UC Berkeley
UC Berkeley Previously Published Works
Title
An investigation of thin-film Ni-Fe oxide catalysts for the electrochemical evolution of oxygen.
Permalink
https://escholarship.org/uc/item/10n0r49b
Journal
Journal of the American Chemical Society, 135(33)
ISSN
0002-7863
Authors
Louie, Mary W
Bell, Alexis T
Publication Date
2013-08-01
DOI
10.1021/ja405351s
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
An Investigation of Thin-Film Ni−Fe Oxide Catalysts for the
Electrochemical Evolution of Oxygen
Mary W. Louie and Alexis T. Bell*
Joint Center for Artificial Photosynthesis, Materials Science Division, Lawrence Berkeley National Laboratory, California 94720,
United States
Department of Chemical and Biomolecular Engineering, University of California , Berkeley, Berkeley, California 94720, United States
*
S
Supporting Information
ABSTRACT: A detailed investigation has been carried out of
the structure and electrochemical activity of electrodeposited
Ni−Fe films for the oxygen evolution reaction (OER) in
alkaline electrolytes. Ni−Fe films with a bulk and surface
composition of 40% Fe exhibit OER activities that are roughly 2
orders of magnitude higher than that of a freshly deposited Ni
film and about 3 orders of magnitude higher than that of an Fe
film. The freshly deposited Ni film increases in activity by as
much as 20-fold during exposure to the electrolyte (KOH);
however, all films containing Fe are stable as deposited. The
oxidation of Ni(OH)
2
to NiOOH in Ni films occurs at
potentials below the onset of the OER. Incorporation of Fe into
the film increases the potential at which Ni(OH)
2
/NiOOH
redox occurs and decreases the average oxidation state of Ni in NiOOH. The Tafel slope (40 mV dec
−1
) and reaction order in
OH
−
(1) for the mixed Ni−Fe films (containing up to 95% Fe) are the same as those for aged Ni films. In situ Raman spectra
acquired in 0.1 M KOH at OER potentials show two bands characteristic of NiOOH. The relative intensities of these bands vary
with Fe content, indicating a change in the local environment of Ni−O. Similar changes in the relative intensities of the bands
and an increase in OER activity are observed when pure Ni films are aged. These observations suggest that the OER is catalyzed
by Ni in Ni−Fe films and that the presence of Fe alters the redox properties of Ni, causing a positive shift in the potential at
which Ni(OH)
2
/NiOOH redox occurs, a decrease in the average oxidation state of the Ni sites, and a concurrent increase in the
activity of Ni cations for the OER.
1. INTRODUCTION
The electrolysis of water to form hydrogen and oxygen offers a
possible means for storing energy obtained from intermittent
sources such as the sun.
1,2
However, the electrolysis of water
requires voltages in substantial excess of the thermodynamic
potential for water-splitting (H
2
O → H
2
+ 1/2O
2
), 1.23 V, due
primarily to the slow kinetics of the oxygen evolution reaction
(OER).
1
At present, the most active catalysts for the OER are
RuO
2
and IrO
2
, but even these operate with overpotentials in
excess of 200 mV (at a current density of 10 mA cm
−2
).
1−3
Moreover, the scarcity of Ru and Ir makes it impractical to use
the metals on a large scale. For these reasons, there has been
considerable interest in the discovery and development of OER
catalysts based on earth-abundant metals.
A review of the literature suggests that Ni−Fe catalysts offer
a promising alternative to catalysts based on precious
metals.
4−12
The lowest overpotential reported, ∼230 mV at
10 mA cm
−2
for electrodeposited Ni−Fe films,
4
is comparable
to overpotentials of 280 and 220 mV for chemically synthesized
IrO
2
and RuO
2
films.
3
It is also notable that Ni−Fe catalysts are
significantly more active for oxygen evolution than either Ni or
Fe alone, which exhibit overpotentials of 350−450 and ∼500
mV, respectively, at 10 mA cm
−2
.
13,14
While the beneficial
effects of Fe on the OER activity of Ni have been reported in a
number of studies,
7,10−15
little is known about the structural or
chemical characteristics of Ni−Fe catalysts, particularly under
conditions where the OER occurs. A small number of in situ
spectroscopic studies have been carried out but yield
contradictory results. Ni−Fe catalysts (9−20 at % Fe)
characterized by either in situ X-ray absorption or Mossbauer
spectroscopy, at potentials relevant for oxygen evolution, have
been found to contain Ni(III), but the oxidation state of Fe is
not clearly defined; some authors conclude that it is Fe(III)
6,15
and others, that it is Fe(IV).
16−18
We report here a structural and electrochemical investigation
of Ni−Fe catalysts used for the OER in alkaline electrolyte. The
structure of the electrodeposited Ni−Fe film was characterized
in situ by Raman spectroscopy as a function of the applied
potential, and the surface compositions were determined by ex
situ X-ray photoelectron spectroscopy (XPS). The observed
Raman characteristics of the Ni−Fe series combined with the
Received: May 28, 2013
Published: July 16, 2013
Article
pubs.acs.org/JACS
© 2013 American Chemical Society 12329 dx.doi.org/10.1021/ja405351s | J. Am. Chem. Soc. 2013, 135, 12329−12337
Ni redox behavior and the kinetic parameters for the OER are
used to correlate structural characteristics of the catalyst under
reaction conditions with the catalyst composition and the
catalytic activity.
2. EXPERIMENTAL SECTION
2.1. Electrocatalyst Preparation. Electrocatalyst films containing
Ni and/or Fe were prepared by electrodeposition onto gold electrodes.
These gold substrates (99.95%, DOE Business Center for Precious
Metals Sales and Recovery (BCPMSR)) were sheathed in Teflon,
leaving exposed surfaces which were either 4 or 5 mm in diameter.
Prior to electrodeposition, the gold substrates were polished with
alumina slurries, from 5 μm down to 50 nm in particle size, with
15 min of sonication in water after each polish. Electrodes which were
used for Raman measurements were additionally roughened, in KCl
(Sigma-Aldrich, P3911), using a previously reported electrochemical
cycling procedure in order to produce substrates that are ideal for
surface-enhanced Raman measurements.
19
(For details see section
S1.1, Supporting Information [SI].) Both smooth and roughened gold
substrates were used for electrochemical measurements; the OER
activity of these substrates was verified to be negligible compared to
that of the films measured in this work (Figure S5, SI).
Electrolyte so lutions for electrodeposition were o btained by
dissolving nickel sulfate hexahydrate (≥99.99% trace metals basis,
Sigma-Aldrich 467901) and/or iron sulf ate heptahydra te (ACS
Reagent ≥99.0%, Sigma-Aldrich 215422) in ultr apure water
(18.2 MΩ, EMD Millipore). The metal concentrations were varied
between 0 and 0.01 M such that deposition baths with less than 50%
Fe contained 0.01 M Ni and a varying Fe concentration, and baths
with greater than 50% Fe contained 0.01 M Fe and a varying Ni
concentration. Prior to dissolution of the appropriate quantities of
sulfate salts, the water was sparged with nitrogen gas for 1 h to prevent
oxidation of Fe(II) to Fe(III) and enable well-controlled depositions.
Films were electrodeposited galvanostatically with a cathodic current
density of 50 μAcm
−2
applied for 1125 s and with nitrogen gas flowing
in the headspace. The potential of the working electrode was
monitored using a Ag/AgCl reference electrode with 4 M KCl filling
solution (Pine RREF0021) or a leakless Ag/AgCl electrode with 3.4 M
KCl filling solution (eDaq ET072). The counterelectrode was a coiled
Pt wire (99.95%, DOE BCPMSR) which was routinely soaked in 5 M
nitric acid to remove any deposited Ni or Fe.
Aged Ni films were prepared by immersion of as-deposited Ni films,
without potential cycling, in 10 M KOH solutions for over 24 h.
Immersion of the films was carried out in Teflon or polypropylene
cells/containers so as to minimize contaminants from glassware.
Electrochemical and in situ Raman characterization of these films was
carried out after the Ni redox peaks and OER currents observed in the
cyclic voltammograms were stable, typically 5−10 cycles.
2.2. Physical Characterization. The compositions and quantities
of the electrodeposited materials were determined by elemental
analysis. Electrodeposited films were dissolved by sonication in high-
purity 5 M HNO
3
(Sigma-Aldrich 84385 or EMD Millipore NX0407).
These solutions were diluted and adjusted such that the final solutions
contained 5% w/w HNO
3
and 1000 ppb of yttrium internal standard
(Sigma Aldrich 01357). Samples were characterized by inductively
coupled plasma optical emission spectroscopy (ICP Optima 7000 DV,
Perkin-Elmer). Calibration solutions contained 5% w/w HNO
3
, 1000
ppb Y, and both Ni and Fe, each with concentrations between 0 and
2000 ppb (Sigma-Aldrich 28944 and 43149 for Ni and Fe sources,
respectively). On the basis of these measurements, we estimate the
thicknesses of Ni−Fe films to be ∼25 and ∼70 nm when deposited on
roughened and polished gold substrates, respectively. (For details see
section S2.2, SI.)
The surface compositions of the electrodeposited Ni−Fe films were
determined by X-ray photoelectron spectroscopy (XPS). Both as-
deposited Ni−Fe films and films which were polarized under OER
conditions (335 mV overpotential in 0.1 M KOH) for 1−2 h were
examined. The XPS measurements were carried out with a Kratos Axis
Ultra spectrometer using a monochromatic Al Kα source (15 mA,
15 kV). The instrument base pressure was 10
−9
Torr, and the charge
neutralizer system was used for all measurements. High-resolution
spectra for Ni 2p, Fe 2p, O 1s, and C 1s were collected using a pass
energy of 20 eV, a step energy of 50 meV, and dwell times of 200−
400 ms. Angle-resolved XPS measurements were carried out by
varying the electron takeoff angle between 0° and 75° with respect to
the sample normal. The resulting spectra were analyzed using
CasaXPS (Casa Software, Ltd.). A standard Shirley baseline with no
offset was used for background correction. In the case of Fe 2p spectra,
an additional correction was necessary due to the presence of a Ni
LMM Auger peak. (For details see sections S4.2−S4.3, SI.) The C 1s
spectrum for adventitious carbon (284.8 eV) was used for charge
correction.
2.3. Electrochemical Characterization. Electrochemical charac-
terization of the Ni−Fe catalysts was carried out in KOH electrolytes
(ACS reagent ≥85%, Sigma-Aldrich 221473) with concentrations of
0.1−4.6 M in ultrapure water. (This KOH source is specified by the
supplier to have ≤0.001% Fe and ≤0.001% Ni.) A Hg/HgO reference
electrode (CH Instruments, ET072) with 1 M KOH filling solution
was used throughout the experiments; the fillin g solution w as
exchanged before each experiment and measured against a second,
unused Hg/HgO reference electrode stored in 1 M KOH. The counter
electrode was a coiled Pt wire, cleaned routinely by nitric acid to
remove any accumulated Ni or Fe deposits. All potentials reported in
this work, unless otherwise noted, are measured against this Hg/HgO
(1 M KOH) reference which has a potential of 0.098 V vs the normal
hydrogen electrode (NHE). The equilibrium potential for oxygen
evolution at any given pH is therefore (1.23 − 0.098 − 0.059 × pH) V.
Two electrochemical cells were used to measure the current−
voltage characteristics of the Ni−Fe catalysts. One was a home-built
Teflon cell
20
which was designed for efficient collection of Raman
signals. The second was a rotating disk electrode (RDE) apparatus
(Pine Instruments) employed for additional electrochemical character-
ization of Ni−Fe films, particularly for acquiring data for analysis of
OER kinetics in the absence of mass transfer effects. Measurements
were carried out using either a Gamry Reference 600 or a BioLogic
VSP potentiostat. IR compensation was applied at 85−95% using the
ohmic resistance determined by AC impedance methods. Specifically,
impedance spectra were obtained at 0 ± 10 mV (vs Hg/HgO)
between 1 MHz and 10 mHz, and the ohmic contribution was
estimated from the Nyquist plots. For 1 and 0.1 M KOH electrolytes,
the ohmic resistances were typically ∼5 and ∼40 Ω, respectively.
When necessary, compensation of the remaining 5−15% of the ohmic
resistance was applied manually to the current−potential data. This
procedure, while having negligible impact on the shape of the redox
features, was necessary for acquiring accurate OER currents and
therefore Tafel slopes, particularly for measurements in electrolytes
with lower KOH concentrations and/or for electrocatalysts with high
OER activity. It should be noted that the compensated resistance is
noticeably higher during Raman spectra acquisition in 0.1 M KOH
(100−200 Ω), since immersion of the objective reduces the effective
cross section for ionic conduction.
2.3.1. Surface Area. The surface areas of the films were estimated
by measuring the electrochemical capacitance of the film−electrolyte
interface in the double-layer regime of the voltammograms. Using
0.1 M KOH, the electrode was potentiostatically cycled, typically
between and 0.1 and 0.16 V vs Hg/HgO (1 M KOH), at scan rates
between 1 and 10 mV s
−1
until the measured voltammograms had
stabilized. The positive and negative capacitance currents at the center
of the potential window were averaged and plotted against the scan
rate to extract the measured capacitance. The surface areas reported
were obtained by using a specific capacitance of 60 μFcm
−2
for
oxides.
21
(Representative plots are provided in section S2.3 of the SI.)
The measured currents were normalized by this area to obtain the
specific current density. The challenges associated with determining
the surface area of catalysts have been reviewed by Trasatti and
Petrii.
21
Extracted values of the specific current density, while reliable
for comparing across the Ni−Fe system, should be used with care
when comparing to other catalysts reported in the literature.
Additional discussion is provided in the SI, section S2.3.
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja405351s | J. Am. Chem. Soc. 2013, 135, 12329−1233712330
2.3.2. Cyclic Voltammograms. Cyclic voltammograms were
recorded between 0 and 0.7−0.8 V vs Hg/HgO in 0.1 M KOH,
with the high-potential limit adjusted so as to minimize the amount of
current driven through the high-activity catalysts. For the same reason,
the potential window was adjusted by ∼60 mV per unit pH, since
higher OER currents were observed at higher KOH concentrations.
Voltammograms were recorded until the redox peaks and the oxygen
evolution currents show negligible change, typically 5−10 cycles, with
the exception of as-deposited Ni films which slowly age/transform in
alkaline electrolytes.
22
While low scan rates are desirable for obtaining
current−voltage curves for kinetic analysis, we chose a scan rate of
10 mV s
−1
to limit the overall time spent under oxygen evolution
conditions. At slower scan rates, increasingly more bubbles formed
under OER conditions and collected on or near the electrode surface.
The accumulation of bubbles on the electrode surface caused a drop in
the measured current due to coverage of the active sites and/or to the
additional ohmic resistance not accounted for by the uncompensated
resistance values measured under non-OER conditions. For experi-
ments carried out in the RDE apparatus, the rotation rate was varied
between 0 and 2400 rotations per minute (RPM).
The activity for the OER was determined from cyclic voltammo-
grams by reading the specific current density at either a constant
overpotential of 300 mV or the overpotential at a constant specific
current density of 10 mA cm
−2
. The Tafel slope was obtained from
data collected by rotating the electrode at 2400 rpm to reduce mass
transport effects and to increase the potential window for Tafel
analysis. In selecting the region for the Tafel fit, we avoided high
potentials at which oxygen bubble evolution causes mass transport
limitations, and low potentials at which the redox transition for Ni(II)/
Ni(III) occurs.
The dependence of the OER current density on the concentrations
of OH
−
and O
2
was determined. KOH concentrations of 0.1, 0.22,
0.46, 1.0, 2.2, and 4.6 M were used; these correspond to pHs of 13,
13.3, 13.7, 14, 14.3 and 14.7, respectively. Cyclic voltammograms were
collected for each electrolyte concentration; measurements were
repeated for the high and low concentrations of 0.1 and 4.6 M (pH 13
and 14.7) at the end of the concentration series to verify stability of
the catalyst films with changing KOH concentration. To examine
effect of O
2
concentration, voltammograms were measured in 0.1 M
KOH electrolyte sparged for one hour with either N
2
or O
2
prior to
measurement. It should be noted that KOH concentrations below pH
12.5 resulted in gradual changes in the nickel redox features and OER
current with time.
2.3.3. Calculation of the Turnover Frequency. The turnover
frequency (TOF) based on Ni sites can be computed in two ways.
Taking the Ni in the entirety of the film to be catalytically active, a
lower limit, TOF
min
, can be calculated using the number of Ni atoms
determined by elemental analysis. On the other hand, the upper limit,
TOF
max
, can be estimated by taking the measured surface area of a film
and computing the number of Ni atoms at the surface using a value of
6.4 × 10
14
Ni atoms per cm
2
area
23
and the surface fraction of Ni as
determined by XPS. Although TOF
min
and TOF
max
differ by 1−2
orders of magnitude, the dependence of TOF on composition is not
affected. Here, we report TOF
min
, but both TOF
min
and TOF
max
are
presented in the SI section S7.
2.4. In Situ Raman Spectroscopy. Raman spectra were acquired
under controlled electrochemical potentials using a homemade Teflon
cell, which contained a working electrode (4 mm Au, Teflon-sheathed)
oriented at the bottom of the cell.
20
We employed a water-immersion
objective (70× mag., N. A. = 1.23, LOMO) which was protected from
the corrosive KOH electrolytes by a 0.001-in. thick fluorinated
ethylene propylene film (McMaster-Carr) or 0.0005-in. thick Teflon
film (American Durafilm); a droplet of water was placed between the
objective lens and the film to retain the high illumination/collection
efficiencies. Additional details are provided in ref 20.
Raman spectra were collected using a confocal Raman microscope
(LabRAM HR, Horiba Yvon Jobin) with a wavelength of 633 nm and
a power of 1−3 mW at the objective. The spot size of the laser beam is
estimated to be between 1 and 2 μm. Acquisition times for Ni−Fe
films were typically 3 s for spectral range of 1100 cm
−1
window. Using
a 600 g/mm grating, the spectral resolution is ∼1cm
−1
. Spectral shifts
were calibrated routinely against the value of 520.7 cm
−1
for a silicon
wafer. Raman spectra were collected at selected potentials as the
potential of working electrode was scanned at a rate of 1 mV s
−1
.
Raman spectra were not background-corrected due to the complexity
of SERS backgrounds. When quantifying relative changes in the peak
heights of Raman bands (carried out for cases when surface-
enhancement contributions are minimized), a linear background
correction was used. The sampling depth/volume of Ni−Fe films by
surface-enhanced Raman spectroscopy is discussed in the SI, section
S5.5.
3. RESULTS AND DISCUSSION
3.1. Electrochemical Characteristics of Ni−Fe Films.
Representative cyclic voltammograms for Ni and Fe films, i.e.,
the end-members of the catalyst series, are shown Figure 1. The
cyclic voltammogram for an as-deposited Ni film in 0.1 M
KOH (Figure 1a) exhibits two primary characteristics, a redox
couple at 0.47 V vs Hg/HgO (1 M KOH) and a positive
(oxidation) current visible at overpotentials greater than 0.65 V.
Both are well-known features for Ni electrodes in alkaline
electrolytes. The redox peaks are attributed to the trans-
formation between Ni(OH)
2
and NiOOH,
14,22
which proceeds
as Ni(OH)
2
+OH
−
↔ NiOOH + H
2
O+e
−
in alkaline
electrolytes. Oxidation currents at higher potentials are due to
the evolution of oxygen, 4OH
−
→ O
2
+2H
2
O+4e
−
.
The cyclic voltammogram for Fe in 0.1 M KOH (Figure 1b)
shows an oxidation current attributed to the OER visible at
Figure 1. Cyclic voltammograms for (a) Ni and (b) Fe films deposited on an Au substrate, measured in 0.1 M KOH at room temperature with
10 mV s
−1
scan rate and 2400 rpm. The equilibrium potential for OER is 0.365 V vs Hg/HgO (1 M KOH).
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja405351s | J. Am. Chem. Soc. 2013, 135, 12329−1233712331
potentials positive of 0.7 V. No redox transitions for Fe are
observed in this potential window, consistent with previous
reports in literature, since the oxidation of metallic Fe to
Fe(II)/Fe(III) oxides/hydroxides occur at potentials between
−0.5 V and −1.2 V vs Hg/HgO (1 M).
13,24
The reduction peak
observed at 0.22 V is that for the underlying gold substrate
(Figure S1, SI); this feature is also present in the case of Ni
films but not apparent in Figure 1a due to the significantly
higher currents observed compared to that of Fe.
The cyclic voltammograms for mixed Ni−Fe films are
characterized by two primary features: one for the Ni(OH)
2
/
NiOOH redox couple and the other for the OER. Two
measures were defined for the OER activity, one being the
specific current density at a constant overpotential of 300 mV
and the other the overpotential at a specific current density of
10 mA cm
−2
. A plot of these two parameters as a function of
the surface composition of the Ni−Fe films, as determined by
X-ray photoelectron spectroscopy (section S.4, SI), reveals that
the OER current density varies across 3 orders of magnitude
with composition (Figure 2). The maximum specific current
density of 20 mA cm
−2
and the minimum overpotential of
280 mV were measured at a composition of 40% Fe. The
observation of this dramatic enhancement of the OER activity
indicates that Ni−Fe catalysts do not behave as simple mixtures
of the end-members and that interaction between Ni and Fe
results in an improvement in the rate of oxygen evolution. It is
notable that, across a composition range of 15−50% Fe, the
specific current density only varies by 2-fold (between 10 and
20 mA cm
−2
) and the overpotential by 20 mV. This behavior is
consistent with the discrepancies found in literature
4−10
regarding the composition of highest OER activity for the
Ni−Fe system; the optimum composition has been reported to
be as low as 10% Fe
6,7
and as high as 50% Fe.
4
The voltammograms for the Ni−Fe series differ noticeably in
their Ni(OH)
2
/NiOOH redox characteristics (Figure 3a). As
more Fe is incorporated into the Ni−Fe film, the Ni(OH)
2
/
NiOOH redox couple shifts to higher potentials, consistent
with previous reports,
4,5,7,10
and the peak area decreases (Figure
3a). At an Fe content of 41% or greater, the oxidation wave is
no longer visible due to its coincidence with the rapid rise in
the OER current, which increases initially with increasing Fe
content (Figure 2). However, films for which both oxidation
and reduction peaks are visible indicate that the two shift in
tandem and have comparable integrated areas (Figure S4a, SI).
Therefore, we used the reduction peak to quantify changes in
the redox characteristics of Ni−Fe films as a function of
composition. A strong linear correlation is observed between
the reduction peak potential and the Fe content (Figure 3b);
the reduction peak shifts to positive potentials by as much as
150 mV relative to that for pure Ni films when the film contains
70% Fe. This shift in the redox potential implies that the
electrochemical oxidation of Ni(OH)
2
to NiOOH is suppressed
by the presence of Fe. The reduction peaks were integrated in
order to determine the extent of Ni reduction/oxidation in the
Ni−Fe films. As shown in Figure 3c for pure Ni films deposited
atop roughened Au substrates, the charge passed during redox
is greater than can be accounted for by assuming a one-electron
redox reaction; we find that 1.2 electrons are transferred per Ni
atom in a redox cycle. (The number of Ni atoms deposited was
measured by elemental analysis.) This result is consistent with
reports that the Ni can exist as γ-NiOOH in which Ni has an
average oxidation state as high as 3.7,
14,25−28
and therefore that
the α-Ni(OH)
2
/γ-NiOOH transformation can involve the
transfer of up to 1.7 electrons per Ni atom. Figure 3c shows
that the number of electrons transferred during Ni(OH)
2
/
NiOOH redox depends strongly on the Fe content, decreasing
Figure 2. Oxygen evolution activity of electrodeposited Ni−Fe
catalysts, taken at 300 mV overpotential and 10 mA cm
−2
specific
current density, as a function of composition in 0.1 M KOH. Filled
and open markers correspond to measurements taken using rotating
and stationary Au working electrode substrates for which electro-
deposited films are ∼70 and ∼25 nm thick, respectively.
Figure 3. Effect of Fe on the Ni(OH)
2
/NiOOH redox couple in 0.1 M KOH. (a) Selected cyclic voltammograms for Ni−Fe films on polished gold
substrates collected at 10 mV s
−1
and 2400 rpm, (b) reduction peak potential, and (c) the number of electrons transferred during redox as a function
of Fe content, shown for films on roughened gold substrates. Lines in (b) and (c) indicate linear fits to the data.
Journal of the American Chemical Society Article
dx.doi.org/10.1021/ja405351s | J. Am. Chem. Soc. 2013, 135, 12329−1233712332
