Cobalt-dithiolene complexes for the photocatalyticand electrocatalytic reduction of protonsin aqueous solutionsWilliam R. McNamara, Zhiji Han, Chih-Juo (Madeline) Yin, William W. Brennessel, Patrick L. Holland1, andRichard Eisenberg1

Department of Chemistry, University of Rochester, Rochester, NY 14627

Edited by* Thomas J. Meyer, University of North Carolina at Chapel Hill, Chapel Hill, NC, and approved May 3, 2012 (received for review December 15, 2011)

Artificial photosynthesis (AP) is a promising method of convertingsolar energy into fuel (H2). Harnessing solar energy to generate H2

from Hþ is a crucial process in systems for artificial photosynthesis.Widespread application of a device for AP would rely on the use ofplatinum-free catalysts due to the scarcity of noblemetals. Herewereport a series of cobalt dithiolene complexes that are exception-ally active for the catalytic reduction of protons in aqueous solventmixtures. All catalysts perform visible-light-driven reduction ofprotons from water when paired with RuðbpyÞ2þ3 as the photosen-sitizer and ascorbic acid as the sacrificial donor. Photocatalysts withelectronwithdrawing groups exhibit the highest activity with turn-overs up to 9,000 with respect to catalyst. The same complexes arealso active electrocatalysts in 1∶1 acetonitrile/water. The electroca-talytic mechanism is proposed to be ECEC, where the Co dithiolenecatalysts undergo rapid protonation once they are reduced toCoL2−

2 . Subsequent reduction and reaction with Hþ lead to H2 for-mation. Cobalt dithiolene complexes thus represent a new groupof active catalysts for the reduction of protons.

electrocatalysis ∣ hydrogen evolution ∣ photocatalysis ∣ redox active ligands

Conversion of energy from the sun into chemically stored en-ergy via artificial photosynthesis (AP) represents a promising

approach to providing renewable energy needed for global devel-opment (1–4). In general, systems for AP are designed to splitwater photochemically, with water oxidation to O2 and aqueousproton reduction to H2. Focusing separately on the oxidative andreductive sides of the water splitting reaction facilitates achievingthis goal while allowing individualized approaches for each com-ponent. For H2 generation from aqueous protons driven by light,both a photosensitizer (PS) and a catalyst are needed, while anelectron source is required to allow the reductive half-reaction tobe studied separately from, and independently of, the oxidativehalf-reaction. In this context, the design and study of new cata-lysts that are composed of inexpensive earth-abundant elementsare highly desired objectives. Despite extensive investigations andanalyses of the components for the reductive side of water split-ting, an active and robust system for using the energy from sun-light to drive molecular hydrogen production in water remains achallenge (5).

Recently, efforts toward new catalyst development havefocused on electrocatalysts for proton reduction (6–17). In thisregard, some cobaloxime catalysts derived from dimethylglyox-ime are active electrocatalytically and operate at particularlylow overpotentials (9, 18, 19). One of these electrocatalysts,CoðdmgHÞ2ðpyrÞCl (dmgH ¼ dimethylglyoximate monoanion)has also been found to be active as a hydrogen generating catalystwhen combined with a suitable photosensitizer and sacrificialelectron donor (6, 10, 20). In recently reported photochemical H2

generation, systems containing cobaloxime catalysts have shownconsiderable activity (21–25) with turnover numbers (TONs) ashigh as 9,000 with respect to chromophore (25–27). However, onelimitation is that the TONs are modest based on catalyst (<300,

with TOF <100∕h), and hydrogen production ceases after 6 hof irradiation. With regard to the limited lifetime, the cessationof H2 production indicated catalyst decomposition, which inturn suggested ligand hydrogenation as a possible decompositionpathway.

One interesting ligand family that is less susceptible to possiblehydrogenation is the unsaturated 1,2-dithiolenes. Complexes ofdithiolenes were examined initially more than four decades agoregarding their molecular and electronic structures and spectro-scopic and physical properties. In the past decade, interest inthese complexes has been rekindled by their bonding and mag-netic properties as studied with newer methods and better instru-mentation, and by the potential use of their delocalized frontierorbitals for multielectron storage in catalysis (28). Dithiolenecomplexes exhibit intense colors and reversible electron transferseries having relatively minor structural changes (28, 29). Basedon these properties, we commenced an examination of dithiolenecomplexes of the 3d transition elements to determine if theywould be active as catalysts for the reductive side of water split-ting. In an initial disclosure, we reported that (Bu4NÞ½CoðbdtÞ2�(1, where bdt ¼ 1; 2-benzenedithiolate) is an active electrocata-lyst and photocatalyst for proton reduction in aqueous/organicmedia (30). Complex 1 exhibited high activity (>2; 700 TONswith respect to catalyst and an initial turnover frequency of>800∕h) when paired with RuðbpyÞ3 2þ as the chromophore andascorbic acid as the sacrificial electron donor under moderatelyacidic conditions. In this paper, we describe a family of cobaltdithiolene complexes that possess expanded photocatalytic activ-ity (>9; 000 TONs). The relative activity of the different com-plexes provides insight into the mechanism of catalysis.

Results and DiscussionScheme 1 shows the structures of the different cobalt dithiolenecatalysts examined in the present study. For complexes 1–3, de-rivatized benzenedithiolate ligands were employed to give a seriesof catalysts with different electronic properties. The synthesis ofthese complexes proceeded by deprotonating the correspondingbenzenedithiol followed by treatment with cobalt(II) tetrafluor-oborate. The resulting deep-blue complexes (1–3) are planar,monoanionic, and paramagnetic (S ¼ 1), and were isolated astheir tetrabutylammonium (TBA) salts. The molecular structuresof 1–3 were confirmed by single crystal X-ray diffraction (see SIText for full details and CIF), and the results are in agreementwith previous structural reports of these complexes with different

Author contributions: W.R.M., P.L.H., and R.E. designed research; W.R.M., Z.H., andC.-J.M.Y. performed research; W.W.B. contributed new reagents/analytic tools; W.R.M.,Z.H., W.W.B., and R.E. analyzed data; and W.R.M., P.L.H., and R.E. wrote the paper .

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1To whom correspondence may be addressed. E-mail: Holland@Chem.Rochester.edu orEisenberg@Chem.Rochester.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1120757109/-/DCSupplemental.

15594–15599 ∣ PNAS ∣ September 25, 2012 ∣ vol. 109 ∣ no. 39 www.pnas.org/cgi/doi/10.1073/pnas.1120757109

Dow

nloa

ded

by g

uest

on

Aug

ust 1

5, 2

021

noncoordinating cations (28). Complex 4, ðTBAÞ½CoðmntÞ2�where mnt ¼ maleonitriledithiolate, is known to reversibly forma dimeric structure in solution, and upon further reduction is re-duced from ½CoðmntÞ2�− to ½CoðmntÞ2�2−, which is predominantlythe monomer (28). In the present study, 4 is characterized struc-turally in the solid state as the dimer shown in Scheme 1. Thedimeric structure for 4 is similar to those found for other Cobis(dithiolene) monoanions in which the dithiolene is electronwithdrawing. The Co-S linkages holding the dimer together arelonger than the Co-S bond distances in the CoS2C2 chelate rings,which is consistent with the notion of facile monomer-dimer equi-libria in solution (28).

Complex 1 was previously reported to be an active catalyst forH2 generation when paired with RuðbpyÞ2þ3 as the photosensiti-zer and ascorbic acid (AA) as the sacrificial electron donor in 1∶1CH3CN∶H2O (30). To probe aspects of the catalysis, the com-plexes 2–4 were examined under conditions for H2 generationsimilar to those used for 1, and cyclic voltammograms of theCo dithiolene complexes were measured. Table 1 shows theexperimentally determined reduction potentials for the ½CoL2�−∕½CoL2�2− couple, which range from −0.04 V to −0.70 V vs. SCE,depending on the electron donating ability of the dithioleneligand. The mnt complex 4 in monomeric form exhibits the leastnegative reduction potential, while CoðtdtÞ2− (2) shows the mostnegative reduction potential. Complex 4 also exhibits a secondreversible redox wave corresponding to the CoðmntÞ2−2 ∕CoðmntÞ3−2 couple at −1.49 V vs SCE, but the other complexesdo not show an analogous reduction before irreversible reductionof solvent/electrolyte.

Photocatalysis for H2 Generation. Under conditions identical tothose used for H2 photogeneration by 1, catalyst 2 promotes H2

formation with RuðbpyÞ2þ3 as PS and AA as the sacrificial elec-tron donor upon 520 nm irradiation (LED light source 0.15 W,0.1 Mascorbic acid in 1∶1CH3CN∶H2O) at 15 °C (Figs. 1 and 2).In these studies, the production of H2 was monitored in real timeby the pressure change in the reaction vessel and confirmed byGC analysis with a TCD detector. Catalyst 2 is very active, evol-ving 0.54 mL H2∕hr using 6.5 × 10−6 M of catalyst (Fig. 1), cor-responding to an initial TOF (turnover frequency) of 690 h−1 per

catalyst while achieving a TON (turnover number) of 2,300 after20 h. However, this is slightly less active than the results pre-viously reported for 1, which exhibited a TOF of 880 h−1 witha TON of 2,700 after 12 h. The dependence of rate on catalystconcentration shows that the process is first order with respect to2, which is in accord with what has been reported before for thebdt catalyst 1.

Fig. 3 shows the photocatalytic activity of the CoðmntÞ−2 cata-lyst 4 under similar conditions with RuðbpyÞ2þ3 and 0.1 Mascorbic acid at pH 4.0 (1∶1 CH3CN∶H2O). Again, hydrogenevolution is dependent on catalyst concentration. At higher cat-alyst concentrations more H2 is evolved, indicating that hydrogenevolution correlates directly with catalyst concentration similarto the other Co dithiolene-containing systems (see SI Text). TheCoðmntÞ−2 catalyst 4 achieves the highest activity of the dithiolenecomplexes, achieving a total TON of approximately 9,000 (mea-sured with respect to cobalt), with an initial turnover frequencyup to 3450 h−1. This level of activity places 4 among themost active H2 generating catalysts reported (26, 30, 31). The sys-tems remain active for 6–10 h before decomposition of catalystand chromophore contribute to a gradual decrease in hydrogenevolution. While system activity declines dramatically duringphotolysis, at higher catalyst concentrations H2 is still evolvedafter 20 h. In general, total system cessation is observed after 24 hof irradiation.

Mechanismof Photocatalysis. In the photochemical system catalyzedby 1, reductive quenching was proposed to be the predominantphotochemical step in which AA quenches excited *RuðbpyÞ2þ3to form reduced RuðbpyÞþ3 . Since RuðbpyÞþ3 is known to be apowerful reducing agent (E1∕2 ¼ −1.26 V vs. NHE) (32), it isthermodynamically capable of reducing each of the Co dithiolenecomplexes to the dianionic complex (16). Thus, a reductivequenching pathway appeared consistent with the results reportedhere for catalysts 2–4. However, the possibility that catalyst com-plexes 2–4 are directly reduced by the excited state of RuðbpyÞ2þ3was examined directly. We have previously noted that 1 doesindeed quench *RuðbpyÞ2þ3 following good Stern-Volmer behaviorwith a quenching rate constant near the diffusion limit of

Scheme 1. Structures of various cobalt dithiolene catalysts presented inthis study.

Table 1. Redox potentials for cobalt complexes (V vs. SCE) determined from cyclic voltammetry(200 mV∕s) in 1∶1 CH3CN∶H2O at a glassy carbon electrode with 0.1 M KNO3

Complex E1∕2ð−1∕ − 2Þ E1∕2ð−2∕ − 3Þ ipc TONWRC (12 h) TOFWRC (h−1)

1 CoðbdtÞ−2 −0.64 −1.21 2,700 8802 CoðtdtÞ−2 −0.70 −1.32 2,300 6903 CoðCl2bdtÞ−2 −0.51 −0.95 6,000 1,4004 CoðmntÞ−2 −0.04 −1.49 −1.37 9,000 3,400

ipc value was taken at top of catalytic wave.

Fig. 1. Hydrogen evolution with 5 × 10−4 M RuðbpyÞ2þ3 and 0.1 M ascorbicacid at pH 4.0 in 1∶1 CH3CN∶H2O with ½2� ¼ 6.4 μM (red), 4.8 μM (green),3.2 μM (blue), 1.6 μM (black).

McNamara et al. PNAS ∣ September 25, 2012 ∣ vol. 109 ∣ no. 39 ∣ 15595

CHEM

ISTR

YSP

ECIALFEAT

URE

Dow

nloa

ded

by g

uest

on

Aug

ust 1

5, 2

021

4 × 1010 M−1s−1 (30). We find that complexes 2–4 also quench*RuðbpyÞ2þ3 following good Stern-Volmer behavior with quenchingrate constants that are near the diffusion limit, ranging from1.25–1.7 × 1010 M−1s−1 (see SI Text).

However, while the rate constants for quenching *RuðbpyÞ2þ3with Co bis(dithiolene) catalysts are two orders of magnitudelarger than that for ascorbic acid (kq of ascorbic acid ¼1.3 × 108 M−1s−1) (30), reductive quenching by ascorbic acid hasa faster rate in the systems under study here because of themuch larger concentration of AA (0.1 M) relative to that ofcatalyst (5 × 10−6 M). Therefore, our evidence indicates that theinitial photochemical step is most likely the formation of reducedRuðbpyÞþ3 using electrons from AA.

Based on the photochemical data described above, it can beconcluded that there is a correlation of H2 generating activity andligand electronic effects for the Co dithiolene catalysts. Specifi-cally, the complexes with the more electron withdrawing subtitu-ents and therefore the more positive reduction potentials forCoL−

2 ∕CoL2−2 are the more active catalysts. On a TOF basis,

the activity order is 4 > 3 > 1 > 2 with CoðmntÞ−2 being the mostactive. The results thus indicate that it is not the reducing powerof the CoL2−

2 complex that is the key determinant in its catalyticactivity. It was therefore of interest to examine aspects of thisconclusion using these complexes as electrocatalysts for Hþ re-duction. To date, most of the well-studied electrocatalysts forproton reduction and H2 generation operate in nonaqueous med-ia with the proton source being added acid. In contrast, photo-chemical efforts on the reductive side of water splitting havegenerally been conducted in aqueous or aqueous/organic media(33–36). Since CoðbdtÞ−2 was previously reported to be an active

electrocatalyst in aqueous/organic solutions (30), the other Codithiolene complexes (2–4) were examined as electrocatalystsfor H2 generation in both dry CH3CN and aqueous organicsolutions.

Electrocatalysis for H2 Generation. Fig. 4 shows cyclic voltammo-grams of complex 3 after successive additions of trifluoroaceticacid (TFA). In contrast to catalyst 1, where the onset of a catalyticwave grows directly at the same potential as the reversible couplefor CoðbdtÞ−2 ∕CoðbdtÞ2−2 , a catalytic wave is observed at apotential 250 mV more cathodic than the reversible couple forCoðCl2bdtÞ−2 ∕CoðCl2bdtÞ2−2 .

Fig. 5 shows the cyclic voltammograms of catalyst 4 in CH3CNupon the addition of TFA. It is also important to mention thatalthough catalyst 4 crystallizes as dianionic dimer, it is in equili-brium with the monoanionic monomer in solution (28). Whenreduced (at −0.04 V vs. SCE) the complex exists entirely as thedianionic monomer (similar to the benzenedithiolate deriva-tives). Although catalyst 4 is the most easily reduced to the dia-nion (−0.04 V vs. SCE), the onset of a catalytic wave occurs at themost cathodic potential (−1.4 V vs SCE) compared to the otherdithiolene catalysts (see Table 1). Hence, the most active photo-catalyst (4) operates as an electrocatalyst at the least favorable(most negative) potential. In contrast, within the group of benze-nedithiol (bdt) derivatized complexes (1–3), the most activephotocatalyst is the complex that functions as an electrocatalystat the least negative (cathodic) potential. Despite the apparentinconsistency with the results for 3 and 4 as electrocatalysts,the results indicate that the path to H2 formation must proceedthrough reduction of the monoanionic Co bis dithiolene catalystto the dianion.

In both 3 and 4, the new wave appears in the potential rangewhere the anion has been reduced to the dianion, and appearsonly upon addition of acid. This suggests an ECEC mechanism(Scheme 2). After reduction of the bis(dithiolene) complex CoL−

2

to the dianion, the catalytic intermediate is rapidly protonated,after which a second reduction occurs. In a final step that isnot observed directly, reaction with Hþ gives H2 and the initialcatalyst CoL−

2 .The electrocatalytic activity of the Co dithiolene complexes

2–4 was also examined in the aqueous solvent mixtures usedfor the photogeneration of hydrogen. All of the complexes areactive as electrocatalysts in 1∶1 CH3CN∶H2O (see SI Text). Fig. 6shows cyclic voltammograms of 3 in this solvent mixture afterthe addition of 0.1 M TFA. A large current enhancement isobserved at a potential that is more negative than the reversibleredox couple for CoL−

2 ∕CoL2−2 , as with the catalyst in pure

Fig. 2. Hydrogen evolved with 5 × 10−4 M RuðbpyÞ2þ3 and 0.1 M ascorbicacid at pH 4.0 in 1∶1 CH3CN∶H2O with ½3� ¼ 6.4 μM (red), 4.8 μM (green),3.2 μM (blue), 1.6 μM (black).

Fig. 3. Hydrogen evolution with 5 × 10−4 M RuðbpyÞ2þ3 and 0.1 M ascorbicacid at pH 4.0 in 1∶1 CH3CN∶H2O with ½4� ¼ 6.4 μM (red), 4.8 μM (green),3.2 μM (blue), 1.6 μM (black).

Fig. 4. Cyclic voltammograms of 0.5 mM 3 in a 0.1 M solution of TBAPF6 inCH3CN (black) upon addition of 2.2 mM TFA (blue), 4.4 mM TFA (red), 6.6 mMTFA (green), and 8.8 mM TFA (purple). Scan rate: 200 mV∕s with a glassycarbon working electrode.

15596 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1120757109 McNamara et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

5, 2

021

CH3CN. This observation is consistent with the conclusion thatthe Co dithiolene catalysts operate via an ECEC mechanismembodied in Scheme 2 with initial reduction to CoL2−

2 , followedby protonation and subsequent reduction. It is reasonable thatthe potentials are more negative in aqueous solution, becausethe acidity of the TFA is decreased by the additional water.

Differences Between Photocatalytic and Electrocatalytic HydrogenGeneration. The photocatalytic and electrocatalytic data for hy-drogen generation indicate a difference in the relative order ofactivity of complexes for photocatalysis vs. electrocatalysis. Spe-cifically, though catalyst 4 is most active photocatalytically, itexhibits the most negative (cathodic) catalytic wave electrochemi-cally. The activity of an electrocatalyst is most commonly mea-sured in terms of the overpotential at which it operates (17).For 4, the overpotential is the largest of the four Co dithiolenesconsidered in this study. On the other hand, the three derivatizedbenzenedithiolate complexes exhibit the same ordering for bothphotochemical and electrochemical activity. The difference be-tween 1–3 and 4 is likely the result of a difference in electronicstructure of the CoL2−

2 complexes, which in turn may lead to adifference in the site of protonation and the consequent reduc-tion potential of the catalytic wave. Hence, the very differentpotentials in Figs. 4 and 5 for reduction of CoHL−

2 may representredox processes for species that are structurally different by virtueof being protonated at different sites. Support for this view comesfrom previously reported electronic structure calculations. ForCoðmntÞ2−2 , the frontier metal and ligand based orbitals havevery similar energies, suggesting that protonation may take placeat either the metal center or the ligand (29, 37, 38). In contrast,the HOMO for CoðbdtÞ2−2 derivatives have only 29% ligand char-acter, thereby favoring protonation at the metal center (39).So far, direct observation of the key protonated species CoHL−

2

for L ¼ mnt and Cl2bdt has not been achieved.The order of photocatalytic hydrogen production activity for

the Co dithiolene catalysts is 4 > 3 > 1 > 2. This ordering corre-lates with the CoL−

2 ∕CoL2−2 reduction potential from least nega-

tive to most negative, indicating that the reducing ability of theCoL2−

2 intermediate is not the key factor in the turnover limiting

step. The turnover limiting step in the photochemical systemmust also be different from that in the electrocatalytic systembecause of differences in the activity of 4 relative to the othercatalysts. One step that correlates with the observed order of re-activity in the photochemical system is the reduction of the Codithiolene monoanion to the dianion. Since the driving forcefor electron transfer from the reduced photosensitizer RuðbpyÞþ3to CoL−

2 correlates with the H2 generation activity, our resultssuggest that this electron transfer may be turnover limiting.However, the possibility that the turnover limiting step involvesa heterolytic hydrogen elimination (involving the more hydridicCoH and the SH proton) is also an explanation for the observedtrend in reactivity. In this case, the electron withdrawing groupswould result in a more acidic SH proton, resulting in a largerdriving force for H2 elimination. This explanation would be simi-lar to what has been observed for recent nickel catalysts (40).

ConclusionsThe cobalt dithiolene complexes presented in this work are activefor the photocatalytic and electrocatalytic reduction of protonswith impressive TONs and TOFs in aqueous/organic media. Thephotocatalytic mechanism is most likely to proceed through areductive quenching pathway in which *RuðbpyÞ2þ3 reacts withascorbic acid to form RuðbpyÞþ3 , followed by turnover-limitingelectron transfer to CoL−

2, giving CoL2−2 . The cobalt dithiolene

catalysts with the least cathodic CoL−2 ∕CoL2−

2 reduction potentialare the most active photocatalyts. The electrocatalytic mechanismis proposed to be ECEC, where the Co dithiolene catalysts under-go rapid protonation once they are reduced to CoL2−

2 . Subsequentreduction and reaction with Hþ lead to H2 formation. Thesecatalysts demonstrate striking activity with photocatalytic TONsand TOFs among the highest reported.

Materials and MethodsMaterials. 1,2-Benzenedithiol, 3,6-dichlorobenzenedithiol, sodium maleoni-triledithiolate, toluenedithiol, cobalt(II) tetrafluoroborate hexahydrate,½RuðbpyÞ3�Cl2, potassium tert-butoxide, and tetrabutylammonium bromidewere purchased from Aldrich and used without further purification. Catalyst1 was synthesized according to a literature procedure (30).

Synthesis of NBu4½CoðtdtÞ2� (2). The procedure was adapted from previouslyreported methods (41). In a Schlenk flask, CoðBF4Þ2 • 6H2O (277 mg,0.81 mmol) and KOtBu (333 mg, 3.0 mmol) were dissolved in 30 mL of dryMeOH under an N2 atmosphere and allowed to stir at room temperature for30 minutes. To this solution, a degassed mixture of toluenedithiol (277 mg,1.67 mmol) in MeOH (5 mL) was added dropwise. The solution was stirred for4 h at room temperature while the color darkened to deep blue. To this solu-tion, a solution of NBu4Br (395.7 mg, 1.23 mmol) in 3 mL ofMeOHwas added,and the solution was allowed to stir at room temperature overnight. The sol-vent volume was reduced under vacuum to approximately 10 mL, and a darkblue precipitate formed. The solid was collected and recrystallized from a

Fig. 5. Cyclic voltammograms of 0.5 mM 4 in a 0.1 M solution of TBAPF6 inCH3CN (black) upon addition of 2.2 mM TFA (blue), 4.4 mM TFA (green). Scanrate: 200 mV∕s with a glassy carbon working electrode.

Scheme 2. Proposed Mechanism for H2 Generation

Fig. 6. Cyclic voltammograms of 0.5 mM 3 in a 0.1 M solution of TBAPF6 in1∶1 CH3CN∶H2O (red) upon addition of 0.1 M TFA (blue). Scan rate: 200 mV∕swith a glassy carbon working electrode.

McNamara et al. PNAS ∣ September 25, 2012 ∣ vol. 109 ∣ no. 39 ∣ 15597

CHEM

ISTR

YSP

ECIALFEAT

URE

Dow

nloa

ded

by g

uest

on

Aug

ust 1

5, 2

021

layered dichloromethane/ether mixture to yield 2 as a blue crystalline solid(312 mg, 0.51 mmol, 63% yield). Crystals for X-ray diffraction were grown bylayering diethyl ether over a dichloromethane solution of 2. Elem. Anal. Cal-culated: C, 59.08; H, 7.93; N, 2.30; Found: C, 58.83; N, 7.97; H, 2.35. ESI MS:m∕z366.92 (expected), 366.15 (found) amu.

Synthesis of NBu4½CoðCl2bdtÞ2� (3). The procedure was adapted from pre-viously reported methods (41). In a Schlenk flask, CoðBF4Þ2 • 6H2O (153 mg,0.45 mmol) and KOtBu (200 mg, 1.8 mmol) were dissolved in 30 mL of dryMeOH under an N2 atmosphere and allowed to stir at room temperaturefor 30 minutes. To this solution, a degassed mixture of 3,6-dichloro-1,2-ben-zenedithiol (250 mg, 1.1 mmol) in MeOH (5 mL) was added dropwise. Thesolution was allowed to stir for 4 h at room temperature while the color dar-kened to deep blue. To this solution, a solution of NBu4Br (275.2 mg,0.85 mmol) in 3 mL of MeOH was added, and the solution was allowed tostir at room temperature overnight. The solvent volume was reduced undervacuum to approximately 10 mL, and a dark blue precipitate formed. Thesolid was collected and recrystallized from a dichloromethane/ether mixtureto yield 3 as a blue crystalline solid (310 mg, 0.72 mmol, 43% yield). Crystalsfor X-ray diffraction were grown by layering diethyl ether over dichloro-methane. Elem. Anal. Calculated (with 3CH2Cl2): C, 38.21; H, 4.76; N, 1.44;Found: C, 38.455; H, 4.76; N, 1.44. ESI MS: m∕z 477.73 (expected), 476.65(found) amu.

Synthesis of ðNBu4Þ2½CoðmntÞ2�2 (4). The procedure was adapted from pre-viously reported methods (41). In a Schlenk flask, CoðBF4Þ2 • 6H2O (289.5 mg,0.85 mmol) was dissolved in 30 mL of dry MeOH under an N2 atmosphere andallowed to stir at room temperature for 30 minutes. To this solution, a de-gassed mixture of sodium maleonitrile dithiolate (186.17 mg, 1.69 mmol) inMeOH (5 mL) was added dropwise. The solution was allowed to stir for 4 h atroom temperature while the color darkened to deep blue. To this solution, asolution of NBu4Br (275.2 mg, 0.85 mmol) in 3 mL of MeOH was added, andthe solution was allowed to stir at room temperature overnight. The solventvolume was reduced under vacuum to approximately 10 mL, and a dark blueprecipitate formed. The solid was collected and recrystallized from a dichlor-omethane/ether mixture to yield 1 as a dark brown crystalline solid (115 mg,0.19 mmol, 22% yield). Crystals for X-ray diffraction were grown by layeringdiethyl ether over dichloromethane. Elem. Anal. Calculated: C, 49.55; N,12.04; H, 6.24; Found: C, 49.406; N, 11.858; H, 6.615. ESI MS: m∕z 338.83 (ex-pected), 338.75 (found) amu.

Hydrogen Evolution Studies. Samples of 5.0 mL were prepared in 40 mL scin-tillation vials. Stock solutions were prepared of: catalyst, a 8.0 × 10−4 M solu-tion in CH3CN; RuðbpyÞ3 2þ, a 1.0 × 10−3 M solution in CH3CN. Varyingamounts of catalyst, RuðbpyÞ2þ3 , and CH3CN were added to obtain a totalvolume of 2.5 mL. To this solution, 2.5 mL of a 0.2M stock solution of aqueousascorbic acid (adjusted to a specific pH by adding NaOH and measured with apH meter) was added, giving a total volume of 5.0 mL. The samples wereplaced into a temperature controlled block at 15 °C and sealed with an air-tight cap fitted with a pressure transducer and a septum. The samples werethen degassed using a mixture of 20% CH4 in N2, with the CH4 being usedlater as an internal reference for GC analysis. The cells were irradiated from

below with high-power Philips LumiLEDs Luxeon Star Hex green (520 nm)700 mA LEDs. The light power of each LED was set to 0.15 W and measuredwith an L30 A Thermal sensor and Nova II power meter (Ophir-Spiricon LLC).The samples were swirled using an orbital shaker. The pressure changes in thevials were recorded using a Labview program from a Freesale semiconductorsensor (MPX4259A series). After irradiation, the headspaces of the vials weresampled by GC to ensure that pressure increases were due to H2 evolutionand to confirm the amount of H2 evolved.

Cyclic Voltammetry. Cyclic voltammetry (CV) measurements were performedwith a PAR 263 A potentiostat/galvanostat cycling at various scan rates, usinga one-compartment cell with a glassy carbon working electrode, Pt auxiliaryelectrode, and a silver wire pseudoreference. Ferrocene was used as an inter-nal standard for all electrochemical experiments. The electrolyte for electro-chemistry in 1∶1 CH3CN∶H2O was 0.1 M KNO3, and in CH3CN was 0.1 Mtetrabutylammonium hexafluorophosphate. Due to the air sensitive natureof the complexes, a blanket of Ar was used during the experiments.

Controlled-Potential Coulometry. A custom-built, airtight electrochemical cellwas used. The working electrode and auxiliary electrode were in separatechambers, separated from the solution using Vycor frits. The solution wasdegassed with N2, and CH4 was added as an internal reference. A solutionof p-toluenesulfonic acid (65 mM) in 0.1 M KNO3 in 1∶1 CH3CN∶H2O waselectrolyzed with 0.2 mM catalyst for 1 h at −1.0 V vs. SCE. Aliquots ofthe headspace were sampled at different times during the electrolysis time,and analyzed by GC using a TCD detector. This analysis gave a Faradaicyield greater than 95%. No hydrogen evolution is observed for catalysts1–4 if the coulometry is carried out at potentials that are more positive than−0.5 V vs. SCE. This is inconsistent with a catalytic reaction at this potentialthat is too slow to be detected by cyclic voltammetry.

Determination of pH in Mixed Solvent Systems. All pH values reported in themanuscript were determined by measuring the pH of an aqueous solutionbefore adding acetonitrile (referred to as w

w pH) (42). The pH of the 1∶1CH3CN∶H2O mixture can also be measured directly to give a value for s

wpH.By using a previously outlined procedure (42), a more accurate value of pHcan be determined for the solvent mixture (ss pH) based on the equationδ ¼ s

wpH − ss pH. For example, for the optimal pH reported for the photoca-

talytic system in this text (ww pH ¼ 4.02) gives a measured swpH in 1∶1

CH3CN∶H2O of 4.84, corresponding to a ss pH of 5.1.

Systems with Ascorbic Acid Sacrificial Donors. When using ascorbic acid as asacrificial donor, the net reaction being driven photochemically can beexpressed by the following equation:

H2A !hvA þH2

The formation of A and H2 from ascorbic acid has been determined tobe thermodynamically unfavorable by 0.41 V at pH 4 (32, 43). Thus, photo-catalytic H2 generation from AA represents a photochemically driven upcon-version of approximately 20 kcal∕mol (43).

1. Lewis NS, Nocera DG (2006) Powering the planet: Chemical challenges in solar energyutilization. Proc Nat Acad Sci USA 103:15729–15735.

2. Bard AJ, Fox MA (1995) Artificial photosynthesis—Solar splitting of water to hydrogenand oxygen. Acc Chem Res 28:141–145.

3. Youngblood WJ, et al. (2009) Photoassisted overall water splitting in a visible light-absorbing dye-sensitized photoelectrochemical cell. J Am Chem Soc 131:926–927.

4. Eisenberg R (2009) Rethinking water splitting. Science 324:44–45.5. Bernholc J, Stiefel EI (1986) Comparison of the electronic properties of bis(disulfido)

dioxidi-.mu.-sulfidodimolybdate(2-) and bis(disulfido)disulfidodi-.mu.-sulfidodimolyb-date(2-). Inorg Chem 25:3876–3879.

6. Razavet M, Artero V, Fontecave M (2005) Proton electroreduction catalyzed by coba-loximes: Functional models for hydrogenases. Inorg Chem 44:4786–4795.

7. Jacques PA, Artero V, Pecaut J, Fontecave M (2009) Cobalt and nickel diimine-dioximecomplexes as molecular electrocatalysts for hydrogen evolution with low overvol-tages. Proc Nat Acad Sci USA 106:20627–20632.

8. Berben LA, Peters JC (2010) Hydrogen evolution by cobalt tetraimine catalystsadsorbed on electrode surfaces. Chem Commun 46:398–400.

9. Hu XL, Brunschwig BS, Peters JC (2007) Electrocatalytic hydrogen evolution at lowoverpotentials by cobalt macrocyclic glyoxime and tetraimine complexes. J Am ChemSoc 129:8988–8998.

10. Hu XL, Cossairt BM, Brunschwig BS, Lewis NS, Peters JC (2005) Electrocatalytic hydro-gen evolution by cobalt difluoroboryl-diglyoximate complexes. Chem Commun37:4723–4725.

11. Mejia-Rodriguez R, Chong DS, Reibenspies JH, Soriaga MP, Darensbourg MY (2004) Thehydrophilic phosphatriazaadamantane ligand in the development of H2 productionelectrocatalysts: Iron hydrogenase model complexes. J Am Chem Soc 126:12004–12014.

12. Zhao X, et al. (2002) Catalysis of H-2/D-2 scrambling and other H/D exchange processesby [Fe]-hydrogenase model complexes. Inorg Chem 41:3917–3928.

13. Gloaguen F, Lawrence JD, Rauchfuss TB (2001) Biomimetic hydrogen evolution cata-lyzed by an iron carbonyl thiolate. J Am Chem Soc 123:9476–9477.

14. Schmidt M, Contakes SM, Rauchfuss TB (1999) First-generation analogues of thebinuclear site in the Fe-only hydrogenases: Fe-2(mu-SR)(2)(CO)(4)(CN)(2)(2-). J AmChem Soc 121:9736–9737.

15. Wilson AD, et al. (2006) Hydrogen oxidation and production using nickel-basedmolecular catalysts with positioned proton relays. J Am Chem Soc 128:358–366.

16. Streich D, et al. (2010) High-turnover photochemical hydrogen production catalyzedby a model complex of the [FeFe]-hydrogenase active site. Chem Eur J 16:60–63.

17. Jacques P-A, Artero V, Pecaut J, Fontecave M (2009) Cobalt and nickel diimine-dioximecomplexes as molecular electrocatalysts for hydrogen evolution with low overvol-tages. Cobalt and nickel diimine-dioxime complexes as molecular electrocatalysts forhydrogen evolution with low overvoltages. Proc Nat Acad Sci USA 106:20627–20632.

18. Artero V, Fontecave M (2005) Some general principles for designing electrocatalystswith hydrogenase activity. Some general principles for designing electrocatalysts withhydrogenase activity. Coord Chem Rev 249:1518–1535.

19. Voloshin YZ, Dolganov AV, Varzatskii OA, Bubnov YN (2011) Efficient electrocatalytichydrogen production from Hþ ions using specially designed boron-capped cobaltclathrochelates. Chem Commun 47:7737–7739.

15598 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1120757109 McNamara et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

5, 2

021

20. Connolly P, Espenson JH (1986) Cobalt-catalyzed evolution of molecular-hydrogen.Inorg Chem 25:2684–2688.

21. Dempsey JL, Brunschwig BS, Winkler JR, Gray HB (2009) Hydrogen evolution catalyzedby cobaloximes. Acc Chem Res 42:1995–2004.

22. Dempsey JL, Winkler JR, Gray HB (2010) Kinetics of electron transfer reactions ofh2-evolving cobalt diglyoxime catalysts. J Am Chem Soc 132:1060–1065.

23. Du PW, Schneider J, Luo GG, Brennessel WW, Eisenberg R (2009) Visible light-drivenhydrogen production from aqueous protons catalyzed by molecular cobaloximecatalysts (vol 48, pg 4962, 2009). Inorg Chem 48:8646–8646.

24. Lazarides T, et al. (2009) Making hydrogen from water using a homogeneous systemwithout noble metals. J Am Chem Soc 131:9192–9214.

25. McCormick TM, et al. (2010) Reductive side of water splitting in artificial photosynth-esis: New homogeneous photosystems of great activity and mechanistic insight. J AmChem Soc 132:15480–15483.

26. Probst B, Kolano C, Hamm P, Alberto R (2009) An efficient homogeneous intermole-cular rhenium-based photocatalytic system for the production of H-2. Inorg Chem48:1836–1843.

27. Probst B, Rodenberg A, Guttentag M, Hamm P, Alberto R (2010) A highly stable rhe-nium-cobalt system for photocatalytic H2 Production: Unraveling the performance-limiting steps. Inorg Chem 49:6453–6460.

28. Stiefel EI (2003) Dithiolene chemistry: Synthesis, properties, and applications. ProgInorg Chem 52:1–738.

29. Sproules S, Wieghardt K (2011) Sulfur K-edge X-ray absorption spectroscopy andHarry’s intuition. Coord Chem Rev 255:837–860.

30. McNamara WR, et al. (2011) A cobalt-dithiolene complex for the photocatalytic andelectrocatalytic reduction of protons. J Am Chem Soc 133:15368–15371.

31. Cline ED, Adamson SE, Bernhard S (2008) Homogeneous catalytic system for photo-induced hydrogen production utilizing iridium and rhodium complexes. Inorg Chem47:10378–10388.

32. Brown GM, Chan SF, Creutz C, Schwarz HA, Sutin N (1979)Mechanism of the formationof dihydrogen from the photoinduced reactions of tris(bipyridine)ruthenium(II) withtris(bipyridine)rhodium(III). J Am Chem Soc 101:7638–7640.

33. Bigi JP, Hanna TE, Harman WH, Chang A, Chang CJ (2010) Electrocatalytic reductionof protons to hydrogen by a water-compatible cobalt polypyridyl platform. ChemCommun 46:958–960.

34. Karunadasa HI, Chang CJ, Long JR (2010) A molecular molybdenum-oxo catalyst forgenerating hydrogen from water. Nature 464:1329–1333.

35. Helm ML, Stewart MP, Bullock RM, DuBois MR, DuBois DL (2011) A synthetic nickelelectrocatalyst with a turnover frequency above 100,000 s(1) for H production. Science333:863–866.

36. Sun Y, et al. (2011) Molecular cobalt pentapyridine catalysts for generating hydrogenfrom water. J Am Chem Soc 133:9212–9215.

37. Wieghardt K, Ray K, George SD, Solomon EI, Neese F (2007) Description of theground-state covalencies of the bis(dithiolato) transition-metal complexes fromX-ray absorption spectroscopy and time-dependent density-functional calculations.Chem Eur J 13:2783–2797.

38. Waters T, Wang XB, Woo HK, Wang LS (2006) Photoelectron spectroscopy of the bis(dithiolene) anions [M(mnt)2]n- (M ¼ Fe − Zn; n ¼ 1, 2): Changes in electronic struc-ture with variation of metal center and with oxidation. Inorg Chem 45:5841–5851.

39. Benedito FL, Petrenko T, Bill E, Weyhermuller T, Wieghardt K (2009) Square planarbis{3,6-bis(trimethylsilyl)benzene-1,2-dithiolato}metal complexes of Cr(II), Co(III),and Rh(II): An experimental and density functional theoretical study. Inorg Chem48:10913–10925.

40. Kilgore UJ, et al. (2011) [Ni((P2N2C6H4X)-N-Ph)(2)](2+) complexes as electrocatalystsfor H-2 production: Effect of substituents, acids, and water on catalytic rates. J AmChem Soc 133:5861–5872.

41. Arumugam K, Bollinger JE, Fink M, Donahue JP (2007) Preparation and isolation ofdithiolene thiophosphoryl molecules as stable, protected forms of dithiolene ligands.Inorg Chem 46:3283–3288.

42. Espinosa S, Bosch E, Roses M (2000) Retention of ionizable compounds on HPLC. 5. pHscales and the retention of acids and bases with acetonitrile-water mobile phases.Anal Chem 72:5193–5200.

43. Clark WM (1960) Oxidation-Reduction Potentials of Organic Systems, (Williams andWilkins Co, Baltimore, MD), 469.

McNamara et al. PNAS ∣ September 25, 2012 ∣ vol. 109 ∣ no. 39 ∣ 15599

CHEM

ISTR

YSP

ECIALFEAT

URE

Dow

nloa

ded

by g

uest

on

Aug

ust 1

5, 2

021