voltage decay (OCVD) measurements were
undertaken to rationalize the role of the deleterious electron
recombination process on device performance trends
(Fig. S7†). The average time an injected electron survives in the
TiO2 film (Tn) is derived from VOC decay according to eqn (1)
(kb is the Boltzmann constant, T is temperature in Kelvin, e is
elementary charge, and t is time).35
ꢀ
ꢁꢀ1
kbT dVOC
Tn ¼ ꢀ
ð1Þ
e
dt
Semilogarithmic plots of Tn versus EF,n were constructed to
compare Tn in DSSCs containing the four mediators (Fig. 8). A
smaller Tn value corresponds to faster VOC decay and faster
recombination. Recombination appears to proceed faster with
higher Emed mediators, with Tn decreasing in the order
[Co-(bpy-DTB)]Z < [Co-bpy]Z < [Co-(bpm-DTB)]Z < [Co-bpm]Z. At
0.2 V vs. NHE the Tn measured for devices containing
[Co-bpm]Z is an order of magnitude smaller than that found in
devices containing [Co-bpy]Z and [Co-(bpm-DTB)]Z.
Discussion
Effect of nitrogen and tert-butyl groups on mediator
properties
Fig. 6 IPCE spectra of Dye-I sensitized DSSCs containing [Co-(bpy-
DTB)]Z (blue), [Co-bpy]Z (gray), [Co-(bpm-DTB)]Z (red), and [Co-bpm]Z
(orange) mediators.
The incorporation of additional nitrogen atoms into the
heteroaromatic ligand is an effective strategy to alter the
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
spin Co(II) complexes20 and is comparable throughout cobalt
mediators with varying numbers of ligand nitrogen atoms.39
The inner-sphere reorganization energies are therefore
assumed to be similar for all complexes described here. The
comparable molecular radii of [Co-bpy]/[Co-bpm] and [Co(bpy-
DTB)]/[Co(bpm-DTB)] similarly leads us to predict comparable
outer-sphere reorganization energies for these complexes.40
Role of Emed on device performance
J–V and OCVD data clearly show that recombination proceeds
faster as the driving force of the reaction increases and this
trend is indicative of Marcus normal-region behaviour for
recombination.8,41–44 Faster interfacial recombination at more
positive Emed is consistent with increased charge interception
and manifests as a decrease in both current density (JSC) and
fill factor (FF). VOC is increasingly difficult to improve as any
increases of Emed are inversely coupled to a decrease of EF,n
.
Fig. 8 Electron lifetime (Tn) vs. quasi-Fermi level (EF,n) graph derived
from open-circuit voltage decay curve from devices containing [Co-
bpm]Z (orange), [Co-(bpm-DTB)]Z (red), [Co-bpy]Z (grey), and [Co-(bpy-
DTB)]Z (blue).
We found the optimal balance between the three device para-
meters with Emed close to 0.57 V vs. NHE. This balance poss-
ibly explains why reports of the best performing DSSCs all
employ cobalt redox couples with an Emed near that value.45–47
It is imperative, however, to ensure that other factors such as
dye regeneration efficiency, light-harvesting ability of the dye,
electrochemical properties of a complex whilst minimally impact- and size-dependent mass transport of mediators are not inter-
ing magnetic properties and physical structure. No appreciable fering with the critical analysis of device performance. We
differences in bond length, coordination geometry, and mass justify in the following text that these factors are not respon-
transport for [Co-bpy]2+/[Co-bpm]2+ and [Co-(bpy-DTB)]2+/ sible for the observed trends in VOC, JSC, and FF.
[Co-(bpm-DTB)]2+ imply that steric effects may be ruled out in
Inefficient dye regeneration is a potential cause for losses
comparing the complexes for DSSC performance metrics. Note in JSC and FF, but it is unlikely the key contributor in this
that tert-butyl groups are known to attenuate electronic coupling study. The driving forces for regeneration of Dye-I are calcu-
interactions at electrode surfaces,36 and the observation of lated to be 0.85, 0.70, 0.38, and 0.20 eV for [Co-(bpy-DTB)]Z,
smaller diffusion coefficients for [Co-(bpy-DTB)]2+ compared to [Co-bpy]Z, [Co-(bpm-DTB)]Z, and [Co-bpm]Z respectively. Feldt
[Co-bpy]2+ is consistent with bulkier complexes experiencing et al. found that a regeneration driving force above 0.39 eV can
slower mass transport in solution.19
sufficiently regenerate more than 91% of the dye molecule.10
Inductive effects are thought to be the primary reason for [Co-(bpm-DTB)] should therefore regenerate Dye-I as effectively
both the positive shift in Emed with increasing number of as [Co-(bpy-DTB)] and [Co-bpy], yet the JSC and FF are compara-
ligand nitrogen atoms and the negative shift of Emed with the tively lower.
addition of tert-butyl groups. The presence of extra nitrogen
It is evident from incident photon to current efficiency
atoms in the aromatic ligand decreases the σ-donor strength (IPCE) spectra that the mediators do not alter the light-harvest-
and increases the π-acceptor strength of bipyrimidine com- ing ability of the dye or the resultant JSC trend (Fig. 6). Any
pared to bipyridine.37,38 Because the former ligands provide synergistic dye-mediator effects would present as a change in
less electron density and stabilization to the Co(II) center a both the intensity and wavelength profile of IPCE spectra, yet
higher potential is required to oxidize the bipyrimidine func- IPCE values were found to decrease uniformly over the wave-
tionalized complexes relative to the bipyridine congeners. In lengths corresponding to Dye-I absorption (Fig. S5†). The peak
contrast the electron-donating tert-butyl groups increase at around 330 nm is associated with direct UV excitation of the
σ-donation from ligand to metal and thus [Co-(bpy-DTB)]2+/ mesoporous TiO2 which becomes more distinct with decreas-
[Co-(bpm-DTB)]2+ are easier to oxidize and exhibit lower Emed ing IPCE.48 Competitive light absorption by the mediator is an
values relative to [Co-bpy]2+/[Co-bpm]2+ respectively.
unlikely contributor to the change in IPCE considering the
The high-spin d7 and low-spin d6 states are maintained minimal visible-light absorption in UV-Vis spectra of the indi-
across all four cobalt complexes regardless of structural vari- vidual mediators (Fig. 2). The limited IPCE performance with
ation. Upon reaction with a photo-oxidized dye all complexes increasingly positive Emed is hence an indication of enhanced
are expected to undergo spin crossover which is known to recombination kinetics.8,18
impact the inner-sphere reorganization energy,20 but the
The molecular size of the cobalt mediators does not affect
extent of the effect was not experimentally or computationally the JSC trend observed from J–V measurements. The initial
determined here. This reorganization energy is predicted to be peak in the photocurrent transient response data corresponds
insensitive to the donor strength of bidentate ligands in high- to JSC in the absence of diffusion effects whereas the final,
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
steady-state JSC is limited by diffusion of the mediator (Fig. 7). represent an ideal series to unambiguously elucidate struc-
A more pronounced difference between the peak and steady- ture–property relationships for DSSC redox mediator perform-
state JSC values in photocurrent transient data for [Co-(bpy- ance. A clear linear relationship between Emed and VOC was
DTB)]Z/[Co-(bpm-DTB)]Z than for [Co-bpy]Z/[Co-bpm]Z is con- determined for DSSCs incorporating each of the four
sistent with slower diffusion of larger mediators.18,19 Under mediators. An imperfect 64 mV gain in VOC for every 100 mV
operating conditions a slower diffusion rate can lead to the increase of Emed is ascribed to a parallel increase in TiO2-
accumulation of Co(III) species at the TiO2-solution interface mediator recombination rates on the basis of IPCE, J–V and
and thereby increase the probability of recombination.49 We OVCD data. Surface state recombination was evident for [Co-
assert that size-dependent diffusion limitations do not affect bpm]Z and [Co-bpy]Z whereas [Co-(bpm-DTB)]Z and [Co-(bpy-
the overall JSC trend observed here as both the peak and DTB)]Z appear to follow a conduction band recombination
steady-state JSC values decrease with increasing Emed
.
mechanism on account of steric screening afforded by the tert-
butyl substituents. This study provides insights for the design
of next generation metal complex redox mediators. The
optimum Emed to maximize both VOC and JSC appears to span
the range 0.57–0.89 V NHE, with higher Emed potentials requir-
ing specific measures to mitigate electron recombination
reactions.
Inhibiting recombination pathways as a function of mediator
size
Our analysis of open-circuit voltage decay plots suggests that
tert-butyl groups hinder electron recombination between the
mediator and surface states of TiO2. The shape of the semi-log-
arithmic Tn curves in Fig. 8 provide insight to differences in
recombination mechanism across the four complexes. The Tn
plots of [Co-bpy]Z (EF,n > 0.0 V vs. NHE) and [Co-bpm]Z (EF,n > Experimental
0.6 V vs. NHE) exhibit nonlinear regions that are indicative of
Materials
recombination mediated by surface states.50 In contrast the
near linear Tn plots for [Co-(bpy-DTB)]Z and [Co-(bpm-DTB)]Z
suggest that the dominant recombination pathway is instead
from the conduction band of TiO2.51 Tert-butyl groups presum-
ably act as spacers that prevent close contact of cobalt
mediators with the TiO2 film and thus attenuate electronic
coupling to the surface states. The electronic coupling factor
decreases exponentially with distance between TiO2 and the
redox mediator and is proportional to the square-root of the
recombination rate constant.52 The steric bulk of tert-butyl
groups relative to methyl groups have previously been shown
to affect a 20-fold decrease in electronic coupling factor and
13–50-fold decrease in charge-transfer rate constant in related
polypyridyl osmium complexes.36 We anticipate a similar
attenuation trend in this work given the estimated 2.2 Å
increase in molecular radii for those cobalt complexes with
tert-butyl groups compared to the unsubstituted complexes. In
summary the inclusion of bulky alkyl groups could be an
effective strategy to deter surface state recombination that is
otherwise enhanced for cobalt mediators with an Emed greater
than 0.85 V vs. NHE.10
2,2′-Bipyridine (bpy) and 4,4′-tert-butyl-2,2′-bipyridine (bpy-
DTB) were obtained from Sigma Aldrich. 2,2′-bipyrimidine
(bpm) was purchased from Ark Pharm Inc. Dry solvents were
commercially purchased and dispensed from an MBRAUN
MB-SPS-800 solvent purification system. Tetra-n-butylammo-
nium hexafluorophosphate (n-NBu4PF6) was purified via recrys-
tallization from 1 : 1 ethanol/water and dried in a vacuum oven
prior to use.53 All other reagents were obtained from commer-
cial sources and used as received.
Compound characterization
NMR spectra were recorded on Bruker Avance II 400 MHz
spectrometer with 5 mm BBO BB/1H Z-GRD probe or Bruker
Avance III HD 400 MHz spectrometer with 5 mm BBO BB/
19F-1H/D Z-GRD probe in deuterated solvent at 298 K.
Chemical shifts are referenced to residual protio-solvent reso-
nances.54 The spectral window was 0 to 120 ppm to measure
paramagnetic proton resonances. High-resolution mass
spectra were recorded using atmospheric pressure chemical
ionization (APCI) on a Waters/Micromass LCT TOF-MS. CHN
elemental analysis were performed on a Thermo Flash 2000
Elemental Analyzer in the UBC Mass Spectrometry Centre or
on a CE Instruments 1100 Elemental Analyzer in the Simon
Fraser University Chemistry Elemental Analysis Lab.
Conclusion
We carried out a systematic study of the structurally analogous
Single-crystal X-ray diffractograms were collected on a
[Co-bpy]Z, [Co-bpm]Z, [Co-(bpy-DTB)]Z, and [Co-(bpm-DTB)]Z Bruker X8 APEX II diffractometer using graphite monochro-
redox mediators to investigate how changes in the Co(III)/Co(II) mated Mo Kα radiation (λ = 0.71073 Å). Crystallographic data
redox couple affect the VOC of representative DSSCs. The in CIF format are available from the Cambridge
inclusion or omission of ring nitrogen atoms and tert-butyl Crystallographic Data Centre.
groups in the bidentate ligand varies the Co(III)/Co(II) redox
Magnetic moments were calculated from the magnetic sus-
couple from 0.42 V to 1.07 V vs. NHE without significantly ceptibility values obtained using the Evans method.26 A sealed
impacting the spin state or light absorption properties, nor the capillary tube containing a drop of 1,4-dioxane in CD3CN was
relative size and diffusion coefficients for those complexes inserted into an NMR tube containing a solution of the Co(II)
with and without tert-butyl groups. The complexes therefore complex (10 mM) and a drop of 1,4-dioxane dissolved in
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
CD3CN. The measured magnetic susceptibilities were cor- three successive freeze–pump–thaw cycles and subsequently
rected and converted to magnetic moments as described in heated to 70 °C under a gently flowing atmosphere of N2. Once
the ESI.†
the temperature was stable a separate solution of 2-chloro-4-
UV-Vis absorption spectra were recorded on a Varian Cary tert-butyl-pyrimidine (1.7 g, 10 mmol) dissolved in DMF
35 UV-vis spectrophotometer with a slit width of 1 nm. (10 mL) (sparged with N2 for 1 h) was added to the heated
Measurements were carried out in MeCN in a 10.00 mm path- nickel suspension. The combined mixture was left to react
length quartz cuvette at ambient temperature. Molar extinction overnight, cooled to ambient temperature, quenched with
coefficients were calculated by Beer’s law using a set of cobalt NH4OH (100 mL, 4 M) and then left to stir for 30 min. After fil-
solutions with known concentrations between 1 μM and tration of excess zinc powder from the mixture the DMF was
5 mM.
evaporated under reduced pressure. The crude product was
Cyclic voltammograms (CV) were obtained using a CH extracted with diethyl ether (100 mL) and washed with H2O
Instruments CHI660D potentiostat. Measurements were taken (1 L) to remove any remaining DMF. Purification by silica
in freshly N2-purged solutions of 0.1 M n-NBu4PF6 dissolved in column chromatography was carried out in 1 : 4 ethyl acetate/
dry MeCN. 0.5 mm diameter Pt wire counter electrode petroleum ether to remove the residual metal catalysts and
(CHI-115, CH Instruments Inc.) and aqueous Ag/AgCl refer- most of the triphenylphosphine followed by 100% ethyl
ence electrode (MF-2052, Bioanalytical Systems Inc.) were used acetate to recover the product. Further purification by tempera-
in combination with a 2 mm diameter Au working electrode ture-gradient recrystallization from hexane yielded the white
(CHI-101, CH Instruments Inc.).
crystalline product (0.82 g, 3.0 mmol, 60%). 1H NMR
Diffusion measurements were taken with a rotating ring- (400 MHz, CDCl3) δ 8.85 (d, J = 5.3 Hz, 2H), 7.40 (d, J = 5.3 Hz,
disk electrode set-up (MSR Electrode Rotator model AFMSRCE 2H), 1.45 (s, 18H). HRMS (APCI) m/z: [M + H+] calcd for
and E6R1 ChangeDisk RRDE tip, Pine Research C16H23N4 + 271.1923, found 271.1924.
Instrumentation) coupled to a potentiostat (CHI920C, CH
General preparation for tris(2,2′-diimine)cobalt(II) salts. The
Instruments Inc.). Pt or Au disk insert of 5 mm diameter procedure was adapted from published methods.9,58
(AFED050P040AU and AFED050P040GC, Pine Research Anhydrous cobalt(II) chloride (CoCl2, 1 equiv.) and the substi-
Instrumentation) was placed in the rotating electrode tip. tuted 2,2′-diimine ligand (3.1–4.3 equiv.) were dissolved separ-
Linear sweep voltammetry was performed under ambient con- ately in MeOH (approximately 10 mL per 10 mg of CoCl2 and
ditions in an electrolyte solution containing 0.1 M n-NBu4PF6 10 mL per 100 mg of ligand). The CoCl2 solution was
and a known concentration of the cobalt(II) species dissolved added dropwise to the ligand solution over 3 min. The com-
in MeCN (approximately 1.5 or 2.5 mM). Pt and Au working bined solution was vigorously stirred and refluxed for 2 h.
disk electrodes of
5
mm diameter (Pine Research Solid potassium hexafluorophosphate (KPF6) or lithium bis(tri-
Instrumentation) were rotated from 100 rpm up to 2500 rpm. fluoromethane)sulfonimide (LiTFSI) was added in excess to
Aqueous Ag/AgCl electrode (MF-2052, Bioanalytical Systems precipitate the product. The precipitate was isolated through
Inc.) or 0.5 mm diameter Pt wire (Strem Chemicals) was used vacuum filtration, rinsed minimally with cold MeOH, and
as the reference electrode. The counter electrode was a 0.5 mm redissolved in MeCN. After filtering the solution through a
diameter Pt wire (CHI-115, CH Instruments Inc.). The glass frit, the solvent was removed in vacuo to yield the pure
measured currents were fitted to the Levich-Koutecky equation product.
to calculate the diffusion coefficients.55
Tris(2,2′-bipyrimidine)cobalt(II) hexafluorophosphate [Co-
bpm](PF6)2. Anhydrous CoCl2 (0.027 g, 0.21 mmol), bpm
(0.13 g, 0.83 mmol), and KPF6 (0.13 g, 0.69 mmol) reacted to
Synthesis
2-Chloro-4-tert-butyl-pyrimidine. Following
a
previously yield the product as an orange powder (0.16 g, 0.19 mmol,
reported procedure,56 2-chloropyrimidine (0.67 g, 5.9 mmol), 90%). 1H NMR (400 MHz, CD3CN): 100.63 (s, 6H), 41.02 (s,
pivalic acid (0.60 g, 5.9 mmol), potassium persulfate (1.6 g, 6H), 19.55 (s, 6H).
5.9 mmol) and silver nitrate (0.20 g, 1.2 mmol) were stirred for
Tris(2,2′-bipyrimidine)cobalt(II) bis(trifluoromethane)sulfo-
16 h in 1 : 1 DCM/H2O (20 mL) at room temperature with the nimide [Co-bpm](TFSI)2. Anhydrous CoCl2 (0.12 g,
exclusion of ambient light. The resulting grey suspension was 0.93 mmol), bpm (0.58 g, 3.7 mmol), and LiTFSI (0.13 g,
dried under reduced pressure, redissolved in ethyl acetate, and 0.69 mmol) reacted to yield the product as an orange powder
1
filtered to separate solid precipitates. The filtrate solution was (0.84 g, 0.76 mmol, 83%). H NMR (400 MHz, CD3CN): 100.81
washed with H2O (3 × 10 mL), dried over Na2SO4, and finally (s, 6H), 41.14 (s, 6H), 19.62 (s, 6H). Anal. Calcd for
purified by silica gel chromatography (1 : 4 ethyl acetate/pet- C28H18CoF12N14O8S4: C, 30.75; H, 1.66; N, 17.93. Found: C,
roleum ether, Rf = 0.43) to afford a pale-yellow oil (0.48 g, 31.00; H,1.58; N, 17.70.
1
2.8 mmol, 48%). H NMR (400 MHz, CD3CN) δ 8.55 (d, J = 5.3
Hz, 1H), 7.43 (d, J = 5.3 Hz, 1H), 1.32 (s, 9H).
Tris(2,2′-bipyridine)cobalt(II) hexafluorophosphate [Co-bpy]
(PF6)2. Anhydrous CoCl2 (0.16 g, 1.3 mmol), bpy (0.63 g,
4,4′-Di-tert-butyl-2,2′-bipyrimidine (bpm-DTB). Adapting a 4.0 mmol), and KPF6 (1.1g, 5.8 mmol) reacted to yield the
literature procedure,57 mixture of NiCl2·6H2O (1.4 g, product as a brown powder (0.19 g, 0.25 mmol, 19%). 1H NMR
a
11 mmol), triphenylphosphine (11 g, 42 mmol), and zinc (400 MHz, CD3CN): 87.80 (s, 6H), 83.80 (s, 6H), 46.00 (s, 6H),
powder (1.0 g, 16 mmol) in DMF (40 mL) were degassed by 14.50 (s, 6H).
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
Tris(4,4′-di-tert-butyl-2,2′-bipyrimidine)cobalt(II) hexafluoro- with a total calculated active area of 0.196 cm2 and total thick-
phosphate [Co-(bpm-DTB)](PF6)2. Anhydrous CoCl2 (0.045g, ness between 10–12 μm. The TiO2 substrates were treated in a
0.35 mmol) and bpy-DTB (0.40 g, 1.5 mmol) were reacted in 40 mM TiCl4(aq) bath at 70 °C for 30 min, and rinsed with H2O
MeOH (50 mL) followed by the addition of KPF6 (0.51 g, then EtOH. After gradual heating to 500 °C over 35 min in air
2.7 mmol). The solvent was evaporated and the crude residue the photoanodes were cooled to 80 °C and then immersed in
redissolved in a minimal volume of dichloromethane (1 mL). an MeCN solution of 0.1 mM dye and 1 mM chenodeoxycholic
Addition of hexane to the solution afforded a precipitate col- acid for 24 h. The stained films were then rinsed and sub-
lected by filtration and washed copiously with additional sequently immersed in MeCN for 30 min before drying in a
hexane. The solid was redissolved in dichloromethane and desiccator. Cathodes were freshly prepared by coating FTO-
again filtered through a glass frit, with the resulting filtrate glass with a 0.2 mM solution of H2PtCl6 dissolved in ethanol
evaporated to afford the product as a pale orange powder followed by curing at 400 °C for 22 min. Cathodes were main-
1
(0.36 g, 0.32 mmol, 90%). H NMR (400 MHz, CD3CN) δ 90.35 tained in a 100 °C environment until ready for immediate cell
(s, 6H), 80.91 (s, 6H), 43.07 (s, 6H), 3.51 (s, 54H). Anal. Calcd assembly. Cells were assembled in a bespoke dry box purged
for C48H66CoF12N12P2: C, 49.70; H, 5.73; N, 14.49. Found: C, with dry air. Electrodes were sealed together with a 30 μm
49.50; H, 5.87; N, 14.39.
surlyn gasket (DuPont) by resistive heating. The electrolyte
General preparation of tris(2,2′-diimine)cobalt(III) salts. contained 0.1 M LiTFSI, 0.2 M TBP, 0.2 M Co(II) and 0.02 M
Excess oxidant (NOPF6) was dissolved in dry MeCN (5 mL) and Co(III) in MeCN. In [Co-bpy]Z, [Co-(bpy-DTB)]Z, and [Co-(bpm-
the corresponding tris(2,2′-diimine) cobalt(II) salt was added DTB)]Z electrolytes, 0.22 M of Co(II) species were dissolved in
as a solid. The product was precipitated by dropwise addition MeCN and Co(III) species were generally generated in situ using
of the reaction mixture into distilled water (50 mL) to form a NOPF6. In [Co-bpm]Z electrolyte, 0.2 M of [Co-bpm](TFSI)2 and
fine powder. The precipitate was redissolved in MeCN and 0.02 M of [Co-bpm](TFSI)3 was dissolved in MeCN and only the
directly evaporated within a storage microvial to recover the soluble portion was used. The electrolyte was injected into the
tris(2,2′-diimine)cobalt(III) product. Although quantitative sandwiched electrodes through a pre-drilled hole on the
yields of the products were not obtained 1H NMR spectro- cathode. The hole was sealed via a 60 μM thin surlyn film
scopic analysis indicated successful oxidization of all cobalt(II) (SX1170-60, Solaronix) and microscope coverslip (18 × 18 ×
tris(2,2′-diimine) complexes to the respective cobalt(III) 0.13–0.17 mm, Fisher Scientific). Silver paint (12684-15,
complex. The purity of isolated powders was inferred from a Electron Microscopy Sciences) was coated on the electrode
combined lack of line-broadening and absence of any detect- edges to ensure electrical contact.
able paramagnetic 1H-NMR spectroscopic resonances corres-
ponding to residual Co(II) complex.
Device characterization. Photovoltaic measurements were
recorded with a OAI TriSOL Class AAA solar simulator (MDL
Tris(2,2′-bipyrimidine)cobalt(III) bistrilfilimide [Co-bpm] 0131-0143-01) equipped with a 300 W xenon short-arc lamp
(TFSI)3. Reaction with NOPF6 and [Co-bpm](PF6)2, followed by (UXL-302-O, Ushio) light source and Class A filter for DSSC
salt metathesis with excess LiTFSI yielded an orange powder. applications to cover from 350–850 nm in 50 nm bandwidth
1H NMR (400 MHz, CD3CN) δ 9.48 (dd, 6H, J = 4.8, 1.6 Hz), increments. The power output of the lamp was calibrated to 1
7.98–7.91 (m, 6H), 7.76 (dd, 6H, J = 6.1, 1.6 Hz).
sun (1000 W m−2) using a certified Newport Oriel Si reference
Tris(2,2′-bipyridine)cobalt(III) hexafluorophosphate [Co-bpy] cell (Model SN 0963). The light output area was 52 mm ×
(PF6)3. Reaction with NOPF6 and [Co-bpy](PF6)2 yielded a 52 mm. All cells were placed and fixed with cellophane tape on
yellow powder. 1H NMR (400 MHz, CD3CN) δ 8.73–8.66 (m, a lab jack covered with aluminum foil prior to undertaking
6H), 8.51–8.43 (m, 6H), 7.73 (t, J = 6.6 Hz, 6H), 7.27 (d, J = 5.7, photovoltaic characterization. The J–V characteristic of each
6H).
Tris(4,4′-di-tert-butyl-2,2′-bipyridine)cobalt(III)
cell was obtained by applying an external potential to the cell
hexafluoro- and measuring the generated photocurrent with a Keithley
phosphate [Co-(bpy-DTB)](PF6)3. Reaction with NOPF6 and [Co- digital source meter (model 2400). Incident photon to current
1
(bpy-DTB)](PF6)2 yielded a yellow powder. H NMR (400 MHz, efficiency (IPCE) spectra were recorded with Newport Oriel IQE
CD3CN) δ 8.63(d, J = 2.1 Hz, 6H), 7.67 (dd, J = 6.4, 2.2 Hz, 6H), 200 system. The system was calibrated with a silicon detector
7.05 (d, J = 6.4 Hz, 6H), 1.45 (s, 54H).
(Newport part number IQE-SAMPLE-SI). The light output was
Tris(4,4′-di-tert-butyl-2,2′-bipyrimidine)cobalt(III) hexafluoro- modulated from 300 nm in direct current mode.
phosphate [Co-(bpm-DTB)](PF6)3. Reaction with NOPF6 and Measurements were taken in 10 nm intervals.
[Co-(bpm-DTB)](PF6)2 yielded an orange powder. 1H NMR
Open-circuit voltage decay was measured by applying zero
(400 MHz, CD3CN) δ 7.91 (d, J = 6.5 Hz, 6H), 7.59 (d, J = 6.5 Hz, current bias with the source meter and measuring the decay of
6H), 1.50 (s, 54H). photovoltage in the dark following illumination at 1 sun.
Device fabrication. Photoanodes were prepared by screen Acquisition time was controlled through Keithley Kickstart
printing a nanoparticulate TiO2 paste on a fluorinated tin software with the following parameters: bias current = 0 A;
oxide (FTO)-coated glass substrate and sintering at 500 °C in delay = 0.0005 s; number of power line cycles (NLPC) = 0.1;
an oven. The photoanodes were printed with one nanoparticle and number of data points = 2500. The illumination sequence
layer (∼18 nm particles, 18NR-T paste, Greatcell Solar) and one was manually controlled by illuminating the cell for ca. 10 s
scattering layer (∼200 nm particles, WER4-O paste, Dyesol) and then turning off the lamp for the duration of the measure-
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
ment. Data acquisition initiated ca. 2 s before the lamp was 19 J. J. Nelson, T. J. Amick and C. M. Elliott, J. Phys. Chem. C,
turned off.
2008, 112, 18255–18263.
20 E. Mosconi, J.-H. Yum, F. Kessler, C. J. Gómez García,
C. Zuccaccia, A. Cinti, M. K. Nazeeruddin, M. Grätzel and
F. De Angelis, J. Am. Chem. Soc., 2012, 134, 19438–19453.
21 Y. Xie and T. W. Hamann, J. Phys. Chem. Lett., 2013, 4, 328–
332.
Conflicts of interest
There are no conflicts of interest to declare.
22 J. Baillargeon, Y. Xie and T. W. Hamann, ACS Appl. Mater.
Interfaces, 2017, 9, 33544–33548.
23 J. England, E. Bill, T. Weyhermüller, F. Neese, M. Atanasov
and K. Wieghardt, Inorg. Chem., 2015, 54, 12002–12018.
Acknowledgements
We are grateful to the Canadian Natural Science and 24 A. Menteş and K. Singh, Acta Crystallogr., Sect. E: Struct.
Engineering Research Council, Canadian Institute for Rep. Online, 2013, 69, m58.
Advanced Research, Canadian Foundation for Innovation, and 25 C. R. Groom, I. J. Bruno, M. P. Lightfoot and S. C. Ward,
the Canada Research Chairs.
Acta Crystallogr., Sect. B: Struct. Sci., Cryst. Eng. Mater.,
2016, 72, 171–179.
26 D. F. Evans, J. Chem. Soc., 1959, 2003–2005.
27 B. N. Figgis, Nature, 1958, 182, 1568–1570.
28 R. R. Ruminski and J. D. Petersen, Inorg. Chim. Acta, 1985,
97, 129–134.
29 S. M. Feldt, E. A. Gibson, E. Gabrielsson, L. Sun,
G. Boschloo and A. Hagfeldt, J. Am. Chem. Soc., 2010, 132,
16714–16724.
Notes and references
1 A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo and H. Pettersson,
Chem. Rev., 2010, 110, 6595–6663.
2 M. Grätzel, J. Photochem. Photobiol., C, 2003, 4, 145–153.
3 K. Kakiage, H. Osada, Y. Aoyama, T. Yano, K. Oya,
S. Iwamoto, J.-I. Fujisawa and M. Hanaya, Sci. Rep., 2016, 6, 30 P. S. Braterman, J. I. Song and R. D. Peacock, Inorg. Chem.,
35888. 1992, 31, 555–559.
4 K. Y. Chen, C. Du, B. O. Patrick and C. P. Berlinguette, 31 D. P. Rillema, G. Allen, T. J. Meyer and D. Conrad, Inorg.
Inorg. Chem., 2017, 56, 2383–2386. Chem., 1983, 22, 1617–1622.
5 R. R. Rodrigues, H. Cheema and J. H. Delcamp, Angew. 32 Y. Wang, E. I. Rogers and R. G. Compton, J. Electroanal.
Chem., Int. Ed., 2018, 130, 5570–5574. Chem., 2010, 648, 15–19.
6 P. G. Bomben, K. C. D. Robson, B. D. Koivisto and 33 N. Elgrishi, K. J. Rountree, B. D. McCarthy, E. S. Rountree,
C. P. Berlinguette, Coord. Chem. Rev., 2012, 256, 1438–1450.
7 P. G. Bomben, T. J. Gordon, E. Schott and
T. T. Eisenhart and J. L. Dempsey, J. Chem. Educ., 2018, 95,
197–206.
C. P. Berlinguette, Angew. Chem., Int. Ed., 2011, 50, 10682– 34 S. J. C. Simon, F. G. L. Parlane, W. B. Swords, C. W. Kellett,
10685.
C. Du, B. Lam, R. K. Dean, K. Hu, G. J. Meyer and
C. P. Berlinguette, J. Am. Chem. Soc., 2016, 138, 10406–
10409.
8 M. J. DeVries, M. J. Pellin and J. T. Hupp, Langmuir, 2010,
26, 9082–9087.
9 S. M. Feldt, G. Wang, G. Boschloo and A. Hagfeldt, J. Phys. 35 A. Zaban, M. Greenshtein and J. Bisquert, ChemPhysChem,
Chem. C, 2011, 115, 21500–21507. 2003, 4, 859–864.
10 S. M. Feldt, P. W. Lohse, F. Kessler, M. K. Nazeeruddin, 36 T. W. Hamann, B. S. Brunschwig and N. S. Lewis, J. Phys.
M. Grätzel, G. Boschloo and A. Hagfeldt, Phys. Chem. Chem.
Phys., 2013, 15, 7087–7097.
11 J. T. Kirner and C. M. Elliott, J. Phys. Chem. C, 2015, 119,
17502–17514.
12 A. B. F. Martinson, T. W. Hamann, M. J. Pellin and
J. T. Hupp, Chemistry, 2008, 14, 4458–4467.
13 S. Zhang, M. Yanagida, X. Yang and L. Han, Appl. Phys.
Express, 2011, 4, 042301.
14 J. W. Ondersma and T. W. Hamann, J. Phys. Chem. C, 2010,
114, 638–645.
Chem. B, 2006, 110, 25514–25520.
37 E. V. Dose and L. J. Wilson, Inorg. Chem., 1978, 17, 2660–
2666.
38 S. Ernst and W. Kaim, J. Am. Chem. Soc., 1986, 108, 3578–
3586.
39 Z.-Z. Sun, K.-M. Zheng, Q.-S. Li and Z.-S. Li, RSC Adv., 2014,
4, 31544–31551.
40 R. Memming, in Semiconductor Electrochemistry, Wiley-VCH
Verlag GmbH & Co. KGaA, Weinheim, Germany, 2015, pp.
127–168.
15 T. W. Hamann, Dalton Trans., 2012, 41, 3111–3115.
16 J. Cong, X. Yang, L. Kloo and L. Sun, Energy Environ. Sci.,
2012, 5, 9180–9194.
41 S. Soman, S. C. Pradhan, M. Yoosuf, M. V. Vinayak,
S. Lingamoorthy and K. R. Gopidas, J. Phys. Chem. C, 2018,
122, 14113–14127.
17 S. A. Sapp, C. M. Elliott, C. Contado, S. Caramori and 42 B. N. DiMarco, L. Troian-Gautier, R. N. Sampaio and
C. A. Bignozzi, J. Am. Chem. Soc., 2002, 124, 11215–11222. G. J. Meyer, Chem. Sci., 2018, 9, 940–949.
18 B. M. Klahr and T. W. Hamann, J. Phys. Chem. C, 2009, 113, 43 Y. Xie, J. Baillargeon and T. W. Hamann, J. Phys. Chem. C,
14040–14045. 2015, 119, 28155–28166.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
44 M. Safdari, P. W. Lohse, L. Häggman, S. Frykstrand, 51 J. W. Ondersma and T. W. Hamann, J. Am. Chem. Soc.,
D. Högberg, M. Rutland, R. A. Asencio, J. Gardner, L. Kloo,
A. Hagfeldt and G. Boschloo, RSC Adv., 2016, 6, 56580–56588.
45 S. Mathew, A. Yella, P. Gao, R. Humphry-Baker,
2011, 133, 8264–8271.
52 J. Bisquert and R. A. Marcus, Top. Curr. Chem., 2014, 352,
325–395.
B. F. E. Curchod, N. Ashari-Astani, I. Tavernelli, 53 W. L. F. Armarego and C. L. L. Chai, in Purification of
U. Rothlisberger, M. K. Nazeeruddin and M. Grätzel, Nat.
Chem., 2014, 6, 242–247.
Laboratory Chemicals, Butterworth-Heinemann, Boston,
2013, pp. 103–554.
46 A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran, 54 G. R. Fulmer, A. J. M. Miller, N. H. Sherden, H. E. Gottlieb,
M. K. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M. Zakeeruddin
A. Nudelman, B. M. Stoltz, J. E. Bercaw and K. I. Goldberg,
and M. Grätzel, Science, 2011, 334, 629–634.
Organometallics, 2010, 29, 2176–2179.
47 K. Kakiage, Y. Aoyama, T. Yano, K. Oya, J.-I. Fujisawa and 55 A. J. Bard and L. R. Faulkner, Electrochemical methods:
M. Hanaya, Chem. Commun., 2015, 51, 15894–15897.
fundamentals and applications, Wiley, New York, 2nd edn,
48 J. Wu, G. Yue, Y. Xiao, J. Lin, M. Huang, Z. Lan, Q. Tang,
2001.
Y. Huang, L. Fan, S. Yin and T. Sato, Sci. Rep., 2013, 3, 56 W.-P. Mai, B. Sun, L.-Q. You, L.-R. Yang, P. Mao, J.-W. Yuan,
1283.
Y.-M. Xiao and L.-B. Qu, Org. Biomol. Chem., 2015, 13,
2750–2755.
49 T. T. Trang Pham, T. M. Koh, K. Nonomura, Y. M. Lam,
N. Mathews and S. Mhaisalkar, ChemPhysChem, 2014, 15, 57 A. J. Downard, I. G. Phillips and P. J. Steel, Aust. J. Chem.,
1216–1221. 2004, 57, 865–868.
50 J. Bisquert, A. Zaban, M. Greenshtein and I. Mora-Seró, 58 R. R. Ruminski and J. D. Petersen, Inorg. Chim. Acta, 1984,
J. Am. Chem. Soc., 2004, 126, 13550–13559.
88, 63–66.
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.