Article

pubs.acs.org/JACS

Benchmarking Heterogeneous Electrocatalysts for the Oxygen

Evolution Reaction
Charles C. L. McCrory,*,§ Suho Jung,§ Jonas C. Peters,*,§,† and Thomas F. Jaramillo*,§,‡
§
Joint Center for Artificial Photosynthesis and †Division of Chemistry and Chemical Engineering, California Institute of Technology,
Pasadena, California 91125, United States
‡
Department of Chemical Engineering, Stanford University, Stanford, California 94305, United States
*
S Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Objective evaluation of the activity of electro-

catalysts for water oxidation is of fundamental importance for
the development of promising energy conversion technologies
Downloaded via UNIV OF HYDERABAD on June 27, 2024 at 06:27:30 (UTC).

including integrated solar water-splitting devices, water

electrolyzers, and Li-air batteries. However, current methods
employed to evaluate oxygen-evolving catalysts are not
standardized, making it difficult to compare the activity and
stability of these materials. We report a protocol for evaluating
the activity, stability, and Faradaic efficiency of electro-
deposited oxygen-evolving electrocatalysts. In particular, we focus on methods for determining electrochemically active surface
area and measuring electrocatalytic activity and stability under conditions relevant to an integrated solar water-splitting device.
Our primary figure of merit is the overpotential required to achieve a current density of 10 mA cm−2 per geometric area,
approximately the current density expected for a 10% efficient solar-to-fuels conversion device. Utilizing the aforementioned
surface area measurements, one can determine electrocatalyst turnover frequencies. The reported protocol was used to examine
the oxygen-evolution activity of the following systems in acidic and alkaline solutions: CoOx, CoPi, CoFeOx, NiOx, NiCeOx,
NiCoOx, NiCuOx, NiFeOx, and NiLaOx. The oxygen-evolving activity of an electrodeposited IrOx catalyst was also investigated
for comparison. Two general observations are made from comparing the catalytic performance of the OER catalysts investigated:
(1) in alkaline solution, every non-noble metal system achieved 10 mA cm−2 current densities at similar operating overpotentials
between 0.35 and 0.43 V, and (2) every system but IrOx was unstable under oxidative conditions in acidic solutions.

■ INTRODUCTION
The efficient electrochemical conversion of H2O to H2 and 1/2
In the case of components for commercial devices such as
photovoltaic cells and polymer-electrolyte membrane fuel cells
O2 catalyzed by materials comprised of earth-abundant (PEMFCs), widely accepted testing protocols and figures of
elements is of fundamental importance to solar-fuels merit exist.15−17 For instance, one figure of merit that has been
devices.1−10 Integrated solar water-splitting devices that couple used for inexpensive Pt-free oxygen reduction catalysts for
light-capturing semiconductors with electrocatalysts to effi- PEMFCs is the operating current per cm3 of catalyst material at
ciently split water show particular promise as a means of direct a fixed overpotential in acidic water at elevated temperature.15
production of fuels from sunlight.9 Identifying efficient In this example, the catalyst testing conditions are dictated by
electrocatalysts for the oxygen evolution reaction (OER) the operating conditions necessary for automotive
remains a key challenge in the production of solar fuel PEMFCs.15,18 While the purpose and operating conditions
generators,6,11,12 and methods for rapidly screening the OER for PEMFCs are significantly different than those for solar
activity of various metal oxides have recently been reported in water-splitting devices, the testing protocols developed for
the context of combinatorial catalyst synthesis.13,14 However, PEMFCs and other commercial devices nevertheless provide
objective evaluation of the efficiency of OER catalysts is inspiration in the development of a benchmarking methodology
complicated by the lack of standardization both in the for electrocatalysts for solar water-splitting devices.
measurement and reporting of electrocatalytic data. Typical Herein we report a procedure for evaluating the activity,
OER catalysts are deposited on a variety of different substrates, stability, and electrochemically active surface area for
and their electrocatalytic activity is measured at a range of pH heterogeneous OER catalysts under standard conditions
values, temperatures, and electrolyte compositions and shown in Figure 1. In particular, the electrochemically active
concentrations, making it difficult to compare the performance surface area (ECSA) of each catalyst is estimated from
and stability of different materials. As such, the development measurements of the double-layer capacitance; the activity
and implementation of a benchmarking methodology to test
the electrocatalytic efficiency of materials for OER remain a Received: July 11, 2013
fundamental challenge in solar fuels research. Published: October 30, 2013

© 2013 American Chemical Society 16977 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987
Journal of the American Chemical Society Article

equipment easily accessible to a typical researcher in the field of

electrocatalysis, and (2) to minimize the time and number of
experiments necessary to evaluate a catalyst’s activity and short-
term stability. This latter criterion is important to facilitate the
screening of large numbers of catalyst materials. The relevant
figure of merit is the overpotential required to achieve 10 mA
cm−2 current density per geometric area at ambient temper-
ature and 1 atm O2. This is approximately the current density
expected at the anode in a 10% efficient solar water-splitting
device under 1 sun illumination.9,19,20
The protocol outlined above has been used to evaluate and
compare the activity and stability of a variety of electro-
deposited catalysts for OER under conditions relevant to an
integrated solar water-splitting device. A representative
sampling of electrodeposited Ni- and Co-based metal oxide
catalysts was chosen for initial evaluation (Table 1).21−23,26−28
Each system was electrodeposited onto an inert, glassy carbon
(GC) electrode and studied in aqueous solutions of 1 M NaOH
and 1 M H2SO4. Glassy carbon was chosen as the electrode
support due to its relative inactivity for OER at moderate
overpotentials.29 1 M NaOH and 1 M H2SO4 were chosen as
the alkaline and acidic electrolytes because their respective
worldwide production volumes greatly exceed those of any
other strong acids and bases,30 making them likely candidates
for use as electrolytes in solar fuel devices.31 The electro-
catalytic efficiency of an IrOx standard25 and the GC
background were also investigated.

Figure 1. Protocol for measuring the electrochemically active surface

area, catalytic activity, stability, and Faradaic efficiency of heteroge-
neous electrocatalysts for OER.
■ EXPERIMENTAL SECTION
Materials. Materials were purchased in the grade indicated and
used as received. Ammonium hydroxide (NH4OH, BioUltra),
ammonium perchlorate (NH4ClO4, 99.999%), ammonium sulfate
((NH4)2SO4, 99.999%), boric acid (H3BO3, BioUltra), cobalt(II)
and stability of each catalyst material are measured using a
nitrate hexahydrate (Co(NO3)2·6 H2O, 99.999%), copper(II) sulfate
combination of voltammetry, chronoamperometry, and chro- pentahydrate (CuSO4·5 H2O, 99.995%), nickel(II) nitrate hexahydrate
nopotentiometry; and the Faradaic efficiency of each material is (Ni(NO3)2·6 H2O, 99.999%), nickel(II) sulfate hexahydrate (NiSO4·6
determined using a rotating ring-disk electrode (RRDE) H2O, 99.99%), iron(II) sulfate heptahydrate (FeSO4·7 H2O, ACS
apparatus. This procedure was developed to meet two specific 99%), potassium carbonate (K 2 CO 3 , 99.995%), potassium
criteria: (1) to use standard electrochemical procedures and hexachloroiridate(IV) dihydrate (K2IrCl6·2 H2O, 99.99%), sodium

Table 1. Catalyst Materials Investigated along with Their Deposition Solutions and Conditions
catalyst deposition solution (in 40 mL H2O) deposition conditions
CoOx-(a)21 0.202 g CoSO4·7 H2O, 0.164 g NH4ClO4, NH4OH to pH 6.8 cathodic deposition at −50 mA cm−2 for 30 sa,
1200 rpm
CoOx-(b) 0.006 g Co(NO3)2·6 H2O, 0.438 g Na2HPO4·2 H2O, 0.240 g NaH2PO4·2 H2O, pH 7 anodic deposition at 1.05 V vs SCE for 8 h in quiescent
(“CoPi”)22,23 solution
CoFeOx21 0.112 g CoSO4·7 H2O, 0.100 g FeSO4·7 H2O, 0.141 g NH4ClO4, NH4OH to pH 5.4 cathodic deposition at −27.5 mA cm−2 for 30 sa,
1200 rpm
IrOx24,25 see Experimental section anodic deposition at 1.40 V vs SCE for 600 s in ice
bath, 1200 rpm
NiOx26 1.047 g Ni(NO3)2·6 H2O cathodic deposition at −16 mA cm−2 for 10 s, 400 rpm

NiCeOx26 1.047 g Ni(NO3)2·6 H2O, 0.174 g Ce(NO3)3·6 H2O cathodic deposition at −16 mA cm−2 for 10 s, 400 rpm

NiCoOx27 2.63 g NiSO4·6 H2O, 2.81 g CoSO4·7 H2O, 6.44 g Na2SO4·10 H2O, 1.24 g H3BO3 cathodic deposition at −50 mA cm−2 for 15 min,
400 rpm
NiCuOx28 0.095 g NiSO4·6 H2O, 0.090 g CuSO4·5 H2O, 0.132 g (NH4)2SO4 Cathodic deposition at −47 mA cm−2 for 50 sa,
1200 rpm
NiFeOx21 0.095 g NiSO4·6 H2O, 0.100 g FeSO4·7 H2O, 0.117 g (NH4)2SO4, cathodic deposition at −50 mA cm−2 for 50 sa,
NH4OH/H2SO4 to pH 2.5 1200 rpm
NiLaOx26 1.047 g Ni(NO3)2·6 H2O, 0.173 g La(NO3)3·6 H2O cathodic deposition at −16 mA cm−2 for 10 s, 400 rpm

a
The reported cathodic deposition current density of these materials on Pt discs was −250 mA cm−2. The large reported deposition current density
may be due to background H2 evolution by the Pt substrate. We were unable to attain this current density when depositing onto GC disks and
instead deposited at the current densities listed.

16978 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987

Journal of the American Chemical Society Article

phosphate dibasic dihydrate (Na2HPO4·2 H2O, BioUltra), sodium meter with a Thermo Scientific Orion refillable Ag/AgCl pH electrode
phosphate monobasic dihydrate (NaH2PO4·2 H2O, BioUltra), and filled with Orion Ag/AgCl reference electrode filling solution. The pH
sodium hydroxide (NaOH, BioUltra) were purchased from Sigma- meter was calibrated with a 5-point calibration curve between pH 1.68
Aldrich. REacton grade cerium(III) nitrate hexahydrate (Ce(NO3)3·6 and 12.45.
H2O, 99.99%), cobalt(II) sulfate heptahydrate (CoSO4·7 H2O, For the deposition of IrOx,24,25 0.0580 g K2IrCl6 was added to 50
99.999%), and lanthanum(III) nitrate hexahydrate (La(NO3)3·6 mL of 0.1 M NaOH solution at pH 13 to form a 2.4 mM solution. The
H2O, 99.999%) were purchased from Alfa Aesar. TraceMetal grade solution was heated to 90 °C for 20 min, and the resulting blue
98% sulfuric acid (H2SO4) was purchased from Fisher Scientific. All solution was placed in an ice bath and allowed to cool. After the
water used was first purified by a Thermo Scientific Barnstead solution temperature reached ∼2 °C, 0.8 mL of 3 M HNO3 was
Nanopure water purification system (18.2 MΩ·cm resistivity). Oxygen rapidly added to the cold IrOx deposition solution.25 The resulting
(O2, Alphagaz-1 grade 99.999%) and argon (Ar, Alphagaz-1 grade dark-blue solution was stirred for 80 min in an ice bath. The cold
99.999%) were purchased from Air Liquide. Nitrogen (N2) was bleed- solution was used for the deposition according to the conditions
off gas from a liquid nitrogen source. reported in Table 1, and the deposition solution was stored in a
Analytical Equipment. All activity, stability, and surface area refrigerator at 5 °C between depositions for up to one week.
measurements were conducted with a Bio-Logic SP200 potentiostat/ Electrochemical Characterization. All activity, stability, and
galvanostat with a built-in electrochemical impedance spectroscopy surface area measurements were conducted in a modified two-chamber
(EIS) analyzer or a Bio-Logic VMP3 multichannel potentiostat/ U-cell in which the first chamber held the working and reference
galvanostat with a built-in EIS analyzer. The working electrodes were 5 electrodes in ∼120 mL of solution, and the second chamber held the
mm diameter disk electrodes mounted in a Pine Instrument Company auxiliary electrode in ∼25 mL of solution. The two chambers were
E6-series ChangeDisk rotating disk electrode assembly in an MSR separated by a fine-porosity glass frit. The cell was purged for ∼20 min
rotator. In the case of rotating ring-disk electrode voltammetry, the 5 with O2 prior to each set of experiments. During static voltammetry
mm disk working electrode was instead mounted in a Pine Instrument measurements, the solution in the first chamber was blanketed under
Company E6-series Pt ChangeDisk rotating ring-disk electrode O2. During rotating disk electrode voltammetry (RDEV) measure-
assembly. The auxiliary electrodes were carbon rods (99.999%, Alfa ments, the solution in the first chamber was continuously bubbled with
Aesar), and the reference electrode was a commercial saturated O2. The uncompensated resistance of the cell was measured with a
calomel electrode (SCE) (CH-Instruments) that was externally single-point high-frequency impedance measurement, and IR drop was
referenced to a solution of ferrocene monocarboxylic acid (Sigma- compensated at 85% through positive feedback using the Bio-Logic
Aldrich) in a 0.2 M phosphate buffer at pH 7 (0.284 V vs SCE).32 Data EC-Lab software. Our typical electrochemical cell had Ru = ∼10 Ω in 1
were recorded using the Bio-Logic EC-Lab and EC-Lab Express M H2SO4 and Ru = ∼20 Ω in 1 M NaOH.
software packages. Electrochemical capacitance measurements were determined using
X-ray photoelectron spectroscopy (XPS) analysis was conducted cyclic voltammetry (CV)33 and EIS.34−36 First, the potential range
using a Surface Science Instruments M-probe spectrometer with a where there is a non-Faradaic current response was determined from
monochromatic 1486.6 eV Al Kα X-ray line source directed 35° with CV. This range is typically a 0.1 V potential window centered on the
respect to the sample surface. The spectrometer was controlled by open-circuit potential (OCP) of the system. CV measurements were
ESCA25 Capture software (version 5.01.04, Service Physics). Spectra conducted in static solution by sweeping the potential across the
were collected with a hemispherical electron analyzer mounted at an nonfaradaic region from the more positive to negative potential and
angle of 35° with respect to the sample surface. The sample chamber back at 8 different scan rates: 0.05, 0.01, 0.025, 0.05, 0.1, 0.2, 0.4, and
was maintained at <5 × 10−9 Torr. Low-resolution survey scans were 0.8 V s−1. The working electrode was held at each potential vertex for
acquired with a 800 μm spot size between the binding energies of 1− 10 s before beginning the next sweep.37 EIS measurements were
1050 eV. Higher-resolution scans with a resolution of ∼0.8 eV were conducted in static solution at three points: the two vertices and the
collected between 270 and 340 eV. Analysis of the spectra was done midpoint potential of the voltammetry measurements. The amplitude
using the CasaXPS Version 2.3.15 software package. of the sinusoidal wave was 10 mV, and the frequency scan range was
Electrode Preparation. Catalysts were deposited onto 5 mm from 100 kHz to 100 Hz. EIS spectra were analyzed with the Bio-Logic
diameter, 4 mm thick Sigradur G glassy carbon (GC) disks (HTW EC-Lab software package.
Hochtemperatur-Werkstoff GmbH). Prior to deposition, the GC disks The Faradaic efficiency of O2 production by each catalyst was
were first polished with 600 grit Carbimet SiC grinding paper measured using a RRDE apparatus. The collection efficiency, N, of the
(Buehler) on a Struers LaboPol-5 polishing wheel at 200 rpm for 2 rotating ring-disk electrode assembly was independently determined to
min, then sonicated for 5 min each in pure water, acetone, be N = 19 ± 1% from reducing K3Fe(CN)6 at a GC electrode and
isopropanol, and again in pure water. reoxidizing it at the Pt ring. RRDE measurements were conducted in a
In addition to GC, a 5 mm diameter Pt disk electrode (Pine cell purged for ∼20 min with N2 and then blanketed with N2 during
Instrument Company) was prepared for surface area studies, and 5 the experiment. Since the freshly prepared electrodeposited catalysts
mm diameter Ni disk electrodes (cut from Ni rod, Alfa Aesar, are generally derived from cathodic depositions, prior to each RRDE
99.995%) were prepared as electrocatalyst surfaces and as additional experiment the electrodeposited catalyst was first pre-anodized at 10
substrates for NiCeOx electrodeposition. The disks were sequentially mA cm−2 for 2 min in 1 M NaOH at 1600 rpm. Before any ring
polished with 9, 6, 3, 1, and 0.1 μm MetaDi Supreme diamond slurries currents were collected, the Pt ring was polished by hand with 1 μm
(Buehler) with an MD-Floc synthetic nap polishing pad (Struers) on a MetaDi Supreme diamond slurry (Buehler) first on a Nylon polishing
Struers LaboPol-5 polishing wheel at 150 rpm for several seconds, cloth (Buehler) and then on a Microcloth polishing pad (Buehler),
then sonicated for 5 min each in pure water, acetone, isopropanol, and followed by rinsing and ∼1 min sonication in pure water. The disk
again in pure water. Sputtered Ni surfaces were prepared as an electrode was then held at open circuit for 2 min, and the ring
electrode surface from a Ni target (99.95%, ACI Alloys). The catalyst electrode was held at −0.7 V vs SCE in N2-saturated 1 M NaOH. This
layer was sputtered onto a polished GC disk from an RF source at 150 was to establish the background current at the ring electrode. The
W at room temperature under a constant flow of 20 sccm Ar while magnitude of the background currents was typically |iring| < 3 μA. The
maintaining an overall pressure of 8.5 mtorr for 35 min. disk electrode was then subjected to sequential 1 min current steps at
The deposition conditions for each catalyst were based on literature 0.5, 1, 2 , 5, and 10 mA cm−2 at 1600 rpm in 1 M NaOH, while the
preparations and are outlined in Table 1. Each electrodeposition was ring was held constant at −0.7 V vs SCE.

■
conducted in a 100 mL cell with 40 mL of deposition solution, and the
auxiliary electrode was separated from the working and reference RESULTS AND DISCUSSION
electrodes in a separate auxiliary chamber by a fine-porosity glass frit
(Bioanalytical Systems Inc.). pH measurements of electrodeposition The general procedure used to evaluate each catalytic material
solutions were conducted with a VWR Symphony multiparameter is highlighted in Figure 1. The techniques used to determine
16979 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987
Journal of the American Chemical Society Article

the electrochemically active surface area, electrocatalytic

activity, stability, and Faradaic efficiency for each OER catalyst
investigated are described in detail below. An important note is
that the goal of this benchmarking study is to propose a useful,
moderate through-put method for screening a variety of OER
catalysts. Each catalyst studied was chosen as representative of
broad classes of catalysts. Catalysts were synthesized by
electrodeposition, following literature procedures with the
possible exception of the electrode support, as outlined in the
Experimental section. In this study, GC was the substrate used
for all catalysts. We note that the benchmarking protocol
reported here was designed specifically for testing OER
electrocatalysts under conditions relevant to an integrated
solar water-splitting device under 1 sun illumination.
Surface Analysis. The qualitative elemental composition of
representative samples for each catalyst was determined using
X-ray photoelectron spectroscopy (XPS) (Figures S1−S11).
Survey scans from 1 to 1050 eV binding energy were used to
confirm the qualitative composition of each sample and to
ensure that only the elements expected in each catalyst sample
were present. High-resolution scans from 270 to 340 eV
binding energy were used to test for photoemission originating
from Pt4d, Ir4d, and Ru3d. Other than in the case of IrOx, no
peaks associated with noble metals appeared in the high-
resolution scans. We thus suspect no noble metal presence in
the electrodeposited catalysts within the ∼0.1 atom % detection
limit of XPS.38,39
Electrochemically Active Surface Area. The electro-
chemically active surface area (ECSA) for each system was Figure 2. Double-layer capacitance measurements for determining
estimated from the electrochemical double-layer capacitance of electrochemically active surface area for an electrodeposited NiOx
the catalytic surface.33 The electrochemical capacitance was catalyst from voltammetry in 1 M NaOH. (a) Cyclic voltammograms
determined using two different methods: (1) by measuring the were measured in a non-Faradaic region of the voltammogram at the
non-Faradaic capacitive current associated with double-layer following scan rate: (purple line) 0.005, (orange line) 0.01, (dark red
charging from the scan-rate dependence of cyclic voltammo- line) 0.025, (cyan line) 0.05, (brown line) 0.1, (blue line) 0.2, (green
grams (CVs)33,37 and (2) from measuring the frequency- line) 0.4, and (red line) 0.8 V/s. The working electrode was held at
dependent impedance of the system using electrochemical each potential vertex for 10 s before the beginning the next sweep. All
impedance spectroscopy (EIS).34−36 current is assumed to be due to capacitive charging. (b) The cathodic
(red open circle) and anodic (blue open square) charging currents
To measure double-layer charging via CV, a potential range measured at −0.05 V vs SCE plotted as a function of scan rate. The
in which no apparent Faradaic processes occur was determined determined double-layer capacitance of the system is taken as the
from static CV. This range is typically a 0.1 V potential window average of the absolute value of the slope of the linear fits to the data.
centered at the open-circuit potential (OCP) of the system. All
measured current in this non-Faradaic potential region is
assumed to be due to double-layer charging. The charging for an electrodeposited NiOx catalyst is shown in Figure 3. In
current, ic, is then measured from CVs at multiple scan rates. the limit of high frequency and under non-Faradaic conditions,
Examples of CVs of the electrodeposited NiOx catalyst in a the electrochemical system is approximated by the modified
non-Faradaic potential range is shown in Figure 2a. The Randles circuit shown in the inset of Figure 3, where Rs is the
working electrode was held at each potential vertex for 10 s solution resistance, CPE is a constant-phase element related to
before beginning the next sweep. The double-layer charging the double-layer capacitance, and Rct is the charge-transfer
current is equal to the product of the scan rate, v, and the resistance from any residual Faradaic processes.42
electrochemical double-layer capacitance, CDL, as given by eq The frequency-dependent impedance of the CPE is given by
1.33,37,40,41 eq 2:34,35
1
ic = vC DL (1) ZCPE =
Q 0(iω)1 − α (2)
Thus, a plot of ic as a function of v yields a straight line with a
slope equal to CDL (Figure 2b). The electrochemical double- where ω is the frequency of the sinusoidal applied potential, i =
layer capacitance measured from the scan-rate dependent CVs (−1)1/2, Q0 is a constant with units of F sa−1, and 1 ≥ a ≥ 0 is
for the electrodeposited NiOx catalyst is CDL = 0.091 mF. related to the phase angle of the frequency response. For the
The double-layer capacitance was also measured using EIS in circuit model used here, it has been suggested that Q0 is related
the same non-Faradaic region. A sinusoidal potential is applied to CDL according to eq 3.34,35
to the system and the frequency-dependent complex ⎡ ⎛ ⎞(a − 1)⎤
1/ a

= ⎢Q 0 ⎜ ⎥
impedance is measured. An example Nyquist plot of the real 1 1
C DL + ⎟
and imaginary components of the electrochemical impedance in ⎢ ⎝ Rs R ⎠ ⎥
a non-Faradaic region measured between 100 Hz and 100 kHz ⎣ ct ⎦ (3)

16980 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987

Journal of the American Chemical Society Article

While ideally one would synthesize smooth, planar surfaces of

each catalyst to measure Cs and estimate ECSA, this is not
practical for most electrodeposited systems. However, specific
capacitances have been measured for a variety of metal
electrodes in acidic and alkaline solutions and typical values
reported range between Cs = 0.015−0.110 mF cm−2 in
H2SO444−53 and Cs = 0.022−0.130 mF cm−2 in NaOH and
KOH solutions.47,50,52,54−63 For our estimates of surface area,
we use general specific capacitances of Cs = 0.035 mF cm−2 in 1
M H2SO4 and Cs = 0.040 mF cm−2 in 1 M NaOH based on
typical reported values.64 The roughness factor (RF) is
calculated by taking the estimated ECSA and dividing by the
geometric area of the electrode, 0.195 cm2. Average ECSA and
RF values for each catalyst investigated in this study along with
Figure 3. Representative Nyquist plots for an electrodeposited NiOx standard deviations from at least three independent measure-
catalyst in 1 M NaOH at −0.1 V (black open square), −0.05 V(red ments are shown in Table 2. Note that the standard deviations
open circle), and 0 V (blue open triangle) vs SCE measured from EIS are an indication of the precision of the measurements and not
in the frequency range 100 kHz to 100 Hz. These potentials fall in a necessarily the accuracy of the calculated ECSA and RF.
potential region in which no Faradaic processes are observed. The Because we use a standard value for specific capacitance in
solid lines are the fits to the data using the simplified Randles circuit our calculations, the error in the accuracy of estimates of ECSA
shown in the inset.
could be as large as a factor of 7 based on the range of reported
Cs for metal electrodes in acidic and alkaline electrolyte
Note that when a = 1, the constant phase element behaves as solutions.33,41 Moreover, in estimating the ECSA from CDL, we
a pure capacitor and CDL = Q0, and when a = 0, the constant do not take into account other possible contributions to the
phase element behaves as a pure resistor and CDL is undefined. measured capacitance including pseudocapacitance due to ion
From the EIS measurement of the electrodeposited NiOx adsorption and intercalation or chemical capacitance due to the
system at E = −0.05 V vs SCE shown in Figure 3, Rs = 15.9 population of electron trap states. The double-layer capacitance
Ω, Rct = 11.2 kΩ, Q0 = 0.151 mF sa−1, and a = 0.815. The measurements assume that the metal oxide catalysts are equally
calculated CDL = 0.079 mF. Note that the double-layer conductive, which is another potential source of error in these
capacitance measured by EIS is within 15% of that measured measurements. In general, we believe the ECSA estimates to be
from the scan rate-dependent CVs. In general, we have found accurate within about an order of magnitude, but caution that
that the CDL measured for a given sample by the two methods the values should be considered only as an approximate guide
tend to agree within ±15%. for comparing surface roughness.
The ECSA of a catalyst sample is calculated from the double- As a validation of the proposed benchmarking methods for
layer capacitance according to eq 4: determining surface area, we have compared the electrochemi-
C DL cally active surface area of a Pt disk electrode determined using
ECSA = our protocols to that determined from estimating the amount
Cs (4)
of adsorbed hydrogen in the H-UPD region of the voltammo-
where Cs is the specific capacitance of the sample or the gram in 1 M H2SO4 under 1 atm N2 (Figures S12−S14), a well-
capacitance of an atomically smooth planar surface of the established method for Pt.33,61,65 The surface area was
material per unit area under identical electrolyte conditions. calculated from the area of the electrochemical peak associated

Table 2. Benchmarking Parameters for Each Catalyst Investigated in 1 M NaOH

catalyst ECSA/cm2 RF ηt=0/V ηt=2h/V jg,η=0.35 V/mA cm−2 js,η=0.35 V/mA cm−2 ε reported ηt=0/V
CoOx-(a) 1.9 ± 0.6 10 ± 3 0.39 ± 0.04 0.42 ± 0.04 0.9 ± 0.3 0.09 ± 0.04 0.94 ± 0.08 0.38a
0.43b
CoOx-(b) (“CoPi”) 5.1 ± 1.7 26 ± 9 0.42 ± 0.02 0.44 ± 0.02 0.4 ± 0.1 0.015 ± 0.006 1.00 ± 0.10
0.41c
CoFeOx 3.6 ± 0.9 19 ± 5 0.37 ± 0.02 0.43 ± 0.03 7±3 0.4 ± 0.2 0.98 ± 0.08 0.34a
0.32d
IrOx 21 ± 10 105 ± 53 0.32 ± 0.04 1.05 ± 0.20 42 ± 13 0.4 ± 0.2 0.92 ± 0.07
0.30e
NiOx 1.8 ± 0.5 9±3 0.42 ± 0.02 0.41 ± 0.05 1.1 ± 0.4 0.12 ± 0.05 0.93 ± 0.08 0.42f
NiCeOx 3.3 ± 1.2 17 ± 6 0.43 ± 0.03 0.44 ± 0.02 1.6 ± 0.7 0.11 ± 0.07 0.90 ± 0.08 0.28f
NiCoOx 5.2 ± 2.1 27 ± 11 0.38 ± 0.03 0.36 ± 0.01 6±3 0.2 ± 0.1 0.92 ± 0.08 0.40g
NiCuOx 1.8 ± 0.3 9±2 0.41 ± 0.01 0.44 ± 0.03 1.4 ± 0.6 0.15 ± 0.07 0.95 ± 0.13 0.42h
NiFeOx 1.1 ± 0.5 6±3 0.35 ± 0.01 0.38 ± 0.02 15 ± 6 3±2 1.00 ± 0.04 0.29a
NiLaOx 1.2 ± 0.3 6±1 0.41 ± 0.03 0.46 ± 0.05 2.5 ± 0.9 0.4 ± 0.2 0.96 ± 0.09 0.36f
GC background 1.9 ± 0.9 10 ± 5 1.19 ± 0.03 1.29 ± 0.02 <0.05 <0.005 0.76 ± 0.07 −

a
Electrodeposited onto Pt and reported at 100 mA cm−2 in air saturated 1 M KOH.21 bElectrodeposited onto FTO glass and reported at 10 mA
cm−2 in 1 M NaOH.43 cElectrodeposited onto FTO glass and reported at 1 mA cm−2 in 0.1 M KH2PO4/K2HPO4 buffer at pH 7.22
d
Electrodeposited onto a GC disk and reported at 10 mA cm−2 in 0.1 M NaOH.24,25 eElectrodeposited onto a GC disk and reported at 10 mA cm−2
in 0.098 M H2SO4.25 fElectrodeposited onto Ni substrate and reported at 16 mA cm−2 in 1 M KOH.26 gElectrodeposited onto Cu substrate and
reported at 10 mA cm−2 in 5 M KOH.27 hElectrodeposited onto Ni substrate and reported at 100 mA cm−2 in 1 M NaOH at 80 °C.28

16981 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987

Journal of the American Chemical Society Article

with the stripping of the adsorbed hydrogen from the Pt surface conditions, all earth-abundant catalysts investigated in this
and the resulting ECSA = 1.9 cm2. The surface area of the same study show similar activity and achieve 10 mA cm−2 per
Pt electrode in the same solution calculated from CV and EIS geometric area current densities within a 0.07 V window
using a specific capacitance of 0.035 mF cm−2 was ECSA = 2.1 between η = ∼0.360−0.430 V. For comparison, IrOx achieves
cm2. The three values agree within ±10%. 10 mA cm−2 current densities for OER at overpotentials of η =
Electrocatalytic Activity. The electrochemical properties 0.32 ± 0.04 V in 1 M NaOH and η = 0.27 ± 0.3 V in 1 M
of each catalyst were investigated using a three-electrode H2SO4. These activities for IrOx in acid and base are similar to
electrochemical cell in a rotating disk electrode configuration. those previously reported for this catalyst.24,25
The OER activity was evaluated primarily by rotating disk The previously reported overpotentials required to achieve
electrode voltammetry (RDEV) at 0.01 V s−1 scan rate and close to 10 mA cm−2 for the various catalysts investigated in
1600 rpm rotation rate. This scan rate is slow enough to ensure this study are also reported in Table 2. In general, the
steady-state behavior at the electrode surface, and the rotation overpotentials reported in this study compare well (within
rate is sufficiently fast to aid in product removal and limit ∼15%) to data already reported in the literature for analogous
bubble formation from evolved O2 at the electrode surface. systems. One notable exception is the electrodeposited NiCeOx
An example voltammogram of NiOx in 1 M NaOH is shown catalyst. This was previously reported as an electrodeposited
in Figure 4. In addition, each catalyst was investigated by a catalyst on a nickel electrode with an overpotential of 0.28 V
required to achieve 16 mA cm−2 for OER.26 This overpotential
is significantly lower than the measured overpotential of 0.43 V
reported in this study to achieve 10 mA cm−2 for a NiCeOx
catalyst deposited onto a GC electrode. This discrepancy
suggests that the substrate may play a role in the activity of the
NiCeOx catalyst. For comparison, NiCeOx was deposited onto
a Ni electrode, and the overpotential required to achieve 10 mA
cm−2 current density was measured to be ηt=0 h = 0.30 V, which
is consistent with the previously reported activity of NiCeOx on
Ni substrates.26 However, the operating overpotential required
to achieve 10 mA cm−2 current density changes to η = 0.35 V
over the course of several minutes and then is stable at η = 0.35
for at least 2 h (Figure S28).
Because the focus of this study is to establish a benchmarking
Figure 4. A representative rotating disk voltammogram of the oxygen protocol, the activity of only a representative sampling of
evolution reaction at an electrodeposited NiOx catalyst at 0.01 V/s electrodeposited catalysts was thoroughly investigated. How-
scan rate and 1600 rpm in O2-saturated 1 M NaOH. The results of 30
ever, the methodology presented here should be applicable to
s chronopotentiometric steps (red open circle) and chronoampero-
metric steps (blue open square) are shown for comparison, and the any heterogeneous electrocatalytic system. As an illustration of
close overlay of the data suggests good approximation of steady-state this, we also investigated the activity of a sputtered Ni film on
conditions. The horizontal dashed line at 10 mA cm−2 per geometric GC and a solid Ni disk electrode in 1 M NaOH. These metallic
area is the current density expected for a 10% efficient solar water- Ni surfaces are oxidized to a NiOx material at potentials
splitting device.9,19,20 The inset is a representative 2 h controlled required for OER.70,71 Both surfaces show roughly equivalent
current electrolysis at 10 mA cm−2 per geometric area for the same activity compared to the electrodeposited NiOx system (Figures
electrodeposited NiOx catalyst. S26−S27).
Another activity metric sometimes reported in the electro-
series of controlled-current chronopotentiometric steps and catalysis literature is the specific activity at a given over-
controlled-potential chronoamperometric steps (Figure S15). potential. The definition of specific activity can vary from study
In such experiments, the current or potential is held constant to studyit may refer to either the specific current density per
for 30 s, and the resulting potential−time or current−time catalyst surface area (js)15,72 or the catalytic turnover frequency
profile should decay to a steady-state value at times >2 s.66,67 (TOF). In turn, the TOF may refer to the rate of electron
Representative steady-state potentials determined from current delivery per surface metal atom per second15 or the rate of
step measurements and steady-state currents determined from product molecules evolved per surface metal atom,73,74 per total
potential step measurements for the electrodeposited NiOx metal atoms including subsurface metal,75,76 or per electro-
system are shown as squares and circles in Figure 4 and show chemically active surface site.24,37,77 In general, a catalyst’s
good agreement when overlaid with the RDV measurement. specific activity can be useful when attempting to compare the
The horizontal dashed line is at 10 mA cm−2 per geometric intrinsic activity of catalysts with different surface areas or
area, the current density expected for a 10% efficient solar loadings. However, due to the various definitions of specific
water-splitting device,9,19,20 and the overpotential required to activity used in the literature, it is important that one be
achieve this current density, ηt=0, is a convenient figure of merit transparent when determining and reporting this parameter.
for electrocatalytic activity.20,68,69 There would be a clear benefit to standardizing how such
Representative rotating disk voltammograms for each specific activities are reported.
material investigated are shown in Figures S16−S25. Average As an illustrative exercise, here we determine the specific
ηt=0 values for each catalyst were calculated from RDV and current density of each catalyst investigated. The method
current step measurements at 10 mA cm−2 per geometric area, employed was chosen because it uses easily measured
and the resulting values in 1 M NaOH are reported along with parameters and requires little knowledge of the catalyst surface
standard deviations measured with at least three independently structure or active site density. The specific current density, js, is
prepared surfaces in Table 2. Note that, under alkaline calculated by dividing the current density per geometric area at
16982 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987
Journal of the American Chemical Society Article

a given overpotential, jg, by the roughness factor of the surface solution prepared with the much higher concentration of 16.1
as shown in eq 5. mM K2IrCl6. As expected, this catalyst shows a less drastic
change in overpotential, achieving 10 mA cm−2 with η = ∼0.41
jg V after 2 h constant polarization (Figure S29), suggesting that
js = activity loss may be due to loss of material.
RF (5)
Controlled current electrolyses for each catalyst in 1 M
The average current density per geometric area at η = 0.35 V, H2SO4 were also investigated. The IrOx system shows a stable
jg,η=0.35 V, and the corresponding specific activity, js,η=0.35 V, are operating overpotential at 10 mA cm−2, changing negligibly
shown in Table 2 for each catalyst investigated. The choice of η from η = 0.27 ± 0.03 V to 0.30 ± 0.02 over a 2 h time period.
= 0.35 V is based on previously reported device models that For every other catalyst investigated there is a dramatic increase
suggest a 10% efficient solar water-splitting device should in the operating potential within only minutes to η = 1.1 V. Not
operate at 10 mA cm−2 with a maximum of ∼0.45 V coincidently, ηt=2 h = 1.11 ± 0.02 V is the overpotential at which
overpotential for OER and HER combined.9,19,78 Assuming a the bare GC background reaches 10 mA cm−2 under these
0.1 V overpotential for HER leaves 0.35 V overpotential conditions, suggesting that the non-noble metal catalysts
available for OER catalysis. Standard deviations from at least investigated here are not stable in acidic solutions under
three experiments with identically prepared samples are oxidizing conditions and that the current density observed is
reported for each jg,η=0.35 V, and standard errors calculated likely arising from the GC substrate itself.
from the standard deviations in the RF and jg,η=0.35 V are The stability measurement protocol outlined above is useful
reported for each js,η=0.35 V. Note that in the case of js, the as a rapid, preliminary screen of catalyst durability. However, it
standard errors reported are measures of error in precision. Due is important to note that a catalyst that shows good 2 h stability
to the inaccuracies inherent in determining ECSA and RF, we may not show the same stability over longer periods of time.
caution that the js values reported should be considered only as The stability measurements outlined here should be augmented
an approximate guide for comparing specific activity and do not by other long-term stability tests for more in-depth studies of
supplant ηt=0 as the primary figure of merit for catalyst activity. promising catalysts. Moreover, additional experiments to study
Electrocatalytic Stability. The short-term stability of each changes in catalyst composition, the mass of the material, and
material under catalytic conditions was determined using catalyst surface area would be useful complements to the short-
controlled-current electrolysis. The catalyst material was held term stability measurements proposed here and should be
at a constant current density of 10 mA cm−2 per geometric area considered as additional studies for promising materials.
for 2 h at a constant 1600 rpm rotation rate, while the operating Faradaic Efficiency. A RRDE apparatus was used to
potential was measured as a function of time. A representative confirm the formation of O2 by each catalyst. RRDE allows for
controlled current electrolysis for NiOx in 1 M NaOH is shown the study of water oxidation at a central electrode disk and the
in the inset of Figure 4 (for other systems, see Figures S16− collection of the dissolved O2 produced at a surrounding Pt-
S25), and the average overpotential required to achieve 10 mA ring electrode. The disk electrode is subjected to sequential 1
cm−2 per geometric area after 2 h of constant electrolysis, ηt=2 h, min current steps from 0.1 to 10 mA cm−2 at a constant
is reported for each catalyst along with standard deviations rotation rate of 1600 rpm under 1 atm N2. The dissolved O2
measured with at least three independently prepared surfaces in generated at the disk electrode is then swept across the
Table 2. If ηt=0 and ηt=2 h are the same, then that is evidence that surrounding Pt ring electrode, which is held at a constant
the catalyst is stable under the operating conditions for at least potential E = −0.7 V vs SCE to rapidly reduce O2 to H2O2
a 2 h period. However, if ηt=2 h > ηt=0, then that is evidence of (Figure S30). Representative disk and ring currents for the
catalyst deactivation over time. Note that the stability electrodeposited NiOx catalyst system in 1 M NaOH are shown
measurement used in this study does not distinguish whether in Figure 5. The Faradaic efficiency was measured from the ring
the deactivation mechanism is due to corrosion, material current collected while the disk electrode was held at a constant
degradation, surface passivation, or other processes. 1 mA cm−2 current density per geometric area; this current
In 1 M NaOH, most catalysts investigated appear stable density is sufficiently large to ensure appreciable O2 production
under operating conditions as the overpotential shifts <0.03 V but sufficiently small to minimize local saturation and bubble
during the 2 h controlled potential electrolysis at 10 mA cm−2 formation at the disk electrode. The Faradaic efficiency of the
per geometric area. However, two catalysts that do show OER system, ε, is proportional to the ratio of the ring current
appreciable shifts in operating overpotential during the course to the disk current and is given by eq 6:
of the stability measurement in 1 M NaOH are CoFeOx, and 2i
IrOx. In the case of CoFeOx, slow catalyst dissolution has been ε= r
idN (6)
previously reported for spinel-type FexCo3−xO4 (0 < x < 1) and
related catalysts,79,80 and this may account for the observed loss where ir is the measured ring current, id = 1.95 mA is constant
of catalytic activity over the course of 2 h. disk current for a 0.195 cm2 disk electrode, and N = 0.19 is the
For IrOx, the overpotential required to achieve 10 mA cm−2 collection efficiency for the RRDE.
increases from η = ∼0.32 V to η = ∼1.05 V during 2 h of The mean Faradaic efficiency measured for three independ-
constant polarization. The general instability of IrOx catalysts in ently prepared samples for each catalyst is shown in Table 2
alkaline solutions under oxidizing condition has been previously with standard deviations. In general, ε ≥ 0.9 for the catalysts
reported,81,82 and may be due to the oxidation of the surface investigated in 1 M NaOH. The low Faradaic efficiency for
IrOx to water-soluble IrO42− or other solvated Ir(VI) ions.43,83 oxygen production by GC may be due to oxidative degradation
If this were the case, one might expect an electrode with a of the carbon surface. Note that small errors in the ring current
higher IrOx loading to maintain a more constant overpotential and collection efficiency can lead to relatively large errors in ε,
over the 2 h experiment. As a simple test, a higher loading IrOx which may contribute to the standard deviations of ∼ ± 0.1 in
catalyst was deposited under standard conditions from a the reported measurements. RRDE measurements are useful
16983 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987
Journal of the American Chemical Society Article

Figure 5. RRDE measurements for the OER by an electrodeposited

NiOx in 1 M NaOH at 1600 rpm rotation rate under 1 atm N2. The
disk electrode was subjected to a series of current steps, and the
operating potential was measured. The dissolved oxygen generated at
the disk was then reduced by 2 electrons at the surrounding Pt ring
electrode. The inset shows a plot of the ratio of the ring and disk
current normalized for collection efficiency and the number of
electrons in the ring and disk reactions as a function of the disk
current. Note that at higher disk currents, the measured current ratio
deviates from the expected value of 1. This is likely due to local O2-
saturation and bubble formation at the disk electrode at higher current
densities; only dissolved O2 can be collected at the ring electrode.
Faradaic efficiency measurements were calculated at id = 0.2 mA, which
is jd = 1 mA cm−2 per geometric area. Figure 6. Comprehensive plots of catalytic activity, stability, and
electrochemically active surface area for OER electrocatalysts in acidic
(top) and alkaline (bottom) solutions. The x-axis is the overpotential
for quickly screening the approximate Faradaic efficiency of required to achieve 10 mA cm−2 per geometric area at time t = 0. The
large numbers of catalysts but should be augmented by more y-axis is the overpotential required to achieve 10 mA cm−2 per
selective and sensitive techniques for more in-depth studies of geometric area at time t = 2 h. The diagonal dashed line is the
promising single catalysts. expected response for a stable catalyst. The color of each point
Comparing Catalytic Performance. A key challenge in represents the roughness factor of the catalyst with a bin size of 1
investigating several parameters related to the performance of order of magnitude with light green representing RF = 1, and dark red
catalytic materials for water oxidation is determining a graphical representing RF > 103. The size of each point is inversely proportional
method of quickly and efficiently disseminating relevant to the standard deviation in the ECSA measurement reported in Table
parameters. While data tables are certainly capable of conveying 2. The region of interest for benchmarking is the unshaded white
large amounts of information, they do not necessarily facilitate region of the plot where the overpotential required to achieve 10 mA
cm−2 per geometric area at time t = 0 and 2 h is <0.5 V. There is a
cross-comparison of large numbers of materials. As a useful break and change in scale in both axes at overpotentials >0.5 V, and
alternative, we present the graphical representation for OER in the corresponding region of the plot is shown in gray. Catalysts whose
1 M NaOH and 1 M H2SO4 shown in Figure 6. The x-axis activity and stability measurements fall inside this gray area are outside
represents the overpotential required to achieve 10 mA cm−2 the region of interest for benchmarking, but their activity and stability
per geometric area at time t = 0, a measure of catalytic activity. measurements are included for completeness.
The y-axis represents the overpotential required to achieve 10
mA cm−2 per geometric area at time t = 2 h, a measure of
catalyst stability. The diagonal dashed line is the expected shows appreciable stability under oxidizing conditions in acidic
response for a stable catalyst; materials whose points on this solutioninstead, each material degrades to what appears to be
plot deviate from that line show catalyst deactivation (or a GC background. Although only a comparatively small subset
activation) over time. The color of the each point is related to of OER catalysts were investigated, this result nevertheless
the roughness factor of the catalyst with a bin size of 1 order of suggests that significant research targeted toward the discovery
magnitude with light green representing RF = 1 and dark red of acid-stable non-noble metal OER catalysts must be pursued
representing RF > 103. The size of each point is inversely if any solar-fuels production or other water electrolysis device is
proportional to the standard deviation in the ECSA measure- to operate under acidic conditions. Another important
ment. Since the ideal catalyst will have a low overpotential, be observation is that, under basic conditions, nearly all earth-
stable over time, and have high specific activity (or low surface abundant catalysts investigated in this study show similar
area), the best catalyst materials for OER will be positioned activity achieving 10 mA cm−2 per geometric area current
toward the bottom left corner of the comprehensive plots and densities within a 0.07 V window between η = ∼0.360 − 0.430
should appear light green in color. Note that there is a break in V. A zoomed-in region of the graphical representation about
each axis at 0.5 V, and there is a scale change following this this potential window in 1 M NaOH is shown in Figure 7.
break for values >0.5 V. Catalysts that operate at 10 mA cm−2 Among the most promising of the earth-abundant catalysts
with overpotentials >0.5 V lie outside the area of interest, but investigated in 1 M NaOH include NiFeOx and NiCoOx, which
their activity and stability measurements are included in this achieve 10 mA cm−2 per geometric area current densities at ηt=0
gray shaded region for completeness. = 0.35 and 0.38 V, respectively, and show good catalyst stability
One key observation prominently illustrated in Figure 6 is over 2 h. Of the two, NiFeOx stands out due to its lower surface
that no non-noble metal catalyst investigated in this study area and hence higher specific activity; at η = 0.35 V, NiFeOx
16984 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987
Journal of the American Chemical Society Article

erations should be taken when applying this catalyst

benchmarking methodology to other systems.

■ CONCLUSIONS
A benchmarking protocol for evaluating the activity, stability,
electrochemically active surface area, and Faradaic efficiency of
heterogeneous OER catalysts under conditions relevant to an
integrated solar water-splitting device has been presented. The
protocol was used to compare the electrocatalytic performance
of 10 heterogeneous OER catalysts deposited onto GC
substrates. In addition, a graphical representation of relevant
electrocatalytic parameters was developed in order to facilitate
Figure 7. Zoomed-in region of interest for alkaline OER from Figure the comparison of the electrocatalytic performance for the
6. The x-axis is the overpotential required to achieve 10 mA cm−2 per various OER catalysts.
geometric area at time t = 0. The y-axis is the overpotential required to By systematically comparing a range of OER catalyst
achieve 10 mA cm−2 per geometric area at time t = 2 h. The diagonal materials using identical methods and procedures, several
dashed line is the expected response for a stable catalyst. The color of important general observations can be made. First, every non-
each point represents the roughness factor of the catalyst with a bin noble metal catalyst investigated herein showed similar OER
size of 1 order of magnitude as shown in Figure 6. The size of each
activity in 1 M NaOH, achieving 10 mA cm−2 current density at
point is inversely proportional to the standard deviation in the ECSA
measurement reported in Table 2 overpotentials between 0.35 and 0.43 V. For comparison, an
electrodeposited IrOx catalyst under the same conditions
achieved 10 mA cm−2 current density at η = 0.32 ± 0.04 V,
has a specific current density that is ∼10 times higher than the although IrOx was unstable during 2 h of constant current
other non-noble metal catalyst investigated. This suggests electrolysis. This suggests that there is still significant room for
NiFeOx has the highest specific activity of the materials improvement in discovering OER catalysts that can operate at
investigated, although it is again important to note the high current density and lower overpotential in a stable manner.
uncertainty inherent in the ECSA and js measurements. In Second, only IrOx showed stability in 1 M H2SO4,
addition to NiFeOx and NiCoOx, CoFeOx also shows maintaining 10 mA cm−2 current density at η = ∼0.30 V for
promising activity achieving 10 mA cm−2 per geometric area 2 h of constant current electrolysis. Every non-noble metal
current density at a comparable overpotential of ηt=0 = 0.37 V, catalyst investigated was unstable under oxidative conditions in
but it lacks the stability of the NiFeOx and NiCoOx systems. 1 M H2SO4, ultimately exhibiting the same catalytic behavior as
The electrodeposited single metal NiOx, CoOx-(a), and the GC substrate. This result highlights the need for non-noble
CoOx-(b) evolve O2 at 10 mA cm−2 per geometric area current metal acid-stable OER catalysts in order for solar water-splitting
densities at only ∼0.05−0.07 V higher overpotential than the devices operating in 1 M H2SO4 to be feasible.
NiFeOx and NiCoOx systems and also show good catalyst As a secondary screen of OER activity, the specific activity of
stability over 2 h. Note that while CoOx-(b) shows the same each electrocatalyst was also determined. NiFeOx showed the
general activity as NiOx and CoOx-(a), its estimated specific highest specific activity, operating at ∼3 mA cm−2 per
activity is ∼6−9 times lower at η = 0.35 V. None of the non- electrochemically active surface area, nearly 10 times higher
noble metal catalysts investigated approach the activity of IrOx than the other catalysts. This suggests that NiFeOx has a higher
in 1 M NaOH, which achieves a 10 mA cm−2 per geometric intrinsic activity compared with the other systems investigated.
area current density at η = 0.32 V in 1 M NaOH, although the Specific activity measurements reported using this method
catalytic activity of IrOx decreases significantly over the course should be taken as an approximate guide rather than an
of 2 h of constant current electrolysis at 10 mA cm−2. absolute value and do not supplant the overpotential required
It is important to note that the benchmarking methodology to achieve 10 mA cm−2 per geometric area as a primary figure
reported here is proposed as a primary screen in evaluating of merit.
catalyst activity and stability under conditions relevant to an We note that the benchmarking protocol reported here is
integrated solar water-splitting device under 1 sun illumination. considered a first and important step in evaluating catalyst
Other parameters not studied as part of this protocol may be materials; further testing will be needed to truly establish
important in catalyst design for specific PEC systems. For catalyst feasibility. We also note that this benchmarking
instance, the thickness and absorptivity of the OER catalyst protocol was specifically designed for testing OER electro-
may have a profound effect on its performance when integrated catalysts under conditions relevant to an integrated solar water-
with a semiconductor as a photoanode in an integrated solar splitting device under 1 sun illumination. Other devices that use
water-splitting device, and therefore examining these traits may OER electrocatalysts such as PEM and alkaline water
be an important secondary screen.84 Moreover, an important electrolyzers or integrated water-splitting cells under multiple-
secondary screen for systems operating at higher current sun illumination may have significantly different operating
densities, such as PEC systems with concentrated solar, may parameters and as such will have different figures of merit and
may require different testing methods than those reported here.

■
include testing the electrocatalytic activity at higher current
densities. It is worth considering that standard electrochemical
methods for studying OER catalysts might be in a mass- ASSOCIATED CONTENT
transport limited regime when drawing ≥100 mA cm−2, and as *
S Supporting Information
such benchmarking catalysts at high current densities may be X-ray photoelectron spectra; discussion of choice of specific
influenced more by concentration overpotentials rather than by capacitance values; rotating disk voltammograms, chronopo-
the electrochemical kinetics of the material. Similar consid- tentiometric steps and chronoamperometric steps for electro-
16985 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987
Journal of the American Chemical Society Article

deposited catalyts; OER activity and stability measurements for (21) Merrill, M. D.; Dougherty, R. C. J. Phys. Chem. C 2008, 112,
NiCeOx on Ni electrodes; confirmation of 2 e− reduction of O2 3655−3666.
on Pt ring electrodes at 0.7 V vs SCE and 1600 rpm rotation (22) Kanan, M. W.; Nocera, D. G. Science 2008, 321, 1072−1075.
rate in 1 M NaOH; rotating disk voltammograms, chronopo- (23) Surendranath, Y.; Dincǎ, M.; Nocera, D. G. J. Am. Chem. Soc.
tentiometric steps and chronoamperometric steps for sputtered 2009, 131, 2615−2620.
(24) Nakagawa, T.; Beasley, C. A.; Murray, R. W. J. Phys. Chem. C
Ni and commercial Ni disks; discussion regarding the use of
2009, 113, 12958−12961.
Tafel plots for comparing electrocatalysts and benchmarking (25) Zhao, Y.; Hernandez-Pagan, E. A.; Vargas-Barbosa, N. M.;
intermediate pH solutions. This material is available free of Dysart, J. L.; Mallouk, T. E. J. Phys. Chem. Lett. 2011, 2, 402−406.
charge via the Internet at http://pubs.acs.org.

■
(26) Corrigan, D. A.; Bendert, R. M. J. Electrochem. Soc. 1989, 136,
723−728.
AUTHOR INFORMATION (27) Ho, J. C. K.; Piron, D. L. J. Appl. Electrochem. 1996, 26, 515−
Corresponding Author 521.
(28) Li, X.; Walsh, F. C.; Pletcher, D. Phys. Chem. Chem. Phys. 2011,
[email protected]
13, 1162−1167.
Notes (29) Noel, M.; Anantharaman, P. N. Surf. Coat. Technol. 1986, 28,
The authors declare no competing financial interest. 161−179.

■ ACKNOWLEDGMENTS
This material is based upon work performed by the Joint
(30) Bomgardner, M. M. Chem. Eng. News 2011, 89, 55−63.
(31) Note that the analytical procedures highlighted in this report
can be used in any pH condition, including neutral pH. One of the
primary challenges in benchmarking systems near pH 7 concerns the
Center for Artificial Photosynthesis, a DOE Energy Innovation choice of buffer and electrolyte. Because the conjugate bases of
Hub, supported through the Office of Science of the U.S. buffering systems tend to be relatively coordinating, they often
Department of Energy under award no. DE-SC0004993. We specifically adsorb to metals and metal oxides. This can have a
are grateful for the many useful insights we have received profound effect on the activity of a given OER catalyst. We believe that
regarding this work from various members of the electro- it would be difficult to define “universal” conditions at intermediate
chemistry community. In particular, we would like to pH and that instead catalysts would need to be studied under more
acknowledge useful discussions with Allen J. Bard, Fred C. than one buffered and perhaps even unbuffered electrolytes. Such a
Anson, Nathan S. Lewis, Carl A. Koval, Manuel P. Soriaga, and study is beyond the scope of the current manuscript. See the
Hans-Joachim Lewerenz. XPS data was collected at the Supporting Information for an expanded discussion.
Molecular Materials Research Center of the Beckman Institute (32) Liaudet, E.; Battaglini, F.; Calvo, E. J. J. Electroanal. Chem. 1990,
of the California Institute of Technology. 293, 55−68.

■
(33) Trasatti, S.; Petrii, O. A. Pure Appl. Chem. 1991, 63, 711−734.
REFERENCES (34) Orazem, M. E.; Tribollet, B. Electrochemical Impedance
Spectroscopy; John Wiley & Sons, Inc.: Hoboken, NJ, 2008, p 233−
(1) Grätzel, M. Acc. Chem. Res. 1981, 14, 376−384. 237.
(2) Koelle, U. New J. Chem. 1992, 16, 157−169. (35) Brug, G. J.; van den Eeden, A. L. G.; Sluyters-Rehbach, M.;
(3) Bard, A. J.; Fox, M. A. Acc. Chem. Res. 1995, 28, 141−145. Sluyters, J. H. J. Electroanal. Chem. 1984, 176, 275−295.
(4) Turner, J. A. Science 2004, 305, 972−974. (36) Huang, V. M.-W.; Vivier, V.; Orazem, M. E.; Pébère, N.;
(5) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, Tribollet, B. J. Electrochem. Soc. 2007, 154, C99−C107.
15729−15735. (37) Benck, J. D.; Chen, Z.; Kuritzky, L. Y.; Forman, A. J.; Jaramillo,
(6) Lewis, N. S. Science 2007, 315, 798−801. T. F. ACS Catalysis 2012, 2, 1916−1923.
(7) Crabtree, G. W.; Dresselhaus, M. S. MRS Bull. 2008, 33, 421− (38) Chusuei, C. C.; Goodman, D. W. In Encyclopedia of Physical
428.
Science and Technology; 3rd ed.; Robert, A. M., Ed.; Academic Press:
(8) Gray, H. B. Nat. Chem. 2009, 1, 7.
New York, 2003, p 921−938.
(9) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi,
(39) Klauber, C.; Smart, R. S. C. In Surface Analysis Methods in
Q.; Santori, E. A.; Lewis, N. S. Chem. Rev. 2010, 110, 6446−6473.
(10) Vesborg, P. C. K.; Jaramillo, T. F. RSC Advances 2012, 2, 7933− Materials Science; 2 ed.; O’Conner, D. J., Sexton, B. A., Smart, R. S. C.,
7947. Eds.; Springer-Verlag: Berlin, 2003, p 3−65.
(11) Eisenberg, R.; Gray, H. B. Inorg. Chem. 2008, 47, 1697−1699. (40) Bockris, J. O. M.; Srinivasan, S. J. Electroanal. Chem. 1966, 11,
(12) Lewis, N. S.; Crabtree, G.; Nozik, A. J.; Wasielewski, M. R.; 350−389.
Alivisatos, A. P. Basic Research Needs for Solar Energy Utilization; (41) Boggio, R.; Carugati, A.; Trasatti, S. J. Appl. Electrochem. 1987,
Department of Energy: Washington, DC, 2005. 17, 828−840.
(13) Gerken, J. B.; Chen, J. Y. C.; Massé, R. C.; Powell, A. B.; Stahl, (42) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
S. S. Angew. Chem., Int. Ed. 2012, 51, 6676−6680. Fundamentals and Applications; John Wiley & Sons, Inc.: Hoboken,
(14) Smith, R. D. L.; Prévot, M. S.; Fagan, R. D.; Trudel, S.; NJ, 2001, p 383−388.
Berlinguette, C. P. J. Am. Chem. Soc. 2013, 135, 11580−11586. (43) Minguzzi, A.; Fan, F.-R. F.; Vertova, A.; Rondinini, S.; Bard, A. J.
(15) Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Chem. Sci. 2012, 3, 217−229.
Catal., B 2005, 56, 9−35. (44) Frumkin, A. N. J. Res. Inst. Catal., Hokkaido Univ. 1967, 15, 61−
(16) Emery, K. A.; Osterwald, C. R. Sol. Cells 1986, 17, 253−274. 83.
(17) Shrotriya, V.; Li, G.; Yao, Y.; Moriarty, T.; Emery, K.; Yang, Y. (45) Pickering, H. W. J. Electrochem. Soc. 1968, 115, 690−694.
Adv. Funct. Mater. 2006, 16, 2016−2023. (46) Rosen, M.; Flinn, D. R.; Schuldiner, S. J. Electrochem. Soc. 1969,
(18) Matsen, D. A.; Bosco, A. D. In Handbook of Fuel Cells: 116, 1112−1116.
Fundamentals, Technology and Applications; Vielstich, W., Lamm, A., (47) Iseki, S.; Ohashi, K.; Nagaura, S. Electrochim. Acta 1972, 17,
Gasteiger, H., Eds.; Wiley: Chichester, U.K., 2003; Vol. 4, p 714−724. 2249−2265.
(19) Weber, M. F.; Dignam, M. J. J. Electrochem. Soc. 1984, 131, (48) Turner, M.; Thompson, G. E.; Brook, P. A. Corros. Sci. 1973, 13,
1258−1265. 985−991.
(20) Gorlin, Y.; Jaramillo, T. F. J. Am. Chem. Soc. 2010, 132, 13612− (49) Glarum, S. H.; Marshall, J. H. J. Electrochem. Soc. 1979, 126,
13614. 424−430.

16986 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987

Journal of the American Chemical Society Article

(50) Badawy, W. A.; Gad-Allah, A. G.; Abd El-Rahman, H. A.; (82) Morimitsu, M.; Murakami, C.; Kawaguchi, K.; Otogawa, R.;
Abouromia, M. M. Surf. Coat. Technol. 1986, 27, 187−196. Matsunaga, M. J. New Mater. Electrochem. Syst. 2004, 7, 323−327.
(51) Reid, J. D.; David, A. P. J. Electrochem. Soc. 1987, 134, 1389− (83) Pourbaix, M. Atlas of Electrochemical Equilibria in Aqueous
1394. Solution; Pergamon Press: Oxford, 1966, p 373−377.
(52) Centeno, T. A.; Stoeckli, F. J. Power Sources 2006, 154, 314− (84) Trotochaud, L.; Mills, T. J.; Boettcher, S. W. J. Phys. Chem. Lett.
320. 2013, 4, 931−935.
(53) Lu, Y.; Xu, H.; Wang, J.; Kong, X. Electrochim. Acta 2009, 54,
3972−3978.
(54) Weininger, J. L.; Breiter, M. W. J. Electrochem. Soc. 1963, 110,
484−490.
(55) Weininger, J. L.; Breiter, M. W. J. Electrochem. Soc. 1964, 111,
707−712.
(56) O’Brien, R. N.; Seto, P. J. Electroanal. Chem. 1968, 18, 219−230.
(57) Hampson, N. A.; Latham, R. J.; Lee, J. B.; Macdonald, K. I. J.
Electroanal. Chem. 1971, 31, 57−62.
(58) Gagnon, E. G. J. Electrochem. Soc. 1973, 120, 1052−1056.
(59) Lasia, A.; Rami, A. J. Electroanal. Chem. 1990, 294, 123−141.
(60) Gu, P.; Bai, L.; Gao, L.; Brousseau, R.; Conway, B. E.
Electrochim. Acta 1992, 37, 2145−2154.
(61) Bai, L.; Gao, L.; Conway, B. E. J. Chem. Soc., Faraday Trans.
1993, 89, 235−242.
(62) Fournier, J.; Brossard, L.; Tilquin, J.-Y.; Cote, R.; Dodelet, J.-P.;
Guay, D.; Menard, H. J. Electrochem. Soc. 1996, 143, 919−926.
(63) Wu, G.; Li, N.; Zhou, D.-R.; Mitsuo, K.; Xu, B.-Q. J. Solid State
Chem. 2004, 177, 3682−3692.
(64) See Supporting Information.
(65) Conway, B. E.; Angerstein-Kozlowska, H. Acc. Chem. Res. 1981,
14, 49−56.
(66) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
Fundamentals and Applications, 2nd ed.; John Wiley & Sons, Inc.:
Hoboken, NJ, 2001, p 353−354.
(67) Prater, K. B.; Bard, A. J. J. Electrochem. Soc. 1970, 117, 207−213.
(68) Matsumoto, Y.; Sato, E. Mater. Chem. Phys. 1986, 14, 397−426.
(69) Another common metric by which to compare electrocatalysts is
their Tafel parameters, including exchange-current density and Tafel
slope. In this report, we chose not to use Tafel parameters as an
activity metric due to the complexity in estimating and understanding
the relevant parameters for multistep, multielectron transfer
mechanisms. While Tafel plots are very useful for deriving mechanistic
and kinetic information from specific electrocatalytic systems, we
believe that determing relevant Tafel parameters for a given system is
system specific and requires mechanistic analysis that is beyond the
scope of this manuscript. For further discussion, please see the
Supporting Information.
(70) Lu, P. W. T.; Srinivasan, S. J. Electrochem. Soc. 1978, 125, 1416−
1422.
(71) McBreen, J. In Modern Aspects of Electrochemistry; Bockris, J. O.
M., Conway, B. E., White, R. E., Eds.; Plenum: New York, 1990; Vol.
21, p 29−63.
(72) Suntivich, J.; May, K. J.; Gasteiger, H. A.; Goodenough, J. B.;
Shao-Horn, Y. Science 2011, 334, 1383−1385.
(73) Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F. Energy
Environ. Sci 2012, 5, 7050−7059.
(74) McKone, J. R.; Sadtler, B. F.; Werlang, C. A.; Lewis, N. S.; Gray,
H. B. ACS Catalysis 2012, 3, 166−169.
(75) Jiao, F.; Frei, H. Energy Environ. Sci. 2010, 3, 1018−1027.
(76) Surendranath, Y.; Kanan, M. W.; Nocera, D. G. J. Am. Chem. Soc.
2010, 132, 16501−16509.
(77) Yagi, M.; Tomita, E.; Kuwabara, T. J. Electroanal. Chem. 2005,
579, 83−88.
(78) Weber, M. F.; Dignam, M. J. Int. J. Hydrogen Energy 1986, 11,
225−232.
(79) Kishi, T.; Takahashi, S.; Nagai, T. Surf. Coat. Technol. 1986, 27,
351−357.
(80) Laouini, E.; Hamdani, M.; Pereira, M. I. S.; Douch, J.;
Mendonça, M. H.; Berghoute, Y.; Singh, R. N. Int. J. Hydrogen Energy
2008, 33, 4936−4944.
(81) Guerrini, E.; Chen, H.; Trasatti, S. J. Solid State Electrochem.
2007, 11, 939−945.

16987 dx.doi.org/10.1021/ja407115p | J. Am. Chem. Soc. 2013, 135, 16977−16987