Electrocatalytic hydrogen evolution at low overpotentials by cobalt macrocyclic glyoxime and tetraimine complexes

Article abstract of DOI:10.1021/ja067876b

Cobalt complexes supported by diglyoxime ligands of the type Co(dmgBF ₂)₂(CH₃CN)₂ and Co(dpgBF ₂)₂(CH₃CN)₂ (where dmgBF₂ is difluoroboryl-dimethylglyoxime and dpgBF₂ is difluoroboryl- diphenylglyoxime), as well as cobalt complexes with [14]-tetraene-N₄ (Tim) ligands of the type [Co(Tim^R)X₂]ⁿ⁺ (R = methyl or phenyl, X = Br or CH₃CN; n = 1 with X = Br and n = 3 with X = CH₃CN), have been observed to evolve H₂ electrocatalytically at potentials between -0.55 V and -0.20 V vs SCE in CH ₃CN. The complexes with more positive Co(II/I) redox potentials exhibited lower activity for H₂ production. For the complexes Co(dmgBF₂)₂(CH₃CN)₂, Co(dpgBF ₂)₂(CH₃CN)₂, [Co(Tim ^Me)Br₂]Br, and [Co(Tim^Me)(CH ₃CN)₂](BPh₄)₃, bulk electrolysis confirmed the catalytic nature of the process, with turnover numbers in excess of 5 and essentially quantitative faradaic yields for H₂ production. In contrast, the complexes [Co(Tim^Ph/Me)Br₂]Br and [Co(Tim^Ph/Me)(CH₃CN)₂](BPh₄) ₃ were less stable, and bulk electrolysis only produced faradaic yields for H₂ production of 20-25%. Cyclic voltammetry of Co(dmgBF₂)₂(CH₃CN)₂, [Co(Tim ^Me)Br₂]⁺, and [Co(Tim^Me)(CH ₃CN)₂]³⁺ in the presence of acid revealed redox waves consistent with the Co(III)-H/Co(II)-H couple, suggesting the presence of Co(III) hydride intermediates in the catalytic system. The potentials at which these Co complexes catalyzed H₂ evolution were close to the reported thermodynamic potentials for the production of H₂ from protons in CH₃CN, with the smallest overpotential being 40 mV for Co(dmgBF ₂)₂(CH₃CN)₂ determined by electrochemistry. Consistent with this small overpotential, Co(dmgBF ₂)₂(CH₃CN)₂ was also able to oxidize H₂ in the presence of a suitable conjugate base. Digital simulations of the electrochemical data were used to study the mechanism of H₂ evolution catalysis, and these studies are discussed.

Full text of DOI:10.1021/ja067876b

Published on Web 06/28/2007
Electrocatalytic Hydrogen Evolution at Low Overpotentials by
Cobalt Macrocyclic Glyoxime and Tetraimine Complexes
Xile Hu, Bruce S. Brunschwig, and Jonas C. Peters*
Contribution from the Arnold and Mabel Beckman Laboratories of Chemical Synthesis and the
Beckman Institute, DiVision of Chemistry and Chemical Engineering, California Institute of
Technology, MC 127-72, Pasadena, California 91125
Received November 3, 2006; E-mail: jpeters@caltech.edu
Abstract: Cobalt complexes supported by diglyoxime ligands of the type Co(dmgBF2)2(CH3CN)2 and Co-
(dpgBF2)2(CH3CN)2 (where dmgBF2 is difluoroboryl-dimethylglyoxime and dpgBF2 is difluoroboryl-diphe-
nylglyoxime), as well as cobalt complexes with [14]-tetraene-N4 (Tim) ligands of the type [Co(TimR)X2]n+
(R ) methyl or phenyl, X ) Br or CH3CN; n ) 1 with X ) Br and n ) 3 with X ) CH3CN), have been
observed to evolve H2 electrocatalytically at potentials between -0.55 V and -0.20 V vs SCE in CH3CN.
The complexes with more positive Co(II/I) redox potentials exhibited lower activity for H2 production. For
the complexes Co(dmgBF2)2(CH3CN)2, Co(dpgBF2)2(CH3CN)2, [Co(TimMe)Br2]Br, and [Co(TimMe)(CH3CN)2]-
(BPh4)3, bulk electrolysis confirmed the catalytic nature of the process, with turnover numbers in excess of
5 and essentially quantitative faradaic yields for H2 production. In contrast, the complexes [Co(TimPh/Me)Br2]Br
and [Co(TimPh/Me)(CH3CN)2](BPh4)3 were less stable, and bulk electrolysis only produced faradaic yields
for H2 production of 20-25%. Cyclic voltammetry of Co(dmgBF2)2(CH3CN)2, [Co(TimMe)Br2]+, and [Co-
(TimMe)(CH3CN)2]3+ in the presence of acid revealed redox waves consistent with the Co(III)-H/Co(II)-H
couple, suggesting the presence of Co(III) hydride intermediates in the catalytic system. The potentials at
which these Co complexes catalyzed H2 evolution were close to the reported thermodynamic potentials for
the production of H2 from protons in CH3CN, with the smallest overpotential being 40 mV for
Co(dmgBF2)2(CH3CN)2 determined by electrochemistry. Consistent with this small overpotential,
Co(dmgBF2)2(CH3CN)2 was also able to oxidize H2 in the presence of a suitable conjugate base. Digital
simulations of the electrochemical data were used to study the mechanism of H2 evolution catalysis, and
these studies are discussed.
hydrogenases3 or on metal complexes having simple macrocyclic
ligands,4-6 catalyze H2 evolution in organic solvents at fairly
negative potentials, from ∼ -1.1 to -2 V vs a saturated calomel
electrode (SCE).7
A select number of molecular systems reported to date show
promise for H2 evolution at more positive potentials. DuBois
et al. have shown that a Cp2Mo2S2 system catalyzes electro-
chemical H2 evolution at -0.26 V vs SCE in CH3CN using
p-cyanoanilinium as the acid source, and a [Ni(PPh2NPh2)(CH3-
1. Introduction
The development of robust hydrogen evolution catalysts that
are made of inexpensive and earth abundant materials is of
considerable current interest.1 Fe-Ni and Fe-Fe hydrogenases
produce dihydrogen in water at close to the thermodynamic
potential of -0.41 V vs the normal hydrogen electrode (NHE)
at pH ) 7.2 However, most known molecular H2 evolution
catalysts, based either on the structure of the active site of
(1) Gratzel, M. Acc. Chem. Res. 1981, 14, 376-384. Bard, A. J.; Fox, M. A.
Acc. Chem. Res. 1995, 28, 141-145. Khaselev, O.; Turner, J. A. Science
1998, 280, 425-427. Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B. Science
2002, 297, 2243-2245. Zou, Z. G.; Ye, J. H.; Sayama, K.; Arakawa, H.
Nature 2001, 414, 625-627. Maeda, K.; Teramura, K.; Lu, D. L.; Takata,
T.; Saito, N.; Inoue, Y.; Domen, K. Nature 2006, 440, 295-295. Dempsey,
J. L.; Esswein, A. J.; Manke, D. R.; Rosenthal, J.; Soper, J. D.; Nocera, D.
G. Inorg. Chem. 2005, 44, 6879-6892. Lewis, N. S.; Nocera, D. G. Proc.
Natl. Acad. Sci. U.S.A. 2006, 103, 15729-15735. US DOE Hydrogen
Program: http://www.hydrogen.energy.gov/; “Basic Research Needs for
the Hydrogen Economy”; available at http://www.sc.doe.gov/bes/reports/
files/NHE_rpt.pdf; Powering the Planet: A Chemical Bonding Center.
Website: http://www.caltechmitsolarpower.caltech.edu/.
(2) Adams, M. W. W. Biochim. Biophys. Acta 1990, 1020, 115-145. Frey,
M. In Structure and Bonding; Hill, H. A. O., Sadler, P. J., Thomson, A. J.,
Eds; Springer-Verlag: Berlin, New York, 1997; Vol. 90, pp 97-126.
Adams, M. W. W.; Stiefel, E. I. Science 1998, 282, 1842-1843. Fontecilla-
Camps, J. C.; Ragsdale, S. W. In AdVances in Inorganic Chemistry; Sykes,
A. G., Ed.; Academic Press: 1999; Vol 47, pp 283-333. Peters, J. W.;
Lanzilotta, W. N.; Lemon, B. J.; Seefeldt, L. C. Science 1998, 282, 1853-
1858. Darensbourg, M. Y.; Lyon, E. J.; Smee, J. J. Coord. Chem. ReV.
2000, 206, 533-561.
(3) Tard, C.; Liu, X. M.; Ibrahim, S. K.; Bruschi, M.; De Gioia, L.; Davies, S.
C.; Yang, X.; Wang, L. S.; Sawers, G.; Pickett, C. J. Nature 2005, 433,
610-613. Gloaguen, F.; Lawrence, J. D.; Rauchfuss, T. B. J. Am. Chem.
Soc. 2001, 123, 9476-9477. Darensbourg, M. Y.; Lyon, E. J.; Zhao, X.;
Georgakaki, I. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 3683-3688.
Ott, S.; Kritikos, M.; Akermark, B.; Sun, L. C.; Lomoth, R. Angew. Chem.,
Int. Ed. 2004, 43, 1006-1009.
(4) Fisher, B.; Eisenberg, R. J. Am. Chem. Soc. 1980, 102, 7361-7363. Beley,
M.; Collin, J. P.; Ruppert, R.; Sauvage, J. P. J. Chem. Soc., Chem. Commun.
1984, 1315-1316. Collin, J. P.; Jouaiti, A.; Sauvage, J. P. Inorg. Chem.
1988, 27, 1986-1990. Efros, L. L.; Thorp, H. H.; Brudvig, G. W.; Crabtree,
R. H. Inorg. Chem. 1992, 31, 1722-1724. Koelle, U. New J. Chem. 1992,
16, 157-169. Collman, J. P.; Wagenknecht, P. S.; Lewis, N. S. J. Am.
Chem. Soc. 1992, 114, 5665-5673. Grass, V.; Lexa, D.; Saveant, J. M. J.
Am. Chem. Soc. 1997, 119, 7526-7532.
(5) Collman, J. P.; Ha, Y. Y.; Wagenknecht, P. S.; Lopez, M. A.; Guilard, R.
J. Am. Chem. Soc. 1993, 115, 9080-9088.
(6) Bhugun, I.; Lexa, D.; Saveant, J. M. J. Am. Chem. Soc. 1996, 118, 3982-
3983.
(7) Unless otherwise noted, all potentials are referenced against SCE. NHE vs
SCE ) -0.24 V in aqueous solution; ferrocenium/ferrocene vs SCE )
0.38 V in acetonitrile.
9
8988
J. AM. CHEM. SOC. 2007, 129, 8988-8998
10.1021/ja067876b CCC: $37.00 © 2007 American Chemical Society

H2 Evolution by Glyoxime and Tetraimine Complexes
A R T I C L E S
Scheme 1. Cobalt Difluoroboryl-Diglyoximate Complexes 1-3
CN)]2+ system catalyzes H2 evolution at -0.46 V vs SCE using
triflic acid as the acid source.8,9 We and others have shown that
Co(diglyoxime) complexes evolve H2 at relatively positive
potentials either catalytically or electrocatalytically.10-12 For
example, Co(dmgBF2)2(CH3CN)2 (1, dmgBF2 ) difluoroboryl-
dimethylglyoxime)11 catalyzes H2 evolution in acidic CH3CN
at -0.55 V vs SCE using CF3COOH or HCl as the acid source,
whereas Co(dpgBF2)2(CH3CN)2 (2, dpgBF2 ) difluoroboryl-
diphenylglyoxime)13 produces H2 at even higher potentials
(-0.28 V vs SCE) in acidic CH3CN using HCl or HBF4‚Et2O
as the acid source (Scheme 1). Similarly, Artero et al. have
shown that [Co(dmgH)2PyCl](dmgH ) dimethylglyoxime)
catalyzes H2 evolution at -1.03 V vs SCE in acidic DMF
solutions using Et3NHCl as the acid source.12
In this work, we report the electrochemical and electrocata-
lytic behavior of a series of cobalt complexes having BF2-
bridged diglyoxime or propane-bridged macrocyclic tetraimine
ligands. The behavior of this series of complexes has allowed
for the formulation of a correlation between the activity of H2
evolution and the CoII/I reduction potentials of the various
complexes. Electrochemical evidence for Co(III) hydride in-
termediates is presented, and a plausible mechanism of the H2
evolution reaction is discussed. Furthermore, complex 1, Co-
(dmgBF2)2(CH3CN)2, is shown to be a molecular catalyst able
to mediate both H2 evolution and oxidation, confirming its low
overpotential for H2 evolution. To the best of our knowledge,
this is the first such demonstration to be reported for a synthetic
catalyst.
catalytic reaction proceeded sufficiently rapidly that the con-
centration of acid was attenuated close to the electrode and that
the current was controlled, at least partially, by diffusion of acid
to the electrode surface.14-16 At higher acid concentrations (e.g.,
9 mM), the catalytic wave approached a plateau shape. The
potential at which catalysis occurred did not change as the acid
concentration was varied. The cyclic voltammetry of a 0.6 mM
solution of 2 (glassy carbon electrode, 0.1 M [nBu4N][ClO4],
CH3CN) was similar to that of 1 (Figure 2), indicating that 2 is
also an electrocatalyst for H2 evolution.
In the absence of a catalyst, the direct reduction of acid at a
glassy carbon electrode in solutions containing tosic acid took
place at potentials more negative than -1.0 V vs SCE, and the
background current was found to be negligible at the potentials
where catalysis occurred using catalyst 1 or 2. The rise in the
catalytic current at -0.6 V vs SCE in Figure 2 is due to the
tail of the catalytic wave originating from a Co(III)-H
intermediate (vide infra).
2.2. Structural Characterization of Complexes 2 and 3.
The solid-state molecular structure of the Co(II) form of catalyst
2 was determined by X-ray crystallography (Figure 3). The
refined structure reveals a six-coordinate cobalt(II) ion that is
ligated by a planar difluoroboryl-diglyoxime ligand and by two
trans-oriented acetonitrile molecules. The cobalt ion is located
on a crystallographically defined inversion center. The observed
axial Co-Nnitrile distance of 2.241(3) Å is much longer than
the 1.886(5) Å averaged Co-Nimine distance in the glyoxime
plane.
The Co(I) form of the complexes can be conveniently
synthesizedbychemicalmethods.Forinstance,[Co(dpgBF2)2(CH3CN)]-
(3) was synthesized by reducing 2 with 1 equiv of Na/Hg or
Co(Cp)2 in CH3CN. The refined solid-state molecular structure
of 3 (Figure 4) contains a five-coordinate cobalt(I) center in a
distorted square-pyramidal ligand environment. The Co-Nnitrile
distance is 1.973(2) Å for 3, 0.268 Å shorter than that in 2.
Because of this strong Co-Nnitrile interaction, the cobalt ion is
displaced 0.27 Å toward the nitrile ligand from the plane defined
by the macrocyclic imine ligand. The average Co-Nimine
distance of 1.851(6) Å for 3 is slightly shorter (0.035 Å) than
that in 2, consistent with expectations for a higher degree of
metal to ligand π-backbonding.
2. Results
2.1. Electrocatalytic H2 Evolution by Cobalt Difluorobo-
ryl-Diglyoximate Complexes 1 and 2. Figure 1 shows the
cyclic voltammogram (CV) of 1 (0.3 mM, 0.1 M [nBu4N][ClO4],
CH3CN) at a glassy carbon electrode. The reversible redox
process shown at -0.55 V vs SCE in the absence of acid was
assigned to the CoII/I couple. Upon addition of the monohydrate
of toluenesulfonic acid (tosic acid, TsOH‚H2O), a catalytic wave
was observed at a potential close to the formal reduction
potential, E°′(CoII/I), for 1. At low acid concentrations, the
catalytic wave exhibited a peak-like shape, indicating that the
(8) Appel, A. M.; DuBois, D. L.; DuBois, M. R. J. Am. Chem. Soc. 2005,
127, 12717-12726.
2.3. Electrocatalytic H2 Evolution by Cobalt Macrocyclic
Tetraimine Complexes 4a-6b. Complexes 4a-6b17-20 (Scheme
(9) Wilson, A. D.; Newell, R. H.; McNevin, M. J.; Muckerman, J. T.; DuBois,
M. R.; DuBois, D. L. J. Am. Chem. Soc. 2006, 128, 358-366.
(10) Hu, X. L.; Cossairt, B. M.; Brunschwig, B. S.; Lewis, N. S.; Peters, J. C.
Chem. Commun. 2005, 4723-4725.
(14) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and
Applications, 2nd ed.; John Wiley & Sons, Inc.: New York, 2001.
(15) Andrieux, C. P.; Blocman, C.; Dumasbouchiat, J. M.; Mhalla, F.; Saveant,
J. M. J. Electroanal. Chem. 1980, 113, 19-40.
(16) Saveant, J. M.; Su, K. B. J. Electroanal. Chem. 1984, 171, 341-349.
(17) Tait, A. M.; Busch, D. E. In Inorganic Syntheses; Douglas, B. E., Ed.;
Wiley-Interscience: New York, 1978; Vol. XVIII, pp 22-26.
(11) Connolly, P.; Espenson, J. H. Inorg. Chem. 1986, 25, 2684-2688.
(12) Artero, V.; Fontecave, M. Coord. Chem. ReV. 2005, 249, 1518-1535;
Razavet, M.; Artero, V.; Fontecave, M. Inorg. Chem. 2005, 44, 4786-
4795.
(13) Tovrog, B. S.; Kitko, D. J.; Drago, R. S. J. Am. Chem. Soc. 1976, 98,
5144-5153.
9
J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007 8989

A R T I C L E S
Hu et al.
Figure 1. Cyclic voltammogram of complex 1 in CH3CN solution
containing 0.1 M [nBu4N][ClO4] in the presence of tosic acid: 0.3 mM 1
and no acid (black), 1.5 mM TsOH‚H2O (red), 4.5 mM TsOH‚H2O (blue),
and 9 mM TsOH‚H2O (green). Scan rate: 100 mV/s; glassy carbon
electrode.
Figure 4. Solid-state structure of the anion of 3‚Na(12-crown-4)2. Hydro-
gen atoms, the Na(12-crown-4)2+ countercation, and cocrystallization solvent
molecules have been omitted for clarity; thermal ellipsoids are displayed
at the 50% probability. Selected bond lengths (Å) and angles (deg): Co1-
N1, 1.853(2); Co1-N2, 1.846(2); Co1-N3, 1.859(2); Co1-N4, 1.846(2);
Co1-N5, 1.973(2); N1-Co1-N2, 80.83(8); N2-Co1-N4, 95.59(8); N3-
Co1-N4, 81.13(8); N3-Co1-N1, 95.34(8); N1-Co1-N5, 96.54(8); N2-
Co1-N5, 102.16(8); N3-Co1-N5, 103.44(8); N4-Co1-N5, 98.39(8).
Scheme 2. Cobalt Complexes of [14]-Tetraene-N4 Ligands 4a-6b
Figure 2. Cyclic voltammogram of 0.6 mM 2 in CH3CN solution
containing 0.1 M [nBu4N][ClO4] in the absence (black) and presence (red)
of 9 mM TsOH‚H2O. Scan rate: 100 mV/s; glassy carbon electrode.
V for 5b, -0.15 V for 6a, and -0.08 V for 6b. The value of
E°′(CoII/I) shifted by ∼60 mV as a result of each substitution
of a methyl group on the Tim ligand by the more electron-
withdrawing phenyl group. This potential shift is comparable
to the ∼65 mV potential shift that results from the substitution
of phenyl for methyl in the cobalt difluoroboryl-diglyoxime
complexes, 1 vs 2. Substitution of two anionic bromide ligands
by two neutral acetonitrile ligands shifted the potential by 20,
50, and 70 mV for 4a to 4b, 5a to 5b, and 6a to 6b, respectively.
The axial bromide ligands in 4a, 5a, and 6a were not replaced
by acetonitrile upon dissolution in acetonitrile, as such solutions
remained green, indicative of the bromide complexes, rather
than turning pale-yellow indicative of the acetonitrile adduct
complexes. All bromide complexes exhibited reversible CoIII/II
Figure 3. Solid-state molecular structure of the anion of 2. Hydrogen atoms
and two CH2Cl2 molecules of cocrystallization have been omitted for clarity;
thermal ellipsoids are displayed at the 50% probability. Selected bond
lengths (Å) and angles (deg): Co1-N1, 1.882(3); Co1-N2, 1.889(3); Co1-
N3, 2.241(3); N1-Co1-N2, 98.33(19); N2-Co1-N1a, 81.78(12); N1-
Co1-N3, 98.33(12); N2-Co1-N3, 81.63(12).
2) each exhibited reversible cyclic voltammetric behavior for
the CoII/I redox couple, with E°′(CoII/I) values vs SCE in CH3-
CN of -0.38 V for 4a, -0.35 V for 4b, -0.25 V for 5a, -0.20
(18) Welsh, W. A.; Reynolds, G. J.; Henry, P. M. Inorg. Chem. 1977, 16, 2558-
2561.
(19) Kildahl, N. K.; Antonopoulos, G. J. Coord. Chem. 1986, 14, 293-307.
(20) Eggleston, D. S.; Jackels, S. C. Inorg. Chem. 1980, 19, 1593-1599.
9
8990 J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007

H2 Evolution by Glyoxime and Tetraimine Complexes
A R T I C L E S
4a-6b when the weak acid CF3COOH (pKa ) 12.7)22,23 was
used as the acid source.
2.4. Stability of Co Macrocyclic Tetraimine Complexes
in the Presence of Acid. The stability of 4a-5b in the presence
of excess acid was evaluated by electronic absorption spectros-
copy (monitored at ∼400 nm for 4a and 5a, and ∼270 nm for
4b and 5b). These complexes showed negligible degradation
over a period of several days in the presence of a 25-fold excess
of TsOH‚H2O (10 mM), suggesting that these complexes are
more acid tolerant than 1 and 2. Under the same conditions, 1
degraded ∼10% in 1 h and 2 degraded ∼10% in 4 h. Complexes
4a-5b degraded in the presence of excess HBF4‚Et2O but more
slowly than 1 and 2. For example, 4a and 4b degraded only
5% in 2 h in the presence of a 30-fold excess of HBF4‚Et2O (3
mM), while under the same conditions 1 and 2 degraded
completelywithinminutes.10TheCo(II)complexes[CoII(TimMe)(CH3-
CN)2]2+ (4c) and [CoII(TimMe/Ph)(CH3CN)2]2+ (5c) had stabilities
that were similar to those for 4b and 5b, respectively.
Figure 5. Cyclic voltammogram of complex 4a in CH3CN solutions
containing 0.1 M [nBu4N][ClO4] in the presence of acid: 0.15 mM 1 and
no acid (black), 1.5 mM TsOH‚H2O (red), and 9 mM TsOH‚H2O (blue).
Scan rate: 100 mV/s; glassy carbon electrode.
couples (Table 2), while all acetonitrile complexes exhibited
no redox waves for their CoIII/II couples between 0.3 V to -1
V vs SCE. This suggests that the bromide ligands of complexes
4a, 5a, and 6a are not substituted by acetonitrile even in their
Co(II) form on the time scale of the CV measurements.
Figure 5 shows the CV of 4a. In CH3CN with 0.1 M [nBu4N]-
[ClO4], reversible reductions were observed for the CoIII/II couple
at -0.06 V vs SCE and for the CoII/I couple at -0.38 V vs
SCE. Addition of TsOH‚H2O to an unstirred solution of 0.15
mM 4a (0.1 M [nBu4N][ClO4] in CH3CN) produced a catalytic
wave at a potential close to the position of E°′(CoII/I) for 4a. At
low acid concentrations, the catalytic wave had a peak-like
shape, but for higher acid concentrations the current approached
a plateau at potentials more negative than E°′(CoII/I). The
potential at which catalysis took place remained nearly constant
with increasing acid concentration. No catalytic waves were
observed at potentials near the E°′(CoIII/II), consistent with the
fact that this potential was significantly more positive than the
thermodynamic potential for hydrogen evolution using tosic acid
as the proton source (vide infra). Similar behavior was observed
when HBF4‚Et2O was used as the proton source (Figure S1,
Supporting Information). Hence, complex 4a is a catalyst for
electrochemical H2 evolution in acidic CH3CN solutions, and
the catalysis is associated with reduction of the Co(II) complex
to its Co(I) state.
Complex 4b was also found to catalyze electrochemical H2
evolution in acidic CH3CN solutions, when either TsOH‚H2O
or HBF4‚Et2O was the acid. As for 4a, the CV (Figure S2)
showed that the catalysis by 4b occurred at the CoII/I potential.
The CV of 5a and 5b in CH3CN solutions did not vary upon
addition of TsOH‚H2O. However, catalytic proton reduction was
observed when HBF4‚Et2O (pKa ) 0.1)21,22 was used as the
proton source (Figure S3). This behavior suggests that the proton
activity obtained with TsOH‚H2O (pKa ) 8.0)22,23 is not high
enough to produce dihydrogen from the Co(I) forms of
complexes 5a and 5b. Complexes 6a and 6b were also probed
for electrochemical hydrogen evolution, but no catalysis was
observed using either TsOH‚H2O or HBF4‚Et2O as the acid
source. In addition, no catalysis was observed for complexes
2.5. Catalytic Efficiency of Co Tetraimine Complexes for
H2 Evolution as Determined by Bulk Electrolysis. Previous
bulk electrolysis experiments indicated that the catalytic currents
observed by cyclic voltammetry for complexes 1 and 2 were
associated with electrocatalytic evolution of H2, with ap-
proximately quantitative faradaic yields for H2 production.
Turnover numbers (TNs) were >5 (TN ) moles of hydrogen
produced divided by moles of catalyst used) were observed,
without appreciable decomposition of the catalysts.10 To evaluate
the behavior of the Co tetraimine complexes, analogous bulk
electrolysis experiments were performed with solutions of 4a-
5b (Table 1). During bulk electrolysis potentials were applied
at -0.58 V (for 4a-b) or -0.48 V (for 5a-b) vs SCE.
Solutions that contained an excess of acid and catalysts 4a-5b
turned from green to red-purple upon electrolysis. This behavior
suggested that under such conditions the catalyst was reduced
toitsCo(II)form,[CoII(TimMe)(CH3CN)2]2+(4c)or[CoII(TimMe)(CH3-
CN)2]2+ (5c).19 The CV of 4c and 5c was identical to that of
4b and 5b.
When a cobalt complex having a TimMe ligand (4a-4c) was
used in bulk electrolysis experiments, H2 was produced (as
determined by GC-MS) in quantitative faradaic yields (Table
1). After five turnovers, no deactivation of the catalyst was
observed. However, when a cobalt complex having the TimMe/Ph
ligand (5a-5c) was used as the catalyst, the faradaic yield for
H2 production was only 20-25%. After two turnovers, ap-
preciable deactivation of the catalyst was observed. The lower
faradaic yields observed for 5a-5c most likely reflect the
decomposition of intermediate species during the course of
electrolysis, rather than the acid sensitivity of the precatalysts,
since 5a-5c are in fact somewhat more stable in the presence
of an excess of acid than are 4a-4c.
2.6. Kinetic and Mechanistic Analysis. CV data for
complexes 1, 2, 4a, and 4b were obtained as a function of acid
and catalyst concentrations.24 The CV of the catalysts, with no
added acid, had well-defined reversible CoII/I reductions (peak
separations of 80 ( 10 mV). The potentials for all the complexes
are shown in Table 2. The diffusion coefficient for each catalyst,
(21) Assuming etherated HBF4 completely dissociates into etherated proton and
-
BF4 in acetonitrile, its pKa roughly equals the pKa for diethyl ether (as
base), which is 0.1. See ref 22.
(24) Due to the instability of 1 and 2 in the presence of a large excess of HBF4-
(Et2O), and due to the nonunity faradaic yield for H2 evolution by 5a and
5b in HBF4(Et2O), the studies were performed using complexes 1, 2, 4a,
and 4b and using TsOH(H2O) as the acid.
(22) Izutsu, K. Acid-Base Dissociation Constants in Dipolar Aprotic SolVents;
Blackwell Scientific Publications: Oxford, 1990.
(23) All pKa values cited in this paper were referred to pKa’s in acetonitrile.
9
J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007 8991

A R T I C L E S
Hu et al.
Table 1. Summary of Bulk Electrolysis Results
concn of
concn of
acid
faradaic
yield for
2 (%)
proton
source
catalyst
(mM)
volume
(mL)
potential
duration
(min)
charge
complex
(mM)
(V vs SCE)
(coulomb)
H
4a
4b/4c
5a
TsOH‚H2O
TsOH‚H2O
HBF4‚Et2O
HBF4‚Et2O
0.9
0.8
0.9
0.7
21
22
20
20
100
100
100
100
-0.58
-0.58
-0.48
-0.48
30
30
30
30
49
54
15
16
∼90
∼100
∼20
5b/5c
∼25
Table 2. Reduction Potentials, Apparent Rate Constants, and
Diffusion Constants for Cobalt Complexes
E
°′(Co(III/II)
E
°′(Co(II/I))
kapp
×
102d
D ×
105 e
cm2/s
1
complex
(V vs SCE)
(V vs SCE)
(M-1 s-
)
1
2
∼0.2a
∼0.3a
-0.06b
NOc
-0.55
-0.28
-0.38
-0.35
-0.25
-0.20
-0.15
-0.08
7 × 10e
0.8
0.14
2.5
2
4a
4b
5a
5b
6a
6b
4
3
1.2
-0.03b
NOc
NA
NA
NA
NA
0.03b
NOc
c
a Irreversible couple. bReversible couple. Not observed. d TsOH‚H2O
was used as the proton source; rates obtained from eq 1, n ) 2, S ) 0.07
cm2. eDiffusion constant determined by simulation of CV in absence of acid.
determined by simulation of its CV in the absence of acid, is
shown in Table 2.
The overall rate constant for H2 evolution may be estimated
from the height of the catalytic plateau current at high
overpotentials.14,15 A plateau shape is only observed when the
rate of the reduction of the catalytic species at the electrode is
equal to the rate of its reoxidation by the acid. Additionally,
the acid concentration in the reaction region must not be
depleted.14,15 Under these conditions, for a reaction that is first
order in both catalyst and acid, the value of the current in the
plateau region is given by
ic ) nFSCp°(DkCs°)1/2
(1)
Figure 6. Dependence of the catalytic plateau current, ic, on the concentra-
tions of catalyst (top) and acid (bottom) for complex 1; TsOH‚H2O was
used as the acid, with a scan rate of 100 mV/s and S ) 0.07 cm2. In the
plots the plateau current was set to zero in the absence of acid, as predicted
by eq 1, even though the catalyst has a nonzero peak current under such
conditions.
where n is the number of electrons, ic is the plateau current
(mA), F is Faraday’s constant (96 485 C mol-1), S is the area
of the electrode surface (cm2), Cp° is the bulk concentration of
redox active species (M), D is the diffusion coefficient (cm2
s-1), k is the reaction rate constant (M-1 s-1), and Cs° is the
bulk concentration of acid (M).14,15
The dependence of ic on the concentrations of catalyst and
acid was measured for the complexes. For H2 evolution
catalyzed by 1 or 4a using TsOH‚H2O as the acid, the catalytic
waves did not exhibit, but approached, a plateau shape under
certain conditions. The peak values of such catalytic currents
were taken as approximations for the values of the plateau
currents. As shown in Figure 6, for catalysis by 1, the catalytic
rate was found to be first order in the concentration of acid and
less than first order in the concentration of catalyst. For catalysis
by 4a, the catalytic rate was found to be first order in the
concentrations of both acid and catalyst (Figure S4).
The catalytic waves reached a plateau shape using 2 or 4b
as the H2 evolution catalyst when the concentration of TsOH‚H2O
was much larger than that of the catalyst. In these cases, the
catalytic rate followed a first-order dependence on the concen-
tration of both acid and catalyst (Figures S5-S6). Approximate
catalytic rate constants, kapp, with TsOH‚H2O were obtained
using eq 1 (Table 2).
Digital simulations of the cyclic voltammetric data were
performed to study the mechanism of electrocatalytic H2
evolution for complexes 1, 2, 4a, and 4b. The CV was simulated
assuming either a monometallic or a bimetallic mechanism
(Scheme 3) for H2 formation.
For catalysis by 1, when a monometallic mechanism (eqs
2-6) was assumed, it was not possible to use a common set of
parameters to simulate the CV for a given concentration of
catalyst in the presence of incremental amounts of TsOH‚H2O.
The simulated currents increased with acid (TsOH‚H2O) much
more rapidly than observed experimentally. In addition, the
simulated CV maintained its peak-type shape at all concentra-
tions of acid used experimentally, while the experimental CV
approached a plateau shape as the acid concentration increased.
Hence, electrocatalytic H2 evolution by 1 is not dominated by
a simple monometallic mechanism.
Simulation of the CV for catalyst 1 assuming a bimetallic
mechanism (eqs 2-4 and 7) gave rise to a more satisfactory
simulation of the experimental data for a given concentration
9
8992 J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007

H2 Evolution by Glyoxime and Tetraimine Complexes
A R T I C L E S
Scheme 3. Two Mechanistic Pathways for Electrochemical Proton Reduction Catalyzed by Cobalt Macrocyclic Complexes
of catalyst in the presence of incremental amounts of TsOH‚
H2O (Figures S7 and S8). However, for different concentrations
of catalysts, the rate constants obtained were different. At lower
catalyst concentration larger rate constants were obtained ([1]
was observed at this reduction potential. In the absence of acid,
no such wave was observed. We tentatively assign this wave
to the “Co(III)-H”/“Co(II)-H” redox couple produced by the
putative [HCo(CH3CN)(dmgBF2)2] intermediate formed in situ
upon reduction of 1 in the presence of acid (see discussion).
The electrochemistry of course cannot distinguish between a
cobalt or ligand centered reduction of the Co(III)-H.
No reversible redox wave for a Co(III)-H/Co(II)-H couple
was observed for 2. Instead, a catalytic wave was observed at
a potential of -0.85 V vs SCE, i.e., 570 mV more negative
than the value of E°′(CoII/I) for 2 (Figure S15).
Redox waves for the Co(III)-H/Co(II)-H couple were also
observed for 4a-b (Figure S16). In contrast, for complexes
5a-b in the absence of acid, at potentials close to those expected
for the Co(III)-H/Co(II)-H couples, redox waves were ob-
served. This reduction of a Co(I) species is likely a ligand-
centered redox process. The intermediacy of a Co(III)-H could
therefore not be ascertained from the cyclic voltammetric data
for these two complexes.
2.8. Hydrogen Oxidation. We have studied the cobalt
complexes 1 and 2 to determine their ability to electrocatalyti-
cally oxidize hydrogen under 1 atm of H2 in CH3CN solutions
buffered with either [nBu4N][CF3CO2] (for 1) or [nBu4N][TsO]
(for 2). No oxidation activity could be detected at scan rates
between 100 and 1000 mV per second near the Co(II/I)
potential; however, we observed that H2 was oxidized by 1 in
the presence of [nBu4N][CF3CO2] in CH3CN over a period of
several days, reaching an equilibrium mixture represented by
eq 8 (Figure 8).
) 0.22 mM; k4 ) 1.5 × 105 M-1 s-1, k7 ) 1.5 × 106 M-1 s-1
)
than at higher catalyst concentration ([1] ) 1.0 mM; k4 ) 7 ×
103 M-1 s-1, k7 ) 9 × 105 M-1 s-1) (Figures S7 and S8).
The changes in calculated rate constants with catalyst
concentration are consistent with Figure 6, which suggests that
the catalysis by 1 is less than first order in catalyst. Therefore
the CV data indicate that the production of H2 most likely
proceeds by a bimetallic pathway; however, direct kinetic
evidence to support this conclusion is needed.
For catalysis by 4a, a monometallic mechanism was also ruled
out because simulations indicated that a monometallic pathway
would give rise to catalytic waves for the oxidation of H2 by
Co(III) (step 5) at the CoIII/II potential. These oxidative catalytic
currents were not observed in our experiments. On the other
hand, a bimetallic mechanism with the same set of rate constants
can be used to simulate the catalytic waves from two different
concentrations of catalyst in the presence of incremental amounts
of acid (Figures S9 and S10).
For catalysis by 2 and 4b, the catalytic waves can be crudely
simulated using either a monometallic or a bimetallic mecha-
nism. Simulation itself therefore does not distinguish between
these two mechanisms without prior knowledge of the equilib-
rium constants and/or reaction rates of the key chemical steps
(Figures S11-S14).
2.7. Evidence for a Cobalt(III) Hydride Intermediate. In
the presence of TsOH‚H2O, a quasi-reversible redox wave for
complex 1 was observed when the potential was scanned
negative of E°′(CoII/I) (Figure 7); however, no catalytic wave
[CoII(dmgBF2)2(CH3CN)2] + 1/2H2 +
[NBu4][CF3CO2] T CF3COOH +
[CoI(dmgBF2)2(CH3CN)][NBu4] + CH3CN (8)
As shown in Figure 8, when 1 atm of H2 was initially added
to a solution of 1.6 mM 1 in the presence of 66 mM [NBu4][CF3-
CO2], the solution maintained its yellow color, indicative of
the [CoII(dmgBF2)2(CH3CN)2] species in solution. With time,
the solution gradually turned dark blue, indicative of the
formation of [CoI(dmgBF2)2(CH3CN)]- in the reaction medium.
After 34 h, the reaction reached equilibrium, and the con-
centrations of [CoII(dmgBF2)2(CH3CN)2] (0.55 mM) and
[CoI(dmgBF2)2(CH3CN)]- (1.05 mM) were determined using
their extinction coefficients at 630 nm (for Co(I)) and 440
(for Co(II)).10 The absence of isosbestic points in Figure 8
suggests
the
possibility
that
some
degradation
Figure 7. Cyclic voltammogram of 1 in the presence of acid showing
a redox couple assigned as Co(III)-H/Co(II)-H; the scan rate was 100
mV/s.
occurs over prolonged periods of time under these con-
ditions, perhaps due to side reactions that reduce the
9
J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007 8993

A R T I C L E S
Hu et al.
The electrochemistry of 1 and 2 (Scheme 1) had been previously
investigated in the presence of CF3COOH (pKa ) 12.7 in CH3-
CN),21,22 HCl‚Et2O (pKa ) 8.9), and HBF4‚Et2O (pKa
)
0.1).21-23 Catalytic waves were observed corresponding to
hydrogen evolution near the value of E°′(CoII/I) (-0.55 V for 1
and -0.28 V for 2 in CH3CN vs SCE), using HBF4‚Et2O, HCl‚
Et2O, and CF3COOH as the acids for 1, and HBF4‚Et2O and
HCl‚Et2O as the acids for 2.10 The presence of a large excess
of coordinated chloride anion had a profound influence on the
shape and position of the observed catalytic waves when HCl‚
Et2O was used as the acid. Thus we have used p-toluenesulfonic
acid (TsOH‚H2O, pKa ) 8.0) in this study,22,23 which is similar
in acid strength to HCl‚Et2O, but features a much more weakly
coordinating counteranion.
A clear correlation was observed between the activity toward
H2 evolution and the CoII/I potentials of the catalysts, with those
complexes having more negative CoII/I potentials being more
active. For the cobalt catalysts investigated here, the complex
with the most negative CoII/I reduction potential, 1, mediated
H2 evolution using the weakest acid CF3COOH (pKa ) 12.7),10
while the complexes with more positive potentials, 2, 4a, and
4b, required stronger acids, such as TsOH‚H2O (pKa ) 8.0).
The complexes with the most positive potentials, 5a and 5b,
required an even stronger acid, such as HBF4‚Et2O (pKa ) 0.1).
For complexes that are more readily reduced than 5b, such as
6a and 6b, no catalysis was observed even with HBF4‚Et2O as
the acid. When using the same acid, the more negative the CoII/I
potential, the faster the catalysis (Table 2). These observations
reflect that the basicity of the active Co(I) species is related to
its CoII/I redox potential. A similar trend has been observed by
Collman and co-workers for cofacial bis(alkyl) diporphyrin
complexes that also catalyze H2 evolution, but at much
more negative potentials (-1.2 to -1.8 V vs SCE) than those
for the Co macrocycles investigated here.5 The more easily
reduced diporphyrin complexes produced H2 at more positive
potentials but required stronger acids to achieve reasonable
reaction rates.5
Figure 8. Time-dependent UV-vis spectra (top) and pictures (bottom) of
solutions initially containing 1.6 mM of Co(dmgBF2)2 (1), 66 mM
[NBu4][CF3CO2], and 1 atm of H2.
unsaturated ligand backbone. The equilibrium constant for eq
8 is approximated to be 0.03 atm-1/2, corresponding to an
overpotential of approximately 90 mV for H2 evolution
by catalyst 1. We were not able to observe similar H2 oxida-
tion by 2 in the presence of [nBu4N][TsO] in CH3CN over a
few weeks, indicating that H2 oxidation by 2 was appreciably
slower.
3. Discussion
3.1. Structural Aspects of Co Diglyoxime Complexes.
Complex 2 is the only crystallographically characterized cobalt-
(II) complex with the (dpgBF2)2 ligand. The salient features of
its structure are similar to those of Co(dmgBF2)2(CH3OH)2,
which has an average Co-N(imine) distance of 1.876(4) Å and
a Co-O(CH3OH) distance of 2.264(4) Å.25
Complex 3 is the only crystallographically characterized
cobalt(I) complex of the (dpgBF2)2 ligand. In general, structur-
ally characterized cobalt(I) complexes with macrocyclic imine
ligands are rare. Espenson and co-workers reported the structur-
ally related complex [Co(dmgBF2)2(py)](nBu4N), in which
the Co(I)-N(imine) bond has an average distance of 1.839(3)
Å, also shorter than the corresponding bond (1.878(4) Å)
in its Co(II) analogue.26 The bond contraction was also
attributed to a higher degree of π-back-bonding in the cobalt(I)
complex.26 The cobalt(I) center is displaced 0.26 Å toward the
pyridine ligand above the macrocyclic plane of the (dmgBF2)2
ligand.
For the electrocatalysts examined here, H2 evolution occurred
at potentials just negative of the E°′(CoII/I) values for the
complexes. Thus H2 production required prior generation of the
Co(I) species. This conclusion is in accord with that of Espenson
and co-workers, who studied the evolution of H2 using Co-
(dmgBF2)2(H2O)2 as the catalyst, CrCl2 as the electron donor,
and HCl(aq) as the proton source.11 Stopped-flow kinetics
experiments in that system allowed identification of a chloride-
bridged complex [(H2O)5Cr-Cl-Co(dmgBF2)2]+ as an inter-
mediate. This intermediate was suggested to decompose to a
cobalt(I) species [LCo(dmgBF2)2]-, which thereafter reacted
with a proton to form a cobalt(III) hydride, [H-Co(L)-
(dmgBF2)2].11 The Co(III) hydride was presumed to react further
to produce H2. Two mechanisms were considered: one that
involved the protonation of the hydride ligand to produce H2,
and the other involving a bimetallic reaction of two Co(III)
hydrides.11
Similar mechanisms can be envisioned for electrocatalytic
H2 evolution by the catalysts 1-2 and 4a-5b, except that the
active Co(I) species is generated electrochemically (Scheme 3).
The CV of the CoII/I couples, with no added acid, was
electrochemically reversible. Thus the rates of interfacial electron
transfer from the electrode to the Co(II) catalysts are faster than
3.2. Mechanistic Aspects of Electrocatalytic H2 Evolution
from the Cyclic Voltammetric Behavior of Co Macrocycles.
(25) Bakac, A.; Brynildson, M. E.; Espenson, J. H. Inorg. Chem. 1986, 25,
4108-4114.
(26) Shi, S.; Daniels, L. M.; Espenson, J. H. Inorg. Chem. 1991, 30, 3407-
3410.
9
8994 J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007

H2 Evolution by Glyoxime and Tetraimine Complexes
A R T I C L E S
Table 3. Equilibrium Constants for the Reduction of an Acid by
the Co(I) Complexes
the rate of diffusion of the Co(II) species to the electrode
surface.14 The electrocatalysis differs from the chemical catalysis
because the reduction of the Co(II) by chromous ion is slow (k
≈ 0.25 s-1 for generation of [Co(dmgBF2)2]- from [(H2O)5-
Cr-Cl-Co(dmgBF2)2]+)11 and rate limiting for H2 evolution.
The electrochemical H2 evolution studied here in CH3CN is
much more rapid.
acid
pKmono
pKbi
CF3COOH
TsOH‚H2O
HBF4‚Et2O
1
1
1
-11.3
-1.9
13.7
1.4
10.8
26.4
CF3COOH
TsOH‚H2O
HBF4‚Et2O
2
2
2
-17.6
-8.1
7.4
-7.8
1.7
17.2
The presumed Co(III)-H complexes that are formed upon
protonation of the Co(I) can in principle be further reduced to
form Co(II)-H species, which could go on to produce H2. We
have not been able to synthesize a genuine Co(III)-H complex
with (dmgBF2)2, (dpgBF2)2, or TimR (R ) Me, Me/Ph, and Ph)
as the supporting ligand. Protonation of the Co(I) species with
tosic acid, CF3COOH, or anisidinium (pKa ) 11.8)22 gave Co-
(II) complexes with the corresponding ligands, with no evidence
for the formation of a long-lived cobalt(III)-H species.
Likewise, reduction of the Co(III) halide complexes with NaBH4
or NaEt3BH gave similar Co(II) products. This would be
consistent with a Co(III)-H species that can bimolecularly
release H2 at a rate that precludes its isolation (eq 7 in Scheme
3). The potentials for reported CoIII/II couples of cobalt(III) alkyl
complexes RCo(dmgBF2)2(H2O) (R ) CH3, C2H5, n-C3H7) are
-1 V vs SCE.27 We expect that the potential for the CoIII/II
couple of cobalt(III) hydride complexes, HCo(dmgBF2)2(L),
should be similar to that of RCo(dmgBF2)2L and be significantly
more negative than E°′(CoII/I) for the parent Co(dmgBF2)2(L)2.
Because electrocatalysis occurs near the measured value of
E°′(CoII/I) for all of the catalysts studied here, the intermediacy
of a Co(II)-H species seems highly unlikely. More negative
potentials, however, could facilitate a pathway for H2 evolution
that is mediated by a Co(II)-H species.
CF3COOH
TsOH‚H2O
HBF4‚Et2O
4a
4a
4a
-9.9
-0.34
15.2
-4.4
5.1
20.6
CF3COOH, TsOH‚H2O, and HBF4‚Et2O, respectively (vide
infra).28,29
The equilibrium constant for the bimetallic pathway can also
be estimated in a similar manner with
2Co(I) +2 HA f 2Co(II) + H2 + 2A-
eq(13) ) 2 × eq(9) - 2 × eq(3)
Kbi (13)
(14)
and
Kbi ) exp{2e(E°′(HA) - E°′(CoII/I))/kBT}
(15)
Calculations for the mono and bimetallic equilibrium constants
are shown in Table 3 for CF3COOH, TsOH‚H2O, and HBF4‚
Et2O. For complex 1 and 2 the monometallic reduction of
acid, Kmono is very unfavorable, except in the presence of the
strongest acid HBF4‚Et2O. However, Kbi is generally favorable
except for complexes 2 and 4a with the weakest acid, CF3-
COOH.
Scheme 3 shows two limiting reaction pathways for the
formation of H2 from a cobalt(III) hydride. The measured
potentials for the reductions of the CoIII/II and CoII/I couples
(eqs 2 and 3) along with the H2 evolution potential22,28-30
The complete reaction for the reduction of acid by the Co(I)
complex for either pathway is given by eq 13. For the
monometallic pathway, the reduction of the acid by Co(I), eq
10, must be driven by the comproportionation reaction of Co-
(III) produced in eq 10 with Co(I). It seems unlikely that a very
unfavorable equilibrium constant of less than 1 × 10-10 can be
compensated by the reduction of Co(III). The monometallic
pathway is only likely to be important for very strong acids
and/or complexes with a relatively negative Co(III/II) potential.
A similar argument has been made by Spiro et al. to rule out a
monometallic mechanism in a cobalt catalyzed hydrogen
evolution system.31
HA + e- f 1/2H2 + A-
E°′(HA)
(9)
yield the equilibrium constant for the following monometallic
reaction:
Co(I) + 2 HA f Co(III) + H2 + 2A-
Kmono
(10)
(11)
We obtain
eq(10) ) 2 × eq(9) - eq(2) - eq(3)
The kinetics of the bimetallic pathway can be derived using
a steady state approach (see Supporting Information for details).
Two limiting rate laws result. One applies when the protonation
step (eq 4) is fast:
and
e(2E°′(HA) - E°′(CoIII/II) - E°′(CoII/I))
Kmono ) exp
{
}
kBT
(28) Felton, G. A. N.; Glass, R. S.; Lichtenberger, D. L.; Evans, D. H. Inorg.
Chem. 2006, 45, 9181-9184.
(29) We chose a potential of -0.14 V vs Fc/Fc+ (i.e., 0.24 V vs SCE using
[nBu4](ClO4) as the electrolyte, see ref 30) as the thermodynamic potential
for H+/H2 in acetonitrile. This potential was determined by Daniele, S., et
al. using platinized platinum as the working electrode. A reversible H+/H2
couple was observed in acetonitrile, and the potential for this couple
corresponds to the true thermodynamic potential for this reaction. It is noted,
however, that Parker, V. D., et al. (Acc. Chem. Res. 1993, 26, 287-294)
and DuBois, D. L., et al. (J. Am. Chem. Soc. 2004, 126, 2738-2743)
estimated values of -0.05 V and -0.07 V vs Fc/Fc+, respectively, for
H+/H2 in acetonitrile, based on the junction potential between aqueous and
acetonitrile solutions.
e[(E°′(HA) - E°′(CoIII/II)) + (E°′(HA) - E°′(CoII/I))]
) exp
{
}
kBT
(12)
where e is the charge on the electron, kB is the Boltzmann
constant, and T is the temperature. The values of E°′(HA) have
been estimated as -0.51, -0.23 and +0.23 V vs SCE for
(30) Connelly, N. G.; Geiger, W. E. Chem. ReV. 1996, 96, 877-910.
(31) Kellett, R. M.; Spiro, T. G. Inorg. Chem. 1985, 24, 2373-2377.
(27) Shi, S.; Bakac, A.; Espenson, J. H. Inorg. Chem. 1991, 30, 3410-3414.
9
J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007 8995

A R T I C L E S
Hu et al.
2
in terms of both current height and shape (Figures S17-18).
These observations, together with the thermodynamic consid-
erations outlined in the previous paragraphs, lead us to conclude
that the bimetallic pathway is the dominant H2 evolution
pathway catalyzed by 1 using p-cyanoanilinium, TsOH‚H2O,
or CF3COOH as the acid. Direct studies of the chemical kinetics
of the H2 evolution catalysis is warranted to further elucidate
the mechanistic details.
[Co(I)][HA]
[A- ]
2
Rate ) K4 k7
[
]
This rate law requires a second-order dependence on the
concentrations of Co(I) and acid and an inverse second-order
dependence on the concentration of the conjugate base of the
acid. The square of the equilibrium constant appears as would
be expected for a rapid pre-equilibrium. Our electrochemical
data do not support this rate law since a first-order dependence
on cobalt and acid is observed, except for complex 1 where the
dependence on cobalt was less than first order. The second
limiting scenario occurs when the formation of Co(III)-H is
rate limiting (eq 4) and the subsequent reactions are fast. For
this case we obtain
3.3. Observation of Co(III)-H Intermediates in H2 Evolu-
tion. Evidence for a cobalt(III)-hydride intermediate was
obtained by the observation of a nearly reversible Co(III)-H/
Co(II)-H couple (E°′ ≈ -1.0 V vs SCE) for the [HCo(CH3-
CN)(dmgBF2)2] intermediate species formed in situ upon
reduction of the Co(II) species in the presence of acid (Figure
7). As mentioned above the E°′ value for this redox wave is
close to the E°′ position of -1.1 V vs SCE that has been
observed for various RCo(dmgBF2)2(H2O) (R ) CH3, C2H5,
n-C3H7) complexes.27
When the catalysis at the CoII/I potential of the original
complex is rapid and the concentration of acid is not too large,
the concentrations of acid and Co(III)-H in the reaction layer
near the electrode at the Co(III)-H/Co(II)-H potential will be
low. Therefore, only small amounts of Co(II)-H are formed at
steady state. The Co(II)-H would then react only slowly due
to the low concentrations of Co(II)-H and acid. This would
enable a reoxidation wave for what is presumably an even more
reactive Co(II)-H species to be observed, as shown in Figure
7. Similar observations have been made by Artero et al. on H2
evolution catalysis by 1 using a p-cyanoanilinium salt as the
acid.34
Rate ) k4[Co(I)][HA]
The rate is first order in both acid and cobalt, and only the
kinetics of the formation of the hydride are important. This case
is consistent with most of the observations.
Digital simulation of the cyclic voltammetric data32 was used
to study the mechanism of the H2 evolution catalysis. The
simulations suggest that catalysis by 1 and 4a proceeds, at least
predominantly, through a bimetallic mechanism. This is con-
sistent with the previous study by Espenson and co-workers on
the mechanism of stoichiometric H2 formation from a kinetically
stabilized Co(III)-H species, [HCo(P(nBu)3)(dmgH)2] (dmgH
) dimethygloxime) in the presence of acid.33 Espenson’s study
suggested that H2 was formed predominantly by the bimetallic
mechanism (eq 7) with a rate constant of 1.7 × 104 M-1 s-1
;
however, H2 was also formed by protonation of [HCo(P(nBu)3)-
(dmgH)2] (eqs 5 and 6), albeit with a much lower rate constant
of 0.42 M-1 s-1. Simulations were not able to distinguish the
bimetallic and monometallic pathways for catalysis by 2 and
4b.
No reversible redox wave for a Co(III)-H/Co(II)-H couple
could be observed for 2. Rather, a catalytic wave was observed
at a potential ∼570 mV more negative than those for the CoII/I
potential. This catalytic wave can be attributed to electrocatalytic
H2 evolution by a Co(II)-H species [HCo(CH3CN)(dpgBF2)2]-,
which can be alternatively formulated as a cobalt(III) hydride
species with a ligand radical anion. Because the catalysis for 2
is slow at the E°′(CoII/I) potential, the concentrations of the
[HCo(CH3CN)(dpgBF2)2] intermediate and acid in the reaction
layer are presumably large enough to trigger a catalytic wave
at the Co(III)-H/Co(II)-H potential. Thus no reoxidation wave
for [HCo(CH3CN)(dpgBF2)2]- was observed. Artero et al.
observed a similar catalytic wave at the Co(III)-H/Co(II)
potential for 1 using a very weak acid, Et3NHCl as the acid in
CH3CN (pKa ) 18.7).22,34
Quasi-reversible redox waves for the Co(III)-H/Co(II)-H
couple could be observed for 4a-b because the catalysis at CoII/I
potentials for 4a-b is efficient and hence the concentrations
of [HCo(Br)(TimMe)]+ intermediates and of acid are small in
the reaction layer at the Co(III)-H/Co(II)-H potential. The
cyclic voltammetric scan is again fast enough to reoxidize Co-
(II)-H prior to its reaction with either the acid or itself.
After submission of this manuscript, a related study by Artero
et al. on the electrocatalytic H2 evolution by 1 appeared.34 Their
simulation studies led them to conclude that H2 evolution
proceeded via a monometallic mechanism at the CoII/I potential
when the p-cyanoanilinium cation (pKa ) 7.6 in CH3CN)22 was
used as the proton source. The bimetallic mechanism was ruled
out by the authors based on the fact that it could not be simulated
to fit the data obtained using different acids (CF3COOH,
p-cyanoanilinium, and Et3NH). Surprised by the discrepancy
between our electrochemical conclusions using tosic acid and
those of Artero and co-workers using p-cyanoanilinium, we
undertook an independent evaluation of H2 evolution catalysis
by 1 using p-cyanoanilinium as the acid in our lab. When
collected under the same conditions as those reported by Artero
et al., our experimental data were consistent. Additionally, we
examined the electrocatalysis with an acid/catalyst ratio larger
than 10, which was not presented in Artero’s paper (Figures
S17-18). We found that both the monometallic and bimetallic
mechanisms can be used to satisfactorily simulate the experi-
mental data when the acid/catalyst ratio is ca. 10 or less.
However, only by assuming a bimetallic mechanism can the
simulated catalytic waves match the experimental waves col-
lected over the entire range of acid/catalyst ratios (1/1-40/1),
These observations of redox/catalytic waves associated with
Co(III)-H/Co(II)-H couples in the H2 evolution catalysis by
complexes 1-2 and 4a-b provide evidence for the intermediacy
of Co(III) hydride species in the electrocatalysis and help to
validate eq 4 in Scheme 3. This study can also be compared to
that of Saveant and co-workers on H2 evolution catalyzed by
an iron porphyrin complex (tetraphenylporphyrin)FeCl.6 The
intermediacy of an Fe(II) hydride species was proposed in their
(32) Rudolph, M. DigiElch 2.0. http://www.digielch.de/.
(33) Chao, T. H.; Espenson, J. H. J. Am. Chem. Soc. 1978, 100, 129-133.
(34) Baffert, C.; Artero, V.; Fontecave, M. Inorg. Chem. 2007, 46, 1817-1824.
9
8996 J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007

H2 Evolution by Glyoxime and Tetraimine Complexes
A R T I C L E S
study due to the appearance of a nearly reversible redox wave
at -1.88 V vs SCE, 280 mV more negative than the E°′ value
for the catalytically active FeI/0 couple. Similar to the present
system, the E°′ potential for the Fe(II)-H/Fe(I)-H couple was
close to those of -1.95 V vs SCE found for the Fe(II)-R/
Fe(I)-R couples, where R is a methyl or an ethyl group.
3.4. Overpotential for Hydrogen Evolution, Hydrogen
Oxidation Activity, and Comparison between Low Overpo-
tential Catalysts. The reduction potential for H2 evolution, E°′-
potentials between -0.55 V and -0.20 V vs SCE in CH3CN.
The overpotentials for the electrocatalysis, relative to the
thermodynamic potentials for H2 evolution in acidic CH3CN,
are estimated to be small. Compared to literature values for the
thermodynamic potential in CH3CN, overpotentials are estimated
to be ∼40 mV for 1 and ∼50 mV for 2. Most significantly, we
have shown experimentally that 1 can mediate both H2 evolution
and oxidation. We estimate from our equilibrium measurements
of a solution containing both H2 and acid that the overpotential
is 90 mV for 1.
1
(HA/H2), from a solution of an acid HA, HA + e- ) /2H2 +
A, varies with acid strength. The potential is related to the proton
The potential at which H2 evolution occurs can be controlled
by modulating the electronic properties of the ligand, with less
electron donating ligands giving rise to a more easily reduced
complex that catalyzes H2 evolution at a more positive potential.
However, for the series of Co complexes described here, the
positive shift of the catalytic wave correlates with a decreased
activity for electrocatalysis. Evidence for the existence of a Co-
(III)-H intermediate has been obtained from the cyclic volta-
mmetric data. Digital simulations of the voltammetry of the
electrocatalysis suggest that H2 evolution catalyzed by 1 and
4a in the presence of moderately weak acids in acetonitrile
proceeds predominantly via a bimetallic mechanism. Ongoing
work will attempt to more directly probe the mechanism of these
H2 evolving systems using stopped-flow methods.
1
reduction potential, E°′(H+/H2) (H+ + e- ) /2 H2), by
E°′(HA/H2) ) E°′(H+/H2) - 0.059 pKa(HA)
E°′(H+/H2) is 0.24 V vs SCE in CH3CN using [nBu4N][ClO4]
as the electrolyte.29,30 Thus, in CH3CN, the value of E°′(HA/
H2) has been estimated to be -0.51, -0.23, and 0.23 V vs SCE
for CF3COOH, TsOH‚H2O, and HBF4‚Et2O, respectively.22,28,29,35
The complexes studied herein appear to catalyze H2 evolution
at measurable rates close to the reversible HA/H2 potential, for
a given acid in CH3CN. The overpotential determined by
electrochemistry is 40 mV using 1 as the catalyst and CF3COOH
as the acid and is 50 mV using 2 as the catalyst and TsOH‚
H2O as the acid. We have studied the cobalt complexes 1 and
2 to determine their ability to electrocatalytically oxidize
hydrogen under 1 atm of H2 in CH3CN solutions buffered with
either [nBu4N][CF3CO2] (for 1) or [nBu4N][TsO] (for 2). No
oxidation activity could be detected at scan rates between 100
and 1000 mV per second near the CoII/I potential. We attributed
this to the slow reaction rates for H2 oxidation catalyzed by 1
and 2. In support of this view, H2 was oxidized by 1 in the
presence of [nBu4N][CF3CO2] in CH3CN over a period of several
days, reaching an equilibrium mixture (Figure 8). The equilib-
5. Experimental Section
5.1. Chemicals and Reagents. All manipulations were carried out
under an inert N2(g) atmosphere using standard Schlenk or glovebox
techniques. Unless otherwise noted, all solvents were deoxygenated
by sparging with N2(g) and were dried by passing through an activated
alumina column. Prior to use, HBF4‚Et2O and CF3COOH were degassed
by standard freeze-pump-thaw procedures. Deuterated solvents were
purchased from Cambridge Isotope Laboratories, Inc., and were
degassed and stored over activated 3 Å molecular sieves prior to use.
All other reagents were purchased from commercial sources and used
without further purification. p-Cyanoanilinium tetrafluorobate and
anisidinium tetrafluoroborate were synthesized following the method
reported by DuBois et al.,9 with the exception that fluoroboric acid in
ether was used as the acid source. Complexes 1,37 2,13 4a,17 4b,19
4c,195a,20 5b,19 5c19, 6a,18 and 6b19 were synthesized according to
literature methods; complex Co(dmgBF2)2(CH3CN) CoCp2 was prepared
as described before.10
5.2. Physical Methods. UV-vis measurements were carried out
using a Varian Cary 50 Bio spectrophotometer controlled by Cary
WinUV software. NMR spectra were recorded using a Varian Mercury
300 spectrometer. 1H NMR chemical shifts were referenced to residual
solvent as determined relative to Me4Si (δ ) 0 ppm). 19F{1H} chemical
shifts were referenced to an external standard (neat BF3‚Et2O, δ ) 0
ppm). X-ray diffraction studies were carried out in the Beckman Institute
Crystallographic Facility on a Bruker Smart 1000 CCD Diffractometer.
Cyclic voltammetric measurements were recorded by a CHI 600B
electrochemical analyzer that was connected to a glassy carbon working
electrode (surface area ) 0.07 cm-2), a platinum wire auxiliary
electrode, and a Ag/AgNO3 (0.01 M) reference electrode filled with
acetonitrile and [nBu4][ClO4] (0.1 M). All potentials were referenced
to Fc/Fc+ as an internal standard and were converted to SCE by adding
0.38 V to the measured potentials.30 The concentrations of HBF4‚Et2O,
TsOH‚H2O, and CF3COOH were calibrated by measuring the pH of
aqueous solutions of known volume/quantity of the acids using a
Beckman 32 pH meter.
rium constant for eq 8 is calculated to be ∼0.03 atm1/2
,
corresponding to an overpotential of ∼90 mV for H2 evolution
by catalyst 1. We did not observe similar H2 oxidation by 2 in
the presence of [nBu4N][TsO] in CH3CN over a few weeks,
indicating that H2 oxidation by 2 was appreciably slower.36
The overpotentials for H2 evolution catalyzed by complexes
1 and 2 are among the smallest established for any synthetic
complex. Considering the thermodynamic potentials of the acid
used in each case,28-30,35 the Cp2Mo2S2 and [Ni(PPh2NPh2)(CH3-
CN)]2+ systems by DuBois et al. have estimated overpotentials
of 120 mV and 570 mV, respectively, and the [Co(dmgH)2PyCl]
system by Artero et al. has an overpotential of 180 mV. To the
best of our knowledge, 1 is the first synthetic complex to be
established to mediate both H2 evolution and oxidation, enabling
the direct measurement of its modest overpotential for H2
evolution.
4. Conclusions
Cobalt complexes supported by macrocyclic glyoxime and
tetraimine ligands catalyzed electrochemical H2 evolution at
(35) Daniele, S.; Ugo, P.; Mazzocchin, G. A.; Bontempelli, G. Anal. Chim. Acta
1985, 173, 141-148. Wayner, D. D. M.; Parker, V. D. Acc. Chem. Res.
1993, 26, 287-294. Ellis, W. W.; Raebiger, J. W.; Curtis, C. J.; Bruno, J.
W.; DuBois, D. L. J. Am. Chem. Soc. 2004, 126, 2738-2743.
(36) One reviewer suggested that the pKa for TsOH(H2O) should be about 7 in
acetonitrile due to a large self-association constant upon deprotonation. If
this is true, the overpotential for 2 will be 109 mV, probably too large for
H2 oxidation by 2 to be observed.
Simulation of the cyclic voltammetry was performed using the
DigiElch program by M. Rudolph.32 The standard potentials for
(37) Bakac, A.; Espenson, J. H. J. Am. Chem. Soc. 1984, 106, 5197-5202.
9
J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007 8997

A R T I C L E S
Hu et al.
electron-transfer reactions (1) and (3) were taken from cyclic voltam-
metric measurements. The rate of charge transfer was set as 105 cm/s
(R ) 0.5). A diffusion coefficient of 10-5 cm2 s-1 was assumed for all
acid present in the reaction mixture, and a semi-infinite 1-D diffusion
model was used in the simulation. The geometry of the electrode was
assumed to be planar. See Supporting Information for detailed
parameters.
Bulk electrolysis experiments were done in a custom-made, gastight,
H-type cell that was equipped with a reticulated vitreous carbon working
electrode (BAS), a Ag/AgNO3 reference electrode, and a coiled Pt
counter electrode. The counter electrode was separated from the working
electrode by a glass frit. The amount of H2 gas evolved was quantified
by analyzing the gas mixture in the headspace using a Thermo Finnigan
DeltaplusXP system equipped with Finnigan trace GC and an Isotope
Ratio Mass spectrometer. The total amount of H2 produced was
calculated as the sum of the H2 in the headspace plus H2 dissolved in
the solvent (calculated using Henry’s Law, with a constant of 275
atm-1).
25.7081(13) Å, c ) 15.1519(6) Å, R ) 90.00 °, â ) 91.535(2)°, γ )
90.00 °, V ) 3833.9(3) Å3, Z ) 4, Fcalcd ) 1.533 Mg/m3.
5.5. Crystallographic Details for [Co(dpgBF2)2(CH3CN)](Na(12-
C-4)2)‚CH3CNEt2O (3‚Na(12-C-4)2). Single crystals suitable for X-ray
diffraction study were grown by diffusion of diethyl ether into a solution
of 3‚Na in CH3CN in the presence of 2 equiv of 12-crown-4. A crystal
of dimensions 0.15 mm × 0.08 mm × 0.05 mm was obtained and was
mounted onto a glass fiber. A total of 79 724 reflections (-16 e h e
16, -19 e k e 19, -39 e l e 38) were collected at T ) 100(2) K in
the range of 1.37° to 28.55° of which 12 962 were unique (Rint
)
0.0830); Mo KR radiation (λ ) 0.710 73 Å). The structure was solved
by the Direct method. All non-hydrogen atoms were refined anisotro-
pically, and hydrogen atoms were placed in calculated idealized
positions. One acetonitrile solvent molecule and one diethyl ether
solvent molecule were found in each asymmetric unit. These solvent
molecules were severely disordered and could not be refined accurately.
The residual peak and hole electron densities were 1.374 and -0.980
eA-3, respectively. The absorption coefficient was 0.402 mm-1. The
least-squares refinement converged normally with residuals of R(F) )
0.0554, wR(F2) ) 0.1016 and a GOF ) 1.042 (I > 2σ(I)).
5.3. Synthesis of [Co(dpgBF2)2]CoCp2 (3‚CoCp2). A solution of
CoCp2 (32 mg, 0.17 mmol) in CH3CN was added to a solution of 2
(113 mg, 0.17 mmol) in CH3CN at room temperature. The reaction
mixture was stirred for 2 h before being evaporated to dryness. The
resulting purple solid was collected on a filter frit, washed with diethyl
C202.18H236.64N23.01O51.04F16Na4Co4B8, space group P2(1)/n, monoclinic,
a ) 12.444(2), b ) 14.709(3), c ) 29.884(6), R ) 90°, â ) 93.319-
(5)°, γ ) 90°, V ) 5461.0(19) Å3, Z ) 5, Fcalcd ) 1.376 Mg/m3.
1
ether, and dried in a vacuum. Yield: 110 mg (75%). H NMR (300
MHz, CD3CN): δ ) 7.35 (bs, 20 H), 5.63 (s, 10 H) ppm. 19F{1H}
NMR (282 MHz, CD3CN) : δ ) 153.2 (m, 4F). Anal. Calcd for
C40H33F4N5B2O4Co2: C, 55.66%; H, 3.85%; N, 8.11%. Found: C,
55.33%; H, 3.93%; N, 8.09%. [Co(dpgBF2)2]CoCp2 (3‚Na) can be
synthesized similarly, using Na/Hg instead of CoCp2 as the reductant.
5.4. Crystallographic Details for [Co(dpgBF2)2(CH3CN)2]‚
2CH2Cl2 (2). Single crystals suitable for X-ray diffraction studies were
grown by diffusion of diethyl ether into a solution of 2 in CH2Cl2 that
had been spiked with a few drops of CH3CN. A crystal of dimensions
0.20 mm × 0.10 mm × 0.05 mm3 was obtained and was mounted
onto a glass fiber. A total of 18 190 reflections (-10 e h e 12, -34
< k e 32, -19 e l e 19) were collected at T ) 100(2) K in the range
of 1.58° to 28.23° of which 4305 were unique (Rint ) 0.1177); Mo KR
radiation (λ ) 0.710 73 Å). The structure was solved by the Direct
method. All non-hydrogen atoms were refined anisotropically, and
hydrogen atoms were placed in calculated idealized positions. The
Acknowledgment. We acknowledge support from an NSF
Chemical Bonding Center (grant CHE-0533150) and from the
Beckman Institute Molecular Materials Research Center. We
thank Prof. Nathan S. Lewis, Prof. Harry B. Gray, and Dr. Jay
Winkler for insightful discussions. We also thank Prof. Alex
Sessions and Dr. Chao Li for their generous help with the gas
chromatography measurements and Neal Mankad and Larry
Henling for help with crystallographic studies.
Supporting Information Available: Steady state analysis for
the rate law of electrocatalysis; electrocatalysis of 4a in HBF4‚
Et2O, of 4b in TsOH‚H2O and HBF4‚Et2O, and of 5a-b in
HBF4‚Et2O; variations of catalytic currents ic on the concentra-
tions of catalyst and acid for 2, 4a, and 4b; simulation of
catalytic waves for 1, 2, 4a, and 4b; catalytic waves of 2 and
4a showing evidence of cobalt(III) hydrides; and crystal-
lographic data for 2 and 3. This information is available free of
charge via the Internet at http://pubs.acs.org.
residual peak and hole electron densities were 1.643 and -1.356 eA-3
,
respectively. The absorption coefficient was 0.793 mm-1. The least-
squares refinement converged normally with residuals of R(F) ) 0.0619,
wR(F2) ) 0.0901, and a GOF ) 1.316 (I > 2σ(I)). C34H30B2Cl4-
CoF4N6O4, space group C2/c, monoclinic, a ) 9.8460(4) Å, b )
JA067876B
9
8998 J. AM. CHEM. SOC. VOL. 129, NO. 29, 2007