Document [original]

Oxygen Reduction Hot Paper

Evidence of Sulfur Non-Innocence in [CoII(dithiacyclam)]2+-

Mediated Catalytic Oxygen Reduction Reactions

Beatrice Battistella, Linda Iffland-Mühlhaus, Maximillian Schütze, Beatrice Cula,

Uwe Kuhlmann, Holger Dau, Peter Hildebrandt, Thomas Lohmiller, Stefan Mebs,

Ulf-Peter Apfel,* and Kallol Ray*

Abstract: In many metalloenzymes, sulfur-containing

ligands participate in catalytic processes, mainly via the

involvement in electron transfer reactions. In a biomi-

metic approach, we now demonstrate the implication of

S-ligation in cobalt mediated oxygen reduction reactions

(ORR). A comparative study between the catalytic

ORR capabilities of the four-nitrogen bound [Co-

(cyclam)]2+(1; cyclam=1,5,8,11-tetraaza-cyclotetrade-

cane) and the S-containing analog [Co(S2N2-cyclam)]2+

(2; S2N2-cyclam=1,8-dithia-5,11-diaza-cyclotetradecane)

reveals improved catalytic performance once the chalc-

ogen is introduced in the Co coordination sphere.

Trapping and characterization of the intermediates

formed upon dioxygen activation at the CoII centers in 1

and 2point to the involvement of sulfur in the O2

reduction process as the key for the improved catalytic

ORR capabilities of 2.

Introduction

Metal-sulfur centers are widespread in biology and are

involved in promoting a number of critical biological

processes.[1,2] For example the active site of galactose oxidase

(GO)[3–5] is composed of a copper center attached to a post-

translationally generated Tyrosine-Cysteine (Tyr-Cys) ligand,

which facilitates the formation of the catalytically relevant

oxidized state at a low potential (Scheme 1). Similarly, in Ni-

dependent superoxide reductase (NiSOD) the Cys residues

are proposed to be crucial for the efficient degradation of

superoxide ion by stabilizing a square planar geometry in the

reduced form of NiSOD and tuning the NiII/NiIII redox

potential.[6,7] Furthermore, the presence of a Cys residue in

Isopenicillin N-synthase (IPNS)[8,9] makes this enzyme a

special class of αKG-dependent FeII/O2activating enzymes,

which exhibits oxidase rather than the usual oxygenase

activity. In addition, the active site of superoxide reductases

(SORs)[10–12] containing a redox-active FeII ion ligated by four

[*] M.Sc. B. Battistella, B.Sc. M. Schütze, Dr. B. Cula, Dr. T. Lohmiller,

Prof. Dr. K. Ray

Institut für Chemie, Humboldt-Universität zu Berlin

Brook-Taylor-Straße 2, 12489 Berlin (Germany)

E-mail: [email protected]

M.Sc. L. Iffland-Mühlhaus, Prof. Dr. U.-P. Apfel

Faculty of Chemistry and Biochemistry, Ruhr-Universität Bochum

Universitätsstraße 150, 44780 Bochum (Germany)

E-mail: [email protected]

Dr. U. Kuhlmann, Prof. Dr. P. Hildebrandt

Institut für Chemie, Fakultät II, Technische Universität Berlin

Straße des 17. Juni 135, 10623 Berlin (Germany)

Prof. Dr. H. Dau, Dr. S. Mebs

Institut für Physik, Freie Universität zu Berlin

Arnimallee 14, 14195 Berlin (Germany)

Dr. T. Lohmiller

EPR4Energy Joint Lab, Department Spins in Energy Conversion and

Quantum Information Science, Helmholtz Zentrum Berlin für

Materialien und Energie GmbH

Albert-Einstein-Str. 16, 12489 Berlin (Germany)

Prof. Dr. U.-P. Apfel

Department for Electrosynthesis, Fraunhofer Institute for Environ-

mental, Safety and Energy Technology UMSICHT

Osterfelder Str. 3, 46047 Oberhausen (Germany)

published by Wiley-VCH GmbH. This is an open access article under

the terms of the Creative Commons Attribution Non-Commercial

License, which permits use, distribution and reproduction in any

medium, provided the original work is properly cited and is not used

for commercial purposes.

Scheme 1. Active sites of various enzymes containing sulfur in the first

or second coordination spheres; the proposed mechanism of cysteine

dioxygenase (CDO) is highlighted in the inset.

Angewandte

Chemie

Research Articles www.angewandte.org

How to cite: Angew. Chem. Int. Ed. 2023, 62, e202214074

International Edition: doi.org/10.1002/anie.202214074

German Edition: doi.org/10.1002/ange.202214074

equatorial histidines and an axial Cys ligand, enables the

selective transformation of superoxides to hydrogen peroxide,

in contrast to Fe-containing superoxide dismutases,[13,14] which

lack the Cys ligand and disproportionate superoxide to afford

both hydrogen peroxide and dioxygen. In cysteine dioxyge-

nase (CDO) the Fe-coordinated thiolate of Cys is proposed

to be directly involved in dioxygen activation, by the

formation of a transient (FeIIO2SCys) moiety, followed by

further oxidation to yield the corresponding sulfonate

(Scheme 1).[15,16]

Understanding the reason for Nature’s choice of employ-

ing cysteinate and thioether residues to promote specific

biological functions in metalloenzymes requires a comprehen-

sive investigation of the influence of thiolate or thioether

ligands on the electronic, magnetic, and reactivity properties

of first-row transition metal ions. Based on detailed spectro-

scopic and theoretical studies the unique reactivities of the

metal-sulfur sites in oxidases and oxygenases has been

attributed to the stabilization of reactive intermediates

formed by the unfavorable one-electron reduction of O2, by

delocalization of the redox active orbitals onto the thioether

or cysteinate sulfur atoms.[4,13] Consistent with this explan-

ation the presence of sulfur ligation is shown to induce novel

dioxygen reactivity in bioinspired iron,[13,17–20] copper,[21–23]

manganese[24–26] and cobalt[26] complexes, which led to the

stabilization of well-defined metal-oxygen intermediates.

However, there is little evidence to support the redox non-

innocence of sulfur in these research works.

In previous studies, we demonstrated that the replace-

ment of the opposing nitrogen atoms in 1,4,8,11-tetraazacy-

clotetradecane (cyclam) with two sulfur atoms in 1,8-dithia-

4,11-diaza-cyclotetradecane (dithiacyclam) enabled the elec-

trochemical reduction of protons and CO2by the correspond-

ing NiII [27] and CoII [28] complexes at more positive potentials.

The lack of any isolable reactive intermediates in these

reactions, however, prevented us from obtaining detailed

mechanistic insights into the role of sulfur in enabling the

[MII(dithiacyclam)(CH3CN)2]2+(M=Co, Ni) mediated CO2

and H+reductions at low overpotentials. In the present study,

we show that the [CoII(dithiacyclam)(CH3CN)2]2+(2) com-

plex (Scheme 2) is also an excellent catalyst for the oxygen

reduction reaction (ORR) which is crucial to diverse proc-

esses ranging from biological respiration[29] and fuel cells[30] to

the selective oxidation of organic molecules.[31] Most impor-

tantly, 2performs the 2e/2H+reduction of dioxygen to

H2O2with high selectivity at an effective overpotential as low

as 66 mV, in contrast to the significantly higher value

(419 mV) determined for [CoII(cyclam)(CH3CN)2]2+(1;

Scheme 2) under analogous conditions. The difference in the

ORR capabilities of [CoII(dithiacyclam)(CH3CN)2]2+, contain-

ing a S2N2macrocyclic ligand, and [CoII(cyclam)(CH3CN)2]2+,

based on the popular N4-cyclam ligand, is demonstrated to

originate from distinct CoO2intermediates formed upon

dioxygen activation in the two cases. Although, diverse

complexes have been previously reported in the literature[30,32]

that exhibit high rates and/or high selectivity for O2reduc-

tions to H2O or H2O2, a particular challenge is the

identification of molecular catalysts that operate with low

effective overpotentials. The mechanistic insights obtained in

this study should, therefore, provide useful and broadly

applicable principles for future design of more efficient ORR

catalysts working at low overpotentials, by making use of the

redox non-innocence of sulfur containing ligands.

Results and Discussion

Combining the tetradentate cyclam and dithiacyclam ligands

with Co(ClO4)2in CH3CN yielded the previously

characterized[28] [CoII(cyclam)(CH3CN)2](ClO4)2(1) and

[CoII(dithiacyclam)(CH3CN)2](ClO4)2(2), respectively. The

molecular structures of 1and 2have been reported

previously, and display a six coordinate geometry with axially

bound CH3CN ligands in both cases (Scheme 2). The S2N2

donor atoms of dithiacyclam in 2and N4-donor atoms in 1

occupy the equatorial coordination sites. The X-band EPR

spectra in frozen CH3CN solutions at 13 K show an axial

signal (gꓕ=2.27, gjj=1.94) for 1and a nearly isotropic signal

at gaverage =2.19 for 2, confirming the S=1/2 spin state in both

cases (Figure S1a, b). Whereas the exchange of the axial

ligands has no significant effect on the coordination sphere or

the spin-state of the central CoII ion in 1, replacement of the

CH3CN ligands with methanol, acetone or trifluoroacetate

ligands leads to an altered geometry with axial sulfur donor

atoms and stabilization of the S=3/2 ground state in [CoII-

(dithiacyclam)(X)2]Y+(Figure S1c; X=MeOH, CH3COCH3,

Y=2+or X=TFA, Y=0, Tables S7, S8, 2-cis). Cyclo-

voltammetric studies on 1and 2in CH3CN reveal a quasi-

reversible oxidation wave corresponding to a CoIII/II couple

(Figure S2). A large anodic shift of the CoIII/II potential is

observed for 2in acetone (Figure S2), likely because of the

gradual replacement of the CH3CN ligands with acetone to

Scheme 2. Schematic structures showing the distinct reaction inter-

mediates formed in the reactions of 1and 2with O2.

Angewandte

Chemie

Research Articles

form 2-cis, as also evident from X-band EPR studies (Fig-

ure S1c).

The ORR capability of 1and 2was further investigated

by UV/Vis spectroscopy. Both 1and 2performed catalytic

ORR even in presence of a weak electron donor like

ferrocene (Fc) using trifluoroacetic acid (TFA) as a proton

source. Catalytic ORR was monitored by a rise in absorbance

at 619 nm, which is attributable to the formation of a

ferrocenium (Fc+) cation.[33] Iodometric titration[34] (Figure 1,

insets) of the final reaction mixture shows that H2O2is

formed with high selectivity in both cases (100% yield of

H2O2for 2and 62% for 1based on the maximum expected

yield of H2O2considering 2ereduction of O2), according to

the mechanism reported in Equation (1). Notably, in absence

of 1and 2, the formation of Fc+could not be detected by

UV/Vis spectroscopy and no H2O2was shown to be formed

under the employed experimental conditions (Figure S3). The

ORR capabilities of 1and 2were then compared under a

uniform set of conditions in acetone and CH3CN using

decamethylferrocene (Fc*) as an alternative chemical

reductant.[32,35,36] O2and iodometric titrations[34] and

1H NMR[37] (Figures S4–S6) studies confirmed the two-elec-

tron reduction of O2to H2O2with high selectivity in both

cases. Turnover frequencies (TOFs) for the conversion of O2

to H2O2were determined from the initial rates of decameth-

ylferrocenium ion (Fc*+) formation by UV/Visible spectro-

scopy under buffered conditions (TFA and NaTFA=sodium

trifluoroacetate, 10 mM each) (Figures S7, S8).[32,38,39] The

collected data were further analyzed to assess the effective

overpotential (ηeff) for the reactions (Scheme S1), which is

defined as the difference between the thermodynamic

potential for O2reduction to hydrogen peroxide (E(O2/

H2O2)) and the E1/2(CoIII/II) values under buffer conditions

[Eq. (2); Figures S9–S11]. Following the reported method-

ology for the determination of the overpotential under non-

aqueous conditions, EO2=H2O2could be determined in the

solvents of use by open-circuit potential (OCP) measure-

ments (EHþ=H2determination, Figures S12, S13, see Support-

ing Information for details).[32,38,39] The thermodynamic analy-

sis yielded an ηeff value of 66 mV for 2in acetone, which is

the lowest overpotential that has been measured (typical

range 150–550 mV) for the cobalt mediated catalytic reduc-

tion of O2to H2O2under buffer conditions.[32] Most

importantly, complex 1showed a significantly lower TOF and

higher overpotential (ηeff =419 mV in acetone, Table S1)

under similar conditions, thereby demonstrating the benefi-

cial influence of sulfur-based ligands on O2reduction

processes.

O2þ2Hþþ2e!H2O2

E�¼0:68 V vs:NHE (1)

heff ¼EðO2=H2O2Þ E1=2ðCoIII=IIÞ(2)

In order to elucidate the catalytic mechanism and to

explain the differences in the O2/H2O2ηeff for 1and 2, we

examined the reactions of 1and 2with O2(in the absence of

TFA/NaTFA and electron donors) to detect the formation of

any cobalt-dioxygen intermediates. Treatment of an acetone

solution of 1with O2at 70°C resulted in the formation of a

metastable (t1=2at 55°C=2.5 h) yellow species 1a with

absorption maxima λmax (ɛmax) at 330 nm (1900 M1cm1) and

455 nm (220 M1cm1) (Figure 2a). Such features are in line

with what has been reported for CoIII-superoxo moieties,

which often present intense absorption in the range 330–

380 nm and a less intense feature in the range 420–

480 nm.[40,49] The conversion of 1to 1a is associated with the

Figure 1. UV/Vis spectral changes in the two-electron reduction of O2

by Fc (3 mM) with 0.1 mM 1(a) and 2(b) in presence of 10 mM TFA

in O2-saturated (12.1 mM) CH3CN (total volume=2.0 mL) at 25°C;

inset: iodometric titration performed on an aliquot (0.1 mL) of the final

reaction mixture confirmed H2O2formation. The yield is based on the

maximum possible yield of H2O2(0.075 mM) according to Equation 1.

See also Figures S4, S5.

Figure 2. a) UV/Vis spectral changes associated with the reaction of 1

with O2at 70°C in acetone. The time trace of the development of the

455 nm band is shown in the inset. b) X-band EPR spectrum (red) of a

1/O2mixture in acetone (1 mM) at 13 K and the corresponding

simulated spectrum (black). Experimental details: perpendicular mode,

ca. 9.35 GHz, 1 mW power; simulation parameters: gjj=2.10, gꓕ=2.00,

Ajj=2.31 mT, Aꓕ=1.35 mT. c) rRaman spectra of a 8 mM 1/O2mixture

in acetone-d6(top) and acetone (bottom) at 70°C upon excitation

with 406 nm laser; the corresponding spectrum in CH2Cl2is shown as

an inset; the spectra in presence of 16O2are shown in red and 18O2in

blue. Asterisks indicate the solvent bands. d) Co K-edge XAS spectra of

1(black) and of a 1/O2mixture (red) in frozen acetone solutions at

20 K.

Angewandte

Chemie

Research Articles

appearance of a new axial EPR spectrum with gjj=2.10, and

gꓕ=2.00 (Figure 2b) and well resolved 59Co hyperfine features

with couplings Ajj=2.31 mT, and Aꓕ=1.35 mT. These EPR

parameters resemble those of CoIII(O2*) superoxo species

reported previously.[40–43] Quantification of the CoII and CoIII-

(O2*) EPR spectra in acetone reveals that only approx-

imately 25% of the total number of spins present in the initial

anaerobic CoII sample are converted to the CoIII(O2*)

intermediate (Figure S14). The low quantity of the CoIII(O2*)

signal may be rationalized by the simultaneous formation of a

peroxide-bridged CoIII dimer 1b (Scheme 2, Co:O2=2:1)

which is EPR-silent under the employed experimental

conditions (X-band, perpendicular mode). Such dimeric

species have been observed with other Co complexes bearing

tetradentate ligands[41,44] and have also been suggested as a

plausible intermediate during the complex 1mediated

electrocatalytic ORR process.[45–48] The resonance Raman

(rR) spectra of the reaction mixture of 1and O2in acetone

and acetone-d6(Figure 2c) display two isotopically sensitive

vibrational bands at 757 and 641 cm1, which in 18O2prepared

samples are downshifted to 727 and 610 cm1, respectively.

The 757 cm1band with an isotopic shift of 30 cm1(calcu-

lated shift 16/18Δcalc. =43 cm1) is assigned to the OO stretch-

ing mode of a peroxo ligand in 1b, and the 641 cm1band (16/

18Δexp. =31 cm1,16/18Δcalc. =29 cm1) is consistent with a CoO

stretching mode. Solvent bands mask the region between

1000–1150 cm1in acetone and acetone-d6, preventing the

observation of the OO stretching mode from the CoIII(O2*)

species 1a under these conditions. Changing the solvent to

dichloromethane allowed the isolation of an additional

18Oisotope sensitive vibrational band at 1099 cm1with an

18O-isotopic shift of 60 cm1(Figures 2c inset, S15), consistent

with its assignment to the OO stretching mode of the

superoxo ligand in 1a. Notably, no CoO vibration corre-

sponding to 1a could be observed, which is presumably

masked by the solvent signal (Figure 2c) or by the intense

CoO vibration band of 1b (at 636 cm1), which is formed in

much higher yield relative to 1a. The DFT calculated OO

and CoO vibrational modes of 1a and 1b are in reasonable

agreement with the experiment (Table S2). Furthermore,

topological analysis of the electron density (AIM) and non-

covalent interaction index (NCI) on the DFT computed

structure of 1a (Figure S16) show that two of the four NH

hydrogens of the cyclam macrocycle point towards the

superoxo ligand and are located at only 2.1 Å distance from

the distal oxygen atom of the end-on bound (O2)*moiety.

The negative charge on (O2)*promotes the hydrogen

bonding interaction and electron density transfer from the

NH groups to (O2)*, stabilizing the superoxo species by

conferring it a higher peroxo-character. This results in a OO

stretching frequency (calculated: 1078 cm1; experimental:

1099 cm1), which is significantly lower compared to the

previously reported CoIII(O2*) species (1153–1123 cm1).[49,50]

In order to probe the oxidation state of cobalt in 1 a/1b, X-

ray absorption spectroscopic (XAS) studies at the Co K-edge

were performed. Figure 2d depicts a comparison of the

normalized Co K-edge XAS spectra of the 1/O2reaction

mixture with the CoII precursor complex 1. A blue shift of ca.

2.5 eV in edge energy from 1(7718.8 eV) to 1 a/1b

(7721.3 eV) and comparison to cobalt standards of known

oxidation states agree with the higher oxidation state of

cobalt in 1a/1b (CoIII) relative to 1(CoII) (Figure S17). The

presence of a CoO2unit in 1a/1b was confirmed by EXAFS

analysis, which yielded the best-fit plot (Figure S17, Table S3)

with an O/N scatterer at 1.86 Å (assigned to the CoO

scatterer) and a further shell of five O/N scatterers at 1.96 Å

(corresponding to the N-donors of cyclam and CH3CN). Fine

structural details are visible in the EXAFS wave, which were

calculated for the DFT models of 1a/1b and compared to the

experimental spectra. The best fit corresponds to the

presence of a mixture of 1a/1b in the O2saturated solution of

1(Figure S18, Table S4), consistent with EPR studies.

Complex 2reacts with O2in CH3CN or acetone at 30°C

to form a deep red species 2a/2b, whose absorption bands at

λmax =495 and 695 nm (ɛmax =1550 and 350 M1cm1, respec-

tively; t1=

2at 10°C=250 s; Figure 3a) are clearly distinct from

the absorption spectrum of 1a/1b and similar to what has

been reported for several CoIII-peroxo moieties, which

generally exhibit absorption features in the range from

500 nm to 700 nm.[50] Furthermore, in contrast to the irrever-

sible conversion of 1to 1a/1b, the formation of 2a/2b is

reversible. Flushing with argon for 30 seconds led to the

decay of 2a/2b, which could then be regenerated upon

Figure 3. a) UV/Vis spectral changes associated with the reaction of 2

with dioxygen at 30°C in CH3CN. The time trace of the development

of the 495 nm band is shown in the inset. b) X-band EPR spectrum

(red) of 2/O2mixture in CH3CN (1 mM) at 13 K and the corresponding

simulated spectrum (black). Experimental details: perpendicular mode,

9.35 GHz, 1 mW power; simulation parameters: geff

x=5.59, geff

y=3.65,

geff

z=1.94). c) Co K-edge XAS spectra of 2(black) and of a 2/O2

mixture (red) in frozen CH3CN solutions at 20 K. d) rRaman spectra of

a 8 mM 2/O2mixture in CD3CN at 30°C upon excitation with

514 nm; the red, blue and black traces correspond to the data recorded

in presence of 16O2,18O2and a statistical mixture of 16O2:16/18O2:18O2

(1:2:1 ratio), respectively. The DFT calculated vibrational modes for 2a

(grey) and 2b (black) in presence of a statistical mixture of 16O2:16/

18O2:18O2(1:2:1 ratio) are shown by the bars. Asterisks indicate the

solvent bands.

Angewandte

Chemie

Research Articles

addition of O2(Figure S19). The EPR spectrum of the

reaction product of 2with O2exhibits a rhombic signal with

geff

x=5.59, geff

y=3.65 and geff

z=1.94, which corresponds to

50% of the total number of cobalt spins (Figures 3b, S20). Its

significant anisotropy evidences a S>1/2 ground state, which

is attributed to the mononuclear CoO2adduct 2a

(Scheme 2), the ground total spin state of which is identified

as 3/2 rather than 5/2 (see below). The remaining 50% of

cobalt is attributed to the EPR-silent peroxo-bridged

dicobalt(III) species (2b), which is characterized by X-ray

crystallography. Single crystals of the metastable intermediate

(Figure 4a) were obtained by Et2O diffusion into a CH3CN

solution of the triflate salt of 2cooled at 25°C and exposed

to O2. As evident from the R-factor (8.1%; Table S5) and

estimated standard deviations, the structure of 2b is of high

quality. The ORTEP diagram in Figure 4a shows a

[(CH3CN)CoIII(dithiacyclam)]2(trans-μ-1,2-O2)(OTf)4(2b)

moiety containing an O2ligand bridging two cobalt ions in an

end-on trans configuration. Each Co ion is ligated by two

sulfur and two nitrogen atoms in the equatorial plane, and an

O2-derived oxygen and CH3CN in the axial sites. Notably, the

OO bond length in 2b (1.508(6) Å, Table S6) is significantly

longer than the typical values (1.43–1.48 Å) determined by X-

ray crystallography for metal-peroxo complexes,[24,44,51] con-

sistent with strong reductive activation of the OO bond.

Furthermore, XRD analysis displays hydrogen bonding

interactions between one of the NH moieties in each

macrocyclic ligand and the bridging peroxo moiety

(NHO1=2.477 Å; NHO2=2.101 Å, Figure 4a), which

presumably contributes to the elongation of the OO bond in

2b. Dioxygen reduction by 2is accompanied by a 1.3 eV blue

shift of the Co K-edge in the XANES spectrum (Figures 3c,

S21; from 7718.2 eV in 2to 7719.5 eV upon oxygenation),

which is significantly smaller than the 2.5 eV blue shift

observed during the oxygenation of 1. Nevertheless, the OO

stretching mode in 2b (vOO(16/18Δexp): 764 (39) cm1) as

obtained from rRaman measurements (Figure 3d) is compa-

rable to that observed for 1 b (vOO(16/18Δexp): 757 (30) cm1).

Thus, although the reductive activation of the OO bond is

comparable in 1b and 2b, the cobalt centers are more

reduced in 2b. This may point to a significantly higher charge

donation to O2from the S2N2donor atoms in 2b relative to

that of the N4centers in 1b.

The rRaman spectrum of the 2/O2mixture shows an

additional 18O-isotope sensitive band at 795 cm1, which shifts

to 764 cm1upon 18O labelling. This is assigned to the OO

stretch of the CoOOS motif of a mononuclear S=3/2

cyclic peroxythioether [(CH3CN)CoIII(dithiacyclam-O2*)]2+

(2a) intermediate (Scheme 2) formed by the plausible attack

of the initially formed CoIII-superoxido species (2a’) to one of

the thioether sulfur atoms of the dithiacyclam ligand. A

rRaman investigation was also done in presence of mixed

isotope O2(“16,18O2”) containing a statistical mixture of 16O2:16/

18O2:18O2(1:2:1 ratio), which further confirms the presence

of both 2a/2b in an oxygenated solution of 2. Since the

795 cm1mode corresponds to an OO stretching of an

asymmetric CoOOS unit in 2 a, four modes are expected

corresponding to the Co16OO16S, Co18OO16S,

Co16OO18S and Co18OO18S moieties in 1:1:1:1 ratio

(grey bars in Figure 3d). In contrast, for the symmetric

CoOOCo moiety in 2b, we should expect three peaks at

764, �746 and 725 cm1in 1:2:1 ratio (black bars in

Figure 3d). A superposition of all the above signals presum-

ably gives rise to the complex pattern with an intense signal

at �764 cm1and shoulders at 725, 745 and 794 cm1in the

experimental spectrum (Figure 3d), which is reproduced by

the optimized DFT structures (see below, Figure S22,

Table S2). Thus, in contrast to the CoII(cyclam) motif, which

acts as a one electron donor, the CoII(dithiacyclam) core

involving a S2N2macrocyclic ligand acts as a two-electron

donor. The non-innocence of the dithiacyclam ligand is

associated with the formation of a SO bond to yield the

CoOOS moiety in 2a. Furthermore, the oxidized sulfur

atom stays away from the coordination sphere of cobalt, as

confirmed by EXAFS analysis (Figure S21). While the

EXAFS spectrum of 2can be best fitted with two CoS

scatterers at 2.26 Å, the corresponding spectrum for an

aerobic solution of 2can be best modelled by a partial loss of

S-shell contribution (1CoS=2.27 Å, 0.5CoS=2.44 Å),

consistent with the presence of equal amount of 2a and 2b

(Table S3), in excellent agreement with the EPR results, and

the presence of only one CoS scatterer in 2a. Geometry

optimization performed by DFT calculation on 2a (Fig-

ure 4b) in the S=3/2 state (involving an S=1CoIII center that

is ferromagnetically coupled to S=1/2 dithiacyclam-O2*) can

nicely reproduce the EXAFS determined metrical parameters

(Table S4) and fine structures (Figure S23).[52] The asymmet-

ric OO stretching of the CoOOS core is calculated at

789 cm1, in excellent agreement with the experiment

(795 cm1). However, the 16/18Δcalc shift is calculated to be

42 cm1, in contrast to the experimentally observed 16/18Δexp

shift of only 30 cm1. This discrepancy can be attributed to

the coupling of the OO stretching coordinate with other

coordinates, which may be underestimated in the calculations,

or because of intermolecular interactions, which are not

considered in DFT.

Figure 4. a) XRD determined molecular structure of [(CH3CN)CoIII-

(dithiacyclam)]2(trans-μ-1,2-O2)(OTf)42b.[53] See Table S6 for bond

lengths in 2b. H-bonding interactions are shown as dotted lines.

b) DFT optimized molecular structure of 2a, showing the loss of the

Co- S(oxidized) atom ligation. The CoS(oxidized) distance is shown

as a dotted line. Color code: C grey, N blue, H white, S yellow, O red,

Co purple.

Angewandte

Chemie

Research Articles

Conclusion

The catalytically relevant, oxidized state of the active site

[CuIIY*

C] of galactose oxidase (GO) is composed of

antiferromagnetically coupled CuII and a post-translationally

generated Tyr-Cys radical cofactor [Y*

C].[4] The thioether

bond of the Tyr-Cys cross-link has been shown experimen-

tally to affect the stability, the reduction potential, and the

catalytic efficiency of the GO active site. Similarly, in various

non-heme iron enzymes the charge donation from the

thiolate ligand is proposed to render the formation of the

FeIII-superoxide complex energetically favorable, driving the

reaction at a single Fe center rather than the energetically

more demanding bimolecular pathway.[14] Although the non-

innocence of thioether and thiolate sulfur atoms is implicated

in these processes, direct evidence is lacking. In a bioinspired

approach we now demonstrate that the presence of thioether

ligation significantly affects the ORR capability of a cobalt

complex; complex 2performs the catalytic reduction of O2to

H2O2at an effective overpotential as low as 66 mV. The high

efficiency of 2can be attributed to the formation of a cyclic

peroxythioether intermediate 2a by the attack of the initially

formed CoIII-superoxide species at one of the thioether sulfur

atoms of the dithiacyclam ligand. As suggested before in

enzymatic systems,[14] this presumably opens up a low-energy

mononuclear pathway for catalytic ORR processes. The

formation of the novel CoOOS moiety in 2a is concluded

based on combined EPR, EXAFS (loss of 1 CoS ligation)

and rRaman (OO vibration at 795 cm1) studies, and it

represents a bioinspired model of the cyclic peroxythiolate

intermediate proposed in cysteine dioxygenase (Scheme 1

inset).[15,16] In the absence of sulfur ligation, the S=1/2 CoIII-

superoxide species 1a can only be reduced by an energetically

more demanding dinuclear mechanism, which may contribute

to the large overpotential (419 mV) required for catalytic

ORR mediated by 1. Notably, the presence of two thioether

ligands does not completely shut down the dinuclear mecha-

nism in 2, as evident from the isolation of the crystals of 2b.

Complex 2b features a highly activated dioxygen center with

a large OO bond length of 1.51 Å, which further signifies

the strong charge donation from the dithiacyclam ligand.

Whether the overpotential for O2reduction can be further

lowered by steric inhibition of the dinuclear mechanism in 2

is now an inherent question, which will be investigated in

future studies.

Acknowledgements

We thank the Deutsche Forschungsgemeinschaft (DFG; Ger-

man Research Foundation) for financial support under

Germany’s Excellence Strategy—EXC-2033 390677874—

“RESOLV” RA/2409/8-1 & AP242/5-1) and EXC-2008—

390540038—“UniSysCat”, as well as for support to T.L.

(Project No. LO 2898/1-1). This work was further supported

by the Fraunhofer Internal Programs and Einstein Center of

Catalysis. Single isotope labelled gas mixture was prepared in

collaboration with the group of Prof. Dr. Christian Limberg

at the Humboldt-University of Berlin, for which we acknowl-

edge Dr. Christin Herwig. We also acknowledge the

Helmholtz Zentrum Berlin (HZB) for providing experimen-

tal infrastructure and allocating beamtime at beamline KMC-

3 of the BESSY synchrotron. Furthermore, we thank Dr.

Daniel Siegmund for the XRD molecular structure analysis

performed at the Ruhr-Universität Bochum. Open Access

funding enabled and organized by Projekt DEAL.

Conflict of Interest

The authors declare no conflict of interest.

Data Availability Statement

The data that support the findings of this study are available

in the Supporting Information of this article.

Keywords: Cobalt ·Macrocyclic Ligands ·OO Activation ·

Reactive Intermediates ·Spectroscopy

[1] A. Krężel, W. Maret, Chem. Rev. 2021,121, 14594–14648.

[2] J. Liu, S. Chakraborty, P. Hosseinzadeh, Y. Yu, S. Tian, I.

Petrik, A. Bhagi, Y. Lu, Chem. Rev. 2014,114, 4366–4469.

[3] J. W. Whittaker, Chem. Rev. 2003,103, 2347–2363.

[4] D. Rokhsana, A. E. Howells, D. M. Dooley, R. K. Szilagyi,

Inorg. Chem. 2012,51, 3513–3524.

[5] S. Itoh, M. Taki, S. Fukuzumi, Coord. Chem. Rev. 2000,198, 3–

20.

[6] a) V. Pelmenschikov, P. E. M. Siegbahn, J. Am. Chem. Soc.

2006,128, 7466–7475; b) N. Lihi, D. Kelemen, N. V. May, I.

Fábián, Inorg. Chem. 2020,59, 4772–4780.

[7] J. Shearer, Acc. Chem. Res. 2014,47, 2332–2341.

[8] J. E. Baldwin, M. Bradley, Chem. Rev. 1990,90, 1079–1088.

[9] P. L. Roach, I. J. Clifton, C. M. H. Hensgens, N. Shibata, C. J.

Schofield, J. Hajdu, J. E. Baldwin, Nature 1997,387, 827–830.

[10] C. Mathé, C. O. Weill, T. A. Mattioli, C. Berthomieu, C.

Houée-Levin, E. Tremey, V. Nivière, J. Biol. Chem. 2007,282,

22207–22216.

[11] E. Tremey, F. Bonnot, Y. Moreau, C. Berthomieu, A. Desbois,

V. Favaudon, G. Blondin, C. Houée-Levin, V. Nivière, JBIC J.

Biol. Inorg. Chem. 2013,18, 815–830.

[12] A. Desbois, J. Valton, Y. Moreau, S. Torelli, V. Nivière, Phys.

Chem. Chem. Phys. 2021,23, 4636–4645.

[13] L. M. Brines, J. A. Kovacs, Eur. J. Inorg. Chem. 2007, 29–38.

[14] a) J. A. Kovacs, L. M. Brines, Acc. Chem. Res. 2007,40, 501–

509; b) C. D. Brown, M. L. Neidig, M. B. Neibergall, J. D.

Lipscomb, E. I. Solomon, J. Am. Chem. Soc. 2007,129, 7427–

7438.

[15] S. Aluri, S. P. de Visser, J. Am. Chem. Soc. 2007,129, 14846–

14847.

[16] E. P. Tchesnokov, A. S. Faponle, C. G. Davies, M. G. Quesne,

R. Turner, M. Fellner, R. J. Souness, S. M. Wilbanks, S. P.

de Visser, G. N. L. Jameson, Chem. Commun. 2016,52, 8814–

8817.

[17] J. Shearer, R. C. Scarrow, J. A. Kovacs, J. Am. Chem. Soc.

2002,124, 11709–11717.

[18] J. A. Kovacs, Chem. Rev. 2004,104, 825–848.

[19] T. Kitagawa, A. Dey, P. Lugo-Mas, J. B. Benedict, W.

Kaminsky, E. Solomon, J. A. Kovacs, J. Am. Chem. Soc. 2006,

128, 14448–14449.

Angewandte

Chemie

Research Articles

[20] L. Wang, M. Gennari, F. G. Cantú Reinhard, J. Gutiérrez, A.

Morozan, C. Philouze, S. Demeshko, V. Artero, F. Meyer, S. P.

de Visser, et al., J. Am. Chem. Soc. 2019,141, 8244–8253.

[21] T. Ohta, T. Tachiyama, K. Yoshizawa, T. Yamabe, T. Uchida,

T. Kitagawa, Inorg. Chem. 2000,39, 4358–4369.

[22] M. Gennari, C. Duboc, Acc. Chem. Res. 2020,53, 2753–2761.

[23] G. Moula, J. Bag, M. Bose, S. Barman, K. Pal, Inorg. Chem.

2022,61, 6660–6671.

[24] M. K. Coggins, X. Sun, Y. Kwak, E. I. Solomon, E. Rybak-

Akimova, J. A. Kovacs, J. Am. Chem. Soc. 2013,135, 5631–

5640.

[25] M. Gennari, D. Brazzolotto, J. Pécaut, M. V. Cherrier, C. J.

Pollock, S. DeBeer, M. Retegan, D. A. Pantazis, F. Neese, M.

Rouzières, et al., J. Am. Chem. Soc. 2015,137, 8644–8653.

[26] a) J. A. Kovacs, Acc. Chem. Res. 2015,48, 2744–2753; b) P.

Kumar, L. Devkota, M. C. Casey, A. A. Fischer, S. V. Linde-

man, A. T. Fiedler, Inorg. Chem. 2022,61, 16664–16677.

[27] P. Gerschel, B. Battistella, D. Siegmund, K. Ray, U.-P. Apfel,

Organometallics 2020,39, 1497–1510.

[28] L. Iffland, D. Siegmund, U.-P. Apfel, Z. Anorg. Allg. Chem.

2020,646, 746–753.

[29] D. Hamdane, H. Zhang, P. Hollenberg, Photosynth. Res. 2008,

98, 657.

[30] a) C. W. Anson, S. S. Stahl, Chem. Rev. 2020,120, 3749–3786;

b) J. Rosenthal, D. G. Nocera, Acc. Chem. Res. 2007,40, 543–

553; c) W. Zhang, W. Lai, R. Cao, Chem. Rev. 2017,117, 3717–

3797; d) M. L. Pegis, C. F. Wise, D. J. Martin, J. M. Mayer,

Chem. Rev. 2018,118, 2340–2391; e) D. Das, Y.-M. Lee, K.

Ohkubo, W. Nam, K. D. Karlin, S. Fukuzumi, J. Am. Chem.

Soc. 2013,135, 2825; f) S. Fukuzumi, S. Mandal, K. Mase, K.

Ohkubo, H. Park, J. Benet-Buchholz, W. Nam, A. Llobet, J.

Am. Chem. Soc. 2012,134, 9906; g) D. Das, Y.-M. Lee, K.

Ohkubo, W. Nam, K. D. Karlin, S. Fukuzumi, J. Am. Chem.

Soc. 2013,135, 4018; h) C. Costentin, H. Dridi, J.-M. Saveant,

J. Am. Chem. Soc. 2015,137, 13535; i) F. Heims, K. Ray, M.

Schwalbe, W. Nam, Curr. Opin. Chem. Biol. 2015,25, 159–171;

j) S. Bhunia, A. Ghatak, A. Dey, Chem. Rev. 2022,122, 12370–

12426.

[31] a) M. Guo, T. Corona, K. Ray, W. Nam, ACS Cent. Sci. 2019,

5, 13–28; b) X.-P. Zhang, A. Chandra, Y.-M. Lee, R. Cao, K.

Ray, W. Nam, Chem. Soc. Rev. 2021,50, 4804–4811; c) K. Ray,

F. F. Pfaff, B. Wang, W. Nam, J. Am. Chem. Soc. 2014,136,

13942–13958.

[32] Y.-H. Wang, B. Mondal, S. S. Stahl, ACS Catal. 2020,10,

12031–12039.

[33] A. Paul, R. Borrelli, H. Bouyanfif, S. Gottis, F. Sauvage, ACS

Omega 2019,4, 14780–14789.

[34] I. Monte-Pérez, S. Kundu, A. Chandra, K. E. Craigo, P.

Chernev, U. Kuhlmann, H. Dau, P. Hildebrandt, C. Greco, C.

Van Stappen, et al., J. Am. Chem. Soc. 2017,139, 15033–15042.

[35] J. P. Hurvois, C. Moinet, J. Organomet. Chem. 2005,690, 1829–

1839.

[36] The ferrocenium (Fc+) cation is known to be unstable in acidic

environment in presence of dioxygen; therefore Fc* was used

as electron donor due to the higher stability of the decameth-

ylferrocenium (Fc*+) cation.

[37] N. A. Stephenson, A. T. Bell, Anal. Bioanal. Chem. 2005,381,

1289–1293.

[38] Y.-H. Wang, M. L. Pegis, J. M. Mayer, S. S. Stahl, J. Am.

Chem. Soc. 2017,139, 16458–16461.

[39] Y.-H. Wang, Z. K. Goldsmith, P. E. Schneider, C. W. Anson,

J. B. Gerken, S. Ghosh, S. Hammes-Schiffer, S. S. Stahl, J. Am.

Chem. Soc. 2018,140, 10890–10899.

[40] a) J. B. Gordon, A. C. Vilbert, M. A. Siegler, K. M. Lancaster,

P. Moënne-Loccoz, D. P. Goldberg, J. Am. Chem. Soc. 2019,

141, 3641–3653; b) A. A. Fischer, S. V. Lindeman, A. T.

Fiedler, Dalton Trans. 2017,46, 13229–13241.

[41] T. Corona, S. K. Padamati, F. Acuña-Parés, C. Duboc, W. R.

Browne, A. Company, Chem. Commun. 2017,53, 11782–11785.

[42] B. Lv, X. Li, K. Guo, J. Ma, Y. Wang, H. Lei, F. Wang, X. Jin,

Q. Zhang, W. Zhang, et al., Angew. Chem. Int. Ed. 2021,60,

12742–12746; Angew. Chem. 2021,133, 12852–12856.

[43] K. Mittra, B. Mondal, A. Mahammed, Z. Gross, A. Dey,

Chem. Commun. 2017,53, 877–880.

[44] M. A. Dedushko, D. Schweitzer, M. N. Blakely, R. D. Swartz,

W. Kaminsky, J. A. Kovacs, J. Biol. Inorg. Chem. 2019,24,

919–926.

[45] C.-L. Wong, J. A. Switzer, K. P. Balakrishnan, J. F. Endicott, J.

Am. Chem. Soc. 1980,102, 5511–5518.

[46] C.-L. Wong, J. F. Endicott, Inorg. Chem. 1981,20, 2233–2239.

[47] K. Kumar, J. F. Endicott, Inorg. Chem. 1984,23, 2447–2452.

[48] a) A. Bakac, J. H. Espenson, J. Am. Chem. Soc. 1990,112,

2273–2278; b) H. Kim, P. J. Rogler, S. K. Sharma, A. W.

Schaefer, E. I. Solomon, K. D. Karlin, J. Am. Chem. Soc. 2020,

142, 3104–3116; c) J. J. D. Sacramento, T. Albert, M. Siegler, P.

Moënne-Loccoz, D. P. Goldberg, Angew. Chem. Int. Ed. 2022,

61, e202111492; Angew. Chem. 2022,134, e202111492.

[49] C.-C. Wang, H.-C. Chang, Y.-C. Lai, H. Fang, C.-C. Li, H.-K.

Hsu, Z.-Y. Li, T.-S. Lin, T.-S. Kuo, F. Neese, et al., J. Am.

Chem. Soc. 2016,138, 14186–14189.

[50] a) A. R. Corcos, O. Villanueva, R. C. Walroth, S. K. Sharma, J.

Bacsa, K. M. Lancaster, C. E. MacBeth, J. F. Berry, J. Am.

Chem. Soc. 2016,138, 1796–1799; b) J. Cho, R. Sarangi, H. Y.

Kang, J. Y. Lee, M. Kubo, T. Ogura, E. I. Solomon, W. Nam, J.

Am. Chem. Soc. 2010,132, 47, 16977–16986; c) Y. Jo, J.

Annaraj, M. S. Seo, Y.-M. Lee, S. Y. Kim, J. Cho, W. Nam, J.

Inorg. Biochem. 2008,102, 2155–2159.

[51] a) A. Brinkmeier, R. A. Schulz, M. Buchhorn, C.-J. Spyra, S.

Dechert, S. Demeshko, V. Krewald, F. Meyer, J. Am. Chem.

Soc. 2021,143, 10361–10366.

[52] Calculations on 2a were performed for the different possible

spin states (S=5/2, 3/2, and 1/2) and orientations (cis and

trans). The calculated energies are summarized in Table S4

(bottom) and the optimized structure in the EPR determined

S=3/2 state for 2a in trans configuration is shown in Fig-

ure S24f. The S=5/2 state lies very high in energy and can be

excluded. Although, the S=3/2 and S=1/2 states of 2a are

isoenergetic within the error of the DFT calculated energies

(typically 5 kcalmol1), its EPR spectrum clearly identifies a

high-spin S=3/2 state. DFT calculations also cannot distinguish

between the two possible isomers (cis and trans) based on

calculated energies; however, the fine structure pattern in the

EXAFS waves and the corresponding Fourier Transform

clearly favor the predominant formation of the trans isomer

(Figure S23). Notably, the trans configuration is also visible in

the XRD structures of 2(ref. [28]) and 2b. The alternate

assignment of S=3/2 2a containing a high spin (S=2) CoIII

center antiferromagnetically coupled to S=1/2 dithiacyclam-

O2*unit is excluded considering the rarity of octahedral high-

spin CoIII centers (a) H. C. Clark, B. Cox, A. G. Sharpe, J.

Chem. Soc. 1957, 4132–4133; b) P. Guetlich, B. R. McGarvey,

W. Klaeui, Inorg. Chem. 1980,19, 3704–3706.) and based on

the observed low-spin S=1/2 ground state for the CoII centre

in 2.

[53] Deposition Number 2193635 contains the supplementary

crystallographic data for this paper. These data are provided

free of charge by the joint Cambridge Crystallographic Data

Centre and Fachinformationszentrum Karlsruhe Access Struc-

tures service.

Manuscript received: September 23, 2022

Accepted manuscript online: November 15, 2022

Version of record online: December 29, 2022

Angewandte

Chemie

Research Articles