Immobilized Cobalt Bis(benzenedithiolate) Complexes: Exceptionally Active Heterogeneous Electrocatalysts for Dihydrogen Production from Mildly Acidic Aqueous Solutions

Article abstract of DOI:10.1021/acs.inorgchem.7b01589

A series of cobalt bis(benzenedithiolate) complexes with varying benzenedithiolate (general abbreviation: bdt^2-) ring substitutions (S₂C₆X₄^2-) were prepared and adsorbed on inexpensive electrodes composed of (a) reduced graphene oxide (RGO) electrodeposited on fluorine-doped tin oxide (FTO) and (b) highly ordered pyrolytic graphite (HOPG). The catalyst-adsorbed electrodes are characterized by X-ray photoelectron spectroscopy. Catalyst loading across the ligand series improved notably with increasing halide substitution [from 2.7 × 10^-11 mol cm^-2 for TBA[Co(S₂C₆H₄)₂] (1) to 6.22 × 10^-10 mol cm^-2 for TBA[Co(S₂C₆Cl₄)₂] (3)] and increasing ring size of the benzenedithiolate ligand [up to 3.10 × 10^-9 mol cm^-2 for TBA[Co(S₂C₁₀H₆)₂] (6)]. Electrocatalytic analysis of the complexes immobilized on HOPG elicits a reductive current response indicative of dihydrogen generation in the presence of mildly acidic aqueous solutions (pH 2-4) of trifluoroacetic acid, with overpotentials of around 0.5 V versus SHE (measured vs platinum). Rate constant (k_obs) estimates resulting from cyclic voltammetry analysis range from 24 to 230 s^-1 with the maximum k_obs for TBA[Co(S₂C₆H₂Cl₂)₂] (2) at an overpotential of 0.59 V versus platinum. Controlled-potential electrolysis studies performed in 0.5 M H₂SO₄ at -0.5 V versus SHE show impressive initial rate constants of over 500 s^-1 under bulk electrolysis conditions; however, steady catalyst deactivation over an 8 h period is observed, with turnover numbers reaching 9.1 × 10⁶. Electrolysis studies reveal that halide substitution is a central factor in improving the turnover stability, whereas the ring size is less of a factor in optimizing the long-term stability of the heterogeneous catalyst manifolds. Catalyst deactivation is likely caused by catalyst desorption from the electrode surfaces.

Full text of DOI:10.1021/acs.inorgchem.7b01589

Article
pubs.acs.org/IC
Immobilized Cobalt Bis(benzenedithiolate) Complexes: Exceptionally
Active Heterogeneous Electrocatalysts for Dihydrogen Production
from Mildly Acidic Aqueous Solutions
Shawn C. Eady, Molly M. MacInnes, and Nicolai Lehnert*
Department of Chemistry, University of Michigan, 930 North University Avenue, Ann Arbor, Michigan 48109, United States
S
* Supporting Information
ABSTRACT: A series of cobalt bis(benzenedithiolate) complexes with varying benzenedithiolate (general abbreviation: bdt2−)
ring substitutions (S2C6X42−) were prepared and adsorbed on inexpensive electrodes composed of (a) reduced graphene oxide
(RGO) electrodeposited on fluorine-doped tin oxide (FTO) and (b) highly ordered pyrolytic graphite (HOPG). The catalyst-
adsorbed electrodes are characterized by X-ray photoelectron spectroscopy. Catalyst loading across the ligand series improved
notably with increasing halide substitution [from 2.7 × 10−11 mol cm−2 for TBA[Co(S2C6H4)2] (1) to 6.22 × 10−10 mol cm−2 for
TBA[Co(S2C6Cl4)2] (3)] and increasing ring size of the benzenedithiolate ligand [up to 3.10 × 10−9 mol cm−2 for
TBA[Co(S2C10H6)2] (6)]. Electrocatalytic analysis of the complexes immobilized on HOPG elicits a reductive current response
indicative of dihydrogen generation in the presence of mildly acidic aqueous solutions (pH 2−4) of trifluoroacetic acid, with
overpotentials of around 0.5 V versus SHE (measured vs platinum). Rate constant (kobs) estimates resulting from cyclic
voltammetry analysis range from 24 to 230 s−1 with the maximum kobs for TBA[Co(S2C6H2Cl2)2] (2) at an overpotential of 0.59
V versus platinum. Controlled-potential electrolysis studies performed in 0.5 M H2SO4 at −0.5 V versus SHE show impressive
initial rate constants of over 500 s−1 under bulk electrolysis conditions; however, steady catalyst deactivation over an 8 h period is
observed, with turnover numbers reaching 9.1 × 106. Electrolysis studies reveal that halide substitution is a central factor in
improving the turnover stability, whereas the ring size is less of a factor in optimizing the long-term stability of the heterogeneous
catalyst manifolds. Catalyst deactivation is likely caused by catalyst desorption from the electrode surfaces.
catalysts immobilized on the electrode surfaces.7−10 The latter
approach has the advantage that the molecular catalysts can be
improved in a systematic manner and derivatized to meet given
application needs, as demonstrated in this work. However, a
INTRODUCTION
■
Devices capable of generating dihydrogen from a renewable
energy source are crucial for meeting the global dihydrogen
consumption demand in a sustainable way and for efforts to
substantial challenge in this regard is to design suitable
replace carbon-based fuel technologies with more sustainable
alternatives.1−4 Large-scale dihydrogen production from
interfaces that allow for the easy, yet stable and affordable,
attachment of the catalysts to the electrode surfaces. In this
protons and electrons, whether directly coupled to water
paper, we further demonstrate that reduced graphene oxide
oxidation photoanodes or electolyzers powered by renewable
energy sources, must inevitably be facilitated by heterogeneous
(RGO) films can serve as inexpensive, yet extremely versatile,
interfaces to adsorb a variety of cobalt bis(benzenedithiolate),
catalysts to allow for application of the catalysts in large-scale
[CoIII(bdt)2]2−, dihydrogen production catalysts to semi-
conductor electrode surfaces, yielding robust and highly active
platforms for electrocatalytic dihydrogen generation.
flow reactors.5−7 To this end, electrocatalysts for dihydrogen
production must be designed that meet the demands of these
systems and, at the same time, are inexpensive and easy to
assemble and that can be prepared on a large scale without
difficulty. These electrocatalysts could be built either from
catalytically active electrode materials or from molecular
Received: June 25, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
In addition to the interface, fast, efficient, and inexpensive
electrocatalysts are necessary to obtain platinum-free manifolds
for dihydrogen production. Previous research in the renewable
energy field has “unearthed” a number of inexpensive iron,
nickel, and cobalt electrocatalysts for dihydrogen produc-
tion.2,4,11−55 Solid-state and molecular catalysts alike have
displayed impressive proton reduction activity, rivaling
platinum in overpotential (η), turnover frequency (TOF),
and turnover number (TON).46,47,56−62 In some cases, these
systems have nearly matched or even exceeded platinum with
regard to stability in poisoning conditions or for dual catalysis
applications, a milestone that should be held in high regard,
considering the decades of catalyst development spent toward
platinum substitution.1,46,47,63 Still, the design of versatile and
inexpensive heterogeneous catalyst manifolds remains a major
challenge for the development of cost-efficient dihydrogen
production systems.48,49,64
Cobalt(III) bis(benzenedithiolate) and corresponding
cobalt(III) bis(dithiolene) complexes, discovered nearly 5
decades ago, have historically been researched for various
applications based on their strong absorption features and
unique electronic structures.26,27,50,65−76 Recently, these
complexes have been found to be active for dihydrogen
production in acidic organic solutions by McNamara et al., with
a maximum TOF of 3400 h−1 and a TON of 9000 in 12 h in
solution.26,27 Later, we reported adsorption of a molecular
cobalt bis(benzenedithiolate) species onto graphitic supports to
provide a heterogeneous proton reduction system with high
activity from mildly acidic aqueous solutions.77 In parallel
studies, these catalyst systems have also been investigated as
two-dimensional polymer materials on graphitic supports, as
well as one-dimensional light-permeable materials on silicon
surfaces for photoelectrocatalysis.67,68 Curiously, these hetero-
geneous manifolds, whether prepared directly by the adsorption
of molecular catalysts onto a support or designed from a
polymer material transferred to a surface, show substantially
higher TOFs than the corresponding catalysts in homogeneous
phase. As demonstrated in this work, cobalt bis-
(benzenedithiolate) based heterogeneous catalyst manifolds
approach even the catalytic activities of nickel bis(diphosphine)
systems (albeit under different conditions), which have become
the gold standard for homogeneous and heterogeneous
dihydrogen production systems in a number of applica-
tions.19,46,47,49,78 Moreover, cobalt bis(benzenedithiolate) com-
plexes provide versatility for incorporation into devices because
both the graphitic supports (e.g., graphene and RGO) used to
adsorb these complexes and their polymeric forms can be
immobilized on a variety of electrode materials. While the
polymeric systems potentially offer more variety in the choice
of electrode materials, one challenge with such materials is
functionalization and thereby the systematic improvement of
their catalytic performance. For this reason, investigation of the
molecular catalysts, which are easily derivatized, is of great
interest to determine the effects of varying the functionality on
the overall performance of the immobilized cobalt bis-
(benzenedithiolate) systems, which is the key point of this
study.
Chart 1. Cobalt Bis(benzenedithiolate) Derivatives
were conducted on RGO electrodeposited on fluorine-doped
tin oxide (FTO), which provides a design strategy of how these
catalysts could be immobilized on a variety of supports,
especially semiconductor electrodes, using RGO as the
universal interface. Quantitative comparisons, based on this
approach, are hampered by the fact that consistency in the
RGO coverage and topography is generally low and there is a
lack of stability of the FTO electrodes at low pH. To
complement these tests, the catalysts were also adsorbed on
highly ordered pyrolytic graphite (HOPG) electrodes. HOPG
electrodes provide a more consistent surface area and
topography, therefore allowing for a quantitative comparison
between different catalyst derivatives in the surface-immobilized
state. The resulting series of heterogeneous electrocatalysts
provides detailed insight into the effect of benzenedithiolate
ligand substitution on the surface adsorption and coverage,
electrocatalytic activity, and lifetime of the catalysts, allowing us
to optimize the catalytic properties of these complexes.
EXPERIMENTAL SECTION
■
General Methods. Chemicals were of the highest purity grade
commercially available and were used without further purification
(unless mentioned). Methanol (anhydrous, ACS grade) was purchased
from Fisher, distilled over calcium hydride, and then degassed via
extended dinitrogen purges prior to use. Acetonitrile (ACS grade),
sodium methoxide, trifluoroacetic acid (TFA), sulfuric acid, potassium
ferricyanide, and potassium ferrocyanide were purchased from Fisher.
1,2-Benzenedithiol, toluene-3,4-dithiol, and 3,6-dichloro-1,2-benzene-
dithiol were purchased from Sigma and used without further
purification. Graphite powder was purchased from MTI Corp. All
procedures were performed under a dinitrogen atmosphere unless
otherwise specified.
Preparation of Dithiol Ligand Derivatives. Tetrahalidedithiol
ligands were prepared using a slightly modified method from that
reported by Gray and co-workers.65 Naphthalenedithiol was prepared
by first following procedures reported by Hart et al. to give o-
dibromonaphthalene, followed by alkylthiol formation and alkyl
cleavage as outlined by Montanucci and co-workers.79,80
3,4,5,6-Tetrachlorobenzenedithiol. A mixture of hexachloroben-
zene (1 g, 3.5 mmol), sodium hydrogen sulfide (0.75 g, 13 mmol),
sulfur (0.08 g, 2.5 mmol), and iron powder (0.18 g, 3.2 mmol) was
heated in 50 mL of N,N-dimethylformamide at 140 °C for 8 h. After
cooling, 100 mL of distilled water was added, and the mixture was
allowed to briefly stir at room temperature, during which time a black
precipitate formed. The precipitate was filtered off, washed with water,
and then dried in vacuo. The dried solid was added to a suspension of
zinc oxide in a 1:1 methanol/1 M aqueous sodium hydroxide solution,
which was refluxed for 1 h. After cooling, the mixture was filtered, and
the yellow filtrate was acidified with a 1 M HCl solution to precipitate
the dithiol product. The solid was dried and recrystallized from
benzene to afford 3,4,5,6-tetrachlorobenzenedithiol 0.44 g (45% yield)
Herein, we report a series of cobalt bis(benzenedithiolene)
derivatives (Chart 1) adsorbed onto graphitic surfaces and
investigations into their dihydrogen production activity in
acidic aqueous solutions, including the quantification of catalyst
loadings and activity data. Because of the ease of preparation
and potential future applications of these systems, initial studies
B
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
1
as a pale-yellow powder. H NMR (400 MHz, CD2Cl2): δH 4.79 (s,
2H).
Tetrabutylammonium Cobalt Bis(benzenedithiolate) (1). Recrys-
tallization of the solid from dichloromethane/ether yielded (TBA)-
[Co(S2C6H4)2] as a dark-blue crystalline solid in 79% yield (0.460 g).
Elem anal. Calcd: C, 57.80; H, 7.62; N, 2.41. Found: C, 58.10; H, 7.69;
N, 2.42. ESI-MS: m/z 339.9 (expected), 338.9 (found). The
spectroscopic data of the product (see Figure S1) match those
available in the literature.27
Tetrabutylammonium Cobalt Bis(3,6-dichlorobenzenedithiolate)
(2). Recrystallization of the solid from dichloromethane/ether yielded
(TBA)[Co(S2C6H2Cl2)2] as a dark-blue crystalline solid in 64% yield
(0.462 g). Elem anal. Calcd (CH2Cl2 1/4 lattice occupancy): C, 45.80;
H, 5.47; N, 1.89. Found: C, 45.73; H, 5.58; N, 1.91. ESI-MS: m/z
477.7 (expected), 476.7 (found). The spectroscopic data of the
product (see Figure S1) match those available in the literature.27
Tetrabutylammonium Cobalt Bis(3,4,5,6-tetrachlorobenzenedi-
thiolate) (3). Recrystallization of the solid from dichloromethane/
ether yielded (TBA)[Co(S2C6Cl4)2] as a dark-blue-black solid in 28%
yield (0.06 g). Elem anal. Calcd: C, 39.22; H, 4.23; N, 1.63. Found: C,
40.23; H, 4.49; N, 1.62. ESI-MS: m/z 615.5 (expected), 614.5 (found).
Tetrabutylammonium Cobalt Bis(3,4,5,6-tetrafluorobenzenedi-
thiolate) (4). Recrystallization of the solid from dichloromethane/
ether yielded (TBA)[Co(S2C6F4)2] as a dark-blue-black solid in 22%
yield. ESI-MS: m/z 483.9 (expected), 483.1 (found).
3,4,5,6-Tetrafluorobenzenedithiol. A mixture of 1,2-dibromo-
3,4,5,6-tetrafluorobenzene (1 g, 3.2 mmol), sodium hydrogen sulfide
(0.75 g, 13 mmol), sulfur (0.08 g, 2.5 mmol), and iron powder (0.18 g,
3.2 mmol) was heated in 50 mL of N,N-dimethylformamide at 120 °C
for 8 h. After cooling, 100 mL of distilled water was added, and the
mixture was allowed to briefly stir at room temperature, during which
time a black precipitate formed. The precipitate was filtered off,
washed with water, and then dried in vacuo. The dried solid was added
to a suspension of zinc oxide in a 1:1 methanol/1 M aqueous sodium
hydroxide solution, which was refluxed for 1 h. After cooling, the
mixture was filtered, and the yellow filtrate was acidified with a 1 M
HCl solution to precipitate the dithiol product. The solid was dried
and recrystallized from benzene to afford 0.12 g (17% yield) of 3,4,5,6-
tetrafluorobenzenedithiol as a pale-yellow powder. 1H NMR (400
MHz, CD2Cl2): δH 4.03 (s, 2H).
6,7-Dibromo-1,4-dihydronaphthalene-1,4-epoxide. To a stirred
solution of 1,2,4,5 tetrabromobenzene (8 g, 20 mmol) and furan (10
mL) in dry toluene (200 mL) at −23 °C under argon was added
dropwise over 3 h n-BuLi (22 mmol in 200 mL of hexane). After the
mixture slowly warmed to room temperature, methanol (1 mL) was
added, the mixture was washed with water and dried, and the solvent
was removed (rotavap). The resulting yellow oily solid was triturated
with hexane to give an off-white solid. This crude product was
chromatographed on silica gel using dichloromethane/hexane (1:1) as
the eluent to give 0.7 g (74% yield) of 6,7-dibromo-1,4-
Tetrabutylammonium Cobalt Bis(toluenedithiolate) (5). Recrys-
tallization of the solid from dichloromethane/ether yielded (TBA)-
[Co(S2C7H7)2] as a dark-blue crystalline solid in 69% yield (0.420 g).
Elem anal. Calcd: C, 59.08; H, 7.93; N, 2.30. Found: C, 58.83; H, 7.97;
N, 2.35. ESI-MS: m/z 366.9 (expected), 366.9 (found). The
spectroscopic data of the product (see Figure S1) match those
available in the literature.27
1
dihydronaphthalene-1,4-epoxide as a white powder. H NMR (400
MHz, CD2Cl2): δH 5.62 (s, 2H), 6.95 (s, 2H), 7.44 (s, 2H).
1,2-Dibromonaphthalene. A suspension of zinc powder (2 g) in 60
mM dry tetrahydrofuran (THF) under a dinitrogen atmosphere was
cooled to 0 °C. Titanium tetrachloride (2 mL) was added dropwise,
and the mixture was heated to reflux for 30 min. The reaction was then
again cooled to 0 °C, and a solution of 6,7-dibromo-1,4-
dihydronaphthalene-1,4-epoxide (1.0 g, 3.3 mmol) in 20 mL of
THF was added dropwise. The mixture was refluxed overnight, cooled,
and poured into 100 mL of cold 10% HCl. The mixture was extracted
with dichloromethane, washed with water, dried with sodium sulfate,
and reduced in vacuo to an off-white powder. The crude product was
chromatographed on silica gel using dichloromethane/hexane (1:1) as
the eluent to afford 0.7 g (74% yield) of 1,2-dibromonaphthalene as a
Tetrabutylammonium Cobalt Bis(naphthalenedithiolate) (6).
Recrystallization of the solid from dichloromethane/ether yielded
(TBA)[Co(S2C10H6)2] as a dark-blue-black crystalline solid in 29%
yield (0.1 g). Crystals suitable for X-ray analysis were obtained by
layering of ether on a dilute solution of 6 in dichloromethane. Elem
anal. Calcd (CH2Cl2 1/4 lattice occupancy): C, 61.91; H, 6.90; N, 1.99.
Found: C, 62.25; H, 7.03; N, 2.02. ESI-MS: m/z 438.9 (expected),
438.9 (found).
Graphene Oxide (GO). GO for graphene depositions was
prepared via Hummer’s method:81
Sodium nitrate (0.5 g) was added to 23 mL of sulfuric acid in a large
beaker and stirred until dissolved. Next, 1 g of graphite powder was
added and stirred. After cooling in an ice bath, potassium
permanganate (3 g) was slowly added to the suspension, instigating
gas formation. The ice bath was then removed, and the temperature of
the suspension was brought up to room temperature. The reaction
beaker was then placed in an oil bath at 40 °C and stirred for 1 h. The
reaction was quenched with the slow addition of 40 mL of deionized
water, causing gas production of brown vapors. After gas production
ceased, the suspension was then further diluted with 40 mL of a 10%
H2O2 solution to convert the remaining manganese oxides into inert
sulfates, causing a change of the solution to a brown color. The
mixture was centrifuged, and the solids were washed extensively with a
mixture of 5% H2SO4 and 5% H2O2. The remaining powder was then
washed with deionized water until neutral pH was reached. The
resulting GO powder was dried in a vacuum oven at 40 °C to give 1.76
g of GO.
RGO Electrode Preparation. A total of 0.5 g of GO (prepared as
described above) was added to 30 mL of a 0.1 M sodium carbonate/
bicarbonate buffered solution (pH 9.2) in Millipore water. The
solution was stirred extensively for a long time to allow for better GO
sheet separation, becoming more viscous over time. Before deposition,
FTO-coated glass was cleaned by sonication in acetone, ethanol, and
water. The GO solution was degassed by a purge with argon gas prior
to deposition. For all depositions, the cleaned FTO-coated glass piece
was used as the working electrode in the GO solution, with a platinum
auxiliary electrode and a Ag/AgCl reference electrode. In a typical
deposition, CV would be initiated at the open-circuit potential (∼0 V)
and scanned cathodically to −1.4 V, with two cathodic scans for all
graphene surfaces used for analysis here. After deposition, the surfaces
1
white powder. H NMR (400 MHz, CD2Cl2): δH 8.15 (s, 2H), 7.75
(m, 2H,), 7.5 (m, 2H).
1,2-Naphthalenedithiol. A solution of 1,2-dibromonaphthalene (1
g, 3.5 mmol) and sodium propanethiolate (1.7 g, 17.5 mmol) in 50 mL
of N,N-dimethylacetamide was stirred under a dinitrogen atmosphere
at 100 °C for 12 h. Sodium metal (0.6 g, 26 mmol) was cut into small
pieces and added to the solution, which was allowed to continue
stirring at 100 °C for 12 h. The resulting mixture was poured into 100
mL of a 0.1 M HCl solution and subsequently extracted three times
with ether. The combined organic layer was washed with water, dried
with sodium sulfate, and reduced in vacuo to give 0.6 g (89% yield) of
1,2-naphthalenedithiol as a yellow solid. 1H NMR (400 MHz,
CD2Cl2): δH 7.9 (s, 2H), 7.42 (m, 2H), 6.64 (m, 2H), 3.95 (s, 2H).
Preparation of Cobalt Bis(benzenedithiolate) Derivatives.
The general procedure for the synthesis of cobalt bis-
(benzenedithiolene) complexes is based on the procedure reported
by Gray and co-workers:65
In a dinitrogen atmosphere glovebox, a solution of the dithiol ligand
(2.05 mmol) and sodium methoxide (0.23 g, 4.10 mmol) is added
dropwise to a suspension of cobalt(II) sulfate hexahydrate (1 mmol) in
30 mL of dry methanol, and the resulting solution is stirred for 2 h. A
solution of tetrabutylammonium bromide (1.05 mmol) in 5 mL of
methanol is added at this time, and the solution is stirred for an
additional 2 h. The solvent volume is reduced by vacuum to <10 mL,
giving a precipitate, which is filtered off, washed with methanol, and
dried. Recrystallization of the solid from dichloromethane/ether yields
the cobalt bis(benzenedithiolate) complexes as dark blue-black
microcrystalline solids.
C
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
were rinsed with deionized water to remove any loose GO from the
surface.
between 0.6 and 5.0 eV for Co 2p spectra. Additional peaks necessary
to fit all spectra used the same fitting parameters.
X-ray Crystallography. Structural Determination of [(C4H9)4N]-
[Co(S2C10H6)2] (6). Blue plates of 6 were grown from a dichloro-
methane solution of the compound layered with diethyl ether at 22 °C.
A crystal of dimensions 0.09 × 0.03 × 0.01 mm was mounted on a
Rigaku AFC10K Saturn 944+ CCD-based X-ray diffractometer
equipped with a low-temperature device and a Micromax-007HF
copper-target microfocus rotating anode (λ = 1.54187 Å) operated at
1.2 kW power (40 kV, 30 mA). The X-ray intensities were measured at
85(1) K with the detector placed at a distance of 42.00 mm from the
crystal. A total of 2028 images were collected with an oscillation width
of 1.0° in ω. The exposure times were 15 s for the low-angle images
and 120 s for the high-angle images. Rigaku d*trek images were
exported to CrysAlisPro for processing and corrected for absorption.
Integration of the data yielded a total of 25589 reflections to a
maximum 2θ value of 139.28°, of which 3213 were independent and
2640 were greater than 2σ(I). The final cell constants (Table S1) were
based on the xyz centroids of 5786 reflections above 10σ(I). Analysis
of the data showed negligible decay during data collection. The
structure was solved and refined with the Bruker SHELXTL (version
2014/6) software package, using the space group C2/c with Z = 4 for
the formula C36H48NS4Co. The cobalt dithiolate complex lies on an
inversion center of the crystal lattice. The tetrabutylammonium cation
lies on a 2-fold axis. All non-hydrogen atoms were refined
anisotropically, with the hydrogen atoms placed in idealized positions.
Full-matrix least-squares refinement based on F2 converged at R1 =
0.0806 and wR2 = 0.2126 based on I > 2σ(I) and R1 = 0.0949 and
wR2 = 0.2567 for all data. Additional details are presented in Table S1
and are given as Supporting Information.
Catalyst-Adsorbed Electrode Preparation. All catalyst-ad-
sorbed electrode surfaces reported in this work were prepared by
soaking the graphitic electrode surfaces (FTO/RGO and HOPG) in
saturated acetonitrile solutions of the cobalt bis(benzenedithiolate)
complexes for 24 h. After this period, surfaces were extensively rinsed
with fresh acetonitrile, acetone, and water. In addition, the surfaces
were electrochemically polished by cycling from 0 to −1 V to ensure
that any loosely bound catalyst was sufficiently removed prior to
electrochemical and X-ray photoelectron spectroscopy (XPS) analysis.
UV−Visible Spectroscopy. Electronic absorption spectra were
collected at room temperature using an Analytic Jena Specord S600
spectrometer with a WinASPECT (V2.2) interface.
Gas Chromatography. Dihydrogen quantification for determi-
nation of Faradaic efficiencies was performed by direct syringe
injection of the cell’s atmosphere into a Trace 1310 Thermo Scientific
gas chromatograph equipped with a thermal conductivity detector and
a flame ionization detector as well as Supelco columns. Dihydrogen in
dinitrogen mixed gases for calibration were purchased from MESA
Specialty Gas.
Electrochemical Measurements. All electrochemical measure-
ments were conducted in 18.2 Millipore water. Cyclic voltammetry
(CV) and controlled-potential coulometry were carried out using a
CHI600 electrochemical analyzer. For RGO measurements, the
working electrode was a piece of FTO-coated glass (purchased from
Pilkington) with thin layers of RGO electrodeposited on the surface by
the method described above. Testing on FTO/RGO surfaces was
performed in triplicate to ensure that no major deviation in the results
due to differences in the RGO coverage occurred. For graphite
measurements, an edge-plane highly ordered pyrolytic graphite
electrode (HOPG; purchased from Pine Instruments) was used. A
platinum wire (CH Instruments) was used as the counter electrode in
all voltammetry experiments, and the reference was an aqueous Ag/
AgCl electrode (CH Instruments; saturated AgCl and a KCl fill
solution). All potentials are reported versus the standard hydrogen
electrode (SHE) by adding 0.205 V to the potential versus Ag/AgCl/
sat. KCl. Bulk electrolysis experiments were conducted in a two-
compartment cell separated by a frit with the same working and
reference electrodes previously mentioned and a carbon felt counter
electrode (purchased from Alfa Aesar). To maintain a steady pH and
provide a more consistent activity profile during electrolysis, the pH
was monitored and adjusted back to the starting pH after increasing by
more than 0.03 units. All solutions were prepared with a 0.1 M
potassium hexafluorophosphate supporting electrolyte, purchased
from Fisher, and subsequently recrystallized. Argon gas was used to
deoxygenate all solutions for a minimum of 30 min prior to data
collection. After every compound was tested on the HOPG electrode,
the electrode was polished until no trace of catalyst activity was
observed in a pH 1.5 sulfuric acid solution and then soaked in the next
catalyst’s soaking solution. The only exception is the bulk electrolysis
of 1, where a different electrode was used because of the fact that the
previous electrode was spent from extensive polishing.
XPS. All XPS spectra were acquired with a Kratos Axis Ultra
analyzer using an Al Kα (1486.6 eV) source with a monochromator.
Spectra were recorded without charge neutralization at a base pressure
of <2.5 × 10−9 Torr. A 8 mA emission current and 14 kV anode HT
were used. Survey scans were acquired at a pass energy of 160 eV.
Because of the low level of catalyst loading, high-resolution XP spectra
of C 1s, Co 2p, Cl 2p, and S 2p recorded at a pass energy of 20 eV
provided very low signals that were difficult to visualize. For this
reason, lower-resolution data recorded at a pass energy of 160 eV are
shown here. The binding energies of all spectra were corrected by
using the difference between the observed C 1s peak energy and the
peak energy of adventitious carbon (284.6 eV).82 Spectra were fit with
a Shirley-type background using the CASAXPS software version 2.313.
Co 2p, Cl 2p, and S 2p spectra were fit using 45% Gaussian and 55%
Lorentzian line shapes. The full width at half-maximum was
constrained between 0.6 and 3.0 eV for Cl 2p and S 2p spectra and
RESULTS
■
Syntheses and Crystal Structure. The cobalt bis-
(benzenedithiolate) complexes 1−6 (general formula:
[CoIII(bdt)2]−) were all synthesized using the same general
procedure originally reported by Gray and co-workers,65
combining a cobalt(II) source with the dithiolate reagent in
the presence of a base (sodium methoxide) and a counterion
source to give the anionic cobalt(III) species as the
tetrabutylammonium salts. The microcrystalline products
were then characterized by mass spectrometry (MS), elemental
analysis, and UV−visible spectroscopy, in addition to the
crystallographic characterization for 6.
The complexes all have strong absorption features in the
near-UV (∼400 nm) and between 500 and 700 nm, with
extinction coefficients as high as 20000 and 14000 M−1 cm−1,
respectively, as shown in Figure S1. The crystal structure of 6
(Figure 1) confirms the expected square-planar structure
observed in other examples of cobalt bis(benzenedithiolate)
complexes, with a perfectly planar structure essentially giving
the molecule the “thickness” of the cobalt diameter. The
packing of the molecules in the crystal shows layers of the flat
molecules with completely eclipsed rings, separated by slightly
offset cations, with a distance of 8.3 Å between the sandwiched
cobalt bis(benzenedithiolate) molecules.
Electrode Surface Immobilization. XPS analysis of FTO
surfaces before and after RGO electrodeposition was reported
in our previous work77 and reflects the change in the carbon
signal in the C 1s core spectrum from having only adventitious
carbon on FTO to containing a variety of C−C and COx
species after RGO deposition. After soaking of the FTO/RGO
electrodes in catalyst solutions, XPS analysis of 3 and 6
adsorbed to FTO/RGO surfaces was performed at a pass
energy of 20 eV to yield high-resolution spectra. However, the
substantially lower loading in these adsorbed catalyst systems
(∼10−10 molCo cm−2) compared to those in the polymeric
D
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Cyclic Voltammetry (CV). Analysis of our cobalt bis-
(benzenedithiolate) complexes on FTO/RGO surfaces by CV
shows redox features in the range of −0.22 to −0.28 V versus
SHE, as shown in Table 1. The features persist upon extended
cycling, indicating the stability of the adsorbed species on the
electrode surface. The redox events generally appear quasi-
reversible in nature, with a higher cathodic current response
than anodic (ipc/ipa values generally below 0.9) and further
decreasing with increasing scan rate for most complexes. Peak
separation was also seen to vary between species, ranging from
just over 0.12 V for 4 to nearly 0.25 V for 6 (Figures S14 and
S21, respectively). Most of the complexes on the FTO/RGO
surfaces have Epc and Epa values that shift slightly with varying
scan rate; however, both this phenomenon and the deviation in
peak separation from the expected reversible value (0.057 V)
are also observed for the ferricyanide internal standard,
suggesting that this is inherent to the experimental conditions,
likely caused by slow electron transfer across the FTO/RGO
interface, and cannot be attributed to the adsorbed catalysts.
CV analysis of the complexes on HOPG electrodes shows
redox features ranging from −0.33 to −0.57 V, consistently
giving more negative E1/2 values compared to the correspond-
ing species on FTO/RGO surfaces. Peak anodic/cathodic
current ratios are close to 1 ( 0.1) for complexes on HOPG,
while a disproportionate increase of the cathodic current
compared to the anodic current with increasing scan rate is still
evident for all complexes with the exception of 4. As seen for
complex 6 in Figure 2, anodic and cathodic peak separations
range from 0.2 to 0.3 V, which is generally consistent with the
peak separation of the ferricyanide internal standard with the
HOPG electrodes.
Quantification of the complex loading was performed via
integration of the cathodic current response at 50 mV s−1 in a
pH 7, 0.1 M KPF6 solution by directly attributing all passed
charges to the one-electron reduction of the cobalt(III) species.
This method gives a range of catalyst concentrations from 2.70
× 10−11 mol cm−2 for 1 to 3.10 × 10−9 mol cm−2 for 6 on
HOPG. While quantitatively not highly accurate, these
estimates, when taken on the same HOPG electrode for very
similar types of complexes, allow for precise comparisons
between catalyst derivatives and thus can inform on the effect
of the ligand system on relative catalyst adsorption. Particularly
interesting are the effects of increasing halide substitution of the
benzenedithiolate ligand. Our CV data in Table 1 show a clear
increase in the catalyst loading from 1 to 2 and 3. The
Figure 1. Crystal structure of (TBA)[Co(S2C10H6)2] (6) with
ellipsoids shown at 50% probability. Hydrogen atoms are omitted
for clarity. Selected bond distances [Å] and angles [deg]: Co−S
2.171(9), S−C 1.764(5), Co---N 6.456; S−Co−S 91.68(4).
systems (10−7−10−6 molCo cm−2)67,68 made high-resolution
XPS less effective because the intensities were generally too low
to distinguish energy differences that would result from the
higher resolution, and often only trace signals were observed, if
any at all.67,68 To better visualize the features present in the
spectra, data were collected at a pass energy of 160 eV. The
data generally show two new features in the Co 2p core spectra,
comprising two sets of Co 2p1/2 and 2p3/2 signals (Figures S2b
and S3b). These signals have been attributed to the
[CoIII(bdt)2]− (778 and 794.1 eV) and [CoII(bdt•)(bdt)]−
(781 and 795 eV) resonance structures reported for cobalt
bis(benzenedithiolate) complexes.66 These results are also
consistent with those reported for both linear and extended
polymer systems of [Co(bdt)2]−.67,68 The S 2p core spectrum
also shows a new signal at 163 eV (Figures S2f and S3d), which
is approximately at a 1.5 eV higher binding energy compared to
CoS materials and consistent with XPS data of phenylthiols
previously reported in the literature.83,84 Analysis of 3 on FTO/
RGO shows similar results for the Co 2p and S 2p spectra, in
addition to showing a new signal for Cl 2p at 201 eV (Figure
S2d), which is in agreement with previous literature reports for
chloro-substituted ring systems.85 These results (especially the
unique, split cobalt XPS signals) directly confirm the molecular
nature of the complexes in the surface-bound state.
Table 1. Redox Potentials (vs SHE) and Selected Catalytic Properties of the Immobilized Cobalt Complexes
a
c
d
e
f
(V) [cat]HOPG (molCo cm−2
)
ηRGO, HOPG (V) icat (μA)
kobs (s−1
)
TOF (s−1
)
TON (time)
RGO,HOPG
1/2
RGO,HOPG
cat/2
complex
E
(V)
E
b
Co(bdt)2 (1)
Co(dcbdt)2 (2)
Co(tcbdt)2 (3)
Co(tfbdt)2 (4)
Co(tdt)2 (5)
−0.28, −0.57
−0.25, −0.52
−0.25, −0.35
−0.22, −0.33
−0.28, −0.45
−0.24, −0.46
−0.65, −0.71
−0.61, −0.67
−0.58, −0.52
−0.57, −0.54
n/a
2.70 × 10−11
2.60 × 10−10
6.22 × 10−10
1.82 × 10−9
1.45 × 10−9
3.10 × 10−9
0.57, 0.63
0.53, 0.59
0.5, 0.44
0.49, 0.46
n/a
224
220
230
120
42
800
320
140
4.30 × 106 (1.5 h)
9.10 × 106 (8 h)
4.02 × 106 (8 h)
2279
2818
2952
n/a
n/a
24
n/a
4
n/a
Co(ndt)2 (6)
n/a, −0.59
n/a, 0.51
2906
7.90 × 103 (1 h)
a
Determined by integration of the cathodic current response in cyclic voltammograms of HOPG/catalyst electrodes at 50 mV s−1 scan rate, based on
b
an HOPG electrode area of 0.2 cm2. Loading determined from CV experiments. CPE experiments for 1 were performed with a different HOPG
c
electrode because of damage to the first electrode, with a measured loading of 4.04 × 10−11 molCo cm−2. Overpotential determined by comparison of
the half-wave potential (Ecat/2) with the open-circuit potential of platinum in the same 1 mM TFA solution under 1 atm of dihydrogen (Figure S13).
d
e
Peak catalytic current observed at pH 1.5 (or at activity saturation pH if higher) on an HOPG electrode. Calculated from CV data by the use of eq
f
1 with icat determined at a scan rate of 50 mV s−1 at 298 K and pH 1.5 (on HOPG). Average TOF from CPE experiments (at pH 0.3), determined
based on the total charge passed prior to reaching the plateau current (TON time), divided by the reductive equivalents for dihydrogen formation
(2) and the total amount of catalyst present on the electrode surface.
E
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 2. Left: CV of 6 adsorbed on an HOPG working electrode in an aqueous 0.1 M KPF6 solution at varying scan rates. Right: Plot of the
cathodic and anodic peak heights as a function of the scan rate. Platinum counter and Ag/AgCl reference electrodes were used, and potentials are
reported versus SHE.
Figure 3. CV plots of 1 (top left) and 3 (bottom left) adsorbed on an HOPG working electrode in an aqueous 0.1 M KPF6 solution with the
addition of TFA at a scan rate of 50 mV s−1. The current response of the bare HOPG electrode in a TFA solution is displayed in the lower panels,
and the inset provides a zoomed view of the catalyst-adsorbed electrode in the absence of acid. The corresponding plots of peak catalytic current
versus acid concentration are presented for 1 (top right) and 3 (bottom right). Platinum counter and Ag/AgCl reference electrodes were used, and
potentials are reported versus SHE.
F
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
tetrafluoro-substituted complex 4 is observed to approximately
double the loading in comparison to the tetrachloro-substituted
species. A comparison of the benzene and naphthalene ring
systems for complexes 1 and 6 shows a substantial increase in
the catalyst loading for 6, with about 2 orders of magnitude
higher loading for the larger ring system (Table 1). A 50-fold
increase in the catalyst loading is observed from 1 to the
toluenedithiolate derivative, 5, indicating that electron-donating
substituents on the ligand ring system can also improve
adsorption. On the basis of these results alone, complex 6 with
the naphthalenedithiolate ligand is the best catalyst for
adsorption on graphitic surfaces, resulting in the best surface
coverage.
catalytic behavior consistent with complexes 1−4, which
contrasts with the inactivity/instability observed for this
complex on FTO/RGO electrodes.
For complexes 2−4 and 6, activity saturation is not observed
even at and below pH values of 1.5, whereas 1 reaches its
activity plateau at approximately 8 mM TFA (pH 2.1; see
Figure 3). Catalytic rate estimates from CV data were
determined using the ratio of the peak cathodic current in
the presence/absence of acid and the catalyst surface coverage
in eq 1, with the icat value at pH 1.5 used as a lower limit for
activity for 2−4 and 6. Here, A is the electrode surface area and
Γ stands for the catalyst surface coverage on the electrode.
icat
nFAΓ
Electrocatalytic Studies. For all complexes adsorbed on
FTO/RGO electrodes, the addition of TFA to the aqueous
solutions resulted in an increase in the cathodic current
coinciding with the disappearance of the anodic wave, indicative
of proton reduction at the electrode surface. The complexes
consistently required a pH lower than 4 to show an increase in
the current response (the exact pH at which the current
increases varies between 3 and 4 depending on the applied
complex). The current response increases linearly with further
addition of acid except in the cases of 5 and 6, which stopped
increasing and subsequently decreased to the background
current upon further acid addition (Figures S19 and S22). In
some of the repeated trials, even the initial current increase was
absent for 5 and 6 upon acid addition, and the original redox
feature was no longer observed, suggesting catalyst desorption
or decomposition. Previous results indicate that the catalysts
simply desorb from the surface over time and that this is the
main reason for the loss of activity.86 For complexes 1−4,
activity saturation was not observed up to a TFA concentration
of 2 mM (pH 2.7). Note that below this pH, the FTO
electrodes are not cathodically durable for testing.77 Ecat/2
values at this pH range from −0.57 to −0.65, with compounds
4 and 1 observed to have the most positive and negative
potentials, respectively. The overpotential is measured as the
difference between the half-wave potential (Ecat/2) of the
catalytic wave [here: in a 1 mM (pH 3) aqueous solution] and
the open-circuit potential of a platinum working electrode in
the same solution under 1 atm of H2 pressure, because this has
been determined in previous literature to be a systematic and
accurate measure of a catalyst’s performance (Figure S13).87,88
This method provides an overpotential range of 0.49−0.57 V
versus platinum for our series of complexes on FTO/RGO
electrodes. For the purpose of comparison, overpotential values
determined by more dated methods using the onset potential of
the catalytic wave range from 0.3 to 0.5 V versus platinum.89
All Ecat/2 values are at least 0.2 V more negative than the E1/2
values observed in the absence of acid, indicating that reduction
proceeds protonation in all cases during electrocatalytic proton
reduction.
kobs
=
(1)
This method provides a range of turnover frequency estimates
from 24 s−1 for 6 to 230 s−1 for 2 (see Table 1), demonstrating
that a vast difference in the activity is achievable by slight
changes in the ligand substitution. On the basis of these
estimates, the most active complexes for surface attachment are
those with partial halide substitution in the ligand system.
Adding halide substituents to the benzenedithiolate complex
(1), leading to the dichlorobenzendithiolate derivative (2),
causes a slight increase in kobs from 220 to 230 s−1; further
increasing the halide substitution in the tetrachlorobenzenedi-
thiolate derivative (3) results in a substantial decrease in the
estimated rate constant to 120 s−1. A significant decrease is also
observed when the tetrachloro-substituted (3) and tetrafluoro-
substituted (4) derivatives are compared, with respective kobs
values of 120 and 42 s−1. The size of the ligand ring system also
has a dramatic effect on the estimated rate constant, leading to
a decrease from 220 to 24 s−1 for the benzenedithiolate and
naphthalenedithiolate derivatives, respectively. While these rate
constant estimates provide insight into the effect of the ligand
substitution on the molecular TOF, it is important to also note
the change in the absolute observed activity, as better measured
by icat without normalization to the catalyst surface coverage.
One pertinent example is the comparison of 1 and 3, in which
the estimated rate of the latter is nearly half that of the former,
but the peak catalytic current is, in fact, more than an order of
magnitude higher for the tetrachloro derivative (icat = 224 vs
2818 μA) due to the drastically higher catalyst loading for 3. A
similar result is observed for 1 and 6, where the
naphthalenedithiolate derivative has a peak catalytic current
more than an order of magnitude higher than that of the
benzenedithiolate derivative (2906 and 224 μA, respectively),
despite having a 10-fold lower estimated rate. These enhance-
ments in the icat values are a direct result of increasing the
catalyst loading, as can most easily be seen in 6 compared to 1
[see Figures 3 (top) vs S22]. In summary, because the apparent
activity observed for each catalyst is dependent on both the
loading and TOF, the highest absolute activity (icat) observed in
the CV studies was not the derivative with the highest kobs (2)
or the highest loading (6) but rather the derivative with a
moderate rate and loading (3). This is best illustrated by a
direct comparison of the active catalyst derivatives on the same
HOPG electrode under the same conditions (pH 2.4 in a TFA
solution), as shown in Figure 4.
TFA titration CV studies for HOPG-adsorbed complexes in
aqueous solutions yield similar results compared to those for
the FTO/RGO electrodes, in all cases eliciting an increasing
cathodic current response upon acid addition. The increase in
the current response is linear with the acid concentration for all
complexes with the exception of 5, which undergoes a change
in response similar to that observed for this compound on
FTO/RGO (Figure S19). The resulting current response for 5
at higher acid concentrations is close to the HOPG background
activity in potential and magnitude, suggesting desorption or
decomposition of the complex. Interestingly, complex 6 exhibits
Ecat/2 values at pH 1.5 (or 2.1 in the case of 1) range from
−0.52 to −0.71, with the most positive and negative potentials
obtained for compounds 3 and 1, respectively. The over-
potential range calculated for the series on HOPG is 0.44−0.63
V versus platinum, with generally good agreement (within 0.05
G
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
complexes 2, 3, and 6, representing the highest activity and
highest catalyst loading on the graphitic surfaces. In addition,
complex 1 was included to serve as a reference point, allowing
an improved understanding of the effect of halide substitution
and increasing ligand ring system size on the catalyst activity
and retention under turnover conditions. Under an applied
potential of −0.5 V in a 0.5 M (pH 0.3) sulfuric acid solution,
complex 3 drew an initial current density of approximately 65
mA cm−2, where the background is notably only 1 mA cm−2.
The current density diminishes over the course of many hours,
reaching a plateau current of approximately 7 mA cm−2 (∼11%
of the original activity) after approximately 8 h. This current
density is maintained for the duration of the experiment
spanning over 14 h, never returning to the background current
level observed for bare HOPG (Figure 5). In addition, in situ
Figure 4. Cyclic voltammograms of active cobalt bis-
(benzenedithiolate) catalyst derivatives adsorbed on the same
HOPG working electrode (after polishing) in an aqueous 0.1 M
KPF6 solution with 4 mM TFA (pH 2.4). Platinum counter and Ag/
AgCl reference electrodes are used, and potentials are reported versus
SHE.
V) with the overpotential values on FTO/RGO (Table 1). The
compounds exhibiting the lowest overpotentials were those
with a high level of halide substitution, as in the case of 3 and 4
(η = 0.44 V), and the one with a larger ligand ring system,
complex 6 (η = 0.51 V). On average, Ecat/2 values are
approximately 0.15 V more negative than the corresponding
E1/2 values. As in the case of the FTO/RGO electrodes, the
onset potentials are more negative than the E1/2 values,
indicating that reduction precedes protonation in the electro-
catalytic mechanism for proton reduction (an EC-type
mechanism). This is in agreement with the mechanistic results
by Eisenberg and co-workers, who showed that protonation of
the formally cobalt(II) complex enables a further reduction that
leads to dihydrogen generation.27
In summary, trends in the Ecat/2 values are consistent across
our series of complexes for both FTO/RGO and HOPG
surfaces. Most significantly, upon an increase in the electron-
withdrawing nature of the benzenedithiolate ligand, a decrease
in kobs is generally observed (with the exception of the dichloro
derivative), while at the same time, the overpotential is lowered
(which is the expected trend). A shift from an Ecat/2 value of
−0.65 to −0.61 and −0.58 V is observed from 1 to 2 and then
3, respectively, on FTO/RGO electrodes, corresponding to 0,
2, and 4 chloro substitutions on each ligand. Similarly, on
HOPG, a shift from −0.71 to −0.67 V and then −0.52 V is
observed from 1 to 3. Neither surface shows a significant
difference in Ecat/2 between the tetrachloro- and tetrafluoro-
substituted ligand systems (3 and 4, respectively), with a 10 mV
lower overpotential for 4 on FTO/RGO and nearly the
opposite trend on HOPG. Extension of the ligand ring system
from benzene to naphthalene induces a shift in Ecat/2 from
−0.71 to −0.59 V, which is notably similar to the shift observed
with the ring halide substitution on going from 1 to 3 or 4.
Bulk Electrolysis. Controlled-potential electrolysis (CPE)
studies were performed on an HOPG electrode after adsorbing
the catalyst in the same manner as that used for the CV studies.
For this purpose, the most active compounds from the CV
studies described above were selected. These correspond to
Figure 5. CPE at −0.5 V versus SHE of several cobalt bis(dithiolene)
catalyst derivatives adsorbed to the same HOPG electrode in an
aqueous 0.1 M KPF6 solution with 0.5 M H2SO4 (pH 0.3). A two-
compartment cell is used with platinum counter and Ag/AgCl
reference electrodes. The arrow indicates when the pH was adjusted
back to 0.3. The inset provides a zoomed view to illustrate the limited
stability of 6.
CV of the electrode after electrolysis, while showing
substantially less current than before electrolysis, still exhibits
notably more current drawn than a bare HOPG electrode
(Figure S24).
Over the first 8 h of turnover for 3, a total of 96 C is passed
by the catalyst-adsorbed HOPG electrode, while the same
electrode without catalyst only passes 5 C in this time. To
provide a TON for comparison, the initial active species is
considered inactive after the plateau current is reached (∼8 h).
This corresponds to 9.95 × 10−4 mol of electrons passed and
approximately 5 × 10−4 mol of dihydrogen formed based on a
near-quantitative (97%) faradaic efficiency determined by gas
chromatography analysis of dihydrogen formation over the first
5 h of CPE. From this value and an HOPG catalyst loading of
1.24 × 10−10 mol determined by CV quantification for 3, we
can calculate a TON of 4.02 × 106 over the 8 h period, as well
as an average TOF of 140 s−1. In contrast, the initial rate of the
system is significantly higher, producing 18 C (9.33 × 10−5 mol
of dihydrogen) over the first 30 min to give a TOF of 418 s−1
under bulk electrolysis conditions.
H
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Further analysis of the HOPG/3 system after CPE was
conducted to ensure that the residual activity was not related to
an uncontrolled cell variable or inaccuracy in the pH
measurement. A separate HOPG electrode (never exposed to
catalyst) was tested in the same solution and confirmed a
definite difference in activity (Figure S24). Further analysis of
the HOPG/3 system after electrolysis in a fresh, pH-neutral
electrolyte solution showed a residual redox signal, albeit at a
more negative potential (E1/2 = −0.55 vs −0.35 V initially) and
with a much larger peak separation, indicative of a new species
or altered electronic environment of the remaining surface-
bound catalyst. Upon the addition of sulfuric acid to reach a pH
of 1.5, a comparison of the HOPG/3 system with the control
HOPG electrode by CV again showed substantially higher
current levels than the background level, with an onset
potential of approximately −0.45 V for the catalytic wave.
These results shown in Figure S25 confirm the presence of an
active species that remains after electrolysis and which exhibits
nearly 10 times lower current densities (at the previously
applied potential) compared to the freshly prepared electrodes.
Complex 2 was observed to have behavior very similar to that
of 3 under CPE conditions, with an initial current density of
approximately 65 mA cm−2 and an initial TOF of 835 s−1 over
the first 30 min. The current density is again seen to diminish
over the course of many hours, reaching a plateau current of
approximately 6 mA cm−2 (∼9% original) after approximately 8
h. Over this time, 91 C are passed, corresponding to 9.43 ×
10−4 mol of electrons passed and 4.71 × 10−4 mol of
dihydrogen formed. From these data and an HOPG catalyst
loading of 5.18 × 10−11 mol for 2, a TON of 9.10 × 106 can be
calculated for the 8 h period, with an average TOF of 320 s−1.
Analysis of complex 6 under electrolysis conditions shows an
initial current density of 15 mA cm−2. Unlike in the case of
complex 3, the current density decreases exponentially over the
course of the first 10 min, reaching a plateau current consistent
with background HOPG within less than an hour of
electrolysis, indicating the complete loss of catalyst activity
via desorption or catalyst degradation, in agreement with the
CV results. After the plateau current was reached at t = ∼30
min, a total of 0.95 C had passed, corresponding to 4.92 × 10−6
mol of dihydrogen produced and a TON of 7.90 × 103 based
on a catalyst loading of 6.22 × 10−10 mol. An average TOF of
4.4 s−1 is determined for the 30 min, although it is important to
note the relatively low rate combined with the short lifetime of
the catalyst produced a significantly smaller amount of current
compared to 3, so both TON and TOF estimates include some
error and are higher-end estimates. Postelectrolysis analysis,
performed analogously to complex 3, showed no substantial
redox signal above the background in a fresh pH-neutral
electrolyte solution, indicating no trace of 6 or a redox-active
decomposition product present on the surface after electrolysis.
CPE of complex 1 shows intermediate behavior compared to
that of the naphthalene and the halide-substituted derivatives.
An initial current density of 30 mA cm−2 was observed, which
also decreases more rapidly than that in the case of the halide
derivatives, although the almost immediate plateau seen for 6 is
not observed. A plateau current of less than 10% of the initial
current was reached after approximately 1.5 h of electrolysis. At
this time, a total of 6.7 C had passed, corresponding to 3.47 ×
10−5 mol of dihydrogen produced and a TON of 4.3 × 106
based on a catalyst loading of 5.39 × 10−12 mol. Despite the
large TON, significantly less dihydrogen is produced for 1 per
electrode area because of the much lower surface coverage
compared to 2 and 3. An average TOF of 800 s−1 is calculated
for the first 1.5 h for 1.
DISCUSSION
■
To design photoelectrodes and fuel-cell systems capable of
efficiently interconverting simple chemical feedstocks into
products that are useful as energy carriers or value-added
industrial feedstocks (especially dihydrogen from water,
methanol from carbon dioxide, etc.), it is a key challenge to
design electrode surfaces with catalytic properties that can then
act as heterogeneous catalysts in large-scale flow reactors. This
can be accomplished either by developing solid-state electrode
materials with catalytically active surfaces or by immobilizing a
molecular catalyst on the electrode surface. With respect to the
latter, we describe a new approach here where catalyst binding
is only based on electrostatic interactions using a thin coating of
RGO as the interface between the catalyst and a metal oxide
electrode. This strategy provides substantial merits over other
approaches in terms of simplicity and flexibility in application
because RGO can be electrodeposited on almost any
conductive material and, alternatively, can be produced by
the chemical reduction of GO and then deposited on other
substrates if necessary. The combination of versatility in
substrate selection and the ability to easily tune molecular
catalysts provides an opportunity to design nearly limitless
systems in terms of activity and overpotential. This manuscript
provides an example of such a system in which catalysts were
easily prepared and functionalized to afford chemically tuned,
highly active yet inexpensive, oxygen-stable heterogeneous
catalyst systems functional in acidic aqueous media for proton
reduction. In this way, this work highlights the advantage of
using immobilized molecular catalysts for heterogeneous
dihydrogen production systems over solid-state approaches
because the overpotential, catalyst loading, and activity can all
be directly controlled by simple ligand modifications in
molecular systems. This is in contrast to solid-state catalysts,
where catalytic sites often correspond to defect sites on the
surface that are hard to identify, characterize, and systematically
improve.
Characterization of our cobalt-catalyst-adsorbed FTO/RGO
surfaces by XPS confirms the presence of a cobalt species with
two distinct resonance states, which is a hallmark of cobalt
bis(dithiolene) complexes.66,75,76 In addition, XPS confirms the
presence of thiol and chloride components. The Co and S 2p
spectra are also consistent with the corresponding XPS signals
observed for the polymeric analogues.67,68 These results
provide strong evidence that the species adsorbed to the
surfaces are not cobalt nanoparticles or solid-state materials
such as cobalt sulfide but rather molecular cobalt complexes
that contain organic phenyl moieties as ligands. CV of the
cobalt bis(benzenedithiolate) complexes on FTO/RGO and
HOPG electrodes provides further insight into catalyst binding
to these electrode surfaces. The consistently more negative E1/2
values observed for 1−6 on HOPG versus FTO/RGO
electrodes suggest that the electronic interaction of the
complexes with the surface is different for the two materials,
which is reasonable considering the relative topography and
surface groups present for edge-plane graphite versus RGO.
More specifically, HOPG shows a regular sp2-hydridized
honeycomb lattice, whereas investigations on RGO films have
revealed the presence of many COx moieties on the surface
(alcohol, aldehyde, and carboxylate groups and epoxides) and
substantially less sp2-hybridized bonding character (∼70% area)
I
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
but also significantly more disorder and the presence of
substantially more holes (∼5% area).90−93 Variances in the
peak separation and broadness of the current signals are
commonly observed for E1/2 values in FTO/RGO/catalyst
systems and can likely be attributed to the fact that each FTO/
RGO-deposited surface varies significantly in topography and in
surface COx speciation.92,93 It is therefore advantageous to use the
more homogeneous HOPG electrodes to quantitatively compare the
catalytic activity of complexes 1−6. However, in terms of future
applications, our FTO/RGO/catalyst system provides a design
blueprint, along with catalyst activity profiles, of how other
types of semiconductors could be surface-functionalized using
an RGO interface.
A comparison of the catalyst loading on the same HOPG
electrode for the various cobalt bis(benzenedithiolate)
complexes investigated here provides an effective internal
standard to determine the effect of ligand substitution on the
surface adsorption and catalyst activity. A substantial increase in
the catalyst loading is observed with the addition of halide
substituents to the benzenedithiolate rings, at least for the
chloro and fluoro derivatives studied here. In fact, whereas the
CV signal of 1 is barely detectable above the non-Faradaic
current, 2 shows a prominent signal an order of magnitude
higher in integral intensity. A comparison of the naphthalene
ligand system with the benzene derivative shows the largest
increase in loading across the different ligand systems, effecting
a change on the scale of over 2 orders of magnitude from the
lowest observed loading in 1 to the highest in the series in 6.
Notably, the estimates for 2 and 3 are comparable to those
reported by Dey and co-workers (∼10−10 mol cm−2) for
adsorption of diiron hydrogenase model complexes on similar
EPG electrodes.24
ligand. This increase is to a large degree caused by the
sequential increase in the catalyst loading from 1 to 3; this is
particularly true in the case of 1, which has the second highest
molecular rate of all of the derivatives but suffers from a
substantially lower loading. So, from a practical point of view,
catalysts 2 and 3 are the champions, leading to the largest
amount of dihydrogen produced per electrode surface area. The
increased ring system size in 6 also results in a decrease in kobs
compared to 1. In fact, despite the large catalyst loading in 6,
this complex is observed to have the lowest catalytic activity for
proton reduction of all of the derivatives tested. On the basis of
all of these considerations, catalyst 3 is the overall champion in
terms of the peak catalytic current, overpotential under application
conditions, and stability on graphitic surfaces. Catalyst 2 provides
close competition, boasting a superior rate from CV studies and
only slightly lower retention under turnover conditions.
Additionally, 2 shines because of its ease of preparation (the
ligand is commercially available).
Analysis of complexes 1−6 adsorbed on HOPG allows for
observation of the catalysts in direct comparison at higher acid
concentrations (low pH conditions), revealing an impressive
activity level for dihydrogen production (kobs values of 24−230
s−1 at pH 1.5). The catalyst-adsorbed electrodes typically reach
current densities of 10 mA cm−2 at pH 1.5 with potentials lower
than −0.5 V versus RHE, and 2 and 3 can attain current
densities higher than 60 mA cm−2 at pH 0.3 with an applied
potential of only −0.48 V versus RHE. Electrocatalytic behavior
of the cobalt complexes is overall very similar on the HOPG
and FTO/RGO electrodes. Key features include a linear
response of the current with increasing acid concentration, a
lack of activity saturation even at higher acid concentrations,
and catalytic current onset potentials that typically occur at a
The electrocatalytic responses of complexes 1−4 adsorbed
on either FTO/RGO or HOPG are generally impressive,
demonstrating the value of the RGO interface in allowing for
fast electron transfer across the interface to a surface-
immobilized catalyst, which promises the potential use of
RGO in a variety of applications interfacing catalysts with
semiconductor electrodes. A comparison of the catalytic
behavior across our ligand series on a HOPG electrode
provides insight into the effect of ligand substitution on the
activity and overpotential. Most apparent is the trend for lower
overpotential with higher halide substitution, as observed for
complexes 1−3. Surprisingly, while the increase in halide
substitution from 1 to 3 causes a decrease of more than 100 mV
in the overpotential (η = 0.63 vs 0.44 V), the effect of
tetrafluoro substitution in 4 compared to tetrachloro
substitution in 3 on the overpotential is seemingly unbeneficial
(η = 0.46 vs 0.44 V). A possible explanation for this surprising
result is the change in the thiolate basicity: after reduction to
cobalt(II), the catalyst is protonated (must be at the thiolate
ligand, which is the only basic site) before reduction to
cobalt(I) occurs. So, at low pH, there is likely a PCET process
occurring, and while the electron-withdrawing substituents help
with the cobalt(III)/cobalt(II) reduction potential, they also
reduce the basicity of the thiolate ligand. This is a possible
explanation that, however, requires further study. The over-
potential also decreases significantly from 1 to 6 (η = 0.63 vs
0.51 V), likely because of the enhanced electron delocalization
and the somewhat electron-withdrawing properties of the larger
aromatic ring system. The peak catalytic current (icat) values are
seen to follow a similar pattern, with over a 10-fold increase
from 1 to 3 resulting from increased halide substitution of the
more negative potential than the observed E1/2.
The TOF values observed during long-term electrolysis, in
particular the initial rates of 1−3, are higher than the CV
estimates determined by eq 1. This suggests that the CV
quantification method can provide a rough estimate of the
catalytic activity but may not generally be accurate for real
application conditions. It is therefore particularly important that
new proton reduction catalysts are tested under bulk
electrolysis conditions to obtain a realistic measure of their
true electrocatalytic activity and, in particular, stability.
Sustained activity above the background current for 2 and 3
after the initial 8 h suggests that either a small percentage of the
catalyst remains active indefinitely or a more stable, less active
species is formed under turnover conditions after the first 6−8
h. These results are particularly interesting in comparison to
our previous studies on bulk (not highly ordered) graphite,
which show no Co 2p signal by XPS after the plateau current is
reached in analogous electrolysis studies.86 A similar result is
obtained here for 2 (see Figure S4). This suggests that either
some catalyst remains on the surface (below XPS detection
limit) or some other active species forms on both the bulk
graphite and HOPG electrodes over time. The possibility of a
nonmetallic active species can also not be excluded. This point
requires further study.
The TOF observed here for 3 is comparable to those
reported for surface-immobilized diiron hydrogenase models by
Dey and co-workers (adsorbed on similar EPG electrodes and
tested under identical conditions). Dey’s system is reported to
have an overpotential of 0.18 V vsersus the thermodynamic
limit of proton reduction for a pH 3 H2SO4 solution. Under
identical CPE conditions, the catalyst-adsorbed EPG disks
J
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
produce ∼5.5 × 10−4 mol of dihydrogen (vs ∼5 × 10−4 mol
obtained here for 3) in the same time frame with a similar
loading (6.22 × 10−10 mol cm−2 for 3 vs ∼10−10 mol cm−2 for
Dey’s catalyst). This ranks our heterogeneous systems among
those with the highest molecular TOFs reported so far in the
literature.24,36,94,95
The disparity in the lifetime on HOPG under CPE observed
for the halide-substituted catalyst derivatives (2 and 3) and the
derivatives with an unsubstituted ring system (1 and 6) is
substantial, with complete loss of activity for both 1 and 6 in
less than 2 h, while the current density slowly decreases for 2
and 3 over approximately 8 h. These results indicate an inherent
difference in the ligand properties that promote the initial
adsorption and those that promote catalyst retention under
turnover conditions. This suggests that larger aromatic ring
systems, such as those in 6, are beneficial for adsorption onto
graphitic surfaces, while more electron-poor ring systems, such
as those used for 2 and 3, are less prone to desorption during
turnover. This is a very interesting finding with respect to the
further optimization of our cobalt catalysts for heterogeneous
electrocatalytic proton reduction in the future.
In addition to closely matching the XPS spectra, the graphite-
adsorbed cobalt bis(benzenedithiolate) systems reported here
show very similar electrocatalytic characteristics compared to
the polymeric forms of these molecules reported by Marinescu
and co-workers.67,68 Onset potentials of approximately −0.44 V
for 3 and 6 on HOPG at pH 1.3 are very similar to those
reported for the two-dimensional (−0.28 V) and one-
dimensional (∼−0.45 V) polymer materials. A current density
of 10 mA cm−2 is reached by 3 at a potential of −0.56 V versus
SHE (−0.48 V vs RHE) at pH 1.3 (Figure 6), compared to
molecular catalyst 3 is either 3−4 orders of magnitude more
active than the polymeric catalysts or that the polymeric
materials form stacked layers on the surface, where only the top
layer(s) is/are catalytically active.
Finally, comparison should be made to nickel bis-
(diphosphine) dihydrogen production catalysts with pendant
amine side chains, which represent the gold standard for proton
reduction electrocatalysis in the homogeneous phase.4,19 Nickel
bis(diphosphine) complexes containing pendant amines
covalently bound to indium−tin oxide-supported multiwalled
carbon nanotubes (ITO/MWCNTs) reported by Le Goff and
co-workers display a catalyst loading similar to that of 6 (∼1.5
× 10−9 vs 3.1 × 10−9 mol cm−2). The covalently bound
catalysts require an exceptionally low overpotential of 20 mV in
aqueous media. CPE of the catalyst-bound ITO/MWCNT
systems in an aqueous pH 0.3 solution at a potential of −0.3 V
versus SHE shows impressive stability with no decrease in the
current density over a 10 h period and quantitative faradaic
efficiency. However, the TOF determined from these experi-
ments is somewhat low at 7 × 10−5 s−1 (∼100000 turnovers in
10 h), and a current density of only 4 mA cm−2 is observed at
−0.3 V versus SHE at pH 0.3.47 A similar system reported by
Tran and co-workers shows a similarly low average TOF of 8.2
s−1 (8.5 × 104 turnovers in 6 h with a catalyst loading of ∼2 ×
10−9 molcat cm−2).46 Excitingly, the overall catalytic properties
of the surface-immobilized cobalt complexes described in this
paper therefore rival those of the best immobilized molecular
catalyst systems reported to date.
CONCLUSIONS
■
In this work, a series of cobalt bis(benzenedithiolate)
derivatives has been prepared and adsorbed on graphitic
surfaces to create heterogeneous electrocatalysts for efficient
proton reduction. These systems exhibit excellent catalytic
performance for dihydrogen production in moderately acidic
aqueous solutions. These catalysts are simple and inexpensive
to prepare and air-stable and can easily be adsorbed onto
graphitic surfaces using a small amount of catalyst material in as
little as 12 h. The catalyst loading increases as more electron-
withdrawing substituents are added to the ligand, although an
even stronger effect is seen as the size of the aromatic ring
system increases. The overpotential decreases with increasing
halide substitution of the benzenedithiolate ligand, although the
differences observed between chloro and fluoro substituents are
small. The inductive effect of the extended ring system is also
seen to decrease the overpotential to a similar extent as the
halide substituents. Activity trends are evident across the ligand
series because turnover frequencies generally decrease with
increasing halide substitution compared to the unsubstituted
compound 1. Under turnover conditions, halide-substituted
derivatives slowly decrease in activity over the course of ∼8 h,
while naphthalene derivatives rapidly deactivate within an hour,
despite higher initial catalyst loading. Total turnovers of ∼107
in 8 h are accomplished in CPE experiments under real
application conditions, with an average TOF of up to 320 s−1
over the 8 h period for the dichloro-substituted catalyst.
Figure 6. Polarization curves of active cobalt bis(benzenedithiolate)
catalyst derivatives adsorbed on the same HOPG working electrode in
an aqueous 0.1 M KPF6 solution with 50 mM TFA (pH 1.3). Platinum
counter and Ag/AgCl reference electrodes were used, and potentials
are reported versus SHE.
−0.34 V versus RHE reported for the two-dimensional MOS1
on glassy carbon and −0.56 V versus RHE reported for the
one-dimensional polymer on glassy carbon. The current density
of 3 is particularly impressive given that the catalyst loading is
approximately 3−4 orders of magnitude lower than that
reported for the polymer systems, while still reaching the 10
mA cm−2 threshold at similar potentials. This implies that the
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.7b01589.
K
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Index of figures, details for overpotential and rate
Article
(12) Canaguier, S.; Artero, V.; Fontecave, M. Modelling NiFe
hydrogenases: nickel-based electrocatalysts for hydrogen production. J.
Chem. Soc., Dalton Trans. 2008, 315−325.
constant calculations, and full crystallographic data sets
(PDF)
́
(13) Capon, J.-F.; Ezzaher, S.; Gloaguen, F.; Petillon, F. Y.;
Schollhammer, P.; Talarmin, J. Electrochemical Insights into the
Mechanisms of Proton Reduction by [Fe2(CO)6{μ-SCH2N(R)-
CH2S}] Complexes Related to the [2Fe]H Subsite of [FeFe]-
Hydrogenase. Chem. - Eur. J. 2008, 14, 1954−1964.
Accession Codes
CCDC 1557448 contains the supplementary crystallographic
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.
(14) Baffert, C.; Artero, V.; Fontecave, M. Cobaloximes as Functional
Models for Hydrogenases. 2. Proton Electroreduction Catalyzed by
Difluoroborylbis(dimethylglyoximato)cobalt(II) Complexes in Organ-
ic Media. Inorg. Chem. 2007, 46, 1817−1824.
(15) Yu, Z.; Wang, M.; Li, P.; Dong, W.; Wang, F.; Sun, L. Diiron
dithiolate complexes containing intra-ligand NH···S hydrogen bonds:
[FeFe] hydrogenase active site models for the electrochemical proton
reduction of HOAc with low overpotential. J. Chem. Soc., Dalton Trans.
2008, 2400−2406.
(16) Roy, S.; Mazinani, S. K. S.; Groy, T. L.; Gan, L.; Tarakeshwar,
P.; Mujica, V.; Jones, A. K. Catalytic Hydrogen Evolution by Fe(II)
Carbonyls Featuring a Dithiolate and a Chelating Phosphine. Inorg.
Chem. 2014, 53, 8919−8929.
(17) McKone, J. R.; Marinescu, S. C.; Brunschwig, B. S.; Winkler, J.
R.; Gray, H. B. Earth-abundant hydrogen evolution electrocatalysts.
Chem. Sci. 2014, 5, 865−878.
(18) Wang, M.; Chen, L.; Sun, L. Recent progress in electrochemical
hydrogen production with earth-abundant metal complexes as
catalysts. Energy Environ. Sci. 2012, 5, 6763−6778.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: lehnertn@umich.edu (N.L.).
ORCID
Nicolai Lehnert: 0000-0002-5221-5498
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
Notes
The authors declare no competing financial interest.
(19) Helm, M. L.; Stewart, M. P.; Bullock, R. M.; DuBois, M. R.;
DuBois, D. L. A Synthetic Nickel Electrocatalyst with a Turnover
Frequency Above 100,000 s−1 for H2 Production. Science 2011, 333,
863−866.
(20) Losse, S.; Vos, J. G.; Rau, S. Catalytic hydrogen production at
cobalt centres. Coord. Chem. Rev. 2010, 254, 2492−2504.
(21) Tard, C.; Pickett, C. J. Structural and Functional Analogues of
the Active Sites of the [Fe]-, [NiFe]-, and [FeFe]-Hydrogenases.
Chem. Rev. 2009, 109, 2245−2274.
ACKNOWLEDGMENTS
■
N.L. acknowledges funding from 3M (Grant NTFG 5286067)
for this work. Further support to S.C.E. was provided by the
Rackham Graduate School (University of Michigan). Acknowl-
edgement is also made for funding from NSF Grants CHE-
0840456, DMR-0420785, and DMR-0320740 for X-ray
crystallography, XPS, and scanning electron microscopy
instrumentation, respectively.
(22) Stewart, M. P.; Ho, M.-H.; Wiese, S.; Lindstrom, M. L.;
Thogerson, C. E.; Raugei, S.; Bullock, R. M.; Helm, M. L. High
Catalytic Rates for Hydrogen Production Using Nickel Electrocatalysts
with Seven-Membered Cyclic Diphosphine Ligands Containing One
Pendant Amine. J. Am. Chem. Soc. 2013, 135, 6033−6046.
(23) Eckenhoff, W. T.; Eisenberg, R. Molecular systems for light
driven hydrogen production. Dalton Trans. 2012, 41, 13004−13021.
(24) Dey, S.; Rana, A.; Dey, S. G.; Dey, A. Electrochemical Hydrogen
Production in Acidic Water by an Azadithiolate Bridged Synthetic
Hydrogenese Mimic: Role of Aqueous Solvation in Lowering
Overpotential. ACS Catal. 2013, 3, 429−436.
(25) Eckenhoff, W. T.; McNamara, W. R.; Du, P.; Eisenberg, R.
Cobalt complexes as artificial hydrogenases for the reductive side of
water splitting. Biochim. Biophys. Acta, Bioenerg. 2013, 1827, 958−973.
(26) McNamara, W. R.; Han, Z.; Alperin, P. J.; Brennessel, W. W.;
Holland, P. L.; Eisenberg, R. A CobaltDithiolene Complex for the
Photocatalytic and Electrocatalytic Reduction of Protons. J. Am. Chem.
Soc. 2011, 133, 15368−15371.
REFERENCES
■
(1) Nocera, D. G. The Artificial Leaf. Acc. Chem. Res. 2012, 45, 767−
776.
(2) Dempsey, J. L.; Brunschwig, B. S.; Winkler, J. R.; Gray, H. B.
Hydrogen Evolution Catalyzed by Cobaloximes. Acc. Chem. Res. 2009,
42, 1995−2004.
(3) Kim, D.; Sakimoto, K. K.; Hong, D.; Yang, P. Artificial
Photosynthesis for Sustainable Fuel and Chemical Production.
Angew. Chem., Int. Ed. 2015, 54, 3259−3266.
(4) Rakowski DuBois, M.; DuBois, D. L. Development of Molecular
Electrocatalysts for CO2 Reduction and H2 Production/Oxidation. Acc.
Chem. Res. 2009, 42, 1974−1982.
(5) Stolley, R. M.; Helm, M. L. Light-harvesting materials: Soft
support for energy conversion. Nat. Chem. 2014, 6, 949−950.
(6) Artero, V.; Chavarot-Kerlidou, M.; Fontecave, M. Splitting Water
with Cobalt. Angew. Chem., Int. Ed. 2011, 50, 7238−7266.
́
(7) Saveant, J.-M. Molecular Catalysis of Electrochemical Reactions.
(27) McNamara, W. R.; Han, Z.; Yin, C. J.; Brennessel, W. W.;
Holland, P. L.; Eisenberg, R. Cobalt-dithiolene complexes for the
photocatalytic and electrocatalytic reduction of protons in aqueous
solutions. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 15594−15599.
(28) Lakadamyali, F.; Kato, M.; Muresan, N. M.; Reisner, E. Selective
Reduction of Aqueous Protons to Hydrogen with a Synthetic
Cobaloxime Catalyst in the Presence of Atmospheric Oxygen.
Angew. Chem., Int. Ed. 2012, 51, 9381−9384.
Mechanistic Aspects. Chem. Rev. 2008, 108, 2348−2378.
(8) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi,
Q.; Santori, E. A.; Lewis, N. S. Solar Water Splitting Cells. Chem. Rev.
2010, 110, 6446−6473.
(9) Valdez, C. N.; Dempsey, J. L.; Brunschwig, B. S.; Winkler, J. R.;
Gray, H. B. Catalytic hydrogen evolution from a covalently linked
dicobaloxime. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 15589−15593.
(10) Sala, X.; Romero, I.; Rodriguez, M.; Escriche, L.; Llobet, A.
Molecular Catalysts that Oxidize Water to Dioxygen. Angew. Chem.,
Int. Ed. 2009, 48, 2842−2852.
(11) Beyler, M.; Ezzaher, S.; Karnahl, M.; Santoni, M.-P.; Lomoth, R.;
Ott, S. Pentacoordinate iron complexes as functional models of the
distal iron in [FeFe] hydrogenases. Chem. Commun. 2011, 47, 11662−
11664.
(29) Gao, W.; Ekstrom, J.; Liu, J.; Chen, C.; Eriksson, L.; Weng, L.;
̈
Åkermark, B.; Sun, L. Binuclear Iron-Sulfur Complexes with Bidentate
Phosphine Ligands as Active Site Models of Fe-Hydrogenase and
Their Catalytic Proton Reduction. Inorg. Chem. 2007, 46, 1981−1991.
(30) Duan, L.; Wang, M.; Li, P.; Na, Y.; Wang, N.; Sun, L. Carbene−
pyridine chelating 2Fe2S hydrogenase model complexes as highly
L
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
active catalysts for the electrochemical reduction of protons from weak
acid (HOAc). J. Chem. Soc., Dalton Trans. 2007, 1277−1283.
(31) Cheah, M. H.; Tard, C.; Borg, S. J.; Liu, X.; Ibrahim, S. K.;
Pickett, C. J.; Best, S. P. Modeling [Fe-Fe] Hydrogenase: Evidence for
Bridging Carbonyl and Distal Iron Coordination Vacancy in an
Electrocatalytically Competent Proton Reduction by an Iron Thiolate
Assembly That Operates through Fe(0)-Fe(II) Levels. J. Am. Chem.
Soc. 2007, 129, 11085−11092.
Complexes: Carbon Monoxide Tolerant Catalysts for Hydrogen
Evolution and Uptake. Angew. Chem., Int. Ed. 2011, 50, 1371−1374.
(47) Le Goff, A.; Artero, V.; Jousselme, B.; Tran, P. D.; Guillet, N.;
Metaye, R.; Fihri, A.; Palacin, S.; Fontecave, M. From Hydrogenases to
́ ́
Noble Metal−Free Catalytic Nanomaterials for H2 Production and
Uptake. Science 2009, 326, 1384−1387.
(48) Krawicz, A.; Yang, J.; Anzenberg, E.; Yano, J.; Sharp, I. D.;
Moore, G. F. Photofunctional Construct That Interfaces Molecular
Cobalt-Based Catalysts for H2 Production to a Visible-Light-Absorbing
Semiconductor. J. Am. Chem. Soc. 2013, 135, 11861−11868.
(49) Seo, J.; Pekarek, R. T.; Rose, M. J. Photoelectrochemical
operation of a surface-bound, nickel-phosphine H2 evolution catalyst
on p-Si(111): a molecular semiconductor|catalyst construct. Chem.
Commun. 2015, 51, 13264−13267.
(50) Rao, H.; Wang, Z.-Y.; Zheng, H.-Q.; Wang, X.-B.; Pan, C.-M.;
Fan, Y.-T.; Hou, H.-W. Photocatalytic hydrogen evolution from a
cobalt/nickel complex with dithiolene ligands under irradiation with
visible light. Catal. Sci. Technol. 2015, 5, 2332−2339.
(51) Fei, H.; Dong, J.; Arellano-Jimenez, M. J.; Ye, G.; Dong Kim, N.;
Samuel, E. L. G.; Peng, Z.; Zhu, Z.; Qin, F.; Bao, J.; Yacaman, M. J.;
Ajayan, P. M.; Chen, D.; Tour, J. M. Atomic cobalt on nitrogen-doped
graphene for hydrogen generation. Nat. Commun. 2015, 6, 8668.
(52) Bullock, R. M.; Appel, A. M.; Helm, M. L. Production of
hydrogen by electrocatalysis: making the H-H bond by combining
protons and hydrides. Chem. Commun. 2014, 50, 3125−3143.
(53) Cardenas, A. J. P.; Ginovska, B.; Kumar, N.; Hou, J.; Raugei, S.;
Helm, M. L.; Appel, A. M.; Bullock, R. M.; O’Hagan, M. Controlling
Proton Delivery through Catalyst Structural Dynamics. Angew. Chem.,
Int. Ed. 2016, 55, 13509−13513.
(54) Stolley, R. M.; Darmon, J. M.; Das, P.; Helm, M. L. Nickel Bis-
Diphosphine Complexes: Controlling the Binding and Heterolysis of
H2. Organometallics 2016, 35, 2965−2974.
(55) Reback, M. L.; Buchko, G. W.; Kier, B. L.; Ginovska-Pangovska,
B.; Xiong, Y.; Lense, S.; Hou, J.; Roberts, J. A. S.; Sorensen, C. M.;
Raugei, S.; Squier, T. C.; Shaw, W. J. Enzyme Design from the Bottom
Up: An Active Nickel Electrocatalyst with a Structured Peptide Outer
Coordination Sphere. Chem. - Eur. J. 2014, 20, 1510−1514.
(56) Zou, X.; Huang, X.; Goswami, A.; Silva, R.; Sathe, B. R.;
(32) Bigi, J. P.; Hanna, T. E.; Harman, W. H.; Chang, A.; Chang, C. J.
Electrocatalytic reduction of protons to hydrogen by a water-
compatible cobalt polypyridyl platform. Chem. Commun. 2010, 46,
958−960.
́ ̀
(33) Pantani, O.; Naskar, S.; Guillot, R.; Millet, P.; Anxolabehere-
Mallart, E.; Aukauloo, A. Cobalt Clathrochelate Complexes as
Hydrogen-Producing Catalysts. Angew. Chem., Int. Ed. 2008, 47,
9948−9950.
(34) McCrory, C. C. L.; Uyeda, C.; Peters, J. C. Electrocatalytic
Hydrogen Evolution in Acidic Water with Molecular Cobalt
Tetraazamacrocycles. J. Am. Chem. Soc. 2012, 134, 3164−3170.
(35) Hu, X.; Brunschwig, B. S.; Peters, J. C. Electrocatalytic
Hydrogen Evolution at Low Overpotentials by Cobalt Macrocyclic
Glyoxime and Tetraimine Complexes. J. Am. Chem. Soc. 2007, 129,
8988−8998.
(36) Berben, L. A.; Peters, J. C. Hydrogen evolution by cobalt
tetraimine catalysts adsorbed on electrode surfaces. Chem. Commun.
2010, 46, 398−400.
(37) Andreiadis, E. S.; Jacques, P.-A.; Tran, P. D.; Leyris, A.;
Chavarot-Kerlidou, M.; Jousselme, B.; Matheron, M.; Pecaut, J.;
Palacin, S.; Fontecave, M.; Artero, V. Molecular engineering of a
cobalt-based electrocatalytic nanomaterial for H2 evolution under fully
aqueous conditions. Nat. Chem. 2012, 5, 48−53.
(38) Carroll, M. E.; Barton, B. E.; Gray, D. L.; Mack, A. E.;
Rauchfuss, T. B. Active-Site Models for the Nickel−Iron Hydro-
genases: Effects of Ligands on Reactivity and Catalytic Properties.
Inorg. Chem. 2011, 50, 9554−9563.
(39) Gartner, F.; Boddien, A.; Barsch, E.; Fumino, K.; Losse, S.;
̈
Junge, H.; Hollmann, D.; Bruckner, A.; Ludwig, R.; Beller, M.
̈
Photocatalytic Hydrogen Generation from Water with Iron Carbonyl
Phosphine Complexes: Improved Water Reduction Catalysts and
Mechanistic Insights. Chem. - Eur. J. 2011, 17, 6425−6436.
(40) Ott, S.; Kritikos, M.; Åkermark, B.; Sun, L. Synthesis and
Structure of a Biomimetic Model of the Iron Hydrogenase Active Site
Covalently Linked to a Ruthenium Photosensitizer. Angew. Chem., Int.
Ed. 2003, 42, 3285−3288.
́
Mikmekova, E.; Asefa, T. Cobalt-Embedded Nitrogen-Rich Carbon
Nanotubes Efficiently Catalyze Hydrogen Evolution Reaction at All
pH Values. Angew. Chem. 2014, 126, 4461−4465.
(57) Chen, W.-F.; Muckerman, J. T.; Fujita, E. Recent developments
in transition metal carbides and nitrides as hydrogen evolution
electrocatalysts. Chem. Commun. 2013, 49, 8896−8909.
(58) Hou, J.; Fang, M.; Cardenas, A. J. P.; Shaw, W. J.; Helm, M. L.;
Bullock, R. M.; Roberts, J. A. S.; O’Hagan, M. Electrocatalytic H2
production with a turnover frequency > 107 s−1: the medium provides
an increase in rate but not overpotential. Energy Environ. Sci. 2014, 7,
4013−4017.
(59) Gao, M.-R.; Liang, J.-X.; Zheng, Y.-R.; Xu, Y.-F.; Jiang, J.; Gao,
Q.; Li, J.; Yu, S.-H. An efficient molybdenum disulfide/cobalt
diselenide hybrid catalyst for electrochemical hydrogen generation.
Nat. Commun. 2015, 6, 5982.
(60) Yin, Y.; Han, J.; Zhang, Y.; Zhang, X.; Xu, P.; Yuan, Q.; Samad,
L.; Wang, X.; Wang, Y.; Zhang, Z.; Zhang, P.; Cao, X.; Song, B.; Jin, S.
Contributions of Phase, Sulfur Vacancies, and Edges to the Hydrogen
Evolution Reaction Catalytic Activity of Porous Molybdenum
Disulfide Nanosheets. J. Am. Chem. Soc. 2016, 138, 7965−7972.
(61) Lassalle-Kaiser, B.; Merki, D.; Vrubel, H.; Gul, S.; Yachandra, V.
K.; Hu, X.; Yano, J. Evidence from in Situ X-ray Absorption
Spectroscopy for the Involvement of Terminal Disulfide in the
Reduction of Protons by an Amorphous Molybdenum Sulfide
Electrocatalyst. J. Am. Chem. Soc. 2015, 137, 314−321.
(62) Karunadasa, H. I.; Montalvo, E.; Sun, Y.; Majda, M.; Long, J. R.;
Chang, C. J. A Molecular MoS2 Edge Site Mimic for Catalytic
Hydrogen Generation. Science 2012, 335, 698−702.
(63) Yang, Y.; Fei, H.; Ruan, G.; Tour, J. M. Porous Cobalt-Based
Thin Film as a Bifunctional Catalyst for Hydrogen Generation and
Oxygen Generation. Adv. Mater. 2015, 27, 3175−3180.
(41) Wiedner, E. S.; Yang, J. Y.; Dougherty, W. G.; Kassel, W. S.;
Bullock, R. M.; DuBois, M. R.; DuBois, D. L. Comparison of Cobalt
and Nickel Complexes with Sterically Demanding Cyclic Diphosphine
Ligands: Electrocatalytic H2 Production by [Co(PtBu2NPh2)-
(CH3CN)3](BF4)2. Organometallics 2010, 29, 5390−5401.
(42) Orthaber, A.; Karnahl, M.; Tschierlei, S.; Streich, D.; Stein, M.;
Ott, S. Coordination and conformational isomers in mononuclear iron
complexes with pertinence to the [FeFe] hydrogenase active site.
Dalton Trans. 2014, 43, 4537−4549.
(43) Eady, S. C.; Breault, T.; Thompson, L.; Lehnert, N. Highly
functionalizable penta-coordinate iron hydrogen production catalysts
with low overpotentials. Dalton Trans. 2016, 45, 1138−1151.
(44) Ibrahim, S. K.; Liu, X.; Tard, C.; Pickett, C. J. Electropolymeric
materials incorporating subsite structures related to iron-only
hydrogenase: active ester functionalised poly(pyrroles) for covalent
binding of {2Fe3S}-carbonyl/cyanide assemblies. Chem. Commun.
2007, 1535−1537.
(45) Chiou, T.-W.; Lu, T.-T.; Wu, Y.-H.; Yu, Y.-J.; Chu, L.-K.; Liaw,
W.-F. Development of a Dinitrosyl Iron Complex Molecular Catalyst
into a Hydrogen Evolution Cathode. Angew. Chem., Int. Ed. 2015, 54,
14824−14829.
(46) Tran, P. D.; Le Goff, A.; Heidkamp, J.; Jousselme, B.; Guillet,
N.; Palacin, S.; Dau, H.; Fontecave, M.; Artero, V. Noncovalent
Modification of Carbon Nanotubes with Pyrene-Functionalized Nickel
M
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(64) Blakemore, J. D.; Gupta, A.; Warren, J. J.; Brunschwig, B. S.;
Gray, H. B. Noncovalent Immobilization of Electrocatalysts on Carbon
Electrodes for Fuel Production. J. Am. Chem. Soc. 2013, 135, 18288−
18291.
(65) Baker-Hawkes, M. J.; Billig, E.; Gray, H. B. Characterization and
Electronic Structures of Metal Complexes Containing Benzene-1,2-
dithiolate and Related Ligands. J. Am. Chem. Soc. 1966, 88, 4870−
4875.
Crystalline Si(111) and Si(100) Surfaces with Br2, I2, and Ferrocenium
in Alcohol Solvents. J. Phys. Chem. B 2002, 106, 3639−3656.
(83) Lindberg, B. J.; Hamrin, K.; Johansson, G.; Gelius, U.; Fahlman,
A.; Nordling, C.; Siegbahn, K. Molecular Spectroscopy by Means of
ESCA II. Sulfur compounds. Correlation of electron binding energy
with structure. Phys. Scr. 1970, 1, 286.
(84) Peng, S.; Li, L.; Han, X.; Sun, W.; Srinivasan, M.; Mhaisalkar, S.
G.; Cheng, F.; Yan, Q.; Chen, J.; Ramakrishna, S. Cobalt Sulfide
Nanosheet/Graphene/Carbon Nanotube Nanocomposites as Flexible
Electrodes for Hydrogen Evolution. Angew. Chem. 2014, 126, 12802−
12807.
(66) Sproules, S.; Wieghardt, K. Dithiolene radicals: Sulfur K-edge X-
ray absorption spectroscopy and Harry’s intuition. Coord. Chem. Rev.
2011, 255, 837−860.
(85) Clark, D. T.; Kilcast, D.; Musgrave, W. K. R. Molecular core
binding energies for some monosubstituted benzenes, as determined
by X-ray photoelectron spectroscopy. J. Chem. Soc. D 1971, 516−518.
(86) Eady, S. C.; MacInnes, M. M.; Lehnert, N. A Smorgasbord of
Carbon: Electrochemical Analysis of Cobalt−Bis(benzenedithiolate)
Complex Adsorption and Electrocatalytic Activity on Diverse
Graphitic Supports. ACS Appl. Mater. Interfaces 2016, 8, 23624−
23634; Correction: ACS Appl. Mater. Interfaces 2017, 9, 15898−
15899.
(87) Fourmond, V.; Jacques, P.-A.; Fontecave, M.; Artero, V. H2
Evolution and Molecular Electrocatalysts: Determination of Over-
potentials and Effect of Homoconjugation. Inorg. Chem. 2010, 49,
10338−10347.
(88) Hoffert, W. A.; Roberts, J. A. S.; Morris Bullock, R.; Helm, M. L.
Production of H2 at fast rates using a nickel electrocatalyst in water-
acetonitrile solutions. Chem. Commun. 2013, 49, 7767−7769.
(89) Felton, G. A. N.; Glass, R. S.; Lichtenberger, D. L.; Evans, D. H.
Iron-Only Hydrogenase Mimics. Thermodynamic Aspects of the Use
of Electrochemistry to Evaluate Catalytic Efficiency for Hydrogen
Generation. Inorg. Chem. 2006, 45, 9181−9184.
(90) Park, S. I.; Quate, C. F. Tunneling microscopy of graphite in air.
Appl. Phys. Lett. 1986, 48, 112−114.
(91) Gwo, S.; Shih, C. K. Site-selective imaging in scanning tunneling
microscopy of graphite: The nature of site asymmetry. Phys. Rev. B:
Condens. Matter Mater. Phys. 1993, 47, 13059−13062.
(92) Dreyer, D. R.; Park, S.; Bielawski, C. W.; Ruoff, R. S. The
chemistry of graphene oxide. Chem. Soc. Rev. 2010, 39, 228−240.
(93) Erickson, K.; Erni, R.; Lee, Z.; Alem, N.; Gannett, W.; Zettl, A.
Determination of the Local Chemical Structure of Graphene Oxide
and Reduced Graphene Oxide. Adv. Mater. 2010, 22, 4467−4472.
(94) Karunadasa, H. I.; Chang, C. J.; Long, J. R. A molecular
molybdenum-oxo catalyst for generating hydrogen from water. Nature
2010, 464, 1329−1333.
(95) Quentel, F.; Passard, G.; Gloaguen, F. Electrochemical hydrogen
production in aqueous micellar solution by a diiron benzenedithiolate
complex relevant to [FeFe] hydrogenases. Energy Environ. Sci. 2012, 5,
7757−7761.
(67) Downes, C. A.; Marinescu, S. C. Efficient Electrochemical and
Photoelectrochemical H2 Production from Water by a Cobalt
Dithiolene One-Dimensional Metal−Organic Surface. J. Am. Chem.
Soc. 2015, 137, 13740−13743.
(68) Clough, A. J.; Yoo, J. W.; Mecklenburg, M. H.; Marinescu, S. C.
Two-Dimensional Metal−Organic Surfaces for Efficient Hydrogen
Evolution from Water. J. Am. Chem. Soc. 2015, 137, 118−121.
(69) Letko, C. S.; Panetier, J. A.; Head-Gordon, M.; Tilley, T. D.
Mechanism of the Electrocatalytic Reduction of Protons with
Diaryldithiolene Cobalt Complexes. J. Am. Chem. Soc. 2014, 136,
9364−9376.
(70) Seidel, W. W.; Hahn, F. E. Chelate complexes of cobalt(III) with
bis(dithiolate) ligands: backbone influence on the electronic properties
and the reactivity of the metal center. J. Chem. Soc., Dalton Trans.
1999, 2237−2242.
(71) Nomura, M.; Fujita-Takayama, C.; Yagisawa, T.; Sugiyama, T.;
Kajitani, M. An organometallic dithiolene complex exhibiting electro-
chemically initiated hydrogen generation. Dalton Trans. 2013, 42,
4764−4767.
(72) Seidel, W. W.; Hahn, F. E. Chances and Limits of the
Coordination Chemistry with Bis(benzene-l,2-dithiolato) Ligands.
Bioinorg. Chem. Appl. 2005, 3, 69−80.
̌
̌ ́
(73) Machata, P.; Herich, P.; Luspai, K.; Bucinsky, L.; Soralova, S.;
Breza, M.; Kozisek, J.; Rapta, P. Redox Reactions of Nickel, Copper,
and Cobalt Complexes with “Noninnocent” Dithiolate Ligands:
Combined in Situ Spectroelectrochemical and Theoretical Study.
Organometallics 2014, 33, 4846−4859.
(74) Solis, B. H.; Hammes-Schiffer, S. Computational Study of
Anomalous Reduction Potentials for Hydrogen Evolution Catalyzed
by Cobalt Dithiolene Complexes. J. Am. Chem. Soc. 2012, 134, 15253−
15256.
(75) Eisenberg, R.; Gray, H. B. Noninnocence in Metal Complexes:
A Dithiolene Dawn. Inorg. Chem. 2011, 50, 9741−9751.
(76) Periyasamy, G.; Burton, N. A.; Hillier, I. H.; Vincent, M. A.;
Disley, H.; McMaster, J.; Garner, C. D. The dithiolene ligand-
’innocent’ or ’non-innocent’? A theoretical and experimental study of
some cobalt-dithiolene complexes. Faraday Discuss. 2007, 135, 469−
488.
(77) Eady, S. C.; Peczonczyk, S. L.; Maldonado, S.; Lehnert, N. Facile
heterogenization of a cobalt catalyst via graphene adsorption: robust
and versatile dihydrogen production system. Chem. Commun. 2014,
50, 8065−8068.
(78) Weingarten, A. S.; Kazantsev, R. V.; Palmer, L. C.; McClendon,
M.; Koltonow, A. R.; Samuel, A. P. S.; Kiebala, D. J.; Wasielewski, M.
R.; Stupp, S. I. Self-assembling hydrogel scaffolds for photocatalytic
hydrogen production. Nat. Chem. 2014, 6, 964−970.
(79) Hart, H.; Bashir-Hashemi, A.; Luo, J.; Meador, M. A. Iptycenes:
Extended triptycenes. Tetrahedron 1986, 42, 1641−1654.
(80) Testaferri, L.; Tiecco, M.; Tingoli, M.; Chianelli, D.;
Montanucci, M. Simple Syntheses of Aryl Alkyl Thioethers and of
Aromatic Thiols from Unactivated Aryl Halides and Efficient Methods
for Selective Dealkylation of Aryl Alkyl Ethers and Thioethers.
Synthesis 1983, 1983, 751−755.
(81) Hummers, W. S.; Offeman, R. E. Preparation of Graphitic
Oxide. J. Am. Chem. Soc. 1958, 80, 1339−1339.
(82) Haber, J. A.; Lewis, N. S. Infrared and X-ray Photoelectron
Spectroscopic Studies of the Reactions of Hydrogen-Terminated
N
DOI: 10.1021/acs.inorgchem.7b01589
Inorg. Chem. XXXX, XXX, XXX−XXX