Spin-Doctoring Cobalt Redox Shuttles for Dye-Sensitized Solar Cells

Article abstract of DOI:10.1021/acs.inorgchem.8b01772

A new low-spin (LS) cobalt(II) outer-sphere redox shuttle (OSRS) [Co(PY5Me₂)(CN)]⁺, where PY5Me₂ represents the pentadentate ligand 2,6-bis(1,1-bis(2-pyridyl)ethyl)pyridine, has been synthesized and characterized for its potential application in dye-sensitized solar cells (DSSCs). Introduction of the strong field CN^- ligand into the open axial coordination site forced the cobalt(II) complex, [Co(PY5Me₂)(CN)]⁺, to become LS based upon the complex's magnetic susceptibility (1.91 ± 0.02 μ_B), determined by the Evans method. Interestingly, dimerization and subsequent cobalt hexacyanide cluster formation of the [Co(PY5Me₂)(CN)]⁺ monomer was observed upon long-term solvent exposure or addition of a supporting electrolyte for electrochemical characterization. Although long-term stability of the [Co(PY5Me₂)(CN)]⁺ complex made it difficult to fabricate liquid electrolytes for DSSC applications, short-term stability in neat solvent afforded the opportunity to isolate the self-exchange kinetics of [Co(PY5Me₂)(CN)]^2+/+ via stopped-flow spectroscopy. Use of Marcus theory provided a smaller than expected self-exchange rate constant of 20 ± 5.5 M^-1 s^-1 for [Co(PY5Me₂)(CN)]^2+/+, which we attribute to a Jahn-Teller effect observed from the collected monomer crystallographic data. When compared side-by-side to cobalt tris(2,2′-bipyridine), [Co(bpy)₃]³⁺, DSSCs employing [Co(PY5Me₂)(CN)]²⁺ are expected to achieve superior charge collection, which result from a smaller rate constant, k_et, for recombination based upon simple dark J-E measurements of the two redox shuttles. Given the negative redox potential (0.254 V vs NHE) of [Co(PY5Me₂)(CN)]^2+/+ and the slow recombination kinetics, [Co(PY5Me₂)(CN)]^2+/+ becomes an attractive OSRS to regenerate near IR absorbing sensitizers in solid-state DSSC devices.

Full text of DOI:10.1021/acs.inorgchem.8b01772

Article
pubs.acs.org/IC
Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX
Spin-Doctoring Cobalt Redox Shuttles for Dye-Sensitized Solar Cells
Josh Baillargeon, Yuling Xie, Austin L. Raithel, Behnaz Ghaffari, Richard J. Staples,
and Thomas W. Hamann*
Department of Chemistry, Michigan State University, East Lansing, Michigan 48824, United States
S
* Supporting Information
ABSTRACT: A new low-spin (LS) cobalt(II) outer-sphere
redox shuttle (OSRS) [Co(PY5Me2)(CN)]+, where PY5Me2
represents the pentadentate ligand 2,6-bis(1,1-bis(2-pyridyl)-
ethyl)pyridine, has been synthesized and characterized for its
potential application in dye-sensitized solar cells (DSSCs).
Introduction of the strong field CN− ligand into the open axial
coordination site forced the cobalt(II) complex, [Co-
(PY5Me2)(CN)]+, to become LS based upon the complex’s
magnetic susceptibility (1.91 0.02 μB), determined by the
Evans method. Interestingly, dimerization and subsequent
cobalt hexacyanide cluster formation of the [Co(PY5Me2)(CN)]+ monomer was observed upon long-term solvent exposure or
addition of a supporting electrolyte for electrochemical characterization. Although long-term stability of the [Co(PY5Me2)-
(CN)]+ complex made it difficult to fabricate liquid electrolytes for DSSC applications, short-term stability in neat solvent
afforded the opportunity to isolate the self-exchange kinetics of [Co(PY5Me2)(CN)]2+/+ via stopped-flow spectroscopy. Use of
Marcus theory provided a smaller than expected self-exchange rate constant of 20 5.5 M−1 −1 for [Co(PY5Me2)(CN)]2+/+
s ,
which we attribute to a Jahn−Teller effect observed from the collected monomer crystallographic data. When compared side-
by-side to cobalt tris(2,2′-bipyridine), [Co(bpy)3]3+, DSSCs employing [Co(PY5Me2)(CN)]2+ are expected to achieve superior
charge collection, which result from a smaller rate constant, ket, for recombination based upon simple dark J−E measurements of
the two redox shuttles. Given the negative redox potential (0.254 V vs NHE) of [Co(PY5Me2)(CN)]2+/+ and the slow
recombination kinetics, [Co(PY5Me2)(CN)]2+/+ becomes an attractive OSRS to regenerate near IR absorbing sensitizers in
solid-state DSSC devices.
opportunities arise for DSSCs to be competitive with existing
silicon PV technologies.
INTRODUCTION
■
Dye-sensitized solar cells, DSSCs, are a promising solar energy
conversion technology due to their ability to achieve high
power conversion efficiencies (PCEs) by way of low-cost
processing. However, for well over a decade, efforts to push
device efficiencies past 10% were stymied by the reliance on a
To date, the best performing DSSC has employed the OSRS
cobalt tris(2,2′-bipyridine), [Co(bpy)3]3+/2+ (bpy = 2,2′-
bipyridine) paired with a Zn-porphyrin sensitizer SM315 to
produce a record PCE of over 13%.11 Although DSSCs
employing [Co(bpy)3]3+/2+ have produced the highest
efficiencies, the performance of these devices are still
suboptimal. The suboptimal performance can be attributed
to the spin change associated with the oxidation of the cobalt
metal center. We have recently shown that the large inner-
sphere reorganization energy associated with the loss of two
antibonding electrons upon oxidation of high-spin (HS)
[Co(bpy)3]2+ to low-spin (LS) [Co(bpy)3]3+ results in
inefficient dye regeneration of the organic dye D35cpdt.9,12
Furthermore, our group and others have shown improved
PCEs in DSSCs containing tandem electrolytes of a fast
exchanging redox shuttle mixed with [Co(bpy)3]3+/2+ provid-
single redox shuttle iodide/triiodide (I−/I3 ) paired with a
−
small subset of ruthenium sensitizers.1,2 The major limitations
of utilizing such dye−electrolyte combinations stems from the
high overpotential (∼0.7 eV) necessary for efficient regener-
ation, the fixed redox potential of iodide/triiodide, as well as
the complicated kinetics associated with charge transfer.1−4 To
avoid such problems, it has been of considerable interest to
seek alternative redox shuttles that are better matched with the
vast library of dyes that already exist. Outer-sphere redox
shuttles (OSRSs) have arisen as a promising solution due to
their tunable ligand framework and simple one-electron
transfer mechanism.5−8 The mechanism for charge transfer
using OSRSs has been successfully modeled by the application
of Marcus theory and has already provided a better
understanding of the regeneration and recombination path-
ways in DSSCs.9,10 With such kinetic predictability provided
by Marcus theory and the fine synthetic control over the
properties of designing new OSRSs, new routes for
optimization of device efficiencies are possible and more
ing even more evidence that regeneration is suboptimal with
12−14
electrolytes containing only [Co(bpy)3]3+/2+
.
Assuming
regeneration can be modeled as a simple cross-exchange
reaction between the dye and redox shuttle, Marcus theory
Received: June 26, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 1. Molecular structures of various cobalt containing redox shuttles for DSSCs: cobalt tris(2,2′-bipyridine), [Co(bpy)3]3+/2+, cobalt bis(1,4,7-
trithiacyclononane), [Co(ttcn)2]3+/2+, cobalt 2,6-bis(1,1-bis(2-pyridyl)ethyl)pyridine acetonitrile, [Co(PY5Me2)(MeCN)]3+/2+, and cobalt 2,6-
bis(1,1-bis(2-pyridyl)ethyl)pyridine cyanide [Co(PY5Me2)(CN)]2+/+
.
would suggest that redox shuttles with faster self-exchange
kinetics should provide faster regeneration kinetics.3 This in
fact is true and has been demonstrated in our lab through the
use of a LS cobalt(II) OSRS [Co(ttcn)2]3+/2+, where ttcn
represents 1,4,7-trithiacyclononane (Figure 1).9 Through
external quantum yield measurements, it was determined
that regeneration was nearly quantitative using [Co-
(ttcn)2]3+/2+ compared to [Co(bpy)3]3+/2+, despite only a
∼60 mV smaller driving force to regenerate the sensitizer
D35cpdt. Unfortunately, DSSCs employing [Co(ttcn)2]3+/2+
suffered from faster recombination compared to [Co-
(bpy)3]3+/2+ which diminished the charge collection efficiency.
In principle, the charge collection can be improved by reducing
the driving force, and thus the rate of recombination, without
sacrificing advantageous regeneration kinetics. Unfortunately,
as the ligand framework of ttcn is comprised of sp3 carbons,
there are no synthetic handles to tune the redox potential, i.e.,
adding substituents onto the carbons or increasing the number
of carbon atoms on the ring system.15−17 Because we are
unaware of any alternative LS cobalt(II) OSRSs to [Co-
(ttcn)2]3+/2+, our effort to investigate the regeneration and
recombination reactions of LS cobalt (II) OSRSs as a function
of driving force has been thwarted.
We have therefore been motivated to exploit alternative
ligand systems and develop a new family of promising LS
cobalt redox shuttles for DSSCs. Ideally, synthesis of new LS
cobalt OSRSs would open up access to more negative redox
potentials than the commonly used cobalt polypyridyl
complexes in order to minimize the driving force for interfacial
charge transfer, i.e., slower recombination kinetics, while
consequently maximizing the charge collection. With such
negative redox potentials and fast exchange kinetics, efficient
dye regeneration is expected at small overpotentials for
sensitizers with smaller optical gaps. This in turn will provide
a viable route for integrating new near IR absorbing sensitizers
into DSSC devices. Inspiration for designing a new family of
OSRSs has come from the groups of Bach and Long where a
pentadentate ligand 2,6-bis(1,1-bis(2-pyridyl)ethyl)pyridine,
PY5Me2, was coordinated to a cobalt center to provide the
parent complex, [Co(PY5Me2)(MeCN)]3+/2+, amenable to
functionalization.18−21 It is expected that the variation of
electron-donating or -withdrawing ligands in the axial site of
these coordination complexes will provide a high degree of
tunability with regards to formal potential and the spin-state of
the cobalt metal center. Analogous studies done by Stack et al.
has already demonstrated such tunability on a series of ferrous
complexes with a structurally similar ligand, 2,6-(bis(bis-2-
pyridyl)methoxymethane)pyridine (PY5).22 In spanning the
spectrochemical series via coordination of various axial ligands
(Cl−, N3, MeOH, CN−, etc.), each Fe(II) complex was highly
susceptible to changes in spin-state and redox potential.
Functionalization of the axial cobalt ligand via displacement of
a weakly coordinated acetonitrile (MeCN) has already been
done using common DSSC electrolyte additives such as 4-tert-
butylpyridine (tBP) and N-methylbenzimidazole (NMBI);
however, use of such pyridine derivatives failed to significantly
modulate the energetics of the resulting OSRSs or the spin−
state of the cobalt metal center.21 To build on Stack and Bach’s
previous studies, we envisioned using a strong field ligand such
as cyanide, CN−, to obtain the desired results. Stack
demonstrated that as a strong donor and anionic ligand
cyanide can push the redox potential more negative than most
ligands in the spectrochemical series. In addition, the strong
field ligand induced a LS Fe(II). With this study in mind, we
reasoned that introduction of the CN− ligand to the sixth
coordination site of the [Co(PY5Me2)(MeCN)]2+ complex
would likewise result in a rare example of a LS Co(II) complex
with a potential more negative of [Co(ttcn)2]3+/2+ and thus
begin a promising new class of OSRS.
In this work, we have prepared and characterized the cobalt
complexes, [Co(PY5Me2)(CN)]2+/+ (Figure 2). Through the
use of Evans method studies, it was determined that
coordination of a cyanide ligand to the parent [Co(PY5Me2)-
(MeCN)]2+ forced the Co(II), [Co(PY5Me2)(CN)]+, to
become LS. Interestingly, addition of cyanide to [Co-
(PY5Me2)(MeCN)]2+ resulted in unexpected side reactions
that were highly dependent on the reaction conditions
imparted. Depending on the equivalents of cyanide, temper-
ature, and overall reaction time, dimerization and subsequent
precipitation of a cobalt cluster complex were observed. The
dimer complex was isolated and characterized; however, due to
solubility issues, the cluster complex was only analyzed by X-
ray crystallography. The instability of the pure [Co(PY5Me2)-
1
(CN)]+ complex was identified by H NMR and electro-
chemistry studies. In neat acetonitrile, the complex remains
stable for kinetic measurements using stopped-flow spectros-
copy; however, dimerization results almost instantaneously
upon addition of a supporting electrolyte. Although stability
appears to be an issue in liquid electrolytes, the short-term
stability in neat solvent makes [Co(PY5Me2)(CN)]2+/+ an
ideal candidate as a solid-state hole conductor for DSSCs.
RESULTS AND ANALYSIS
■
Synthesis. Synthesis of the parent [Co(PY5Me2)-
(MeCN)](OTf)2 was carried out following a modified
literature procedure.18 Coordination of the neutral PY5Me2
ligand to the acidic cobalt metal center yielded a stable halide
product, [Co(PY5Me2)I]I, using the proper metal salt and the
B
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
robust solvato complex, [Co(PY5Me2)(MeCN)](OTf)2, upon
metathesis in a coordinating solvent such as acetonitrile. In
principal, synthesis of the [Co(PY5Me2)(CN)](OTf) complex
from the parent [Co(PY5Me2)(MeCN)](OTf)2 should be
facile and clean via the addition of a cyanide source.
Interestingly, however, the cyanide ligand was observed to be
much more labile than anticipated. Careful control over the
reaction conditions was necessary in order to mitigate
dimerization and/or subsequent cluster formation of a cobalt
hexacyanide complex.
S8). Single crystals suitable for X-ray crystallography were
again obtained via slow vapor diffusion of ether into a
concentrated acetonitrile solution of [Co(PY5Me2)(CN)]-
(OTf)2 at room temperature (Figure 2b). A full character-
ization of the structure, spin-state, optical and kinetic
properties of both [Co(PY5Me2)(CN)]2+/+ complexes were
carried out and are described in detail below.
Dimerization was observed when the reaction conditions
were modified such that only one or less equivalents (≤1
equiv) of CN− are added to the [Co(PY5Me2)(MeCN)]-
(OTf)2 reaction mixture. Even at low temperature (ice bath)
and under an inert atmosphere, the predominant product that
was obtained was a dimer complex. Single crystals were readily
grown by slow vapor diffusion of diethyl ether into a
concentrated acetonitrile solution containing the dimer, as in
Figure 3. Structural information and vibrational properties
Isolation of the pure [Co(PY5Me2)(CN)](OTf) complex
was obtained by ensuring the reaction between [Co(PY5Me2)-
(MeCN)](OTf)2 and CN− was carried out using a slight
excess (∼1.2−1.5 equiv) of CN−, at low temperature (ice
bath), under an inert atmosphere, and in a noncompetitive
solvent such as methanol. By visual inspection, the reaction
was deemed complete within seconds as the solution changed
from yellow to a dark reddish/brown upon cyanide addition.
Single crystals were obtained via slow vapor diffusion of ether
into a concentrated acetonitrile solution of [Co(PY5Me2)-
1
(CN)](OTf) at room temperature (Figure 2a). H NMR of
Figure 3. Single crystal representation of the dimer complex provided
by Olex2 and structurally refined by ShelXL software. The solvent and
counterions are excluded for clarity in the crystal structure above.
Depicted ellipsoids are at the 50% probability level.
were measured from single crystals of the dimer complex. Due
to the lability of the complex upon solvation and/or supporting
electrolyte addition (Figures S6 and S12), solution measure-
ments were avoided.
Interestingly, along with the dimer complex a second side
product was observed. This came in the form of an insoluble
precipitate that would crash out if the dimer solution or a
[Co(PY5Me2)(CN)](OTf) solution that was allowed to sit for
extended periods of time. The insoluble product was also
readily obtained during the synthesis of the [Co(PY5Me2)-
(CN)](OTf) complex if the reaction mixture was carried out
at room temperature or allowed to stir for several minutes in a
cold bath under conditions where excess cyanide (>1.5 equiv)
was present. The thermodynamically stable species was
determined to be a cluster complex whose structure can be
found in Figure S1. Single crystals were difficult to isolate as
the cluster complex was only soluble in DMSO; however, tiny
single crystals were obtained from an attempt to grow
[Co(PY5Me2)(CN)](OTf) crystals over the course of several
days.
Figure 2. Single crystal representations of (a) [Co(PY5Me2)(CN)]-
(OTf) and (b) [Co(PY5Me2)(CN)](OTf)2 provided by Olex2 and
structurally refined by ShelXL software. The solvent and counterions
are excluded for clarity in each of the crystal structures above.
Depicted ellipsoids are at the 50% probability level.
the pure paramagnetic [Co(PY5Me2)(CN)](OTf) indicated
rather upfield chemical shifts that ranged from 4 to 20 ppm
with a single broad signal around 58 ppm that was only
identifiable at high concentrations (Figure S4). Oxidation of
[Co(PY5Me2)(CN)](OTf) to produce the stable Co(III)
product, [Co(PY5Me2)(CN)](OTf)2, was obtained using
silver triflate, AgOTf, in an acetonitrile solution. The short-
term stability of [Co(PY5Me2)(CN)](OTf) in neat acetoni-
trile coupled with the rapid reaction upon addition of Ag+
1
yielded a clean [Co(PY5Me2)(CN)](OTf)2 product via H
X-ray Crystallography. The crystal structures of [Co-
(PY5Me2)(CN)](OTf) and [Co(PY5Me2)(CN)](OTf)2 are
depicted in Figure 2. Refinement data for both complexes, as
well as the dimer complex discussed below, are summarized in
NMR (Figure S5). As expected, long-term stability is
maintained for [Co(PY5Me2)(CN)](OTf)2 in neat solution
as well as upon the addition of a supporting electrolyte (Figure
C
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
a
Table 1. Selected Bond Lengths and Angles for Co(PY5Me2)(CN)](OTf) and Co(PY5Me2)(CN)](OTf)2
bond distances and bond angles
[Co(PY5Me2)(CN)](OTf) A
Co(PY5Me2)(CN)](OTf) B
Co(PY5Me2)(CN)](OTf)2
Co−N1
Co−N2
Co−N3
Co−N4
Co−N5
Co−C30
C30−N6
1.977(2)
2.127(3)
2.066(3)
2.138(3)
2.088(3)
1.913(3)
1.128(4)
1.964(3)
2.074(3)
2.123(3)
2.097(3)
2.115(3)
1.917(4)
1.133(4)
1.992(3)
1.981(3)
1.981(3)
1.980(3)
1.973(3)
1.891(3)
1.151(4)
N2−Co−N3
N2−Co−N5
N3−Co−N4
N4−Co−N5
N2−Co−N4
N3−Co−N5
N1−Co−C30
N6−C30−Co
81.58(11)
99.27(10)
94.74(10)
84.21(10)
175.10(11)
176.37(11)
178.38(13)
177.7(3)
82.77(11)
95.64(11)
98.31(11)
83.15(11)
177.23(11)
176.54(11)
177.86(13)
177.3(3)
83.63(11)
96.65(11)
95.76(11)
83.99(10)
178.92(12)
178.66(11)
179.68(14)
178.9(3)
a
Bond lengths are reported in Angstroms (Å) and bond angles are in degrees (deg). The standard deviations of each value are shown in
parentheses. Each of the N1−Co−Nx (x = 2−5) bond angles are not listed since each value is nearly 90° ( 1−2°).
Table S2. The atom labeling is kept consistent for the
[Co(PY5Me2)(CN)](OTf) and [Co(PY5Me2)(CN)](OTf)2
complexes in order to make side-by-side structural compar-
isons. Selected bond lengths and angles for each structure can
be found in Table 1. Around the equatorial plane, nitrogen
atoms N2−5 of their respective pyridine subunits have been
appropriately assigned with the nitrogen atom N1 being
associated with the pyridine unit axial to the exogenous
cyanide ligand labeled C30 and N6.
labeled to make side-by-side comparisons to the monomeric
complex. Both of the axial bonds Co1−N1 and Co1−C30 appear
to be longer than the monomer’s by ∼0.07 and ∼0.05 Å,
respectively. Around the equatorial plane, the average Co−
N2−5 bond lengths are nearly identical between the two
complexes, deviating only by ∼0.009 Å. Interestingly, the
average equatorial Co−N8−11 bond length around the Co2 is
actually ∼0.017 Å shorter than the average bond length for
nitrogens bound to Co1. Although the formal negative charge
of the CN− resides on the carbon atom, suggesting a tighter
bond between Co1 and C30, the bond length between these
two atoms is only ∼0.024 Å shorter than the Co2−N6 bond.
The delocalized charge throughout the cyanide bridge also
causes the C30N6 triple bond to weaken and expand. Both
the Co1−C30−N6 and Co2−N6−C30 bond angles are measured
to be the same (∼177.6°), however, the N1−Co1−C30 bond
angle is slightly more acute (∼0.8°) than the N6−Co2−N7
bond angle. As with the monomeric complex, both cobalt
metal centers lie slightly above the equatorial plane as each of
the equatorial pyridines are less than 90° to the axial pyridines.
Even though it is not shown in Figure 3, three triflates were
found per dimer molecule, which would imply that each metal
center is in its reduced state, i.e., Co(II).
Upon inspection of the [Co(PY5Me2)(CN)]2+/+ crystal
structures, it appears that coordination of the exogenous
cyanide to the sixth coordination site of the parent [Co-
(PY5Me2)(MeCN)](OTf)2 yields a distorted octahedral
structure. In the case of the [Co(PY5Me2)(CN)](OTf)
complex, two independent molecules make up the asymmetric
unit and are both represented above. Superposition of these
two molecules leads to nearly indistinguishable structures with
minor bond angle and/or bond length changes, given in Table
1. The cobalt(II) metal center resides slightly above the
equatorial plane as each of the pyridine units (N2−N5) are
slightly less (∼1°) than 90° from the axial pyridine (N1).
Constrictive bond angles are observed for the pyridine units
bound through the ethyl bridge and are rather acute for the
cobalt(II) (81.6−82.8°). Oxidation of the cobalt(II) leads to a
contraction of the equatorial pyridines, which widens the N2−
Co−N3 bond angle creating a more symmetric complex. The
average Co−N bond length change of the four pyridines in the
equatorial plane is ∼0.125 Å. A minor bond length change is
observed for the axial pyridine unit upon oxidation. Also, as the
axial pyridine (N1) of the PY5Me2 expands, the more
electropositive Co(III) causes the Co−C bond length of the
cyanide (C30) to contract by nearly the same distance.
Consequently, the shorter Co−C bond length causes the C−
N triple bond to become slightly longer (∼0.02 Å).
Magnetic Properties. Magnetic susceptibility of the
paramagnetic [Co(PY5Me2)(CN)](OTf) complex was meas-
1
ured in acetonitrile-d3 by H NMR (Figure S9) using the
Evans method.23,24 Measurements were collected using a
regular NMR tube containing a known concentration of
[Co(PY5Me2)(CN)](OTf) dissolved in acetonitrile-d3 along
with a capillary insert filled with a saturated solution of
ferrocene (diamagnetic standard) also dissolved in acetonitrile-
d3. The concentration of [Co(PY5Me2)(CN)](OTf) was
varied at room temperature to provide the standard deviation
in the calculated effective magnetic moment, μeff. The μeff for
[Co(PY5Me2)(CN)](OTf) was calculated using eqs 1 and 2,
where χM is the molar susceptibility of the solute, Δν is the
observed frequency shift of the reference resonance (hertz), ν0
is the spectrometer frequency (s−1), c is the concentration of
[Co(PY5Me2)(CN)](OTf) (mol/L), χp is the paramagnetic
contribution to the molar susceptibility of the solute, and T is
the temperature (K) of the sample.
As mentioned in the synthesis section above, the labile CN−
ligand of the [Co(PY5Me2)(CN)](OTf) complex in acetoni-
trile leads to the formation of a dimer complex (Figure 3).
Dark red crystals of this complex were readily obtained for
single crystal X-ray analysis, and the resulting bond lengths and
bond angles for this complex are reported in Table S3. Each of
the nitrogen atoms, N1−5, bound to Co1 are appropriately
D
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 4. Raman spectra using single crystals of (a) [Co(PY5Me2)(CN)](OTf) (red line) and Co(PY5Me2)(CN)](OTf)2 (green line) as well as
(b) the dimer complex (blue line).
Table 2. Summary of Cyano, CN, Stretching Frequencies Using Raman Spectroscopy and Measuring Single Crystals of
[Co(PY5Me2)(CN)](OTf), [Co(PY5Me2)(CN)](OTf) and the Dimer Complex
method
[Co(PY5Me2)(CN)](OTf)
[Co(PY5Me2)(CN)](OTf)2
2140
dimer complex
2113
2106
2114
Raman (cm−1
)
3000Δν
(OTf) complex, splitting of the cyanide signal is also observed.
Given that there are two independent molecules in the
asymmetric unit of the [Co(PY5Me2)(CN)](OTf) crystal, we
speculate that the slight differences in localized environment,
i.e., solvent, counterions, bond distances, or bond angles, could
change the polarizability of the CN bond, which results in the
two distinct vibrational signals at very similar wavenumbers.
UV−Vis Spectroscopy. The UV−vis of the monomeric
[Co(PY5Me2)(CN)](OTf) (red line) and [Co(PY5Me2)-
(CN)](OTf)2 (green line) complexes are shown in Figure 5.
χM
=
4πν0c
(1)
(2)
μ
= 2.828 χpT
eff
Given that the diamagnetic contribution, χd, is generally small
and negligible compared to the overall paramagnetic
contribution, χp, μeff was therefore determined directly from
χM providing a μeff = 1.91 0.02 μB.25 The spin-only magnetic
moment, μso, for LS cobalt(II) complexes is calculated to be
μso = 1.73 μB suggesting that the experimentally determined
value supports a LS cobalt(II) complex.
Vibrational Spectroscopy. Raman spectroscopy was used
to further characterize the monomeric and dimer complexes,
Figure 4. In both measurements, single crystals of each
complex were used. Of particular interest was identifying the
CN vibrational frequencies for each complex. The dimer
showed a single strong signal for the CN stretch at 2113 cm−1
(Figure 4b), which was slightly blue-shifted from the primary
CN vibrational signal of the Co(II) monomer at 2106 cm−1.
Oxidation of the Co(II) monomer also resulted in a blue-
shifted cyano stretch frequency at 2140 cm−1. Looking at the
crystal structures of the monomeric complexes, it appears that
the CN bond length elongates upon oxidation from Co(II) to
Co(III). This may seem counterintuitive to a general statement
that the longer bond length would result in a lower vibrational
frequency at the same bond order; however, previous
investigations concluded that the CN vibrational frequency
increase is due to the force constant increase of the bond.26−28
Oxidation of the Co(II) metal center leads to a higher energy
signal due to the decreased π backbonding ability of the metal
center leading to less antibonding character on the cyanide
ligand (Table 2).
Figure 5. UV−vis spectra of [Co(PY5Me2)(CN)](OTf) (red line)
and [Co(PY5Me2)(CN)](OTf)2 (green line) measured under air free
conditions in acetonitrile. Inset: the d−d transition of the [Co-
(PY5Me2)(CN)](OTf)2 complex.
The cobalt(II) complex, [Co(PY5Me2)(CN)](OTf), contains
two metal-to-ligand charge transfer (MLCT) bands in
acetonitrile between 300−400 nm (ε > 2000 M−1 cm−1).
However, the only significant absorption feature of the
[Co(PY5Me2)(CN)](OTf)2 UV−vis is a weak multifeatured
d−d transition band at ∼442 nm; see Figure 5 inset.
Presumably, this band resides in the [Co(PY5Me2)(CN)]-
Two additional CN signals from the expected rise in the
Raman for both monomeric complexes (Figure 4a). These CN
signals, at 2254 cm−1, can be attributed to trapped acetonitrile
in the crystal lattice and agree well with the crystal structures
which show one molecule of acetonitrile per molecule of
monomeric complex. In the case of the [Co(PY5Me2)(CN)]-
E
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(OTf) UV−vis spectrum but is obscured by the onset of the
MLCT band. In both spectra, below 300 nm a strong
absorption band is present and is attributed to a metal-
independent ligand-based π−π* transition.22,29 Such an
assignment is made due to the absorption of the free
PY5Me2 ligand in acetonitrile (Figure S10), which has a λmax
at 263 nm and an ε > 15 000 M−1 cm−1.
Electrochemical Properties. Cyclic voltammetry (CV)
was used to probe the redox behavior of each cobalt
pentapyridine complex synthesized. To avoid degradation or
potential oxidation, each CV was measured under an N2
atmosphere. Similar to previous literature reports, [Co-
(PY5Me2)(MeCN)](OTf)2 reveals a quasi-reversible redox
wave at 0.822 V vs NHE, Figure 6 (dark blue dashed line),
that the molecule is redox inactive, while the cobalt(I)/(II)
transition is seen at −0.845 V vs NHE, ruling out the
possibility of these being the origin of the observed
shoulder.19,20 Given the simplicity of the system, it is
speculated that the shoulder could be attributed to a
counterion (CI) coordinated complex, [Co(PY5Me2)(OTf)],
or the five coordinate [Co(PY5Me2)]2+ complex vacant of a
species bound to the sixth coordination site.19
Upon isolation of the [Co(PY5Me2)(CN)](OTf) complex,
three quasi-reversible redox waves are observed when perform-
ing the electrochemistry in acetonitrile with 0.1 M TBAPF6,
Figure 6 (red line). The largest and most negative wave at
0.254 V vs NHE is assigned to the cobalt(II)/(III) redox
potential of the [Co(PY5Me2)(CN)]+/2+ complex. Even with
the use of single crystals, it appears that after solvation and
addition of a supporting electrolyte small amounts of the
[Co(PY5Me2)(MeCN)]2+ and what we speculate to be the
counterion (CI) coordinated, [Co(PY5Me2)(OTf)]+, exist in
solution. This phenomenon becomes more evident upon
overlaying the two CVs as in Figure 6. The equilibrium of
these two species is observed in the electrochemistry of the
dimer complex as well (Figure S12). Although the cobalt(II)/
(III) wave of the two metal centers is predominant and occurs
successively between 0.2−0.4 V vs NHE, upon further anodic
sweeping the redox waves for both the [Co(PY5Me2)-
(MeCN)]2+ and [Co(PY5Me2)(OTf)]+ also appear. Unlike
[Co(PY5Me2)(CN)]+, measuring the initial CV of the
[Co(PY5Me2)(CN)]2+ complex resulted in the observation
of a single [Co(PY5Me2)(CN)]2+/+ redox process, indicating
the stability of the oxidized monomer in supporting electrolyte.
However, successive CVs of the [Co(PY5Me2)(CN)]2+
complex led to the observed side-product formation shown
in Figure 6 as the concentration of [Co(PY5Me2)(CN)]+ built
up in solution.
Figure 6. CV data of [Co(PY5Me2)(CN)](OTf) (red line) and
[Co(PY5Me2)(MeCN)](OTf)2 (dark blue dashed line) measured in
acetonitrile with 0.1 M TBAPF6 using a platinum disk working
electrode, a platinum mesh counter electrode, and a lab-made Ag/
AgNO3 (0.1 M TBAPF6 acetonitrile) reference electrode.
Cross-Exchange Kinetics. Stopped-flow spectroscopy was
used to determine the homogeneous electron-transfer self-
exchange rate constant for [Co(PY5Me2)(CN)]2+/+. Although
stability has been demonstrated to be an issue with
[Co(PY5Me2)(CN)]+ while conducting electrochemical stud-
ies, short-term stability of the complex was confirmed in neat
corresponding to the cobalt(II)/(III) oxidation.20 Interestingly
however, a small shoulder is also observed in the [Co-
(PY5Me2)(MeCN)](OTf)2 CV, though it is difficult to see in
Figure 6 and never mentioned in prior reports. Previous
electrochemical studies done on the free PY5Me2 ligand reveal
1
acetonitrile over the course of a day via H NMR studies. By
neglecting the use of a supporting electrolyte, 1H NMR studies
validated the reliability of carrying out stopped-flow studies
Figure 7. (a) Plot of absorbance at 505 nm vs time, corresponding to the growth of the [Co(terpy)2]2+ species (red dots) and the resulting single
exponential fit (black line) for the reduction of [Co(terpy)2]3+ (4.0 × 10−5 M) by [Co(PY5Me2)(CN)]+ (1.2 × 10−3 M). (b) Pseudo-first-order
rate constants, kobs, versus the excess concentration of [Co(PY5Me2)(CN)]+ for the reactions between [Co(PY5Me2)(CN)]+ and [Co(terpy)2]3+.
F
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 8. Current density (J) versus applied potential (E) plots for [Co(PY5Me2)(CN)](OTf)2 (green dots) and [Co(bpy)3](PF6)3 (light blue
dots) OSRSs measured using a mesoporous TiO2 working electrode, a platinum mesh counter electrode, and a lab-made Ag/AgNO3 (0.1 M
TBAPF6) reference electrode in an acetonitrile solution with 0.1 M LiOTf.
over the course of a few hours in neat acetonitrile. To isolate
k12
≅
k11k22K12
(4)
the self-exchange rate constant for [Co(PY5Me2)(CN)]2+/+
using stopped-flow, a series of cross-exchange reactions
between [Co(PY5Me2)(CN)]+ and cobalt bis(2,2′:6′,2′′-
terpyridine), [Co(terpy)2]3+, were performed, which provided
the cross-exchange rate constant, k12, for reaction R1 below:
where k22 is the self-exchange rate constant of [Co-
(terpy)2]3+/2+, and K12 is the equilibrium constant for the
electron-transfer reaction. The Marcus cross-relation shown
above has been modified to neglect the nonlinear correction
term, f12, and the electrostatic work term, W12, which are a
function of the reaction mixture’s ionic strength. Given the
issue of CN− dissociation upon introduction of a supporting
electrolyte, stopped-flow solutions were made neat and the f12
and W12 terms were ignored. Without supporting electrolyte,
however, it is expected that reduced coupling will result, and an
underestimate of the self-exchange rate constant, k11, for
[Co(PY5Me2)(CN)]+ will be made. The equilibrium constant,
K12, for the electron-transfer reaction can be determined based
on the free-energy difference of reaction R1 and can be
described by
[Co(PY Me2)(CN)]+ + [Co(terpy)2]3+
5
k12
⎯⎯→⎯ [Co(PY Me2)(CN)]2+ + [Co(terpy)2]2+
(R1)
5
Selection of [Co(terpy)2]3+ for the cross-exchange with
[Co(PY5Me2)(CN)]+ was based on the complex’s known
outer-sphere one-electron transfer mechanism and slow
electron-transfer kinetics.29 Given the large potential difference
between [Co(terpy)2]3+/2+ and [Co(PY5Me2)(CN)]2+/+
(Table S4), the reaction was assumed to reach completion
without an appreciable back reaction. Although the large
driving force facilitates faster electron-transfer kinetics, low
concentrations of the reactants provided sufficient signal and
observable decays on the time scale of stopped-flow, which was
a result of the large extinction coefficient of the [Co(terpy)2]2+
species formed in solution (Figure S11).30,31 Figure 7a shows a
−nFΔE = −RT ln K12
(5)
where n is the number of electrons transferred (n = 1), F is
Faraday’s constant, ΔE is the formal potential difference
between the oxidant and reductant in solution, R is the gas
constant, and T is the temperature. CVs shown in Figure S13
and summarized in Table S4 indicate a 285 mV formal
potential difference between [Co(PY5Me2)(CN)]2+/+ and
[Co(terpy)2]3+/2+. The calculated equilibrium constant for
cross-exchange reaction R1 is therefore (6.6 0.9) × 104. The
self-exchange rate constant, k22, for [Co(terpy)2]3+/2+ was
measured independently by crossing the complex with 1,1′-
dimethylferrocene [Fe(C5H4CH3)] (reaction S1) under
similar conditions, i.e., neat acetonitrile. Details regarding the
reaction mixtures, experimental design and the resulting
equilibrium and kinetic rate constants can be found in the
Supporting Information and the Experimental Section below.
On the basis of these stopped-flow studies, the self-exchange
rate constant, k22, for [Co(terpy)2]3+/2+ was calculated to be 41
9.9 M−1 s−1. Such a value matches well with those of prior
literature reports.34,35 Using this experimentally determined
self-exchange rate constant, k22, for [Co(terpy)2]3+/2+ and the
calculated equilibrium constant, k12, the self-exchange rate
constant for [Co(PY5Me2)(CN)]2+/+, k11, was calculated to be
obst
single exponential fit, A = A∞+(Ao−A∞) e−k , to a plot of the
absorbance at 505 nm versus time, which corresponds to the
growth of the [Co(terpy)2]2+ species due to the reduction of
[Co(terpy)2]3+ by [Co(PY5Me2)(CN)]+. In all reactions, the
[Co(PY5Me2)(CN)]+ species was held in excess of [Co-
(terpy)2]3+, which allowed the observed rate constants, kobs, to
be expressed by
kobs = k12[Co(PY Me2)(CN)]+
(3)
5
Figure 7b shows a straight line fit of the kobs values plotted as a
function of the [Co(PY5Me2)(CN)]+ concentration and
provided the value for the forward rate constant, k12, from
the slope, respectively. The initial concentrations for the
[Co(PY5Me2)(CN)]+ and [Co(terpy)2]3+ reaction mixtures,
as well as the observed pseudo-first order rate constants for
each of these electron-transfer reactions can be found in Table
S5.
Using the experimentally determined cross-exchange rate
constant, k12, for reaction R1 above, the Marcus cross-relation,
eq 4, was used to calculate the self-exchange rate constant, k11,
for [Co(PY5Me2)(CN)]+:32,33
20
5.5 M−1 s−1. This measured self-exchange value for
[Co(PY5Me2)(CN)]2+/+ is surprisingly small. Isoelectronic
cobalt(II) complexes such as [Co(ttcn)2]2+ sustain self-
G
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
exchange rate constants several orders of magnitude larger than
these measurements indicate that the kinetics for interfacial
recombination is significantly faster for [Co(bpy)3]3+ com-
pared to that for [Co(PY5Me2)(CN)]2+. While the [Co-
(PY5Me2)(CN)]2+/+ cells outperformed the [Co(bpy)3]3+/2+
cells, the uncertainty in the solution potential and concen-
trations of the various redox active species present at the TiO2
surface prevented us from being confident in making solid
conclusions; these data were therefore omitted.
9,36
that of [Co(PY5Me2)(CN)]2+/+
.
A few explanations could
reconcile this observed phenomenon and will be discussed in
detail below. One explanation is that the slow self-exchange
kinetics are due to a large inner-sphere reorganization that the
complex undergoes upon oxidation. Looking at the crystal
structures above, it appears that the equatorial pyridine units of
the PY5Me2 ligand significantly contract (∼ 0.1 Å) when going
from cobalt(II) to cobalt(III). The large structural change
could inhibit the expected fast electron self-exchange, which
results in a much smaller self-exchange rate constant.
Recombination Kinetics. To mimic the recombination
reactions at a TiO2 interface, as in operating DSSCs, half-cells
were constructed as in Figure S16. The three-electrode setup
provided the opportunity to conduct dark recombination
studies to better understand the kinetics of interfacial charge
transfer. As depicted, the three-electrode setup was constructed
such that a thin mesoporous film of TiO2, deposited onto an
FTO substrate, acted as a working electrode, while a platinum
mesh was used as a counter electrode and a lab-made Ag/
AgNO3 electrode was used as a reference. Use of a reference
electrode was important to these studies since it afforded a way
to compare the current density, J, for each redox shuttle, at the
same TiO2 potential, regardless of any differences in solution
potential there may have been. By directly measuring J the
differences in recombination rate constants, ket, to the
dissolved acceptor, at any given applied potential, E, could
be identified, assuming the density of conduction band
electrons, ns, at any given applied potential and the initial
acceptor/redox shuttle concentrations, [A], were the same:36
DISCUSSION
■
Rational design of new OSRSs is pivotal to the development of
next generation DSSCs. Control over the coordination
environment of cobalt OSRSs provides the ability to finely
tune the charge-transfer kinetics of these complexes, which
dictate the overall rates of regeneration and recombination in
dye cells. Use of the pentadentate PY5Me2 ligand affords a
ligand periphery that enables the binding of a multitude of
exogenous ligands, which can control the redox chemistry and
spin-state of the cobalt metal center. In an effort to synthesize a
new LS Co(II) OSRS that was expected to sustain fast
electron-transfer kinetics and a more negative formal potential
to inhibit recombination, cyanide, CN−, was chosen as the
exogenous ligand.
In theory, the synthesis of [Co(PY5Me2)(CN)](OTf) from
the parent [Co(PY5Me2)(MeCN)](OTf)2 was expected to be
facile and clean via the addition of a cyanide source. To our
surprise, however, the cyanide ligand was observed to be much
more labile than anticipated. Careful control over the reaction
conditions was necessary in order to mitigate dimerization
and/or subsequent cluster formation of a cobalt hexacyanide
complex. If the reaction between [Co(PY5Me2)(MeCN)]-
(OTf)2 and CN− (>1.5 equiv) was allowed to stir for an
extended period of time in an ice bath or was attempted at
room temperature, then the thermodynamically stable cluster
complex was observed to crash out of solution. Structural
support of the complex came from X-ray crystallography,
Figure S1. Similar structures, referred to as “star-like clusters”,
have been reported in the literature by the Long group in an
effort to study magnetic exchange.37 Through the reaction of
[(PY5Me2)V(MeCN)]2+ with [M(CN)6]3− (M = Cr, Mo), a
cis cyano cluster [(PY5Me2)V4M(CN)6]5+ was obtained,
which is structurally equivalent to the [(PY5Me2)4Co4Co-
(CN)6]4+ cluster complex that is isolated from our experi-
J(E) = −qket[A]ns
(6)
The measured current density plots as a function of applied
potential, J−E, for the half-cells employing [Co(PY5Me2)-
(CN)](OTf)2 (green dots) and [Co(bpy)3](PF6)3 (light blue
dots) OSRSs can be found in Figure 8 above.
Selection of [Co(bpy)3]3+ has become the benchmark for
our side-by-side comparisons of new OSRSs as it has emerged
as the champion redox shuttle, along with the fact that it is
expected to employ the same one electron-transfer mechanism
as [Co(PY5Me2)(CN)]2+. To ensure accurate measurements
were being acquired, CVs were measured before and after the
dark recombination studies, which indicated the reference was
stable and the redox shuttles were well-behaved. We note that
carrying out a single dark recombination measurement using
[Co(PY5Me2)(CN)]2+ resulted in a negligible concentration
of side-products as [Co(PY5Me2)(CN)]+ was produced.
However, successive CV scans, as mentioned above, resulted
in an obvious evolution of side-products, similar to the
phenomena observed in Figure 6. On the basis of the CVs, the
formal potentials of [Co(PY5Me2)(CN)]2+/+ and [Co-
(bpy)3]3+/2+ were determined to be 0.254 and 0.590 V vs
NHE, respectively. Given the dark J−E curve for [Co-
(bpy)3]3+/2+ was characteristic of that previously measured
for [Co(Me2bpy)3]3+/2+ DSSCs further validated the J−E
behavior measured for [Co(PY5Me2)(CN)]2+.8 Although rapid
stirring was imparted during each measurement, the dark
current density did deviate from ideal behavior at more
negative potentials due to solution resistance. A comparison of
both J−E curves indicates that the onset for recombination and
the magnitude of dark current for [Co(bpy)3]3+ is more
positive and significantly larger than [Co(PY5Me2)(CN)]2+.
Since the current density is a direct measure of recombination,
1
ments. During H NMR measurements, used to probe the
stability of [Co(PY5Me2)(CN)](OTf) over a couple day
period, precipitation of what we expect to be free ligand and
the cluster complex is observed (Figure S15) in neat
acetonitrile and in acetonitrile with supporting electrolyte,
i.e., 0.1 M TBAPF6. Interestingly, decomposition of [Co-
(PY5Me2)(CN)](OTf) is facilitated more rapidly upon
addition of the TBAPF6 supporting electrolyte, and after
1
several days the H NMR provides chemical shifts for the free
PY5Me2 ligand, the dimer complex, as well as the [Co-
(PY5Me2)(CN)](OTf) complex originally being studied
(Figure S7). Formation of the dimer complex indicates free
cyanide being liberated into solution as well as the presence of
the solvato complex, [Co(PY5Me2)(MeCN)]2+. Given the
lability of Co(II) and the steric strain of the equatorial
pyridines (N2−Co−N3) (Table 1), the formation of [Co-
(CN)6]4− seems feasible by way of excess cyanide displacing
the PY5Me2 ligand. Any available [Co(PY5Me2)(MeCN)]2+ is
then expected to coordinate via the accessible lone pair of the
nitrogen Lewis base on the cyanide ligands of the [Co-
H
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(CN)6]4− complex formed in solution. This would provide an
explanation for the observed precipitate and the lack of
reduction supported a LS Co(II) to LS Co(III) transition,
which corroborate the data collected for magnetic suscepti-
bility. The average bond length change of the Co−N bonds of
the pyridine units is only ∼0.104 Å, which is significantly
smaller than those reported for well-known HS Co(II) redox
shuttles such as [Co(bpy)3]2+ and [Co(phen)3]2+ that have
known Co−N bond length changes of ∼0.19 Å upon
oxidation.16,38,39 Previously reported Co(II) redox shuttles
1
appreciable [Co(PY5Me2)(MeCN)]2+ found in the H NMR
spectrum.
Modification of the reaction conditions such that only one
or less equivalents (≤1 equiv) of CN− are added to the
[Co(PY5Me2)(MeCN)](OTf)2 reaction mixture at low
temperature results in the predominant isolation of a dimer
complex. Single crystals were easily grown and the structural
integrity was confirmed via X-ray crystallography (Figure 3).
Vibrational studies of the dimer complex revealed a single
sharp CN signal in the Raman spectrum that was slightly blue-
shifted from the monomeric Co(II) complex. 1H NMR studies
using single crystals of the dimer complex (Figure S6)
provided a spectrum different from that measured for
[Co(PY5Me2)(CN)](OTf) (Figure S4). With chemical shifts
as far downfield as 80 ppm, we speculate that the dimer
complex is HS; however, any attempt to measure the magnetic
susceptibility via the Evans method would be difficult as there
are clearly multiple species that form in solution upon
solvation of the pure complex. When overlaid, the chemical
shifts of both the [Co(PY5Me2)(MeCN)](OTf)2 and [Co-
(PY5Me2)(CN)](OTf) (Figures S3 and S4) align well with
the various chemical shifts found in the dimer spectrum
(Figure S6). Electrochemical measurements of the dimer
with similar structures such as [Co(PY5Me2)(NMBI)]/2+
,
where NMBI represents N-methylbenzimiazole, also contained
larger average Co−N bond length changes of ∼0.150 Å.21
With such small average Co−N bond length changes of the
[Co(PY5Me2)(CN)]2+/+ complex, one would expect intrinsi-
cally fast self-exchange kinetics; however, this does not appear
to be the case. Using stopped-flow spectroscopy, the calculated
self-exchange rate constant, k11, for [Co(PY5Me2)(CN)]2+/+
was determined to be only 20 5.5 M−1 s−1. This is orders of
magnitude lower than the value (∼1.1 × 104 M−1 s−1)
determined for another isoelectronic LS Co(II) complex,
[Co(ttcn)2]3+/2+, also measured using stopped-flow spectros-
copy.9,16 Slower self-exchange kinetics were expected for the
[Co(PY5Me2)(CN)]2+/+ cross-exchange reactions given the
lack of supporting electrolyte. However, such a drastic
difference in self-exchange rates is unlikely to be strictly due
to the increased work function associated with electron-
transfer. Thus, we reasoned from the crystal structures and
UV−vis data that the slower observed kinetics were likely due
to a Jahn−Teller distortion of the [Co(PY5Me2)(CN)]+
complex. As mentioned above, the equatorial pyridines go
through a rather large contraction (∼0.125 Å) upon oxidation,
while the axial bonds change minimally, suggesting the
complex undergoes a Jahn−Teller compression, as in Scheme
1.
1
complex (Figure S12) rectify the phenomena observed by H
NMR. Though it is difficult to assess the abundance of
[Co(PY5Me2)(CN)](OTf), as there are two successive redox
waves atop the expected formal potential of [Co(PY5Me2)-
(CN)](OTf), it is clear that [Co(PY5Me2)(MeCN)](OTf)2
and what is speculated to be the [Co(PY5Me2)(OTf)](OTf)
complex are both present, which confirms the lability of the
dimer complex in solution.
Isolation of the pure [Co(PY5Me2)(CN)](OTf) complex
was achieved by ensuring the reaction between [Co(PY5Me2)-
(MeCN)](OTf)2 and CN− was carried out using a slight
excess of CN−, at low temperature and in a noncompetitive
solvent. After complexation of the exogenous cyanide, the
complex was determined to be stable for several hours in neat
acetonitrile, which enabled magnetic susceptibility measure-
ments to be carried out. However, introduction of any
supporting electrolyte immediately induced the dissociation
of cyanide, Figure 6, and the conversion to the dimer and/or
cluster complex. Surprisingly, even the synthesis of the
oxidized [Co(PY5Me2)(CN)](OTf)2 complex needed to be
carefully completed. Initial attempts to oxidize the parent
[Co(PY5Me2)(MeCN)](OTf)2 complex with Ag+ resulted in
unwanted side products that readily formed purple crystals
suitable for X-ray crystallography, Figure S2. As was previously
observed, oxidation of the parent [Co(PY5Me2)(MeCN)]-
(OTf)2 resulted in chemistry occurring with the counterion in
solution, which liberated fluoride to produce a [Co(PY5Me2)-
(F)](OTf)2 complex.21 Therefore, to obtain the pure [Co-
(PY5Me2)(CN)](OTf)2 complex, the synthesis was carried
out by oxidizing [Co(PY5Me2)(CN)](OTf) with AgOTf. This
yielded a clean product since any potential AgCN that
precipitated out of solution was filtered off with the
precipitated silver solid and any unreacted [Co(PY5Me2)-
(CN)](OTf) or [Co(PY5Me2)(MeCN)](OTf)2 could be
neatly removed by washing the crude [Co(PY5Me2)(CN)]-
(OTf)2 powder with dichloromethane. An in-depth analysis of
the [Co(PY5Me2)(CN)]2+/+ crystal structures appeared to
suggest the changes in bond distances upon oxidation or
Scheme 1. Splitting of the d-Orbitals Based on the
Hypothesized Jahn−Teller Compression of the
[Co(PY5Me2)(CN)]+ Complex
It is difficult, however, to verify such phenomena using
simple UV−vis measurements of the [Co(PY5Me2)(CN)]+
species. The strong broad visible absorption coupled with the
MLCT transitions between 300 and 400 nm masks any
noticeable d−d transitions. Interestingly, even though LS
Co(III) is not supposed to Jahn−Teller distort, the multi-
featured d−d transition, Figure 5 inset, implies nondegenerate
d-orbitals with more than one electronic transition. This is
expected to be more obvious in the [Co(PY5Me2)(CN)]+
UV−vis but again remains hidden by the broad visible
absorption band. Aside from the complex’s ability to go
I
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
through a Jahn−Teller compression, another possible explan-
ation for the smaller observed self-exchange rate constant
could be that the complex transfers charge via an inner-sphere
mechanism rather than an outer-sphere mechanism. In all
assumptions above, it is thought that the activation complexes
are two separate entities; however, it is difficult to rule out the
possibility that [Co(PY5Me2)(CN)]2+/+ does not complex to
transfer electrons. The accessible lone pair of the nitrogen
Lewis base has been shown to dimerize in solution, which
would suggest it could pair with the acidic cobalt metal center
of the [Co(terpy)2]3+/2+ complex when carrying out the cross-
exchange reaction.
Although the self-exchange kinetics of [Co(PY5Me2)-
(CN)]2+/+ are slower than expected, it still remains an
interesting redox shuttle for regenerating IR absorbing
sensitizers. Simple dark J−E measurements, shown in Figure
8 above, compare [Co(PY5Me2)(CN)]2+ to [Co(bpy)3]3+ and
qualitatively demonstrate that the recombination rates to
[Co(PY5Me2)(CN)]2+ are much slower than those to
[Co(bpy)3]3+. According to eq 6, the dark current that is
measured for each shuttle is directly proportional to the rate
constant for recombination, ket, at any given applied bias of the
TiO2 electrode.37 Thus, after comparing the dark currents
between the two different redox shuttles at the same applied
bias, it appears that at the formal potential of [Co(PY5Me2)-
(CN)]2+ (0.254 V vs NHE) the recombination rate constant is
over 3 orders of magnitude larger for [Co(bpy)3]3+ than
[Co(PY5Me2)(CN)]2+. Additionally, nearly 0.380 V must be
applied past the formal potential of [Co(PY5Me2)(CN)]2+,
while only ∼0.270 V must be applied past the formal potential
of [Co(bpy)3]3+ in order to reach the same magnitude of dark
current, −1 mA cm−2. From these dark recombination
measurements, it seems clear that [Co(bpy)3]3+ is a much
better acceptor than [Co(PY5Me2)(CN)]2+. Such a conclusion
can be attributed to the reduced driving force for
recombination of conduction band electrons and the intrinsi-
cally small self-exchange rate constant of [Co(PY5Me2)-
(CN)]2+. Given the observed kinetic behavior, if introduced
into DSSCs, superior charge collection is expected for cells
containing [Co(PY5Me2)(CN)]2+/+ coupled with the properly
integrated IR absorbing sensitizer.
exchange rate constant (k11 = 20
5.5 M−1 s−1) was
hypothesized to arise from either a Jahn−Teller compression
observed by collecting single crystals of the [Co(PY5Me2)-
(CN)]2+/+ complexes and/or a more complicated inner-sphere
mechanism via complexation through the nitrogen lone pair of
the exogenous cyanide ligand. Dark J−E measurements
suggested that rates of recombination to the oxidized redox
shuttle, [Co(PY5Me2)(CN)]2+ are actually slower than the
champion redox shuttle, [Co(bpy)3]3+, when compared side-
by-side. As a result, the [Co(PY5Me2)(CN)]2+/+ redox shuttle
becomes an attractive candidate as a solid-state hole conductor
for DSSCs with the promise of achieving quantitative charge
collection, while also having the ability to successfully
regenerate near IR and/or IR absorbing sensitizers.
EXPERIMENTAL SECTION
■
Materials. All reagents were obtained from commercial suppliers
(Oakwood Chemical, Sigma-Aldrich, Alfa Aesar or Strem Chemicals)
and used as received unless otherwise stated. Solvents used in the
synthesis, characterization, and kinetics studies of all cobalt complexes
were dried prior to being stored in a glovebox (MBRAUN Labmaster
SP). Tetrahydrofuran (Fisher Chemical, Optima) and diethyl ether
(Anhydrous, ACS Reagent, ≥99.0%) were distilled over sodium/
benzophenone. Methanol was dried by reacting magnesium turnings
and iodine, then distilling under nitrogen and storing over 3 Å
molecular sieves. Acetonitrile (Fisher Chemical Certified ACS,
≥99.5%) and dichloromethane (Macron Fine Chemicals AR ACS)
were purified by being passed through an activated alumina column.
The supporting electrolytes, tetrabutylammonium hexafluorophos-
phate (Sigma-Aldrich, 98%, TBAPF6) and lithium triflate (Sigma-
Aldrich, 99.995% trace metals basis, LiOTf), were stored in a
glovebox under moisture free conditions prior to use. However,
before storing in the glovebox, TBAPF6 was recrystallized from
ethanol/diethyl ether and dried under vacuum.
Instrumentation. CHN analysis was conducted at Michigan State
University. UV−vis spectra were measured with a PerkinElmer
Lambda 35 UV−vis spectrometer using 1 cm path length quartz
cuvettes. High-resolution mass spectra (HRMS) were obtained at the
Michigan State University Mass Spectrometry Service Center using a
Waters GCT Premier instrument run on electron ionization (EI)
direct probe or a Waters QTOF Ultima instrument run on
electrospray ionization (ESI+). Infrared spectroscopy (IR) was
obtained using a JASCO FT/IR-6600 spectrometer. Raman spectros-
copy was collected using a Renishaw inVia Raman microscope
employing a RL532C100 laser source. 1H NMR spectra were
measured at room temperature (25 °C) on an Agilent DirectDrive2
500 MHz spectrometer and referenced to residual solvent signals. All
coupling constants are apparent J values measured at the indicated
field strengths in Hertz (s = singlet, d = doublet, t = triplet, q =
quartet, dd = doublet of doublets, ddd = doublet of doublet of
doublets, td = triplet of doublets, m = multiplet). Cyclic voltammetry
(CV) measurements were performed with a μAutolabIII/FRA2
potentiostat using a platinum disk working electrode, platinum
mesh counter electrode, and a lab-made Ag/AgNO3 (0.1 M TBAPF6
acetonitrile) reference electrode. Ferrocene was used as an internal
reference. The error associated with each redox shuttle’s formal
potential, E°, is based on the standard deviation of the formal
potentials measured over three separate days. Reference conversion to
NHE was done assuming the potential of ferrocene in acetonitrile is
0.40 V vs SCE.40 Dark J−E measurements were obtained for both
[Co(PY5Me2)(CN)]2+ and [Co(bpy)3]3+ in acetonitrile with 0.1 M
lithium triflate, LiOTf, using a three-electrode setup interfaced with
the μAutolab mentioned above. The three-electrode setup contained a
mesoporous thin film of TiO2 nanoparticles attached to an FTO
substrate (fabrication described below) which acted as a working
electrode and a lab-made Ag/AgNO3 reference (described above),
along with a high surface area platinum mesh counter electrode.
Figure S16 gives a pictorial illustration of the setup.8
CONCLUSION
■
A new LS Co(II) redox shuttle has been synthesized and fully
characterized for its potential application in DSSCs. The new
class of cobalt redox shuttles shares the caveat that
coordination of the pentapyridine ligand, PY5Me2, affords
the opportunity to functionalize the sixth site of the cobalt
metal center with a variety of exogenous ligands that can not
only modulate the redox potential of the shuttle but also
manipulate the spin-state of the complex. In an effort to force
Co(II) to become LS, cyanide was chosen as the exogenous
ligand. Magnetic susceptibility measurements were used to
confirm the Co(II) complex, [Co(PY5Me2)(CN)](OTf), was
in fact LS upon isolating of the pure product. Interestingly, the
cyanide ligand was much more labile than expected in a
competitive coordinating solvent such as acetonitrile. Dis-
sociation of the cyanide resulted in dimerization and the
thermodynamically stable cluster complex. Without the use of
a supporting electrolyte to help facilitate cyanide dissociation,
the [Co(PY5Me2)(CN)](OTf) complex was stable enough in
neat acetonitrile to collect kinetic measurements using
stopped-flow spectroscopy. The unexpectedly slow self-
J
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
X-ray Crystallography. Crystals were mounted on a nylon loop
with paratone oil on a Bruker APEX-II CCD diffractometer. The
crystal was kept at T = 173(2) K during data collection. Using Olex2
(Dolomanov et al., 2009), the structure was solved with the ShelXS
(Sheldrick, 2008) structure solution program, using the Direct
Methods solution method. The model was refined with version
2014/6 of XL (Sheldrick, 2008) using Least Squares minimization. All
non-hydrogen atoms were refined anisotropically. Hydrogen atom
positions were calculated geometrically and refined using the riding
model. There are two independent molecules in the asymmetric unit
of the [Co(PY5Me2)(CN)](OTf) crystals. Structure and refinement
data are summarized in Table S2 for [Co(PY5Me2)(CN)](OTf),
[Co(PY5Me2)(CN)](OTf)2, the dimer complex, the Cluster
Complex and [Co(PY5Me2)(F)](OTf)2. The structures of the
Cluster Complex and [Co(PY5Me2)(F)](OTf)2 can be found in
Figures S1 and S2.
found (Calcd) for C31H25CoF3N6O3S: C, 54.47 (54.95); H, 3.71
(3.72); N, 11.90 (12.40).
[Co(PY5Me2)(CN)](OTf)2. [Co(PY5Me2)(CN)](OTf) (0.133
mmol, 89.9 mg) was dissolved in a small amount (∼5 mL) of
acetonitrile. A second acetonitrile solution (∼5 mL) of silver triflate,
AgOTf, (0.132 mmol, 34.0 mg) was made and slowly added to the
first. Fine gray silver particulate formed immediately after AgOTf
addition, and the reaction mixture turned from a dark reddish/brown
to a light orange solution. The mixture was allowed to stir for 2 h
before the silver particulate was syringe-filtered, and the crude
product was crashed with diethyl ether. The supernatant was decanted
and the pure light orange product (yield: 52.8%) was isolated by
washing with dichloromethane and ether. Crystals suitable for single
crystal X-ray diffraction analysis were obtained by slow vapor diffusion
of ether into a concentrated acetonitrile solution of [Co(PY5Me2)-
(CN)](OTf)2 at room temperature. Mass spectroscopy, elemental
analysis, and 1H NMR were also used to characterize the
[Co(PY5Me2)(CN)](OTf)2 complex. Intense peaks for [Co-
(PY5Me2)(CN)]2+ at 264.1 and [Co(PY5Me2)(CN)](OTf)+ at
677.1 were observed by mass spectrometry (M+). However, it
appeared even with pure material the mass spectroscopy also showed
peaks for the reduced complex: [Co(PY5Me2)]2+ at 251, [Co-
Synthesis of Parent Cobalt Complex. [Co(PY5Me2)(MeCN)]-
(OTf)2. Unless otherwise noted, all synthesis procedures were
performed under inert N2 atmosphere using Schlenk line or standard
glovebox techniques. The ligand PY5Me2 (2,6-bis(1,1-bis(2-pyridyl)-
ethyl)pyridine) was synthesized according to a procedure previously
reported in the literature.41 The synthesized ligand was characterized
1
(PY5Me2)(OTf)]+ at 651.1 and [Co(PY5Me2)(CN)]+ at 528.1. H
1
by H NMR and the resulting chemical shifts were matched to the
1
NMR (500 MHz, acetonitrile-d3) δ 9.91 (dd, J = 6.2, 1.4 Hz, 4H),
8.40 (dd, J = 8.7, 7.3 Hz, 1H), 8.31 (d, J = 8.0 Hz, 2H), 8.11 (ddd, J =
8.6, 7.3, 1.3 Hz, 4H), 8.04 (dd, J = 8.2, 1.7 Hz, 4H), 7.71 (ddd, J =
7.6, 6.2, 1.7 Hz, 4H), 2.81 (s, 6H). Elemental analysis: found (calcd)
for C32H25CoF6N6O6S2: C, 45.67(46.50); H, 3.17(3.05); N,
9.59(10.17).
literature report: H NMR (500 MHz, acetonitrile-d3) δ 8.45 (ddd, J
= 4.8, 1.9, 1.0 Hz, 4H), 7.61 (t, J = 7.9 Hz, 1H), 7.52 (ddd, J = 8.1,
7.5, 1.9 Hz, 4H), 7.14 (ddd, J = 7.4, 4.8, 1.1 Hz, 4H), 7.04 (d, J = 7.9
Hz, 2H), 6.91 (dt, J = 8.1, 1.0 Hz, 4H), 2.13 (s, 6H). The
[Co(PY5Me2)(MeCN)](OTf)2 complex (MeCN represents acetoni-
trile) was synthesized using a modified literature procedure. First,
[Co(PY5Me2)(I)]I was prepared, but only allowed to stir overnight
(∼12 h) before being collected. Finally, metathesis of [Co(PY5Me2)-
(I)]I to yield [Co(PY5Me2)(MeCN)](OTf)2 was done using
thallium(I) triflate (TlOTf) and allowed to stir overnight (∼12 h).
Characterization of the parent [Co(PY5Me2)(MeCN)](OTf)2
complex was carried out by way of elemental analysis, Table S1,
mass spectrometry, and electrochemistry, i.e., cyclic voltammetry (see
Figure 6 above). During the mass spectroscopy measurements (M+),
it was observed that each of the parent complexes lost their −MeCN
ligand. This resulted in intense peaks for Co(PY5Me2)2+ and
Co(PY5Me2)(OTf)+ at 251.07 and 651.09. Elemental analysis:
found (calcd) for C33H28CoF6N6O6S2: C, 45.33(47.09); H,
3.03(3.35); N, 8.56(9.99).
[Co(PY5Me2)(CN)](OTf). In a glovebox, [Co(PY5Me2)(MeCN)]-
(OTf)2 (0.178 mmol, 149.9 mg) was dissolved in (∼5 mL) methanol
and a separate methanolic solution (∼3 mL) of KCN (0.264 mmol,
17.2 mg) was made before being pulled out and placed in an ice bath
to cool. After allowing the mixtures to equilibrate to the temperature
of the ice bath, the KCN solution was slowly charged to the stirring
solution of [Co(PY5Me2)(MeCN)](OTf)2, which immediately
turned from bright yellow/orange to a dark reddish/brown. To
avoid the accumulation of side-products, the reaction mixture was
only allowed to stir for 1 min before being precipitated with dry
diethyl ether. Dissolution of the crude [Co(PY5Me2)(CN)](OTf)
yielded a brown powder. The supernatant was decanted in the
glovebox and the pure product was obtained after recrystallizing in
dichloromethane and washing with diethyl ether (yield: 65.4%).
(Insoluble particulate in dichloromethane was syringe-filtered before
being crashed with diethyl ether.) Crystals suitable for single crystal
X-ray diffraction analysis were obtained by slow vapor diffusion of
ether into a concentrated acetonitrile solution of [Co(PY5Me2)-
(CN)](OTf) at room temperature. Mass spectrometry, elemental
analysis, and 1H NMR were also used to characterize the
[Co(PY5Me2)(CN)](OTf) complex; see the Supporting Information.
An intense peak for [Co(PY5Me2)(CN)]+ at 528.2 was observed in
the mass spectra (M+), along with peaks for complexes that lost their
exogenous CN ligand ([Co(PY5Me2)]2+ at 251 and [Co(PY5Me2)-
(OTf)]+ at 651.1). Interestingly, even with pure material, peaks for
the oxidized complex were also observed ([Co(PY5Me2)(CN)]2+ at
264.1 and [Co(PY5Me2)(CN)](OTf)+ at 677.1). Elemental analysis:
Synthesis of Cross-Exchange Redox Shuttles. OSRSs used in
the stopped-flow studies were either purchased from commercial
suppliers or synthesized from previous literature reports. Synthesis of
the [Co(terpy)2]3+/2+ complexes, where terpy represents 2,2’:6′,2″-
terpyridine, was carried out using a modified literature procedure.7
Briefly, the appropriate stoichiometric ratio (∼2.1 equiv) of the terpy
(Alfa Aesar, 97%) ligand was reacted with (1 equiv) cobalt dichloride
hexahydrate (CoCl2·6H2O) in methanol. The reaction was brought to
reflux and stirred in air for ∼2 h. Upon cooling, the reaction mixture
was concentrated, and an excess of TBAPF6 (∼6−8 equiv) dissolved
in methanol was added. A brownish/orange solid precipitated out of
solution after sonication. The pure product was vacuum-filtered and
washed with copious amounts of methanol and diethyl ether before
being collected and dried. The isolated paramagnetic [Co(terpy)2]-
1
(PF6)2 species was characterized by H NMR containing chemical
shifts up to ca. 100 ppm. Oxidation of the [Co(terpy)2](PF6)2
complex was carried out using 1.1 equiv of nitrosonium hexafluor-
ophosphate (Strem Chemicals, min. 97%, NOPF6) dissolved in a
minimal amount of acetonitrile. The reaction was stirred in air
overnight (∼12 h.) to ensure the reaction reached completion.
Isolation of the crude [Co(terpy)2](PF6)3 product was carried out via
precipitation from acetonitrile using diethyl ether. The solid was
vacuum-filtered and washed with dichloromethane, methanol, and
diethyl ether. Recrystallization in acetonitrile yielded the pure
[Co(terpy)2](PF6)3 product confirmed via 1H NMR. 1H NMR
(500 MHz, acetonitrile-d3) δ 9.16−9.08 (m, 2H), 9.02 (d, J = 8.1 Hz,
4H), 8.61 (d, J = 7.8 Hz, 4H), 8.25 (t, J = 7.8 Hz, 4H), 7.44 (t, J = 6.8
Hz, 4H), 7.25 (d, J = 5.8 Hz, 4H). 1,1′-dimethylferrocene (Sigma-
Aldrich, 95%), [Fe(C5H4CH3)2], was used as received. Oxidation of
[Fe(C5H4CH3)2] to obtain the ferrocenium salt, [Fe(C5H4CH3)2]-
(BF4), was carried out using a procedure reported in the literature.41
Cross-Exchange Kinetics. Stopped-flow measurements were
performed using a similar methodology to that previously
reported.9,12 Briefly, samples were measured using an Olis RSM
1000 DeSa rapid-scanning spectrophotometer with dual-beam UV−
vis recording to Olis SpectralWorks software. The instrument
contained a quartz cell with a 1 cm path length. Scans were taken
once every millisecond with 1 nm resolution. The 150 W xenon arc
lamp was controlled using an LPS-220B Lamp Power Supply and held
to within 79−81 W during each measurement. The temperature was
also held constant at 25 0.1 °C using a NESLAB RTE-140 chiller/
K
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
circulator. Two cross-exchange reactions were measured and
described in detail below. All [Co(PY5Me2)(CN)](OTf), [Co-
(terpy)2](PF6)3, and [Fe(C5H4CH3)2]+/0 solutions were prepared
neat using dry acetonitrile.
Data derived from CHN analysis, X-ray diffraction,
NMR, UV−vis and electrochemistry measurements,
plots of stopped-flow results (PDF)
Pseudo-first-order conditions were implemented in both cross-
exchange reactions, which maintained at least a 10-fold excess of a
single reactant species. In the case of the cross-reaction between
[Fe(C5H4CH3)2] and [Co(terpy)2](PF6)3, however, a 10-fold excess
of both a single reactant and product species was maintained since the
reaction was expected to reach equilibrium. The concentrations of
[Co(PY5Me2)(CN)](OTf) were varied and held in excess for the
reactions with [Co(terpy)2](PF6)3. However, both the concentrations
of [Fe(C5H4CH3)2] and [Fe(C5H4CH3)2](BF4) were held in excess,
while the [Fe(C5H4CH3)2] concentration was varied for the reactions
with [Co(terpy)2](PF6)3. In both cross-exchange reactions the
spectral changes were monitored at 505 nm, following the growing
absorbance of the [Co(terpy)2]2+ species. Scientific Data Analysis
Accession Codes
CCDC 1851784−1851788 contain the supplementary crys-
tallographic data for this paper. These data can be obtained
free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by
emailing data_request@ccdc.cam.ac.uk, or by contacting The
Cambridge Crystallographic Data Centre, 12 Union Road,
Cambridge CB2 1EZ, UK; fax: +44 1223 336033.
AUTHOR INFORMATION
Corresponding Author
*E-mail: hamann@chemistry.msu.edu.
■
ORCID
Software provided fits for the pseudo-first-order rate constants, kobs
,
Richard J. Staples: 0000-0003-2760-769X
Thomas W. Hamann: 0000-0001-6917-7494
Notes
using a nonlinear least-squares regression. Seven independent trials
were averaged to provide the measured kobs values. Absorbance plots
for each pseudo-first order reaction were fit using A = A∞ + (Ao −
obst
A∞) e−k . The second-order rate constants were calculated from the
The authors declare no competing financial interest.
slope of the kobs versus the excess concentration of either
[Co(PY5Me2)(CN)](OTf) or [Fe(C5H4CH3)2], and each had a
goodness of fit, R2 > 0.996. The error associated with measured kobs
values were taken to be the standard deviation of the seven
independent trials. The negligible error in concentration was
propagated based on prepared stock solutions of each reaction
mixture. It was assumed that uniform mixing led to minimal deviation
in the reactants initial concentrations.
ACKNOWLEDGMENTS
■
T.W.H. thanks the Chemical Sciences, Geosciences, and
Biosciences Division, Office of Basic Energy Sciences, Office
of Science, the U.S. Department of Energy Grant No. DE-
SC0017342. The CCD-based X-ray diffractometer at Center
for Crystallographic Research, MSU, were upgraded and/or
replaced by departmental funds.
Semiconductor Anode Fabrication. High surface area thin
films of titanium dioxide (TiO2) on fluorine-doped tin oxide (FTO)
glass substrates (TEC 15, Hartford), 12 Ω cm−2, were made to
conduct recombination studies to the oxidized redox shuttles:
[Co(PY5Me2)(CN)]2+ and [Co(bpy)3]3+. The glass substrates were
cleaned in an ultrasonic bath using (in order) soap water, deionized
water, acetone and isopropyl alcohol. To burn off any organic residue
the substrates were then baked at 400 °C for 30 min. After cooling, a
blocking layer was deposited on the FTO substrates by way of atomic
layer deposition (ALD). A Savannah 200 instrument (Cambridge
Nanotech Inc.) deposited 1000 cycles of titanium isopropxide
(99.999% trace metals basis, Sigma-Aldrich) at 225 °C and water
using reactant exposure times of 0.3 and 0.015 s, respectively.
Between each exposure, nitrogen was purged for 5 s. After ALD, a
transparent thin film (∼5−6 μm) of ∼30 nm TiO2 nanoparticles was
prepared by doctor-blading a commercial paste (DSL 30NR-D,
DYESOL) on the FTO glass substrates coated with the TiO2 blocking
layer. The doctor-bladed films were allowed to relax for 10 min at
room temperature on benchtop, then for another 15 min in the oven
at 100 °C. The electrodes were annealed by heating in air to 325 °C
for 5 min, 375 °C for 5 min, 450 °C for 5 min, and 500 °C for 15 min.
A post TiCl4 treatment was completed after cooling the sintered TiO2
anodes to ∼70 °C. The post TiCl4 treatment was carried out using
∼40 mL of a 40 mM stock solution of TiCl4 dissolved in Milli-Q
water. The solution was heated to 70 °C for 10 min in an oven before
the TiO2 sintered films were immersed for 30 min. After 30 min, the
films were rinsed with Milli-Q water and baked again at 500 °C for
another 30 min. Upon cooling, electrical contact was made using
copper wire leads coated in silver epoxy. Before deposition of the
epoxy on the FTO substrates, part of the blocking layer was manually
scraped off. As the epoxy dried gently on a hot plate, the TiO2 films
were covered to protect against any organic residue from diffusing
into the mesopores.
REFERENCES
■
(1) Klahr, B. M.; Hamann, T. W. Performance Enhancement and
Limitations of Cobalt Bipyridyl Redox Shuttles in Dye-Sensitized
Solar Cells. J. Phys. Chem. C 2009, 113 (31), 14040−14045.
̈
(2) Clifford, J. N.; Palomares, E.; Nazeeruddin, M. K.; Gratzel, M.;
Durrant, J. R. Dye Dependent Regeneration Dynamics in Dye
Sensitized Nanocrystalline Solar Cells: Evidence for the Formation of
a Ruthenium Bipyridyl Cation/iodide Intermediate. J. Phys. Chem. C
2007, 111 (17), 6561−6567.
(3) Hamann, T. W.; Ondersma, J. W. Dye-Sensitized Solar Cell
Redox Shuttles. Energy Environ. Sci. 2011, 4 (2), 370.
(4) Boschloo, G.; Hagfeldt, A. Characteristics of the Iodide/
Triiodide Redox Mediator in Dye-Sensitized Solar Cells. Acc. Chem.
Res. 2009, 42 (11), 1819−1826.
(5) Hattori, S.; Wada, Y.; Yanagida, S.; Fukuzumi, S. Blue Copper
Model Complexes with Distorted Tetragonal Geometry Acting as
Effective Electron-Transfer Mediators in Dye-Sensitized Solar Cells. J.
Am. Chem. Soc. 2005, 127 (26), 9648−9654.
(6) Daeneke, T.; Mozer, A. J.; Kwon, T.-H.; Duffy, N. W.; Holmes,
A. B.; Bach, U.; Spiccia, L. Dye Regeneration and Charge
Recombination in Dye-Sensitized Solar Cells with Ferrocene
Derivatives as Redox Mediators. Energy Environ. Sci. 2012, 5,
7090−7099.
(7) Sapp, S. A.; Elliott, C. M.; Contado, C.; Caramori, S.; Bignozzi,
C. A. Substituted Polypyridine Complexes of Cobalt (II/III) as
Efficient Electron-Transfer Mediators in Dye-Sensitized Solar Cells. J.
Am. Chem. Soc. 2002, 124 (37), 11215−11222.
(8) Ondersma, J. W.; Hamann, T. W. Measurements and Modeling
of Recombination from Nanoparticle TiO2 Electrodes. J. Am. Chem.
Soc. 2011, 133 (21), 8264−8271.
(9) Xie, Y.; Baillargeon, J.; Hamann, T. W. Kinetics of Regeneration
and Recombination Reactions in Dye-Sensitized Solar Cells Employ-
ing Cobalt Redox Shuttles. J. Phys. Chem. C 2015, 119 (50), 28155−
28166.
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
chem.8b01772.
(10) Feldt, S. M.; Lohse, P. W.; Kessler, F.; Nazeeruddin, M. K.;
̈
Gratzel, M.; Boschloo, G.; Hagfeldt, A. Regeneration and Recombi-
nation Kinetics in Cobalt Polypyridine Based Dye-Sensitized Solar
L
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
4‑
Cells, Explained Using Marcus Theory. Phys. Chem. Chem. Phys. 2013,
15 (19), 7087−7097.
Assignments for Fe(CN)6 and Cs2MgFe(CN)6. Inorg. Chem. 1976,
15 (2), 249−252.
(28) Jones, L. H. Nature of Bonding in Metal Cyanide Complexes as
Related to Intensity and Frequency of Infrared Absorption Spectra.
Inorg. Chem. 1963, 2 (4), 777−780.
(29) Klein Gebbink, R. J. M.; Jonas, R. T.; Goldsmith, C. R.; Stack,
T. D. P. A Periodic Walk: A Series of First-Row Transition Metal
Complexes with the Pentadentate Ligand PY5. Inorg. Chem. 2002, 41
(18), 4633−4641.
(30) Krivokapic, I.; Zerara, M.; Daku, M. L.; Vargas, A.; Enachescu,
C.; Ambrus, C.; Tregenna-piggott, P.; Amstutz, N.; Krausz, E.;
Hauser, A. Spin-Crossover in Cobalt (II) Imine Complexes. Coord.
Chem. Rev. 2007, 251 (3−4), 364−378.
(31) Enachescu, C.; Krivokapic, I.; Zerara, M.; Real, J. A.; Amstutz,
N.; Hauser, A. Optical Investigation of Spin-Crossover in Cobalt (II)
Bis-Terpy Complexes. Inorg. Chim. Acta 2007, 360 (13), 3945−3950.
(32) Marcus, R. A.; Sutin, N. Electron Transfers in Chemistry and
Biology. Biochim. Biophys. Acta, Rev. Bioenerg. 1985, 811 (3), 265−
322.
(33) Sutin, N. Nuclear, Electronic, and Frequency Factors in
Electron Transfer Reactions. Acc. Chem. Res. 1982, 15 (9), 275−282.
(34) Beattie, J. K.; Elsbernd, H. Electron Exchange Reaction of Bis
(terpyridine)cobalt3+/2+ Re-Examined by NMR. Inorg. Chim. Acta
1995, 240 (1−2), 641−644.
(35) Chow, H. S.; Constable, E. C.; Housecroft, C. E.; Kulicke, K. J.;
Tao, Y. When Electron Exchange Is Chemical Exchange−assignment
of 1H NMR Spectra of Paramagnetic cobalt(II)-2,2’:6’,2″-Terpyridine
Complexes. Dalt. Trans. 2005, 2, 236−237.
(36) Hamann, T. W.; Gstrein, F.; Brunschwig, B. S.; Lewis, N. S.
Measurement of the Dependence of Interfacial Charge-Transfer Rate
Constants on the Reorganization Energy of Redox Species at N-ZnO/
H2O Interfaces. J. Am. Chem. Soc. 2005, 127 (40), 13949−13954.
(37) Freedman, D. E.; Jenkins, D. M.; Long, J. R. Strong Magnetic
Exchange Coupling in the Cyano-Bridged Coordination Clusters
[(PY5Me2)4V4M(CN)6]5+ (M = Cr, Mo). Chem. Commun. 2009,
219 (32), 4829−4831.
(38) Du, M.; Zhao, X.-J.; Cai, H. Crystal Structure of tris(2,2’-
bipyridine)cobalt(iii) Triperchlorate Dihydrate, [Co(C10H8N2)3]-
(CIO4)3·2H2O. Z. Kristallogr. - New Cryst. Struct. 2004, 219, 463−
465.
(11) Mathew, S.; Yella, A.; Gao, P.; Humphry-Baker, R.; Curchod, B.
F. E.; Ashari-Astani, N.; Tavernelli, I.; Rothlisberger, U.; Nazeeruddin,
̈
M. K.; Gratzel, M. Dye-Sensitized Solar Cells with 13% Efficiency
Achieved through the Molecular Engineering of Porphyrin Sensitizers.
Nat. Chem. 2014, 6 (3), 242−247.
(12) Baillargeon, J.; Xie, Y.; Hamann, T. W. Bifurcation of
Regeneration and Recombination in Dye-Sensitized Solar Cells via
Electronic Manipulation of Tandem Cobalt Redox Shuttles. ACS
Appl. Mater. Interfaces 2017, 9 (39), 33544−33548.
(13) Hao, Y.; Yang, W.; Zhang, L.; Jiang, R.; Mijangos, E.; Saygili, Y.;
̈
Hammarstrom, L.; Hagfeldt, A.; Boschloo, G. A Small Electron Donor
in Cobalt Complex Electrolyte Significantly Improves Efficiency in
Dye-Sensitized Solar Cells. Nat. Commun. 2016, 7, 13934.
(14) Yang, W.; Vlachopoulos, N.; Hao, Y.; Hagfeldt, A.; Boschloo,
G. Efficient Dye Regeneration at Low Driving Force Achieved in
Triphenylamine Dye LEG4 and TEMPO Redox Mediator Based Dye-
Sensitized Solar Cells. Phys. Chem. Chem. Phys. 2015, 17, 15868−
15875.
(15) Chandrasekhar, S.; McAuley, A. Syntheses, Structure, and
Reactivity of Cobalt and Nickel Complexes of 1,4,7-Trithiacyclode-
cane: Self-Exchange Rates for Co(S6)2+/3+ Couples from 59Co
NMR Spectroscopy. Inorg. Chem. 1992, 31 (3), 480−487.
(16) Kuppers, H. J.; Neves, A.; Pomp, C.; Ventur, D.; Wieghardt, K.;
Nuber, B.; Weiss, J. Electron-Transfer Barriers in cobalt(III) and
cobalt(II) Bis Complexes of 1,4,7-Triazacyclononane(tacn) and 1,4,7-
Trithiacyclononane (Ttcn). Crystal Structures of [CoII(tacn)2]-
I2.2H2O and of [CoIII(ttcn)2](ClO4)3. Inorg. Chem. 1986, 25 (14),
2400−2408.
(17) Osvath, P.; Sargeson, A. M.; Skelton, B. W.; White, A. H.
Cobalt (II) and cobalt(III) Complexes of a Novel Homoleptic
Thioether Cage. J. Chem. Soc., Chem. Commun. 1991, 15, 1036−1038.
(18) Zadrozny, J. M.; Freedman, D. E.; Jenkins, D. M.; Harris, T. D.;
̀
́
Iavarone, A. T.; Mathoniere, C.; Clerac, R.; Long, J. R. Slow Magnetic
Relaxation and Charge-Transfer in Cyano-Bridged Coordination
Clusters Incorporating [Re(CN)7]3-/4-. Inorg. Chem. 2010, 49
(19), 8886−8896.
(19) Sun, Y.; Bigi, J. P.; Piro, N. A.; Tang, M. L.; Long, J. R.; Chang,
C. J. Molecular Cobalt Pentapyridine Catalysts for Generating
Hydrogen from Water. J. Am. Chem. Soc. 2011, 133 (24), 9212−9215.
(20) King, A. E.; Surendranath, Y.; Piro, N. A.; Bigi, J. P.; Long, J. R.;
Chang, C. J. A Mechanistic Study of Proton Reduction Catalyzed by a
Pentapyridine Cobalt Complex: Evidence for Involvement of an
Anation-Based Pathway. Chem. Sci. 2013, 4 (4), 1578−1587.
(21) Kashif, M. K.; Axelson, J. C.; Duffy, N. W.; Forsyth, C. M.;
Chang, C. J.; Long, J. R.; Spiccia, L.; Bach, U. A New Direction in
Dye-Sensitized Solar Cells Redox Mediator Development: In Situ
Fine-Tuning of the cobalt(II)/(III) Redox Potential through Lewis
Base Interactions. J. Am. Chem. Soc. 2012, 134 (40), 16646−16653.
(22) Goldsmith, C. R.; Jonas, R. T.; Cole, A. P.; Stack, T. D. P. A
Spectrochemical Walk: Single-Site Perturbation within a Series of Six-
Coordinate Ferrous Complexes. Inorg. Chem. 2002, 41 (18), 4642−
4652.
(39) Yao, J. C.; Ma, L. F.; Yao, F. J. Crystal Structure of Tris (2,2’-
bipyridine)cobalt(II) Diperchlorate, [Co(C10H8N2)3][ClO4]2. Z.
Kristallogr. - New Cryst. Struct. 2005, 220, 483−484.
(40) Connelly, N. G.; Geiger, W. E. Chemical Redox Agents for
Organometallic Chemistry. Chem. Rev. 1996, 96 (2), 877−910.
(41) Bechlars, B.; D’Alessandro, D. M.; Jenkins, D. M.; Iavarone, A.
T.; Glover, S. D.; Kubiak, C. P.; Long, J. R. High-Spin Ground States
via Electron Delocalization in Mixed-Valence Imidazolate-Bridged
Divanadium Complexes. Nat. Chem. 2010, 2 (5), 362−368.
(23) Evans, D. F. The Determination of the Paramagnetic
Susceptibility of Substances in Solution by Nuclear Magnetic
Resonance. J. Chem. Soc. 1959, 2003−2005.
(24) Schubert, E. M. Utilizing the Evans Method with a
Superconducting NMR Spectrometer in the Undergraduate Labo-
ratory. J. Chem. Educ. 1992, 69 (1), 62.
(25) Bain, G. A.; Berry, J. F. Diamagnetic Corrections and Pascal’s
Constants. J. Chem. Educ. 2008, 85 (4), 532−536.
(26) Kettle, S. F. A.; Aschero, G. L.; Diana, E.; Rossetti, R.;
Stanghellini, P. L. The Vibrational Spectra of the Cyanide Ligand
Revisited: Terminal Cyanides. Inorg. Chem. 2006, 45 (13), 4928−
4937.
(27) Swanson, B. I.; Rafalko, J. J. Perturbation of Intramolecular
Vibrations by Strong Interionic Forces. Vibrational Spectra and
M
DOI: 10.1021/acs.inorgchem.8b01772
Inorg. Chem. XXXX, XXX, XXX−XXX