MetalminusLigand Cooperativity via Exchange Coupling Promotes Iron-Catalyzed Electrochemical CO2 Reduction at Low OverpotentialsJeffrey S Derrick Matthias Loipersberger Ruchira Chatterjee Diana A Iovan Peter T SmithKhetpakorn Chakarawet Junko Yano Jeffrey R Long Martin Head-Gordon and Christopher J Chang

Cite This J Am Chem Soc 2020 142 20489minus20501 Read Online

ACCESS Metrics amp More Article Recommendations sı Supporting Information

ABSTRACT Biological and heterogeneous catalysts for theelectrochemical CO2 reduction reaction (CO2RR) often exhibit ahigh degree of electronic delocalization that serves to minimizeoverpotential and maximize selectivity over the hydrogen evolutionreaction (HER) Here we report a molecular iron(II) system thatcaptures this design concept in a homogeneous setting through theuse of a redox non-innocent terpyridine-based pentapyridine ligand(tpyPY2Me) As a result of strong metalminusligand exchange coupling between the Fe(II) center and ligand [Fe(tpyPY2Me)]2+

exhibits redox behavior at potentials 640 mV more positive than the isostructural [Zn(tpyPY2Me)]2+ analog containing the redox-inactive Zn(II) ion This shift in redox potential is attributed to the requirement for both an open-shell metal ion and a redox non-innocent ligand The metalminusligand cooperativity in [Fe(tpyPY2Me)]2+ drives the electrochemical reduction of CO2 to CO at lowoverpotentials with high selectivity for CO2RR (gt90) and turnover frequencies of 100 000 sminus1 with no degradation over 20 h Thedecrease in the thermodynamic barrier engendered by this coupling also enables homogeneous CO2 reduction catalysis in waterwithout compromising selectivity or rates Synthesis of the two-electron reduction product [Fe(tpyPY2Me)]0 and characterizationby X-ray crystallography Mossbauer spectroscopy X-ray absorption spectroscopy (XAS) variable temperature NMR and densityfunctional theory (DFT) calculations support assignment of an open-shell singlet electronic structure that maintains a formal Fe(II)oxidation state with a doubly reduced ligand system This work provides a starting point for the design of systems that exploitmetalminusligand cooperativity for electrocatalysis where the electrochemical potential of redox non-innocent ligands can be tunedthrough secondary metal-dependent interactions

INTRODUCTION

The electrochemical carbon dioxide reduction reaction(CO2RR) provides opportunities to synthesize value-addedproducts from this greenhouse gas in a sustainable manner12

Efficient catalysts for this reaction are required to selectivelydrive CO2 reduction over the thermodynamic and kineticallycompetitive hydrogen evolution reaction (HER)3 Biological4

heterogeneous5 and molecular3 catalysts have been studied forCO2 reduction and molecular systems in particular offerpotential advantages such as their small size relative to enzymesand the fact that they can be tuned with a level of atomicprecision that is inaccessible with heterogeneous materials6

However despite advances in the development of molecularcatalysts with highly specialized ligand scaffolds includingcyclams7minus10 porphyrins and phthalocyanines11minus15 and tricar-bonyl-bipyridines1617 it remains a challenge to designmolecules that exhibit low overpotentials high turnovernumbers and compatibility with aqueous electrolytes to avoidoff-pathway HER processes In this regard one key advantage ofbiological and heterogeneous catalysts is their ability tominimize overpotential and maximize selectivity for CO2reduction via electronic delocalization either through elec-tron-tunneling pathways1819 or a high local density of

states20minus22 respectively Such systems can thereby achieveorganization and separation of reducing equivalents in a way thatis challenging to design in molecular systemsPrevious work from our laboratories has explored the

chemistry of the PY5Me2 ligand (26-bis(11-bis(2-pyridyl)-ethyl)pyridine) and its molybdenum23minus25 and cobalt2526

derivatives as catalysts for the HER We reasoned that byincorporating a redox-active pendant such as terpyridine27minus29

within this ligand scaffold it would be possible to increasemetalminusligand orbital mixing and delocalization of electrondensity away from the metal center to favor CO2 reduction(Figure 1) Precedent for the use of redox non-innocent ligandsin molecular electrochemical CO2RR has shown the viability ofthis strategy161727minus39 along with classic systems bearing strongmetalminusligand cooperativity such as metal-dithiolenes forHER40minus43 Indeed the suppression of metal-centered reduc-

Received October 7 2020Published November 18 2020

ArticlepubsacsorgJACS

copy 2020 American Chemical Society20489

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

Dow

nloa

ded

via

UN

IV O

F C

AL

IFO

RN

IA B

ER

KE

LE

Y o

n D

ecem

ber

3 2

020

at 0

352

03

(UT

C)

See

http

sp

ubs

acs

org

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s

tions can limit the formation of off-pathway metal hydrideintermediates necessary for hydrogen evolution3544 and therebyfavor CO2 reduction catalysis As part of a larger program inelectrocatalysis via bioinorganic mimicry6242545minus48 oursynthetic approach was inspired by mononuclear iron hydro-genase enzymes which catalyze the reduction of methenyl-H4MPT+ to methylene-H4MPT in methanogenic archaea49 Inthese systems heterolysis of H2 and hydride shuttling is enabledby a single redox-innocent iron(II) site that synergisticallyinteracts with a redox non-innocent guanyl pyridinol cofactor inthe primary coordination sphere to provide reducing equivalents(Figure 1)Here we report an iron polypyridine complex [Fe-

(tpyPY2Me)(CH3CN)]2+ ([Fe1]2+) (tpyPY2Me = 6-(11-

di(pyridin-2-yl)ethyl)-22prime6prime2Prime-terpyridine) which features aredox-active terpyridine fragment and demonstrates thisprinciple of electronic delocalization In particular metalminusligand cooperativity promotes strong exchange couplingbetween the redox-active tpyPY2Me ligand and iron(II) centerwhere the [Fe(tpyPY2Me)]2+ complex is able to accept twoelectrons at reduction potentials that are 640 mV more positivethan the isostructural [Zn(tpyPY2Me)]2+ analog containing theredox-inactive Zn(II) ion showing that a redox non-innocentligand alone is not sufficient to promote multielectron redoxchemistry with low activation barriers Upon two-electronreduction of [Fe(tpyPY2Me)]2+ strong antiferromagneticcoupling facilitates electrocatalytic reduction of CO2 to CO atlow overpotentials Under these conditions [Fe1]2+ exhibitshigh selectivity for CO production (greater than 90) at highturnover frequencies Synthetic reduction of the [Fe-(tpyPY2Me)]2+ complex enables access to the hypothesizedcatalytically active species [Fe(tpyPY2Me)]0 which has been

characterized by X-ray crystallography Mossbauer spectrosco-py X-ray absorption spectroscopy (XAS) variable temperatureNMR and density functional theory (DFT) calculations Thecollective data support assignment of an open-shell singletground state that maintains a formal Fe(II) oxidation statecomprised of an intermediate-spin iron(II) (SFe = 1)antiferromagnetically coupled to a doubly reduced tripletterpyridine ligand (Stpy = 1) This electronic structure alsoenables the molecular catalyst to operate at low overpotentials inboth organic and neutral aqueous electrolytes resulting in highselectivity for CO2RR over HER

RESULTS AND DISCUSSIONDesign Synthesis and Characterization of the Redox

Non-Innocent tpyPY2Me Ligand and Its Metal Coordi-nation Complexes Showing Deviations from IdealizedOctahedral Geometry The tpyPY2Me ligand and corre-sponding iron(II) complex [Fe(tpyPY2Me)]2+ abbreviated as[Fe1]2+ were prepared following modifications of previouslyreported procedures for the pentapyridine PY5Me2 ligand andits metal complexes (see Supporting Information) Briefly thepentapyridine ligand tpyPY2Me bearing a redox non-innocentterpyridine fragment was formed by a lithium-mediatedcoupling between 6-bromo-22prime6prime2Prime-terpyridine and 22prime-(ethane-11-diyl)dipyridine A series of control coordinationcomplexes were synthesized and evaluated to disentangle thecontributions of the redox-active ligand from the metal center(Figure 2) In order to decipher the role of the iron center inelectrochemical behavior and subsequent catalysis we preparedthe diamagnetic isostructural zinc(II) analog [Zn-(tpyPY2Me)]2+ ([Zn1]2+) that is equipped with the sameredox-active tpyPY2Me ligand but with a redox-silent zinc

Figure 1 Bioinspired design of a molecular iron polypyridyl complex for electrochemical carbon dioxide reduction catalysis The molecular CO2reduction catalyst [Fe1]2+ was designed to feature a mononuclear iron center supported by a redox-active pendant in a robust synthetic polypyridylscaffold inspired by the active site of mononuclear iron hydrogenase

Figure 2Chemical structures of [Fe1]2+ and the coordination complexes used as controls to disentangle the contributions of the ligand and the metalcenter in catalysis (a) Chemical structure of [Fe1]2+ highlighting cooperativity between the open-shell iron center and redox non-innocent tpyPY2Meligand (b) The diamagnetic analog containing the redox-silent Zn(II) center [Zn1]2+ enables the role of the iron center to be probed with the sameredox non-innocent ligand scaffold Likewise (c) [Fe(PY5Me2)(CH3CN)]

2+ and (d) [Zn(PY5Me2)(CH3CN)]2+ both feature the redox-innocent

PY5Me2 as a related pentapyridine ligand and thus aid in identifying the contributions of the redox-active terpyridine fragment within the polypyridylscaffold

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20490

center To aid in identifying the role of the redox-activeterpyridine fragment of the polypyridyl scaffold [Fe(PY5Me2)-(CH3CN)]

2+ 50 and [Zn(PY5Me2)(CH3CN)]2+ 26 complexes

bearing the related pentapyridine PY5Me2 ligand withoutpolypyridine conjugation were synthesized Indeed irreversiblereduction of the PY5Me2 ligand occurs at very negativepotentials (between minus20 and minus25 V vs FcFc+) and is coupledto catalyst decomposition thus it serves as a polypyridyl redox-innocent ligand control (Figure 2)A comparison of the solid-state structures of [Fe1]2+ and

[Fe(PY5Me2)(CH3CN)]2+(50) obtained via single-crystal X-ray

diffraction (XRD) revealed that the enhanced rigidity of thetpyPY2Me ligand enforces a more distorted pseudo-octahedral(Oh) geometry in [Fe1]2+ relative to [Fe(PY5Me2)-(CH3CN)]

2+ which adopts an almost ideal Oh geometry(Figure 3a Tables S1 and S2) In particular occupation of theiron equatorial plane by the three terpyridine nitrogen atomsand coordination of the pyridine arms to axial and equatorialpositions induces severe axial distortions in [Fe1]2+ that areabsent in [Fe(PY5Me2)(CH3CN)]

2+ (Table S2) This dis-tortion also significantly compresses the axial bond angle in

[Fe1]2+ (17364(6)deg) whereas the corresponding angle in[Fe(PY5Me2)(CH3CN)]2+ is nearly linear (17853(7)deg)(Figure 3a) Distortion of the primary coordination sphere asa result of the enhanced ligand rigidity of tpyPY2Me is alsoobserved for the control analog bearing the redox-silent Zn(II)center [Zn1]2+ The solid-state structure of [Zn1]2+ (Figure S1Tables S3 and S4) shows that tpyPY2Me enforces a severelydistorted square pyramidal coordination geometry with theterpyridine nitrogen atoms occupying equatorial positions Thisstructure departs quite significantly from the Oh geometryobserved for [Zn(PY5Me2)(CH3CN)]

2+26

The structure of [Fe1]2+ remains more distorted than that of[Fe(PY5Me2)(CH3CN)]

2+ in solution as verified by 1H NMRspectra collected for both complexes in CD3CN at 295 K The1H NMR spectrum of Fe(PY5Me2)(CH3CN)]

2+ features fivesharp signals consistent with a C2v symmetric complex with low-spin Fe(II) (S = 0) whereas the 1H NMR spectrum of [Fe1]2+

features slightly more broadened resonances with no well-defined symmetry (Figure S2) Alternatively [Zn1]2+ shows a1H NMR spectrum with all expected ligand resonancesaccounted for consistent with its diamagnetic ground state(see Supporting Information) Using the Evans method wemeasured an effective magnetic moment (μeff) of 127 μB for[Fe1]2+ at 293 K which is indicative of a low-spin Fe(II) groundstate (S = 0) with a population of thermally accessible spinexcited states at room temperature (Figure S3) Because the[Fe1]2+ bond lengths measured at 100 K are in agreement withthose of other low-spin Fe(II) polypyridyl complexes50 thecollective data suggest that [Fe1]2+ undergoes a spin-statetransition between 100 K and room temperature To furtherprobe the spin-state transition of [Fe1]2+ variable-temperature(VT) 1H NMR spectra were collected in CD3CN across atemperature range of 233 to 353 K (Figure S4) At 293 Kresonances at 947 773 712 and 674 ppm appear as verybroad singlets and upon heating the sample to 353 K thesepeaks become undetectable due to broadening into the baseline(Figure S4b) Upon stepwise cooling the resonances start tobecome fully resolved At 233 K integration of the aryl regionaccounts for the 18 proton signals expected for [Fe1]2+ Thistemperature-dependent signal broadening is in line with ourinterpretation of a population of thermally accessible spinexcited states Due to the absence of systematic temperature-dependent paramagnetic shifts of the [Fe1]2+ resonances wewere unable to quantify the spin equilibrium from the 1H NMRdata alone Nevertheless variable temperature Evans methoddata further revealed a gradual increase in μeff upon heating inagreement with the occurrence of a spin transition process insolution (Figure S5) The limited temperature range available inacetonitrile and limited solubility of [Fe1]2+ in alternativesolvents to expand this temperature range prevents isolation of apure population of low-spin Fe(II) (μeff of 120 μB at 233 K) andthe transition is incomplete at the highest available temperature(μeff of 133 μB at 353 K) precluding accurate determination ofthe spin-equilibrium transition (T12)Zero-field 57Fe Mossbauer spectra collected for [Fe1]2+ at 5

and 295 K further support such a spin state transition for theFe(II) center (Figure S6 and Tables S5) At 5 K the spectrumfeatures a symmetric Lorentzian doublet with isomer shift (δ)and quadrupole splitting (|ΔEQ|) values of 03562(4) and10203(8) mms respectively consistent with a low-spiniron(II) (Figure S6a and Tables S5) The density functionaltheory (DFT) computed Mossbauer parameters closely matchthe experimental values within the well-established uncertainty

Figure 3 Structural characterization of iron polypyridyl complexes (a)Solid state structures of [Fe(PY5Me2)(CH3CN)]

2+ featuring theredox-innocent PY5Me2 ligand and [Fe1]2+ as determined via X-raydiffraction analysis Thermal ellipsoids are plotted at 80 probabilitylevel orange blue and gray ellipsoids represent Fe N and C atomsrespectively Non-coordinating solvent and anions are omitted forclarity (b) Simplified Walsh diagram illustrating the effect of axialdistortion away from the near ideal Oh symmetry in [Fe(PY5Me2)-(CH3CN)]

2+ toward C2v symmetry in [Fe1]2+

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20491

ranges (sim010 mms for δ and sim03 mms for |ΔEQ|)5152 thus

corroborating the low-spin Fe(II) assignment (Figure S7 andTable S6) In contrast the Mossbauer spectrum measured at295 K features an asymmetric quadrupole doublet that can be fitto two different iron sites (Figure S6b and Table S5) Theparameters of the major site are nearly identical to the 5 Kspectrum when accounting for second-order Doppler shifts(SODS) (δ = 0285(1) mms |ΔEQ| = 1056(2) mms) and aretherefore consistent with the same low-spin Fe(II) speciesobserved at 5 K The isomer shift of the minor site is significantlymore positive (062(1) mms) and is indicative of a higher spiniron(II) species DFT predictedMossbauer parameters for thesehigher-spin Fe(II) species correspond closely to the exper-imental values for the minor site measured at 295 K furthersupporting this assignment (Figure S7 and Table S6)The distinct iron(II) spin states in [Fe1]2+ and [Fe(PY5Me2)-

(CH3CN)]2+ can be understood from the simplified Walsh

diagram presented in Figure 3b Axial distortions engendered bythe rigid tpyPY2Me ligand lead to a decrease in symmetry fromidealizedOh to C2v and this distortion significantly stabilizes thedz2 orbital and destabilizes the dxz through z-component mixingof the t1u and t2g orbitals The stabilization of the dz2 orbital anddisruption of t2g degeneracy can then facilitate the formation ofthermally accessible spin excited states near room temperatureWe hypothesize that these marked differences in the ground-state electronic structures of [Fe1]2+ and Fe(PY5Me2)-(CH3CN)]

2+ contribute to their distinct electrochemical andcatalytic behavior (vide inf ra)Cyclic Voltammetry of tpyPY2Me Fe(II) and Zn(II)

Complexes and Comparison to PY5Me2 Analogs Lacking

the Redox Non-Innocent Terpyridine Fragment Thepresence of a redox-active terpyridine moiety in tpyPY2Me alsogives rise to unique electrochemical behavior for [Fe1]2+ and[Zn1]2+ relative to their PY5Me2 analogues The room-temperature cyclic voltammogram (CV) of [Fe1]2+ collectedin CH3CN under Ar exhibits three reversible features a one-electron oxidation at E12 = 066 V (versus FcFc+) assigned tothe iron(IIIII) couple and two closely spaced one-electronredox processes that are centered at minus143 V (Figure 4a)Further chemical characterization of this redox behaviorindicates that these redox events are two sequential ligand-based reductions (tpyPY2Me01minus and tpyPY2Me1minus2minus) (videinf ra) During constant potential coulometry of 76 times 10minus3

mmol of [Fe1]2+ conducted at minus160 V vs FcFc+ under Aratmosphere 158 C of charge was passed equating to 21electrons per molecule of [Fe1]2+ supporting this assignment astwo closely spaced one-electron redox waves centered at minus143V vs FcFc+ (Figure S8) No additional features were observedwhen scanning to more negative potentialsIn contrast under the same conditions the cyclic voltammo-

gram of [Zn1]2+ exhibits two well-separated reversible one-electron ligand-centered reductions with E12 of minus156 andminus207 V as well as an additional irreversible redox event(tentatively assigned as tpyPY2Me2minus3minus) atminus241 V (Figure 4a)The first ligand centered reduction of [Zn1]2+ occurs at acomparable potential to that of [Fe1]2+ but the second ligandcentered reduction is shifted 640 mVmore negative and it is thisdisparate electrochemical behavior between [Fe1]2+ and[Zn1]2+ that indicates a degree of communication betweenthe reduced tpyPY2Me ligand and the open-shell iron center

Figure 4 Electrochemical characterization Cyclic voltammograms collected under Ar of (a) [Fe1]2+ (red trace) and [Zn1]2+ (black trace) complexesand (b) [Fe(PY5Me2)(CH3CN)]

2+ (red trace) and [Zn(PY5Me2)(CH3CN)]2+ (black trace) complexes supported by redox-innocent PY5Me2 These

data show that the unique combination of Fe and redox non-innocent tpyPY2Me ligand lead to electrochemically reversible two-electron chemistrywith the Zn-tpyPY2Me analog exhibiting two reversible one-electron reduction waves with the second reduction at potentials 640 mV more negativethan the iron analog The Fe(II) and Zn(II) PY5Me2 analogs both exhibit irreversible electrochemical reductions at potentials more negative thanminus20V vs FcFc+ Cyclic voltammograms of (c) [Fe1]2+ and (d) [Fe(PY5Me2)(CH3CN)]

2+ collected under Ar (black) CO2 (red) and with added phenol(35 M blue) Voltammograms were collected with a scan rate of 100 mVs with an electrolyte of 010 M TBAPF6 dissolved in CH3CN

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20492

that is not possible in the isostructural complex containing theclosed-shell redox-silent zinc center In contrast ligand-basedreductions in [Fe(PY5Me2)(CH3CN)]

2+ and [Zn(PY5Me2)-(CH3CN)]

2+ are electrochemically irreversible and occur atpotentials negative of minus20 V (Figure 4b) highlighting thecrucial role of the terpyridyl fragment in achieving electro-chemically reversible reduction events at relatively positivepotentialsElectrochemical Reduction of CO2 Catalyzed by

[Fe(tpyPY2Me)]2+ in Organic Solution Under an atmos-phere of CO2 CV data collected for [Fe1]2+ exhibit only a subtleincrease in current relative to scans under Ar however whenexcess phenol (35 M) is added as a proton source two largecatalytic waves appear at ca minus150 and minus20 V versus FcFc+

(Figure 4c) The PY5Me2 analog [Fe(PY5Me2)(CH3CN)]2+

also displays catalytic currents upon addition of CO2 andphenol but the onset current for catalysis is gt500 mV morenegative than that observed for [Fe1]2+ (Figure 4d) Phenoltitrations conducted with solutions of [Fe1]2+ under a CO2atmosphere revealed a dose-dependence on phenol concen-tration (Figure S9a) The electrochemical response is mitigatedwhen control phenol titrations are collected under an Aratmosphere suggesting that [Fe1]2+ selectively catalyzeselectrochemical CO2 reduction (CO2RR) over the hydrogenevolution reaction (HER) even at high concentrations oforganic acid (Figure S9b) As anticipated the isostructuralZn(II) analog [Zn1]2+ does not show similar increases incatalytic current at such positive potentials (Figure S10)establishing that the redox non-innocent tpyPY2Me ligand alonewith a Lewis acid center is not sufficient for promoting CO2RRactivityControlled Potential Electrolysis Studies for Direct

Product Measurements of Electrochemical CO2RR Cata-lyzed by [Fe(tpyPY2Me)]2+ in Organic Solution Prepara-tive-scale controlled potential electrolysis (CPE) experimentswere conducted in CO2-saturated acetonitrile with 35M phenolacross a range of applied potentials (minus135 to minus198 V vs FcFc+) and the products were quantified by gas chromatographyWe find that CO is the major product of CO2RR Importantlywe note that subsequent turnover number and turnover ratevalues reported in this work are derived from these directmeasurements of product as opposed to indirect methodsderived from transient CV currents with electrochemicalassumptionsBecause the utility of a catalyst is benchmarked by its rate and

overpotential (η) the applied potentials were also convertedinto overpotentials following the methods reported by Mayer53

and Matsubara and co-workers5455 Accurate conversionnecessitates that the thermodynamic potential for CO2reduction to CO is known under the experimental conditionsWhereas the aqueous standard potential is known (minus0106 V vsSHE56 minus0730 V vs FcFc+ 57) it is difficult to experimentallydetermine in organic solvents where the identity and amount ofthe proton source complicate the measurements This caveatoften leads to the use of estimated values that do not match theexperimental conditions or rely on thermodynamic assumptionsthat do not rigorously apply such as taking the CO2 and COstandard states to be 1M1314 or assuming anhydrous conditionswithout reporting measured water concentrations or usinganhydrous organic acids58 leading to potential inconsistenciesin the literature with regard to reported overpotentials With thegoal of increasing standardization and transparency in molecularelectrocatalysis we report the applied potentials from our CPE

experiments in both overpotential and versus FcFc+ and use aconservative estimate of ECO2CO

0 (minus127 V vs FcFc+)55

obtained from measured pKa values for acetonitrileminuswatermixtures (see Supporting Information for additional details) Itshould be noted that this value is 770 mVmore positive than forldquodryrdquo acetonitrile and 70 mV more positive than the previouslyestimated standard redox potential for ldquowetrdquo acetonitrile derivedby assigning unusual gas standard states1314 Tafel plots thatcompare catalyst activity are plotted to show both the calculatedoverpotential as well as the directly measured potential versusthe reference electrode used in the experiments We reason thatreporting both values should lead to more clarity inbenchmarking catalysts[Fe1]2+ is highly selective for CO2 reduction over hydrogen

evolution across the entire potential window examined (Figure5a and Figures S11 and S12 Table S7) with averaged specificcurrent densities for CO production (jCO) reaching as high as 36mAcm2 (Figure 5a and Figures S11 and S12 Table S7) The

Figure 5 Electrochemical CO2 reduction performance of [Fe1]2+ (a)Selectivity of [Fe1]2+ for CO2 reduction to CO at varying over-potentials and specific current densities for CO (jCO) production(averages of three experiments) (b) Catalytic Tafel plot for [Fe1]2+

obtained from CPE experiments in acetonitrile with added phenol (35M) CPE experiments were collected in an electrolyte of 010 MTBAPF6 dissolved in CH3CN

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20493

Faradaic efficiency for CO production (FECO) reaches 94 at anoverpotential of 190 mV (minus146 V vs FcFc+) and increases to100 with the application of overpotentials between 320 and710 mV (minus159 and minus198 V vs FcFc+) (Figure 5a and FigureS11 Table S7) This high activity and selectivity for COproduction at applied potentials as low asminus146 V vs FcFc+ is asubstantial decrease in the energy requirements compared toother molecular catalysts and sits among the most efficientreported catalysts1427minus2959minus61 Significantly no CO and onlysmall amounts of H2 (FEH2

= 44) were detected in analogousCPE experiments with [Fe(PY5Me2)(CH3CN)]

2+ at muchlarger overpotentials (Figure S13) Moreover variable potentialCPE experiments with [Zn1]2+ conducted betweenminus166 V andminus241 V vs FcFc+ produced only trace amounts of CO (FECO =1minus6) and a small amount of H2 (FEH2

= 1minus20) furtherillustrating the importance of metalminusligand cooperativity forefficient CO2RR catalysis (Figure S14)Control experiments strongly support the homogeneous

molecular nature of [Fe1]2+ First CO2 reduction was notobserved in the absence of [Fe1]2+ (Figure S15) Furthermorecyclic voltammetry and UVminusvis data collected before and after 1h CPE experiments are indistinguishable attesting to the bulkstability of [Fe1]2+ under electrolysis conditions (Figure S16)Finally the possible formation of an electrode-adsorbed activecatalyst was excluded through a series of control experimentsFirst peak currents for [Fe1]2+ and [Zn1]2+ scale linearly withthe square root of the scan rate indicative of freely diffusingspecies in solution (Figures S17 and S18) Second CPE rinsetests62 which probe for electrode deposition give complete

inversion of product selectivity with H2 as the predominantproduct and negligible generation of CO (Figure S19)

Kinetic Analysis of Electrochemical CO2RR Catalyzedby [Fe(tpyPY2Me)]2+ in Organic Solution Encouraged bythe high selectivity of [Fe1]2+ for electrochemical CO2RR overcompeting HER processes we sought to evaluate its kineticperformance Cyclic voltammograms acquired for [Fe1]2+

under CO2-saturated electrolyte in the presence of 35 Mphenol (Figure S20) exhibit scan rate dependence andnoncanonical peak-shaped waves indicative of competitivenonideal processes (eg substrate consumption productinhibition etc) As a result kobs (TOFmax) obtained fromcatalytic plateau current analysis is not appropriate here and willlead to underestimations of the rates of CO2 reduction63

Therefore we instead extracted kinetic parameters directly fromour variable potential CPE experiments (Figure S11) asdescribed by Saveant and co-workers6465 where the products(CO and H2) were detected and quantified by gaschromatography (Figures S21 and S22) The observed rateconstants (kobs) at each applied potential were extracted fromthe averaged specific current density for CO production (jCO)taken across the entire 1 h CPE experiments (Figure S11) thatwere conducted in triplicate (see Supporting Information foradditional details)A catalytic Tafel plot was constructed from our CPE

experiments with direct product quantification (Figure 5b)and is summarized in Table S7 Tafel analysis identifies [Fe1]2+

as one of the most active molecular electrocatalysts reported todate with it achieving a turnover frequency (TOF) of 75000 sminus1

at an applied overpotential as low as 190 mV (minus146 V vs Fc

Figure 6 Synthetic preparation and characterization of the two-electron reduced species [Fe1]0 The collective structural and spectroscopic datasupport assignment to an open-shell singlet species that maintains a formal Fe(II) oxidation state (a) Two-electron reduction of [Fe1]2+ generates the[Fe1]0 complex that is a putatively active species for CO2 reduction (b)

1H NMR spectrum of [Fe1]2+ (black) and 1H NMR spectrum of the productresulting from chemical reduction of [Fe1]2+with CoCp2 (purple) both spectra were in CD3CN at room temperature (c) Zero-field 57FeMossbaueranalysis of [Fe1]2+ (black) and [Fe1]0 (purple) measured at 5 K (d) Solid-state structure of [Fe1]0 from single-crystal X-ray diffraction analysisThermal ellipsoids are plotted at 80 probability level orange blue and gray ellipsoids represent Fe N and C atoms respectively Residual solventand co-crystallized CoCp2 are omitted for clarity

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20494

Fc+) and reaching TOFs gt 500 000 sminus1 between overpotentialsof 560minus710 mV (minus181 and minus198 V vs FcFc+) (Figure S23)Notably this rapid homogeneous CO2 reduction catalysis isachieved without any modifications to the secondary coordina-tion sphere to introduce hydrogen bondproton-relay function-alities to break electronic scaling relationships13144766minus69

In parallel we also attempted to apply foot-of-the-waveanalysis (FOWA) as pioneered by Saveant6465 to estimate thekinetics based solely on CVs collected in the presence andabsence of CO2 FOWA estimates the idealized catalyticreactivity by considering the initial portion of the catalyticwave where the effects of side phenomena are expected to beminimal Unfortunately FOWA does not find clear applicationhere because the linear region of the FOW is very small and is illdefined due to the appearance of a pre-wave feature at minus136 Vvs FcFc+ and a multi-segmented catalytic wave with apparentmaxima at minus156 and minus196 V vs FcFc+ (Figures 4c and S24)The existence of this multi-segmented catalytic wave maysuggest that the catalytic resting state is changing as morenegative potentials are applied Therefore the extraction ofkinetic data based on our CPE experiments results in moreaccurate analysis as they are determined from preparative scaleexperiments that directly detect and measure CO rather thanrelying on estimations of observed rate constants determinedfrom CV dataFinally to assess the long-term catalytic stability and

recyclability of [Fe1]2+ four consecutive 5 h CPE experimentswere conducted at minus198 V (versus FcFc+) the highestpotential investigated without addition of fresh [Fe1]2+

between each experiment (Figure S25) [Fe1]2+ is remarkablystable with 221C of charge passed over the 20 h windowwithoutany loss in selectivity for CO (average FECO = 87)Chemical Reduction and Characterization of [Fe-

(tpyPY2Me)]0 ([Fe1]0) We reasoned that the observedelectrocatalytic efficiency of [Fe1]2+ for CO2RR stems from asynergistic effect between the iron center and the redox-activetpyPY2Me ligand as both components are necessary to achieveselective CO2RR reactivity To establish a molecular-levelframework for understanding this metalminusligand cooperativitywe sought to synthesize isolate and characterize the two-electron reduced complex [Fe(tpyPY2Me)]0 ([Fe1]0) whichwe hypothesize has putative catalytic relevance Chemicalreduction of [Fe1]2+ with 22 equiv of decamethylcobaltocene(Figure 6a) produced [Fe1]0 as a dark purple solidInterestingly room temperature 1H NMR analysis of [Fe1]0

in CD3CN revealed that the compound is diamagnetic with thearyl and aliphatic signals shifted upfield (Figure 6b) In supportof this result a nearly identical 1H NMR spectrum was obtainedfor the product [Zn1]0 generated from the chemical reductionof [Zn1]2+ with potassium graphite (see Figure S26 and detailsin the Supporting Information) These data are consistent withour assignment of the reduction features for both [Fe1]2+ and[Zn]2+ as ligand-centered events Frozen solution Fe K-edge X-ray absorption near-edge spectroscopy (XANES) was used tocharacterize metal oxidation states (Figure 7) Absorption edgeenergies were determined from the second derivative zerocrossings giving the following values (eV) 71269 ([Fe1]0) and71274 ([Fe1]2+) These values are similar to compounds in theliterature70 leading to the assignment of the oxidation states ofboth species as Fe(II)Given the unique diamagnetic ground state of [Fe1]0 two

major electronic structures are possible The first possibility is alow-spin iron(II) center (SFe = 0) coordinated to a reduced

closed-shell singlet ligand (StpyPY2Me = 0) generated by bothelectrons occupying the same π orbital (ie [Fe-(tpyPY2Me)2minus]0) The other possibility is an intermediate-spin iron(II) center (SFe = 1) antiferromagnetically coupled to areduced ligand with an open-shell triplet configuration (StpyPY2Me= 1) through half occupation of both low-lying π terpyridineorbitals leading to an overall metalminusligand spin-coupledelectronic structure (ie [Fe(tpyPY2Mebullbull)2minus]0) The twoelectronic configurations are more clearly distinguished hereby specifying the spin state of the ligand as either a singlet(tpyPY2Me)2minus or a triplet (tpyPY2Mebullbull)2minus in the chemicalformula Because of the large anodic shift (640 mV) in theligand-based reductions of [Fe1]2+ relative to [Zn1]2+ (Figure4a) the electrochemistry indicates a strong influence of themetal center and supports assignment to a [Fe-(tpyPY2Mebullbull)2minus]0 electronic configuration In further supportof this model Mossbauer data for [Fe1]0 revealed an isomershift of δ = 02608(2) mms and a quadrupole splitting of |ΔEQ|= 07637(4) mms consistent with the formation ofintermediate-spin iron(II) (Figures 6c and S27)71 Theantiferromagnetic coupling between the reduced biradicalligand (tpyPY2Mebullbull)2minus and the intermediate-spin iron(II) isexpected to be quite strong given that a diamagnetic 1H NMRspectrum is observed at room temperature Indeed DFTcalculations predict a large coupling constant of minus1023 cmminus1giving rise to a singlet-quintet gap of 4400 cmminus1 (see theSupporting Information)Further insights into the electronic structure of [Fe1]0 can be

gleaned from the solid-state structure of [Fe1]0 Single crystalsof [Fe1]0 were obtained by slow evaporation of a saturatedacetonitrile solution and X-ray diffraction analysis revealed adistorted square pyramidal geometry (Figure 6d Tables S8 andS9) resulting from the loss of the axial acetonitrile ligand in[Fe1]2+ This structure is in good agreement with the reductionpathway predicted by DFT (vide inf ra) The bond metrics of theterpyridine moiety unambiguously supports the two-electronreduction of the tpyPY2Me ligand in the solid state and thus theiron(II) oxidation state assignment In particular the con-traction of the intra-pyridine bond lengths (CpyminusCprimepy) from anaverage of 1474(3) Aring in [Fe1]2+ to 1434(4) Aring in [Fe1]0 isconsistent with the metrics with which the oxidation levels ofterpyridine can be established as determined through structuralstudies on a series of iron(II) and chromium(III) bis-terpyridinecomplexes reported by Wieghardt and co-workers7273

Furthermore average CminusN bond lengths of the terpyridine

Figure 7 Experimental Fe K-edge XAS spectra of [Fe1]2+ (black) andtwo-electron reduced [Fe1]0 (red) supporting a formal Fe(II)oxidation state assignment for both species

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20495

moiety in [Fe1]0 (1384(3) Aring) depart quite significantly fromwhat would be expected for typical neutral aromatic pyridinerings (135plusmn 01 Aring respectively)72 and what was observed in thesolid state structure of [Fe1]2+ (1353(2) Aring)Distinguishing between the two possible electronic config-

urations of [Fe1]0 based on the single crystal X-ray diffractiondata is more challenging The DFT predicted structure for theopen-shell singlet ([Fe(tpyPY2Mebullbull)2minus]0) more closelymatches the experimental bond metrics of [Fe1]0 than doesthe predicted structure of the closed-shell singlet ([Fe-(tpyPY2Me)2minus]0) however the solid state structures aresusceptible to additional packing and dispersion forces that arenot accounted for by DFT calculations (Table S10) In theorythe unique bonding interactions of the two terpyridineLUMOs72minus74 should enable us to directly distinguish betweenthe two possible electronic configurations of [Fe1]0 but thechange in bond metrics between the dianionic singlet ligand(tpyPY2Me2minus) and the triplet π radical dianion (tpy-PY2Mebullbull2minus) is expected to be quite subtle given the metricsreported by Wieghardt and co-workers on a series of reducedhomoleptic bis(terpyridine) complexes ([M(tpy)2]

n+ M = CrFe)7273 Attempts to crystallize [Zn1]0 have been unsuccessfulto date thus preventing direct crystallographic comparison to[Fe1]0Although absolute determination of the electronic structure

should not be based solely on X-ray diffraction when it is takentogether with the electrochemical NMR andMossbauer resultsthe collective data support the iron(II) oxidation stateassignment of [Fe1]0 as an open-shell singlet ground state

([Fe(tpyPY2Mebullbull)2minus]0) composed of an intermediate-spiniron(II) center (SFe = 1) that is antiferromagnetically coupledto a doubly reduced tpyPY2Me (Stpy = 1) ligand This strongexchange coupling shifts the tpyPY2Me-based reductions of[Fe1]2+ to more positive potentials relative to the analog withthe redox-silent Zn(II) center by 640mV which we hypothesizeultimately promotes highly selective CO2 reduction by [Fe1]

2+To further elucidate the electronic structures of [Fe1]2+0 wesubsequently applied extensive DFT and multireferencemethods

Electronic Structure Calculations Support StrongMetalminusLigand Exchange Coupling Between Fe andtpyPY2Me Units We carried out DFT calculations using theωB97X-D functional to generate a more accurate molecularorbital picture of our [Fe1]2+ system Optimized structures andpredicted redox potentials for [Fe1]2+ and [Zn1]2+ are in goodagreement with experimental data (Tables S11 and 12)Interestingly although the DFT data indicate the metal centerin [Fe1]2+ is best described as low-spin iron(II) the calculatedsinglet-quintet gap is small explaining the observed thermalpopulation of a higher spin-state at room temperature (TablesS13 and 14) The first reduction of [Fe1]2+ to [Fe1]+ ispredicted to generate a ground state doublet composed of a low-spin iron(II) and a tpyPY2Me-based radical (tpyPY2Mebull)1minus

with a predicted redox potential of minus146 V (Figure 8aminusc) Thespin density plot of this [Fe1]+ species indicates a ligand-basedreduction with almost no excess spin on the iron center (Figure8b) The second reduction populates a second low lying πorbital of tpyPY2Me and induces loss of the axial acetonitrile

Figure 8 DFT analysis and calculated molecular orbitals for [Fe1]+ and [Fe1]0 species (a) DFT predicted pathway for reduction of [Fe1]2+ to theopen-shell singlet [Fe1]0 [Fe(tpyPY2Mebullbull)2minus]0 The computed reduction potentials are given for reduction of [Fe1]2+ to [Fe1]+ and [Fe1]+ to[Fe1]0 See Table S11 for the predicted reduction pathways for [Zn(1)]2+0 and for the closed-shell singlet [Fe1]0 [Fe(tpyPY2Me)2minus]0 (b) Spindensity plot and (c) simplified molecular orbital diagram of the one electron reduced intermediate [Fe1]+ (d) Spin density plot and (e) simplifiedmolecular orbital diagram of [Fe1]0 ([Fe(tpyPY2Mebullbull)2minus]0) Localized metal orbitals are depicted in the Supporting Information

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20496

molecule providing access to an intermediate-spin iron(II) state(Figure 8ade) The SFe = 1 iron center is stronglyantiferromagnetically coupled to the two electrons in thetpyPY2Me π orbitals (ie [Fe(tpyPY2Mebullbull)2minus]0 Figure 8e)Localized orbital bonding analysis confirmed that iron remainsin the 2+ oxidation state during the second reduction processwhich has a predicted reduction potential of minus151 V Thispredicted redox potential is in good agreement with thatobserved experimentallyIn contrast the other feasible closed-shell singlet config-

uration a low-spin iron(II) center (SFe = 0) coordinated to areduced closed-shell doubly reduced singlet ligand (StypPY2Me =0) (ie [Fe(tpyPY2Me)2minus]0) does not match the experimentaldata The doubly reduced tpy ligand is strongly π basic and thusdelocalizes some electron density into the empty metal orbitals(distorted eg-type) The computed reduction potential of minus203V vs FcFc+ using the [Fe(tpyPY2Me)2minus]0 electronic config-uration does not match the experimental value ofminus143 V vs FcFc+ This difference can be rationalized by the large free energygap (12 kcalmol) between the broken symmetry solution([Fe(tpyPY2Mebullbull)2minus]0) and the closed-shell singlet ([Fe-(tpyPY2Me)2minus]0) consisting of low-spin iron(II) (SFe = 0)and doubly reduced ligand (StpyPY2Me = 0) (Tables S13 and 14)It should also be noted that a large variety of different hybriddensity functionals all predict the broken symmetry config-uration as the ground state on the singlet surface (see TableS14)In addition multi-reference calculations using complete

active space self-consistent field (CASSCF) confirmed theantiferromagnetically coupled ground state (see the SupportingInformation) This calculation reveals a strong entanglement ofthe two ligand-based orbitals and the two metal d-orbitals Inparticular the axial distortion away from octahedral symmetryallows the dz2 and dxy metal orbitals (or superposition of the twoorbitals) to interact with the two tpyPY2Me π orbitals in a weakπ-type interaction (Figure S28) Hence the novel ligandframework with two low-lying π orbitals a distortedcoordination geometry and a moderate ligand field incombination with the iron metal center stabilizes two excesselectrons effectively The spin transition alters both theenergetics of the d-orbitals and local spin on the central metalto facilitate an effective antiferromagnetic coupling In contrastin the alternative closed-shell singlet electronic configuration([Fe(tpyPY2Me)2minus]0 SFe = 0 and Stpy = 0) the stabilization ofthe doubly reduced tpyPY2Me π (Stpy = 0) by the high lying egof the low-spin iron(II) (SFe = 0) is marginal This feature isillustrated by the almost identical second ligand reductionpotentials of [Zn1]2+ and [Fe1]2+ with a closed-shell singletelectronic structure (minus202 V for [Zn1]2+ and minus203 V vs FcFc+ for [Fe1]2+ with SFe = 0 and Stpy = 0)We note that a similar electronic structure involving this

intramolecular antiferromagnetic coupling was proposed byNeese and co-workers for the electronic structure of anotherCO2 reduction catalyst the doubly reduced iron tetraphenylpor-phryin complex [Fe(TPP)]2minus75 Through quantum mechanicalcalculations and spectroscopic data (Mossbauer and XAS) theyassign both reductions of [FeTPP]2minus to the porphyrin ligandframework and show that it is coupled to an intermediate-spiniron center (SFe = 1) giving an overall singlet state

75 It should benoted that the iron center in the un-reduced iron(II) complex[Fe(TPP)]0 is intermediate-spin and therefore the metal centerdoes not undergo a spin transition upon reduction Thisbehavior can be contrasted to the group VII CO2 reduction

catalyst [Re(bpy)(CO)3]minus (and its Mn derivative) where the

quantum chemical calculations and spectroscopic data revealedthat one electron is placed into the π orbital of the non-innocent bpy ligand and the second electron is placed in themetal dz2 orbital but is strongly delocalized into the threecarbonyl ligands Both electrons form a metalminusligand bond(delocalized metalminusligand orbital) with a singlet ground statewhich is 15 kcalmol lower in energy than the uncoupled tripletstate32 Thus the ldquoinitialrdquo d-orbitals are not involved in thestabilization of the excess electrons as they are in [Fe1]2+ and[Fe(TPP)]

Electrochemical CO2RR Catalyzed by [Fe-(tpyPY2Me)]2+ in Water Given the low overpotentialsrequired for [Fe1]2+ to catalyze homogeneous electrochemicalCO2 reduction in organic solvent we sought to evaluate itsactivity in water (Figure 9) For this purpose we prepared themore water-soluble nitrate analog of [Fe1]2+ (see theSupporting Information) and X-ray diffraction analysis revealedthat anion exchange does not alter the catalyst structure (FigureS29 Tables S15 and S16) As a cheap abundant and benignsolvent and proton source water is an attractive medium forCO2 reduction However homogeneous molecular CO2

Figure 9 Aqueous CO2 reduction performance for the Fe-tpyPY2Meelectrocatalyst (a) Faradaic efficiencies for CO (blue) and H2 (red)formation at varying applied potentials for 60 min (averages of threeexperiments) The average FECO is indicated by the black dashed line(b) Catalytic Tafel plot for [Fe1]2+ obtained from CPE experiments in010 M NaHCO3

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20497

reduction catalysts typically require mercury working electrodesto operate effectively in aqueous electrolytes because their onsetpotentials are too negative to be compatible with more desirablecarbon-based electrodes that preferentially catalyze competingwater reduction1076minus79 Owing to the 640mV positive shift in 2-electron reduction potential upon moving from Zn(II) toFe(II) we envisioned that CO2RR could be favored over HEReven with water as both a solvent and proton sourceAgainst this backdrop we were pleased to observe that cyclic

voltammograms collected for [Fe1]2+ dissolved in 010 MNaHCO3 with a carbon paste working electrode show theformation of a catalytic wave when saturated with CO2 (FigureS30) Notably variable potential CPE experiments (FiguresS31minusS35) revealed that [Fe1]2+ maintains its activity andselectivity for CO2 reduction to CO in water (average FECO =92 and FEH2

lt 1) with maximum turnover frequenciesdetermined from direct product detection reaching gt50 000 sminus1

at overpotentials of 500 mV (Figure 9b) Importantly controlrinse tests passed negligible current during 1 h CPE experimentsand the selectivity was inverted such that H2 is themajor product(Figures S31minusS36) UVminusvisible spectra collected before andafter electrolysis are nearly indistinguishable however cyclicvoltammograms change quite drastically following the 1 h CPE(Figure S37) indicative of some degree of catalyst depositionand decomposition on the glassy carbon electrode In line withthis observation electrochemical stripping behavior is alsoobserved which may contribute to slower catalysis in waterrelative to organic solvent (Figure S38) Nevertheless these datashow that [Fe1]2+ is an effective molecular catalyst forhomogeneous CO2 reduction in water Crucially given itsmolecular nature the activity stability and overpotential of thiscatalyst can in principle be tuned through ligand modificationsaccessible by synthetic chemistry Further the study of thiscatalyst and related systems has the potential to shed importantinsights on the behavior of materials catalysts bearing atomicallydispersed iron sites80

CONCLUDING REMARKS

In summary inspired by the beneficial feature of electronicdelocalization exhibited by biological and heterogeneoussystems for electrochemical CO2RR catalysis to minimizeoverpotentials and favor CO2RR over competing HER path-ways even in aqueous media we report a molecular system thatcaptures this design concept A new ligand framework thatincorporates a redox-active terpyridine moiety into a modularpolypyridyl scaffold can achieve highly selective and efficientiron-catalyzed electrochemical CO2 reduction to CO in bothorganic and aqueous solutions with high specificity overcompeting hydrogen evolution pathways Importantly thecombination of both an open-shell metal ion using an earth-abundant first-row transition metal in conjunction with a redoxnon-innocent ligand enables metal-cooperativity through strongexchange coupling that shifts electrochemical reductionpotentials to promote CO2RR catalysis Specifically the Fe(II)tpyPYMe2 analog exhibits a 640 mV positive shift in the 2-electron reduction potential relative to the isostructural analogbearing a redox-silent Zn(II) center poising it for multielectronelectrocatalysis with minimized overpotentials Control experi-ments with Fe(II) and Zn(II) complexes bearing a relatedpentapyridine ligand lacking the redox-active terpyridinefragment and the same redox non-innocent ligand but with aredox-silent Zn(II) center show that CO2RR relies on

synergistic coupling between both metal and ligand compo-nentsThe combination of distorted coordination geometry low-

lying π orbitals and moderate ligand field engendered by therigid redox non-innocent tpyPY2Me ligand promotes strongmetalminusligand cooperativity with the iron(II) center [Fe1]2+ isshown to catalyze CO2 reduction at extremely mild redoxpotentials (η = 190 mV minus 146 V vs FcFc+) relative to thecontrol polypyridyl complexes presented in this study Thepositive redox potentials at which CO2RR catalysis occurs isinterpreted through extensive experimental spectroscopic(NMR Mossbauer and XAS) XRD and electrochemicalstudies along with electronic structure calculations whichcollectively point to a model in which the two-electron reducediron complex [Fe1]0 possesses an open-shell singlet groundstate electronic structure ([Fe(tpyPY2Mebullbull)2minus]0) resultingfrom a strong intramolecular antiferromagnetic couplingbetween a triplet reduced ligand (tpyPY2Me tpyPY2Mebullbull2minusStpyPY2Me = 1) and an intermediate-spin Fe(II) center (SFe = 1)We propose that the multielectron chemistry of [Fe1]2+ atrelatively positive potentials is facilitated through this electronicstructure which in turn promotes its exceptional selectivity(FECO gt 90) through suppression of HER pathwaysFurthermore variable potential CPE experiments with directproduct detection show that [Fe1]2+ operates at fast rates inboth organic (TOF gt 100 000 sminus1) and aqueous electrolytes(TOF gt 50 000 sminus1) with robust long-term stability andrecyclabilityThe ability to create molecular compounds with strong

exchange coupling between synergistic redox non-innocentligands and metal centers has broad implications for catalystdesign including enabling control over the density andarrangement of functional active sites Indeed redox balancingthrough electronic delocalization is pivotal for biological andmaterials catalysts to achieve complex chemical transformationsat or near thermodynamic potentials However discretemolecular compounds offer a level of tunability via coordinationchemistry that remains unparalleled in protein design andnanostructure engineering Thus inorganic molecules areuniquely positioned to open access to a range of new propertiesand offer a desirable alternative strategy for lowering energybarriers for redox catalysis The observed 640 mV electro-chemical potential shift and associated turn-on of CO2RRcatalysis upon substituting Zn(II) for Fe(II) for the same redoxnon-innocent ligand emphasizes metalminusligand cooperativity forredox balancing that goes beyond redox activity dominated ateither ligand or metal sites alone

ASSOCIATED CONTENT

sı Supporting InformationThe Supporting Information is available free of charge athttpspubsacsorgdoi101021jacs0c10664

X-ray crystallographic data for [Fe1]2+ [Zn1]2+ [Fe1]0and [Fe(tpyPY2Me)(NO3)][NO3] (CIF)

Experimental and computational details compoundcharacterization supplemental electrochemical and spec-troscopic data and DFT geometry optimized atomic xyzcoordinates (PDF)

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20498

AUTHOR INFORMATION

Corresponding AuthorChristopher J Chang minus Department of Chemistry andDepartment of Molecular and Cell Biology University ofCalifornia Berkeley California 94720 United StatesChemical Sciences Division Lawrence Berkeley NationalLaboratory Berkeley California 94720 United Statesorcidorg0000-0001-5732-9497 Email chrischang

berkeleyedu

AuthorsJeffrey S Derrick minus Department of Chemistry University ofCalifornia Berkeley California 94720 United StatesChemical Sciences Division Lawrence Berkeley NationalLaboratory Berkeley California 94720 United States

Matthias LoipersbergerminusDepartment of Chemistry Universityof California Berkeley California 94720 United Statesorcidorg0000-0002-3648-0101

Ruchira Chatterjee minus Molecular Biophysics and IntegratedBioimaging Division Lawrence Berkeley National LaboratoryBerkeley California 94720 United States

Diana A Iovan minus Department of Chemistry University ofCalifornia Berkeley California 94720 United Statesorcidorg0000-0001-9889-7183

Peter T Smith minus Department of Chemistry University ofCalifornia Berkeley California 94720 United StatesChemical Sciences Division Lawrence Berkeley NationalLaboratory Berkeley California 94720 United States

Khetpakorn Chakarawet minus Department of ChemistryUniversity of California Berkeley California 94720 UnitedStates orcidorg0000-0001-5905-3578

Junko Yano minusMolecular Biophysics and Integrated BioimagingDivision Lawrence Berkeley National Laboratory BerkeleyCalifornia 94720 United States orcidorg0000-0001-6308-9071

Jeffrey R Long minus Department of Chemistry and Department ofChemical and Biomolecular Engineering University ofCalifornia Berkeley California 94720 United StatesMaterials Sciences Division Lawrence Berkeley NationalLaboratory Berkeley California 94720 United Statesorcidorg0000-0002-5324-1321

Martin Head-Gordon minus Department of Chemistry Universityof California Berkeley California 94720 United StatesChemical Sciences Division Lawrence Berkeley NationalLaboratory Berkeley California 94720 United Statesorcidorg0000-0002-4309-6669

Complete contact information is available athttpspubsacsorg101021jacs0c10664

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTSThis research was supported by the Director Office of ScienceOffice of Basic Energy Sciences and the Division of ChemicalSciences Geosciences and Bioscience of the US Departmentof Energy at Lawrence BerkeleyNational Laboratory (Grant NoDE-AC02-05CH11231 to CJC Grant No DE-AC02-05CH11231 to JY) and the US Department of EnergyOffice of Science Office of Advanced Scientific Computing andthe Office of Basic Energy Sciences via the Scientific Discoverythrough Advanced Computing (SciDAC) program (toMH-G)

Collection and interpretation of Mossbauer spectra weresupported by National Science Foundation CHE-1800252 toJRL CJC is a CIFAR Fellow PTS acknowledges the NSFfor a graduate research fellowship We thank Dr Hasan Celikand UC Berkeleyrsquos NMR facility in the College of Chemistry(CoC-NMR) and Dr Ping Yu and UC Davisrsquos NMR facility forspectroscopic assistance Instruments in the CoC-NMR aresupported in part by NIH S10OD024998 We thank Dr NSettineri for assistance with X-ray crystallography We thank RMurphy C Gould C Stein and Dr A Wuttig for fruitfuldiscussions and Dr K R Meihaus for editorial assistance Wededicate this manuscript to Prof Dick Andersen

REFERENCES(1) Lewis N S Nocera D G Powering the Planet ChemicalChallenges in Solar Energy Utilization Proc Natl Acad Sci U S A2006 103 (43) 15729minus15735(2) De Luna P Hahn C Higgins D Jaffer S A Jaramillo T FSargent E H What Would It Take for Renewably PoweredElectrosynthesis to Displace Petrochemical Processes Science 2019364 (6438) 3506(3) Francke R Schille B Roemelt M Homogeneously CatalyzedElectroreduction of Carbon DioxideMethods Mechanisms andCatalysts Chem Rev 2018 118 (9) 4631minus4701(4) Can M Armstrong F A Ragsdale S W Structure Functionand Mechanism of the Nickel Metalloenzymes CO Dehydrogenaseand Acetyl-CoA Synthase Chem Rev 2014 114 (8) 4149minus4174(5) Corbin N Zeng J Williams K Manthiram K HeterogeneousMolecular Catalysts for Electrocatalytic CO2 Reduction Nano Res2019 12 (9) 2093minus2125(6) Smith P T Nichols E M Cao Z Chang C J Hybrid Catalystsfor Artificial Photosynthesis Merging Approaches from MolecularMaterials and Biological Catalysis Acc Chem Res 2020 53 (3) 575minus587(7) Fisher B J Eisenberg R Electrocatalytic Reduction of CarbonDioxide by Using Macrocycles of Nickel and Cobalt J Am Chem Soc1980 102 (24) 7361minus7363(8) Beley M Collin J-P Ruppert R Sauvage J-P Nickel(II)-Cyclam An Extremely Selective Electrocatalyst for Reduction of CO2in Water J Chem Soc Chem Commun 1984 19 1315minus1316(9) Schneider J Jia H Kobiro K Cabelli D E Muckerman J TFujita E Nickel(II) Macrocycles Highly Efficient Electrocatalysts forthe Selective Reduction of CO2 to CO Energy Environ Sci 2012 5(11) 9502minus9510(10) Froehlich J D Kubiak C P Homogeneous CO2 Reduction byNi(Cyclam) at a Glassy Carbon Electrode Inorg Chem 2012 51 (7)3932minus3934(11) Hammouche M Lexa D Saveant J M Momenteau MCatalysis of the Electrochemical Reduction of Carbon Dioxide byIron(ldquo0rdquo) Porphyrins J Electroanal Chem 1988 249 (1) 347minus351(12) Meshitsuka S Ichikawa M Tamaru K Electrocatalysis byMetal Phthalocyanines in the Reduction of Carbon Dioxide J ChemSoc Chem Commun 1974 No 5 158minus159(13) Costentin C Drouet S Robert M Saveant J-M A LocalProton Source Enhances CO2 Electroreduction to CO by a MolecularFe Catalyst Science 2012 338 (6103) 90minus94(14) Azcarate I Costentin C Robert M Saveant J-M Through-Space Charge Interaction Substituent Effects in Molecular CatalysisLeading to the Design of the Most Efficient Catalyst of CO2-to-COElectrochemical Conversion J Am Chem Soc 2016 138 (51) 16639minus16644(15) Ren S Joulie D Salvatore D Torbensen K Wang MRobert M Berlinguette C P Molecular Electrocatalysts CanMediateFast Selective CO2 Reduction in a Flow Cell Science 2019 365 (6451)367minus369(16) Hawecker J Lehn J-M Ziessel R Electrocatalytic Reductionof Carbon Dioxide Mediated by Re(bipy)(CO)3Cl (bipy = 22rsquo-Bipyridine) J Chem Soc Chem Commun 1984 328minus330

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20499

(17) Smieja J M Kubiak C P Re(bipy-tBu)(CO)3Cl-improvedCatalytic Activity for Reduction of Carbon Dioxide IR-Spectroelec-trochemical and Mechanistic Studies Inorg Chem 2010 49 (20)9283minus9289(18) Reda T Plugge C M Abram N J Hirst J ReversibleInterconversion of Carbon Dioxide and Formate by an ElectroactiveEnzyme Proc Natl Acad Sci U S A 2008 105 (31) 10654minus10658(19) Evans R M Siritanaratkul B Megarity C F Pandey KEsterle T F Badiani S Armstrong F A The Value of Enzymes inSolar Fuels Research - Efficient Electrocatalysts through EvolutionChem Soc Rev 2019 48 (7) 2039minus2052(20) Nitopi S Bertheussen E Scott S B Liu X Engstfeld A KHorch S Seger B Stephens I E L Chan K Hahn C Noslashrskov JK Jaramillo T F Chorkendorff I Progress and Perspectives ofElectrochemical CO2 Reduction on Copper in Aqueous ElectrolyteChem Rev 2019 119 (12) 7610minus7672(21) Todorova T K Schreiber M W Fontecave M MechanisticUnderstanding of CO2 Reduction Reaction (CO2RR) TowardMulticarbon Products by Heterogeneous Copper-Based CatalystsACS Catal 2020 10 (3) 1754minus1768(22) Rodriguez P Koper M T M Electrocatalysis on Gold PhysChem Chem Phys 2014 16 (27) 13583minus13594(23) Karunadasa H I Chang C J Long J R A MolecularMolybdenum-Oxo Catalyst for Generating Hydrogen from WaterNature 2010 464 (7293) 1329minus1333(24) Karunadasa H I Montalvo E Sun Y Majda M Long J RChang C J A Molecular MoS2 Edge Site Mimic for CatalyticHydrogen Generation Science 2012 335 (6069) 698minus702(25) Zee D Z Chantarojsiri T Long J R Chang C J Metal-Polypyridyl Catalysts for Electro- and Photochemical Reduction ofWater to Hydrogen Acc Chem Res 2015 48 (7) 2027minus2036(26) Sun Y Bigi J P Piro N A Tang M L Long J R Chang CJ Molecular Cobalt Pentapyridine Catalysts for Generating Hydrogenfrom Water J Am Chem Soc 2011 133 (24) 9212minus9215(27) Chen Z Chen C Weinberg D R Kang P Concepcion J JHarrison D P Brookhart M S Meyer T J ElectrocatalyticReduction of CO2 to CO by Polypyridyl Ruthenium Complexes ChemCommun 2011 47 (47) 12607minus12609(28) Gonell S Massey M D Moseley I P Schauer C KMuckerman J T Miller A J M The Trans Effect in ElectrocatalyticCO2 Reduction Mechanistic Studies of Asymmetric RutheniumPyridyl-Carbene Catalysts J Am Chem Soc 2019 141 (16) 6658minus6671(29) Gonell S Assaf E A Duffee K D Schauer C K Miller A JM Kinetics of the Trans Effect in Ruthenium Complexes ProvideInsight into the Factors That Control Activity and Stability in CO2

Electroreduction J Am Chem Soc 2020 142 (19) 8980minus8999(30) Li T-T Shan B Xu W Meyer T J Electrocatalytic CO2

Reduction with a Ruthenium Catalyst in Solution and on Nanocrystal-line TiO2 ChemSusChem 2019 12 (11) 2402minus2408(31) Liu Y Fan X Nayak A Wang Y Shan B Quan X MeyerT J Steering CO2 Electroreduction toward Ethanol Production by aSurface-Bound Ru Polypyridyl Carbene Catalyst on N-Doped PorousCarbon Proc Natl Acad Sci U S A 2019 116 (52) 26353minus26358(32) Benson E E Sampson M D Grice K A Smieja J MFroehlich J D Friebel D Keith J A Carter E A Nilsson AKubiak C P The Electronic States of Rhenium Bipyridyl Electro-catalysts for CO2 Reduction as Revealed by X-Ray AbsorptionSpectroscopy and Computational Quantum Chemistry AngewChem Int Ed 2013 52 (18) 4841minus4844(33) Clark M L Cheung P L Lessio M Carter E A Kubiak CP Kinetic and Mechanistic Effects of Bipyridine (bpy) SubstituentLabile Ligand and Broslashnsted Acid on Electrocatalytic CO2 Reductionby Re(bpy) Complexes ACS Catal 2018 8 (3) 2021minus2029(34) Keith J A Grice K A Kubiak C P Carter E A Elucidationof the Selectivity of Proton-Dependent Electrocatalytic CO2 Reductionby Fac-Re(bpy)(CO)3Cl J Am Chem Soc 2013 135 (42) 15823minus15829

(35) Su X McCardle K M Panetier J A Jurss J WElectrocatalytic CO2 Reduction with Nickel Complexes Supported byTunable Bipyridyl-N-Heterocyclic Carbene Donors UnderstandingRedox-Active Macrocycles Chem Commun 2018 54 (27) 3351minus3354(36) Su X McCardle K M Chen L Panetier J A Jurss J WRobust and Selective Cobalt Catalysts Bearing Redox-Active Bipyridyl-N-Heterocyclic Carbene Frameworks for Electrochemical CO2

Reduction in Aqueous Solutions ACS Catal 2019 9 (8) 7398minus7408(37) Lin S Diercks C S Zhang Y-B Kornienko N Nichols EM Zhao Y Paris A R Kim D Yang P Yaghi O M Chang C JCovalent Organic Frameworks Comprising Cobalt Porphyrins forCatalytic CO2 Reduction in Water Science 2015 349 (6253) 1208minus1213(38) Diercks C S Lin S Kornienko N Kapustin E A Nichols EM Zhu C Zhao Y Chang C J Yaghi O M Reticular ElectronicTuning of Porphyrin Active Sites in Covalent Organic Frameworks forElectrocatalytic Carbon Dioxide Reduction J Am Chem Soc 2018140 (3) 1116minus1122(39) Gonell S Lloret-Fillol J Miller A An Iron Pyridyl-CarbeneCatalyst for Low Overpotential CO2 Reduction to CO MechanisticComparisons with the Ruthenium Analogue and PhotochemicalPromotion ChemRxiv Preprint 2020 11911383(40) Eisenberg R Gray H B Noninnocence in Metal Complexes ADithiolene Dawn Inorg Chem 2011 50 (20) 9741minus9751(41) McNamara W R Han Z Alperin P J Brennessel W WHolland P L Eisenberg R A Cobalt-Dithiolene Complex for thePhotocatalytic and Electrocatalytic Reduction of Protons J Am ChemSoc 2011 133 (39) 15368minus15371(42) McNamara W R Han Z Yin C-J Brennessel W WHolland P L Eisenberg R Cobalt-Dithiolene Complexes for thePhotocatalytic and Electrocatalytic Reduction of Protons in AqueousSolutions Proc Natl Acad Sci U S A 2012 109 (39) 15594minus15599(43) Letko C S Panetier J A Head-Gordon M Tilley T DMechanism of the Electrocatalytic Reduction of Protons withDiaryldithiolene Cobalt Complexes J Am Chem Soc 2014 136(26) 9364minus9376(44) Wu Y Jiang J Weng Z Wang M Broere D L J Zhong YBrudvig G W Feng Z Wang H Electroreduction of CO2 Catalyzedby a Heterogenized Zn-Porphyrin Complex with a Redox-InnocentMetal Center ACS Cent Sci 2017 3 (8) 847minus852(45) Proppe A H Li Y C Aspuru-Guzik A Berlinguette C PChang C J Cogdell R Doyle A G Flick J Gabor N M vanGrondelle R Hammes-Schiffer S Jaffer S A Kelley S O LeclercM Leo K Mallouk T E Narang P Schlau-Cohen G S ScholesG D Vojvodic A Yam V W-W Yang J Y Sargent E HBioinspiration in Light Harvesting and CatalysisNat Rev Mater 20205 828(46) Jurss J W Khnayzer R S Panetier J A El Roz K A NicholsE M Head-Gordon M Long J R Castellano F N Chang C JBioinspired Design of Redox-Active Ligands for MultielectronCatalysis Effects of Positioning Pyrazine Reservoirs on Cobalt forElectro- and Photocatalytic Generation of Hydrogen from WaterChem Sci 2015 6 (8) 4954minus4972(47) Nichols E M Derrick J S Nistanaki S K Smith P TChang C J Positional Effects of Second-Sphere Amide Pendants onElectrochemical CO2 Reduction Catalyzed by Iron Porphyrins ChemSci 2018 9 (11) 2952minus2960(48) Smith P T Weng S Chang C J An NADH-Inspired RedoxMediator Strategy to Promote Second-Sphere Electron and ProtonTransfer for Cooperative Electrochemical CO2 Reduction Catalyzed byIron Porphyrin Inorg Chem 2020 59 (13) 9270minus9278(49) Dey S Das P K Dey A Mononuclear Iron HydrogenaseCoord Chem Rev 2013 257 (1) 42minus63(50) Chantarojsiri T Sun Y Long J R Chang C J Water-SolubleIron(IV)-Oxo Complexes Supported by Pentapyridine Ligands AxialLigand Effects on Hydrogen Atom and Oxygen Atom TransferReactivity Inorg Chem 2015 54 (12) 5879minus5887

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20500

(51) Romelt M Ye S Neese F Calibration of Modern DensityFunctional Theory Methods for the Prediction of 57Fe MossbauerIsomer Shifts Meta-GGA and Double-Hybrid Functionals InorgChem 2009 48 (3) 784minus785(52)McWilliams S F Brennan-Wydra E MacLeod K C HollandP L Density Functional Calculations for Prediction of 57Fe MossbauerIsomer Shifts and Quadrupole Splittings in β-Diketiminate ComplexesACS Omega 2017 2 (6) 2594minus2606(53) Pegis M L Roberts J A S Wasylenko D J Mader E AAppel A M Mayer J M Standard Reduction Potentials for Oxygenand Carbon Dioxide Couples in Acetonitrile and NN-Dimethylforma-mide Inorg Chem 2015 54 (24) 11883minus11888(54) Matsubara Y Grills D C Kuwahara Y ThermodynamicAspects of Electrocatalytic CO2 Reduction in Acetonitrile and with anIonic Liquid as Solvent or Electrolyte ACS Catal 2015 5 (11) 6440minus6452(55)Matsubara Y Standard Electrode Potentials for the Reduction ofCO2 to CO in Acetonitrile-Water Mixtures Determined Using aGeneralized Method for Proton-Coupled Electron-Transfer ReactionsACS Energy Lett 2017 2 (8) 1886minus1891(56) Bard A J Parsons R Jordan J Standard Potentials in AqueousSolution M Dekker New York 1985(57) Pavlishchuk V V Addison A W Conversion Constants forRedox Potentials Measured versus Different Reference Electrodes inAcetonitrile Solutions at 25degC Inorg Chim Acta 2000 298 (1) 97minus102(58) Hooe S L Dressel J M Dickie D A Machan C W HighlyEfficient Electrocatalytic Reduction of CO2 to CO by a MolecularChromium Complex ACS Catal 2020 10 (2) 1146minus1151(59) Ngo K T McKinnonMMahanti B Narayanan R Grills DC Ertem M Z Rochford J Turning on the Protonation-FirstPathway for Electrocatalytic CO2 Reduction by Manganese BipyridylTricarbonyl Complexes J Am Chem Soc 2017 139 (7) 2604minus2618(60) Dey S Ahmed M E Dey A Activation of Co(I) State in aCobalt-Dithiolato Catalyst for Selective and Efficient CO2 Reduction toCO Inorg Chem 2018 57 (10) 5939minus5947(61) Ahmed M E Rana A Saha R Dey S Dey A HomogeneousElectrochemical Reduction of CO2 to CO by a Cobalt PyridineThiolate Complex Inorg Chem 2020 59 (8) 5292minus5302(62) Lee K J McCarthy B D Dempsey J L On DecompositionDegradation and Voltammetric Deviation The Electrochemistrsquos FieldGuide to Identifying Precatalyst Transformation Chem Soc Rev 201948 (11) 2927minus2945(63) Rountree E S McCarthy B D Eisenhart T T Dempsey J LEvaluation of Homogeneous Electrocatalysts by Cyclic VoltammetryInorg Chem 2014 53 (19) 9983minus10002(64) Costentin C Drouet S Robert M Saveant J-M TurnoverNumbers Turnover Frequencies and Overpotential in MolecularCatalysis of Electrochemical Reactions Cyclic Voltammetry andPreparative-Scale Electrolysis J Am Chem Soc 2012 134 (27)11235minus11242(65) Costentin C Drouet S Robert M Saveant J-M Correctionto Turnover Numbers Turnover Frequencies and Overpotential inMolecular Catalysis of Electrochemical Reactions Cyclic Voltammetryand Preparative-Scale Electrolysis J Am Chem Soc 2012 134 (48)19949minus19950(66) Galan B R Schoffel J Linehan J C Seu C Appel A MRoberts J A S Helm M L Kilgore U J Yang J Y DuBois D LKubiak C P Electrocatalytic Oxidation of Formate by [Ni-(PR

2NRprime2)2(CH3CN)]

2+ Complexes J Am Chem Soc 2011 133(32) 12767minus12779(67) Sung S Kumar D Gil-Sepulcre M Nippe M ElectrocatalyticCO2 Reduction by Imidazolium-Functionalized Molecular Catalysts JAm Chem Soc 2017 139 (40) 13993minus13996(68) Haviv E Azaiza-Dabbah D Carmieli R Avram L Martin JM L Neumann R A Thiourea Tether in the Second CoordinationSphere as a Binding Site for CO2 and a Proton Donor Promotes theElectrochemical Reduction of CO2 to CO Catalyzed by a Rhenium

Bipyridine-Type Complex J Am Chem Soc 2018 140 (39) 12451minus12456(69) Chapovetsky A Welborn M Luna J M Haiges R Miller TF Marinescu S C Pendant Hydrogen-Bond Donors in CobaltCatalysts Independently Enhance CO2 Reduction ACS Cent Sci 20184 (3) 397minus404(70) Rudd D J Goldsmith C R Cole A P Stack T D PHodgson K O Hedman B X-Ray Absorption SpectroscopicInvestigation of the Spin-Transition Character in a Series of Single-Site Perturbed Iron(II) Complexes Inorg Chem 2005 44 (5) 1221minus1229(71) Gutlich P Eckhard B Trautwein A XMossbauer Spectroscopyand Transition Metal Chemistry Fundamentals and ApplicationsSpringer Berlin 2011(72) Scarborough C C Lancaster K M DeBeer S WeyhermullerT Sproules S Wieghardt K Experimental Fingerprints for Redox-Active Terpyridine in [Cr(tpy)2](PF6)n (n = 3minus0) and the RemarkableElectronic Structure of [Cr(tpy)2]

1‑ Inorg Chem 2012 51 (6) 3718minus3732(73) England J Scarborough C C Weyhermuller T Sproules SWieghardt K Electronic Structures of the Electron Transfer Series[M(bpy)3]

n [M(tpy)2]n and [Fe(t bpy)3]n (M= Fe Ru n = 3+ 2+ 1+

0 1-) A Mossbauer Spectroscopic and DFT Study Eur J Inorg Chem2012 4605minus4621(74) Garino C Gobetto R Nervi C Salassa L Rosenberg ERoss J B A Chu X Hardcastle K I Sabatini C Spectroscopic andComputational Studies of a Ru(II) Terpyridine Complex TheImportance of Weak Intermolecular Forces to Photophysical Proper-ties Inorg Chem 2007 46 (21) 8752minus8762(75) Romelt C Song J Tarrago M Rees J A van Gastel MWeyhermuller T DeBeer S Bill E Neese F Ye S ElectronicStructure of a Formal Iron(0) Porphyrin Complex Relevant to CO2Reduction Inorg Chem 2017 56 (8) 4745minus4750(76) Kang P Meyer T J Brookhart M Selective ElectrocatalyticReduction of Carbon Dioxide to Formate by a Water-Soluble IridiumPincer Catalyst Chem Sci 2013 4 (9) 3497minus3502(77) Costentin C Robert M Saveant J-M Tatin A Efficient andSelective Molecular Catalyst for the CO2-to-CO ElectrochemicalConversion in Water Proc Natl Acad Sci U S A 2015 112 (22)6882minus6886(78) Taheri A Thompson E J Fettinger J C Berben L A AnIron Electrocatalyst for Selective Reduction of CO2 to Formate inWater Including Thermochemical Insights ACS Catal 2015 5 (12)7140minus7151(79) Nakada A Ishitani O Selective Electrocatalysis of a Water-Soluble Rhenium(I) Complex for CO2 Reduction Using Water As anElectron Donor ACS Catal 2018 8 (1) 354minus363(80) Gu J Hsu C-S Bai L Chen H M Hu X AtomicallyDispersed Fe3+ Sites Catalyze Efficient CO2 Electroreduction to COScience 2019 364 (6445) 1091minus1094

Journal of the American Chemical Society pubsacsorgJACS Article

httpsdxdoiorg101021jacs0c10664J Am Chem Soc 2020 142 20489minus20501

20501