Synthesis and electrochemical studies of cobalt(III) monohydride complexes containing pendant amines

Article abstract of DOI:10.1021/ic401232g

Two new tetraphosphine ligands, P nC-PPh_{2 2}N ^Ph₂ (1,5-diphenyl-3,7-bis((diphenylphosphino)alkyl)-1,5- diaza-3,7-diphosphacyclooctane; alkyl = (CH₂)₂, n = 2 (L2); (CH₂)₃, n = 3 (L3)), have been synthesized. Coordination of these ligands to cobalt affords the complexes [Co ^II(L2)(CH₃CN)]²⁺ and [Co^II(L3) (CH₃CN)]²⁺, which are reduced by KC₈ to afford [Co^I(L2)(CH₃CN)]⁺ and [Co^I(L3) (CH₃CN)]⁺. Protonation of the Co^I complexes affords [HCo^III(L2)(CH₃CN)]²⁺ and [HCo ^III(L3)(CH₃CN)]²⁺. The cyclic voltammetry of [HCo^III(L2)(CH₃CN)]²⁺, analyzed using digital simulation, is consistent with an E_rC_rE_r reduction mechanism involving reversible acetonitrile dissociation from [HCo^II(L2)(CH₃CN)]⁺ and resulting in formation of HCo^I(L2). Reduction of HCo^III also results in cleavage of the H-Co bond from HCo^II or HCo^I, leading to formation of the Co^I complex [Co^I(L2)(CH₃CN)] ⁺. Under voltammetric conditions, the reduced cobalt hydride reacts with a protic solvent impurity to generate H₂ in a monometallic process involving two electrons per cobalt. In contrast, under bulk electrolysis conditions, H₂ formation requires only one reducing equivalent per [HCo^III(L2)(CH₃CN)]²⁺, indicating a bimetallic route wherein two cobalt hydride complexes react to form 2 equiv of [Co ^I(L2)(CH₃CN)]⁺ and 1 equiv of H₂. These results indicate that both HCo^II and HCo^I can be formed under electrocatalytic conditions and should be considered as potential catalytic intermediates.

Full text of DOI:10.1021/ic401232g

Article
pubs.acs.org/IC
Synthesis and Electrochemical Studies of Cobalt(III) Monohydride
Complexes Containing Pendant Amines
Eric S. Wiedner,*,‡ John A. S. Roberts,‡ William G. Dougherty,† W. Scott Kassel,† Daniel L. DuBois,‡
and R. Morris Bullock*,‡
‡Physical Sciences Division, Pacific Northwest National Laboratory, P.O. Box 999, K2-57, Richland, Washington 99352, United States
†Department of Chemistry, Villanova University, 800 East Lancaster Avenue, Villanova, Pennsylvania 19085, United States
S
* Supporting Information
ABSTRACT: Two new tetraphosphine ligands, PnC‑PPh 2NPh
2
2
(1,5-diphenyl-3,7-bis((diphenylphosphino)alkyl)-1,5-diaza-
3,7-diphosphacyclooctane; alkyl = (CH2)2, n = 2 (L2);
(CH2)3, n = 3 (L3)), have been synthesized. Coordination
of these ligands to cobalt affords the complexes [CoII(L2)-
(CH3CN)]2+ and [CoII(L3)(CH3CN)]2+, which are reduced
by KC8 to afford [CoI(L2)(CH3CN)]+ and [CoI(L3)-
(CH3CN)]+. Protonation of the CoI complexes affords
[HCoIII(L2)(CH3CN)]2+ and [HCoIII(L3)(CH3CN)]2+. The
cyclic voltammetry of [HCoIII(L2)(CH3CN)]2+, analyzed
using digital simulation, is consistent with an ErCrEr reduction
mechanism involving reversible acetonitrile dissociation from
[HCoII(L2)(CH3CN)]+ and resulting in formation of HCoI(L2). Reduction of HCoIII also results in cleavage of the H−Co bond
from HCoII or HCoI, leading to formation of the CoI complex [CoI(L2)(CH3CN)]+. Under voltammetric conditions, the
reduced cobalt hydride reacts with a protic solvent impurity to generate H2 in a monometallic process involving two electrons per
cobalt. In contrast, under bulk electrolysis conditions, H2 formation requires only one reducing equivalent per
[HCoIII(L2)(CH3CN)]2+, indicating a bimetallic route wherein two cobalt hydride complexes react to form 2 equiv of
[CoI(L2)(CH3CN)]+ and 1 equiv of H2. These results indicate that both HCoII and HCoI can be formed under electrocatalytic
conditions and should be considered as potential catalytic intermediates.
intermediate, yet the high reactivity of catalytically active
HCoIII intermediates often precludes their experimental
characterization.3,16,32−35 Two HCoIII complexes containing a
tetraamine ligand motif have been isolated:
HCoIII(dmgH)2(PnBu3) (dmgH = dimethylglyoximate)36,37
and [HCoIII(tmen)(OH2)]2+ (tmen = 2,3-dimethylbutane-2,3-
diamine),38 though neither of these complexes have been
reported to form H2 electrocatalytically. Recent reports have
indicated that electrochemical reduction of HCoIII(HMP)-
INTRODUCTION
■
As the use of renewable energy sources becomes more
widespread, conversion of electricity into chemical bonds will
become increasingly important to store the energy produced by
intermittent power sources such as solar and wind.1,2 The H−H
bond of H2 is the simplest chemical bond, but H2 production
and oxidation will require inexpensive catalysts to be
economically attractive on a large scale. A wide range of
molecular cobalt complexes have been reported for electro-
catalytic H2 production.3−11 In particular, cobaloxime (1) and
other related cobalt tetraimine complexes have been estimated
to operate at overpotentials as low as 100 mV or less.12−20
Cobalt complexes have also been studied as photocatalysts for
production of H2.10,11,21−31
(CH3 CN) (HMP
=
hangman porphyrin)6 and
[HCoIII(triphos)(CH3CN)2]2+ (triphos = CH3C-
(CH2CH2PPh2)3)7 in the presence of strong acids leads to
electrocatalytic H2 production, though these HCoIII complexes
were generated in situ and have not been studied under
noncatalytic conditions, that is, in the absence of acid.
We have reported cobalt complexes having a single bidentate
PR NR′ ligand (P2N2 = 1,5-diaza-3,7-diphosphacyclooctane),
2
2
[CoII(PR NR′2)(CH3CN)3]2+ (2), that catalyze the formation of
2
H2 at overpotentials of 160−280 mV with turnover frequencies
of 90−160 s−1.39,40 These compounds undergo ligand exchange
to afford [Co(PR NR′2)2(CH3CN)]n+ complexes containing two
2
Formation of a CoIII monohydride intermediate, hereafter
designated HCoIII, is typically invoked as a catalytic
Received: May 16, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
PR NR′ ligands at either the CoII oxidation state (R = R′ =
2
2
41
Ph)40 or the CoI oxidation state (R = Bu, R′ = Ph).
t
Accordingly no HCoIII species containing a single PR NR′
2
2
ligand has been isolated for this class of catalysts. We report
here the synthesis of new tetradentate phosphine ligands built
around a P2N2 moiety. Cobalt complexes bearing these ligands
do not undergo ligand exchange, and they afford isolable HCoIII
species that are structural analogs of the intermediates
proposed to form during H2 production by [CoII(PR NR′2)-
2
(CH3CN)3]2+. The electrochemistry of these hydride com-
plexes under noncatalytic conditions provides insight into the
redox behavior of these compounds and pathways of H2
formation from the reduced hydride species. These reduced
cobalt hydrides are important intermediates for electrocatalytic
H2 production; their catalytic performance will be detailed in a
subsequent paper.
RESULTS
■
1
determined by integration of the H and 19F NMR spectra
relative to NBu4PF6 that was added as an internal reference,
and elemental analysis was in agreement with this formulation.
Acetonitrile solutions of [CoI(L2)(CH3CN)]+ are orange, but
THF solutions of this complex are deep purple. An orange
solution is obtained again if the THF solvent is removed and
the product is redissolved in acetonitrile. This solvent-
dependent color change suggests that the CH3CN ligand of
[CoI(L2)(CH3CN)]+ is weakly bound, and in THF solution, a
four-coordinate species, [CoI(L2)]+, is formed upon loss of the
acetonitrile ligand. Only two PCH2N resonances are observed
Synthesis and Characterization of Ligands and
Complexes. The new tetradentate ligands were prepared
from the asymmetric diphosphines Ph2P(CH2)2PH2 and
Ph2P(CH2)3PH2.42,43 Treatment of these terminal phosphines
with 10 equiv of paraformaldehyde afforded the desired
bis(hydroxymethyl)phosphines (eq 1). Subsequent treatment
in the H NMR spectrum of [CoI(L2)(CH3CN)]+ in CD3CN,
1
which indicates that each face of the L2 ligand is equivalent on
the NMR time scale and that the acetonitrile ligand is
exchanging rapidly. Two broad doublets are observed at 77.1
and 73.5 ppm in the 31P{1H} NMR spectrum of [CoI(L2)-
(CH3CN)]+. This splitting pattern is consistent with an
AA′BB′ spectrum that is broadened due to coupling between
phosphorus and the quadrupolar 59Co nucleus (I = 7/2, 100%).
The complex [CoI(L3)(CH3CN)]+ can be generated in a
similar manner, with 31P{1H} NMR resonances shifted upfield
to 20.1 and 11.2 ppm. This 60 ppm upfield shift results from
the expansion of the chelate rings formed by the linker arms
from five to six atoms.44
of these compounds with 1 equiv of aniline in hot acetonitrile
afforded the ligands L2 and L3, where the number denotes the
length of the alkyl chain linking the terminal phosphines to the
1
cyclic phosphines. These ligands were characterized by H and
31P{1H} NMR spectroscopy and elemental analysis, and the
data are consistent with the assigned structures. Full
experimental details and characterization are given in the
Experimental Section.
Treatment of L2 with 1 equiv of [CoII(CH3CN)6](BF4)2 in
CH3CN affords [CoII(L2)(CH3CN)](BF4)2, as shown in eq 2.
Protonation of in situ generated [CoI(L2)(CH3CN)]+ with
p-bromoanilinium tetrafluoroborate affords the hydride com-
plex [HCoIII(L2)(CH3CN)]2+ (eq 3), which can be separated
from the KBF4 byproduct by extraction with CH2Cl2. The
hydride complex is indefinitely stable in the solid state and in
acetonitrile solution at room temperature under N2. Isolated
[HCoIII(L2)(CH3CN)]2+ typically contains small amounts
1
(<10%) of [CoII(L2)(CH3CN)]2+ as determined by H NMR
spectroscopy. The hydride resonance appears as a broad singlet
at −21.2 ppm (Δν1/2 = 144 Hz) in the 1H NMR spectrum, and
a resonance corresponding to the acetonitrile ligand is found at
1.75 ppm. In CD3CN solution, the peak intensity of the
coordinated acetonitrile ligand gradually diminishes over the
course of days and is accompanied by a corresponding increase
in the resonance for free acetonitrile at 1.96 ppm, which
indicates that dissociation of the acetonitrile ligand is slow. A
very broad resonance is observed in the 31P{1H} NMR
spectrum. Lineshape analysis of this resonance reveals the
The complex [CoII(L3)(CH3CN)](BF4)2, with a (CH2)3
linker, was obtained in a similar fashion when L3 was used as
the ligand. Magnetic moments of 2.30 and 2.20 μB were
determined for [CoII(L2)(CH3CN)]2+ and [CoII(L3)-
(CH3CN)]2+, respectively, using the Evans method. These
values are consistent with low-spin CoII complexes with one
unpaired spin.
Reduction of [CoII(L2)(CH3CN)]2+ with KC8 afforded the
CoI complex [CoI(L2)(CH3CN)]+, as shown in eq 3. The
isolated complex contained 0.5 equiv of the KBF4 byproduct as
presence of two broad signals at approximately 80 (Δν1/2 ≈
930 Hz) and 76 ppm (Δν1/2 ≈ 1500 Hz); the spectrum and
details of the fitting are given in the Supporting Information.
B
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 1. X-ray crystal structure diagrams of the dications of [Co(L2)(CH3CN)](BF4)2 (a), [Co(L3)(CH3CN)](BF4)2 (b), [HCo(L2)(CH3CN)]-
(BF4)2 (c), and [HCo(L3)(CH3CN)](BF4)2 (d). All hydrogen atoms are omitted except for the hydride ligands attached to cobalt in panels c and d.
Only the ipso carbons of the phenyl rings are shown for clarity. Thermal ellipsoids are shown at the 50% probability level.
Table 1. Selected Bond Distances (Å)
compound
Co−P(1)
Co−P(2)
Co−P(3)
Co−P(4)
Co−N(3)
[CoII(L2)(CH3CN)]2+
[CoII(L3)(CH3CN)]2+
[HCoIII(L2)(CH3CN)]2+
[HCoIII(L3)(CH3CN)]2+
2.2241(4)
2.2940(4)
2.2329(4)
2.2746(4)
2.1863(4)
2.2440(5)
2.1805(4)
2.2267(4)
2.1744(4)
2.2452(4)
2.1763(4)
2.1982(4)
2.2602(4)
2.3138(4)
2.2356(4)
2.2697(4)
2.0031(11)
2.0041(12)
1.9434(13)
1.9444(11)
The hydride complex [HCoIII(L3)(CH3CN)]2+ can be
prepared in a similar manner. The hydride resonance of
[HCoIII(L3)(CH3CN)]2+ appears as a pentet at −19.66 ppm
(2JHP = 46.7 Hz), which is in contrast to the broad singlet
observed for the hydride resonance of [HCoIII(L2)-
(CH3CN)]2+. Additionally, the 31P{1H} NMR spectrum of
[HCoIII(L3)(CH3CN)]2+ displays resonances at 21.8 ppm
(Δν1/2 = 234 Hz) and 18.4 ppm (Δν1/2 = 211 Hz) that are
much narrower than those observed for [HCoIII(L2)-
(CH3CN)]2+. These differences in the NMR line widths
suggest that quadrupolar relaxation of the hydrogen and
phosphorus atoms bound to cobalt is more efficient in
[HCoIII(L3)(CH3CN)]2+ than in [HCoIII(L2)(CH3CN)]2+.
Structural Studies. Crystals of [CoII(L2)(CH3CN)]2+,
[CoII(L3)(CH3CN)]2+, [HCoIII(L2)(CH3CN)]2+, and
[HCoIII(L3)(CH3CN)]2+ suitable for X-ray diffraction were
grown by layering ether onto an acetonitrile solution of each
complex. A drawing of each cation is shown in Figure 1, and
selected bond distances and angles are given in Table 1 and
Table S2, Supporting Information. The structure of [CoII(L2)-
(CH3CN)]2+ has a square pyramidal geometry with the four
phosphorus atoms of the L2 ligand occupying the basal
positions and an acetonitrile ligand occupying the axial position.
One six-membered chelate ring of the cyclic P2N2 fragment
adopts a chair conformation, minimizing steric interactions with
the acetonitrile ligand, and the other six-membered ring has a
boat conformation. Each arm of the L2 ligand forms a five-
membered chelate ring containing one phosphorus from the
cyclic P2N2 fragment (PC) and one of the terminal
diphenylphosphino fragments (PT). The P(2)−Co−P(3) bite
C
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
angle of the P2N2 fragment (PC−Co−PC) is small
(81.096(14)°) as typically observed for [MII(PR NR′2)-
2
(CH3CN)]2+ (M = Co, Ni) and [NiII(PR NR′2)2(CH3CN)]2+
2
complexes,39,41,45−47 and the average PC−Co−PT bite angle of
85.880(20)° is typical for a five-membered diphosphine ligand.
The average Co−PC bond length of 2.1804(6) Å is shorter than
the average Co−PT bond length of 2.2422(6) Å. The structure
of [CoII(L3)(CH3CN)]2+ is very similar to that of [CoII(L2)-
(CH3CN)]2+, with the major difference being the larger average
PC−Co−PT bite angle of 90.640(22)° for [CoII(L3)-
(CH3CN)]2+ due to the larger six-membered chelate ring of
this complex.
The solid state structure of [HCoIII(L2)(CH3CN)]2+ is
similar to that of [CoII(L2)(CH3CN)]2+, with the exception
that the hydride ligand, which was located in the difference
map, occupies the sixth coordination site trans to the
acetonitrile ligand. The Co−P bond lengths and P−Co−P
bite angles of [HCoIII(L2)(CH3CN)]2+ are similar to those
observed for [CoII(L2)(CH3CN)]2+. In the structure for
[HCoIII(L2)(CH3CN)]2+, the axial acetonitrile ligand has a
Co(1)−N(3) bond length of 1.9434(13) Å, while the
corresponding bond length in [CoII(L2)(CH3CN)]2+ is
significantly longer at 2.0031(11) Å. This is consistent with
2
removal of an electron from the dz orbital that is antibonding
with respect to the acetonitrile ligand upon transitioning from
CoII to CoIII. As expected, the solid state structure of
[HCoIII(L3)(CH3CN)]2+ is similar to that of [HCoIII(L2)-
(CH3CN)]2+ with the exception of the larger average PC−Co−
PT bite angle (90.001(18)°) for [HCoIII(L3)(CH3CN)]2+.
The equatorial P atoms show a dihedral twist about the Co
atom that is greater for [HCoIII(L3)(CH3CN)]2+ than for
[HCoIII(L2)(CH3CN)]2+. This distortion is most evident
through comparison of the two equatorial dihedral angles
(α), one of which is defined by the opposing PC−Co−PT
planes and the other by the PC−Co−PC and PT−Co−PT
planes. These angles are smaller for [HCoIII(L2)(CH3CN)]2+
(11.4° and 12.3°, respectively) than for [HCoIII(L3)-
(CH3CN)]2+ (18.3° and 18.6°). This structural detail
influences the redox properties of these complexes, as discussed
below. In contrast, the dihedral angles for [CoII(L2)-
(CH3CN)]2+ and [CoII(L3)(CH3CN)]2+ are very similar
(13.8−15.2°).
Figure 2. (A) Cyclic voltammogram of [CoII(L2)(CH3CN)]2+ at υ =
0.1 V s−1, (B) cyclic voltammogram of [HCoIII(L2)(CH3CN)]2+ at υ
= 0.1 V s−1, and (C) cyclic voltammograms (three cycles) of
[HCoIII(L2)(CH3CN)]2+ at 50 V s−1. Conditions: 2 mM analyte, 0.2
M NBu4PF6 acetonitrile solution, glassy carbon working electrode, 23
2 °C.
Electrochemical Studies. The cyclic voltammogram of
[CoII(L2)(CH3CN)]2+ in acetonitrile (Figure 2a) shows a
reversible wave having a half-wave potential E1/2 of −0.82 V vs
Cp2Fe+/0 (the reference couple for all potentials in this paper),
s i m i l a r t o t h e C o I I / I c o u p l e s o f o t h e r
[CoII(diphosphine)2(CH3CN)]2+ complexes.48,49 The anodic
and cathodic peak currents (ipa and ipc) show a linear
dependence on the scan rate υ1/2, indicating diffusion control.
An irreversible reduction wave with cathodic peak potential Epc
= −2.52 V and with ipc roughly double that of the CoII/I wave is
assigned to the reduction of CoI to Co−I. [CoII(L3)-
Reduction of [HCoIII(L2)(CH3CN)]2+ (Figure 2b) is
quasireversible at υ = 0.1 V s−1, with Epc = −1.57 V and Epa
= −1.47 V. The peak-to-peak separation (ΔEp) of 100 mV is
larger than that for Cp2Fe+/0 under the same conditions (68
mV; not shown). Cyclic voltammograms of [HCoIII(L3)-
(CH3CN)]2+ were similar in appearance, with Epc = −1.29 V
and Epa = −1.21 V at υ = 0.1 V s−1 (Figure S3, Supporting
Information). A linear dependence of ipc on υ1/2 indicates
diffusion control for reduction of [HCoIII(L2)(CH3CN)]2+
(Figure S4, Supporting Information). A new wave is observed
on the return sweep at Epa′ = −0.80 V, where the prime (′)
indicates a voltammetric property associated with this
oxidation. This wave is consistent with oxidation of [CoI(L2)-
(CH3CN)]+, which could be formed by net loss of either H• or
H− in a H−Co bond cleavage reaction. The mechanism of the
H−Co bond cleavage reaction is investigated subsequent to the
following analysis of the quasireversible HCoIII reduction wave.
Figure 2c shows the voltammograms obtained when the
applied potential was cycled three times through the
[HCoIII(L2)(CH3CN)]2+ reduction wave at υ = 50 V s−1. In
(CH3CN)]2+ also shows a reversible CoII/I wave with E1/2
=
−0.82 V, and in this case shows two partially resolved
irreversible reduction waves with Epc = −2.20 and −2.32 V,
assigned to the CoI/0 and Co0/−I couples, respectively (Figure
S2, Supporting Information). A small reduction wave at
approximately −1.1 V is observed occasionally in voltammo-
grams of [CoII(L2)(CH3CN)]2+ and frequently in voltammo-
grams of [CoII(L3)(CH3CN)]2+; these reductions are
attributed to minor coordination isomers and are discussed in
more detail in the Supporting Information.
D
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Scheme 1. ErCrEr Mechanism for Reduction of [HCoIII(L2)(CH3CN)]2+
the first cycle (black trace), the peak potential for reduction of
[HCoIII(L2)(CH3CN)]2+ is observed at −1.73 V, and a new
shoulder appears at approximately −1.53 V on the second and
third cathodic sweeps (red and blue traces). This indicates that
reduction of [HCoIII(L2)(CH3CN)]2+ on the cathodic sweep
followed by oxidation on the return anodic sweep does not
completely regenerate [HCoIII(L2)(CH3CN)]2+. Therefore
reduction of [HCoIII(L2)(CH3CN)]2+ proceeds by an ErCrEr
mechanism having two reversible electron transfers (Er) and an
intervening reversible chemical step (Cr). We propose that the
chemical step of this ErCrEr sequence is the reversible
dissociation of CH3CN from [HCoII(L2)(CH3CN)]+ (Scheme
1) to form [HCoII(L2)]+. The final reduction product of this
ErCrEr sequence is HCoI(L2), and the rate of formation of this
complex depends on the rate at which CH3CN dissociates from
[HCoII(L2)(CH3CN)]+. In this case, the potential of the first
electron transfer to [HCoIII(L2)(CH3CN)]2+ (E1°) is more
negative than the potential of the second electron transfer to
[HCoII(L2)]+ (E°2). A similar reduction mechanism is observed
for [HCoIII(dppe)2(CH3CN)]2+, and the analogues
[HCoII(dppe)2]+ and HCoI(dppe)2 have been independently
synthesized and characterized.48
With redox reactions involving chemical steps, the evolution
of peak potentials with scan rate depends on the
mechanism.50,51 An ideal reversible one-electron redox reaction
followed by an irreversible chemical reaction (an ErCi
mechanism) has a characteristic slope of ΔEp/Δlog(υ) = 30
mV. For an ErCiEr pathway, the peak potential changes with
log(υ) precisely as for the ErCi process.52 Plots of
experimentally measured Epc values versus log(υ) between υ
= 0.05 and 5 V s−1 afford a slope of 34 mV for
[HCoIII(L2)(CH3CN)]2+ and 41 mV for [HCoIII(L3)-
(CH3CN)]2+ (Figure S5, Supporting Information), which
supports the ErCrEr mechanism proposed in Scheme 1. These
slopes increase to 86−89 mV with scan rates greater than υ = 5
V s−1, which is consistent with a transition from reversible to
quasireversible electron transfer. Unlike Epc, the slope of Epa
versus log(υ) is only 11 mV between υ = 0.05 and 50 V s−1 and
can be attributed to solution resistance. This suggests that
binding of acetonitrile to [HCoII(L2)]+ is slow on the
voltammetry time scale; otherwise a greater variation in Epa
would be observed.
Scheme 2. Mechanisms for Bimetallic and Monometallic
H−Co Bond Cleavage
different combinations of HCoIII, HCoII, and HCoI can
participate in bimetallic and monometallic H−Co bond
cleavage. In the present case, [HCoIII(L2)(CH3CN)]2+ and
[HCoIII(L3)(CH3CN)]2+ are stable in acetonitrile solution,
which eliminates the possibility of either a bimetallic pathway
involving two HCoIII or a monometallic reaction of HCoIII with
a trace protic impurity.
The monometallic and bimetallic pathways both produce CoI
at the electrode and may be differentiated by examining the
voltammetry as a function of HCoIII concentration. The ratio
ipa′/ipc provides a qualitative comparison of the amount of CoI
formed (ipa′) relative to the total concentration of cobalt
species (ipc), that is, the initial HCoIII concentration, and ipa′/ipc
will increase with [HCoIII]o only if bimetallic routes are
operative. As shown in Figure 3, ipa′/ipc actually decreases with
Figure 3. Plot of ipa′/ipc vs initial concentration of [HCoIII(L2)-
(CH3CN)]2+, where ipa′ is ipa for the [CoI(L2)(CH3CN)]+ oxidation.
Conditions: 0.2 M NBu4PF6 acetonitrile solution, glassy carbon
working electrode, 23 2 °C.
As mentioned above, H−Co bond cleavage can occur upon
reduction of [HCoIII(L2)(CH3CN)]2+, and this process results
in the formation of [CoI(L2)(CH3CN)]+. We consider two
types of mechanisms, either or both of which may be operative:
a bimetallic reaction requiring one electron per Co and forming
H2 by net loss of H• and a monometallic reaction requiring two
electrons per Co and involving net loss of H− to a protic
solvent impurity such as H2O. These reaction pathways are
illustrated for HCoII in Scheme 2, though it is important to
note that the electron stoichiometry is independent of the
oxidation state of the reactive cobalt hydride. In principle
increasing [HCoIII]o from 0.5 to 2.0 mM at υ ≥ 1 V s−1,
indicating that more CoI is formed at lower initial
concentrations of cobalt. This supports a monometallic
mechanism in which HCoII, HCoI, or both lose H− to a protic
solvent impurity. The likely protic impurity is water, which can
serve as a proton source to produce 1 equiv of H2 per Co (e.g.,
Scheme 2, monometallic pathway). Acetonitrile dried as
E
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
a
Table 2. Optimized Simulation Parameters for Cyclic Voltammograms of [HCoIII(L2)(CH3CN)]2+
b
Electron Transfer Reactions and Parameters
redox couple
E° (V vs Cp2Fe+/0
)
k° (cm s−1
)
c
[HCoIII/II(L2)(CH3CN)]2+/+
[HCoII/I(L2)]+/0
[CoII/I(L2)(CH3CN)]2+/+
−1.5614(3)
−1.4742(5)
−0.82
0.025
0.28(2)
d
0.1
Chemical Reactions and Parameters
e
reaction
Keq (M)
kfwd (s−1
)
krev (M−1 s−1
)
[HCoII(L2)(CH3CN)]+ ⇌ [HCoII(L2)]+ + CH3CN
[HCoII(L2)(CH3CN)]+ → [CoII(L2)(CH3CN)]2+ + H−
32(2)
1030(20)
128(2)
32
a
b
Uncertainties are given at the 2σ (95.4%) confidence level and are generated in the least-squares refinement. Butler−Volmer transfer coefficient
c
d
(α) was fixed at 0.5 for all electron transfer reactions. Determined by manual iteration as described in the Experimental Section. Determined by
e
voltammogram simulation of isolated [CoII(L2)(CH3CN)]2+. krev = kfwd/Keq.
described in the Experimental Section typically contains 10−15
ppm water as determined by Karl Fischer titration (10 ppm
water by mass = 0.4 mM water in acetonitrile). Deliberate
addition of excess water (140 mM) to an acetonitrile solution
of [HCoIII(L2)(CH3CN)]2+ (2 mM) rendered its reduction
completely irreversible at scan rates of 1 V s−1 or less (Figure
S6, Supporting Information), though partial reversibility was
observed at higher scan rates (Figure S7, Supporting
Information).
concentrations of 1.0, 1.5, and 2.0 mM were employed in the
refinements of the digital simulations.
One conclusion drawn above is that reactions between two
HCo complexes to form H2 with net loss of H• from each,
while relevant in bulk electrolysis, are slow on the cyclic
voltammetry time scale. Candidate reactants for a bimetallic
pathway include [HCoIII(L2)(CH3CN)]2+, [HCoII(L2)-
(CH3CN)]+, [HCoII(L2)]+, and HCoI(L2), and several
combinations of these are reasonable (see the Supporting
Information). The second-order rate constants, which will be
independent of analyte concentration, were estimated by
refinement against only the 1.0 mM data, and separately
against only the 2.0 mM data, to assess the validity of the
different bimetallic models. The estimated rate constants were
consistently larger by a factor of 1.7 to 12.7 when the 1.0 mM
data set was used (see the Supporting Information for
tabulation of simulation parameters). This supports our
conclusion that these bimetallic reactions, while favorable
thermodynamically, do not contribute significantly to the
observed voltammetric responses.
Bulk electrolysis of [HCoIII(L2)(CH3CN)]2+ (2 mM) at a
potential of −1.81 V vs Cp2Fe+/0 resulted in the passage of 1.1
e− per cobalt, and voltammetry of the solution after electrolysis
indicated quantitative conversion to [CoI(L2)(CH3CN)]+.
These results indicate a net loss of H• following reduction
and a bimetallic mechanism for H2 formation under these
conditions. The slight excess of 0.1 electrons observed in the
bulk electrolysis suggests that HCoII or HCoI may consume the
trace water impurity in a monometallic pathway prior to or
concurrent with bimetallic H2 formation. A bulk electrolysis
experiment with [HCoIII] = 2.0 mM and [H2O] = 0.4 mM
would afford 1.2 electrons per Co if the water were completely
consumed in a monometallic reaction, in agreement with the
result obtained within experimental error (see Supporting
Information).
Two differences exist between the reactivity observed in the
bulk electrolysis and cyclic voltammetry experiments. First, the
extent of monometallic H−Co bond cleavage is limited by the
water concentration in a bulk electrolysis experiment, whereas
water can be recruited by diffusion from the bulk solution in a
cyclic voltammetry experiment. Second, and perhaps more
importantly, bimetallic H−Co bond cleavage is observed in the
bulk electrolysis experiment, but not in the cyclic voltammetry
experiments. This difference in apparent reactivity indicates that
bimetallic H2 formation pathways are prevalent on the time
scale of an electrolysis experiment (ca. 20 min), but occur to a
smaller extent on the time scale of a cyclic voltammetry
experiment (≤1 s when υ ≥ 1 V s−1).
The monometallic routes were modeled as pseudo-first-order
reactions; the concentration of the protic impurity is unknown
but is continually replenished at the electrode by diffusion from
the bulk solution. Acetonitrile solutions of [HCoIII(L2)-
(CH3CN)]2+ are stable prior to reduction, so the candidate
hydride donors are [HCoII(L2)(CH3CN)]+, [HCoII(L2)]+, and
HCoI(L2). Monometallic routes for formation of H2 leading to
[CoI(L2)(CH3CN)]+ were considered for each; however, only
the [HCoII(L2)(CH3CN)]+ model afforded both a first-order
rate constant that was independent of cobalt concentration and
an increased ipa′ for oxidation of [CoI(L2)(CH3CN)]+ at all
scan rates. An unconstrained simultaneous refinement on all
parameters of the monometallic [HCoII(L2)(CH3CN)]+ model
was performed against all of the experimental data sets.
Parameter estimates obtained from this refinement are given in
Table 2, and selected simulated voltammograms are shown in
Figure 4. The uncertainties provided in Table 2 (and in Tables
S3−S14, Supporting Information) are given at the 2σ (95.4%)
confidence level and are generated in the least-squares
refinement. These error estimates indicate only how well the
experimental data is fit by the chosen model and do not reflect
the precision of the cyclic voltammetry technique, systematic
errors in the experimental data, or the validity of the chosen
model. Therefore we assume a larger uncertainty in the real
values.
Modeling of Voltammograms. The above analysis
provides a qualitative description of the reactions that occur
following reduction of [HCoIII(L2)(CH3CN)]2+. Digital
simulations of the cyclic voltammograms were performed to
examine the relative rates of mono- and bimetallic H−Co bond
cleavage reactions, acetonitrile binding and dissociation, and the
influence of electron transfer kinetics. Each of these reactions is
relevant to electrocatalytic proton reduction. Experimental
voltammograms of [HCoIII(L2)(CH3CN)]2+ were recorded at
five scan rates ranging from 1 to 50 V s−1, and analyte
While the monometallic [HCoII(L2)(CH3CN)]+ model
provided the best fit to the experimental data, this model is
not necessarily a unique solution. The reaction of water with
F
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
satisfactory modeling of the experimental voltammograms.
Mono- and bimetallic models not containing a reversible
acetonitrile dissociation step predicted a severely diminished
current for ipa, ipa′, or both. As seen in Table 2, dissociation of
acetonitrile from [HCoII(L2)(CH3CN)]+ (k = 1030 s−1)
appears to be somewhat faster than acetonitrile coordination
to [HCoII(L2)]+ (k = 576 s−1 in 18 M acetonitrile solvent). A
simulation of the cyclic voltammogram cycling experiment
(Figure 2c) using the refinement parameters shown in Table 2
did not display the expected shoulder at −1.53 V (Figure S8,
Supporting Information). This indicates that the estimated
lifetime of [HCoII(L2)]+ is too short, and the optimized rate
constant for acetonitrile coordination is too large. Manually
decreasing this rate constant from k = 576 to 247 s−1 affords
simulated voltammograms showing this shoulder feature
(Figure S9, Supporting Information). The difference in these
barrier heights is 0.5 kcal mol−1, which suggests that the
optimized simulation parameters are reasonably accurate.
The simulations presented above reveal that the value of Epc
is predominantly a function of the rates of acetonitrile
dissociation and electron transfer relative to the scan rate. At
low scan rates, acetonitrile dissociation and electron transfer are
both fast relative to the potential sweep, and the cathodic wave
corresponds to the two-electron reduction of [HCoIII(L2)-
(CH3CN)]2+ to HCoI(L2). In this case Epc (−1.57 V at υ = 0.1
V/s) approaches within −20 mV of the formal potential for the
reduction of [HCoIII(L2)(CH3CN)]2+ to [HCoII(L2)]+,
calculated at E0 = −1.55 V (eqs 4−6; subscript S indicates
S
the reduction is coupled to dissociation of a solvent ligand). As
the scan rate is increased, acetonitrile dissociation and electron
transfer become slow relative to the potential sweep and Epc
shifts negative of the reversible [HCoIII/II(L2)(CH3CN)]2+/+
couple at −1.56 V.
Figure 4. Experimental (black line) and simulated (red line)
voltammograms for reduction of [HCoIII(L2)(CH3CN)]2+ (2 mM)
at (A) υ = 1 V s−1, (B) υ = 10 V s−1, and (C) υ = 50 V s−1. Simulations
use the parameters listed in Table 2.
Analysis of the cyclic voltammograms of [HCoIII(L2)-
(CH3CN)]2+ reveal that at a scan rate of 0.1 V s−1, the
experimentally measured values of Epc (−1.57 V) and Epa
(−1.46 V) are very close to the formal potentials of the
[HCoIII(L2)(CH3CN)]2+/[HCoII(L2)]+ (−1.55 V) and
[HCoII/I(L2)]+/0 (−1.47 V) redox couples obtained from
digital simulations. By applying this correlation to a cyclic
voltammogram of [HCoIII(L3)(CH3CN)]2+ recorded at 0.1 V
s−1, one can estimate potentials of −1.29 and −1.21 V for the
[HCoIII(L3)(CH3CN)]2+/[HCoII(L3)]+ and [HCoII/I(L3)]+/0
couples, respectively. This estimate of the [HCoIII(L3)-
(CH3CN)]2+/[HCoII(L3)]+ couple assumes that the equili-
brium constant for acetonitrile dissociation from [HCoII(L3)-
(CH3CN)]+ is similar to the equilibrium constant of 32 M that
was determined (eq 5) for [HCoII(L2)(CH3CN)]+.
HCoI(L2) is likely more favorable than reaction with
[HCoII(L2)(CH3CN)]+, in accord with the experimental
hydride donor abilities of [HCoII(dppe)2]+ and HCoI(dppe)2.48
Therefore a model was considered in which both [HCoII(L2)-
(CH3CN)]+ and HCoI(L2) could participate in monometallic
H−Co bond cleavage. However, refinement on this model
afforded a rate constant for monometallic H−Co bond cleavage
of HCoI(L2) that was 11 orders of magnitude smaller than that
obtained for [HCoII(L2)(CH3CN)]+. This result is inconsistent
with the reactivity predicted from consideration of the HCoII
and HCoI hydride donor abilities. We attribute this result to a
limitation of the least-squares refinement procedure to handle a
model containing two H−Co bond cleavage steps, and we
believe that both [HCoII(L2)(CH3CN)]+ and HCoI(L2) react
with water during the course of recording a cyclic voltammo-
gram.
DISCUSSION
■
Cobalt electrocatalysts for H2 production typically display a
catalytic current enhancement at the potential of the CoII/I
couple, at the potential of the CoI/0 couple, or at a potential
between these two couples.4 Comparison of the reduction
potentials of possible catalytic intermediates to the observed
catalytic wave can provide information about the mechanism,
and mechanistic studies often focus on pathways for formation
As mentioned above, a voltammogram collected at υ = 50 V
s−1 displayed a shoulder at −1.53 V on the second cathodic
sweep when the potential was cycled repeatedly about the
[HCoIII(L2)(CH3CN)]2+ reduction (Figure 2c). This data
motivated inclusion of explicit acetonitrile dissociation and
binding steps, and these were found to be necessary for
G
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
of HCoIII and HCoII.7,32,33 Knowledge of the chemical
reactivity of hydride intermediates on the voltammetry time
scale in the absence of added protons can provide information
about the mechanism(s) of electrocatalytic H2 formation. In
cases where reduction of a cobalt hydride intermediate
determines the electrocatalytic overpotential, examination of
the structural factors that control their redox properties can aid
the development of more energy-efficient electrocatalysts.
This study was designed to investigate the chemical and
electrochemical reactivity of possible cobalt hydride inter-
mediates in electrocatalytic hydrogen production by cobalt
complexes containing pendant amines. The previously reported
H2 formation and the passage of 1 e− per [CoI(L2)(CH3CN)]+
generated. This is relevant in the context of experiments
performed in the absence of acid but not necessarily for
electrocatalysis experiments conducted in the presence of
substantial amounts of acid.
Relationships between Structure and Redox Poten-
tial. The complexes [CoII(L2)(CH3CN)]2+ and [CoII(L3)-
(CH3CN)]2+ both display a CoII/I redox couple at −0.82 V.
This redox potential is intermediate between the E1/2 values of
−0.70 and −0.95 V observed for the cobalt complexes
[CoII(dppe)2(CH3CN)]2+ and [CoII(dedpe)2(CH3CN)]2+
(dedpe = Et2PCH2CH2PPh2).48,49 The qualitative ordering of
these redox potentials follows the electron donor ability of the
phosphine ligands, which is expected since the reduction of a
square pyramidal CoII complex (d7) to a square pyramidal CoI
complex (d8) does not involve a significant structural
rearrangement of the phosphine coordination geometry.
In contrast, the CoI/0 couple is a d8 to d9 reduction process
that is isoelectronic to the NiII/I couple, and the ability of the
phosphine ligands to adopt a tetrahedral geometry strongly
affects this redox couple. Previous studies on
[MII(diphosphine)2]2+ (M = Ni, Pd, Pt) complexes demon-
strated that the MII/I redox couples are affected significantly by
the diphosphine bite angle.55−58 Computational investigations
revealed that this phenomenon is a result of structural
distortion from a square planar to a tetrahedral geometry as
the ligand bite angle increases.55 This tetrahedral distortion
lowers the energy of the metal LUMO, which results in a more
positive reduction potential. The complex [CoI(L2)-
(CH3CN)]+ is constrained to a square planar arrangement of
electrocatalysts, [CoII(PR NR′2)(CH3CN)3]2+, underwent li-
2
gand exchange reactions that complicated mechanistic inves-
tigations.39−41 New tetradentate phosphine ligands containing
pendant amines (L2 and L3) were found to greatly increase the
stability of these intermediates. The CoII and CoI complexes,
[CoII(L2/L3)(CH3CN)]2+ and [CoI(L2/L3)(CH3CN)]+, are
similar to previously studied [CoII(diphosphine)2(CH3CN)]2+
and [CoI(diphosphine)2(CH3CN)]+ complexes.48,49 Addition-
ally, the HCoIII complexes [HCoIII(L2/L3)(CH3CN)]2+ are
structurally analogous to [HCoIII(dppe)2(CH3CN)]2+,48 and
these hydride complexes all display similar spectroscopic
features and stability toward H2 elimination. Importantly,
[CoII(L2)(CH3CN)]2+ and [CoII(L3)(CH3CN)]2+ are also
electrocatalysts for H2 production, and their catalytic activity
will be reported separately.
The voltammetry of [HCoIII(L2)(CH3CN)]2+ reflects a
composite of several processes: formation of HCoI through an
ErCrEr pathway (Scheme 1), monometallic reaction of HCoII
(and possibly HCoI) with a trace solvent impurity, and the
kinetics of the HCoIII reduction step. While simulation models
for various bimetallic H−Co bond cleavage reactions could
approximate individual experimental voltammograms reason-
ably well, significant discrepancies were observed for the
bimetallic rate constants obtained from refinements of
voltammograms recorded at different analyte concentrations.
On the other hand, refinement of models based on
monometallic H−Co bond cleavage resulted in nearly identical
rate constants at different analyte concentrations. These
observations strongly suggest that monometallic reaction of
the reduced cobalt hydride species is kinetically relevant on the
voltammetry time scale, while bimetallic reactions are not
kinetically important for cyclic voltammetry. In contrast, Koelle
and Ohst reported cyclic voltammetry studies in which
reduction of HCoIII(Cp)(dppv) (dppv = Ph2PCHCHPPh2)
in the absence of acid produces H2 through a bimetallic
reaction mechanism, with an estimated lower limit of 5000 M−1
s−1 for the second-order rate constant.53 Recent results from
Kaim and co-workers show that electrochemical oxidation of
HCoI(CO)2(dippf)2 (dippf = 1,1′-bis(diisopropylphosphino)-
ferrocene) results in H2 formation, which is proposed to occur
through a bimetallic pathway on the voltammetry time scale.54
However, bimetallic pathways are kinetically relevant for
[HCoIII(L2)(CH3CN)]2+ on the time scale of an electrolysis
experiment, ca. 25 min, as evidenced by the passage of 1.1 e−
per [CoI(L2)(CH3CN)]+ generated. This is in stark contrast to
[HCoII(dppe)2]+, which is stable in solution for greater than 18
h despite a favorable driving force for bimetallic H2 elimination
(ΔG = −3.6 kcal mol−1).48 One explanation for this difference
in reactivity is that the pendant amine of [HCoIII(L2)-
(CH3CN)]2+ can facilitate an intermolecular proton transfer
to [HCoII(L2)(CH3CN)]+ or HCoI(L2), which will result in
the L2 ligand, and it has a very negative CoI/0 couple of Epc
=
−2.52 V. With the larger bite angle of the L3 ligand, the CoI/0
couple of [CoI(L3)(CH3CN)]+ is shifted positive to Epc
=
−2.20 V. However, the CoI/0 couple of [CoI(L3)(CH3CN)]+ is
still more negative than the CoI / 0 couple of
[CoI(dedpe)2(CH3CN)]+ (−1.91 V), which is free to adopt a
tetrahedral geometry upon reduction to Co0.49
The HCoIII/II redox potentials, which are influenced by
acetonitrile dissociation, become more negative in the order
[HCoIII(dppe)2(CH3CN)]2+ (−0.83 V) > [HCoIII(L3)-
(CH3CN)]2+ (−1.29 V) > [HCoIII(L2)(CH3CN)]2+ (−1.55
V). This sequence follows the order of the analogous CoI/0
couples and suggests that the five-coordinate HCoII reduction
products are also stabilized by a dihedral twist of the phosphine
l i g a n d s f r o m
a p l a n a r g e o m e t r y . W h i l e n o
[HCoII(diphosphine)2]+ complexes were found in a search of
the Cambridge Crystallographic Database, many
HCoI(diphosphine)2 and related complexes have been
crystallographically characterized. In particular, HCoI(dppe)2,
HCoI(dedpe)2, and HCoI(dmpe)2 have P−Co−P angles of
approximately 90°,48,49 which is similar to that of the
tetradentate L2 and L3 ligands. These HCoI complexes are
best described as square pyramids with the hydride ligand
occupying a basal position, and similar geometries are expected
for the HCoII analogues of these species. Therefore the
reduction from [HCoI I I (dppe)2 (CH3 CN)]2 + to
[HCoII(dppe)2]+ induces a dihedral twist similar to the
reduction of CoI to Co0. However, it is difficult for the rigid
tetradentate phosphine ligands of [HCoII(L2)]+ and
[HCoII(L3)]+ to undergo a dihedral twist, so the hydride
ligands of these complexes will possess more axial character
than [HCoII(dppe)2]+. This geometric constraint leads to a
destabilization of the HCoII species and results in sequentially
H
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
more negative HCoIII/II couples for [HCoIII(L3)(CH3CN)]2+
and [HCoIII(L2)(CH3CN)]2+. A similar trend is observed for
the HCoII/I redox potentials, where the [HCoII/I(L2)]+/0 couple
(−1.47 V) is more negative than the [HCoII/I(L3)]+/0 couple
(−1.21 V).
Relevance to Cobalt H2 Production Electrocatalysts.
The above analysis of the influence of molecular geometry on
various cobalt redox potentials is useful for predicting trends of
cobalt tetraphosphine systems. We anticipate that these
concepts can be extended to other cobalt systems, which
would be valuable toward understanding and predicting the
redox chemistry of cobalt hydride species that are intermediates
of electrocatalytic H2 production. However, factors such as
molecular charge can also influence the redox potentials of
HCoIII species, and it is difficult to identify geometric trends
across a range of different ligand types given the limited
number of HCoIII complexes that have been characterized by
cyclic voltammetry.
If the potential of the HCoII/I couple is negative of the CoI/0
couple in a given catalytic system, then a HCoI intermediate
can be excluded from mechanistic consideration. This
possibility requires the open-shell (d7) HCoII species to be
more stable toward homolysis, that is, larger ΔG°H•, than the
closed-shell (d8) HCoI species as illustrated by eq 7. However,
the opposite trend in metal−hydride bond stability is typically
observed: closed-shell metal hydrides have larger ΔG°• values
than their open-shell counterparts. This trend hHas been
observed experimentally for hydride complexes of chromium,63
molybdenum,64 tungsten,64,65 iron,66 cobalt,48 rhodium,67 and
platinum,68 and has been predicted for nickel using DFT
computations.69 Additionally, the S−H bonds of the closed-
shell complexes Cp*2Mo2S3(SH), Cp*2Mo2S2(SH)2, and
Cp*2Mo2S2(SMe)(SH) all have larger ΔGH°• values than their
open-shell counterparts obtained upon either oxidation or
reduction by one electron.70 An example of the reverse trend
has been reported: the W−H bond of open-shell [Cp*(dppe)-
WH3]+ has a larger ΔG°• value than closed-shell Cp*(dppe)-
WH3 as determined frHom infrared spectroscopy and DFT
computations.71,72 It is reasonable to expect that most cobalt
systems will follow the typically observed trend, and the H−Co
bond of an open-shell HCoII complex will have a smaller ΔGH°•
value than its closed-shell HCoI counterpart. As a result, a
HCoII intermediate will likely be reduced to HCoI at potentials
positive of the CoI/0 couple for most cobalt H2 production
electrocatalysts according to the relationship described by eq 7.
Consequently a HCoI species should be considered as a
potential catalytic intermediate in mechanistic investigations.
Consideration of a HCoI intermediate increases the
mechanistic complexity of electrocatalytic H2 production
using molecular cobalt complexes. Depending on the
experimental conditions, cobalt hydride intermediates can be
formed in three metal oxidation states by a variety of
protonation and reduction sequences. Additionally, reduction
of HCoII to HCoI may require prior ligand dissociation from
HCoII, and the kinetics of this step may vary widely among
different catalyst platforms. These steps directly affect the
catalytic overpotential, which is determined by the strength of
the acid and by the potential of electrocatalysis. Additionally, a
HCoI intermediate is likely to be a stronger hydride donor than
a HCoII intermediate,48,49 which is potentially beneficial for
electrocatalytic proton reduction. However, as recently
indicated by Gray and co-workers,7 a HCoI intermediate may
be detrimental to catalysis if it can be protonated to form a
stable (H)2CoIII complex48,49,73,74 under electrocatalytic con-
ditions. Knowledge of the thermodynamic properties of the
possible intermediates is expected to facilitate catalyst develop-
ment. In a subsequent paper, we will describe the electro-
catalytic formation of H2 by [CoII(L2)(CH3CN)]2+ and
[CoII(L3)(CH3CN)]2+ and investigate the role of various
intermediates in the catalytic cycle.
The possibility of a HCoI intermediate is not commonly
considered in mechanistic investigations of electrocatalytic H2
production by molecular cobalt complexes. The present work
demonstrates that a HCoI complex, HCoI(L2), is formed at the
same potential as the HCoII complex, [HCoII(L2)]+, and prior
voltammetry studies48 demonstrated overlapping reduction
waves for the HCoIII/II and HCoII/I couples of
[HCoIII(dppe)2(CH3CN)]2+. These HCoI species are formed
at potentials of relevance to electrocatalytic H2 production, that
is, positive of the CoI/0 couple; therefore, it is of interest to
consider whether similar behavior can be generally expected of
other cobalt systems. For a given cobalt system, the CoI/0 and
HCoII/I redox couples are thermodynamically related to the
solution homolytic bond dissociation free energies (ΔG°H•) of
both the HCoII and HCoI species as illustrated in Scheme 3.
Scheme 3. Thermochemical Relationship Between
Reduction Potential and Cobalt−Hydride Solution
Homolytic Bond Dissociation Free Energy
This relationship can be expressed mathematically as shown in
eq 7 using known thermochemical cycles48,59−62 and will be
true even in cases where the formal CoI/0 or HCoII/I reductions
are ligand-centered.6,18 Equation 7 reveals that the potential of
the HCoII/I couple can be estimated to be either positive or
negative of the CoI/0 couple if the relative values of ΔG°H• for
HCoII and HCoI are known.
SUMMARY AND CONCLUSIONS
■
By employing a tetradentate phosphine ligand, we isolated and
characterized the HCoIII complexes [HCoIII(L2)(CH3CN)]2+
and [HCoIII(L3)(CH3CN)]2+, which each possess pendant
amines. Electrochemical reduction of these complexes activates
the H−Co bond toward cleavage. Analysis of the voltammo-
grams suggests a monometallic reaction between the reduced
cobalt hydride and a solvent impurity, while bulk electrolysis
experiments suggest a bimetallic reaction between two cobalt
hydride species. Digital simulations confirmed that the
E°(HCoII/I
= E°(CoI/0) +
)
I
II
°
•
°
•
ΔGH (HCo ) − ΔGH (HCo )
(7)
23.06
I
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
at 0 V. Controlled potential electrolyses were performed using a CH
Instruments 1100A power potentiostat.
monometallic pathway is dominant on the voltammetry time
scale. Additionally, digital simulations revealed the presence of
an ErCrEr reduction mechanism for [HCoIII(L2)(CH3CN)]2+,
which ultimately forms HCoI(L2), a HCoI complex, upon
dissociation of an acetonitrile ligand from [HCoII(L2)-
(CH3CN)]+, a HCoII complex. These results are significant
because HCoIII and HCoII complexes are believed to be
important intermediates in electrocatalyic H2 production, yet
they can rarely be analyzed due to their high reactivity.
Furthermore, these results indicate that HCoI complexes, while
seldom considered as catalytic intermediates, may also be
important intermediates in electrocatalytic and photocatalytic
H2 production.
Syntheses. Bis(hydroxymethyl)(2-diphenylphosphinoethyl)-
phosphine. Diphenyl(2-phosphinoethyl)phosphine (3.04 g, 0.0123
mol), paraformaldehyde (3.63 g, 0.121 mol, 9.8 equiv), and CH3CN
(100 mL) were combined in a 250 mL Schlenk flask. The mixture was
heated overnight in a 65 °C oil bath, then the mixture was filtered, and
the filtrate was concentrated to dryness. The resulting oil was
evacuated while heating at 100 °C for 3 h, which afforded
bis(hydroxymethyl)(2-phosphinoethyl)phosphine (2.20 g) as a white
solid. 31P{1H} NMR spectroscopy indicated the product was
approximately 66% pure. Less product decomposition occurred by
conducting the evacuation step at lower temperatures for longer
periods of time. However, the product could not be obtained
consistently in this manner on larger scales, presumably due to poor
mass transport of CH2O within the viscous product medium. 1H NMR
(500 MHz, CDCl3): δ 7.47−7.42 (m, 4H, ArH), 7.37−7.32 (m, 6H,
ArH), 4.32 (dd, 2H, 2JHH = 13.4 Hz, 2JHP = 10.0 Hz, P(CHAHBOH)2),
EXPERIMENTAL SECTION
■
Methods and Materials. All manipulations were carried out
under N2 using standard vacuum line, Schlenk, and inert-atmosphere
glovebox techniques. THF (Alfa-Aesar, anhydrous, nonstabilized),
acetonitrile (Alfa-Aesar, anhydrous, amine-free), dichloromethane
(Fisher), hexane (Honeywell), and diethyl ether (Burdick & Johnson)
were purified by sparging with N2 and passage through neutral
alumina, and ethanol (Pharmco-Aaper absolute anhydrous) was
purified by sparging with N2 and passage through calcium sulfate,
using an Innovative Technology, Inc., Pure Solv solvent purification
system. Water was dispensed from a Millipore Milli-Q purifier and
sparged with nitrogen. Acetonitrile-d3 (Cambridge Isotope Laborato-
ries) was vacuum-distilled from P2O5. Chloroform-d (Cambridge
Isotope Laboratories) was degassed by three freeze−pump−thaw
cycles and dried over 4 Å molecular sieves. Tetrabutylammonium
hexafluorophosphate (Fluka, ≥99%) was used as received. Tetraethy-
lammonium tetrafluoroborate (Alfa-Aesar) was recrystallized twice
from CH3CN/Et2O and dried under vacuum at room temperature. p-
Bromoanilinium tetrafluoroborate was prepared by reaction of the
substituted aniline with 1.5 equiv of HBF4·Et2O followed by
recrystallization from CH3CN/Et2O. Ferrocene (Aldrich) was
sublimed under vacuum before use. Diphenylphosphine (Strem,
99%), diethyl vinylphosphonate (Acros, 97%), diethyl allylphospho-
nate (Acros, 97%), 2,2′-azobis(2-methylpropionitrile) (AIBN, Aldrich,
97%), lithium aluminum hydride (Aldrich, 95%), chlorotrimethylsilane
(Aldrich, 99%), paraformaldehyde (Aldrich, 95%), aniline (Aldrich,
2
2
4.12 (dd, 2H, JHH = 13.4 Hz, JHP = 6.7 Hz, P(CHAHBOH)2), 2.62
(br s, 2H, OH), 2.23 (m, 2H, P(CH2)2P), 1.91 (m, 2H, P(CH2)2P).
3
31P{1H} NMR (CDCl3): δ −12.74 (d, 1P, JPP = 31.4 Hz, PPh2),
3
−18.79 (d, 1P, JPP = 31.4 Hz, P(CH2OH)2).
Bis(hydroxymethyl)(2-diphenylphosphinopropyl)phosphine. This
compound was prepared in a manner analogous to that described for
diphenyl(2-phosphinoethyl)phosphine, using 5.44 g (0.0209 mol) of
diphenyl(2-phosphinopropyl)phosphine and 6.30 g (0.210 mol, 10
equiv) of paraformaldehyde. The crude oil was evacuated while
heating at 70 °C for 8 h, to afford the product as a white solid (6.33 g).
31P{1H} NMR spectroscopy indicated the product was approximately
1
87% pure. H NMR (300 MHz, CDCl3): δ 7.47−7.39 (m, 4H, ArH),
7.38−7.30 (m, 6H, ArH), 4.32 (m, 2H, P(CHAHBOH)2), 4.07 (m, 2H,
P(CHAHBOH)2), 3.00−2.62 (br s, 2H, OH), 2.23 (m, 2H, P(CH2)3P),
1.97 (m, 2H, P(CH2)3P), 1.71 (m, 2H, P(CH2)3P). 31P{1H} NMR
(CDCl3): δ −16.75 (s, 1P, PPh2), −23.68 (s, 1P, P(CH2OH)2).
L2. Crude bis(hydroxymethyl)(2-phosphinoethyl)phosphine (2.99
g) was dissolved in EtOH (60 mL) in a 100 mL Schlenk flask. Aniline
(0.88 mL, 9.7 mmol) was added to the solution via syringe, then the
flask was immersed overnight in a 65 °C oil bath, causing a white
precipitate to form. After cooling to room temperature, the mixture
was filtered, and the solid was rinsed with EtOH (20 mL) and Et2O (2
× 20 mL) to afford L2 (1.69 g, 2.33 mmol, 48% based on aniline) as a
1
white powder. H NMR (500 MHz, CDCl3): δ 7.47−7.41 (m, 8H,
ArH), 7.40−7.31 (m, 12H, ArH), 7.19 (m, 4H, ArH), 6.68 (m, 6H,
99%), and p-bromoaniline (Aldrich, 97%) were used as received.
2
2
75,76
ArH), 4.14 (dd, 4H, JHH = 15.0 Hz, JHP = 13.6 Hz, PCHeqHaxN),
[Co(CH3CN)6](BF4)2
was prepared according to literature
2
2
42 43
3.36 (dd, 4H, JHH = 15.0 Hz, JHP = 4.2 Hz, PCHeqHaxN), 2.29 (m,
4H, P(CH2)2P), 1.44 (m, 4H, P(CH2)2P). 31P{1H} NMR (CDCl3): δ
−11.90 (d, 2P, 3JPP = 24.8 Hz, PPh2), −45.67 (br d, 2P, 3JPP = 24.8 Hz,
P2N2). Anal. Calcd for C44H46N2P4: C, 72.72; H, 6.38; N, 3.85. Found:
C, 72.65; H, 6.24; N, 3.85.
procedures. Ph2PCH2CH2PH2 and Ph2PCH2CH2CH2PH2 were
prepared using modified literature procedures, as detailed in the
Supporting Information.
1
Instrumentation. H and 31P NMR spectra were recorded on a
Varian Inova spectrometer (500 MHz for 1H) at 25 °C. All 1H
chemical shifts have been internally calibrated to the monoprotio
impurity of the deuterated solvent. The 31P NMR spectra were
referenced to external phosphoric acid at 0 ppm. NMR simulations
were performed using the line-fitting feature of MestReNova v6.0.4.
Magnetic moments were determined by Evans method77−79 in
CD3CN solution. Elemental analyses were performed by Atlantic
Microlab, Inc.
L3. Crude bis(hydroxymethyl)(2-phosphinopropyl)phosphine (6.33
g) was dissolved in CH3CN (125 mL) inside of a 250 mL Schlenk
flask. Aniline (1.80 mL, 19.8 mmol) was added to the solution via
syringe; then the flask was immersed in a 65 °C oil bath for 15 h,
leading to the formation of a white precipitate. After cooling to room
temperature, the solvent was removed under vacuum, and the solid
was washed with Et2O (3 × 30 mL). The crude product was
precipitated twice from CH2Cl2/Et2O at −35 °C and rinsed with Et2O
(20 mL) to afford L3 (2.70 g, 3.58 mmol, 36% based on aniline) as a
Voltammetry measurements were performed using a CH Instru-
ments 620D potentiostat equipped with a standard three-electrode
cell. Experiments were performed in a glovebox at ambient
1
white powder. H NMR (500 MHz, CDCl3): δ 7.47−7.41 (m, 8H,
3
ArH), 7.37−7.28 (m, 12H, ArH), 7.14 (dd, 4H, JHH = 8.6, 7.3 Hz,
temperature, 23
2 °C, using a 4−5 mL conical glass vial fitted
3
3
ArH), 6.64 (t, 2H, JHH = 7.3 Hz, ArH), 6.59 (d, 4H, JHH = 8.6 Hz,
with a polysilicone cap having openings sized to closely accept each
electrode. The working electrode (1 mm PEEK-encased glassy carbon,
Cypress Systems EE040) was polished using diamond paste (Buehler,
0.25 μm) on a polishing pad wet with purified H2O, then the electrode
was rinsed with neat acetonitrile. A glassy carbon rod (Structure Probe,
Inc.) was used as the counterelectrode, and a silver wire suspended in a
solution of Bu4NPF6 (0.2 M) in acetonitrile and separated from the
analyte solution by a Vycor frit (CH Instruments 112) was used as a
pseudoreference electrode. Ferrocene was used as an internal standard,
and all potentials are referenced to the ferrocenium/ferrocene couple
2
2
ArH), 4.09 (dd, 4H, JHH = 15.0 Hz, JHP = 12.9 Hz, PCHeqHaxN),
2
2
3.39 (dd, 4H, JHH = 15.0 Hz, JHP = 3.9 Hz, PCHeqHaxN), 2.22 (m,
4H, P(CH2)3P), 1.75 (m, 4H, P(CH2)3P), 1.53 (m, 4H, P(CH2)3P).
31P{1H} NMR (CDCl3): δ −16.67 (s, 2P, PPh2), −50.88 (br s, 2P,
P2N2). Anal. Calcd for C44H46N2P4: C, 73.20; H, 6.68; N, 3.71. Found:
C, 73.29; H, 6.54; N, 3.63.
[CoII(L2)(CH3CN)](BF4)2. A solution of [Co(CH3CN)6](BF4)2
(0.692 g, 1.45 mmol) in CH3CN (5 mL) was added to a suspension
of L2 (1.049 g, 1.443 mmol) in CH3CN (50 mL). The resulting dark
J
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
2
2
brown solution was stirred for 1 h, then the solvent was removed
under vacuum. The resulting residue was stirred overnight in hexane
(40 mL), then the mixture was filtered, and the light brown powder
was dried under vacuum to afford [CoII(L2)(CH3CN)](BF4)2 (1.255
4.39 (m, 2H, PCH2N), 4.21 (dd, 2H, JHH = 14.5 Hz, JHP = 3.0 Hz,
2
PCH2N), 3.98 (d, 2H, JHH = 14.5 Hz, PCH2N), 3.73 (m, 2H,
PCH2N), 3.21 (m, 2H, P(CH2)2P), 2.65 (m, 4H, P(CH2)2P), 2.49 (m,
2H, P(CH2)2P), 1.75 (s, 3H, CH3CN), −21.2 (br s, Δν1/2 = 144 Hz,
1H, CoH). 31P{1H} NMR (CD3CN): δ 80 (br s, Δν1/2 = 930 Hz, 2P),
76 (br s, Δν1/2 = 1500 Hz, 2P). Anal. Calcd for C46H50B2CoF8N3P4:
C, 55.17; H, 5.03; N, 4.20. Found: C, 54.53; H, 5.07; N, 4.47.
[HCoIII(L3)(CH3CN)](BF4)2. This complex was prepared in a manner
analogous to that described for [HCoIII(L2)(CH3CN)](BF4)2 and was
isolated as a yellow powder in 77% yield. The isolated product
contained trace quantities of [CoII(L3)(CH3CN)](BF4)2 as deter-
1
g, 1.255 mmol, 87%). H NMR (500 MHz, CD3CN): δ 8.68 (br s,
Δν1/2 = 63 Hz), 7.54 (br s, Δν1/2 = 26 Hz), 6.92 (br s, Δν1/2 = 89 Hz),
5.58 (br s, Δν1/2 = 73 Hz). μeff = 2.30 μB. Anal. Calcd for
C46H49B2CoF8N3P4: C, 55.23; H, 4.94; N, 4.20. Found: C, 54.54; H,
5.11; N, 4.36.
[CoII(L3)(CH3CN)](BF4)2. Over a time period of 1 h, 30 min, a
solution of [Co(CH3CN)6](BF4)2 (0.301 g, 0.629 mmol) in CH3CN
(12 mL) was added to a suspension of L3 (0.478 g, 0.634 mmol) in
CH3CN (30 mL). The resulting orange-brown solution was stirred for
10 min, then the solvent was removed under vacuum. The resulting
residue was triturated overnight in hexane (30 mL), then the mixture
was filtered, and the orange-brown powder was dried under vacuum to
1
1
mined by H NMR spectroscopy. H NMR (500 MHz, CD3CN): δ
7.57 (t, 2H, 3JHH = 7.6 Hz, ArH), 7.49−7.37 (m, 6H, ArH), 7.27(t, 4H,
3JHH = 7.7 Hz, ArH), 7.24−7.11 (m, 13H, ArH), 7.07 (t, 1H, JHH
=
3
3
3
7.3 Hz, ArH), 6.59 (dd, 2H, JHP = 9.7 Hz, JHH = 7.4 Hz, ArH), 4.24
2
(d, 2H, JHH = 14.6 Hz, PCH2N), 4.01−3.87 (m, 6H, PCH2N), 2.70
1
afford [CoII(L3)(CH3CN)](BF4)2 (0.535 g, 0.520 mmol, 83%). H
(m, 2H, P(CH2)3P), 2.48−2.27 (m, 4H, P(CH2)3P), 2.43 (s,
NMR (500 MHz, CD3CN): δ 8.68 (br s, Δν1/2 = 65 Hz), 7.53 (br s,
Δν1/2 = 25 Hz), 6.56 (br s, Δν1/2 = 111 Hz), 5.91 (br m). μeff = 2.20
μB. Anal. Calcd for C48H53B2CoF8N3P4: C, 56.06; H, 5.19; N, 4.09.
Found: C, 55.22; H, 5.12; N, 4.34.
2
CH3CN), 2.20−2.00 (m, 6H, P(CH2)3P), −19.66 (p, 1H, JHP
=
46.7 Hz, CoH). 31P{1H} NMR (CD3CN): δ 21.8 (br s, Δν1/2 = 234
Hz, 2P), 18.4 (br s, Δν1/2 = 211 Hz, 2P). Anal. Calcd for
C48H54B2CoF8N3P4: C, 56.00; H, 5.29; N, 4.08. Found: C, 55.54; H,
5.20; N, 4.18.
[CoI(L2)(CH3CN)]BF4. Solid [CoII(L2)(CH3CN)](BF4)2 (0.244 g,
0.244 mmol) was added to a stirring suspension of KC8 (0.033 g, 0.24
mmol) in CH3CN (12 mL). The mixture was stirred for 45 min, then
the solvent was removed. The residue was stirred with THF (10 mL)
for 1 h, then the mixture was allowed to settle overnight. The mixture
was filtered through Celite, and the Celite was rinsed with excess THF.
The purple filtrate was concentrated to dryness, then the residue was
dissolved in acetonitrile (2 mL) to afford an orange solution. Diethyl
ether (20 mL) was added to the acetonitrile solution, causing an
orange powder to precipitate. The mixture was cooled to −35 °C, then
the powder was collected by filtration, washed with Et2O (2 × 5 mL),
and dried under vacuum to afford [CoI(L2)(CH3CN)]BF4·(KBF4)0.5
(103 mg, 0.105 mmol, 43%). The amount of KBF4 present was
Bulk Electrolysis of [HCoIII(L2)(CH3CN)](BF4)2. A three-necked
25 mL round-bottom flask was used as the bulk electrolysis vessel, and
the electrolysis was performed in a glovebox. A 1 mm PEEK-encased
glassy carbon electrode attached to copper wire was used as the
working electrode for cyclic voltammogram measurements, and a
cylinder of reticulated vitreous carbon attached to a copper wire was
used as the working electrode for bulk electrolysis. The reference
(silver wire) and counter electrode (Nichrome wire) were placed into
5 mm glass tubes terminating in Vycor fritted disks and filled with
acetonitrile (0.2 M NEt4BF4). A solution of [HCoIII(L2)(CH3CN)]-
(BF4)2 (0.020 g, 0.020 mmol, 2 mM) and ferrocene (0.004 g, 0.02
mmol, 2 mM) in acetonitrile (0.2 M NEt4BF4, 10 mL) was transferred
to the electrolysis vessel. Bulk electrolysis was performed at −1.81 V
versus the ferrocenium/ferrocene internal reference until the current
decreased to 4.5% of its initial value. A total of 2.0197 C of charge was
passed, which corresponds to 1.1 mol of electrons per mole of
[HCoIII(L2)(CH3CN)](BF4)2. A cyclic voltammogram was recorded
upon completion of the electrolysis, and an oxidation wave was
observed at a potential corresponding to the [CoII/I(L2)-
(CH3CN)]2+/+ couple.
1
determined by integration of the H and 19F NMR spectra relative to
added [NBu4]PF6. 1H NMR (500 MHz, CD3CN): δ 7.32 (t, 4H, 3JHH
= 7.7 Hz, ArH), 7.24 (t, 4H, 3JHH = 7.0 Hz, ArH), 7.10 (d, 4H, 3JHH
=
8.0 Hz, ArH), 7.02 (m, 8H, ArH), 6.99−6.91 (m, 10H, ArH), 3.85 (br
2
2
d, 4H, JHH = 13.1 Hz, PCH2CN), 3.52 (d, 4H, JHH = 13.1 Hz,
PCH2N), 2.52 (m, 4H, P(CH2)2P), 2.19 (m, 4H, P(CH2)2P). 31P{1H}
NMR (CD3CN): δ 77.1 (br d, 2P, 2JPP = 70 Hz), 73.5 (br d, 2P, 2JPP
=
70 Hz). Anal. Calcd for C46H49B1.5CoF6K0.5N3P4: C, 56.58; H, 5.06; N,
4.30. Found: C, 56.80; H, 5.03; N, 4.08.
[CoI(L3)(CH3CN)](BF4). A solution of [CoII(L3)(CH3CN)](BF4)2
(0.014 g, 0.014 mmol) in CD3CN (ca. 1 mL) was added to solid KC8
(3 mg, 0.02 mmol, ca. 1.4 equiv). The mixture was stirred for 1 h and
filtered through a small plug of Celite, and the orange filtrate was
Digital Simulations. DigiElch version 7.F (build 7.023) was used
for digital simulations and nonlinear least-squares refinement of
models against experimental data. Voltammograms were collected
using a pre-equilibrium time of 2 s, and pre-equilibrium was enabled
for all refinements. Background subtraction was employed for all data
sets, and the double-layer capacitance (Cdl) was set to 0 for all
refinements. A reasonable estimate for the solution resistivity, ρ (50
Ω), was made by manual iteration through comparison of the
simulated and experimental Ep values for the [CoII/I(L2)-
(CH3CN)]2+/+ couple, and this value was used for all refinements.
Self-diffusion of acetonitrile was taken as 4.35 × 10−5 cm2 s−1;80 the
activity of acetonitrile in the electrolyte solutions was taken to be 18
M, in accord with Tilset’s estimate.81 Current normalization was used
for all refinements.
1
analyzed by NMR spectroscopy. H NMR (500 MHz, CD3CN): δ
3
7.31 (t, 4H, JHH = 7.9 Hz, ArH), 7.20−7.12 (m, 12H, ArH), 7.06 (d,
3
4H, JHH = 8.2 Hz, ArH), 6.98−6.91 (m, 10H, ArH), 3.38 (s, 8H,
PCH2CN), 2.34 (m, 4H, P(CH2)3P), 2.04−1.95 (br m, 9H, P(CH2)3P
+ CH3CN). 31P{1H} NMR (CD3CN): δ 20.1 (m, 2P), 11.2 (m, 2P).
[HCoIII(L2)(CH3CN)](BF4)2. Solid KC8 (0.052 g, 0.39 mmol, 1.6
equiv) was added to a stirring solution of [CoII(L2)(CH3CN)](BF4)2
(0.251 g, 0.250 mmol) in CH3CN (12 mL). After being stirred for 2.5
h, the mixture was filtered through Celite, and solid p-bromoanilinium
tetrafluoroborate (0.072 g, 0.28 mmol, 1.1 equiv) was added to the
orange filtrate. The resulting yellow solution was stirred for 45 min,
then the solvent was removed under vacuum. The residue was
extracted with CH2Cl2 (2 × 5 mL) and filtered through Celite, and the
filtrate was concentrated under vacuum. The residue was triturated
overnight in Et2O (10 mL), then the mixture was filtered, and the
resulting yellow powder was rinsed with Et2O (2 × 5 mL) and dried
under vacuum to afford [HCoIII(L2)(CH3CN)](BF4)2 (0.174 g, 0.174
mmol, 70%). The isolated product contained trace quantities of
The diffusion coefficient (D0) of [CoII(L2)(CH3CN)]2+ was
measured using chronoamperometry experiments and determined to
be 9.6 × 10−6 cm2 s−1 using the Cottrell equation, and this value was
used for all cobalt species included in the digital simulations.
Simulations of [CoII(L2)(CH3CN)]2+ were performed on five
voltammograms recorded from 1 to 50 V s−1 with an analyte
concentration of 2 mM, and a heterogeneous electron transfer rate
constant (k°) of 0.1 cm s−1 was obtained for this redox couple. In the
experimental voltammograms of [HCoIII(L2)(CH3CN)]2+, a small
amount of [CoII(L2)(CH3CN)]2+ was observed as a contaminant on
the initial cathodic sweep, and its bulk concentration was estimated to
be 7% of the [HCoIII(L2)(CH3CN)]2+ concentration by manual
iteration. A second unknown contaminant was observed at Epc ≈ −1.1
1
[CoII(L2)(CH3CN)](BF4)2 as determined by H NMR spectroscopy.
3
1H NMR (500 MHz, CD3CN): δ 7.59 (t, 2H, JHH = 7.4 Hz, ArH),
3
7.49−7.37 (m, 6H), 7.35 (t, 4H, JHH = 7.2 Hz, ArH), 7.27 (d, 2H,
3
3JHH = 7.9 Hz, ArH), 7.18−7.10 (m, 11H, ArH), 7.06 (t, 1H, JHH
=
7.5 Hz, 7.3 ArH), 6.59 (dd, 2H, 3JHP = 10.3 Hz, JHH = 7.2 Hz, ArH),
3
K
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
V and was not modeled since the concentration of this species was
approximately 2% of the [HCoIII(L2)(CH3CN)]2+ concentration.
Three sets of refinements were performed for each monometallic
and bimetallic reaction model: (1) free refinement on the 2 mM data
set, (2) free refinement on the 1 mM data set, and (3) free refinement
on the 1 mM data set with Keq and kfwd for reversible acetonitrile
dissociation fixed to the values obtained from refinement on the 2 mM
data set. Tables containing these refinement parameters are given in
the Supporting Information. For refinement on the full data set, the
heterogeneous electron transfer rate constant (k°) for all of the
electron transfer steps was fixed initially at 0.1 cm s−1. With k° fixed,
the simulated Epc values were 40 mV positive of the experimental
values at υ = 50 V s−1, though better agreements were observed at
lower scan rates. In subsequent refinements, k° for the
[HCoII/I(L2)]+/0 couple was allowed to refine freely, and k° for the
[HCoIII/II(L2)(CH3CN)]2+/+ couple was manually decreased until the
experimental and simulated peak potentials differed by less than 15
mV at all scan rates.
X-ray Diffraction Studies. For each of the crystal structure
studies, a single crystal grown from a CH3CN solution layered with
Et2O was mounted using NVH immersion oil onto a nylon fiber and
cooled to the data collection temperature of 100(2) K. Data were
collected on a Bruker-AXS Kappa APEX II CCD diffractometer with
0.71073 Å Mo Kα radiation. Unit cell parameters were obtained from
90 data frames at 0.3° Φ from three different sections of the Ewald
sphere. The data sets were treated with SADABS absorption
corrections based on redundant multiscan data (Sheldrick, G.,
Bruker-AXS, 2001). All non-hydrogen atoms were refined with
anisotropic displacement parameters. All hydrogen atoms were treated
as idealized contributions except for the hydride ligand, which was
located from the difference map and allowed to refine freely.
the modeling of this minor disorder was ignored in favor of a less-
restrained model.
[HCoIII(L3)(CH3CN)](BF4)2. The data was consistent with the
centrosymmetric, triclinic space group P1. The asymmetric unit
̅
contains one [HCoIII(L3)(CH3CN)]2+ cation, two [BF4]− anions, and
two molecules of acetonitrile solvent all located on general positions
yielding Z = 2 and Z′ = 1. There is two-part disorder in one of the
[BF4]− anions and one propyl bridge on the ligand. In each case, the
disorder was located from the difference map and refined using SAME,
SIMU, and DELU commands to keep the two parts similar.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental details of phosphine precursor synthesis,
simulated 31P{1H} NMR spectrum of [HCoIII(L2)(CH3CN)]2+
and parameters, selected bond angles from X-ray diffraction,
discussion of minor coordination isomers, calculation of
predicted electron stoichiometry for a bulk electrolysis
experiment, cyclic voltammetry plots, refinement parameters
from digital simulation of cyclic voltammograms, and CIF files
containing crystallographic data. This material is available free
of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail addresses: eric.wiedner@pnnl.gov; morris.bullock@
pnnl.gov.
[CoII(L2)(CH3CN)](BF4)2. The systematic absences in the data were
consistent with the centrosymmetric, monoclinic space group P21/n.
The asymmetric unit contains one [CoII(L2)(CH3CN)]2+ cation, two
[BF4]− anions and 2.5 molecules of acetonitrile solvent, all located on
general positions yielding Z = 4 and Z′ = 1. The [BF4]− anions were
disordered over two positions, which were located from the difference
map and refined. The [BF4]− bond lengths and angles were set
equivalent to one another using the SAME command in an attempt to
keep them chemically equivalent. A SQUEEZE analysis suggested 2.5
molecules of CH3CN solvent per asymmetric unit. The difference map
suggests three separate sites in the asymmetric unit for the acetonitrile.
One is fully occupied and refines well. The other two are likely
disordered over those two positions in a ratio that would lead to an
average of 1.5 equiv of solvent. These did not refine well, and their
geometries were bent instead of linear. Therefore, all three solvent
molecules were removed using SQUEEZE. There was also two-part
disorder in one of the phenyl rings of the ligand, which was located
from the difference map and refined. The disordered rings were fixed
using an AFIX66 command to help stabilize the refinement.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank the U.S. Department of Energy, Office of Basic
Energy Sciences, Division of Chemical Sciences, Geosciences
and Biosciences, for support of the initial parts of this study.
Current work is supported by the Center for Molecular
Electrocatalysis, an Energy Frontier Research Center funded by
the U.S. Department of Energy, Office of Science, Office of
Basic Energy Sciences. Pacific Northwest National Laboratory
is operated by Battelle for the U.S. Department of Energy.
REFERENCES
■
(1) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U. S. A. 2006,
103, 15729−15735.
[CoII(L3)(CH3CN)](BF4)2. The data was consistent with the
(2) Cook, T. R.; Dogutan, D. K.; Reece, S. Y.; Surendranath, Y.;
Teets, T. S.; Nocera, D. G. Chem. Rev. 2010, 110, 6474−6502.
(3) Artero, V.; Chavarot-Kerlidou, M.; Fontecave, M. Angew. Chem.,
Int. Ed. 2011, 50, 7238−7266.
centrosymmetric, triclinic space group P1. The asymmetric unit
̅
contains one [CoII(L3)(CH3CN)]2+ cation, two [BF4]− anions, and
two molecules of acetonitrile solvent all located on general positions
yielding Z = 2 and Z′ = 1. There is two-part disorder in one of the
BF4− anions, one phenyl ring, and one propyl bridge. In each case, the
disorder was located from the difference map, and the two parts were
restrained to be equivalent using the SAME command. SIMU and
DELU commands were attempted to keep the thermal parameters of
the disordered parts similar. However, these resulted in some NPD
atoms in the disordered parts so more rigid EADP commands were
used.
(4) Dempsey, J. L.; Brunschwig, B. S.; Winkler, J. R.; Gray, H. B. Acc.
Chem. Res. 2009, 42, 1995−2004.
(5) Wang, M.; Chen, L.; Sun, L. Energy Environ. Sci. 2012, 5, 6763−
6778.
(6) Lee, C. H.; Dogutan, D. K.; Nocera, D. G. J. Am. Chem. Soc. 2011,
133, 8775−8777.
(7) Marinescu, S. C.; Winkler, J. R.; Gray, H. B. Proc. Natl. Acad. Sci.
U. S. A. 2012, 109, 15127−15131.
[HCoIII(L2)(CH3CN)](BF4)2. The systematic absences in the data were
consistent with the centrosymmetric, monoclinic space group P21/c.
The asymmetric unit contains one [HCoIII(L2)(CH3CN)]2+ cation,
two [BF4]− anions, and one molecule of acetonitrile solvent all located
on general positions yielding Z = 4 and Z′ = 1. The thermal ellipsoid
plot suggested that there was positional disorder in one of the phenyl
rings; however, even after modeling and restraining the structure, the
thermal parameters did not show significant improvement. Therefore,
(8) Bigi, J. P.; Hanna, T. E.; Harman, W. H.; Chang, A.; Chang, C. J.
Chem. Commun. 2010, 46, 958−960.
(9) Sun, Y.; Bigi, J. P.; Piro, N. A.; Tang, M. L.; Long, J. R.; Chang, C.
J. J. Am. Chem. Soc. 2011, 133, 9212−9215.
(10) Singh, W. M.; Baine, T.; Kudo, S.; Tian, S.; Ma, X. A. N.; Zhou,
H.; DeYonker, N. J.; Pham, T. C.; Bollinger, J. C.; Baker, D. L.; Yan,
B.; Webster, C. E.; Zhao, X. Angew. Chem., Int. Ed. 2012, 51, 5941−
5944.
L
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(11) McNamara, W. R.; Han, Z.; Alperin, P. J.; Brennessel, W. W.;
Holland, P. L.; Eisenberg, R. J. Am. Chem. Soc. 2011, 133, 15368−
15371.
(12) Razavet, M.; Artero, V.; Fontecave, M. Inorg. Chem. 2005, 44,
4786−4795.
(44) Garrou, P. E. Chem. Rev. 1981, 81, 229−266.
(45) Kilgore, U. J.; Roberts, J. A. S.; Pool, D. H.; Appel, A. M.;
Stewart, M. P.; DuBois, M. R.; Dougherty, W. G.; Kassel, W. S.;
Bullock, R. M.; DuBois, D. L. J. Am. Chem. Soc. 2011, 133, 5861−5872.
(46) Galan, B. R.; Schoffel, J.; Linehan, J. C.; Seu, C.; Appel, A. M.;
̈
(13) Hu, X. L.; Cossairt, B. M.; Brunschwig, B. S.; Lewis, N. S.;
Peters, J. C. Chem. Commun. 2005, 4723−4725.
(14) Baffert, C.; Artero, V.; Fontecave, M. Inorg. Chem. 2007, 46,
1817−1824.
Roberts, J. A. S.; Helm, M. L.; Kilgore, U. J.; Yang, J. Y.; DuBois, D. L.;
Kubiak, C. P. J. Am. Chem. Soc. 2011, 133, 12767−12779.
(47) Kilgore, U. J.; Stewart, M. P.; Helm, M. L.; Dougherty, W. G.;
Kassel, W. S.; DuBois, M. R.; DuBois, D. L.; Bullock, R. M. Inorg.
Chem. 2011, 50, 10908−10918.
(15) Jacques, P. A.; Artero, V.; Pecaut, J.; Fontecave, M. Proc. Natl.
Acad. Sci. U. S. A. 2009, 106, 20627−20632.
(48) Ciancanelli, R.; Noll, B. C.; DuBois, D. L.; DuBois, M. R. J. Am.
Chem. Soc. 2002, 124, 2984−2992.
(16) Hu, X. L.; Brunschwig, B. S.; Peters, J. C. J. Am. Chem. Soc.
2007, 129, 8988−8998.
(49) Mock, M. T.; Potter, R. G.; O’Hagan, M. J.; Camaioni, D. M.;
Dougherty, W. G.; Kassel, W. S.; DuBois, D. L. Inorg. Chem. 2011, 50,
11914−11928.
(17) McCrory, C. C. L.; Uyeda, C.; Peters, J. C. J. Am. Chem. Soc.
2012, 134, 3164−3170.
(18) Stubbert, B. D.; Peters, J. C.; Gray, H. B. J. Am. Chem. Soc. 2011,
133, 18070−18073.
(50) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
Fundamentals and Applications, 2nd ed.; John Wiley & Sons: Hoboken,
NJ, 2001.
(19) Berben, L. A.; Peters, J. C. Chem. Commun. 2010, 46, 398−400.
(20) Valdez, C. N.; Dempsey, J. L.; Brunschwig, B. S.; Winkler, J. R.;
Gray, H. B. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 15589−15593.
(21) Du, P.; Schneider, J.; Luo, G.; Brennessel, W. W.; Eisenberg, R.
Inorg. Chem. 2009, 48, 4952−4962.
(51) Nicholson, R. S.; Shain, I. Anal. Chem. 1964, 36, 706−723.
́
(52) Saveant, J. M. Elements of Molecular and Biomolecular
Electrochemistry: An Electrochemical Approach to Election Transfer
Chemistry; Wiley: Hoboken, NJ, 2006.
(22) Lazarides, T.; McCormick, T.; Du, P.; Luo, G.; Lindley, B.;
Eisenberg, R. J. Am. Chem. Soc. 2009, 131, 9192−9194.
(23) Du, P.; Eisenberg, R. Energy Environ. Sci. 2012, 5, 6012−6021.
(24) Zhang, P.; Jacques, P. A.; Chavarot-Kerlidou, M.; Wang, M.;
Sun, L.; Fontecave, M.; Artero, V. Inorg. Chem. 2012, 51, 2115−2120.
(25) Probst, B.; Rodenberg, A.; Guttentag, M.; Hamm, P.; Alberto, R.
Inorg. Chem. 2010, 49, 6453−6460.
(53) Koelle, U.; Ohst, S. Inorg. Chem. 1986, 25, 2689−2694.
(54) Krafft, M. J.; Bubrin, M.; Paretzki, A.; Lissner, F.; Fiedler, J.;
Zalis, S.; Kaim, W. Angew. Chem., Int. Ed. 2013, 52, 6781−6784.
́ ̌
(55) Nimlos, M. R.; Chang, C. H.; Curtis, C. J.; Miedaner, A.; Pilath,
H. M.; DuBois, D. L. Organometallics 2008, 27, 2715−2722.
(56) Berning, D. E.; Miedaner, A.; Curtis, C. J.; Noll, B. C.; DuBois,
M. C. R.; DuBois, D. L. Organometallics 2001, 20, 1832−1839.
(57) Berning, D. E.; Noll, B. C.; DuBois, D. L. J. Am. Chem. Soc.
1999, 121, 11432−11447.
(58) Raebiger, J. W.; Miedaner, A.; Curtis, C. J.; Miller, S. M.;
Anderson, O. P.; DuBois, D. L. J. Am. Chem. Soc. 2004, 126, 5502−
5514.
(59) Wayner, D. D. M.; Parker, V. D. Acc. Chem. Res. 1993, 26, 287−
294.
(60) Parker, V. D.; Handoo, K. L.; Roness, F.; Tilset, M. J. Am. Chem.
Soc. 1991, 113, 7493−7498.
(61) Tilset, M.; Parker, V. D. J. Am. Chem. Soc. 1990, 112, 2843−
2843.
(62) Tilset, M.; Parker, V. D. J. Am. Chem. Soc. 1989, 111, 6711−
6717.
(26) Probst, B.; Guttentag, M.; Rodenberg, A.; Hamm, P.; Alberto, R.
Inorg. Chem. 2011, 50, 3404−3412.
(27) Li, L.; Duan, L.; Wen, F.; Li, C.; Wang, M.; Hagfeldt, A.; Sun, L.
Chem. Commun. 2012, 48, 988−990.
(28) Huang, J.; Mulfort, K. L.; Du, P.; Chen, L. X. J. Am. Chem. Soc.
2012, 134, 16472−16475.
(29) Hawecker, J.; Lehn, J. M.; Ziessel, R. New J. Chem. 1983, 7,
271−277.
(30) Fihri, A.; Artero, V.; Pereira, A.; Fontecave, M. Dalton Trans.
2008, 5567−5569.
(31) Fihri, A.; Artero, V.; Razavet, M.; Baffert, C.; Leibl, W.;
Fontecave, M. Angew. Chem., Int. Ed. 2008, 47, 564−567.
(32) Solis, B. H.; Hammes-Schiffer, S. Inorg. Chem. 2011, 50, 11252−
11262.
(63) Tilset, M. J. Am. Chem. Soc. 1992, 114, 2740−2741.
(64) Skagestad, V.; Tilset, M. J. Am. Chem. Soc. 1993, 115, 5077−
5083.
(33) Dempsey, J. L.; Winkler, J. R.; Gray, H. B. J. Am. Chem. Soc.
2010, 132, 16774−16776.
(34) King, A. E.; Surendranath, Y.; Piro, N. A.; Bigi, J. P.; Long, J. R.;
Chang, C. J. Chem. Sci. 2013, 4, 1578−1587.
(65) Roberts, J. A. S.; Appel, A. M.; DuBois, D. L.; Bullock, R. M. J.
Am. Chem. Soc. 2011, 133, 14604−14613.
(35) Muckerman, J. T.; Fujita, E. Chem. Commun. 2011, 47, 12456−
12458.
(66) Tilset, M.; Hamon, J. R.; Hamon, P. Chem. Commun. 1998,
765−766.
(36) Schrauzer, G. N.; Holland, R. J. J. Am. Chem. Soc. 1971, 93,
(67) Raebiger, J. W.; DuBois, D. L. Organometallics 2005, 24, 110−
118.
1505−1506.
(37) Bhattacharjee, A.; Chavarot-Kerlidou, M.; Andreiadis, E. S.;
Fontecave, M.; Field, M. J.; Artero, V. Inorg. Chem. 2012, 51, 7087−
7093.
(68) Miedaner, A.; Raebiger, J. W.; Curtis, C. J.; Miller, S. M.;
DuBois, D. L. Organometallics 2004, 23, 2670−2679.
(69) Chen, S.; Rousseau, R.; Raugei, S.; Dupuis, M.; DuBois, D. L.;
Bullock, R. M. Organometallics 2011, 30, 6108−6118.
(70) Appel, A. M.; Lee, S. J.; Franz, J. A.; DuBois, D. L.; DuBois, M.
R. J. Am. Chem. Soc. 2009, 131, 5224−5232.
(71) Pleune, B.; Morales, D.; Meunier-Prest, R.; Richard, P.;
Collange, E.; Fettinger, J. C.; Poli, R. J. Am. Chem. Soc. 1999, 121,
2209−2225.
(38) Rahman, A. F. M. M.; Jackson, W. G.; Willis, A. C.; Rae, A. D.
Chem. Commun. 2003, 2748−2749.
(39) Wiedner, E. S.; Yang, J. Y.; Dougherty, W. G.; Kassel, W. S.;
Bullock, R. M.; DuBois, M. R.; DuBois, D. L. Organometallics 2010, 29,
5390−5401.
(40) Jacobsen, G. M.; Yang, J. Y.; Twamley, B.; Wilson, A. D.;
Bullock, R. M.; DuBois, M. R.; DuBois, D. L. Energy Environ. Sci. 2008,
1, 167−174.
(72) Pleune, B.; Poli, R.; Fettinger, J. C. J. Am. Chem. Soc. 1998, 120,
3257−3258.
(41) Wiedner, E. S.; Yang, J. Y.; Chen, S. T.; Raugei, S.; Dougherty,
W. G.; Kassel, W. S.; Helm, M. L.; Bullock, R. M.; DuBois, M. R.;
DuBois, D. L. Organometallics 2012, 31, 144−156.
(42) King, R. B.; Cloyd, J. C.; Kapoor, P. N. J. Chem. Soc., Perkin
Trans. 1 1973, 2226−2229.
(73) Bianchini, C.; Laschi, F.; Peruzzini, M.; Ottaviani, F. M.; Vacca,
A.; Zanello, P. Inorg. Chem. 1990, 29, 3394−3402.
(74) Heinekey, D. M.; Liegeois, A.; Vanroon, M. J. Am. Chem. Soc.
1994, 116, 8388−8389.
(75) Hathaway, B. J.; Holah, D. G.; Underhill, A. E. J. Chem. Soc.
1962, 2444−2448.
(43) Meier, U. W.; Hollmann, F.; Thewalt, U.; Klinga, M.; Leskela,
̈
M.; Rieger, B. Organometallics 2003, 22, 3905−3914.
M
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(76) Heintz, R. A.; Smith, J. A.; Szalay, P. S.; Weisgerber, A.; Dunbar,
K. R. Inorg. Synth. 2002, 33, 75−83.
(77) Evans, D. F. J. Chem. Soc. 1959, 2003−2005.
(78) Schubert, E. M. J. Chem. Educ. 1992, 69, 62−62.
(79) Baker, M. V.; Field, L. D.; Hambley, T. W. Inorg. Chem. 1988,
27, 2872−2876.
(80) Hurle, R. L.; Woolf, L. A. J. Chem. Soc., Faraday Trans. 1 1982,
78, 2233−2238.
(81) Tilset, M. Inorg. Chem. 1994, 33, 3121−3126.
N
dx.doi.org/10.1021/ic401232g | Inorg. Chem. XXXX, XXX, XXX−XXX