Electroreduction of CO2 to Formate with Low Overpotential using Cobalt Pyridine Thiolate Complexes

Article abstract of DOI:10.1002/anie.202006269

Electrocatalytic CO₂ reduction to value-added products provides a viable alternative to the use of carbon sources derived from fossil fuels. Carrying out these transformations at reasonable energetic costs, for example, with low overpotential, remains a challenge. Molecular catalysts allow fine control of activity and selectivity via tuning of their coordination sphere and ligand set. Herein we investigate a series of cobalt(III) pyridine-thiolate complexes as electrocatalysts for CO₂ reduction. The effect of the ligands and proton sources on activity was examined. We identified bipyridine bis(2-pyridinethiolato) cobalt(III) hexaflurophosphate as a highly selective catalyst for formate production operating at a low overpotential of 110 mV with a turnover frequency (TOF) of 10 s^?1. Electrokinetic analysis coupled with density functional theory (DFT) computations established the mechanistic pathway, highlighting the role of metal hydride intermediates. The catalysts deactivate via the formation of stable cobalt carbonyl complexes, but the active species could be regenerated upon oxidation and release of coordinated CO ligands.

Full text of DOI:10.1002/anie.202006269

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Electroreduction of CO2 to Formate with low overpotential using
Cobalt Pyridine Thiolate Complexes
Authors: Subal Dey, Tanya Todorova, Marc Fontecave, and Victor
Mougel
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202006269
Link to VoR: https://doi.org/10.1002/anie.202006269

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
Electroreduction of CO2 to Formate with low overpotential using
Cobalt Pyridine Thiolate Complexes
Subal Dey,[a],[b] Tanya K Todorova,[b] Marc Fontecave,[b]* and Victor Mougel[a],[b]*
[a]
[b]
Department of Chemistry and Applied Biosciences, ETH Zürich, Vladimir-Prelog-Weg 1-5, 8093 Zürich (Switzerland)
Laboratoire de Chimie des Processus Biologiques, UMR 8229 CNRS, Collège de France, Paris, Sorbonne Université, PSL Research University, 11 Place
Marcelin Berthelot, 75231 Paris, Cedex 05 (France)
E-mail: marc.fontecave@college-de-france.fr (M. Fontecave)
E-mail: mougel@inorg.chem.ethz (V. Mougel)
Supporting information for this article is given via a link at the end of the document.
Abstract: Electrocatalytic CO2 reduction to value-added products
provides a viable alternative to the use of carbon sources derived from
fossil fuels. Nevertheless, the ability to carry out these transformations
at reasonable energetic costs, e.g. with low overpotential, remains a
significant challenge. Molecular catalysts offer a great option in this
context, as fine control of their activity and selectivity can be obtained
via the tuning of their coordination sphere and ligand set. To this end,
we investigated here a series of cheap cobalt(III) pyridine-thiolate
complexes as electrocatalysts for CO2 reduction. The effect of the
ligands and proton sources on activity was examined. We were able
ligands are soft Lewis bases with σ and πdonation properties,
providing increased electron density on reduced metals while
maintaining strong metal-ligand covalency. These two factors are
beneficial to the reactivity of the reduced complexes with CO2.
However, because of their anionic character, thiolate ligands also
disfavor the reduction of the complex, as the active species
require more cathodic potential to be generated. This can, in turn,
result in substantial overpotentials as well as less effective
catalysts in terms of turnover numbers (TONs). Here, we show
that highly active Co-based CO2 reduction molecular catalysts
can be obtained via the modulation of the electronic influence of
anionic thiolate ligands with neutral nitrogen ligands. We prepared
a series of 2-pyridinethiolate Co complexes notably combined
with substituted bipyridine (bpy) ligands. These complexes,
obtained in one synthetic step from cheap commercially available
reagents, are capable of catalyzing electrochemical CO2
reduction at very low overpotential with high selectivity towards
carbon-based products. We report the investigation of their
catalytic activity in the presence of different proton sources and
ligand sets and propose a mechanistic explanation to the
observed activity supported by theoretical calculations.
to
identify
[bipyridine-bis-(2-pyridinethiolato)-cobalt(III)-
hexaflurophosphate] as a highly selective catalyst for formate
production operating at a very low overpotential of 110 mV to achieve
a TOF of 10 s-1. Detailed electrokinetic analysis coupled with density
functional theory allowed establishing a mechanistic pathway for
these catalysts, highlighting the role of key metal hydride
intermediates. The catalysts deactivate via the formation of stable Co
carbonyl complexes, but we demonstrated that the active species
could be regenerated upon oxidation and release of coordinated CO
ligands.
Introduction
Results and Discussion
Carbon dioxide can be used as a carbon source alternative to
fossil-fuels for the production of energy-dense chemicals by
electrolysis utilizing renewable energy sources. The inherent
stability of CO2 has triggered a large number of studies aimed at
developing active catalysts to promote its reduction.[1] Early
studies have focused on low-valent rare metal catalysts (e.g.
Re,[2] Ru,[3] Ir,[4] Rh,[5]) and have been more recently extended to
earth-abundant transition metals (e. g. Mn,[6] Fe,[7] Co,[8] Ni[9]). This
large number of molecular catalysts has enabled understanding
the mechanistic pathways for CO2 reduction and elucidating the
electronic factors responsible for the catalyst selectivity to either
CO or HCOOH.[1] However, the vast majority of them still suffer
from high overpotentials (generally 300-800 mV).[1, 10]
Synthesis and Characterization
We investigated here the series of Co complexes presented in
Scheme
hexaflurophosphate] (1) and [4,4´-bismethoxy-bipyridine-bis-2-
pyridinethiolato-cobalt(III)-hexaflurophosphate] (2) were
1.
[bipyridine-bis-(2-pyridinethiolato)-cobalt(III)-
synthetized in 67 and 83% yield respectively by addition of 2
equivalents of sodium pyridine-2-thiolate to a methanol solution of
[bipyridyl-dichloro-cobalt(II)], by analogy with reported procedures
for the corresponding nickel(II) complexes[14] and subsequent
aerobic oxidation. The perchlorate analogue of 1 had been
previously prepared using a different synthetic route but had not
been extensively characterized.[15] Complex
3
[tris-2-
The major advantage of molecular catalysts is that they offer an
easy modulation of their catalytic properties by tuning the
electronics of the ligands. Yet surprisingly, in the context of CO2
reduction a vast majority of the reported catalysts show very little
variation in their 1st coordination sphere, where nitrogen-based
donor ligands are used in most cases. Despite being ubiquitous
in the enzymatic systems competent for CO2 transformation
pyridinethiolato-cobalt(III)] was synthesized in 61% yield via the
salt metathesis reaction of 3 equivalents of sodium pyridine-2-
thiolate and Co(II)Cl2 in methanol followed by aerobic oxidation.
+
R
S
Co
S
N
Co
N
S
S
S
N
N
N
N
N
R
(carbon
monoxide
dehydrogenase[11]
and
formate
dehydrogenase[12]), thiolate ligands had not been used for the
design of CO2 reduction catalysts until recently.[13] Thiolate
1, R = H
2, R = OMe
3
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
Scheme 1: Co pyridine-thiolate complexes 1-3.
between -0.77 V and -1.1 V (all potentials are reported here vs.
Fc/Fc+) and peak to peak separation values (Ep) between 75 and
97 mV. By analogy with previously reported data for complex 1,
we assigned this redox wave to a CoIII/II process.[21]
Complexes 1-3 were fully characterized by elemental analysis,
UV-Vis, 1H-NMR, HR-MS and single crystal X-ray crystallography.
The X-ray single crystal structures of the complexes 1-3 are
shown in Figure 1. Both complexes 1 and 2 crystallize in a
distorted octahedral geometry with the S ligands trans to each
other in the axial positions (S-Co-S angle = 165°) and the four N
positioned in the equatorial plane. The Co-S bond distances of
2.304(4) Å and 2.288(1) Å and the average Co-N bond distances
of 1.932(2) Å and 1.929(2) Å for 1 and 2 are in good agreement
with previously reported low-spin Co(III) complexes bearing
thiolate[16] and pyridine[17] ligands. The main difference between
the structures of 1 and 2 resides in the C-C bond distance
between the two pyridine rings of the bpy moiety of 1.467 and
1.486 Å for 1 and 2 respectively, correlating with the stronger
electron donating properties of the -OMe groups. Due to the
asymmetry induced by its ligand set, complex 3 deviates slightly
more from octahedral geometry, having two S ligands in axial
positions (S-Co-S angle = 164°) and an equatorial plane defined
by the third sulfur and the three nitrogen ligands. The notable
difference of this new series of complexes with previously
reported Fe(II)[18] and Ni(II) analogues[19] is the much shorter M-S
bond distances consistent with the +III oxidation state of Co.
Figure 2: Electrochemical behaviour of the complexes. a. Cyclic
voltammograms of complex 1, 2 and 3 under argon. CV scans for various
potential windows are overlaid (dashed lines) for complex 1 only. b, c, d are the
overlay of the CV responses for 1, 2 and 3 respectively under catalytic condition,
under argon in the absence (black) or presence (blue) of TFE, in the presence
of CO2 (red) and in the presence of TFE and CO2 (green). All data were collected
using 1 mM solution of the complexes in CH3CN/0.1 M (nBu)4NPF6 at 100 mVs-
1 scan rates using GC working electrodes.
Table 1. Electrochemical characteristics of complex 1-3.
Complex
E1/2 II/I (V)a
EpIII/IIb (mV)
DcatIII/II,b cm2s-1
E1/2 III/II (V)
Figure 1: Solid-state molecular structure of complexes 1-3 (left to right) Counter
anion, interstitial solvent molecules and H atoms were omitted for clarity. Carbon
(grey), cobalt (pink), nitrogen (blue) and sulfur (yellow) atoms are represented
with 50% probability ellipsoids.
1
2
3
-0.77
-0.81
-1.02
-1.85
-2.02
-2.35
75
82
97
3.3 10-6
3.1 10-6
3.0 10-6
1H NMR spectra confirm the diamagnetism of these complexes,
as expected for low spin Co(III) complexes. UV-Vis data (Figure
S1) showed three main absorptions peaks for 1 at 313, 377 and
613 nm respectively. Very similar absorption features are found
for in the UV-Vis spectrum of complex 2 (326, 373 and 607 nm),
while that of complex 3 shows a strong absorption at 280 mm
along with three consecutive shoulders at 298, 324, 365 and a
broad band at 700 nm. The broad low energy maxima with lower
intensity (ε < 100 L mol-1 cm-1) at ~610 nm for 1 and 2 and at 700
nm for 3 are attributed to spin-allowed d-d transitions, as expected
for pseudo-octahedral low-spin Co(III) complexes.[20] The red shift
of the thiolate-to-metal charge transfer in 3 suggests an increased
electron density at the metal center, in agreement with the
presence of the three anionic thiolate ligands.
a The peak to peak separation (ΔEp) for CoII/I processes are not provided due to
large irreversibility of the CVs. b The diffusion coefficients are mentioned for the
CoIII/II processes, in cathodic scan.
On further cathodic scan, two successive irreversible one-
electron redox processes appear for 1 at -1.85 V and -2.05 V
respectively (Figure 2a, dashed line). The redox process at -1.85
V is assigned as reduction of Co(II) to Co(I) species, while the
process at -2.05 V is attributed to the reduction of the bpy ligand,
in agreement with previously reported values for Ni or Fe
complexes containing bpy ligands.[22] We ascribed the increased
irreversibility of the CoII/I process at lower scan rates (Figure S2)
to the dissociation of a pyridine-thiolate ligand, as observed for
corresponding Ni analogues where dissociation of the pyridine
moiety of 2-pyridinethiolate ligand had been established (Scheme
2).[14] This hypothesis is further supported by the appearance of a
new pre-wave to the Co(II) to Co(III) oxidation feature at - 0.95 V
on anodic scan and assigned to the oxidation of this five
coordinated Co(I) species. Complex 2 presents a similar redox
behavior with an overall cathodic shift of the E1/2 values (Figure
2a green, Figure S3 and Table 2), in agreement with the electron-
donating properties of the –OMe groups. This has a stronger
impact on the most reduced state, a cathodic shift of 170 mV
being observed for E1/2 (CoII/I). The tris pyridine-thiolate complex
Additional high energy absorptions with high ε observed for all
complexes can be attributed to the charge transfers from thiolate
to metal and the inter ligand charge transfer (π – π*).
Electrochemical Data:
a. Cyclic voltammetry under Argon: The electrochemical
properties of the complexes were investigated by cyclic
voltammetry (CV) in 0.1 M (nBu)4NPF6 anhydrous acetonitrile
solution under an Argon atmosphere (Figure 2a, and Table 1). All
complexes 1-3 presented a first reversible reduction wave at a
half-wave potential (E1/2) (determined as the average of the
cathodic peak potential Ep and the anodic peak potential Ea)
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
3 shows a reversible redox peak at a more cathodic value of -1.04
V attributed to the CoIII/II couple (Figure 2a red and Figure S3). An
irreversible feature was observed at the very cathodic potential of
-2.35 V. We hypothesize that this irreversible feature is associated
with a permanent ligand dissociation from the Co(I) center. This
hypothesis is further confirmed by the appearance of an additional
oxidation wave at ca. -0.4V at the higher scan rates for 1 and 2
(Figure 2 and Figures S2-S3), assigned to the oxidation process
of the thus generated pentacoordinated Co(II) complexes. The
observation of ligand dissociation events for all complexes
suggests that open coordination sites are generated upon
reduction, a prerequisite for binding an exogeneous ligand (e. g.
CO2, proton etc.). All the complexes show free diffusion behavior
in solution according to Randles-Sevcik plot, which allows
homogeneous electrocatalytic investigations to be performed
(see ESI for details).
saturated (~0.23 M) CH3CN solution. The electrochemical data for
all complexes are given in Figure 2. In the presence of CO2,
irreversible features with current enhancement were observed for
complexes 1-3 at the potential of the CoII/I redox couple (red
traces of Figure 2b, 2c and 2d and half-peak potentials given in
Table 2). Half-peak potential values suggest that CO2 activation
occurs on Co(I) complexes. The modest current enhancement
encouraged us to explore the use of proton sources to enhance
the rates and lower the overpotential of the reaction.[6a, 23] Using
complex 1 as a prototypical catalyst to identify the best reaction
conditions, we observed that the Icat/Ip value was substantially
increased upon addition in the electrochemical solution of 2,2,2-
trifluoroethanol (TFE), H2O, or phenol (PhOH). The largest value
was observed using TFE (see Figures S6-S8 and ESI for details).
The catalytic performances of 1 were hence investigated by 2 h
controlled potential electrolysis (CPE) at -1.83 V in the presence
of these proton sources (Figure S9-S10 and Table S3). The
highest overall FY for CO2 reduction value of 64 % (7 % of CO
and 57 % of HCOOH, together with 19 % for H2) was obtained
using TFE, which was thus chosen for the investigation of the
catalytic activity of the other complexes described here. With
respect to complex 1, 2 h CPE at -1.98V of a CO2-saturated
solution of complex 2 in the presence of TFE (Figure S13a)
revealed an increase of the selectivity for CO2 reduction (70 %
FY) accompanied by a lower undesired H2 evolution (10 % FY),
yet at the cost of a more cathodic potential (Ecat/2 = -1.93 V).
Product analysis after 2 h CPE for 3 at -2.33 V (Figure S13b)
revealed high selectivity for HCOOH (57 % FY) with low CO and
H2 generation (2.5 % and <1% respectively, Table 2).
-
N
N
S
Co
e-
S
N
N
N
N
Co
S
N
S
N
Scheme 2: Proposed redox assisted pyridine dissociation for the complexes.
DFT optimization of complex 1 showed that it has a singlet ground
state and its computed geometry is in excellent agreement with
the experimental X-ray single crystal structure (bond distances
are within 0.01 Å and bond angles are within 1 of the
corresponding values in the crystal structure) (Table S2). Its one-
The high selectivity of complex 3 is nevertheless counterbalanced
by its low total FY of ca. 60 % and very cathodic catalytic potential
(Ecat/2 = -2.35 V). This low total FY goes along with lower turnover
numbers (TONs) and a fast decrease of the current during CPE
(Figure S13b and table S3), highlighting the lower stability of
complex 3 with respect to complexes 1 and 2 (Table 2).
For all three catalysts and all proton sources, the total FYs
reported above are below unity and vary between 61 and 83 %.
These low total FYs result from the deactivation of the catalysts
in catalytic conditions, further discussed in the section below
“catalyst deactivation mechanism”.
electron reduction leads to the formation of a CoII species, 1red1
,
which has a quartet spin state that is 2.4 kcal mol-1 lower in energy
than the doublet state. Our DFT calculations show that addition of
an extra electron in 1 results in ~0.2 Å elongation of all six Co–L
distances in the high-spin structure 1red1, while in the low-spin 1red1
the Co–S bonds are elongated to as much as 0.39 Å, and only
~0.01 Å for the Co–N bonds. The electron resides on the Co
center (Figure S5B), which has a Mulliken spin density of 2.63,
implying an oxidation state of +2. One-electron reduction of 1red1
generates a high-spin triplet CoI species 1red2 (open-shell singlet
is 14 kcal mol-1 higher in energy) as illustrated by the orbital
analysis (Figure S5C). Computations indicate the existence of two
different geometries, i.e., 1red2 and 1red2’ that are less than 1 kcal
mol-1 different in energy. 1red2 preserves the coordination sphere
around the CoI center, but the two Co–S bond lengths are further
elongated by 0.24 Å, while the Co–N distances are shortened by
0.17 Å (Table S2). In contrast, in 1red2’ one Co–NPyS bond is
cleaved, while the other Co–N bond distances are almost
unchanged, and Co–S ones are shortened by ~0.2 Å, in
agreement with the ligand dissociation process mentioned above.
b. Catalytic behavior under CO2: The electrochemical CO2
reduction ability of these complexes was investigated in CO2
c. Benchmarking of the Catalysts: In order to compare their
catalytic performances, we evaluated the CO2 reduction kinetics
of complexes 1-3 by determining their maximum turnover
frequency (TOFmax, see ESI for details). TOFmax values of 27.5 s-
and 29.5 s-1 were obtained for 1 and 2 respectively, indicating
1
that the introduction of electron-donating substituent on the bpy
moiety does not significantly increase the rate of the reaction
despite driving the operating potential to more cathodic values. By
contrast, the presence of a third pyridine-thiolate ligand strongly
influenced the reaction kinetics, with a TOFmax value of 178 s-1
determined for complex 3. We benchmarked the catalytic
performances of catalysts 1-3 against other molecular catalysts
Table 2. Comparative catalytic parameters for complex 1-3 in presence of CO2 and TFE.
a
e
Catalyst
Ecat/2
-1.74
-1.93
-2.35
TFE conc.b
0.11 M
TOFmax (s-1)c
27.5
Total TONd
FYCO (%)d
FYformate (%)d
FYH2 (%)d
η10s-1
110
280
600
1
2
3
5.4
5.1
2.4
7
57
64
57
19
10
< 1
0.07 M
29.5
6
0.07 M
178
2.5
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
aThe half catalytic peak potentials, Ecat/2, were determined from the potential corresponding to the value at half of the catalytic current. boptimal TFE concentrations
were determined by CV studies presented in Figures S10 and S14-15. cTOFmax was determined according to equation (1). dFYs and TONs were determined after
2h CPE at -1.83 V for 1, -1.98 V for 2 and -2.33 V for 3. e η10s-1 is the overpotential required to achieve a TOF of 10 s-1.
using a catalytic Tafel plot, presented in Figure 3 (see ESI for
details on the construction of the Tafel plot).[24] The catalytic Tafel
plot allows examining the catalytic performances against both
S18). This points to the formation of a Co(I) carbonyl complex
during catalytic turnover,[29] accumulating in electrocatalytic
conditions and leading to deactivation. Such
a catalyst
kinetic (TOFmax) and thermodynamic (overpotential, η) descriptors. deactivation by formation of a stable carbonyl complex has been
It should be noted here that the TOFmax and Tafel plot could be
confounding as the complexes reported in this manuscript show
selectivity for HCOOH below unity. Nevertheless, plotting the
turnover frequencies for formate determined from the CPE at
different applied overpotentials on the Tafel plot highlights their
overall match with the logTOF-η plots determined by CV
measurements (Figure 3, see ESI for details). The TOF values
determined from the CPE data are slightly higher than those
determined by CV measurements, in agreement with the
additional convection resulting from the stirring applied during
CPE experiments. This confirms the validity of the Tafel plot for
complexes 1-3 and enables comparison with previously reported
catalysts for the reduction of CO2 to HCOOH, namely
previously reported for several molecular CO2 reduction
electrocatalysts.[29-30]
To further test that hypothesis, we attempted to regenerate the
active catalyst by re-oxidation promoting the CO release. A 30 min
controlled potential electrolysis at – 0.23 V was carried out
immediately after 2 h CPE of complex 1 at -1.83 V under a CO2
atmosphere in optimized catalytic conditions (0.11 M TFE). The
decay of current in this re-oxidation step occurred concomitantly
with a release of a substantial amount of CO in the headspace.
Considering this amount and the charge stored under the form of
the Co(I)-CO complex allows assigning over 95% of the passed
current during electrolysis and a substantial increase of the total
FY of 16% (Table S5). Regeneration of the active catalyst was
confirmed by the CV of complex 1 after this re-oxidation step that
is indistinguishable from that obtained from freshly prepared
solutions (Figure S19).
2- [13a]
+ [26]
Ir(POCOP),[4] HFe4N(CO)12,[25] Ni(qpdt)2
,
FeN5Cl2 ,
Pt(dmpe)2
,
2+ [27] IminobpyCo3+[28] and CpCo(P2N2)2+.[8c]
Similar inhibition behavior was observed for complex 2, also
resulting in FYs close to unity (Figure S20; Table S5). Complex 2
however presented a faster decrease of current over CPE,
suggesting that the CO inhibition was much faster than for
complex 1 (Figure S20 and Figure S9). Despite that faster catalyst
inhibition, we could also demonstrate the integrity of catalyst 2
after the oxidation step: initial catalytic activity could be essentially
restored for a second CPE at -1.98 V carried out after the
oxidation step (Figure S20 and table S6; see ESI for details).
The same procedures were applied for complex 3. Contrary to
what observed with complexes 1 and 2, only trace amounts of CO
were released (<1 % of FY) for 3 upon re-oxidation after 2 h CPE
and catalytic activity could not be restored.
These data indicate that the deactivation is of different origin for
complexes 1-2 and 3: the former two are inhibited by the
reversible binding of CO on the reduced catalyst, while the latter
undergoes irreversible degradation. This can be explained by the
electronic properties of the ligands. In the case of complex 3, the
strong electron density provided by the three thiolate ligands likely
destabilizes the Co(I) state, as highlighted by the irreversibility of
the CoII/I process in the CV. This results in a fast degradation of
the catalyst, in agreement with the fast drop of current during
electrolysis and low TONs. This deactivation was further
confirmed by high-resolution SEM images and EDX analysis of
the electrode surface after electrolysis, showing the deposition of
Figure 3: Catalytic Tafel plot for complex 1 (black line), 2 (red line) and 3 (blue
line). The data are presented in comparison with other selective molecular
catalysts for the reduction of CO2 to HCOOH (dashed lines, see inset). Crosses
represent the TOFs for formate determined from CPE data. Values and
conditions are given in ESI.
It clearly appears that complex 1 stands among the best reported
CO2 to HCOOH catalysts: achieving a TOF of 10 s-1 requires less
than 110 mV overpotential for 1, which is the lowest value
reported so far. While the catalysts Ir(POCOP), CpCo(P2N2)+ and
2-
Ni(qpdt)2 show relatively higher TOFmax values, in the same
range than that for 3, this comes at the cost of a substantial
overpotential (~600 mV, Table 2).[4, 8c, 13a]
d. Catalyst deactivation mechanism: As described above, the
complexes reported here all present total FYs significantly below
unity. This fact, combined with the gradual decay of the catalytic
current density over time during electrolysis (Figure S9, S13a and
S13b), points toward a deactivation of the catalysts in operating
conditions. In the case of complexes 1 and 2, we observed a
gradual deepening of the color of the electrolytic solution upon
electrolysis. On the other hand, no precipitate and no particle
formation could be observed during electrolysis, and a rinse test
(see Figure S14 and ESI for details) proved negative. In addition,
high resolution SEM images and EDX analysis of the electrode
surface confirmed the absence of deposit at the electrode after
electrolysis (Figure S15-S16). Monitoring complex 1 in CO2
electrocatalytic conditions with in-situ IR-SEC revealed the
gradual appearance of a new νCO stretch at 1893 cm-1 (Figure
nanoparticles containing Co and
S
(Figure S17). The
decomposition of complex 3 to such a cobalt-sulfide material
necessitates at least 3 electrons per complex decomposed,
potentially explaining the lower FYs observed. In contrast, the
presence of neutral bipyridine ligands in complexes 1 and 2 allows
lowering the overall charge of the metal center and increased
stabilization of the Co(I) species, which gets inhibited via the
formation of a CO adduct. The faster inhibition reaction observed
with complex 2 can be rationalized by the electron-donating -OMe
substituents on the bipyridine ligand, which induce an increased
electron density on the Co center and strengthen the metal to CO
backdonation.
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
Reaction mechanism: In all cases HCOOH was the main
product of the reduction, together with small amounts of CO and
H2. Our discussion will here mainly concern complex 1, which is
showing the best catalytic performances.
The anodic potential shift in the CV in the presence of a proton
source (Figure 2a and table 1) shows that the generation of the
main active species involves a proton transfer. Our calculations
indicate that the most favorable protonation site of both singly-
(1red1) and doubly- (1red2) reduced complexes is the nitrogen atom
of the pyridine-thiolate ligand (NPyS). This is in full agreement with
the reduced Co center favoring the thione form of the pyridine-
thiolate ligand, confirmed by the shortening of the C-S bond of the
ligand upon protonation by ca. 0.5 Å. The resulting species are
Two possible reaction pathways can lead to formate, via (i) the
2-
coordination of CO2 to the Co(I) center affording a Co(III)-CO2
intermediate followed by the reaction with weak Bronsted acids
leading to formate after reorganization[1, 26, 31] or (ii) via the direct
reaction of the Co(I) center with H+ forming a Co(III)-H species,
which can react with CO2 to generate formate. To identify which
reaction path is operational here, we investigated the kinetic
parameters of this reaction in the case of complex 1. The catalytic
current correlates linearly with proton and catalyst concentration
(Figure S6-S7 and S21), in agreement with a second-order
dependency of the rate with proton concentration and a first-order
dependency with catalyst concentration (see ESI for details).
Furthermore, a first-order dependency of the rate with CO2
concentration was observed (Figure S12). A primary kinetic
isotope effect (KIE) of 8.2 was determined using water as a proton
source, indicating that the rate-determining step (RDS) involves a
proton transfer (Figure S23). This fairly large primary KIE
suggests that the reaction takes place via the formation of Co-H
species,[8c] in further agreement with the observation of H2 as the
second main product of the reaction. In that case, the employed
proton source plays a key role in the competitive protonation of
the M-H species affording H2.[1, 31]
Yang and Kubiak have recently established a thermodynamic
framework using the hydricity of the metal complex and the pKa of
the proton source as descriptors to rationalize the reactivity of
metal hydrides with CO2 and H+ and the associated selectivity of
the catalyst.[27, 32] We determined a hydricity (ΔGH-) value of 38
kcal mol-1 for the Co(III)-H species generated from 1 using the
empirical equation developed by Kubiak et. al.[32] and standard
reduction potential (E1/2II/I) (see ESI for details). This ΔGH- value is
lower than that for formate in acetonitrile (44 kcal mol-1), showing
that the hydride transfer to CO2 leading to formate is exergonic.
The pKa of TFE in CO2-saturated acetonitrile of 25.1 exactly
matches the pKa defining the initial pH point from which selective
CO2 reduction can be achieved.[27] According to the
thermodynamic product diagram illustrating metal hydride
reactivity with protons and CO2 in acetonitrile,[27] complex 1
operates in a zone where a competitive H2 evolution occurs
concomitantly to formate production. This also explains the larger
amount of H2 production for complex 1 in the presence of more
acidic PhOH or water (i.e. carbonic acid in the presence of CO2).
Nevertheless, the proximity to the selective CO2 reduction to
formate area of the diagram ensures formate being the main
product of the reaction. In the case of complexes 2 and 3 the
hydricity of the corresponding Co(III)-H species (ΔGH- = 33 and 26
kcal mol-1, respectively) is increased by the more electron-
donating nature of the ligands and the selectivity for formate is
thus increased, but at the expense of a higher thermodynamic
cost.
1
red1–NH in a quartet state and 1red2–NH in a triplet state (the
corresponding low-spin structure is 15.4 kcal mol-1 higher in
energy, see Table S7). Intramolecular proton transfer from NPyS
to the Co center has a computed transition state (TS) barrier of
34.8 kcal mol-1, indicating that the formation of Co(III)-H hydride
species cannot occur via this pathway, in line with the stabilization
of the pyridine-thiolate ligand in the thione form upon protonation
mentioned above. Instead, direct protonation of the CoI site by
TFE is more favorable to occur with a TS barrier of 20.1 kcal mol-
1, affording the Co(III)–H hydride species 1-CoH (MOs shown in
Figure S24). The high activation energy is in agreement with the
hydride formation being the RDS step. The transition state
structure for this second proton transfer highlights the stabilization
of the incoming TFE by hydrogen bonding with the protonated
pyridine (Figure S25). The CoIII–H can promote a hydride transfer
to CO2 to generate the Co-HCO2 complex 1-HCO2, with a
transition state barrier of 14.5 kcal mol-1 (TS1 structure in Figure
S23a), which upon CO2 insertion converts to intermediate 1-
OCHO (G = -3.2 kcal mol-1). Further protonation leads to the
release of HCOOH and the recovery of catalyst 1.
The competitive formation of hydrogen is proposed to occur via
the direct protonation of the Co(III)–H bond by a TFE molecule,
leading to an H–H bond in a CoH2 adduct. The computed
transition state barrier is 20.2 kcal mol-1, in agreement with the
findings of formate being the major product. The TFE interacts via
hydrogen-bonding with the protonated NPyS, facilitating the proton
transfer to the Co–H unit (TS2 structure in Figure S26b).Our
calculations could not locate a transition state involving the Co–H
and N–H units and implying an intramolecular pathway for the H2
formation.
Last, CO2 binding on complex 1red2-NH in a 1-CO2(C) fashion is
endergonic by 11 kcal mol-1, affording the species 1red2-CO2.
Binding in a 2-CO2(C,O) or 1-OCO manner is 10.4 and 17.2 kcal
mol-1 energetically less favored. Intramolecular proton transfer
from the protonated NPyS ligand yields a carboxylate intermediate
1-COOH with a TS barrier of only 2.5 kcal mol-1. The easy
intramolecular proton transfer observed here is consistent with
the destabilization of the thione form of the ligands upon oxidation
of the Co center to Co(III). This further highlights the adaptability
of the pyridine-thiolate ligand and its influence on the pyridine
nitrogen protonation: low Co oxidation states favor the thione form
and the protonation of the nitrogen while high oxidation states
favor the thiolate form and deprotonated pyridine.
In the presence of TFE, 1-COOH undergoes a heterolytic
cleavage of the C–O bond (G‡ = 25.6 kcal mol-1), to produce the
CO-bound complex 1-CO and water (MOs are shown in Figures
S27-S28). 1-CO can release CO and regenerate complex 1 or
alternatively be further reduced by 2 electrons to afford the Co(I)-
CO complex 1red2-CO. This species has a characteristic CO
stretching at 1907 cm-1, in close agreement with the deactivation
species observed in IR-SEC experiments.
The observation of CO as a minor product suggests that the
catalyst can follow another reaction pathway involving the initial
binding of CO2 to reduced Co(I) center followed by a C-O bond
cleavage mediated by protonation.[2b, 33]
These experimental findings along with DFT computations on
complex 1, allowed us to propose the overall reaction mechanism
of CO2 reduction to HCOOH, H2 and CO as shown in Scheme 3.
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
H
[5]
[6]
a) C. Caix, S. Chardon-Noblat, A. Deronzier, J. Electroanal. Chem. 1997,
434, 163-170; b) T. K. Todorova, T. N. Huan, X. Wang, H. Agarwala, M.
Fontecave, Inorg. Chem. 2019, 58, 6893-6903.
H
N
O
N
CO
Co
N
H+
O
Co
N
CO
N
S
S
N
N
S
HCOOH
N
S
H2
O
-H+
N
G = -3.2
1-CO
1-OCHO
a) M. Bourrez, F. Molton, S. Chardon-Noblat, A. Deronzier, Angew.
Chem. Int. Ed. 2011, 50, 9903-9906; b) M. D. Sampson, A. D. Nguyen,
K. A. Grice, C. E. Moore, A. L. Rheingold, C. P. Kubiak, J. Am. Chem.
Soc. 2014, 136, 5460-5471; c) B. Reuillard, K. H. Ly, T. E. Rosser, M. F.
Kuehnel, I. Zebger, E. Reisner, J. Am. Chem. Soc. 2017, 139, 14425-
14435.
S
N
G = 25.6
N
S
Co
H+
N
1
O
O
H
N
H2
HO
O
N
H
N
S
S
N
Co
S
N
N
Co
2e-
H+
S
N
N
G = 20.2
1-COOH
1-HCO2
NH
[7]
[8]
[9]
a) C. Costentin, S. Drouet, M. Robert, J.-M. Savéant, Science 2012, 338,
90; b) M. Hammouche, D. Lexa, M. Momenteau, J. M. Saveant, J. Am.
Chem. Soc. 1991, 113, 8455-8466; c) C. Costentin, M. Robert, J.-M.
Savéant, Acc. Chem. Res. 2015, 48, 2996-3006; d) A. W. Nichols, S.
Chatterjee, M. Sabat, C. W. Machan, Inorg. Chem. 2018, 57, 2111-2121.
a) D. C. Lacy, C. C. L. McCrory, J. C. Peters, Inorg. Chem. 2014, 53,
4980-4988; b) N. Elgrishi, M. B. Chambers, M. Fontecave, Chem. Sci.
2015, 6, 2522-2531; c) S. Roy, B. Sharma, J. Pécaut, P. Simon, M.
Fontecave, P. D. Tran, E. Derat, V. Artero, J. Am. Chem. Soc. 2017, 139,
3685-3696.
G = 14.5
G = 2.5
N
S
Co
N
S
H
H
N
G = 20.1
H
N
N
G = 11.0
CO2
O
O
N
S
N
S
CO2
H+
Co
Co
N
N
S
N
S
1red2-NH
N
1-CO2
1-CoH
Scheme 3: Proposed reaction mechanism for the generation of HCOOH, H2
and CO. The relative Gibbs free energies (G, kcal mol-1) and transition state
barriers (G‡, kcal mol-1) are given relative to the preceding intermediate
a) M. F. Kuehnel, K. L. Orchard, K. E. Dalle, E. Reisner, J. Am. Chem.
Soc. 2017, 139, 7217-7223; b) D. Hong, Y. Tsukakoshi, H. Kotani, T.
Ishizuka, T. Kojima, J. Am. Chem. Soc. 2017, 139, 6538-6541.
Conclusion
This work allowed isolating a series of highly active Co-based
CO2 electroreduction catalysts. We achieved the conversion of
CO2 to HCOOH as the major product with very low overpotential
by tuning the electronic properties of the metal centers using
pyridine-thiolate ligands. Benchmarking of these complexes
allowed identifying complex 1 as one of the best catalysts
amongst all CO2 to HCOOH molecular electrocatalysts to date.
We identified here that the formation of a stable carbonyl complex
was at the origin of the catalyst deactivation. We demonstrated
the possibility to regenerate the active catalyst by re-oxidation of
such carbonyl resting states, promoting the CO ligand release.
The electronic and mechanistic understanding provided by this
study highlights the interest of using thiolate-based ligands in the
context of CO2 reduction.
[10] H. Takeda, C. Cometto, O. Ishitani, M. Robert, ACS Catal. 2017, 7, 70-
88.
[11] a) M. Can, F. A. Armstrong, S. W. Ragsdale, Chem. Rev. 2014, 114,
4149-4174; b) R. Hille, J. Hall, P. Basu, Chem. Rev. 2014, 114, 3963-
4038; c) J.-H. Jeoung, H. Dobbek, Science 2007, 318, 1461; d) H.
Dobbek, L. Gremer, R. Kiefersauer, R. Huber, O. Meyer, Proc. Nat. Acad.
Sci. 2002, 99, 15971; e) H. Dobbek, L. Gremer, O. Meyer, R. Huber, Proc.
Nat. Acad. Sci. 1999, 96, 8884.
[12] a) M. J. Romão, Dalton Trans. 2009, 4053-4068; b) L. B. Maia, J. J. G.
Moura, I. Moura, J. Biol. Inorg. Chem. 2015, 20, 287-309.
[13] a) T. Fogeron, T. K. Todorova, J.-P. Porcher, M. Gomez-Mingot, L.-M.
Chamoreau, C. Mellot-Draznieks, Y. Li, M. Fontecave, ACS Catal. 2018,
8, 2030-2038; b) T. Fogeron, P. Retailleau, M. Gomez-Mingot, Y. Li, M.
Fontecave, Organometallics 2019, 38, 1344-1350; c) S. Dey, M. E.
Ahmed, A. Dey, Inorg. Chem. 2018, 57, 5939-5947; d) M. E. Ahmed, A.
Rana, R. Saha, S. Dey, A. Dey, Inorg. Chem. 2020, 59, 5292-5302.
[14] Z. Han, L. Shen, W. W. Brennessel, P. L. Holland, R. Eisenberg, J. Am.
Chem. Soc. 2013, 135, 14659-14669.
[15] M. Kita, K. Yamanari, Y. Shimura, Bull. Chem. Soc. Jpn. 1989, 62, 3081-
3088.
Acknowledgements
[16] K. J. Franz, L. H. Doerrer, B. Spingler, S. J. Lippard, Inorg. Chem. 2001,
40, 3774-3780.
This work was supported by the ANR JCJC project NitroCOCa
(ANR-17-CE05-0021). The calculations were performed using the
HPC resources of GENCI (TGCC) through Grant 2019-810082.
We thank Lise-Marie Chamoreau for assistance with the X-ray
crystal structure determination of complex 3 and Yuan-Zi Xu for
SEM measurements.
[17] a) A. Chapovetsky, T. H. Do, R. Haiges, M. K. Takase, S. C. Marinescu,
J. Am. Chem. Soc. 2016, 138, 5765-5768; b) S. Fukuzumi, S. Mandal, K.
Mase, K. Ohkubo, H. Park, J. Benet-Buchholz, W. Nam, A. Llobet, J. Am.
Chem. Soc. 2012, 134, 9906-9909.
[18] S. G. Rosenfield, P. K. Mascharak, S. K. Arora, Inorg. Chim. Acta 1987,
129, 39-46.
[19] S. G. Rosenfield, H. P. Berends, L. Gelmini, D. W. Stephan, P. K.
Mascharak, Inorg. Chem. 1987, 26, 2792-2797.
Keywords: metal-thiolate complex, electrocatalysis,
homogeneous CO2 reduction, overpotential, density functional
theory, reaction mechanism
[20] M. Gennari, D. Brazzolotto, S. Yu, J. Pécaut, C. Philouze, M. Rouzières,
R. Clérac, M. Orio, C. Duboc, Chem. Eur. J. 2015, 21, 18770-18778.
[21] M. Okuno, M. Kita, K. Kashiwabara, J. Fujita, Chem. Lett. 1989, 18,
1643-1646.
[22] a) D. Brazzolotto, M. Gennari, N. Queyriaux, T. R. Simmons, J. Pécaut,
S. Demeshko, F. Meyer, M. Orio, V. Artero, C. Duboc, Nat. Chem. 2016,
8, 1054; b) S. Roy, T. L. Groy, A. K. Jones, Dalton Trans. 2013, 42, 3843-
3853.
References
[1]
[2]
R. Francke, B. Schille, M. Roemelt, Chem. Rev. 2018, 118, 4631-4701.
a) J. M. Smieja, M. D. Sampson, K. A. Grice, E. E. Benson, J. D.
Froehlich, C. P. Kubiak, Inorg. Chem. 2013, 52, 2484-2491; b) J.
Hawecker, J.-M. Lehn, R. Ziessel, J. Chem. Soc., Chem. Commun. 1984,
328-330.
[23] K. T. Ngo, M. McKinnon, B. Mahanti, R. Narayanan, D. C. Grills, M. Z.
Ertem, J. Rochford, J. Am. Chem. Soc. 2017, 139, 2604-2618.
[24] a) C. Costentin, S. Drouet, M. Robert, J.-M. Savéant, J. Am. Chem. Soc.
2012, 134, 11235-11242; b) C. Costentin, M. Robert, J.-M. Savéant,
Chem. Soc. Rev. 2013, 42, 2423-2436.
[3]
[4]
a) C. W. Machan, M. D. Sampson, C. P. Kubiak, J. Am. Chem. Soc. 2015,
137, 8564-8571; b) B. A. Johnson, S. Maji, H. Agarwala, T. A. White, E.
Mijangos, S. Ott, Angew. Chem. Int. Ed. 2016, 55, 1825-1829.
P. Kang, C. Cheng, Z. Chen, C. K. Schauer, T. J. Meyer, M. Brookhart,
J. Am. Chem. Soc. 2012, 134, 5500-5503.
[25] A. Taheri, E. J. Thompson, J. C. Fettinger, L. A. Berben, ACS Catal. 2015,
5, 7140-7151.
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
[26] L. Chen, Z. Guo, X.-G. Wei, C. Gallenkamp, J. Bonin, E. Anxolabéhère-
Mallart, K.-C. Lau, T.-C. Lau, M. Robert, J. Am. Chem. Soc. 2015, 137,
10918-10921.
[27] B. M. Ceballos, J. Y. Yang, Proc. Nat. Acad. Sci. 2018, 115, 12686.
[28] F.-W. Liu, J. Bi, Y. Sun, S. Luo, P. Kang, ChemSusChem 2018, 11, 1656-
1663.
[29] S. Fernández, F. Franco, C. Casadevall, V. Martin-Diaconescu, J. M.
Luis, J. Lloret-Fillol, J. Am. Chem. Soc. 2020, 142, 120-133.
[30] a) J. Schneider, H. Jia, J. T. Muckerman, E. Fujita, Chem. Soc. Rev.
2012, 41, 2036-2051; b) E. Fujita, C. Creutz, N. Sutin, D. J. Szalda, J.
Am. Chem. Soc. 1991, 113, 343-353; c) J. D. Froehlich, C. P. Kubiak, J.
Am. Chem. Soc. 2015, 137, 3565-3573.
[31] B. Mondal, A. Rana, P. Sen, A. Dey, J. Am. Chem. Soc. 2015, 137,
11214-11217.
[32] K. M. Waldie, A. L. Ostericher, M. H. Reineke, A. F. Sasayama, C. P.
Kubiak, ACS Catal. 2018, 8, 1313-1324.
[33] M. L. Clark, P. L. Cheung, M. Lessio, E. A. Carter, C. P. Kubiak, ACS
Catal. 2018, 8, 2021-2029.
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202006269
Angewandte Chemie International Edition
RESEARCH ARTICLE
8
This article is protected by copyright. All rights reserved.