Mechanism-Based Development of a Low-Potential, Soluble, and Cyclable Multielectron Anolyte for Nonaqueous Redox Flow Batteries

Article abstract of DOI:10.1021/jacs.6b07638

The development of nonaqueous redox flow batteries (NRFBs) has been impeded by a lack of electroactive compounds (anolytes and catholytes) with the necessary combination of (1) redox potentials that exceed the potential limits of water, (2) high solubilit

Full text of DOI:10.1021/jacs.6b07638

Article
pubs.acs.org/JACS
Mechanism-Based Development of a Low-Potential, Soluble, and
Cyclable Multielectron Anolyte for Nonaqueous Redox Flow
Batteries
Christo S. Sevov,†,§ Sydney L. Fisher,‡,§ Levi T. Thompson,*,‡ and Melanie S. Sanford*,†
†Department of Chemistry, University of Michigan, 930 North University Avenue, Ann Arbor, Michigan 48109, United States
‡Department of Chemical Engineering, University of Michigan, 2300 Hayward Street, Ann Arbor, Michigan 48109, United States
S
* Supporting Information
ABSTRACT: The development of nonaqueous redox flow
batteries (NRFBs) has been impeded by a lack of electroactive
compounds (anolytes and catholytes) with the necessary
combination of (1) redox potentials that exceed the potential
limits of water, (2) high solubility in nonaqueous media, and
(3) high stability toward electrochemical cycling. In addition,
ideal materials would maintain all three of these properties
over multiple electron transfer events, thereby providing a
proportional increase in storage capacity. This paper describes
the mechanism-based design of a new class of metal-
coordination complexes (MCCs) as anolytes for NRFBs.
The tridentate bipyridylimino isoindoline (BPI) ligands of these complexes were designed to enable multielectron redox events.
These molecules were optimized using a combination of systematic variation of the BPI ligand and the metal center along with
mechanistic investigations of the decomposition pathways that occur during electrochemical cycling. Ultimately, these studies led
to the identification of nickel BPI complexes that could undergo stable charge-discharge cycling (<5% capacity loss over 200
cycles) as well as a derivative that possesses the previously unprecedented combination of high solubility (>700 mM in CH3CN),
multiple electron transfers at low redox potentials (−1.7 and −1.9 V versus Ag/Ag+), and high stability in the charged state for
days at high concentration. Overall, the studies described herein have enabled the identification of a promising anolyte candidate
for NRFBs and have also provided key insights into chemical design principles for future classes of MCC-based anolytes.
there are currently no electroactive species that possess the
properties necessary for this application. Specifically, NRFBs
require electroactive molecules with a combination of three
features: (1) redox potentials that exceed the potential limits of
water, (2) high solubility, and (3) high stability toward
electrochemical cycling. Furthermore, ideal materials would
maintain these three properties while cycling over multiple
redox couples, thereby providing a proportional increase in
storage capacity.
Metal-coordination complexes (MCCs) bearing acetylaceto-
nate (acac),18−22 dithiolate,23,24 2,2′-bipyridine (bpy),25−34 and
cyclopentadienyl35−37 ligands have recently gained attention as
electrolytes for NRFBs. However, to date, none of these
materials possess all three of the features discussed above. For
example, ferrocene-based MCCs exhibit high solubility in
nonaqueous media as well as stable single-electron electro-
chemical cycling. However, their redox potentials are modest
(i.e., within the range of aqueous media).35,37 In contrast,
several bipyridine-33,34 and acac-derived22,38 MCCs exhibit
multiple redox couples over a much wider potential window, as
INTRODUCTION
■
Redox flow batteries (RFBs) are energy storage devices that use
solvated redox-active species to interconvert chemical and
electrical energy.1−3 During RFB operation, solutions contain-
ing the electroactive species (termed anolytes and catholytes)
are pumped from external reservoirs over inert electrodes to
charge and discharge the battery.4,5 In established RFBs, the
electroactive species are aqueous solutions of redox-active
transition-metal salts or organic molecules.6−13 Despite many
recent advances, the achievable cell voltages in these systems
are inherently limited by the small ∼1.5 V electrochemical
stability window of the aqueous medium.5 As such, aqueous
RFBs suffer from moderate energy densities and thus require
large quantities of solvent, supporting salts, and electroactive
compounds to achieve high storage capacity.
In principle, this limitation could be addressed through the
development of electrolytes that are compatible with non-
aqueous solvents such as acetonitrile, which is electrochemically
stable over a 5 V window.2 Despite their promise, nonaqueous
redox flow batteries (NRFBs) remain underdeveloped for two
main reasons. First, the separators and cell components
employed in aqueous RFBs are incompatible with nonaqueous
systems.14−17 Second (and directly relevant to this paper),
Received: July 24, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.6b07638
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
well as high solubility in nonaqueous media. However, none of
these complexes undergo stable electrochemical cycling
(typically showing >50% capacity fade after <10 cycles). In
these latter systems, the mechanistic origin of the poor
cyclability is not well understood, and there have been no
systematic evaluations of MCC decomposition pathways during
bulk charging. Ultimately, fundamental chemical information
about degradation rates and mechanisms are expected to
inform the rational design of MCCs that exhibit the
combination of cyclability, solubility, and redox potential
required for application to NRFBs.
Furthermore, the BPI ligands can be prepared via a modular
synthesis, which provides rapid access to a diverse library of
complexes (Figure 2a).
We first evaluated the known compound Ni(L1)2 bearing an
unsubstituted BPI ligand.44 As expected for a complex
containing a polydentate ligand of high symmetry,45,46 Ni(L1)2
exhibited extremely low solubility (0.42 mM) in CH3CN,
which precluded CV studies. Our strategy for improving
solubility involved decreasing the symmetry of the complexes
to disfavor crystal packing47−50 as well as incorporating alkoxy
ether functional groups, which have been shown to enhance the
solubility of other MCCs in polar, aprotic solvents (Figure
2b).22,33 In addition to improved solubility, the BPI derivatives
functionalized with electron-donating alkoxy substituents
exhibit more negative reduction potentials than the parent
MCCs, which is advantageous for applications as anolytes.
The solubilities of nickel complexes Ni(L1)2−Ni(L6)2 were
measured in CH3CN and are summarized in Figure 2c. Modest
improvements to solubility were observed upon the incorpo-
ration of methoxy substituents [complexes Ni(L2)2 and
Ni(L3)2]. Furthermore, derivatives bearing the diethylenegly-
coxy (ODG) substituent [complexes Ni(L4)2−Ni(L6)2]
exhibited solubility enhancements of more than 3 orders of
magnitude versus the parent Ni(L1)2. Most notably, Ni(L6)2
was isolated as a viscous oil that is miscible with CH3CN to
form a free-flowing solution of greater than 700 mM
concentration. Moreover, a 250 mM solution of Ni(L6)2
could be prepared in the presence of 500 mM TBABF4, the
concentration of supporting electrolyte required for bulk
electrolysis.51 The dramatic effect of minor perturbations of
the peripheral ligand structure on solubility is exemplified by
the low solubilities of Ni(L1)2−Ni(L3)2, moderate solubilities
of Ni(L4)2−Ni(L5)2, and excellent solubility of Ni(L6)2.
We also assessed the impact of metal on MCC solubility
using a common ligand framework. Comparative studies were
performed on complexes of ligands L4 and L5. These were
selected because their Ni complexes are only moderately
soluble in CH3CN (5−15 mM). As such, small perturbations in
solubility as a function of metal cation were easier to detect
than with other analogues. MCCs of L4 and L5 containing
Mg2+, Mn2+, Fe2+, Co2+, Ni2+, and Zn2+ were prepared. Single-
crystal X-ray structures of the isolated Mn(L4)2, Fe(L4)2,
Ni(L4)2, and Zn(L4)2 derivatives show pseudo-octahedral
geometries with nearly identical unit cells (see the Supporting
Information). Although these complexes are nearly indistin-
guishable in the solid-state, their solubilities vary from 28 to 8
mM (Figure 2d). A similar trend was observed for complexes of
L5, for which the Zn and Ni complexes were less soluble than
the Mn and Fe analogues by nearly an order of magnitude. The
impact of the metal center on the solubility of this MCC series
can be rationalized on the basis of the extent of charge shielding
by the ligand.52 Shielding of the metal ion is typically weaker for
early versus late transition metals.53 As a result, MCCs of earlier
metals are expected to be more polar and thus exhibit higher
solubility in polar solvents like CH3CN.
This paper describes the design, synthesis, and electro-
chemical evaluation of anolyte MCCs bearing tridentate
bipyridylimino isoindoline (BPI) ligands (Figure 1). With the
Figure 1. Design strategy of next-generation MCCs for NRFBs.
appropriate selection of ligand substituents and metal center,
complexes were identified that exhibit multiple redox couples at
very negative potentials as well as high solubility in acetonitrile.
Systematic investigations of MCC stability during bulk
electrolysis revealed that ligand dissociation is a major
degradation pathway. These studies ultimately guided the
identification of a highly stable Ni(BPI)2 anolyte that
undergoes more than 200 charge−discharge cycles through
multiple redox couples at potentials well beyond those of
aqueous anolytes.
RESULTS AND DISCUSSION
■
Ligand Design and MCC Solubility Measurements.
Our previous work in this area had focused on developing
soluble MCCs that exhibit multiple reversible electron transfers
by cyclic voltammetry (CV).33,34 While promising compounds
were identified, these molecules (along with nearly all of the
MCCs reported to date) exhibited low stability during charge−
discharge cycling in nonaqueous media.19,20,22,24,25,27,28,34,39 We
hypothesized that the poor cyclability of these complexes
resulted from dissociation of their bidentate ligands during
electrochemical cycling.40−42 To combat this decomposition
pathway, we pursued a tridentate ligand scaffold (BPI, Figure
1), which is expected to bind more strongly to the metal,
thereby suppressing ligand dissociation and yielding complexes
with enhanced stability and cyclability. MCCs of BPI ligands
offer the additional advantage that they undergo reduction at
the ligand rather than at the metal.43,44 As such, these redox-
active ligands can serve as reservoirs for multiple electrons and
can stabilize the MCCs (since the oxidation state of the metal
center remains unchanged during reductive redox processes).
Cyclic voltammetry (CV). We next evaluated the M(L5)2
series via CV in CH3CN.54 Each complex exhibits two quasi-
reversible reductions at approximately −1.7 and −1.9 V vs Ag/
Ag+ (Figure 3). These occur at nearly identical redox potentials
for all of the MCCs, regardless of the metal center, which is
consistent with reduction involving a ligand-based orbital.
Coulometry experiments confirm that each couple corresponds
to a single electron transfer. Voltammetry to positive potentials
B
DOI: 10.1021/jacs.6b07638
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 2. (a) General procedure for the modular synthesis of BPI ligands and their metalation. (b) Prepared MCCs of BPI derivatives with varying
metals. (c) Effect of the ligand on the solubility of Ni complexes Ni(L1)2−Ni(L6)2. (d) Effect of the metal center on the solubility of complexes with
ligand HL4 (left) or HL5 (right).
Bulk Electrolysis. While the solubility and CV data suggest
that the BPI MCCs are promising anolyte candidates, bulk
cycling is required to assess their electrochemical stability. To
this end, galvanostatic charging and discharging of the M(L5)2
series was performed in a glass H-cell with an ultrafine glass frit
as the separator. A representative data set for Zn(L5)2 is shown
in Figure 4. Voltage cutoffs were selected on the basis of CV
(Figure 4a) to ensure that only the desired redox couples were
accessed. Charge/discharge cycling was performed through two
negative redox couples to 100% state-of-charge (SOC) at 1C
(0.54 mA) using 2 mM solutions of the MCCs. The potential
curve (Figure 4b) during charging exhibits two plateaus (I and
II) that are consistent with the redox potentials measured by
CV. Discharge occurs with a high voltaic efficiency (90%) at
plateaus III and IV. The average Coulombic efficiency (CE) is
94 3%, with minor losses likely resulting from degradation
and/or crossover of the charged material through the glass frit.
Higher CE (97%) was observed using a faster charge rate (2C),
Figure 3. Cyclic voltammograms of M(L5)2 analogues. CV
conditions: 2 mM M(L5)2 in CH3CN with 0.5 M TBABF4 at 100
which is likely due to the reduced time for degradation and/or
mV/s scan rate. Gray dashed lines represent the axis of origin.
crossover of the charged species.
We next systematically assessed the features that contribute
to the charge/discharge stability of M(BPI)2 complexes.
Although reduction of the MCCs is proposed to occur at a
ligand-based orbital, variation of the ligand had minimal impact
on the cycling profiles of these MCCs (see the Supporting
Information). In marked contrast, the identity of the metal had
a profound effect (Figure 4d). For example, the filled shell
revealed quasi-reversible redox couples for Mn(L5)2, Fe(L5)2,
Co(L5)2, and Ni(L5)2 at varying potentials. These redox
processes are consistent with metal-based M2+/M3+ oxidations.
As expected, no analogous couples at positive potentials were
observed for complexes of metals that have filled valence shells
[e.g., Mg(L5)2 and Zn(L5)2].
C
DOI: 10.1021/jacs.6b07638
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. (a) CV of Zn(L5)2. Dashed lines represent the cutoff potentials during bulk cycling. (b) Potential curves for cycle 2 of Zn(L5)2 with
charge and discharge plateaus labeled I−IV for comparison to the CV half-wave potentials. (c) Charge and discharge capacities for Zn(L5)2 along
with Coulombic efficiency as a function of cycle. (d) Discharge capacity, normalized to the theoretical number of electrons charged, for each complex
in the M(L5)2 series.
complexes Zn(L5)2 and Mg(L5)2 showed modest stability
(∼50% decomposition over 50 cycles). The early-transition-
metal complex Mn(L5)2 showed extremely rapid fade of energy
storage capacity (complete decomposition in less than 10
cycles). In contrast, complexes of the later transition metals
Ni(L5)2, Co(L5)2, and Fe(L5)2 exhibited no detectable loss of
storage capacity over 50 charge−discharge cycles to 100% SOC
through two electrons. To our knowledge, these complexes
represent the first examples of stable, multielectron cycling of MCC
electrolytes for NRFBs. The dramatic variation in stability as a
function of metal is particularly noteworthy because (i) the
negative redox events are believed to be solely ligand-based and
(ii) these complexes exhibit nearly identical electrochemical
behavior at negative potentials when evaluated by CV.
Degradation Mechanisms. To gain further insight into
the decomposition pathways, CVs of the relatively unstable Zn
and Mg complexes were measured before (dashed traces) and
after (solid traces) 25 bulk electrolysis cycles. After cycling,
both CVs exhibit an irreversible oxidation at approximately
+0.5 V (Figure 5a). It is highly unlikely that this new peak
results from a metal-based oxidation of a decomposed Mg2+ or
Zn2+ species.55 Instead, we hypothesize that it involves
oxidation of the free ligand or a fragment thereof.
supporting electrolyte, which is expected to form TBA+L5−
with concomitant loss of H2. Complete consumption of HL5
was confirmed by CV on the basis of the disappearance of the
cathodic peak at −1.7 V (traces iv and v). CV of the species
formed after bulk electrolysis shows a peak at −2.3 V, which we
attribute to reduction of L5− to L52−. The corresponding
oxidation of L52− occurs at a much higher potential of −1.3 V.
Most importantly, an additional irreversible peak is observed at
positive potential (+0.5 V), likely corresponding to aL5−/L5•
couple. This peak is the same as that observed in the CVs
following cycling of the Zn and Mg MCCs. Furthermore, when
a CV of L5− was measured with an identical voltage window to
the MCCs, only the irreversible peak at +0.5 V was recorded
(trace vi). Collectively, these data suggest that the oxidative
peak observed after MCC decomposition results from oxidation
of the free ligand L5− to neutral L5•, while oxidation of the
neutral HL5 requires potentials beyond the measured CV
window (trace iv).
These results provide evidence for the presence of free ligand
in solution and thus support the hypothesis that ligand
shedding plays a major role in MCC decomposition. The
high stability of the neutral ML2 complexes (stable for days at
room temperature in CH3CN without ligand loss) suggests that
ligand shedding occurs from a reduced form of the MCC. To
test this hypothesis, bulk electrolysis of Mg(L5)2 was
conducted without a voltaic cutoff while charging galvanostati-
cally to 2e− per molecule (Figure 5b). Charging curves during
this experiment initially plateaued at −1.8 and −1.9 V, which
correspond to the first and second reductions of Mg(L5)2,
respectively, but decreased to −2.3 V before complete charging
Ligand shedding is a possible pathway to generate new
redox-active species. The immediate organic product of ligand
dissociation under the anhydrous conditions of bulk electrolysis
would be anionic L5− with a TBA+ (TBA = tetrabutylammo-
nium) counterion derived from the supporting electrolyte. To
generate an authentic sample of this material, we subjected the
free ligand (HL5) to reductive electrolysis in a TBA+BF4
−
D
DOI: 10.1021/jacs.6b07638
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
because the predominant mechanisms of ligand substitution at
different metals have been extensively studied,56,59−61 these
insights provide a guide to improve the electrochemical stability
of a given MCC by targeting and inhibiting the known pathway
for ligand exchange at that specific metal.
Extended Cycling and Shelf-Life Stability. We next
conducted extended cycling studies to more thoroughly assess
the optimal anolyte candidate. Our stability studies provided
invaluable guidance in selecting the MCC for extended cycling.
Ligand exchange at octahedral Ni2+ occurs with relatively slow
rates via a dissociative mechanism, which suggests that
Ni(BPI)2 complexes should exhibit particularly high cycling
stability. Thus, Ni(L5)2 was chosen for extended cycling to the
commercial standard of 80% SOC.62 The results from these
studies are summarized in Figure 6a and demonstrate that this
Figure 5. (a) CVs of unstable MCCs before (dashed, black traces) and
after (solid, colored traces) cycling studies compared to free ligand.
Gray dashed lines represent the axis of origin. Arrows denote the
starting point and scanning direction. (b) Charging curves from
electrolysis of a Mg(L5)2 solution.
through 2e−. This negative charging potential corresponds to
the reduction of free ligand L5−, as revealed by CV trace v. The
detection of free L5− within the short time frame of the first
charging cycle (∼10 min) indicates that decomposition to
release free ligand occurs rapidly once Mg(L5)2 is reduced.
Subsequent charging revealed that the complex continues to
decompose and that electron transfer occurs primarily at the
potential required for reduction of L5−.
Figure 6. (a) Capacity retention of Ni(L5)2. (b) Potential curves for
cycles 100−105.
complex can be cycled through two electrons with <5%
capacity fade after 200 cycles. Charge and discharge curves after
100 cycles, illustrated in Figure 6b, indicate that the Ni(L5)2
complex is charging through two couples at the expected
potentials and discharging with high CE (97%).
An understanding of the decomposition mechanism and rate
as a function of metal will guide the design of future
generations of anolyte materials. Our cycling studies reveal
that BPI complexes of Fe2+, Co2+, and Ni2+ are significantly
more stable anolytes compared to those of Mg2+, Mn2+, and
Zn2+. The large effect of metal on stability was initially
unexpected, given the similar CVs. In retrospect, we note that
the observed trends for cycling stabilities of the M(BPI)2
complexes correlate well with the rates of ligand exchange at
divalent transition and main-group metals.56 Metals with filled
shells, such as Mg2+ or Zn2+, lack ligand-field interactions,
resulting in rapid rates of dissociative ligand substitution.57
Conversely, metals with partially filled orbitals, such as Mn2+,
Fe2+, Co2+, or Ni2+, form complexes that are far less susceptible
to dissociative ligand substitution. However, unlike the
complexes of the late transition metals, ligand substitution at
many octahedral complexes of Mn2+ has been shown to occur
predominantly by an associative mechanism.58 As a result, the
In parallel with these multielectron cycling studies, we
investigated the shelf-life stability of the charged MCC at high
concentration. A 100 mM solution of Ni(L6)2 in CH3CN was
prepared with 0.5 M TBABF4 supporting electrolyte.63 This
concentration is comparable to high concentration studies
performed on NRFB chemistries in flow.37,64 Galvanostatic
charging of this solution was performed with a 50-fold greater
current than the standard bulk-electrolysis experiments. The
charging occurred at potentials that correlate well with the
peaks observed by CV (−1.8, −2.0 V). An ohmic resistance
contribution of 4 Ω was measured by electrochemical
impedance spectroscopy. This is consistent with the low
overpotentials observed during electrolysis. Once charged, the
solution was removed from the H-cell and stored in a separate
glass container under inert conditions at room temperature.
The concentration of the charged MCC as a function of time
was determined by CV. As plotted in Figure 7b, the doubly
reduced MCC remained stable and soluble in the electrolyte
solution for days at high concentration (0.1 M) without
measurable degradation or precipitation.
2+
exchange of CH3CN at Mn(CH3CN)6 is 10 000 times faster
than that at Ni(CH3CN)62+, and the rates of ligand exchange
increase in the order Ni2+ < Co2+ < Fe2+ ≪ Mn2+.56 Overall,
our results highlight an inverse correlation between the rate of
ligand exchange and stability during cycling. More generally,
E
DOI: 10.1021/jacs.6b07638
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
*mssanfor@umich.edu
Author Contributions
§C.S.S. and S.L.F. contributed equally to this work.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the Joint Center for Energy
Storage Research (JCESR), a grant from the Department of
Energy, Energy Innovation Hub, and a grant from the National
Science Foundation, Sustainable Energy Pathways (SEP)
(NSF-1230236). S.L.F. was supported by the National Science
Foundation Graduate Research Fellowship Program (DGE
1256260). We thank Dr. J. M. Kampf for assistance with X-ray
crystallography, as well as funding from the National Science
Foundation (grant CHE-0840456) for X-ray instrumentation.
Figure 7. (a) Charging potential curve for a 2e− reduction of a 100
mM solution of Ni(L6)2. (b) Concentration of the charged Ni(L6)2 as
a function of time. The red triangle represents the concentration of the
neutral solution before cycling.
CONCLUSIONS
■
REFERENCES
In summary, this paper describes the development of a series of
MCCs containing earth-abundant metals and tridentate BPI
ligands as potential anolyte materials for NRFBs. Systematic
modification of the BPI ligand framework and the metal center
enabled the identification of anolytes with high solubilities and
low redox potentials. Studies of the decomposition of these
complexes during bulk-electrolysis cycling revealed that ligand
shedding is a major decomposition pathway. As such, we
observe a strong correlation between MCC stability and the
known rates of ligand substitution at divalent metal centers.
Specifically, BPI complexes of nickel were found to undergo
200 charge and discharge cycles through two reductions with
<5% capacity fade. Ultimately these studies led to the
identification of Ni(L6)2, a complex that possesses the
previously unprecedented combination of high solubility,
multiple electron transfers at low redox potentials, and high
stability in the charged state, even at high concentration.
Overall, the studies described herein have delivered a promising
anolyte candidate for NRFBs and have also provided key
insights into chemical design principles for future classes of
MCC-based anolytes. Our ongoing work is focused on the
design of second-generation complexes that address the
remaining challenges associated with these materials. In
particular, MCC-based anolytes with lower molecular weight
per mole of electron transferred are important future targets.54
Such materials should enable an increase in the effective
concentration of the anolyte, thereby providing the high energy
densities that are required for practical NRFB electrolytes.54
■
(1) Soloveichik, G. L. Chem. Rev. 2015, 115, 11533.
(2) Shin, S.-H.; Yun, S.-H.; Moon, S.-H. RSC Adv. 2013, 3, 9095.
(3) Wang, W.; Luo, Q.; Li, B.; Wei, X.; Li, L.; Yang, Z. Adv. Funct.
Mater. 2013, 23, 970.
(4) Hou, Y.; Vidu, R.; Stroeve, P. Ind. Eng. Chem. Res. 2011, 50, 8954.
(5) Yang, Z.; Zhang, J.; Kintner-Meyer, M. C. W.; Lu, X.; Choi, D.;
Lemmon, J. P.; Liu, J. Chem. Rev. 2011, 111, 3577.
(6) Sun, H.; Park, J. W.; Hwang, S. S.; Lee, D.; Lee, M. J. US Patent
US8481192, 2013.
(7) Ding, C.; Zhang, H.; Li, X.; Liu, T.; Xing, F. J. Phys. Chem. Lett.
2013, 4, 1281.
(8) Winsberg, J.; Janoschka, T.; Morgenstern, S.; Hagemann, T.;
Muench, S.; Hauffman, G.; Gohy, J.-F.; Hager, M. D.; Schubert, U. S.
Adv. Mater. 2016, 28, 2238.
(9) Janoschka, T.; Martin, N.; Martin, U.; Friebe, C.; Morgenstern,
S.; Hiller, H.; Hager, M. D.; Schubert, U. S. Nature 2015, 527, 78.
(10) Janoschka, T.; Hager, M. D.; Schubert, U. S. Adv. Mater. 2012,
24, 6397.
(11) Lin, K.; Chen, Q.; Gerhardt, M. R.; Tong, L.; Kim, S. B.;
Eisenach, L.; Valle, A. W.; Hardee, D.; Gordon, R. G.; Aziz, M. J.;
Marshak, M. P. Science 2015, 349, 1529.
(12) Huskinson, B.; Marshak, M. P.; Suh, C.; Er, S.; Gerhardt, M. R.;
Galvin, C. J.; Chen, X.; Aspuru-Guzik, A.; Gordon, R. G.; Aziz, M. J.
Nature 2014, 505, 195.
(13) Liu, T.; Wei, X.; Nie, Z.; Sprenkle, V.; Wang, W. Adv. Energy
Mater. 2016, 6, 1501449.
(14) Darling, R. M.; Weber, A. Z.; Tucker, M. C.; Perry, M. L. J.
Electrochem. Soc. 2016, 163, A5014.
(15) Darling, R.; Gallagher, K.; Xie, W.; Su, L.; Brushett, F. J.
Electrochem. Soc. 2016, 163, A5029.
(16) Su, L.; Darling, R. M.; Gallagher, K. G.; Xie, W.; Thelen, J. L.;
Badel, A. F.; Barton, J. L.; Cheng, K. J.; Balsara, N. P.; Moore, J. S.;
Brushett, F. R. J. Electrochem. Soc. 2016, 163, A5253.
(17) Nagarjuna, G.; Hui, J.; Cheng, K. J.; Lichtenstein, T.; Shen, M.;
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.6b07638.
■
S
́
Moore, J. S.; Rodríguez-Lopez, J. J. Am. Chem. Soc. 2014, 136, 16309.
(18) Zhang, D.; Lan, H.; Li, Y. J. Power Sources 2012, 217, 199.
(19) Liu, Q.; Sleightholme, A. E. S.; Shinkle, A. A.; Li, Y.; Thompson,
L. T. Electrochem. Commun. 2009, 11, 2312.
Crystallographic data of Fe(L4)2 in CIF format (CIF)
Crystallographic data of HL4 in CIF format (CIF)
Crystallographic data of Mn(L4)2 in CIF format (CIF)
Crystallographic data of Ni(L4)2 in CIF format (CIF)
Crystallographic data of Zn(L4)2 in CIF format (CIF)
Experimental procedures and characterization of all new
compounds, including spectroscopic data and potentio-
metric data (PDF)
(20) Sleightholme, A. E. S.; Shinkle, A. A.; Liu, Q.; Li, Y.; Monroe, C.
W.; Thompson, L. T. J. Power Sources 2011, 196, 5742.
(21) Yamamura, T.; Shiokawa, Y.; Yamana, H.; Moriyama, H.
Electrochim. Acta 2002, 48, 43.
(22) Suttil, J. A.; Kucharyson, J. F.; Escalante-Garcia, I. L.; Cabrera, P.
J.; James, B. R.; Savinell, R. F.; Sanford, M. S.; Thompson, L. T. J.
Mater. Chem. A 2015, 3, 7929.
(23) Eisenberg, R.; Gray, H. B. Inorg. Chem. 2011, 50, 9741.
(24) Cappillino, P. J.; Pratt, H. D.; Hudak, N. S.; Tomson, N. C.;
Anderson, T. M.; Anstey, M. R. Adv. Energy Mater. 2014, 4, 1300566.
(25) Xing, X.; Zhang, D.; Li, Y. J. Power Sources 2015, 279, 205.
AUTHOR INFORMATION
Corresponding Authors
*ltt@umich.edu
■
F
DOI: 10.1021/jacs.6b07638
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(26) Park, M.-S.; Lee, N.-J.; Lee, S.-W.; Kim, K. J.; Oh, D.-J.; Kim, Y.-
J. ACS Appl. Mater. Interfaces 2014, 6, 10729.
(27) Chakrabarti, M. H.; Dryfe, R. A. W.; Roberts, E. P. L.
(62) Li, L.; Kim, S.; Wang, W.; Vijayakumar, M.; Nie, Z.; Chen, B.;
Zhang, J.; Xia, G.; Hu, J.; Graff, G.; Liu, J.; Yang, Z. Adv. Energy Mater.
2011, 1, 394.
(63) Because this complex undergoes two reversible reductions, it has
twice the storage capacity of a one-electron anolyte. As such, a 100
mM solution is equivalent to reduction of a 200 mM solution
containing a one-electron anolyte.
(64) Wei, X.; Xu, W.; Vijayakumar, M.; Cosimbescu, L.; Liu, T.;
Sprenkle, V.; Wang, W. Adv. Mater. 2014, 26, 7649.
Electrochim. Acta 2007, 52, 2189.
(28) Chakrabarti, M. H.; Lindfield Roberts, E. P.; Saleem, M. Int. J.
Green Energy 2010, 7, 445.
(29) Morita, M.; Tanaka, Y.; Tanaka, K.; Matsuda, Y.; Matsumura-
Inoue, T. Bull. Chem. Soc. Jpn. 1988, 61, 2711.
(30) Matsuda, Y.; Tanaka, K.; Okada, M.; Takasu, Y.; Morita, M.;
Matsumura-Inoue, T. J. Appl. Electrochem. 1988, 18, 909.
(31) Mun, J.; Lee, M.-J.; Park, J.-W.; Oh, D.-J.; Lee, D.-Y.; Doo, S.-G.
Electrochem. Solid-State Lett. 2012, 15, A80.
(32) Kim, J.-H.; Kim, K. J.; Park, M.-S.; Lee, N. J.; Hwang, U.; Kim,
H.; Kim, Y.-J. Electrochem. Commun. 2011, 13, 997.
(33) Cabrera, P. J.; Yang, X.; Suttil, J. A.; Brooner, R. E. M.;
Thompson, L. T.; Sanford, M. S. Inorg. Chem. 2015, 54, 10214.
(34) Cabrera, P. J.; Yang, X.; Suttil, J. A.; Hawthorne, K. L.; Brooner,
R. E. M.; Sanford, M. S.; Thompson, L. T. J. Phys. Chem. C 2015, 119,
15882.
(35) Zhao, Y.; Ding, Y.; Song, J.; Li, G.; Dong, G.; Goodenough, J.
B.; Yu, G. Angew. Chem., Int. Ed. 2014, 53, 11036.
(36) Hwang, B.; Park, M.-S.; Kim, K. ChemSusChem 2015, 8, 310.
(37) Wei, X. L.; Cosimbescu, L.; Xu, W.; Hu, J. Z.; Vijayakumar, M.;
Feng, J.; Hu, M. Y.; Deng, X. C.; Xiao, J.; Liu, J.; Sprenkle, V.; Wang,
W. Adv. Energy Mater. 2015, 5, 1400678.
(38) Escalante-García, I. L.; Wainright, J. S.; Thompson, L. T.;
Savinell, R. F. J. Electrochem. Soc. 2015, 162, A363.
(39) Liu, Q.; Shinkle, A. A.; Li, Y.; Monroe, C. W.; Thompson, L. T.;
Sleightholme, A. E. S. Electrochem. Commun. 2010, 12, 1634.
(40) Shinkle, A. A.; Sleightholme, A. E. S.; Griffith, L. D.; Thompson,
L. T.; Monroe, C. W. J. Power Sources 2012, 206, 490.
(41) Nawi, M. A.; Riechel, T. L. Inorg. Chem. 1981, 20, 1974.
(42) Kitamura, M.; Yamashita, K.; Imai, H. Bull. Chem. Soc. Jpn. 1976,
49, 97.
(43) Csonka, R.; Speier, G.; Kaizer, J. RSC Adv. 2015, 5, 18401.
(44) Gagne, R. R.; Marritt, W. A.; Marks, D. N.; Siegl, W. O. Inorg.
Chem. 1981, 20, 3260.
(45) Freeman, D. H.; Swahn, I. D.; Hambright, P. Energy Fuels 1990,
4, 699.
(46) Huang, X. F.; Nakanishi, K.; Berova, N. Chirality 2000, 12, 237.
(47) Ozaki, S.; Nakagawa, Y.; Shirai, O.; Kano, K. J. Pharm. Sci. 2014,
103, 3524.
(48) Docherty, R.; Pencheva, K.; Abramov, Y. A. J. Pharm. Pharmacol.
2015, 67, 847.
(49) Takeuchi, T.; Oishi, S.; Kaneda, M.; Ohno, H.; Nakamura, S.;
Nakanishi, I.; Yamane, M.; Sawada, J.-i.; Asai, A.; Fujii, N. ACS Med.
Chem. Lett. 2014, 5, 566.
(50) Ishikawa, M.; Hashimoto, Y. J. Med. Chem. 2011, 54, 1539.
(51) This is the maximum concentration of a two-electron anolyte
that could be electrochemically supported by the given electrolyte
stoichiometry.
(52) Teoh, W. H.; Mammucari, R.; Foster, N. R. J. Organomet. Chem.
2013, 724, 102.
(53) Aschenbrenner, O.; Kemper, S.; Dahmen, N.; Schaber, K.;
Dinjus, E. J. Supercrit. Fluids 2007, 41, 179.
(54) Darling, R. M.; Gallagher, K. G.; Kowalski, J. A.; Ha, S.;
Brushett, F. R. Energy Environ. Sci. 2014, 7, 3459.
(55) Oxidation of Mg(II) or Zn(II) is unlikely to occur at such low
potentials.
(56) Helm, L.; Merbach, A. E. Chem. Rev. 2005, 105, 1923.
(57) Nakamura, S.; Meiboom, S. J. Am. Chem. Soc. 1967, 89, 1765.
(58) Inada, Y.; Sugata, T.; Ozutsumi, K.; Funahashi, S. Inorg. Chem.
1998, 37, 1886.
(59) Lincoln, S. F.; Merbach, A. E. In Advances in Inorganic Chemistry;
Sykes, A. G., Ed.; Academic Press, 1995; Vol. 42, p 1.
(60) Hogg, R.; Wilkins, R. G. J. Chem. Soc. 1962, 0, 341.
(61) Holyer, R. H.; Hubbard, C. D.; Kettle, S. F. A.; Wilkins, R. G.
Inorg. Chem. 1965, 4, 929.
G
DOI: 10.1021/jacs.6b07638
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX