of 18
Crystalline Nickel Manganese Antimonate as a Stable Water-Oxidation Catalyst
in
Aqueous 1.0 M
H
2
SO
4
Ivan
A. Moreno-Hernandez
1
, Clara A. MacFarland
1
, Carlos G. Read
1
, Kimberly
M.
Papadantonakis
1
, Bruce S. Brunschwig
2
, Nathan
S. Lewis
1,2,3
1
Division
of Chemistry and Chemical
Engineering,
127-72, California Institute of Technology,
Pasadena, CA
91125, USA
2
Beckman Institute Molecular Materials
Research
Center, California Institute of Technology,
Pasadena, CA
91125, USA
3
Kavli Nanoscience Institute, California Institute of Technology, Pasadena,
CA 91125, USA
*Correspondence
to: nslewis@caltech.edu
1
Electronic
Supplementary
Material
(ESI)
for
Energy
&
Environmental
Science.
This
journal
is
©
The
Royal
Society
of
Chemistry
2017
Supplementary Information
Materials and Methods
Chemicals
All chemicals
were used as received, including sulfuric acid (H
2
SO
4
, Fisher Scientific,
TraceMetal Grade,
93-98%), antimony(III) chloride
(SbCl
3
, Alfa Aesar, ACS, 99.0% min),
tin(IV) chloride hydrate
(SnCl
4
·xH
2
O, Alfa
Aesar,
98%), antimony standard
for ICP (Sigma
Aldrich, TraceCERT), multielement
standard
solution
1 for
ICP (Sigma Aldrich, TraceCERT),
potassium sulfate (K
2
SO
4
,
Macron Chemicals,
ACS),
and
gallium-indium eutectic
(Alfa
Aesar,
99.99%). A Millipore deionized water system was used to
obtain
water
with
resistivity of 18.2
cm.
Sample Preparation
A conductive film of antimony-doped tin oxide (ATO) was
deposited onto quartz slides
via a spray pyrolysis procedure.
1
Briefly,
the process consisted of heating
a quartz slide at 550
°C on
a hot plate,
and
using a
spray gun to spray a
0.24 M
ethanolic solution of SnCl
4
doped
with 3 mol% SbCl
3
onto the quartz slide. The thickness of the ATO film
was controlled by
changing the
duration of the spray.
Films with a
sheet
resistance of 13-20
Ω
sq
-1
were used for
this study.
Sputter depositions were
performed with an AJA Orion sputtering
system. To
make a
direct contact
to the
ATO
after film preparation,
prior to
depositing
metal films the ATO-coated
quartz
slides
were
partially covered
with Kapton
tape. Multi-metal
films were co-sputtered from
three metal
targets in an Ar plasma: Antimony (ACI Alloys,
99.95%), Nickel (ACI
Alloys,
99.95%), and Manganese (ACI Alloys, 99.95%). An
Ar
flow rate of 20 sccm was used to
sustain
2
the plasma, and the power applied to the three metal targets was varied
to
obtain
films that had
different compositions. The chamber pressure
was 5 mTorr during the deposition,
and the base
pressure of
the chamber was < 10
-7
Torr prior
to use. The sample was
not
heated intentionally
during the deposition
process. The stoichiometry reported in
the
sample
name was obtained by
dissolving the deposited
unannealed metal films in 1.0 M H
2
SO
4
(aq), followed by determination
of the concentration of dissolved ions using ICP-MS.
After metal film deposition, the films were annealed in a muffle furnace (Thermolyne
F48020-80) to form
the
oxides.
The temperature was
increased at a ramp rate of 10 °C min
-1
until the
temperature set point was reached. The temperature was then held for 6 h and allowed
to return to room temperature. The
temperature
set
point
was 700 °C unless otherwise specified.
Samples were
cleaved into
pieces, and In-Ga eutectic (Aldrich) was scribed on the ATO. Tinned
Cu wire was
threaded through a glass tube that
had been cleaned with
aqua
regia, which
consisted of
a 3:1 v/v solution of
concentrated
hydrochloric
acid
and nitric acid, respectively.
Ag paint (SPI, Inc.) was used
to bond the wire to the
portion
of ATO that had
been
covered with
In-Ga.
The
contact
was
allowed
to dry for
at least 2 h at room temperature or for 15 min at
85
°C in
an
oven. Hysol 9460 epoxy was used
to insulate
the
contact
and define the electrode area,
and the epoxy
was allowed
to cure for at
least 12 h at
room
temperature.
An optical scanner
(Epson perfection
V360)
was used to image
the electrode area and ImageJ was used to
quantify
the area.
Electrode
areas
were between 1 and
11 mm
2
unless otherwise
specified.
Materials Characterization
X-ray diffraction (XRD) analysis was
performed
with a Bruker
D8
Discover instrument
equipped with a
2-dimensional Vantec-500 detector. Copper
radiation (1.54 Å)
was
3
generated with
a tube voltage of 50 kV and a tube
current of 1000
μA.
The incident beam was
focused with a 0.5 mm
diameter mono-capillary collimator. An aligned laser
beam
was used to
ensure that the sample
was placed
at the correct depth for diffraction
measurements. Coupled
theta/two theta mode was
used,
with a
θ
angle that
was
half
of the
angle. The scattered x-ray
radiation was
collected
by the Vantec-500 detector with
an
angular resolution < 0.04°, which
enabled the collection
of
diffraction from a
range
of 20°. Four
scans were performed in
the
range of 25° to 85°
2θ,
and radiation
was counted for a total duration
of
4 h to obtain the XRD
profile. The collected data were analyzed using Bruker EVA
software. The peaks were indexed
to reference patterns
of
SnO
2
and CoSb
2
O
6
.
2, 3
The
preferred orientation of the ATO crystals was
not controlled during the ATO fabrication
process,
thus XRD data were not
used
for quantitative
analysis. Transmission-electron microscopy (TEM) samples of the films were prepared using a
focused Ga-ion beam (FIB)
on a
FEI Nova-600
Nanolab FIB/FESEM, with Pt and C protection
layers being applied before
being exposed to the FIB. High-resolution TEM
(HRTEM) and
scanning-transmission electron microscopy energy-dispersive X-ray spectroscopy
(STEM-EDS)
data were obtained
using a Tecnai Polara
(F30) TEM at an accelerating
voltage of 300 keV.
X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy (XPS)
data
were collected
using
a Kratos Axis
NOVA
(Kratos Analytical, Manchester, UK) at a background pressure of 10
-9
Torr. A monochromatic
Al
source
at 1486.6 eV was used
for excitation. Survey scans were collected at 1 eV
resolution, whereas 0.05 eV resolution was used for
high-resolution
scans.
The peak energies
were calibrated
against
the binding energy of the adventitious C 1s peak,
which
was set at 284.8
eV.
4
The XPS 2p
3/2
spectra of transition metal oxides are fit by multiple peaks, often 6, with
the intensity of the
individual peaks varying with the
compound.
Thus the energy for the
intensity maximum of
the
observed spectra can significantly shift
without a
change in
metal
oxidation state. For example both NiO and Ni(OH)
2
are formally Ni(II)
but the observed peak
maxima differ by more than 1.5eV.
4
The Ni2p3/2 spectra taken for Ni
0.5
Mn
0.5
Sb
1.7
O
y
before
electrochemistry (Figure 4A) are similar to XPS spectra of Ni(OH)
2
and dissimilar to
those of
γ-
NiOOH or
β-NiOOH
which have
a maxima at ~
856 eV.
5
Also
no sharp low energy peak
for Ni
metal was observed in Ni
0.5
Mn
0.5
Sb
1.7
O
y
(Figure 4A).
5
The peak
maximum
observed for
Ni(OH)
2
is
~855.3 eV compared to a maximum
observed here of 855.3 eV.
4
NiO has an
observed maximum
at 853.8 eV with a side
peak
at ~855.8,
4
two peaks of almost
equal height at
853.7 and 855.4
eV
are used to fit these
lines.
6
The higher energy
peak has an area of about 3
times that
of the
lower energy peak
but the narrow
width of the lower energy peak
is responsible
for the maximum
in the observed spectra.
Thus we assign the Ni in the Ni
0.5
Mn
0.5
Sb
1.7
O
y
compound to Ni(II).
The
Mn spectra taken
before electrochemical operation
qualitatively
looked
more like MnO then
the higher oxides (Mn
2
O
3
, MnOOH, or MnO
2
).
5
The peak maximum
observed here is 641.3 eV while that of MnO is ~ 641.5 eV. All
the other Mn oxides have peak
maxima close to 642 eV
and have
a sharp feature in the spectra at low energy. Thus we assign
the Mn as
Mn(II).
The XP spectrum of
Sb
3d
5/2
overlaps with that of the O
1s.
The 3d
3/2
XP
spectra of the
oxides
of Sb
shows
peak maxima
at 539.7, 540.3, and
540.6 eV
for Sb
2
O
3
, Sb
2
O
4
,
and Sb
2
O
5
, respectively.
7
The Ni
0.5
Mn
0.5
Sb
1.7
O
y
had a 3d
3/2
maxima
at 539.7 eV and
was
assigned as Sb(III).
After electrochemical operation
the Ni and Sb signals exhibited a
shift towards higher
binding energies
by 0.39 and 0.33 eV,
respectively.
5
The shift
of the Ni 2p
3/2
maximum suggests
5
that Ni near the surface is partially oxidized.
5
The
surfaces of transition metal
antimonates
usually consist of a
mixture of Sb
3+
and
Sb
5+
. The observed 0.33 eV
shift of the Sb
3d
3/2
peak
suggests that
the
surface Sb
was
oxidized, indicating
that the surface changed from
having
a
substantial Sb(III)
component to adopting more
Sb(V)
character
Electrochemical Testing
Sulfuric acid
(TraceMetal
grade,
Fisher Chemical) was used to make 1.0 M solutions,
unless otherwise indicated. Digital potentiostats
(SP-200,
MPG-2,
Bio-Logic Science
Instruments) were
used to acquire
electrochemical
data. Mercury/mercury
sulfate
electrodes
(MSEs) were
calibrated with a reversible hydrogen
electrode (RHE).
The RHE consisted of a
platinum disk
(CH Instruments)
submerged
in hydrogen-saturated
1.0 M
sulfuric acid, with
H
2
(g)
bubbled underneath
the Pt
disk.
The
potentials of the
MSEs were
between
0.676 V and 0.691 V
vs. RHE.
Glass electrochemical
cells
were
cleaned with
aqua
regia
prior to
use. The working,
reference, and counter electrodes
consisted of the sample, a
calibrated MSE,
and
a carbon rod
(Strem) separated from the main
compartment by a glass frit, respectively. Approximately 25
mL of 1.0 M
H
2
SO
4
(aq) was
used as the electrolyte,
unless
otherwise
specified. The solution
resistance was determined to be ~
10 Ohm from
impedance measurements on a
platinum disk,
and the electrochemical data
were corrected for the solution resistance (i.e. 10 Ohm). Cyclic
voltammetric
data were
collected at 10 mV s
-1
unless
otherwise specified.
The electrochemically
active surface
area (ECSA) was determined
using a previously
reported procedure.
8
After
cyclic voltammetry, the electrodes were held at open circuit for 1
min, and subsequently cyclic voltammetry
scans were
collected
at ± 50 mV from
open circuit at
scan rates
of
100, 75, 25, and 10 mV s
-1
. The
current
steps for
the cyclic voltammetry
were 100
6
μV
and
100% of the current was
acquired
during each time step. The anodic and
cathodic
currents at the center of the cyclic voltammetry scans were plotted versus scan
rate, and a linear
fit was used to determine
the capacitance. For all linear fits
the
R
2
value
was greater than 0.99.
Atomic-force microscopy was used to determine
the roughness factor of the ATO
(RF
= 1.32),
and the geometric area-normalized capacitance
of ATO
(0.0254 mF cm
-2
) was
divided by the
roughness factor to determine the capacitance normalized to
the electrochemical surface area
(0.0192 mF
cm
-2
).
The
roughness factor for antimonates
was
calculated by dividing
the
electrochemical surface area determined by capacitance measurements
by the geometric
area
of
the electrode. Impedance
measurements were performed
in
a static electrolyte of 1.0 M H
2
SO
4
(aq) in the frequency
range of 50 kHz – 50 Hz.
A potential of 1.63 V vs. RHE was applied and
the amplitude
of
the sinusoidal wave
amplitude
was 10 mV.
Inductively-coupled plasma mass spectrometry
Inductively coupled plasma mass spectrometry (ICP-MS)
data were
collected
using an
Agilent 8800 Triple Quadrupole ICP-MS
system. Calibration solutions were prepared
by
diluting the multielement and antimony standard
solutions for ICP with
18.2
cm resistivity
water.
To
determine the
loading of the catalysts,
metal films were sputter deposited on glass
slides and cut into ~
1 cm
2
pieces, and the geometric area of the pieces was measured with
an
optical scanner
(Epson perfection
V360)
and ImageJ. The unannealed
metal
films on glass were
placed in 10 mL of 1.0 M H
2
SO
4
(aq),
and the
films were allowed to dissolve for several days.
The total amount of Ni, Mn, and Sb in solution was determined from ICP-MS measurements,
and the results were
normalized to the geometric area of the pieces used.
7
A two-compartment electrolysis cell with a Nafion
membrane separating the reference electrode
and working electrode (~ 0.9
cm
2
) from the counter electrode was used to determine the amount
of Ni, Mn, and
Sb from
the working compartment (25 mL) in
the electrolyte during
chronopotentiometry at
10 mA cm
-2
. At various time intervals 1 mL of electrolyte was
taken
from the working
compartment, and immediately replenished with 1 mL of fresh
electrolyte.
The concentration of Ni, Mn, and Sb in the working compartment was
determined for every
time
interval, and was corrected for the dilution that occurred
each time a
sample was taken.
Electrolyzer Efficiency
Calculation and Solar Fuel Device Efficiency Discussion
The expected efficiency of an electrolyzer was calculated
with equation 1. The
overpotential
for
the hydrogen-evolution
reaction (HER) was
taken to be 80 mV (for example,
Ni
2
P and CoP),
9
and the voltage
drop from solution resistance was
taken to be 100 mV. The
efficiency for Ni
0.5
Mn
0.5
Sb
1.7
O
y
(~ 745 mV overpotential) was
57.7%, and for IrO
x
(~350 mV)
the efficiency was 69.9%.
The maximum
efficiency expected when an optimized tandem cell is coupled with
a set
of electrocatalysts has been determined previously.
10
Briefly,
the procedure
consists of
determining the overall overpotential of the water splitting reaction, in
this case
825 mV at 10
mA cm
-2
(745 mV for Ni
0.5
Mn
0.5
Sb
1.7
O
y
and
80 mV for
an HER catalyst) and finding a
combination of semiconductors operating at the Shockley-Queisser limit that maximize the
efficiency of the device. For semiconductors operating
at the Shockley-Queisser
limit
an
efficiency above
20% can be obtained
for devices with overall overpotentials and resistance
losses < 960 mV.
10
8
(1)
퐸푓푓푖푐푖푒푛푐푦
=
1.229 푉
1.229 푉
+
푂퐸푅
+
퐻퐸푅
+
푖푅
Turnover Frequency and
Turnover Number
Calculations
The turnover frequency
and
turnover
number of Ni
0.5
Mn
0.5
Sb
1.7
O
y
were determined by
calculating the
amount of catalyst
with the total number of transition-metal atoms in
the film or
by estimating
the
number of exposed transition
metals
based on the electrochemically active
surface area
(ECSA). The
ECSA should give an upper bound to the values, and the total number
of transition
metals should
give a lower bound. An estimate
for the surface
atom density
(4.6x10
14
atoms
cm
-2
) based on the crystal structure of transition-metal antimonates was used to
convert the ECSA
to the number of transition-metal
atoms
exposed.
9
Supplementary Information Figures
Figure S1.
X-ray diffraction pattern for Ni
x
Mn
1-x
Sb
1.6-1.8
O
y
thin films on an ATO substrate. The
peaks were indexed to reference patterns of SnO
2
and CoSb
2
O
6
.
2, 3
Figure S2.
(a) Surface nickel-to-transition
metal
ratio and (b) surface antimony-to-transition
metal ratio,
as determined
by XPS
measurements,
as a function of the bulk composition
10
determined by
ICP-MS. Error
bars
are
included
in each data point and
are
one
standard
deviation from
at least 3 samples.
Figure S3.
(a) Cyclic voltammetry of Ni
x
Mn
1-x
Sb
1.7
O
y
, SbO
y
, and ATO
electrodes
without
resistance compensation. (b) Overpotential
at 0.1 mA cm
-2
of electrochemically
active surface
area for Ni
x
Mn
1-x
Sb
1.7
O
y
. (c) Impedance measurements of Ni
x
Mn
1-x
Sb
1.7
O
y
, SbO
y
in the
frequency range of
50 Hz
– 50 kHz, 10 mV amplitude in
contact with 1.0 M H
2
SO
4
(aq) at 1.63
V vs. RHE
.
Electrode areas were
~0.2
cm
2
for impedance measurements. Impedance
data were
fit to a
model (insert) that
consisted of the series resistance (R
s
) associated with
the electrode
and
electrolyte resistance, in series with a
constant-phase
element
(CPE) in parallel with
the
contact
resistance (R
ct
) associated with the OER kinetics.
11
Figure S4.
(a) Tafel
slopes determined
from cyclic voltammetry in
the current density range of
0.3-3 mA
cm
-2
. (b) Current
density (
J
) at
η
= 650 mV for
Ni
0.5
Mn
0.5
Sb
1.7
O
y
at various loadings
determined by
calibration
of the deposition rate of the sputterer.
The
loading
is defined as the
amount of
Ni
and
Mn present in the as-synthesized film.
A
loading
of 0.48
mol cm
-2
corresponded to a
film thickness of ~ 300 nm. Error bars are one standard deviation from at least
3 samples. (c) Area-normalized electrode
differential capacitance
at various loadings. Higher
differential capacitance indicates
more
electrochemically active
surface area.
Figure S5.
Potential
of a
Ni
0.5
Mn
0.5
Sb
1.7
O
y
electrode being
held at 10 mA cm
-2
for 8 h, followed
by 16 h at open circuit and
subsequently
being held at 10 mA cm
-2
for 8 h.
12
Figure S6.
(a) Cyclic voltammetry of NiSb
1.8
O
y
and
MnSb
1.7
O
y
after 50 h of
chronopotentiometry at
10 mA cm
-2
based on the
electrode geometric
area. (b)
Chronopotentiometry of Ni
0.5
Mn
0.5
O
y
, SbO
y
, ATO,
and Ni
0.5
Mn
0.5
Sb
1.7
O
y
at 10 mA cm
-2
based
on the electrode geometric area.
(c)
Area-normalized differential capacitance of a
Ni
0.5
Mn
0.5
Sb
1.7
O
y
electrode determined
between chronopotentiometry tests
at 10 mA cm
-2
of
geometric area.
Figure S7.
Cumulative
dissolution of a Ni
0.5
Mn
0.5
Sb
1.7
O
y
electrode during
chronopotentiometry
at 10 mA cm
-2
. The active electrode geometric
area was 0.920 cm
2
.
13
Figure S8.
Diffraction pattern of Ni
0.5
Mn
0.5
Sb
1.7
O
y
before and after chronopotentiometry for
144 h at 10 mA cm
-2
of geometric area. The
peaks were indexed to reference patterns of SnO
2
and CoSb
2
O
6
.
2, 3
Figure S9.
High-angle
annular dark-field
(HAADF) images of (a)
Ni
0.5
Mn
0.5
Sb
1.7
O
y
before
electrochemical operation, and (b)
Ni
0.5
Mn
0.5
Sb
1.7
O
y
after chronopotentiometry
for
144 h at 10
mA cm
-2
. The lower portion of the
image
was identified as the
ATO
support, and a
porous film
14
on top of the ATO was identified as the
catalyst film. The film
thickness
for each image was
measured as the
distance
between the compact ATO layer and the dark portion of the image
corresponding to regions without material.
Figure S10.
SEM images of
(a) ATO substrate,(b) Ni
0.5
Mn
0.5
Sb
1.7
O
y
before electrochemical
operation, and (c) Ni
0.5
Mn
0.5
Sb
1.7
O
y
after chronopotentiometry for 144 h at 10 mA cm
-2
.
Figure S11.
STEM-EDS data of
as-synthesized
Ni
0.5
Mn
0.5
Sb
1.7
O
y
showing
that only
Ni, Mn,
Sb
were detectable in the
film. Cu, C,
Ga,
N, and Si are
from the TEM
grid and TEM
sample
preparation.
15
Figure S12.
High-resolution
XPS data
of
as-synthesized Ni
x
Mn
1-x
Sb
1.6-1.8
O
y
and
Ni
0.5
Mn
0.5
Sb
1.7
O
y
after chronopotentiometry for
144 h at 10 mA cm
-2
. The spectral regions are:
(a) Ir
4f and (b) Ru 3d. No Ir
or Ru was detected
on the surface by XPS.
Figure S13.
Chronopotentiometry at 10 mA cm
-2
of Ni
0.5
Mn
0.5
Sb
1.7
O
y
films
annealed
at 500 °C
and 700 °C.
16
Figure S14.
X-ray diffraction of Ni
0.5
Mn
0.5
Sb
1.7
O
y
films annealed at 500 °C,
700 °C, and
900
°C.
Reference patterns are SnO
2
(red
circles) and
CoSb
2
O
6
(blue triangles).
Table
S1.
Composition of electrocatalysts determined with ICP-MS and initial roughness
factor
(RF) determined from
capacitance measurements.
Sample
Ni
Mn
Sb
RF
(μmol
cm
-2
)
(μmol
cm
-2
)
(μmol
cm
-2
)
MnSb
1.7
O
y
0.000
0.475 ± 0.007
0.797 ± 0.005
18 ± 4
Ni
0.2
Mn
0.8
Sb
1.6
O
y
0.111 ± 0.003
0.392 ± 0.005
0.808 ± 0.012
17 ± 2
Ni
0.5
Mn
0.5
Sb
1.7
O
y
0.228 ± 0.004
0.251 ± 0.004
0.813 ± 0.008
29 ± 2
Ni
0.7
Mn
0.3
Sb
1.7
O
y
0.342 ± 0.003
0.143 ± 0.002
0.819 ± 0.012
12 ± 1
NiSb
1.8
O
y
0.470 ± 0.005
0.000
0.830 ± 0.003
4 ± 1
Table
S2.
Comparison of
X-Ray Photoelectron spectroscopy peak positions
for various samples.
Sample
Ni 2p
3/2
Mn 2p
3/2
Sb 3d
3/2
Reference
(eV)
(eV)
(eV)
Ni
0.5
Mn
0.5
Sb
1.7
O
y
Before
855.3 ± 0.1
641.3 ± 0.1
539.7 ± 0.1
Ni
0.5
Mn
0.5
Sb
1.7
O
y
After
855.6 ± 0.1
641.5 ± 0.1
540.0 ± 0.1
Ni(OH)
2
855.3 ± 0.1
-
-
4
NiO
853.8 ± 0.1
-
-
4
γ-NiOOH
856.2 ± 0.2
-
-
4
β-NiOOH
856.1 ± 0.1
-
-
4
17