Redox-Induced Structural Reorganization Dictates Kinetics of Cobalt(III) Hydride Formation via Proton-Coupled Electron Transfer

Article abstract of DOI:10.1021/jacs.0c11992

Two-electron, one-proton reactions of a family of [CoCp(dxpe)(NCCH3)]2+ complexes (Cp = cyclopentadienyl, dxpe = 1,2-bis(di(aryl/alkyl)phosphino)ethane) form the corresponding hydride species [HCoCp(dxpe)]+ (dxpe = dppe (1,2-bis(diphenylphosphino)ethane), depe (1,2-bis(diethylphosphino)ethane), and dcpe (1,2-bis(dicyclohexylphosphino)ethane)) through a stepwise proton-coupled electron transfer process. For three [CoCp(dxpe)(NCCH3)]2+ complexes, peak shift analysis was employed to quantify apparent proton transfer rate constants from cyclic voltammograms recorded with acids ranging 22 pKa units. The apparent proton transfer rate constants correlate with the strength of the proton source for weak acids, but these apparent proton transfer rate constants curiously plateau (kpl) as the reaction becomes increasingly exergonic. The absolute apparent proton transfer rate constants across both these regions correlate with the steric bulk of the chelating diphosphine ligand, with bulkier ligands leading to slower kinetics (kplateau,depe = 3.5 × 107 M-1 s-1, kplateau,dppe = 1.7 × 107 M-1 s-1, kplateau,dcpe = 7.1 × 104 M-1 s-1). Mechanistic studies were conducted to identify the cause of the aberrant kPTapp-ΔpKa trends. When deuterated acids are employed, deuterium incorporation in the Cp ring is observed, indicating protonation of the CoCp(dxpe) species to form the corresponding hydride proceeds via initial ligand protonation. Digital simulations of cyclic voltammograms show ligand loss accompanying initial reduction gates subsequent PCET activity at higher driving forces. Together, these experiments reveal the details of the reaction mechanism: reduction of the Co(III) species is followed by dissociation of the bound acetonitrile ligand, subsequent reduction of the unligated Co(II) species to form a Co(I) species is followed by protonation, which occurs at the Cp ring, followed by tautomerization to generate the stable Co(III)-hydride product [HCoCp(dxpe)]+. Analysis as a function of chelating disphosphine ligand, solvent, and acid strength reveals that the ligand dissociation equilibrium is directly influenced by the steric bulk of the phosphine ligands and gates protonation, giving rise to the plateau of the apparent proton transfer rate constant with strong acids. The complexity of the reaction mechanism underpinning hydride formation, encompassing dynamic behavior of the entire ligand set, highlights the critical need to understand elementary reaction steps in proton-coupled electron transfer reactions.

Full text of DOI:10.1021/jacs.0c11992

pubs.acs.org/JACS
Article
Redox-Induced Structural Reorganization Dictates Kinetics of
Cobalt(III) Hydride Formation via Proton-Coupled Electron Transfer
Daniel A. Kurtz, Debanjan Dhar, Noémie Elgrishi, Banu Kandemir, Sean F. McWilliams,
William C. Howland, Chun-Hsing Chen, and Jillian L. Dempsey*
Cite This: J. Am. Chem. Soc. 2021, 143, 3393−3406
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: Two-electron, one-proton reactions of a family of
[CoCp(dxpe)(NCCH3)]2+ complexes (Cp = cyclopentadienyl,
dxpe = 1,2-bis(di(aryl/alkyl)phosphino)ethane) form the corre-
sponding hydride species [HCoCp(dxpe)]+ (dxpe = dppe (1,2-
bis(diphenylphosphino)ethane), depe (1,2-bis(diethylphosphino)-
ethane), and dcpe (1,2-bis(dicyclohexylphosphino)ethane))
through a stepwise proton-coupled electron transfer process. For
three [CoCp(dxpe)(NCCH3)]2+ complexes, peak shift analysis
was employed to quantify apparent proton transfer rate constants
from cyclic voltammograms recorded with acids ranging 22 pKa
units. The apparent proton transfer rate constants correlate with
the strength of the proton source for weak acids, but these
apparent proton transfer rate constants curiously plateau (kpl) as
the reaction becomes increasingly exergonic. The absolute apparent proton transfer rate constants across both these regions correlate
with the steric bulk of the chelating diphosphine ligand, with bulkier ligands leading to slower kinetics (kplateau,depe = 3.5 × 107 M−1
s−1, kplateau,dppe = 1.7 × 107 M−1 s−1, kplateau,dcpe = 7.1 × 104 M−1 s−1). Mechanistic studies were conducted to identify the cause of the
app
PT
aberrant k −ΔpKa trends. When deuterated acids are employed, deuterium incorporation in the Cp ring is observed, indicating
protonation of the CoCp(dxpe) species to form the corresponding hydride proceeds via initial ligand protonation. Digital
simulations of cyclic voltammograms show ligand loss accompanying initial reduction gates subsequent PCET activity at higher
driving forces. Together, these experiments reveal the details of the reaction mechanism: reduction of the Co(III) species is followed
by dissociation of the bound acetonitrile ligand, subsequent reduction of the unligated Co(II) species to form a Co(I) species is
followed by protonation, which occurs at the Cp ring, followed by tautomerization to generate the stable Co(III)-hydride product
[HCoCp(dxpe)]+. Analysis as a function of chelating disphosphine ligand, solvent, and acid strength reveals that the ligand
dissociation equilibrium is directly influenced by the steric bulk of the phosphine ligands and gates protonation, giving rise to the
plateau of the apparent proton transfer rate constant with strong acids. The complexity of the reaction mechanism underpinning
hydride formation, encompassing dynamic behavior of the entire ligand set, highlights the critical need to understand elementary
reaction steps in proton-coupled electron transfer reactions.
the application of electrochemical techniques to quantify the
catalytic performance metrics of first-row transition metal
hydrogen evolution catalysts,3,5−8 limited work has focused on
the specific parameters that influence the formation of metal
hydride reaction intermediates. In this context, our group
recently reported a systematic study on the effects of acid
strength and structure on the observed rate constant of proton
transfer (kaPpTp) of an electrochemically generated cobalt(I)
INTRODUCTION
■
Just like a chain is only as strong as its weakest link, catalysis is
only as fast as its slowest step(s). Consequently, understanding
the nature of the rate-limiting step(s) in a catalytic cycle and
the factors that influence it is crucial to improving the overall
efficiency of a catalyst. In fuel-forming reactions such as
hydrogen production or carbon dioxide reduction, the
formation of transition metal−hydride bonds is a common
elementary step that often has significant impact on the overall
rates of product formation.1−4 A transition metal hydride
complex is often formed through a proton-coupled electron
transfer process, via either a stepwise pathway (proton transfer
followed by electron transfer, PT-ET, or vice versa, ET-PT) or
a single concerted step (concerted proton−electron transfer,
CPET).4 While there have been numerous reports describing
Received: November 16, 2020
Published: February 23, 2021
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406
© 2021 American Chemical Society
3393

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
complex to form a stable cobalt(III) hydride complex from
[CoCp(dppe)(NCCH3)][PF6]2 (1, Cp = cyclopentadienyl,
dppe = 1,2-bis(diphenylphosphino)ethane).9 We observed a
linear free-energy relationship for relatively weak acids and
noted that sterically encumbered acids result in dramatically
slower protonation when compared to acids with similar pKa
geometric and structural rearrangements can occur during
acid/base reactions involving metal complexes, imparting
substantial reorganization energies that attenuate rate con-
stants of these reactions.
In this work, we interrogate the PCET reaction kinetics for a
series of three cobalt complexes with cyclopentadienyl and
chelating bisphosphine ligands that react via proton-coupled
electron transfer to form stable metal hydride complexes.
While not active catalysts, these model complexes enable the
PCET process associated with hydride formation to be
decoupled from catalyst turnover, and their sterics and
electronics are readily tunable through ligand substitution.
Here, by varying the identity of the phosphine substituents, the
steric bulk of the chelating bisphosphine ligand (generically
termed dxpe) was systematically perturbed, enabling us to
interrogate the role of structural changes in governing the
observed kinetics of hydride formation and gain deep insight
into the reaction pathway through which the hydride
complexes form. All three [CoCp(dxpe)(NCCH3)]2+ com-
plexes react via a stepwise electron transfer−electron transfer−
proton transfer pathway to form the corresponding [HCoCp-
(dxpe)]+ and exhibit a linear free-energy relationship (LFER)
between the apparent proton transfer rate constant (kaPpTp) and
values and no steric bulk near the acidic proton. More
app
PT
intriguingly, we also observed a distinct plateau of k values
with stronger acids that extends over 14.5 kcal/mol of driving
force for proton transfer. The second-order rate constants were
3 orders of magnitude lower than the calculated diffusion-
limited rate constant. This plateau region alluded to underlying
structural reorganization factors that limited the kinetics of
hydride formation, as the pKa of the added acid no longer
app
PT
affected the rate constant of proton transfer, but k remained
second order.
Aberrant rate constant−pKa relationships such as the one
highlighted above underscore the complexity of metal hydride
formation reactions. Both structural reorganization and
chemical steps such as ligand loss prior to formation of the
active catalyst can dictate the kinetics associated with the
formation of key intermediates in catalysis. This has previously
been observed for the CO2 reduction catalyst Re(bpy)-
(CO)3Cl. Upon reduction to [Re(bpy)(CO)3Cl]•−, slow
catalytic CO2 reduction is initiated, but the observed kinetics
are limited by slow chloride loss to allow for CO2
coordination.10 Similarly, the mechanism of oxidative addition
of haloarenes to trialkylphosphine palladium(0) complexes, a
key step in a suite of palladium-catalyzed cross-coupling
reactions, was found to be crucially dependent on the nature of
the halide, indicating that associative displacement of the
ligand in the precatalyst is a key step in this multistep catalytic
process.11,12 Additionally, in examples of olefin polymerization
using high-efficiency metallocene catalysts, backbone ligand
reorganization has been proposed as the rate-limiting step of
the reaction.13
proton source pKa at low driving force. As the proton transfer
app
PT
process becomes mildly exergonic (ΔpKa ≈ − 2), k becomes
driving force-independent. The phosphine substituents indeed
have a significant impact on absolute magnitude of the
observed proton transfer kinetics across both these regions, but
the general observation of a LFER and a plateau region in the
Brønsted plots remains invariant.
Reaction kinetics quantified in noncoordinating solvents
reveal that ligand loss accompanying initial reduction gates
subsequent PCET activity at higher driving forces. Compre-
hensive kinetics simulations incorporating this elementary step
and its independently measured rate constants and equilibrium
app
PT
constants rationalize the aberrant k −ΔpKa trends. The
ligand dissociation processwhich is accompanied by
substantial geometric rearrangementdirectly correlates to
the steric bulk of the bisphosphine ligand. Further, we unveil
evidence that protonation proceeds through the Cp ligand,
followed by tautomerization to form the hydride species, rather
than directly to the metal center, indicating a substantial
kinetic barrier to direct metal protonation. These data reveal a
reaction mechanism of hydride formation that illustrates the
potential for latent complexity in PCET reactions as the cobalt
complex negotiates a series of elementary reaction steps that
implicate the dynamic behavior of the entire ligand set despite
both the proton and electron ultimately residing on the metal
center. Ultimately, these conclusions underscore how
deceptively simple reaction schemes can veil complex
mechanisms and showcase how mechanistic studies can
elucidate the key roles both metal and ligand play in mediating
substrate transformations.
Additional nuanced behavior has been detected in proton
transfer reactions involving transition metal complexes
supported by pentamethylcyclopentadienyl and other sub-
stituted cyclopentadienyl ligands. In several examples of low-
valent complexes of cobalt,14−16 rhodium,17,18 and iron19,20
with pentamethylcyclopentadienyl (Cp*) or indenyl ligands,
initial protonation at the ring is preferential to protonation at
the metal center, even when the metal hydride complex is
thermodynamically more stable than the ring-protonated
species. In some cases, the participation of the Cp* ligand in
the initial protonation event and subsequent tautomerization
to form the metal hydride complex is proposed to be crucial for
hydrogen evolution catalysis.21 This observed proton relay
reactivity is analogous to that observed with intentionally
placed amine-based pendant bases in nickel-based hydrogen
evolution electrocatalysts.2 Peters and co-workers have
reported that protonation of Cp* rings may be important for
nitrogen reduction to NH3 where [Cp*Co(η4-C5Me5H)]+ is a
highly reactive PCET donor. Blakemore and co-workers have
shown that accompanying ligands can dramatically affect
whether Cp*-rhodium-based complexes are protonated or
not.22−24 The observation of kinetically facile ligand
protonation over more sluggish metal protonation is best
contextualized in the pioneering works of Norton and Pearson.
Protonation/deprotonation reactions involving metal com-
plexes are noted to be more similar to those involving carbon
acids than oxygen- or nitrogen-based acids.25−29 Large
RESULTS AND DISCUSSION
■
Synthesis and Characterization of the [CoCp(dxpe)-
(NCCH3)]2+ Complexes. Three cobalt(III) complexes with
bis(phosphino)ethane ligands generically termed dxpe were
synthesized through routes described previously involving a
common [CoCp(SMe2)3][PF6]2 precursor (1−3, Scheme
1).30 In comparison to the previously studied complex
[CoCp(dppe)(NCCH3)][PF6]2 (1), [CoCp(depe)-
(NCCH3)][PF6]2 (2, depe = 1,2-bis(diethylphosphino)-
3394
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
more upfield (−16.7 ppm for 5 and −16.5 ppm for 6) than the
aryl phosphine complex (−15.3 ppm for 4).
Scheme 1. Synthetic Routes to
[CoCp(dxpe)(NCCH3)][PF6]2 and [HCoCp(dxpe)][PF6]
Complexes.
Electrochemical Analysis in the Absence of Acid. To
probe the effects of the ligand electronics on the metal-
centered redox events, cyclic voltammograms were recorded
for complexes 1−3 in 0.25 M [Bu4N][PF6] acetonitrile
solution. Voltammograms of 2 and 3 exhibit two reversible
waves assigned to Co(III/II) and Co(II/I) couples, similar to
that reported previously for 1 (Figure 1, Table 1).9 For 2, both
ethane) has a less sterically bulky phosphine ligand, while
[CoCp(dcpe)(NCCH3)][PF6]2 (3, dcpe = 1,2-bis-
(dicyclohexylphosphino)ethane) is coordinated by a sterically
encumbered phosphine. The cobalt(III) hydride complexes
(4−6) were prepared and isolated by first combining
CoCp(CO)2 and dxpe in toluene and heating to generate
the corresponding cobalt(I) complexes CoCp(dxpe), followed
immediately by protonation with [NH4][PF6]. All complexes
were characterized using multinuclear NMR spectroscopy
(Figures S1−S16) and UV−visible absorbance spectroscopy
(Figures S17 and S18). Single crystals of the two new hydride
complexes 5 and 6 were grown and characterized by single-
crystal X-ray diffraction (SC-XRD) analysis (Figures S73 and
S74, bond distances and bond angles in Tables S10, S11, S13,
and S14). The hydride ligand for 5 was placed in its ideal
position and was first refined using individual relative isotropic
displacement parameters and later in fixed position to reduce
the number of parameters, and a Co−H bond length of 1.413
Å was determined without esds. The hydride ligand in 6 was
obtained from the Fourier difference map, its position was
refined, and a Co−H bond length of 1.35(5) Å was obtained.
While the SC-XRD method cannot yield accurate hydrogen
positions,31 the bond lengths are broadly in line with other
reported metal hydride bond distances, and the SC-XRD data
support the overall connectivity shown in Scheme 1.32
To probe the effects of the electronic perturbations caused
by altering the supporting bisphosphine ligand, the pKa values
of hydrides 5 and 6 were quantified via spectrophotometric
titration (Figures S19 and S20) and compared to the
previously reported pKa value of 4.9 The pKa values of 5 and
6, 23.6 and 22.6, respectively, indicate complexes with more
strongly donating alkyl phosphines are significantly more basic
than the previously reported aryl-substituted 4 (pKa = 18.4).9
Notably, these electronic effects are also manifested in the
hydride resonance in the 1H NMR spectrum of each complex;
the alkyl phosphine complexes exhibit hydride resonances
Figure 1. Cyclic voltammograms of 1−3 (1 mM) recording in 0.25 M
[NBu4][PF6] CH3CN at υ = 100 mV s−1. The internal standard Fc is
present in each scan, and all voltammograms are referenced the Fc+/
Fc redox couple.
Table 1. Electrochemical Properties of 1−3 in CH3CN
potential (V vs
Fc+/Fc)
diffusion coefficient
(D, cm2 s−1
)
ks (cm s−1
)
complex
(ligand)
Co(III/ Co(II/ Co(III/
Co(III/ Co(II/
II)
I)
II)
Co(II/I)
II)
I)
a
a
b
a
1 (dppe)
−0.51
−0.93 3.4 ×
5.2 ×
0.051
0.11
10−6
10−6
a
2 (depe)
−0.71
−0.39
−1.21 4.4 ×
6.7 ×
0.022
0.36
0.19
10−6
10−6
a
3 (dcpe)
−1.25
7.0 ×
10−6
a
b
E°′. E1/2, apparent potential, shifted from E° by ligand equilibrium.
the Co(III/II) (−0.71 V vs Fc+/0) and Co(II/I) (−1.21 V vs
Fc+/0) couples are shifted to more negative potentials
compared to 1 (−0.51 and −0.93 V vs Fc+/0, respectively),
as expected for the stronger alkyl phosphine donor ligand in
comparison to the arylphosphine.33 While the Co(II/I) couple
of the cyclohexyl-substituted phosphine derivative 3 (E1/2 =
−1.25 V vs Fc+/0) is similarly shifted to more negative
potentials in comparison to 1, the Co(III/II) couple appears at
a significantly more positive potential (E1/2 = −0.39 V vs
Fc+/0). Additionally, when voltammograms of the analyte
solution are recorded at various time delays (0−4.5 h) after
dissolution in CH3CN under an N2 atmosphere, the Co(III/II)
wave for 3 broadens and assumes a more quasi-reversible form
with ΔE > 57 mV (Figure S21). This anomalous behavior is
discussed below. The diffusion coefficients and heterogeneous
electron transfer rate constants for complexes 2 and 3 are
3395
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Scheme 2. Solvent Loss upon One-Electron Reduction of the Cobalt(III) Complexes
Scheme 3. Net Electrochemical Reaction Scheme in the Presence of Acid
similar to those reported previously for 1 (Figures S22−S25,
Table 1).9
The electrochemical behavior of 3 in both DFB and CH2Cl2
is starkly different from that observed for 1 and 2. The Co(III/
II) feature is irreversible as anticipated (Figure S29), but it
appears at a substantially more negative potential (Ep,c ≈ −0.56
V vs Fc+/0 at 200 mV s−1) when compared to that in CH3CN
(E1/2 = −0.39 V vs Fc+/0). This observation contradicts the
shift to more positive potentials that would be consistent with
the EC reactivity resulting from coordinated CH3CN loss upon
one-electron reduction. Further, titration of CH3CN into a
DFB solution of 3 does not result in enhanced reversibility of
the Co(III/II) voltammetric feature, indicating that CH3CN
loss from Co(II) is irreversible for 3 (Figure S29). We
hypothesize that for 3 the bulky dcpe ligand places CH3CN
coordination in equilibrium in the Co(III) oxidation state
([CoCp(dcpe)(NCCH3)]2+ ⇌ [CoCp(dcpe)]2+ + CH3CN),
consistent with speciation of related cobalt complexes.37 Slow
CH3CN coordination/dissociation explains the observed
broadening of the Co(III/II) redox feature in neat CH3CN
over time, as described above. In order to explore this
hypothesis, we compared computationally predicted free-
energy changes associated with dissociation of CH3CN in
both the Co(III) and Co(II) oxidation states for complexes 1−
3 (Table S1). These data show that while the loss of CH3CN
from Co(III) is significantly uphill for 1 and 2 (calculated ΔG
values of +19.6 and +27.1 kcal/mol, respectively), this reaction
is uphill only by 6.4 kcal/mol for 3, and thus at room
temperature [CoCp(dcpe)(NCCH3)]2+ is in equilibrium with
[CoCp(dcpe)]2+ (K ≈ 2 × 10−5 M). In contrast, the
corresponding reactions involving CH3CN loss in the Co(II)
oxidation state are all predicted to be exergonic, with a
substantially larger driving force for 3 (−15.4 kcal/mol) than 1
and 2 (−5.9 and −6.5 kcal/mol, respectively). Unfortunately,
owing to this complex ligation equilibrium of 3, we are unable
to determine kd for 3 using eq 1, as we do not know the true
E1/2 for 3 in the absence of any CH3CN loss.
Interrogating CH3CN Ligand Loss. For 1, reduction to
the Co(II) species leads to a concomitant dissociation of the
bound CH3CN (Scheme 2), which is detected by a loss of
reversibility for the Co(III/II) couple in voltammograms
recorded in a noncoordinating solvent such as CH2Cl2.9 While
previous experiments using 1 were performed in CH2Cl2, the
analogous measurement of 2 in CH2Cl2 was precluded by
noticeable reactivity, presumably with the solvent (Figure
S26). Therefore, ligand loss was studied in another non-
coordinating solvent (1,2-difluorobenzene, DFB), compatible
with both 1 and 2. Using the scan rate-dependent peak shift of
E
p,c (eq 1, where R, T, and F are the gas constant, temperature,
and Faraday’s constant, respectively),34 the rate constant for
ligand loss (kd) for 2 in DFB was found (kd = 6.8 × 104 s−1,
Figure S27). For 1, a kd value of 1.8 × 104 s−1 in DFB was
determined (Figure S28). Notably, the rate constant for 1 is ∼
3 orders of magnitude smaller in DFB compared to that in
CH2Cl2 (kd = 4 × 107 s−1). Comparison of the ethyl- and
phenyl- substituted complexes indicates that CH3CN dissoci-
ation is ∼4 times faster for 2 than 1 in DFB.
ikdRT y
RT
F
RT
2F
j
j
z
z
Epc = E1/2
−
(0.78) +
ln
j
j
z
z
Fυ
(1)
k
{
For both 1 and 2, titration of CH3CN to the analyte solution
results in regaining of reversibility of the Co(III/II) feature,
indicating that CH3CN loss is reversible (Figure S29). We
define Keq by eq 2, noting that for practical purposes the
equilibrium constant has units of M.35,36 With this expression
for Keq, the apparent E1/2 (E0′) relates to the true E1/2 of the
Co(III/II) couple through eq 3:
[Co(II)][CH3CN]
[Co(II) − NCCH3]
kd
k−d
Keq
=
=
(2)
Voltammetric Responses in the Presence of a Proton
Source. When cyclic voltammograms of 2 and 3 in CH3CN
are recorded in the presence of added acid, the Co(II/I) redox
feature becomes irreversible and the cathodic peak potential
shifts more positive, as was previously observed for 1.9 The
Co(II/I) peak position evolves as a function of scan rate,
consistent with an EC process in which an irreversible
chemical (C) step follows a fast electron transfer (E) (Scheme
3).
i
Keq
[CH3CN]
y
z
RT
F
j
E0′ = E1/2
+
lnj1 +
z
j
z
j
j
z
z
(3)
k
{
For 1, analysis of the E1/2 peak shift in DFB as a function of 1/
[CH3CN] (eq 3) yields an equilibrium constant value for
solvent dissociation (Keq) of 25 M (Figure S30), which is
slightly lower than the previously reported value in CH2Cl2
(Keq = 76 M).9 Similar analysis for 2 in DFB yields a Keq value
of 50 M (Figure S31). These results indicate that this
equilibrium constant for solvent dissociation is dependent on
both the nature of the bisphosphine ligand and the solvent.
Consistent with previously reported reactivity of 1,9 we
assign the chemical step following reduction of Co(II) as
protonation of the Co(I) species to form the corresponding
Co(III) hydride complex. The peak potential shifts as a
3396
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
app
PT
Figure 2. Brønsted plots for the apparent proton transfer rate constant k of Co(I) species using (left) nonbulky acids and (right) highlighting
sterically encumbered acids.
function of both acid concentration ([HA]) and scan rate (υ)
acids, we observe different linear free-energy relationships
according to
within the group of bulky acids for each complex with a lower
LFER slope (α ≈ 0.1−0.2), albeit with fewer data points than
app
i
j
j
y
z
z
app
kPT [HA]RT
Fυ
RT
F
RT
2F
the nonbulky LFER regions. Additionally, for k values that
PT
Epc = E1/2
−
(0.78) +
ln
j
z
z
j
are too low (in this case, when Ep,c < E1/2), the peak-shift
analysis breaks down, and thus for complex 3 with bulky acids,
the data are not reliable. Thus, we refrain from quantitative
(4)
k
{
where E1/2 is the half-wave potential of the one-electron
reversible electrochemical event in the absence of a coupled
the apparent rate constant for
protonation. The evolution of the peak potential with acid
conclusions from these bulky acid data.
For all complexes, second-order k
app
PT
chemical step, and k
app
PT
values become
independent of acid strength when the difference between
the pKa of the hydride complex and the acid exceeds ca. 3 pKa
units, leading to a distinct plateau in the Brønsted plot. Like
the LFER region, the maximum proton transfer rate constant
defining this plateau (kplateau) qualitatively correlates with the
steric bulk of the phosphine substituent; the ethyl-substituted
phosphine ligand (3.5 × 107 M−1 s−1) has the largest plateau
rate constant, followed by the phenyl-substituted ligand (1.7 ×
107 M−1 s−1) and the cyclohexyl-substituted ligand (7.1 × 104
34
app
PT
concentration or scan rate allows ready determination of k
.
Apparent proton transfer rate constants were determined for
2 and 3 for a range of acids spanning 22 pKa units in
acetonitrile (Figure 2, Tables S2 and S3, illustrative data in
Figures S33−S58). Rate constants are unique for each acid−
cobalt complex combination. These differences are anticipated,
as the driving force for proton transfer is quantified by the
difference in pKa values of the hydride complex and the acid.
To account for this, Brønsted plots are presented as k vs
ΔpKa, where ΔpKa is the difference between the pKa of the
added acid and the pKa of the Co(III) hydride species
[HCoCp(dxpe)]+ in order to facilitate comparison across the
three different complexes studied (Figure 2). The three
app
PT
M−1 s−1).
app
PT
To better quantify the observed correlations between k
values and steric bulk of 1, 2, and 3 in the LFER and plateau
regions, we sought to gain additional quantification of how
ligand sterics influence kinetics of metal protonation. Norton
has previously described the intrinsic barrier to protonation of
reduced metal complexes to form metal hydride com-
plexes.25,38−40 These intrinsic barriers are well described by
reorganization energies, which can be extracted from self-
Brønsted plots are similar in shape, each with three distinct
regions: (1) a linear free-energy relationship region where k
app
PT
varies with ΔpKa, (2) a region where proton transfer rate
constants plateau at mildly exergonic proton transfers (starting
at ∼ΔpKa = −2) with a “plateau rate constant” value (kplateau),
and (3) data points that align with neither region 1 or 2, but
share a commonality that the acids have steric bulk about the
acidic proton. The LFER region in each plot has a similar
Brønsted slope (α), with no significant trend in α values across
the series: α1 = 0.55, α2 = 0.51, and α3 = 0.45. These α values
are similar to those reported for protonation to form other
transition metal hydride complexes.29 Across the three
complexes, the absolute rate constants are consistently highest
for the depe ligand, followed by the dppe and the dcpe ligand,
qualitatively correlating with the steric bulk of the ligand.
Consistent with previous observations for 1, protonation
reactions of 2 and 3 with acids with steric bulk around the
acidic proton are slower than expected based on the acid pKa.
Using a series of para-substituted 2,6-dimethylpyridinium
1
exchange rate constants for proton transfer (kself). While H
NMR techniques such as variable-temperature NMR and line
broadening have previously been used to determine kself
values,29 no appreciable amount of broadening was observed
in the cyclopentadienyl resonances of mixtures of the studied
cobalt(III) hydride and cobalt(I) complexes up to 70 °C,
indicating the exchange was slower than could be quantified by
these tools. Two-dimensional exchange spectroscopy (2D-
1
EXSY) H NMR can measure rate constants on slower time
scales (0.01−100 s−1), although this technique has seen limited
application to measure rate constants for atom transfer
reactions.41,42 For the self-exchange reaction between the
cobalt(III) hydride and its conjugate base, kinetic information
can be obtained using appropriate mixing times and the
3397
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
relative integrals of the off-diagonal EXSY peaks in the 2D
NMR. To minimize the effects of longitudinal relaxation,
which are rather large in these cobalt complexes (T1 < 0.3 s),
the smallest mixing time that gave appreciable off-diagonal
EXSY peaks was used, typically 50−100 ms.41 Following these
protocols, kself values were quantified for [HCoCp(dppe)]+/
CoCp(dppe) and [HCoCp(depe)]+/CoCp(depe) in CD3CN/
C6D6 mixtures (1:1, for solubility, Figures S59 and S60). As
predicted, the kself value for the ethyl-substituted complexes
(kself‑2 = 61 M−1 s−1) was higher than that of the bulkier
To contextualize the [HCoCp(dxpe)]+/CoCp(dxpe) self-
exchange rate constants described above with this mechanistic
finding, we sought to assess whether self-exchange proceeds via
direct metal-to-metal proton transfer or via an initial ligand
protonation. When the deuteride reagent NaBD4 was added to
[CoCp(dppe)(NCCH3)]2+, [DCoCp(dppe)]+ was observed
as the exclusive product, with no deuterium incorporation in
the Cp ring detected (Figure S62). This result indicates that
Cp protons and the hydride proton do not exchange with each
other. However, when equimolar amounts of CoCp(dppe) and
[DCoCp(dppe)]+ were mixed, deuterium incorporation in the
Cp ring was observed (Figure S62), indicating that the proton
transfer self-exchange reaction proceeds via a two-step process
through initial Cp protonation followed by tautomerization to
yield the [DCoCp(dppe)]+. These data inform us that the self-
exchange proton transfer rate constants obtained via EXSY are
an apparent self-exchange rate constant and are not an accurate
measure of the direct proton transfer barrier.
While the deuterium incorporation data do not support a
change in the overall mechanism between the LFER and
plateau regions, we recognized that a change in rate-limiting
step(s) as a function of driving force is possible, given the
indirect, two-step hydride formation process elucidated. We
first hypothesized that in the LFER region (endergonic to
mildly exergonic proton transfer), the observed reaction
kinetics would be dominated by intermolecular proton transfer
from the acid to the cyclopentadienyl ring of CoCp(dxpe),
while in the plateau region (significantly exergonic proton
transfer) tautomerization from the [Co(CpH)(dxpe)]+ to the
[HCoCp(dxpe)]+ would control the rate law. If the rate of
phenyl-substituted complexes (kself‑1 = 3.1 M−1 s−1),
app
PT
supporting the conclusion that k values are influenced by
steric bulk. Extremely low solubility of CoCp(dcpe) thwarted
attempts to obtain the value of kself‑3 using this method.
Understanding the Plateau Region in the Brønsted
Plots. While kself values provide support and quantification for
app
PT
the observation that the absolute k values trend with ligand
sterics in the LFER region and the plateau region, they do not
app
PT
provide a rationale for the observation that k becomes
independent of the acid strength when the proton transfer
reaction becomes mildly exergonic. To understand this abrupt
plateau of rate constants, we considered how either a change in
mechanism or a change in rate law might give rise to driving
app
force-independent k values.
PT
Protonation of Co(I) Proceeds Indirectly via Ligand
Protonation. Preferential protonation of pentamethyl-
cyclopentadienyl and other cyclopentadienyl-based ligands
over a metal ion has previously been reported as discussed
above.14−17 We hypothesized that a change in mechanism
protonation at the metal center with weaker acids and
protonation at the cyclopentadienyl ligand with stronger
acids followed by a intramolecular proton transfer to the
metal centercould explain the two distinct regimes observed
in the Brønsted plots. To evaluate whether protonation
proceeded through the cyclopentadienyl ring, we treated
CoCp(dppe) with 1 equiv of d3-anilinium (pKa = 10.41). If the
metal was directly protonated, deuterium incorporation would
be observed exclusively in the metal hydride resonance.
However, deuterium incorporation in both the Cp ring and
tautomerization dictated the reaction kinetics, the acid pKa
app
PT
independence of k reflects the intrinsic barrier associated
with this intramolecular tautomerization. However, this model
is not consistent with the observation of an acid concentration-
dependent peak shift in the plateau region.9
Interrogation of the Rate Law. We next evaluated whether
the rate law differs between the LFER region and the plateau
app
region by quantifying the kinetic isotope effect (KIE) of k
.
PT
app
k
values were quantified with proteo (kH) and deuterated
PT
2
the hydride was observed in the H NMR spectrum of the
(kD) acids for two proton sources, one with a pKa value in the
LFER region (acetic acid, pKa = 23.51) and a second in the
plateau region of the Brønsted plot (anilinium, pKa = 10.41),
to compare KIE values of 1 in CH3CN. A normal, primary KIE
was observed for acetic acid (kH/kD = 4.5, Figure S63),
indicating that proton transfer is involved in the rate-limiting
step(s) in the LFER, while an inverse KIE (kH/kD = 0.75,
Figures S64 and S65) was quantified for anilinium, suggesting
that the rate-controlling process is different with stronger acids
that fall in the plateau region. Inverse KIE values often indicate
a pre-equilibrium process prior to a rate-limiting step involving
the proton transfer. For instance, the reductive elimination of
RH from the tungsten hydride complex [Me2Si(C5Me4)2]W-
(R)H has an inverse KIE, attributed to an inverse equilibrium
isotope effect in the multistep reaction and not due to kinetic
differences in a single step.43
reaction product, indicating that protonation proceeds with
participation of the cyclopentadienyl ligand (Figure S61). The
observation of both Cp-d1 and DCo(III) suggests either
protonation is unselective for the endo vs exo positions or the
two-step protonation is reversible. When the same reaction was
carried out using 10 equiv of the weaker acid CD3COOD (pKa
= 23.51), deuterium incorporation was again observed in both
the cyclopentadienyl and hydride resonances. These data
indicate that protonation of the CoCp(dppe) to form
[HCoCp(dppe)]+ proceeds via initial ligand protonation in
both the LFER and plateau regions, followed by intramolecular
proton transfer. DFT calculations indicate that both steps are
substantially exergonic (−17.1 and−18.6 kcal mol−1, respec-
tively; Scheme S2 and Table S8, Supporting Information).
Similar relative energetics are computed for 2 and 3, and we
thus expect this mechanism is operative for these complexes as
well. While these data do not support the hypothesis that a
In studies interrogating the formation of [HCoCp(dxpe)]+
described in this work, the Co(I) synthon CoCp(dxpe) is
accessed in situ via the two-electron reduction of the
[CoCp(dxpe)(NCCH3)]2+ precursor (Scheme 3). The reduc-
tion of the Co(III) species to Co(II) is accompanied by a
concomitant dissociation of the coordinated acetonitrile ligand
for 1 and 2 (see above); subsequent electron transfer occurs to
the unligated Co(II) complex. The reduction-induced ligand
change in initial protonation site is responsible for the abrupt
app
PT
change in k −pKa relationships observed at the junction of
the LFER and plateau regions, they do suggest a large kinetic
barrier for direct, metal-centered protonation of CoCp(dxpe),
a finding that has substantial ramifications for understanding
PCET reactions that form metal hydride complexes.
3398
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Scheme 4. Overall Mechanism for the Electrochemical Reaction
dissociation differs for the cyclohexyl-substituted complex 3, as
coordination of CH3CN to the Co(III) species is under
equilibrium for this complex (see above). Whereas the
app
PT
quantification of k discussed above is based on a simple
EC model (eq 4) and assumes the ligand dissociation does not
influence proton transfer kinetics, KIE data suggest ligand
dissociation from the electrochemically generated Co(II) may
play a crucial role in the observed kinetics of protonation. To
test this hypothesis, we performed digital simulations of cyclic
voltammograms based on an EC1EC2C3 reaction mechanism
that includes CH3CN ligand dissociation from the Co(II)
species (C1) and initial protonation on the Cp ring of the
Co(II) species (C2), followed by a tautomerization step (ktaut
)
that yields the Co(III) hydride (C3) (Scheme 4, detailed
expressions included in the Supporting Information).
Digital simulations of cyclic voltammograms were carried
out for the EC1EC2C3 PCET reactivity of 1 in bulk CH3CN
using the reaction mechanism depicted in Scheme 4 and
experimentally determined values of diffusion coefficients and
heterogeneous rate constants (Table 1). We are unable to
experimentally determine the rate constant (kd) and
equilibrium constant (Keq) for CH3CN dissociation in neat
CH3CN (see above), and therefore approximated those values
using the parameters quantified in CH2Cl2 (kd = 4 × 107 s−1
and Keq = 76 M) for this qualitative model.9 The final step of
the intramolecular proton tautomerization is assumed to be
irreversible. The specific rate constant for tautomerization
(ktaut) was found not to influence the simulations (104−1014
s−1) and generally set to 1 × 1014 s−1 in all simulations. Initial
simulations were conducted at a fixed scan rate (200 mV s−1)
for 0.5 mM 1 and 50 mM acid; the driving force and the rate
constant for protonation of CoCp(dppe) were varied by
changing the pKa of the acid (Figure S66).
Figure 3. Simulated cyclic voltammograms for 1 (0.5 mM) in
CH3CN in the presence of 50 mM acid at 200 mV s−1 with varying
input kPT values: 1 × 104 M−1 s−1 (red), 1 × 106 M−1 s−1 (yellow), 1 ×
109 M−1 s−1 (green), 5 × 109 M−1 s−1 (blue), and 1 × 1010 M−1 s−1
(purple). The dashed black trace is the simulated voltammogram in
the absence of any added acid. Simulations are carried out using the
reaction scheme and parameters listed in Scheme S1.
agreement with experimental observations.9 Importantly,
app
PT
when k values are determined from the Ep,c values of the
simulated data and eq 4, a reasonable agreement between the
app
PT
input kPT values and the calculated k is found (e.g., when the
k
k
PT input values are 100 and 1000 M−1 s−1, the corresponding
app
values are 98 and 640 M−1 s−1, respectively). This
First, cyclic voltammograms were simulated for acids aligned
with the LFER region of the Brønsted plots (8 ≥ ΔpKa ≥ −2).
On the basis of the slope of the Brønsted plot in the LFER
region (∼0.5), we assume that a 2-unit decrease in the ΔpKa is
PT
app
PT
indicates that in the LFER region the rate constants (k
)
obtained using the EC peak shift analysis discussed above are
indeed representative of the intrinsic proton transfer rate
constant kPT. On this basis, we attribute the differences
accompanied by a 10-fold increase in kPT (102 M−1 s−1≤ kPT
≤
107 M−1 s−1). In these simulations, the Ep,c value of the Co(II/
I) reduction wave shifts positive as the input value of kPT is
increased, consistent with experimental data which has been
interpreted by an increase in the apparent proton transfer rate
constant with stronger acids (Figure 3). In this region,
simulations predict that Ep,c is dependent on the scan rate
(Figure S67) and acid concentration (Figure S68), in
app
app
PT
observed in k for 1, 2, and 3 in the LFER regionwhere k
PT
values trend with the bulk of the ligandto differences in the
steric profile of the dxpe ligands. Steric bulk around the cobalt
center dictates the approach of molecular acids to the
nucleophilic CoCp(dxpe) in the bimolecular proton transfer
process, much like steric bulk around the acidic proton of the
acid does (see above).
3399
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 4. (Left) Simulated voltammograms for 0.5 mM 1 with varying amounts of acid at 200 mV s−1, using a kPT value of 1 × 109 M−1 s−1. (Right)
Variation of Ep,c in simulated voltammograms with log(acid concentration) input. The red line is a linear fit (R2 = 0.9989) with a slope of 30 mV
decade−1.
Next, voltammograms were simulated with the ΔpKa value
in the plateau region (−2 > ΔpKa ≥ −8), but with kPT values
extrapolated from the LFER region (assuming a theoretical
Brønsted slope of 0.5 that extends across the entire ΔpKa
range) until the diffusion-limited rate constant of ∼1 × 1010
M−1 s−1 is reached. The diffusion limit for these complexes in
this solvent system was previously estimated using the Debye−
Smoluchowski equation.9 In these simulated voltammograms,
Ep,c varies with scan rate, as anticipated for an EC reaction
(Figure S69) and observed experimentally. However, in
contrast to the simulations performed in the LFER region (8
≥ ΔpKa ≥ −2), the observed Ep,c of the Co(II/I) reduction
A simulated Brønsted plot was constructed from the
apparent rate constants (kaPpTp) obtained from the Ep,c values
of voltammograms simulated using the reaction mechanism
described in Scheme 4 for the reaction of 1 with acids spanning
a ΔpKa range of 16 units (8 ≥ ΔpKa ≥ −8) in CH3CN. This
app
PT
plot exhibits a plateau in k values for exergonic proton
transfer reactions (kplateau ≈ 8.5 × 106 M−1 s−1) and a Brønsted
slope of α = 0.42 in the LFER region (Figure 5). This
feature in the simulations do not vary as a function of the kPT
app
PT
values input (Figure 3), such that k is independent of the
input kPT values (and thus ΔpKa). Peak shift analysis (eq 4) of
app
the Ep,c values from the simulated voltammograms yields k
=
≤
PT
9 × 106 M−1 s−1 (for theoretical kPT input values 108 M−1 s−1
kPT ≤ 1010 M−1 s−1). The discrepancies between k and the
app
PT
input kPT indicate that in the plateau region the apparent
proton transfer rate constants are not a direct measurement of
the elementary proton transfer step. Surface concentration
profiles for these simulations explain this finding, showing that
app
PT
with large intrinsic kPT values, k values are dictated by the
parameters that describe both ligand dissociation from
[CoCp(dppe)(NCCH3)]+ and protonation of CoCp(dppe)
(Figure S72, Supporting Information).
Upon varying the acid concentration (0.3−100 mM) at a
fixed scan rate (200 mV s−1) with kPT = 1 × 109 M−1 s−1, the
evolution of Ep,c for the Co(II/I) feature shows intriguing
behavior in the simulated voltammograms (Figure 4). For acid
concentrations < 20 mM, the Ep,c shifts to more positive
potentials with increasing acid concentration, varying linearly
with log([HA]) with the slope of ∼30 mV decade−1 as
predicted by eq 4 for a canonical EC reaction (Figure 4). By
contrast, Ep,c becomes independent of acid concentration when
it exceeds 20 mM, implicating a change in the rate law. To
verify this behavior experimentally, we collected voltammo-
grams of 0.5 mM 1 recorded in CH3CN with high
concentrations of anilinium (pKa = 10.41); indeed Ep,c(Co-
(II/I)) becomes invariant with acid concentrations > 30 mM
(Figure S70).
Figure 5. Simulated Brønsted plots as a function of Keq for CH3CN
dissociation from [CoCp(dppe)(NCCH3)]+.
simulated Brønsted plot is in qualitative agreement with the
experimental plot (Figure 3, kplateau ≈ 1.7 × 107 M−1 s−1; α1 =
0.55); differences are attributed to the approximated kd and Keq
values used for the simulations. These simulations suggest that
dissociation of CH3CN from [CoCp(dppe)(NCCH3)]+ gates
protonation of CoCp(dppe), leading to the observed plateau
app
PT
of k values with stronger acids. To evaluate this further,
Brønsted plots were simulated as a function of both kd and Keq
values. Varying kd from 103 to 1014 s−1 led to no changes in the
app
calculated k values. However, Keq influences the absolute
PT
app
values of k in both the LFER and plateau regions of the
PT
3400
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Brønsted plot; higher Keq values correspond to higher values of
kaPpTp, with differences most pronounced in the plateau region
(Figure 5). With Keq = 10 M, kplateau is ∼4 × 105 M−1 s−1; a 10-
fold increase in Keq increases kplateau by nearly 2 orders of
magnitude (∼1 × 107 M−1 s−1). The observation that Keq, but
app
PT
not kd, influences k indicates that the thermodynamics (and
not kinetics) of this acetonitrile dissociation [CoCp(dppe)-
(NCCH3)]+ dictate the observed kinetics of CoCp(dppe)
protonation. Owing to the complex kinetics described by
Scheme 4, involving multiple competing chemical and
electrochemical reactions, we refrain from deriving a
quantitative rate law to describe this reaction. Nevertheless,
these comprehensive kinetics simulations support the con-
app
PT
clusion that the observed plateau in k values with stronger
acids reflects gating of CoCp(dppe) generation (and its
subsequent protonation) by CH3CN dissociation from
[CoCp(dppe)(NCCH3)]+.
app
Figure 6. k values for the protonation of CoCp(dppe) with selected
PT
acids in CH3CN, CH2Cl2, and DFB, plotted as a function of acid pKa
in CH3CN.
Protonation Kinetics in Noncoordinating Solvents in
the Plateau Region. As detailed above, the equilibrium
constant for the CH3CN dissociation (Keq) from [CoCp-
(dppe)(NCCH3)]+ is solvent dependent. On the basis of the
strong interdependence between the kplateau and Keq in the
simulated Brønsted plots (Figure 5), we hypothesized that
experimental Brønsted plots recorded in different solvents
app
PT
midium triflate; Tables S4 and S6). Indeed, k values were
independent of acid strength for these strong acids for both 1
and 2, and from these data kplateau values were determined
(Figure 7). The ratio of kplateau‑2/kplateau‑1 ≈ 3.3 in DFB
would enable us to support the conclusion that the kaPpTp, and
app
PT
the associated kplateau values, are dictated by Keq. k values for
1 with three acids in the plateau region (4-CF3-2,3,5,6-
tetrafluorophenol, p-toluenesulfonic acid, and dimethylforma-
midinium triflate, Tables S4 and S5) were determined from
voltammograms recorded as a function of scan rate and acid
concentration in both DFB and CH2Cl2. The selection of acids
was dictated by common solubility of acids in these two
solvents. As in CH3CN, the Co(II/I) couple becomes
irreversible in DFB and CH2Cl2 upon addition of acid and
the peak potential shifts as a function of acid concentration and
app
PT
scan rate, enabling ready determination of the apparent k via
eq 4. Notably, Ep,c becomes insensitive to acid concentration at
higher acid concentrations, as observed in CH3CN. For
instance, with a 1 mM solution of 1 in DFB, the Ep,c stops
changing when the concentration of 4-CF3-2,3,5,6-tetrafluor-
ophenol exceeds 50 mM (Figure S71).
app
PT
Comparison of k values for 1 in DFB and CH2Cl2 in the
plateau region (Figure 6) reveals that the observed plateau rate
constant (kplateau) in DFB (6 × 105 M−1s−1) is ∼7 times lower
than that in CH2Cl2 (4 × 106 M−1 s−1). This aligns with the
differences in experimentally determined Keq values for these
Figure 7. Side-by-side comparison of Brønsted plots for complexes 1
and 2 (left) in DFB with 3 acids in the plateau region (4-CF3-2,3,5,6-
tetrafluorophenol, p-toluenesulfonic acid, and dimethylformamidium
triflate) and (right) in CH3CN with all the plateau region acids shown
in Figure 2.
solvents (Keq = 76 M in CH2Cl2, 25 M in DFB), supporting
app
PT
the conclusions that k is influenced by CH3CN dissociation
from the electrochemically generated Co(II) species. Of note,
app
(kplateau‑2 = 2 × 106 M−1 s−1, kplateau‑1 = 6 × 105 M−1 s−1), in
qualitative agreement with their relative Keq values in DFB
(Keq‑2 = 50 M, Keq‑1 = 25 M). Collectively, these data indicate
that the steric profile of the chelating phosphine ligands
impacts the Keq for CH3CN dissociation.
The observation that the bulkier dppe ligand of 1
corresponds to a lower propensity (smaller Keq) for ligand
dissociation than the less bulky depe ligand of 2 initially seems
counterintuitive. However, recognizing that structural reorgan-
ization associated with ligand dissociation (which involves the
dxpe ligands moving into the Cp−Co plane) should influence
the equilibrium described by Keq, a smaller barrier for
geometric reorganization should correspond to a larger Keq.
Taking the self-exchange rate constants for 1 and 2 (kself‑1 = 3.1
k
values for weaker acids (acetic acid, 2,3,4,6-tetrafluor-
PT
ophenol, and salicylic acid) determined in DFB indicate an
LFER region is accessible, similar to CH3CN data (Figure 6,
Table S5).
Returning to the observation that the kplateau values for 1, 2,
and 3 span nearly 3 orders of magnitude in CH3CN, we
hypothesized that the differences in chelating phosphine
ligands influence the Keq values for acetonitrile dissociation
from the corresponding [CoCp(dxpe)(NCCH3)]+ species, and
this in turn gives rise to the dramatic differences in kplateau
values. Because Keq values are not tractable in neat CH3CN, we
app
PT
compared k values for 1 and 2 in DFB with three acids
expected to fall in the plateau region (4-CF3-2,3,5,6-
tetrafluorophenol, p-toluenesulfonic acid, and dimethylforma-
3401
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
M−1 s−1; kself‑2 = 61 M−1 s−1; see above) as a proxy for
geometric reorganization energies, the observed trends
between Keq and ligand sterics are readily justified. While
neither Keq nor kself‑3 was experimentally accessible for the dcpe
complex 3, we predict that the steric bulk of the cyclohexyl
substituents will lead to a kself‑3 that is substantially smaller than
sources with steric bulk around the acidic proton are utilized,
app
k
is dramatically attenuated for all three complexes, with acid
PT
pKa values spanning across both regions.
KIE experiments in the LFER region and the plateau region
reveal that there is a different rate-limiting step in the two
regimes, a curious finding for a seemingly simple proton
transfer reaction. When deuterium-labeled acids were used to
protonate isolated Co(I) complexes, deuterium incorporation
into the Cp ring was observed, indicating CpH formation
followed by tautomerization to form HCo(III). This indicates
an intrinsic barrier to direct metal protonation, a finding that
has ramifications on the design of catalysts for fuel production
that proceeds via metal hydride intermediates. However, the
ligand-based protonation observed in this and other cyclo-
pentadienyl−metal complexes suggests there may be oppor-
tunities to exploit metal−ligand cooperativity to overcome
these intrinsic barriers through alternative reaction pathways,
and future work in this area should examine these
opportunities.
kself‑1 and kself‑2, and in turn a smaller Keq. Thus, the observed
app
PT
differences in k for 1, 2, and 3 across both the LFER and
plateau regions are best described by differences in the relative
K
eq values, which are dictated by the geometric reorganization
accompanying CH3CN dissociation, such that less bulky
ligands have larger kplateau values.
Together, observed trends between k and ligand sterics;
app
PT
successful digital simulation of experimental voltammograms as
a function of acid concentration, scan rate, kPT, and Keq;
correlation between Keq and kplateau values as a function of both
solvent and ligand sterics; and deuterium exchange reactions
support the conclusion that the proton-coupled electron
transfer reaction to generate [HCoCp(dxpe)]+ is well-
described by the multistep reaction depicted in Scheme 4. In
this mechanism, a bound acetonitrile must dissociate from the
electrochemically generated [CoCp(dxpe)(NCCH3)]+ species
before undergoing subsequent reduction and protonation.
Further, the initial protonation occurs at the Cp ligand,
followed by an intramolecular proton transfer to form the
product [HCoCp(dxpe)]+. Crucially, these data support the
conclusion that CH3CN dissociation from [CoCp(dxpe)-
(NCCH3)]+ gates the proton transfer to CoCp(dxpe) when
Digital simulations revealed further mechanistic nuances;
dissociation of a bound CH3CN upon reduction of Co(III) to
Co(II) gates subsequent electron and proton transfer. When
app
PT
the PCET reaction is endergonic or mildly exergonic, k
reflects the intrinsic rate constant for protonation of Co(I).
However, when the rate constant for protonation increases, the
apparent kinetics of protonation are influenced by the
equilibrium constant for ligand dissociation from Co(II)
(described by Keq). These mechanistic findings are supported
app
PT
the intermolecular proton transfer rate constants are very fast,
by correlations between experimental Keq and k values in the
app
leading to a plateau in k with stronger acids. In the LFER
plateau region as a function of both solvent and chelating
phosphine ligand. These findings emphasize that seemingly
simple proton-coupled electron transfer reactions may proceed
via unexpected elementary reaction steps, and a detailed
understanding of reaction mechanisms is critical to advance
our understanding and the development of catalysts that
mediate proton-coupled electron transfer processes.
PT
app
PT
region, the k values determined are reflective of the intrinsic
rate constant for protonation of CoCp(dxpe) (kPT), but as kPT
app
PT
becomes larger, k
is dictated by the Keq for CH3CN
dissociation (supported by correlations between kplateau and Keq
values as a function of ligand and solvent) and kPT (supported
by the acid concentration influence on Ep,c, which gives rise to
app
second-order k values at low to moderate acid concen-
PT
app
PT
trations). At very high acid concentrations, k becomes acid-
EXPERIMENTAL METHODS
■
independent and is solely dictated by Keq.
General Considerations. All experiments were performed in a
nitrogen-filled glovebox, unless otherwise noted. All glovebox solvents
were degassed with argon and dried using a Pure Process Technology
solvent purification system. Tetrabutylammonium hexafluorophos-
phate (TCI, 98%) was recrystallized from ethanol. 1,2-Bis-
(diphenylphosphino)ethane (Ark Pharm, 97%), 1,2-bis-
(diethylphosphino)ethane (Sigma-Aldrich, 97%), 1,2-bis-
(dicyclohexylphosphino)ethane (Acros Organics, 98%), dimethyl
sulfide (Sigma-Aldrich, anhydrous ≥ 99%), and cyclopentadienyl-
cobalt dicarbonyl (Strem, >95%) were used as received. Acetonitrile-
d6 (99.8% D) and benzene-d6 was purchased from Cambridge Isotope
Laboratories.
CONCLUSIONS
■
A series of Co(III) complexes with the general formula
[CoCp(dxpe)(NCCH3)]2+ (1−3) were prepared with a series
of phosphine ligands with substituents that vary in their steric
bulk (ethyl, phenyl, and cyclohexyl). Two-electron, one-proton
PCET reactivity of these complexes leads to the formation of
the corresponding hydride species [HCoCp(dxpe)]+. Through
cyclic voltammetry measurements recorded in the presence of
molecular acids of varying strength, reactivity was determined
to proceed via two-electron reduction of the Co(III)
precursors followed by protonation to form the corresponding
1H, 13C{1H}, and 31P{1H} NMR spectra were collected on either a
Bruker 600 or Bruker 500 MHz spectrometer at 295 K. 2D-EXSY
NMR spectra were collected on a Bruker 700 MHz spectrometer at
295 K. Chemical shifts are reported relative to residual proteo solvent
signals. UV−vis absorption spectra were collected using either an
Agilent Cary 60 UV−vis absorbance spectrophotometer or an
OceanOptics FLAME-S-UV−vis spectrometer with a DH-mini light
source.
Samples were analyzed with a Q Exactive HF-X (ThermoFisher,
Bremen, Germany) mass spectrometer. Samples were introduced via a
heated electrospray source (HESI) at a flow rate of 10 μL/min. One
hundred time domain transients were averaged in the mass spectrum.
HESI source conditions were set as follows: nebulizer temperature
100 °C, sheath gas (nitrogen) 15 arb, auxiliary gas (nitrogen) 5 arb,
sweep gas (nitrogen) 0 arb, capillary temperature 250 °C, RF voltage
HCo(III) complexes. From electrochemical peak shift analysis
app
of the Co(II/I) reduction step, k values were determined for
PT
app
PT
each complex. Brønsted plots correlating k vs ΔpKa show
stark differences based on the steric bulk of the phosphine
ligand: the ethyl-substituted complex 2 has higher k values in
app
PT
comparison to the phenyl-substituted complex 1, and the bulky
app
cyclohexyl-substituted complex 3 has much lower k values.
PT
app
PT
For all complexes, k vs ΔpKa exhibited a linear free-energy
relationship for endergonic to mildly exergonic protonation
reactions (when ΔpKa > −2), and k values plateaued for
protonation reactions with larger driving forces at values well
app
PT
below the diffusion limit for proton transfer. When proton
3402
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
100 V. The mass range was set to 600−2000 m/z. All measurements
were recorded at a resolution setting of 120 000. Solutions were
analyzed at 0.1 mg/mL or less based on responsiveness to the ESI
mechanism. Xcalibur (ThermoFisher, Breman, Germany) was used to
analyze the data. Molecular formula assignments were determined
with Molecular Formula Calculator (v 1.2.3). All observed species
were singly charged, as verified by unit m/z separation between mass
spectral peaks corresponding to the 12C and 13C12Cc‑1 isotope for each
elemental composition.
Sample Preparation for 2D EXSY Experiments. To a
suspension/solution of [HCoCp(dxpe)][PF6] in THF was added
an excess of NaH (10−100 equivalents). The suspension was stirred
overnight at room temperature in the glovebox, which was
accompanied by a color change from yellow to red. The mixture
was dried under vacuum and resuspended in toluene. The mixture
was stirred overnight, followed by filtration to get rid of excess NaH,
unreacted [HCoCp(dxpe)][PF6], and NaPF6. The filtrate was then
dried under vacuum to afford CoCp(dxpe). This was then dissolved
in the NMR solvent along with a known amount of [HCoCp(dxpe)]-
[PF6] solid and a known amount of diethyl ether as an internal
standard.
confirmed to be minima on the potential energy surface by the
absence of imaginary frequencies after numerical frequency
calculations on the optimized structures.
Synthesis of [CoCp(depe)(NCCH3)][PF6]2 (2). To a stirring solution
of [CoCp(SMe2)3][PF6]2 (202 mg, 0.34 mmol) in 5 mL of CH3CN
was added slowly depe (80 μL, 0.34 mmol) with an accompanying
color change from pink to dark orange. The solution was stirred at
295 K for 48 h, filtered, and then evaporated to dryness.
Dichloromethane (10 mL) was added to the residue, followed by
sonication until a free-flowing powder was obtained. The powder was
collected and redissolved in minimal CH3CN, and the solution was
added dropwise to stirred diethyl ether to obtain a bright orange
powder. The orange powder was dried under reduced pressure and
was obtained in 60% yield. 1H NMR (600 MHz, CD3CN): δ 5.71 (s,
5H), 2.45 (dp, J = 15.0, 7.4 Hz, 2H), 2.35−2.20 (m, 4H), 2.11−1.98
(m, 6H), 1.97 (s, 3H), 1.34 (dt, J = 19.0, 7.7 Hz, 6H), 1.27 (td, J =
16.4, 7.4 Hz, 6H). 13C{1H} NMR (151 MHz, CD3CN): δ 137.29,
90.27, 21.87 (dd, J = 21.4, 18.3 Hz), 20.43 (dd, J = 15.6, 12.8 Hz),
17.88 (dd, J = 15.0, 12.2 Hz), 7.78, 4.59 (p, J = 21.4 Hz). 31P{1H}
NMR (243 MHz, CD3CN): δ 89.72 (2P), −145.00 (hept, J = 705.7
Hz, 2P). Anal. Calcd for C17H32CoF12NP4: C, 30.88; H, 4.88, N, 2.12.
Found: C, 30.79; H, 4.98, N, 2.03.
Electrochemical Methods. Electrochemical measurements were
performed using published procedures.9 All measurements were
performed in a N2-filled glovebox with a WaveDriver (Pine Research)
potentiostat using a 3 mm diameter carbon working electrode, a 3 mm
diameter glassy carbon counter electrode, and a silver wire pseudo-
reference electrode. A 20 mL scintillation vial was used as an
electrochemical cell, fitted with a custom-made Teflon cap to hold the
three electrodes. The electrode leads in the glovebox were connected
to the potentiostat with a custom shielded electrode cable feed-
through. All scans were referenced to the ferrocenium/ferrocene
couple at 0 V. Acetonitrile (Fisher Scientific, HPLC grade, >99.9%)
for electrochemical experiments was dried and degassed using a Pure
Process Technology solvent purification system. Ferrocene was
present in each scan unless otherwise noted. Ohmic drop was
minimized using a high electrolyte concentration (0.25 M tetrabutyl-
ammonium hexafluorophosphate, [NBu4][PF6]), through minimiza-
tion of the distance between the working and reference electrodes,
and through manual iR compensation. Glassy carbon electrodes (CH
Instruments, 3 mm diameter disk) were polished with 0.05 μm
alumina powder (CH Instruments, contained no agglomerating
agents) Milli-Q water slurries, rinsed, and ultrasonicated briefly in
Milli-Q water to remove residual polishing powder. The silver wire
pseudo-reference electrode was submerged in a glass tube containing
electrolyte (0.25 M [NBu4][PF6] in acetonitrile, CH3CN) and
separated from the solution with a porous glass Vycor tip. The
working electrode was pretreated with cyclical scans from
approximately 2 to −2 V (the exact value varied in accordance with
the silver wire pseudo-reference) at 250 mV s−1 in 0.25 M
[NBu4][PF6] until cycles were superimposable (typically achieved
within three cycles).
Synthesis of [CoCp(dcpe)(NCCH3)][PF6]2 (3). To a stirring solution
of [CoCp(SMe2)3][PF6]2 (87.9 mg, 0.14 mmol) in 5 mL of CH3CN
was added dcpe (62 mg, 0.15 mmol) with an accompanying color
change from pink to dark orange. The solution was stirred at 295 K
overnight, after which is was filtered. The filtrate was added dropwise
to diethyl ether, and the resulting solid was collected and dried via
filtration. The solid was then suspended in chloroform, sonicated,
filtered, and washed with chloroform, then dried under high vacuum
to give 3 as an orange solid in 70% yield. 1H NMR (600 MHz,
CD3CN): δ 5.88 (s, 5H), 2.26−2.11 (m, 4H), 2.07−1.98 (m, 4H),
1.96 (s, 3H), 1.93−1.85 (m, 14H), 1.79−1.50 (m, 10H), 1.46−1.29
(m, 16H). 13C{1H} NMR (151 MHz, CD3CN): δ 139.53, 90.00,
40.96 (dd, J = 11.4, 9.3 Hz), 39.33 (dd, J = 11.5, 9.2 Hz), 30.22, 30.10
(td, J = 60.6, 2.7 Hz), 29.65, 28.05 (t, J = 6.1 Hz), 27.79 (t, J = 5.3
Hz), 27.64 (t, J = 5.4 Hz), 27.55 (t, J = 5.7 Hz), 26.05 (d, J = 33.0
Hz), 21.34 (dd, J = 19.4, 15.5 Hz). 31P{1H} NMR (243 MHz,
CD3CN): δ 92.49 (2P), −145.00 (hept, J = 705.7 Hz, 2P). Anal.
Calcd for C33H56CoF12NP4: C, 45.16; H, 6.43, N, 1.60. Found: C,
45.80; H, 6.76, N, 0.48.
Synthesis of [HCoCp(depe)][PF6] (5). CoCp(CO)2 (0.25 mL, 1.87
mmol) and depe (0.44 mL, 1.88 mmol) were combined in 20 mL of
toluene in a bomb flask. The bomb flask was put on the Schlenk line
under an N2 atmosphere and heated at 105 °C for 30 min, wherein
the color changed from light red to dark red with bubbling for the first
10 min. After the reaction flask was cooled to room temperature, the
bomb flask was brought back into the glovebox and the reaction
mixture was diluted with 20 mL of methanol. Solid NH4PF6 (618 mg,
3.79 mmol) was added to the mixture, and it was stirred overnight,
wherein the color changed from dark red to dark orange with a small
amount of yellow precipitate. The entire mixture was then added to
200 mL of diethyl ether with vigorous stirring to yield a pale yellow
powder. The yellow solid was filtered, rinsed with ether, and dried
under high vacuum (79%). Vapor diffusion of diethyl ether into a
concentrated solution of the solid in CH3CN yielded large orange
Digital Simulations. Digital simulations of cyclic voltammograms
were performed with DigiElch 8.FD electrochemical simulation
software (ElchSoft through Gamry Instruments). Simulation param-
eters are detailed in Scheme S1 (Supporting Information).
Computational Details. DFT calculations were performed with
the ORCA program package, version 4.0.1.44 Optimized geometries
were computed using the BP86 functional.45,46 Atom-pairwise
dispersion correction with the Becke−Johnson damping scheme
(D3BJ),47,48 the scalar relativistic zero-order regular approximation
(ZORA),49 and the scalar relativistically recontracted version of the
Aldrichs triple-ζ basis set, def2-TZVP, were used on all atoms with an
auxiliary basis set of def2/J.50 The conductor-like polarizable
continuum model51 (CPCMC) was used to simulate an acetonitrile
solution (ε = 36.6). Resolution of identity (RI) was used to
approximate two electron integrals for geometry optimizations and
the numerical frequency calculations. The SCF calculations were
tightly converged (TightSCF). Optimizations were tightly converged
(TightOpt). Numerical integrations during all DFT calculations were
done on a dense grid (ORCA grid4). The calculated structures were
1
crystals of 5. H NMR (500 MHz, CD3CN): δ 5.12 (s, 5H), 2.07
(ddt, J = 15.0, 10.2, 7.6 Hz, 2H), 1.99−1.90 (m, 5H), 1.88−1.76 (m,
4H), 1.74−1.57 (m, 2H), 1.12 (td, J = 16.3, 7.6 Hz, 6H), 1.07 (td, J =
18.2, 7.6 Hz, 6H), −16.73 (t, J = 69.7 Hz, 1H). 13C{1H} NMR (126
MHz, CD3CN): δ 85.00, 24.68 − 24.44 (m), 24.35 (dd, J = 7.4, 3.9
Hz), 24.19 (d, J = 3.4 Hz), 8.80 (t, J = 1.4 Hz), 8.68 (t, J = 2.7 Hz).
31P{1H} NMR (202 MHz, CD3CN): δ 95.58 (1P), −145.00 (hept, J
= 706.5 Hz, 2P). HRMS: (HESI/Orbitrap) m/z [M − PF6]+ calcd for
C15H30P2Co 331.11547; found 331.11472.
Synthesis of [HCoCp(dcpe)][PF6] (6). CoCp(CO)2 (0.23 mL, 1.72
mmol) and dcpe (0.704 mg, 1.67 mmol) were combined in 10 mL of
toluene in a bomb flask. The bomb flask was put on the Schlenk line
under an N2 atmosphere and heated at 100 °C for 3 h, wherein the
color changed from light red to dark red with bubbling for the first 10
3403
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
min. After the flask was cooled to room temperature, it was brought
back into the glovebox and NH4PF6 (0.285 mg, 1.75 mmol) was
added to the toluene solution, followed by 10 mL of methanol. The
suspension was stirred vigorously in the glovebox for 16 h, which
resulted in the precipitation of a yellow solid. The pale yellow solid
was filtered, rinsed with diethyl ether, and dried under high vacuum
(74% yield). Vapor diffusion of diethyl ether into a concentrated
solution of the solid in CH3CN yielded orange crystals of 6. 1H NMR
(500 MHz, CD3CN): δ 5.20 (s, 5H), 2.08−1.96 (m, 6H), 1.94−1.76
(m, 14H), 1.76−1.68 (m, 4H), 1.56−1.44 (m, 4H), 1.43−1.23 (m,
20H), −16.54 (t, J = 69.8 Hz, 1H). 13C{1H} NMR (126 MHz,
CD3CN): δ 84.83, 38.85 (td, J = 18.1, 8.1 Hz), 37.01 (td, J = 18.5, 8.1
Hz), 29.52, 29.47, 29.04 (t, J = 2.4 Hz), 28.94, 27.72 (td, J = 12.6, 6.1
Hz), 27.21 (dt, J = 19.2, 5.9 Hz), 26.64 (d, J = 2.3 Hz), 22.62 (dd, J =
23.4, 17.0 Hz). 31P{1H} NMR (202 MHz, CD3CN): δ 107.26,
−145.00 (hept, J = 706.5 Hz). HRMS: (HESI/Orbitrap) m/z [M −
PF6]+ calcd for C31H54P2Co 547.30327; found 547.30263.
27599-3290, United States; orcid.org/0000-0003-0150-
9557
Complete contact information is available at:
https://pubs.acs.org/10.1021/jacs.0c11992
Author Contributions
All authors have given approval to the final version of the
manuscript.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the U.S. Department of Energy,
Office of Science, Office of Basic Energy Sciences, under
Award No. DE-SC0015303 and the University of North
Carolina at Chapel Hill. J.L.D. acknowledges support from a
Packard Fellowship in Science and Engineering. D.A.K.
acknowledges the support of the Kenan Graduate Fellowship,
and W.C.H. acknowledges the support of the Taylor
Fellowship. We thank the University of North Carolina’s
Department of Chemistry NMR Core Laboratory for their use
of their NMR spectrometers, supported by the National
Science Foundation under Grant No. CHE-0922858. We
thank the University of North Carolina’s Department of
Chemistry Mass Spectrometry Core Laboratory, especially
Brandie Ehrmann, for their assistance with mass spectrometry
analysis, supported by the National Science Foundation under
Grant No. CHE-1726291. We thank the University of North
Carolina Medical Biomolecular NMR Laboratory Core Facility
for the use of their NMR spectrometer, supported by the
National Cancer Institute of the National Institutes of Health
under Award No. P30CA016086. The content is solely the
responsibility of the authors and does not necessarily represent
the official views of the National Institutes of Health. We thank
Shannon Stahl, Clark Landis, James Gerkin, and Alexander
Miller for insightful discussions.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/jacs.0c11992.
■
sı
Further experimental details, characterization data,
additional experimental and simulated cyclic voltammo-
grams (PDF)
Accession Codes
CCDC 2040905−2040906 contain the supplementary crys-
tallographic data for this paper. These data can be obtained
free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by
emailing data_request@ccdc.cam.ac.uk, or by contacting the
Cambridge Crystallographic Data Centre, 12 Union Road,
Cambridge CB2 1EZ, UK; fax: +44 1223 336033.
AUTHOR INFORMATION
Corresponding Author
■
Jillian L. Dempsey − Department of Chemistry, University of
North Carolina at Chapel Hill, Chapel Hill, North Carolina
27599-3290, United States; orcid.org/0000-0002-9459-
4166; Email: dempseyj@email.unc.edu
REFERENCES
Authors
■
Daniel A. Kurtz − Department of Chemistry, University of
North Carolina at Chapel Hill, Chapel Hill, North Carolina
27599-3290, United States; orcid.org/0000-0002-0082-
1699
Debanjan Dhar − Department of Chemistry, University of
North Carolina at Chapel Hill, Chapel Hill, North Carolina
27599-3290, United States
Noémie Elgrishi − Department of Chemistry, University of
North Carolina at Chapel Hill, Chapel Hill, North Carolina
27599-3290, United States; Department of Chemistry,
Louisiana State University, Baton Rouge, Louisiana 70803,
United States; orcid.org/0000-0001-9776-5031
Banu Kandemir − Department of Chemistry, University of
North Carolina at Chapel Hill, Chapel Hill, North Carolina
27599-3290, United States
Sean F. McWilliams − Department of Chemistry, University of
North Carolina at Chapel Hill, Chapel Hill, North Carolina
27599-3290, United States
William C. Howland − Department of Chemistry, University
of North Carolina at Chapel Hill, Chapel Hill, North
Carolina 27599-3290, United States
(1) Dempsey, J. L.; Brunschwig, B. S.; Winkler, J. R.; Gray, H. B.
Hydrogen Evolution Catalyzed by Cobaloximes. Acc. Chem. Res. 2009,
42 (12), 1995−2004.
(2) Helm, M. L.; Stewart, M. P.; Bullock, R. M.; DuBois, M. R.;
DuBois, D. L. A Synthetic Nickel Electrocatalyst with a Turnover
Frequency Above 100,000 s−1 for H2 Production. Science 2011, 333
(6044), 863−866.
(3) Bediako, D. K.; Solis, B. H.; Dogutan, D. K.; Roubelakis, M. M.;
Maher, A. G.; Lee, C. H.; Chambers, M. B.; Hammes-Schiffer, S.;
Nocera, D. G. Role of Pendant Proton Relays and Proton-Coupled
Electron Transfer on the Hydrogen Evolution Reaction by Nickel
Hangman Porphyrins. Proc. Natl. Acad. Sci. U. S. A. 2014, 111 (42),
15001−15006.
(4) Dempsey, J. L. Proton-Coupled Electron Transfer: Metal
Hydrides Find the Sweet Spot. Nat. Chem. 2015, 7 (2), 101−102.
(5) Connor, G. P.; Mayer, K. J.; Tribble, C. S.; McNamara, W. R.
Hydrogen Evolution Catalyzed by an Iron Polypyridyl Complex in
Aqueous Solutions. Inorg. Chem. 2014, 53 (11), 5408−5410.
(6) Lewandowska-Andralojc, A.; Baine, T.; Zhao, X.; Muckerman, J.
T.; Fujita, E.; Polyansky, D. E. Mechanistic Studies of Hydrogen
Evolution in Aqueous Solution Catalyzed by a Tertpyridine-Amine
Cobalt Complex. Inorg. Chem. 2015, 54 (9), 4310−4321.
(7) Elgrishi, N.; Rountree, K. J.; McCarthy, B. D.; Rountree, E. S.;
Eisenhart, T. T.; Dempsey, J. L. A Practical Beginner’s Guide to
Cyclic Voltammetry. J. Chem. Educ. 2018, 95 (2), 197−206.
Chun-Hsing Chen − Department of Chemistry, University of
North Carolina at Chapel Hill, Chapel Hill, North Carolina
3404
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
(8) Elgrishi, N.; McCarthy, B. D.; Rountree, E. S.; Dempsey, J. L.
Reaction Pathways of Hydrogen-Evolving Electrocatalysts: Electro-
chemical and Spectroscopic Studies of Proton-Coupled Electron
Transfer Processes. ACS Catal. 2016, 6 (6), 3644−3659.
(9) Elgrishi, N.; Kurtz, D. A.; Dempsey, J. L. Reaction Parameters
Influencing Cobalt Hydride Formation Kinetics: Implications for
Benchmarking H2-Evolution Catalysts. J. Am. Chem. Soc. 2017, 139
(1), 239−244.
(10) Johnson, F. P. A.; George, M. W.; Hartl, F.; Turner, J. J.
Electrocatalytic Reduction of CO2 Using the Complexes [Re(Bpy)-
(CO)3L]n (n = + 1, L = P(OEt)3, CH3CN; n = 0, L = Cl−,Otf− ; Bpy
= 2,2′-Bipyridine; Otf− = CF3SO3) as Catalyst Precursors: Infrared
Spectroelectrochemical Investigation. Organometallics 1996, 7333 (9),
3374−3387.
(11) Barrios-Landeros, F.; Carrow, B. P.; Hartwig, J. F. Effect of
Ligand Steric Properties and Halide Identity on the Mechanism for
Oxidative Addition of Haloarenes to Trialkylphosphine Pd(0)
Complexes. J. Am. Chem. Soc. 2009, 131 (23), 8141−8154.
(12) McMullin, C. L.; Fey, N.; Harvey, J. N. Computed Ligand
Effects on the Oxidative Addition of Phenyl Halides to Phosphine
Supported Palladium(0) Catalysts. Dalt. Trans. 2014, 43 (36),
13545−13556.
(13) Zaccaria, F.; Cipullo, R.; Budzelaar, P. H. M.; Busico, V.; Ehm,
C. Backbone Rearrangement during Olefin Capture as the Rate
Limiting Step in Molecular Olefin Polymerization Catalysis and Its
Effect on Comonomer Affinity. J. Polym. Sci., Part A: Polym. Chem.
2017, 55 (17), 2807−2814.
(14) Chalkley, M. J.; Oyala, P. H.; Peters, J. C. Cp* Noninnocence
Leads to a Remarkably Weak C-H Bond via Metallocene Protonation.
J. Am. Chem. Soc. 2019, 141 (11), 4721−4729.
(15) Chalkley, M. J.; Del Castillo, T. J.; Matson, B. D.; Peters, J. C.
Fe-Mediated Nitrogen Fixation with a Metallocene Mediator:
Exploring pKa Effects and Demonstrating Electrocatalysis. J. Am.
Chem. Soc. 2018, 140 (19), 6122−6129.
(16) Chalkley, M. J.; Del Castillo, T. J.; Matson, B. D.; Roddy, J. P.;
Peters, J. C. Catalytic N2 -to-NH3 Conversion by Fe at Lower Driving
Force: A Proposed Role for Metallocene-Mediated PCET. ACS Cent.
Sci. 2017, 3 (3), 217−223.
(17) Pitman, C. L.; Finster, O. N. L.; Miller, A. J. M.
Cyclopentadiene-Mediated Hydride Transfer from Rhodium Com-
plexes. Chem. Commun. 2016, 52 (58), 9105−9108.
(18) Johnson, S. I.; Gray, H. B.; Blakemore, J. D.; Goddard, W. A.
Role of Ligand Protonation in Dihydrogen Evolution from a
Pentamethylcyclopentadienyl Rhodium Catalyst. Inorg. Chem. 2017,
56 (18), 11375−11386.
(19) Jung, S.; McCrory, C. C. L.; Ferrer, I. M.; Peters, J. C.;
Jaramillo, T. F. Benchmarking Nanoparticulate Metal Oxide Electro-
catalysts for the Alkaline Water Oxidation Reaction. J. Mater. Chem. A
2016, 4 (8), 3068−3076.
(20) Schild, D. J.; Drover, M. W.; Oyala, P. H.; Peters, J. C.
Generating Potent C-H PCET Donors: Ligand-Induced Fe-to-Ring
Proton Migration from a Cp*FeIII-H Complex Demonstrates a
Promising Strategy. J. Am. Chem. Soc. 2020, 144 (44), 18963−18970.
(21) Quintana, L. M. A.; Johnson, S. I.; Corona, S. L.; Villatoro, W.;
Goddard, W. A.; Takase, M. K.; VanderVelde, D. G.; Winkler, J. R.;
Gray, H. B.; Blakemore, J. D. Proton-Hydride Tautomerism in
Hydrogen Evolution Catalysis. Proc. Natl. Acad. Sci. U. S. A. 2016, 113
(23), 6409.
(22) Hopkins, J. A.; Lionetti, D.; Day, V. W.; Blakemore, J. D.
Chemical and Electrochemical Properties of [Cp*Rh] Complexes
Supported by a Hybrid Phosphine-Imine Ligand. Organometallics
2019, 38 (6), 1300−1310.
(23) Boyd, E. A.; Lionetti, D.; Henke, W. C.; Day, V. W.; Blakemore,
J. D. Preparation, Characterization, and Electrochemical Activation of
a Model [Cp*Rh] Hydride. Inorg. Chem. 2019, 58 (6), 3606−3615.
(24) Lionetti, D.; Day, V. W.; Blakemore, J. D. Structural and
Chemical Properties of Half-Sandwich Rhodium Complexes Sup-
ported by the Bis(2-Pyridyl)Methane Ligand. Dalt. Trans. 2019, 48
(33), 12396−12406.
(25) Weberg, R. T.; Norton, J. R. Kinetic and Thermodynamic
Acidity of Hydrido Transition-Metal Complexes. 6. Interstitial
Hydrides. J. Am. Chem. Soc. 1990, 112 (3), 1105−1108.
(26) Walker, H. W.; Kresge, C. T.; Ford, P. C.; Pearson, R. G. Rates
of Deprotonation and pKa Values of Transition Metal Carbonyl
Hydrides. J. Am. Chem. Soc. 1979, 101 (24), 7428−7429.
(27) Kresge, A. J. What Makes Proton Transfer Fast. Acc. Chem. Res.
1975, 8 (10), 354−360.
(28) Kramarz, K. W.; Norton, J. R. Slow Proton-Transfer Reactions
in Organometallic and Bioinorganic Chemistry. Progress in Inorganic
Chemistry 2007, 42, 1−65.
(29) Edidin, R. T.; Sullivan, J. M.; Norton, J. R. Kinetic and
Thermodynamic Acidity of Hydrido Transition-Metal Complexes. 4.
Kinetic Acidities toward Aniline and Their Use in Identifying Proton-
Transfer Mechanisms. J. Am. Chem. Soc. 1987, 109 (13), 3945−3953.
̈
(30) Kuhn, N.; Winter, M. Synthese Und Reaktivitat von
Dienylmetall-Verbindungen IV. Verbindungen Des Typs C5H5Ni-
(R2PX)X. J. Organomet. Chem. 1982, 232 (1), C48−C50.
(31) Harris, R. K.; Hodgkinson, P.; Pickard, C. J.; Yates, J. R.; Zorin,
V. Chemical Shift Computations on a Crystallographic Basis: Some
Reflections and Comments. Magn. Reson. Chem. 2007, 45 (S1),
S174−S186.
(32) Ciancanelli, R.; Noll, B. C.; DuBois, D. L.; DuBois, M. R.
Comprehensive Thermodynamic Characterization of the Metal-
Hydrogen Bond in a Series of Cobalt-Hydride Complexes. J. Am.
Chem. Soc. 2002, 124 (12), 2984−2992.
(33) Berning, D. E.; Miedaner, A.; Curtis, C. J.; Noll, B. C.;
Rakowski DuBois, M. C.; DuBois, D. L. Free-Energy Relationships
between the Proton and Hydride Donor Abilities of [HNi-
(Diphosphine)2]+ Complexes and the Half-Wave Potentials of Their
Conjugate Bases. Organometallics 2001, 20 (9), 1832−1839.
(34) Savéant, J.-M. Elements of Molecular and Biomolecular
Electrochemistry; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2006.
(35) Sampson, M. D.; Nguyen, A. D.; Grice, K. A.; Moore, C. E.;
Rheingold, A. L.; Kubiak, C. P. Manganese Catalysts with Bulky
Bipyridine Ligands for the Electrocatalytic Reduction of Carbon
Dioxide: Eliminating Dimerization and Altering Catalysis. J. Am.
Chem. Soc. 2014, 136 (14), 5460−5471.
(36) Fujita, E.; Creutz, C.; Sutin, N.; Szalda, D. J. Carbon Dioxide
Activation by Cobalt(I) Macrocycles: Factors Affecting Carbon
Dioxide and Carbon Monoxide Binding. J. Am. Chem. Soc. 1991,
113 (1), 343−353.
(37) Mock, M. T.; Potter, R. G.; O’Hagan, M. J.; Camaioni, D. M.;
Doughtery, W. G.; Kassell, W. S.; DuBois, D. L. Synthesis and
Hydride Transfer Reactions of Cobalt and Nickel Hydride Complexes
to BX3 Compounds. Inorg. Chem. 2011, 50 (23), 11914−11928.
(38) Kristjansdottir, S. S.; Loendorf, A. J.; Norton, J. R. Kinetic and
Thermodynamic Acidity of Hydrido Transition-Metal Complexes. 9.
A Sterically Hindered Cationic Hydride, [H4Re(PMe2Ph)4]+. Inorg.
Chem. 1991, 30 (23), 4470−4471.
(39) Jordan, R. F.; Norton, J. R. Kinetic and Thermodynamic
Acidity of Hydrido Transition-Metal Complexes. 1. Periodic Trends
in Group VI Complexes and Substituent Effects in Osmium
Complexes. J. Am. Chem. Soc. 1982, 104 (5), 1255−1263.
(40) Jordan, R. F.; Norton, J. R. Proton Transfer Reactions in
Organometallic Chemistry. ACS Symp. Ser. 1982, Ch. 17, 403−423,
DOI: 10.1021/bk-1982-0198.ch017.
(41) Nikitin, K.; O’Gara, R. Mechanisms and Beyond: Elucidation of
Fluxional Dynamics by Exchange NMR Spectroscopy. Chem. - Eur. J.
2019, 25, 1−40.
(42) Huang, T.; Rountree, E. S.; Traywick, A. P.; Bayoumi, M.;
Dempsey, J. L. Switching between Stepwise and Concerted Proton-
Coupled Electron Transfer Pathways in Tungsten Hydride Activation.
J. Am. Chem. Soc. 2018, 140 (44), 14655−14669.
(43) Churchill, D. G.; Janak, K. E.; Wittenberg, J. S.; Parkin, G.
Normal and Inverse Primary Kinetic Deuterium Isotope Effects for C-
H Bond Reductive Elimination and Oxidative Addition Reactions of
Molybdenocene and Tungstenocene Complexes: Evidence for
3405
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Benzene σ-Complex Intermediates. J. Am. Chem. Soc. 2003, 125 (5),
1403−1420.
(44) Neese, F. The ORCA Program System. Wiley Interdiscip. Rev.:
Comput. Mol. Sci. 2012, 2 (1), 73−78.
(45) Becke, A. D. Density Functional Calculations of Molecular
Bond Energies. J. Chem. Phys. 1986, 84 (8), 4524−4529.
(46) Perdew, J. P. Density-Functional Approximation for the
Correlation Energy of the Inhomogeneous Electron Gas. Phys. Rev.
B: Condens. Matter Mater. Phys. 1986, 33 (12), 8822−8824.
(47) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and
Accurate Ab Initio Parametrization of Density Functional Dispersion
Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010,
132 (15), 154104.
(48) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the Damping
Function in Dispersion Corrected Density Functional Theory. J.
Comput. Chem. 2011, 32 (7), 1456−1465.
(49) van Wullen, C. Molecular Density Functional Calculations in
̈
the Regular Relativistic Approximation: Method, Application to
Coinage Metal Diatomics, Hydrides, Fluorides and Chlorides, and
Comparison with First-Order Relativistic Calculations. J. Chem. Phys.
1998, 109 (2), 392−399.
(50) Weigend, F. Accurate Coulomb-Fitting Basis Sets for H to Rn.
Phys. Chem. Chem. Phys. 2006, 8 (9), 1057.
̈ ̈
(51) Klamt, A.; Schuurmann, G. COSMO: A New Approach to
Dielectric Screening in Solvents with Explicit Expressions for the
Screening Energy and Its Gradient. J. Chem. Soc., Perkin Trans. 2
1993, No. 5, 799−805.
3406
https://dx.doi.org/10.1021/jacs.0c11992
J. Am. Chem. Soc. 2021, 143, 3393−3406