Nucleation, growth, and repair of a cobalt-based oxygen evolving catalyst

Article abstract of DOI:10.1021/ja3000084

The mechanism of nucleation, steady-state growth, and repair is investigated for an oxygen evolving catalyst prepared by electrodeposition from Co²⁺ solutions in weakly basic electrolytes (Co-OEC). Potential step chronoamperometry and atomic force microscopy reveal that nucleation of Co-OEC is progressive and reaches a saturation surface coverage of ca. 70% on highly oriented pyrolytic graphite substrates. Steady-state electrodeposition of Co-OEC exhibits a Tafel slope approximately equal to 2.3 × RT/F. The electrochemical rate law exhibits a first order dependence on Co²⁺ and inverse orders on proton (third order) and proton acceptor, methylphosphonate (first order for 1.8 mM ≤ [MeP_i] ≤ 18 mM and second order dependence for 32 mM ≤ [MeP_i] ≤ 180 mM). These electrokinetic studies, combined with recent XAS studies of catalyst structure, suggest a mechanism for steady state growth at intermediate MeP_i concentration (1.8-18 mM) involving a rapid solution equilibrium between aquo Co(II) and Co(III) hydroxo species accompanied with a rapid surface equilibrium involving electrolyte dissociation and deprotonation of surface bound water. These equilibria are followed by a chemical rate-limiting step for incorporation of Co(III) into the growing cobaltate clusters comprising Co-OEC. At higher concentrations of MeP_i ([MeP_i] ≥ 32 mM), MePO ₃^2- equilibrium binding to Co(II) in solution is suggested by the kinetic data. Consistent with the disparate pH profiles for oxygen evolution electrocatalysis and catalyst formation, NMR-based quantification of catalyst dissolution as a function of pH demonstrates functional stability and repair at pH values >6 whereas catalyst corrosion prevails at lower pH values. These kinetic insights provide a basis for developing and operating functional water oxidation (photo)anodes under benign pH conditions.

Full text of DOI:10.1021/ja3000084

Article
pubs.acs.org/JACS
Nucleation, Growth, and Repair of a Cobalt-Based Oxygen Evolving
Catalyst
Yogesh Surendranath, Daniel A. Lutterman, Yi Liu, and Daniel G. Nocera*
Department of Chemistry, 6-335, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139-4307, United States
S
* Supporting Information
ABSTRACT: The mechanism of nucleation, steady-state
growth, and repair is investigated for an oxygen evolving
catalyst prepared by electrodeposition from Co2+ solutions in
weakly basic electrolytes (Co-OEC). Potential step chro-
noamperometry and atomic force microscopy reveal that
nucleation of Co-OEC is progressive and reaches a saturation
surface coverage of ca. 70% on highly oriented pyrolytic
graphite substrates. Steady-state electrodeposition of Co-OEC
exhibits a Tafel slope approximately equal to 2.3 × RT/F. The
electrochemical rate law exhibits a first order dependence on Co2+ and inverse orders on proton (third order) and proton
acceptor, methylphosphonate (first order for 1.8 mM ≤ [MePi] ≤ 18 mM and second order dependence for 32 mM ≤ [MePi] ≤
180 mM). These electrokinetic studies, combined with recent XAS studies of catalyst structure, suggest a mechanism for steady
state growth at intermediate MePi concentration (1.8−18 mM) involving a rapid solution equilibrium between aquo Co(II) and
Co(III) hydroxo species accompanied with a rapid surface equilibrium involving electrolyte dissociation and deprotonation of
surface bound water. These equilibria are followed by a chemical rate-limiting step for incorporation of Co(III) into the growing
cobaltate clusters comprising Co-OEC. At higher concentrations of MePi ([MePi] ≥ 32 mM), MePO32− equilibrium binding to
Co(II) in solution is suggested by the kinetic data. Consistent with the disparate pH profiles for oxygen evolution electrocatalysis
and catalyst formation, NMR-based quantification of catalyst dissolution as a function of pH demonstrates functional stability and
repair at pH values >6 whereas catalyst corrosion prevails at lower pH values. These kinetic insights provide a basis for
developing and operating functional water oxidation (photo)anodes under benign pH conditions.
solutions of Co2+ salts are electrolyzed in the presence of
INTRODUCTION
■
phosphate (Co−Pi), borate (Co−Bi), or methylphosphonate
(Co-MePi) (collectively termed Co-OEC);36,37 more recently,
we have used a similar strategy to prepare a Ni−Bi catalyst.38
These catalysts are of interest because they (1) form in situ
under mild conditions on a variety of conductive sub-
strates;36−39 (2) exhibit high activity in pH 7−9 water at
room temperature;36,37 (3) are functional in impure water;37,39
(4) are composed of inexpensive, earth-abundant materials;36,37
(5) self-heal by reversing catalyst corrosion at open circuit upon
application of a potential;40 (6) are functional models of the
oxygen-evolving complex of Photosystem II;41 and (7) can be
interfaced with light absorbing and charge separating
materials.42−48 The ability to perform the latter under benign
conditions has enabled the development of the artificial
leaf.49−51 In the architecture of the leaf, the Co-OEC is
interfaced to a Si junction in a wireless configuration to allow
for the direct splitting of water under illumination with 1 sun of
AM 1.5 simulated sunlight.
The electrochemical splitting of water into hydrogen and
oxygen is a high energy density method for storing solar energy
in the form of chemical fuels.1−4 This endergonic electro-
chemical conversion stores 1.23 V and consists of the four
electron, four proton oxidation of water to oxygen and the
reduction of the produced protons to hydrogen. Of these two
half-reactions, the oxygen evolution reaction (OER) is
particularly demanding because it requires the coupling of
multiple proton and electron transfers,5−7 the distribution of
four redox processes over a narrow potential range, and the
formation of two oxygen−oxygen bonds.8−14 The efficiency
and conditions required for this reaction are key determinants
of the overall viability of energy storage via water-splitting,4 and
accordingly, the continued development of effective OER
catalysts15−34 stands as a central scientific and technological
challenge in energy conversion.
We have emphasized the development of inexpensive, highly
manufacturable water-splitting catalysts that may be easily
integrated with photovoltaic substrates. The simple operation
of the catalyst from conventional water sources under benign
conditions is an important step toward providing distributed
solar energy storage at low-cost.35 We have recently described
the self-assembly of a highly active cobalt-based oxygen
evolving catalyst that forms as a thin film when aqueous
In developing photoelectrochemical and photovoltaic-based
architectures for water splitting, such as the artificial leaf,
attention must be paid to the (semi)conductor|catalyst|
Received: January 1, 2012
Published: March 6, 2012
© 2012 American Chemical Society
6326
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
electrolyte interface where efficiency determining processes
such as photovoltage generation, current rectification, light
transmission, and catalysis must take place simultaneously.
Accordingly, the continued development and improvement of
devices such as the artificial leaf provide an imperative for
understanding catalyst deposition and morphology of the
catalyst on surfaces so as to permit synthetic control over the
microstructure of the (semi)conductor|catalyst| electrolyte
interface.42,43,52,53 Such studies of Co-OEC are enabled by
the ability to form Co-OEC at potentials significantly lower
than that required for OER. We now report mechanistic details
of the pathway for catalyst nucleation and steady-state growth,
and its relationship to catalyst repair processes. The progressive
nature of the nucleation process of Co-OEC has been isolated
by using chronoamperometry and AFM imaging. In addition,
hydrodynamic electrokinetic methods have been employed to
furnish the rate law for steady-state catalyst growth, and the pH
regimes of functional catalyst stability and repair have been
defined from electrokinetic profiles. These kinetic insights
provide a coherent picture of the electrodeposition process of
Co-OEC and provide a rational framework for refining the
properties of the (semi)conductor|catalyst|solution interface for
the development of improved devices such as the artificial leaf.
RESULTS
■
To probe the initial stages of catalyst formation, chronoam-
perograms of freshly polished glassy carbon electrodes were
recorded in quiescent solutions of Co2+ (0.4 mM) containing
0.02 M MePi and 1.97 M KNO3 electrolyte (pH 7.5). We note
that an active OER catalyst may be electrodeposited from
solutions containing phosphate or borate electrolytes.36,37,54
For the studies reported here, methylphosphonate (MePi) was
chosen as the electrolyte for catalyst growth, as opposed to
phosphate or borate, because low millimolar concentrations of
Co2+ are indefinitely stable and no discernible precipitate is
observed in solution over days. In the case of Pi or Bi,
precipitate will form over the course of several hours.
Notwithstanding, the results reported here are generalizable
as previous studies have shown that the nature of the catalyst is
the same whether electrodeposition proceeds from Pi, Bi, or
MePi.37 The chronoamperograms of this solution are shown in
Figure 1 along with a slow scan (5 mV/s) cyclic voltammogram
(CV) recorded under identical conditions. The CV exhibits an
anodic wave at 0.99 V (all potentials are referenced to the
normal hydrogen electrode (NHE)) characteristic of catalyst
formation. Following initial polarization at 0.75 V (omitted in
Figure 1), during which time only double-layer charging
occurred, the electrode potential was stepped to various
voltages spanning the catalyst formation wave in the CV. At
all the step potentials explored, the chronoamperograms display
three features. The current rapidly decays from an initial spike
immediately following the potential step followed by a rapid
rise to attain a peak followed by a gradual decay toward a
nonzero limiting value. For higher potentials, the magnitude of
the peak current density, jmax, systematically increases whereas
the time corresponding to the peak, tmax, systematically
decreases.
Figure 1. (top) Cyclic voltammogram (scan rate = 5 mV/s) and
(bottom) potential step chronoamperograms of a freshly polished
glassy carbon disk electrode in 0.4 mM Co2+, and 0.02 M MePi, 1.97
M KNO3 electrolyte at pH 7.5. Chronoamperometry data recorded
with a step voltage of 1.01 (magenta, −··), 0.99 (green, −·−·), 0.97
(blue, ···), 0.95 (red, −−−), and 0.93 V (black, ) following a 100 s
pulse at 0.75 V (not shown). Step voltages used to collect
chronoamperograms are indicted with vertical lines overlaid on the
CV trace.
the Supporting Information. Similar to the behavior observed
using a glassy carbon working electrode, potential step
polarization from 0.75 to 0.97 V in 0.4 mM Co2+, 0.02 M
MePi, and 1.97 M KNO3 electrolyte (pH 7.5) leads to a
nucleated growth trace with tmax = 10.1(1) s. Using these same
conditions, an array of catalyst-coated HOPG electrodes were
prepared following potential step polarization that was ceased
at ca. 0.2, 0.5, 1, 2, 4, and 8 × tmax. Following the termination of
electrolysis, the electrodes were rinsed thoroughly with reagent
grade water, dried, and imaged by AFM. Representative images
are shown in Figures 2 and S2. The relatively smooth HOPG
electrode surface is visible underneath discontinuous islands of
Co-OEC. The bare HOPG surface is punctuated randomly by
long lines due to step edges between graphite layers. Islands of
catalyst are distributed randomly on the highly oriented
pyrolytic graphite surface, excluding a higher density of islands
that localize at step edges. At 0.2 × tmax, the island sizes range
from 35 to 55 nm and catalyst covers 9(1)% of the surface. At
0.5 × tmax, the island sizes range from 65 to 85 nm, and catalyst
surface coverage increases to 27(2)%. At 1 × tmax, catalyst
To further interrogate the nucleation process, partially
nucleated catalyst films were imaged by atomic force
microscopy (AFM). Potential step chronoamperograms were
recorded using highly oriented pyrolytic graphite (HOPG)
working electrodes cleaved from a bulk slab prior to the
experiment; details of the electrode preparation are provided in
6327
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
catalyst formation from 0.4 mM Co2+, 0.02 M MePi, and 1.97
M KNO3 electrolyte (pH 7.5) were measured at a variety of
rotation rates, ω, and applied potentials. For each potential
examined, a Koutecky−Levich plot of j−1 versus ω−1/2 (Figure
́
S3) was extrapolated to infinite rotation rate (ω−1/2 = 0) to
determine the activation controlled current density for catalyst
formation, jac, in the absence of mass-transport limitations.55
For all data points, ohmic potential losses amounted to <1 mV
and were ignored. A Tafel plot of the applied potential versus
the log of jac is shown in Figure 3. The data points exhibit
linearity over 0.12 V and a slope of 60 mV/decade.
Figure 2. Representative 5 × 5 μm2 AFM images of a highly oriented
pyrolytic graphite electrode after being subjected to potential step
polarization from 0.75 to 0.97 V for (a) 0.2, (b) 0.5, (c) 1, (d) 2, (e) 4,
and (f) 8 × tmax (=10.1(1) s). Bars to the right of each image indicate
the depth with full scale values of (a) 20, (b) 30, (c) 50, (d) 75, (e) 75,
and (f) 50 nm. Electrolyte conditions: 0.4 mM Co2+ and 0.02 M MePi,
1.97 M KNO3 electrolyte at pH 7.5 (2 M ionic strength). Plot displays
coverage percentage of catalyst versus the normalized duration of
Figure 3. Tafel plot of Co-OEC catalyst film formation from 0.4 mM
Co2+ and 0.02 M MePi, 1.97 M KNO3 electrolyte at pH 7.5 onto a Pt
rotating disk electrode. Activation controlled current density values
(j ) were derived from Koutecky−Levich analysis of steady-state
́
ac
current densities measured at multiple rotation rates (Figure S3). The
Tafel slope is 60 mV/decade.
The dependence of the catalyst formation rate on Co2+
concentration was interrogated by collecting Tafel plots of
steady-state growth of catalyst using the procedure described
above. Figure S4 shows these Tafel plots for catalyst growth
from solutions containing [Co2+] = 0.1, 0.18, 0.32, 0.56, and 1
mM and 0.02 M MePi and 1.97 M KNO3 (pH 7.5). The Tafel
plot exhibits linearity over >1.5 decades in activation controlled
current density and an average Tafel slope of 62(8) mV/
decade. Interpolation of these Tafel plots at 0.84, 0.86, 0.88,
and 0.90 V yields steady-state activation controlled current
densities for catalyst growth as a function of Co2+ concentration
(Figure 4). The data exhibit linearity over the decade range of
Co2+ concentrations with an average slope of 1.1(1). These
results establish that the electrodeposition process is first order
in Co2+.
The role of proton activity on the electrodeposition rate was
defined from constructing Tafel plots over the pH range 7−
8.25 (Figure S5). Because acid is generated during catalyst
formation and the buffer capacity was low, concentrated KOH
was periodically added to maintain the bulk solution pH to
within 0.02 pH units for the duration of Tafel data collection
(see SI). All Tafel plots exhibit linearity over ∼2 decades in
activation controlled current density and an average Tafel slope
of 60(4) mV/decade (Figure S5). Unlike the case described
above for variations in the Co2+ concentration, the Tafel plots
in Figure S5 are distributed dramatically in potential preventing
potential step polarization, t/tmax
.
coverage rises to 39(2)%, and catalyst islands have begun to
coalesce. Catalyst coverage continues to increase to 48(2)% at 2
× tmax and 65(5)% at 4 × tmax, after which a plateau is reached at
8 × tmax (65(3)%.) At 8 × tmax, the agglomerated catalyst islands
range in size from 70 to 140 nm. The saturation behavior of the
surface coverage of catalyst is clearly apparent from the plot of
percent catalyst coverage versus t/tmax (Figure 2). Beyond 0.5 ×
tmax, a marked increase in particle number density does not
accompany the further increase in surface coverage.
In order to deconvolute steady-state catalyst growth kinetics
from nucleation kinetics, the following studies of catalyst
growth were all conducted on preformed catalyst films
exhibiting saturation surface coverage. Specifically, prior to
the collection of steady-state kinetic data, a 200 s galvanostatic
pulse at 50 μA/cm2 was applied to the electrode, rotated at
2500 rpm. This pulse corresponds to 10 mC/cm2 charge passed
for catalyst deposition, which is ∼10 fold higher than the typical
charge passed at 8 × tmax (<1 mC/cm2) in the nucleated growth
studies described above. This ensures that the catalyst film
completely covers the surface of the Pt electrode and that the
subsequent kinetic data is not influenced by nucleation
processes. Using preformed 10 mC/cm2 catalyst films as a
substrate for steady-state catalyst growth, current densities for
6328
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
Thus, the −172(5) mV/pH unit slope observed in Figure 5,
when divided by negative of the 60(4) mV/decade Tafel slope,
yields the reaction order in pH of 2.9(2). Thus, these data
establish third order dependence on pH and, therefore, inverse
third order dependence on proton activity.
The dependence of the electrodeposition rate on the
concentration of the proton-accepting electrolyte species was
ascertained by collecting Tafel plots in 0.4 mM Co2+ (pH 7.5)
over a MePi concentration range from 1 to 178 mM (Figure
S6). All solutions contained sufficient KNO3 background
electrolyte to maintain a constant 2 M ionic strength. For
each MePi concentration, Tafel data were collected using
independently prepared 10 mC/cm2 catalyst films and the bulk
solution pH was maintained at 7.5
0.02 with periodic
addition of concentrated KOH (see SI). The Tafel data exhibit
linearity over a ∼2 decade range in current density and an
average slope of 64(8) (Figure S6). Interpolation of these Tafel
plots at jac = 10, 32, and 100 μA/cm2 (log (jac/A/cm2) = −5,
−4.5, and −4 on Figure S6) yields the required potential as a
function of the log of the MePi concentration (Figure 6). The
Figure 4. [Co2+] dependence of steady-state activation controlled
current density for catalyst film formation in 0.02 M MePi, 1.97 M
KNO3 electrolyte at pH 7.5 (2 M ionic strength) at E = 1.04 (green
◆
■
●
▲
), 1.06 (blue ), 1.08 (red ), and 1.1 V (black ). Average slope
is 1.1(1). Data were interpolated from Tafel plots in Figure S4.
a straightforward interpolation at constant potential over the
entire pH range. However, interpolation of these Tafel plots at
jac = 3.2, 10, 32, and 100 μA/cm2 (log (jac/A/cm2) = −5.5, −5,
−4.5, and −4 on Figure S5) yields the required potential as a
function of pH (Figure 5). The data are linear over the 1.25 pH
Figure 6. [MePi] dependence of the potential for catalyst film
formation in 0.4 mM Co2+ at pH 7.5 (2 M ionic strength with KNO3
■
●
background electrolyte) at jac = 10 (black ), 32 (red ), and 100
2
▲
(blue ) μA/cm . Average slopes are 60(2) and 120(7) mV/decade
for the low and high buffer strength regions, respectively. Data were
interpolated from Tafel plots in Figure S6.
data exhibit two distinct linear regions. For MePi concen-
trations between 32 and 178 mM, a slope of 120(7) mV/
decade is observed, whereas a slope of 60(2) mV/decade is
observed for MePi concentrations between 1.8 and 17.8 mM.
As encountered above, the data in Figure 6 correspond to
constant jac, and thus, the potential dependence of the current
density (the Tafel behavior) is convoluted with the slope of the
potential versus log([MePi]) plot.56 For this case
Figure 5. pH dependence of the potential for catalyst film formation in
0.4 mM Co2+ and 0.02 M MePi, 1.97 M KNO3 electrolyte at jac = 3.2
2
■
◆
●
▲
(black ), 10 (red ), 32 (blue ), and 100 (green ) μA/cm .
Average slope is −172(5) mV/pH unit. Data were interpolated from
Tafel plots in Figure S5.
unit range explored here and exhibit an average slope of
−172(5) mV/pH unit. In this experiment, the Tafel data were
interpolated at constant current density, jac, and therefore, the
potential dependence of the current density (the Tafel
behavior) is convoluted with the slope of the potential versus
pH plot.56 Specifically
⎛
⎜
⎝
⎞
⎟
⎠
⎛
⎜
⎝
⎞
⎟
⎠
⎛
⎜
⎝
⎞
⎟
⎠
∂log(j)
∂log([MeP ])
∂E
∂E
∂log(j)
=
∂log([MeP ])
i
i
j
[MeP ]
E
i
(2)
Therefore, the 120(7) and 60(2) mV/decade slopes observed
in Figure 6 correspond to reaction orders of −1.9(3) and
−0.9(1). These results establish inverse first order and inverse
second order dependence on MePi concentration over the
⎛
⎜
⎝
⎞
⎟
⎠
⎞
⎛
⎞
⎛
∂log(j)
∂pH
∂E
∂pH
∂E
∂log(j)
= −
⎜
⎝
⎟
⎠
⎜
⎝
⎟
⎠
j
E
pH
(1)
6329
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
ranges 1.8−17.8 mM and 32−178 mM, respectively. Below 1.8
mM MePi, the potential appears to be invariant with buffer
strength. However, activation controlled current density data
could not be obtained below 1 mM MePi because local and
bulk pH changes at very low buffer strength caused drastic
the latter, 0.17, and a greater initial slope of the dissolution
profile for the former. A similar phenomenon is also observed
at pH 5 with a systematic decrease in the overall fraction
dissolved after 1 h for both the 0 and 1 mA/cm2 samples, but
no dissolution is detected at pH 6 and 7. A plot of the total
fraction, f, of Co dissolved after 1 h of electrolysis versus the
solution pH is shown in Figure 8. This data establishes that Co-
systematic errors in the extrapolation of Koutecky−Levich plots
́
to infinite rotation rate. Thus, while there appears to be a
region of zero order dependence below 1.8 mM MePi, the lack
of data at very low buffer strength makes this assignment
tenuous at best.
As catalyst dissolution is expected to be the microscopic
reverse of the catalyst formation process, the kinetics of catalyst
dissolution were also quantified over a wide pH range. For this
study, a series of 100 mC/cm2 catalyst films were prepared on
fluorine−tin oxide (FTO) working electrodes by electro-
deposition at 1.05 V from quiescent 0.1 M phosphate
electrolyte (Pi), pH 7 solutions containing 0.5 mM Co2+.
Under these conditions, negligible water oxidation catalysis is
observed. Thus, the 100 mC/cm2 charge passed provides an
accurate estimate of the Co content of the film, ∼1 μmol Co/
cm2. After film preparation, each electrode was rinsed with
reagent grade water and operated for water oxidation at 1 and 0
mA/cm2 in stirred electrolyte solutions of 0.04 M Britton−
Robinson buffer (see SI for details) that contained no Co2+ ion.
At various times over the course of one hour, aliquots of the
electrolyte solution were removed from the working compart-
ment of the electrolysis cell. The Co2+ concentration in each
aliquot was determined by measuring the magnitude of Co2+-
induced paramagnetic line broadening of the 31P NMR peak for
inorganic phosphate and comparing to a standard curve (see SI
and Figure S7). From these concentration measurements, the
fraction of Co dissolved from the film versus time was
calculated and is shown for pH 4 in Figure 7. For electrodes
Figure 8. Fraction (f) of catalyst film dissolved after 1 h vs pH for
electrodes poised at 1 (blue ) and 0 (red ) mA/cm in 0.04 M
Britton−Robinson buffer. Lines are presented to guide the eye.
2
■
●
OEC exhibits functional stability at > pH 6 but is subject to
progressive dissolution below this pH.
DISCUSSION
■
Chronoamperometry affords valuable insight into the initial
stages of catalyst formation. Important to these studies, a
characteristic sharp anodic prefeature wave several hundred
millivolts prior to the onset of water oxidation catalysis is
observed in cyclic voltammograms of Co2+ in a buffering
electrolyte at pH 7−9 (Figure 1).36,37 A single CV scan of a
freshly polished inert electrode with a switching potential
beyond this prefeature wave but prior to the onset of catalysis is
necessary and sufficient for the electrodeposition of a film
active for water oxidation catalysis. We utilize this separation in
potentials between water oxidation and catalyst formation to
isolate the kinetics for catalyst formation.
Chronoamperograms with step potentials spanning the
anodic prefeature wave (see Figure 1) reveal that the process
is not merely a diffusion controlled oxidation of Co2+ ions.
Whereas the potential step chronoamperogram of a electro-
chemically and chemically reversible one-electron redox couple
exhibits a simple decay of current proportional to t−1/2
immediately following the potential step,57 the chronoampero-
grams of Co-OEC formation shown in Figure 1 are more
complicated and reminiscent of the nucleated growth curves
observed for the electrodeposition of metal films.58 Following
the potential step, the current decays rapidly from an initial
spike due to currents associated with double-layer charging and
adsorption/desorption events on the electrode surface.58 This
initial decay is followed by a rise in current that is attributed to
the nucleation of the catalyst in the form of islands, which in
effect serve to increase the surface area of the electrode. Quasi-
three-dimensional growth of these catalyst islands gives rise to a
Figure 7. Fraction (f) of catalyst film dissolved versus time for an
2
■
●
electrodes poised at 1 (blue ) and 0 (red ) mA/cm in an initially
Co2+-free electrolyte solution of 0.04 M Britton−Robinson buffer, pH
4. Lines are presented to guide the eye.
operated at 0 and 1 mA/cm2, catalyst corrosion prevails
throughout the entire duration of the experiment. However, for
the electrode held under open-circuit conditions (0 mA/cm2),
the rate of catalyst corrosion is greater than that for the
electrode operated at 1 mA/cm2. This is evidenced by both a
greater fraction dissolved after 1 h for the former, 0.33, versus
6330
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
rapid increase in current until a maximum is reached. At this
point, the reactant-depleted diffusion zones surrounding each
growing catalyst island begin to overlap, causing the onset of
semi-infinite linear diffusion of Co2+ ions and/or buffer
constituents and an associated t−1/2 decay of the current.
Notably, at very long times, the current reaches a steady-state
plateau rather than decaying to zero. This is attributed to
convectional mass-transport of Co2+ to the surface at these long
times. Importantly, the step potential has a marked impact on
the time, tmax, and current density, jmax, at which the
chronoamperogram exhibits its characteristic peak. Increasing
the potential increases the driving force, and, thus, the rate of
nucleation and catalyst growth, resulting in a systematic
increase in jmax. The enhanced growth rate causes the diffusion
fields to overlap more rapidly, and, consequently, leads to a
systematic decrease in tmax with increasing driving force. Thus,
control of the electrodeposition potential provides a convenient
method for modulating tmax and jmax, enabling control over the
surface coverage of catalyst islands with fidelity (vide infra). All
of these characteristic features, together, establish that Co-OEC
formation occurs via nucleation followed by diffusion limited
growth.
Comparing the experimental chronoamperometric transients
to theoretical predictions provides insight into the mechanism
of nucleated growth, which can be characterized by one of two
limiting extremes, progressive as opposed to instantaneous
nucleation. In the instantaneous case, all potential nucleation
sites, N0, on the surface are assumed to give rise to nuclei at
time zero following the potential step. In contrast, for the
progressive case, the rate of nucleation is given by the following
equation59
Figure 9. Normalized 0.97 V potential step chronoamperogram
(black, ) of a freshly polished glassy carbon disk electrode 0.4 mM
Co2+ and 0.02 M MePi, 1.970 M KNO3 electrolyte at pH 7.5.
Theoretical normalized chronoamperograms for the cases of
instantaneous (blue, ···) nucleated growth and progressive nucleated
growth with 0 (green, −·−·) and 4th (red, −−−) order nucleation rate
laws.
possible, the point of Figure 9 is to show the differences
between instantaneous and progressive nucleation and the
impact of a nonzero order. The simulation of a fourth order
rate law was chosen simply to represent qualitatively the
deviations expected for a high, nonzero nucleation order and is
not intended to quantitatively isolate the reaction order for
nucleation. Given this limitation, Figure 9 clearly distinguishes
between instantaneous and progressive nucleation. Instanta-
neous nucleation is characterized by a very sharp increase in
current at early times and a broad peak, both of which are not
observed in the experimental data. In contrast, the normalized
experimental trace exhibits a slow rise, following double layer
charging at early times, and a sharp peak, in agreement with
that expected for zero order progressive nucleation. However,
the experimental data deviates positively from theory for zero
order progressive nucleation at times well beyond tmax. In
contrast, the entire duration of the experimental transient is
well modeled by the simulated curve for fourth order
progressive nucleation except at early times. At early times,
the current is dominated by non-Faradaic double-layer charging
and adsorption/desorption events on the surface that are not
accounted for by the theoretical models.58,59 Good agreement
between the experimental data and the theoretical trace is also
observed for higher step potentials (1.01 and 0.99 V), but for
the lower potential steps (0.95 and 0.93 V), the experimental
current exhibits a positive deviation relative to the theory for
long time points (Figure S1). In these cases, tmax > 10 s after the
potential step, and thus, convective mass transport62 is expected
to elevate the current at long times relative to the theoretical
models, which all assume purely diffusion-limited mass
transport.58,59 While this agreement between theory and
experiment in Figure 9 is insufficient for determining a precise
reaction order for the nucleation rate law, it does point to a
progressive mechanism of nucleation with a nonzero order
which is consistent with the complexity of the catalyst
n
⎛
⎜
⎝
⎞
⎟
⎠
∂N
∂t
c
= k
(N − N)
0
N
c
(3)
0
where kN is the nucleation rate constant, and the term, N0 − N,
represents the number of unoccupied nucleation sites on the
electrode surface. In eq 3, c and c0 represent the local and bulk
concentrations of all solution species upon which the
nucleation rate depends, with the exponent, n, characterizing
the overall reaction order. For the most elementary cases of
instantaneous nucleation and progressive nucleation with a zero
order nucleation rate law, n = 0, several theoretical models have
been proposed to explain the chronoamperometric transi-
ent.58,60,61 Additionally, for the case of progressive nucleation, a
theoretical model has been elaborated to account for the
influence of nonzero nucleation order on the expected
chronoamperometric transient.58 Distinguishing between the
different mechanistic possibilities is most conveniently provided
by comparing the theoretically predicted and experimental
transients in a normalized form. In Figure 9, the experimental
chronoamperometric transient for the 0.97 V step potential is
plotted with the ordinate and abscissa normalized to the peak
current density, (j/jmax)2, and peak time, (t/tmax), respectively.
The normalized representation in Figure 9 serves to
deconvolute changes in nucleation mechanism from changes
in kN and N0 (eq 3) which are both substrate and potential
dependent (vide supra). Overlaid on this normalized
experimental data are three theoretical traces for the limiting
cases of instantaneous nucleation followed by diffusion-limited
growth58 and progressive nucleation followed by diffusion-
limited growth with the assumption of a zero order58 and
fourth order59 nucleation rate law. Whereas other orders are
6331
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
formation process relative to simple metal film deposition and
the third order rate law determined for steady-state film growth
(vide infra).
shows, a clear first order dependence is observed across a 60
mV range of potential values spanning the linear Tafel region.
Taken together with the 60 mV/decade Tafel slope, this result
implies that the reversible one-electron equilibrium step also
involves a single Co2+ ion. This result shows that a mechanism
involving multinuclear solution species is not required for
catalyst self-assembly.
Further evidence for the progressive nature of the nucleation
process is afforded by AFM images of electrode surfaces subject
to partial nucleation (Figure 2). To meet the stringent surface
smoothness requirements of AFM, highly oriented pyrolytic
graphite electrodes were employed. Consistent with the
behavior predicted for high-order progressive nucleation,59
the number of nuclei increase dramatically at early times,
between 0.2 and 1 × tmax, after which relatively few new nuclei
are formed. Instead, the already formed nuclei grow and
coalesce to coat a greater proportion of the surface between 1
and 4 × tmax. The surface coverage plateaus after 4 × tmax at ca.
65%. At this point, the diffusion fields of each growing catalyst
island have completely overlapped thereby slowing the increase
of surface coverage;58 AFM shows that the tops of catalyst
islands, which protrude into the undepleted bulk solvent, grow
preferentially. We fully anticipate that prolonged electrolysis
would lead to full surface coverage as the growing particles
coalesce and form the continuous films observed by SEM in
previous studies.36,37,39 Although similar chronoamperometric
transients are observed on different electrode substrates
including glassy carbon (Figure 1), ITO, FTO, and HOPG,
we expect that catalyst island morphology and surface coverage
will depend on the nature of the electrode and its interaction
with growing catalyst particles. Notwithstanding, these AFM
results confirm that catalyst nucleation is progressive, and
importantly, they point to a convenient method for patterning
surfaces with catalyst islands. The appropriate choice of
(photo)potential allows for control over the absolute value of
tmax (vide supra), and termination of electrodeposition at the
appropriate time relative to tmax provides a convenient handle
with which to control the surface coverage of catalyst islands.
The growth of catalyst islands is subject to mass-transport
limitations. In order to isolate the mechanistic details of catalyst
growth, one must determine the current−potential relationship
in the absence of mass-transport limitations and without
influence from nucleation kinetics. To ensure the latter, steady
state kinetic data were obtained by monitoring the rate of film
deposition on electrodes on which catalyst was deposited after
10 mC/cm2 was passed. This pretreatment of the electrode
ensured complete coverage of the electrode surface with
catalyst (vide supra), and therefore, the data should be
negligibly impacted by nucleation and instead report on the
kinetics of steady-state catalyst growth. Activation controlled
current densities, jac, under any given set of potential and
electrolyte conditions were obtained by extrapolating current
densities measured at variable rotation rates on a rotating disk
As the Co(aq)2+/3+ redox couple is 1.92 V,64 it is expected
that one or more protons are involved in the one-electron
oxidation of Co2+ to Co3+ and catalyst assembly. Insight into
the number of protons exchanged in equilibrium prior to the
rate limiting chemical step is provided by interpolation of pH
dependent Tafel plots. Below pH 7, reliable data cannot be
obtained because the deposition process begins to overlap with
water oxidation thus convoluting the Tafel behavior. Above pH
8.25, the Tafel slope begins to rise significantly above 60 mV/
decade thus preventing a straightforward interpretation of the
reaction order. Hence Tafel data were collected over the pH
range 7−8.25 (Figure S5). The potential versus pH plot of
Figure 5 exhibits good linearity over the pH range explored for
a 1.5 decade range of jac values. The slope of these plots, taken
together with the 60 mV/decade Tafel slope, establishes an
inverse third order dependence on proton activity. Given that
the pKa = 7.0 for MePi, under these ionic strength conditions
(see SI for details of pKa calculation at 2 M ionic strength), the
2−
concentration of MePO3 does not change in a logarithmic
fashion over the range of pH interrogated here. Equilibria
involving this species (vide infra), therefore, will not
appreciably influence the apparent reaction order in proton.
Taken together, these results suggest that the oxidation of Co2+
to Co3+ is coupled to three protons, which permits access to
Co3+ at low potentials and accordingly enables catalyst self-
assembly from solution.
The high degree of proton coupling to the reversible one-
electron transfer establishes the key role for MePi (or Pi or Bi)
as the proton acceptor in catalyst formation. The anion
additionally has the proclivity to bind cobalt centers of nascent
catalyst clusters. Interpolation of Tafel plots collected over a
wide range of buffer strengths (1−178 mM) indicates that the
potential required to sustain a given jac increases as buffer
strength increases (Figure 6). The requirement for additional
driving force with increasing buffer strength indicates an
inhibitory effect of MePi on the catalyst growth process.
Furthermore, the potential versus log([MePi]) plot exhibits two
distinct regions of different slopes, at buffer strengths beyond
32 mM (log([MePi]) = 1.5), a −120 mV/decade slope is
observed corresponding to inverse second order dependence in
MePi. At intermediate buffer strengths, 1.8−17.8 mM (log-
([MePi]) = 0.25−1.25), a −60 mV/decade slope is observed
corresponding to inverse first order dependence in MePi.
Importantly, the transition between inverse first and second
order behavior corresponds well to the established binding
constant for MePi to Co2+ in solution.65 At 50 mM MePi, 50%
electrode (RDE) to infinite rotation speed (Figure S3).57
A
Tafel plot of the applied potential versus the log of jac exhibits
good linearity over an ∼2 decade range in activation-controlled
current density and a slope of 60 mV/decade (Figure 3). A 60
mV/decade slope implies a catalyst formation mechanism
involving a one-electron reversible equilibrium preceding a
chemical rate-limiting step for catalyst formation.56
XAS studies have established that as-deposited Co-OEC
catalyst films comprise molecular cobaltate clusters.63 The
clusters form by self-assembly following the oxidation of Co2+
to Co3+. The reaction order in Co2+ is obtained directly from
interpolation of Tafel plots collected at various Co2+
concentrations. Interpolation at constant applied potential
yields jac as a function of the Co2+ concentration. As Figure 4
2−
of the Co2+ in solution is estimated to be bound to MePO3
under the pH and ionic strength conditions of this study (see SI
for details of binding calculation). This observation suggests
that the observed inverse second order behavior at high buffer
strength is due, in part, to MePO3 binding to Co2+ in
2−
solution. On the basis of phosphate binding constants,66
protonated forms of MePi are not expected to bind Co2+, and
the impact of these species on catalyst growth can be safely
ignored. The underlying first order behavior, which spans 1.8−
178 mM (log([MePi]) = 0.25−2.25) buffer strength, suggests a
6332
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
Figure 10. Proposed mechanism of Co-OEC catalyst film formation.
second mode of catalyst growth inhibition, which we attribute
to MePi binding to clusters in the growing catalyst film. Thus,
equilibrium dissociation of MePi from the surface must precede
incorporation of newly oxidized Co centers.
Similarly, θA represents the fraction of active sites for film
growth in the form of intermediate A. As such, the ratio of θB ×
[{Co3+}] to the product of resting state species, θA × [{Co2+}],
is given by the Nernst equation
The data in Figures 3−6 establish the following electro-
3+
θ [{Co }]
⎡
⎤
FE
RT
chemical rate law for the growth of Co-OEC catalyst films
B
−3
−1
= K (a +) ([MeP ]) exp
1
i
⎢
⎣
⎥
⎦
H
2+
θ [{Co }]
(8)
⎡
⎤
FE
RT
A
2+
−3
−1
j
= k ([Co ])(a +) ([MeP ]) exp
0
i
⎢
⎣
⎥
⎦
H
ac
(4)
where K1 is the equilibrium constant at E = 0. This equation
represents a quasi-equilibrium assumption for the conversion
described by eq 5. Substituting for θB × [{Co3+}] in eq 7 yields
where k0 is a potential-independent rate constant. This constant
is proportional to the exchange current density for the
electrodeposition process. The rate expression carries the
observed first order dependence on [Co2+] (Figure 4), inverse
⎡
⎤
FE
RT
2+
−3
−1
+
) ([MeP ]) exp
H
v = θ K k ([{Co }])(a
A 1 2
i
⎢
⎣
⎥
⎦
+
(9)
third order dependence on proton activity, aH (Figure 5),
inverse first order dependence on [MePi] at intermediate buffer
strength (Figure 6), and the exponential relationship with
potential, E (Figure 3). Rearrangement of the log form of eq 4
yields a Tafel slope, ∂E/∂log(i), of 59 mV/decade that is also
consistent with the experimental data shown in Figure 3.
The electrochemical rate law in eq 4 is consistent with the
following mechanistic sequence:
If the surface coverage of the resting state species, A, is
appreciable (>0.9) over the potential and current range in
which Tafel and other kinetic data are collected, θA will not vary
appreciably with potential and can be set as a constant. Under
these conditions, eq 9 approximates the experimental rate law,
eq 4, with
k = F × θ K k
(10)
0
A 1 2
2+
3+
−
+
A + {Co } ⇌ B + {Co } + e + 3H + MeP
(5)
(6)
i
Information about the structure of resting state species, A, is
provided by X-ray absorption spectroscopy (XAS),63 which
reveals Co-OEC to be composed of Co-oxo/hydroxo clusters
composed of edge-sharing CoO6 octahedra, which are the basic
structural components of the extended planes of alkali metal
cobaltates, particularly Co(O)OH.67 Within the framework of
this structural model, it is reasonable to invoke that growth of
the clusters, and thus, of the catalyst film itself proceeds via
attachment of new Co fragments to the exposed edges of these
3+
B + {Co } → C
Here, a reversible one electron, three proton equilibrium is
followed by a chemical rate limiting step for catalyst formation.
In this sequence, A, B, and C designate surface species whereas
Co ions in solution are given in brackets. The nature of the
species is discussed below. This mechanistic sequence defined
by eqs 5 and 6 implies that phase transfer of Co ions from
solution to the film occurs as part of the rate-limiting step, a
contention that is supported by the following kinetics analysis.
The velocity of the reaction, v, is given by
clusters. Given that Co2+ binding to MePO3 cannot account
2−
for the inverse first order dependence on MePi at intermediate
buffer strengths, these exposed edge sites are proposed to be
coordinated to MePi buffering species in the resting state. This,
3+
v = k θ [{Co }]
(7)
2 B
2+
taken together with the first pKa of Co(OH2)6 of 9.2 (see SI
where k2 is the rate constant for the chemical step, [{Co3+}] is
the concentration of the {Co3+} species in solution, and θB is
the surface coverage of the solid phase intermediate
participating in the rate-limiting chemical transformation. In
this formalism, θB is the surface concentration of B (Γ, in mol/
cm2) divided by the maximum surface concentration (Γmax).
We note that the number of active sites for film growth on the
surface may represent a small proportion of the total number of
solvent exposed Co centers in the film. Thus, Γmax does not
equal the number of exposed Co centers per cm2, but rather the
number of active sites for film growth per cm2, and θB represents
the fraction of these active sites in the form of intermediate B.
for details of calculation), establishes that Co(OH2)6 is the
2+
resting-state species in solution.
Figure 10 shows a proposed mechanistic model for Co-OEC
film growth that is consistent with the surface and solution
resting-state species and the electrochemical rate law given by
eq 4. A portion of the growing cobaltate cluster is shown to
present the surface active sites from which growth may occur.
In this mechanistic model, surface and solution phase reactions
exist in equilibrium prior to the rate limiting phase transfer of
Co to the growing catalyst surface. Given that the net reaction
to form the catalyst consists of a three proton, one electron
oxidation, Co2+ to Co(O)OH clusters, and that all of these
6333
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
proton and electron transfers exist in equilibrium, we believe it
is unlikely that phase transfer, which is expected to be
irreversible, occurs prior to the completion of each equilibrium
step. Thus, in solution, the Co(OH2)62+ is proposed to undergo
a one electron, two proton minor equilibrium PCET reaction
to form a Co(OH)2(OH2)4+ intermediate. It is known that the
corresponding Co(III) hexaaquo complex has a pKa of 2.9.68
Inasmuch as deposition is performed at pH 7−9, it is
reasonable to invoke a second deprotonation of this species.
A surface equilibrium involving MePO32− dissociation accounts
for the inverse first order dependence on MePi concentration.
Since the cobaltate cluster most closely corresponds to an array
of CoO(OH) units, a third deprotonation is expected and is in
line with the overall inverse third order dependence of the
experimental rate law. Whether this occurs from a species in
solution or on the growing catalyst cluster is not known; in
Figure 10, we show it to occur from a surface intermediate
+
followed by rate-limiting binding of the Co(OH)2(OH2)4 in
solution to effect catalyst growth by one Co center (highlighted
in pink in Figure 10). We note that alternative atomistic models
for the deprotonation and buffer dissociation equilibria may be
envisioned which are kinetically indistinguishable. Notwith-
standing, the mechanistic sequence presented in Figure 10 can
be repeated with adjacent surface sites to permit steady state
growth in a fashion that is consistent with the experimental rate
law.
Figure 11. pH dependence of the potential for catalyst film formation
(black ) and oxygen evolution (red ) at jac = 30 μA/cm2. Data for
film formation were interpolated from Tafel plots in Figure S5. Data
for oxygen evolution are reproduced from ref 54.
●
▲
dissolution experiments would serve to increase the potential
required for electrodeposition by ∼60 mV and would cause the
curves in Figure 11 to meet at ∼pH 5.7, consistent with the
experimental corrosion data. This predicted dependence of the
pH regime of catalyst stability on Co2+ concentration also
serves to explain the recent observation in the literature69 that 1
mM Co2+ must be added to electrolyte solutions during catalyst
operation to maintaining functional stability at pH 3.7. These
observations indicate that control over the deposition potential
through appropriate choice of electrolyte, pH, and Co2+
concentration is critical for enabling water oxidation and self-
repair at intermediate pH.
The foregoing mechanism not only provides insight into the
in situ catalyst formation but also sheds light on the self-repair
mechanism of Co-OEC. Both formation and repair processes
arise from the fact that, at > pH 7, the potentials necessary to
sustain catalyst film formation and growth are below that
required for water oxidation. Thus, upon application of a
potential sufficient to generate oxygen, ample driving force
exists to redeposit any Co in solution that may have leached
from the film between catalyst operation cycles. Here, we find
that the electrodeposition process exhibits an inverse third
order dependence on proton activity, which is much greater
than the inverse first order dependence observed for water
oxidation catalysis mediated by Co-OEC catalyst films.54 This
disparity in proton order suggests that, as the pH is lowered,
the potential necessary for catalyst formation will increase more
quickly than the corresponding rise in potential required for
water oxidation catalysis. This trend is shown graphically in
Figure 11, where the potential versus pH profiles for water
oxidation catalysis54 and catalyst formation, both proceeding at
the same activation controlled rate, are overlaid. While a sizable
separation of 0.31 V is observed for the two processes at pH 8,
extrapolation of the catalyst formation profile indicates that the
two processes occur at the same potential at pH 5.2. Under the
assumption that the rate law for catalyst formation remains
unchanged down to pH 5.2, this crossover point provides a
crude estimate of the pH regime in which in situ formation and
catalyst self-repair remains operative. Indeed, direct measure-
ment of the catalyst dissolution over a wide pH range (Figure
8) is qualitatively consistent with the prediction offered by
Figure 11. Experimentally, no dissolution is detected at pH 6−7
whereas catalyst corrosion prevails at pH 5 and pH 4. In the
catalyst corrosion experiments, complete catalyst dissolution
would only give rise to a solution concentration of 50 μM Co2+,
approximately an order of magnitude lower than the 0.4 mM
Co2+ concentration at which the data in Figure 11 were
collected. Noting the first order Co2+ concentration depend-
ence in the rate law, the more dilute conditions found in the
CONCLUSION
■
A self-consistent model of the nucleation, growth, and repair
processes of Co-OEC catalyst films has been constructed by
using a combination of electrochemical, NMR, and AFM
techniques. The nucleation of the catalyst proceeds by a
progressive mechanism giving rise to islands, whose surface
density can be controlled by modulating the electrodeposition
potential and the duration of electrodeposition relative to the
characteristic chronoamperometric peak. While future studies
are needed to extend these findings to catalyst nucleation and
growth on specific semiconductor substrates, these results
provide a mechanistic foundation that may be leveraged to
control the microstructure of Co-OEC functionalized photo-
anodes. The steady-state data for film growth point to a three
proton, one electron PCET equilibrium prior to the rate
limiting chemistry of Co association to the catalyst surface. The
electrolyte assumes countering roles of facilitating rapid
multiproton-coupled electron transfer while simultaneously
inhibiting catalyst growth through surface adsorption. Togeth-
er, the kinetic profiles of catalyst formation define the pH and
electrolyte regimes of catalyst stability and self-repair, establish-
ing a rational framework for future studies aimed at controlling
the properties of the semiconductor−solution interface for the
development of improved direct solar-to-fuels materials and
devices.
6334
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
(20) Gao, Y.; Akermark, T.; Liu, J. H.; Sun, L. C.; Akermark, B. J. Am.
Chem. Soc. 2009, 131, 8726−8727.
ASSOCIATED CONTENT
* Supporting Information
Full experimental details, additional nucleated growth traces,
AFM images, Tafel plots, Koutecky−Levich plots, and NMR
line broadening calibration curves. This material is available free
of charge via the Internet at http://pubs.acs.org.
■
S
(21) Dogutan, D. K.; McGuire, R. Jr.; Shao-Horn, Y.; Nocera, D. G.
J. Am. Chem. Soc. 2011, 133, 9178−9180.
́
(22) Brimblecombe, R.; Swiegers, G. F.; Dismukes, G. C.; Spiccia, L.
Angew. Chem., Int. Ed. 2008, 47, 7335−7338.
(23) Hocking, R. K.; Brimblecombe, R.; Chang, L.-Y.; Singh, A.;
Cheah, M. H.; Glover, C.; Casey, W. H.; Spiccia, L. Nat. Chem. 2011,
3, 461−466.
AUTHOR INFORMATION
Corresponding Author
nocera@mit.edu
■
(24) Geletii, Y. V.; Botar, B.; Kogerler, P.; Hillesheim, D. A.; Musaev,
̈
D. G.; Hill, C. L. Angew. Chem., Int. Ed. 2008, 47, 3896−3899.
(25) Geletii, Y. V.; Besson, C.; Hou, Y.; Yin, Q.; Musaev, D. G.;
Notes
Quinonero, D.; Cao, R.; Hardcastle, K. I.; Proust, A.; Kogerler, P.; Hill,
̃
̈
The authors declare no competing financial interest.
C. L. J. Am. Chem. Soc. 2009, 131, 17360−17370.
(26) Yin, Q.; Tan, J. M.; Besson, C.; Geletti, Y. V.; Musaev, D. G.;
Kuznetsov, A. E.; Luo, Z.; Hardcastle, K. I.; Hill, C. L. Science 2010,
328, 342−345.
ACKNOWLEDGMENTS
■
We thank D. Kwabena Bediako for many productive
discussions. This research was supported by a Center for
Chemical Innovation of the National Science Foundation (CCI
Powering the Planet, Grants CHE-0802907), the DOE
Catalysis and Solar Photochemistry programs, and a grant
from the Chesonis Family Foundation. Y.S. gratefully acknowl-
edges the National Science Foundation for a predoctoral
fellowship. D.A.L. gratefully acknowledges the Jane Coffin
Childs Memorial Fund for Medicinal Research for a
postdoctoral fellowship.
(27) Stracke, J. J.; Finke, R. G. J. Am. Chem. Soc. 2011, 133, 14872−
14875.
(28) Chen, Z.; Concepcion, J. J.; Luo, H.; Hull, J. F.; Paul, A.; Meyer,
T. J. J. Am. Chem. Soc. 2010, 132, 17670−17673.
(29) Concepcion, J. J.; Jurss, J. W.; Templeton, J. L.; Meyer, T. J. J.
Am. Chem. Soc. 2008, 130, 16462−16463.
(30) Concepcion, J. J.; Jurss, J. W.; Norris, M. R.; Chen, Z.;
Templeton, J. L.; Meyer, T. J. Inorg. Chem. 2010, 49, 1277−1279.
(31) Bernet, L.; Lalrempuia, R.; Ghattas, W.; Mueller-Bunz, H.;
Vigara, L.; Llobet, A.; Albrecht, M. Chem. Commun. 2011, 47, 8058−
8060.
REFERENCES
■
(32) Romain, S.; Vigara, L.; Llobet, A. Acc. Chem. Res. 2009, 42,
(1) Nocera, D. G. Inorg. Chem. 2009, 48, 10001−10017.
(2) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2006, 103,
15729−15735.
1944−1953.
(33) Wasylenko, D. J.; Ganesamoorthy, C.; Henderson, M. A.;
Koivisto, B. D.; Osthoff, H. D.; Berlinguette, C. P. J. Am. Chem. Soc.
2010, 132, 16094−16106.
(3) Abbott, D. Proc. IEEE 2010, 98, 42−66.
(4) Cook, T. R.; Dogutan, D. K.; Reece, S. Y.; Surendranath, Y.;
Teets, T. S.; Nocera, D. G. Chem. Rev. 2010, 110, 6474−6502.
(5) Cukier, R. I.; Nocera, D. G. Annu. Rev. Phys. Chem. 1998, 49,
337−369.
(34) Wasylenko, D. J.; Ganesamoorthy, C.; Koivisto, B. D.;
Henderson, M. A.; Berlinguette, C. P. Inorg. Chem. 2010, 49, 2202−
2209.
(35) Nocera, D. G. ChemSusChem 2009, 2, 387−390.
(6) Huynh, M. H. V.; Meyer, T. J. Chem. Rev. 2007, 107, 5004−5064.
(7) Hammes-Schiffer, S. Acc. Chem. Res. 2009, 42, 1881−1889.
(8) Siegbahn, P. E. M. J. Am. Chem. Soc. 2009, 131, 18238−18239.
(9) Concepcion, J. J.; Jurss, J. W.; Brennaman, M. K.; Hoertz, P. G.;
Patrocinio, A. O. T.; Murakami Iha, N. Y.; Templeton, J. L.; Meyer, T.
J. Acc. Chem. Res. 2009, 42, 1954−1965.
(10) Betley, T. A.; Wu, Q.; Van Voorhis, T.; Nocera, D. G. Inorg.
Chem. 2008, 47, 1849−1861.
(11) Betley, T. A.; Surendranath, Y.; Childress, M. V.; Alliger, G. E.;
Fu, R.; Cummins, C. C.; Nocera, D. G. Phil. Trans. R. Soc., B 2008,
363, 1293−1303.
(36) Kanan, M. W.; Nocera, D. G. Science 2008, 321, 1072−1075.
(37) Surendranath, Y.; Dinca,
2009, 131, 2615−2620.
(38) Dinca, M.; Surendranath, Y.; Nocera, D. G. Proc. Natl. Acad. Sci.
̆
M.; Nocera, D. G. J. Am. Chem. Soc.
̆
U.S.A. 2010, 107, 10337−10341.
(39) Esswein, A. J.; Surendranath, Y.; Reece, S. R.; Nocera, D. G.
Energy Environ. Sci. 2011, 4, 499−504.
(40) Lutterman, D. A.; Surendranath, Y.; Nocera, D. G. J. Am. Chem.
Soc. 2009, 131, 3838−3839.
(41) Kanan, M. W.; Surendranath, Y.; Nocera, D. G. Chem. Soc. Rev.
2009, 38, 109−114.
(12) Dempsey, J. L.; Esswein, A. J.; Manke, D. R.; Rosenthal, J.;
Soper, J. D.; Nocera, D. G. Inorg. Chem. 2005, 44, 6879−6892.
(13) Chang, C. J.; Chang, M. C. Y.; Damrauer, N. H.; Nocera, D. G.
Biophys. Biochim. Acta 2004, 1655, 13−28.
(42) Zhong, D. K.; Sun, J.; Inumaru, H.; Gamelin, D. R. J. Am. Chem.
Soc. 2009, 131, 6086−6087.
(43) Zhong, D. K.; Gamelin, D. R. J. Am. Chem. Soc. 2010, 132,
4202−4207.
(14) Surendranath, Y.; Nocera, D. G. Prog. Inorg. Chem. 2011, 57,
505−560.
(44) Steinmiller, E. M. P.; Choi, K. S. Proc. Natl. Acad. Sci. U.S.A.
2009, 106, 20633−20636.
(15) McDaniel, N. D.; Coughlin, F. J.; Tinker, L. L.; Bernhard, S. J.
Am. Chem. Soc. 2008, 130, 210−217.
(45) Seabold, J. A.; Choi, K. S. Chem. Mater. 2011, 23, 1105−1112.
(46) Zhong, D. K.; Cornuz, M.; Sivula, K.; Gratzel, M.; Gamelin, D.
̈
(16) Schley, N. D.; Blakemore, J. D.; Subbaiyan, N. K.; Incarvito, C.
D.; D’Souza, F.; Crabtree, R. H.; Brudvig, G. W. J. Am. Chem. Soc.
2011, 133, 10473−10481.
R. Energy Environ. Sci. 2011, 4, 1759−1764.
(47) McDonald, K. J.; Choi, K.-S. Chem. Mater. 2011, 23, 1686−
1693.
(17) Blakemore, J. D.; Schley, N. D.; Balcells, D.; Hull, J. F.; Olack,
G. W.; Incarvito, C. D.; Eisenstein, O.; Brudvig, G. W.; Crabtree, R. H.
J. Am. Chem. Soc. 2010, 132, 16017−16029.
(48) Zhong, D. K.; Choi, S.; Gamelin, D. R. J. Am. Chem. Soc. 2011,
133, 18370−18377.
(49) Pijpers, J. J. H.; Winkler, M. T.; Surendranath, Y.; Buonassisi, T.;
Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 10056−10061.
(50) Reece, S. Y.; Hamel, J. A.; Sung, K.; Jarvi, T. D.; Esswein, A. J.;
Pijpers, J. J. H.; Nocera, D. G. Science 2011, 334, 645−648.
(51) Nocera, D. G. Acc. Chem. Res. 2012, in press.
(18) Hull, J. F.; Balcells, D.; Blakemore, J. D.; Incarvito, C. D.;
Eisenstein, O.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 2009,
131, 8730−8731.
(19) Grotjahn, D. B.; Brown, D. B.; Martin, J. K.; Marelius, D. C.;
Abadjian, M.-C.; Tran, H. N.; Kalyuzhny, G.; Vecchio, K. S.; Specht, Z.
G.; Cortes-Llamas, S. A.; Miranda-Soto, V.; van Niekerk, C.; Moore, C.
E.; Rheingold, A. L. J. Am. Chem. Soc. 2011, 133, 19024−19027.
(52) Kay, A.; Cesar, I.; Gratzel, M. J. Am. Chem. Soc. 2006, 128,
̈
15714−15721.
6335
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336

Journal of the American Chemical Society
Article
(53) Su, J.; Feng, X.; Sloppy, J. D.; Guo, L.; Grimes, C. A. Nano Lett.
2010, 11, 203−208.
(54) Surendranath, Y.; Kanan, M. W.; Nocera, D. G. J. Am. Chem. Soc.
2010, 132, 16501−16509.
(55) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
Fundamentals and Applications; Wiley: New York, 2001; pp 340−344.
(56) Gileadi, E. Electrode Kinetics for Chemists, Chemical Engineers, and
Materials Scientists; Wiley-VCH: New York, 1993; pp 127−184.
(57) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
Fundamentals and Applications; Wiley: New York, 2001; Chapter 5.
(58) Scharifker, B.; Hills, G. Electrochim. Acta 1983, 28, 879−889.
(59) Zheng, M.; West, A. C. J. Electrochem. Soc. 2004, 151, C502−
C507.
(60) Mirkin, M. V.; Nilov, A. P. J. Electroanal. Chem. 1990, 283, 35−
51.
(61) Sluyters-Rehbach, M.; Wijenberg, J. H. O. J.; Bosco, E.; Sluyters,
J. H. J. Electroanal. Chem. 1987, 236, 1−20.
(62) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
Fundamentals and Applications; Wiley: New York, 2001; p 35.
̌
(63) Kanan, M. W.; Yano, J.; Surendranath, Y.; Dinca, M.; Yachandra,
V. K.; Nocera, D. G. J. Am. Chem. Soc. 2010, 132, 13692−13701.
(64) Haynes, W. M.; Lide, D. R. CRC Handbook of Chemistry and
Physics, 91st ed., Internet Version; CRC Press: Boca Raton, FL, 2011;
pp 8−21.
(65) Sigel, H.; Chen, D.; Corfu, N. A.; Gregan, F.; Holy, A.; Strasak,
M. Helv. Chim. Acta 1992, 75, 2634−2656.
(66) Martell, A. E.; Smith, R. M.; Motekaitis, R. J. Database 46: NIST
Critically Selected Stability Constants of Metal Complexes, Version 8;
Texas A&M University: College Station, TX.
(67) Poltavets, V. V.; Croft, M.; Greenblatt, M. Phys. Rev. B 2006, 74,
125103/1−8.
(68) Sisley, M. J.; Jordan, R. B. Inorg. Chem. 2006, 45, 10758−10763.
(69) Gerken, J. B.; Landis, E. C.; Hamers, R. J.; Stahl, S. S.
ChemSusChem 2010, 3, 1176−1179.
6336
dx.doi.org/10.1021/ja3000084 | J. Am. Chem. Soc. 2012, 134, 6326−6336