S
1
Supporting Information
for
:
Catalysis of the oxygen
-
evolution reaction in 1.0 M sulfuric acid by manganese
antimonate
films synthesized via chemical vapor deposition
Jacqueline A. Dowling
†,
⊥
, Zachary P. Ifkovits
†,
⊥
, Azhar I. Carim
†,‡
,
J
ake
M. Evans
†
,
Madeleine C.
Swint
†
, Alexander Z. Ye
†
, Matthias H. Richter
†
, Anna X. L
i
†
, and Nathan S. Lewis
†,‡,
*
†
Division
of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena,
California 91125, United States
‡
Beckman
Institute,
California Institute of Technology, Pasadena, California 91125, United States
⊥
J. A. D. and Z. P. I. contributed equally.
*
E
mail:
nslewis@caltech.edu
Experimental Methods
Materials and Chemicals
The c
hemical vapor deposition (CVD) precursor
,
bis(ethylcyclopentadienyl)manganese
(98%
-
Mn, STREM) (Mn(EtCp)
2
)
,
was
preloaded in a CVD cylinder
and
used as received.
The
a
tomic layer deposition (ALD) precursor tris(
dimethylamino)antimony (99.99%
-
Sb, STREM)
(TDMA
-
Sb) was loaded
into a new ALD cylinder in a glovebox under N
2
(g)
. TEC 8 fluorine
-
doped tin oxide (FTO, Sigma Aldrich) substrates were used as received. The electrolyte, H
2
SO
4
(
95.0
-
98.0%
,
ACS reagent grade,
JT Baker
)
was diluted to 1
.0
M using 18.2 M
Ω
cm resistivity
water obtained from a Thermo Scientific Nanopure system. In
-
Ga eutectic (99.99%, metals basis,
Alfa Aesar)
,
PELCO conductive Ag paint (Ted Pella, Inc)
, and
Loctite
epoxy
(
EA 9460
)
were used
as received. O
2
(g)
(ultra
-
high purity grade, Airgas) was used to produce O
3
(g)
,
which was used as
a co
-
reactant in chemical vapor deposition. O
2
(g)
(industrial
grade
, Airgas)
was used to
purge the
electrolyte solution during durability experiments.
12.1 M
HCl and
15.6 M
HNO
3
(
ACS
reagent
grade
,
Sigma Aldrich
) were used to clean glassware.
S
2
Sample preparation
Chemical vapor deposition was
performed
using a
Cambridge Nanotech
Savannah S200
Atomic Layer Deposition (ALD) System. After
the
precursors and co
-
reactants were pulsed
,
a
waiting period
allow
ed
the ALD chamber (set at 150 °C) to return to
the
base vacuum pressure
(~0.5 torr
)
at
a
20 s
c
cm N
2
flow rate
. P
recursor cylinder jackets were heated
as noted
(
Table S
1
)
.
A
fter consecutive
rinsing
with isopropyl alcohol, acetone, and H
2
O
,
and drying with N
2
(g)
,
the
TEC8 fluorine
-
doped tin oxide (FTO) substrates were loaded into the ALD chamber
.
P
rior to
deposition, a glass slide was positioned
in
the ALD chamber
on top of the FTO substrate
.
The slide
covered
part of the surface,
allowing for
subsequent
top
-
facing electrical contact directly to the
substrate during
the
fabrication
of electrodes
(Figure S
1
).
To
anneal the
samples, the t
emperature
of
a
Thermolyne muffle furnace
was
increased
from room temperature to 600 °C at a rate of 10 °C
min
-
1
,
and then
was
held for 6 h
at
600 °C
before
being allowed to return
to room temperature.
Table S
1
.
Chemical vapor deposition recipe for Mn
0.63
Sb
0.37
O
x
.
Figure S
1
.
Electrode preparation schematic.
A
glass slide partially cover
ed
the FTO substrate
to
allow
for
subsequent
electrical contact directly to
the
substrate.
S
3
A
nnealed Mn
0.63
Sb
0.37
O
x
samples were cleaved
,
and
In
-
Ga
was
scribed directly onto the
bare FTO substrate
that had been
covered in the ALD chamber
by a glass slide
(Figure S
1
a).
The
sample was affixed to
tinned Cu wire
using
Pelco conductive
Ag
paint (Ted Pella, Inc.)
,
and the
assembly was then
dried in
an
oven
for 1 h
at 95 °
C
.
E
xcept
for
a small area where the catalyst
was exposed
,
the entire electrode
was encased in epoxy
,
and
the entire electrode was
dried in
an
oven
for
~
8 h
at 95 °C. The
exposed
catalyst area (Figure S
1
d) was measured with an optical
scanner (Epson perfection V360)
,
and the
geometric
area
(
0.10
-
0.15 c
m
2
)
was quantified
using
ImageJ software.
Electrochemical
measurements
Electrochemical analysis was performed using a
B
iologic SP
-
200 potentiostat
with
EC
-
Lab software.
Experiments were performed in a
two
-
compartment, three
-
electrode configuration
in a
50 mL pyrex flask.
The potential of the
SCE reference
electrode
(
CHI150, CH Instruments
)
relative
to
a
reversible hydrogen
electrode (
RHE) was measured by
monitoring the
voltage
difference between the SCE and
an
unannealed
Pt mesh electrode
in 1
.0
M H
2
SO
4
(aq)
that was
exposed to a continuous stream
of
H
2
(g)
.
The
Ti
-
mesh counter electrode was isolated from the
working and reference electrodes using a porous glass frit (porosity of 10
-
20 μm, Ace glass). All
glassware
,
as well as the Pt mesh
,
were
cleaned by
immersion
for several
h
ours
in a freshly
prepared 3:1
(
by volume
)
solution
of
12
.1 M
HCl
(aq)
and
15.6
M
HNO
3
(aq)
,
followed by rinsing
with H
2
O before
use
for electrochemistry.
During all
experiments, the 1
.0
M H
2
SO
4
(aq)
electrolyte
solution (50 mL) was continuously stirred
and maintained at room temperature
. Water
-
saturated
O
2
(g)
was
continuously bubbled into the
solution during electrochemistry
.
The electrochemical durability experiment was
performed
under galvanostatic conditions
(
J
=
10 mA cm
-
2
) for seven 24 h intervals (168 h total).
All electrochemical data
are
presented
S
4
using
the IUPAC con
vention
. At 24 h intervals during durability tests at 10 mA cm
-
2
, three
voltammetr
ic
cycles
were
recorded
at a scan rate of
ν
= 40
mV s
-
1
,
with an initial potential,
E
0
=
1.04 V vs. RHE and
an
upper
-
limit potential of
E
1
= 1.94 V vs. RHE. Impedance data were
collected at open circuit (~1.4 V vs RHE)
by applying a
10 mV
amplitude
sinusoidal wave
and
measuring
10 data points per decade at frequencies that ranged from 10 Hz to 10 kHz. Impedance
data were fit to a
modified
Randles
circuit
(
R
1
+
Q
2
/
R
2
) with EC
-
lab software to determine the
solution resistance (
R
1
)
,
the
capacitance
of a
constant
-
phase element
(
a “leaky” capacitor,
Q
2
)
, and
the charge
-
transfer resistance
associated with
the
OER kinetics
(
R
2
)
.
The u
ncompensated
resistance
correction was set to equal
90% of the solution resistance (10
-
15
Ω)
measured by
impedance.
The value of
Q
2
was used to evaluate
the
electrochemically active surface area (ECSA)
as well as the
change in surface
roughness during the durability experiment. For ECSA
calculations, the roughness factor
(
RF
)
of the
TEC 8 FTO substrate
was
assumed to be equal to
that of antimony
-
doped tin oxide (ATO) (
RF
= 1
.32).
1
T
he 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
).
1
0.2 mL a
liquots
of
electrolyte were
taken during
galvanostatic operation
and were diluted
by 5 mL of 5%
w/w
HNO
3
(aq)
(
0.8 M
)
before analysis
by
inductively coupled plasma mass
spectrometry (ICP
-
MS) using an Agilent 8800 Triple Quadrupole ICP
-
MS system. Standards of
known concentration were
produced from
10
μ
g
mL
-
1
Mn and
999
±
2
μ
g mL
-
1
Sb standards
(Sigma Aldrich
) via
serial dilution with 5%
w/w HNO
3
(aq)
(
0.8 M).
E
udiometr
ic measurements indicated
>
97% Farad
a
ic efficiency for O
2
production
for
a
Mn
0.63
Sb
0.37
O
x
electrode
during 93 h
of continuous operation at
J
=
10 mA cm
-
2
.
S
5
Materials characterization
Ellipsometric
analysis was used to measure the g
rowth rates of individual oxides on Si
substrates
.
D
ata
were
collected using
65, 70, and 70 ° angles
across a wavelength range of 380
–
890
nm
with a
J.A. Woolam Co.
ellipsometer
. Ellipsometry data
were
analyzed
using the
CompleteEASE software
package
.
The
stoichiometry
of the catalyst material
was
determined
by dissolving deposited
unannealed metal films on Si substrates
for several days
in 10 mL of 1.0 M H
2
SO
4
(aq)
. The
concentration of dissolved metals
(Sb and Mn)
was then determined
using ICP
-
MS.
Films from
the same
deposition
batch were annealed
,
assembled
into electrodes (Figure S1)
,
and
subjected to
galvanostatic
testing
(Figure 2
, Figure 3)
.
Samples of the anolyte were taken periodically during
galvanostatic testing
and assessed via ICP
-
MS to quantify
the amount of
material
that had
dissolved into the electrolyte
.
The values from this ICP
-
MS analysis were normalized by the
concentrations
measured
when
the annealed metal films
were fully dissolved
, as described above,
to yield
the
percentages
of metals in the film that had dissolved at any point in time
.
The g
eometric area of the Si
substrates was measured with an optical scanner (using the
same procedure used to define
the
electrode area)
,
to determine the
area
-
normalized mass
-
loading
(23.53 μg Mn cm
-
2
, 32.3
0
μg Sb cm
-
2
). A mass loading of 0.42 μmol Mn cm
-
2
corresponded to
an
~100 nm thick
oxide
(Table S
2
).
Metal
Metal fraction
(%)
Mass loading
(μg cm
-
2
)
Mass loading
(μmol cm
-
2
)
Oxide thickness
(nm)
Sb
63%
32.3
0
0.27
35.7
Mn
37%
23.53
0.43
65.1
Total metal
100%
55.83
0.70
~100
Table S
2
.
M
ass loading and metal fraction
of a
s
-
prepared Mn
0.63
Sb
0.37
O
x
catalysts
determined by
ICP
-
MS
analysis of unannealed films digested
in 1
.0
M H
2
SO
4
(aq)
for several days
.
The o
xide
thickness
was
based on individual oxide growth rates
deter
mined from ellipsometry
and
the
total
subcycles
used during film growth
(Scheme 1).
S
6
Scanning
-
electron micrographs (SEMs) were obtained with a FEI Nova NanoSEM 450 at
an accelerating voltage of 10.00 kV with a working distance of 5 mm and an in
-
lens secondary
electron detector
.
Micrographs were acquired with a resolution of 688 pixels μm
-
1
over ~ 2 μm
2
areas. Energy dispersive X
-
ray (EDX) spectroscopy was performed in the SEM using an
accelerating voltage of 15.00 kV and a working distance of 5 mm
, with an
Oxford Instruments X
-
Max silicon drift detector. Spectra were collected in the range
of 0 to 10 keV
,
and quantitative
compositions
of the deposited material
were
obtained
using the “INCA” software package (Oxford
Instruments).
C
ompositions
were reported for
the average of
n
= 6 independent measurements
from different locations on the sample.
X
-
ray diffraction (XRD) data
were
collected using a Bruker D8 Discover diffractometer
with a Cu K
α
source and a two
-
dimensional Vantec detector. XRD data
were
acquired from the
FTO substrate in a Bragg
-
Brentano geometry. Grazing incidence X
-
ray diffraction (GIXRD)
data
were
collected from Mn
y
Sb
1
-
y
O
x
on a FTO substrate with the X
-
rays directed at a grazing angle
ω
= 0.3° above the plane of the sample surface
, with
the detector swept throughout the entire 2
θ
range.
X
-
ray photoelectron spectroscopy (XPS) was performed using a Kratos Axis Ultra system
with a base pressure of
3
× 10
-
9
Torr in the analysis chamber. A 150 W monochromatic Al K
α
source was used to irradiate the sample with X
-
rays (1486.6 eV).
F
or maximum depth sensitivity
,
the
hemispherical analyzer
was
oriented for detection along the sample surface normal. The data
were analyzed using CasaXPS software. A Shirley background was used for Mn spectra and a U 2
Tougaard background was used
to more accurately ca
pture the background signal
for Sb, O, and
C spectra. All peaks were referenced to adventitious C at a binding energy of 284.8 eV. The Mn
2p
3/2
peak was fit to standards of pure oxides
,
to estimate the Mn oxidation state and the
S
7
contribution of multiple oxidation states to the peak.
2
Mn 3p spectra were used in
conjunction
with
the
Mn 2p
3/2
emissions to
more accurately determine the Mn oxidation state.
2
–
6
The Sb 3d
5/2
peak
overlap
ped
the
O 1s
emission
, so
the
Sb oxidation state
was determined from
the Sb 3d
3/2
peak,
which has no overlap with O
.
The Sb
3d
3/2
peak was
fit with literature standards for
the
Sb
oxidation state
,
and the Sb 3d
5/2
contribution was calculated by constraining the spin
-
orbit peak
splitting (
Δ
Sb 3d = 9.38 eV),
taking
the full width at half max
imum
(FWHM) to be equivalent
between the peaks,
constraining the peak
area ratio
to be
3d
5/2
:3d
3/2
=
(
3:2)
,
and assuming
that the
remaining signal was due to O 1s.
Additional Data
Figure S
2
.
(A)
Cyclic voltammograms acquired
with a
Mn
0.63
Sb
0.37
O
x
electrode
in 1
.0
M
H
2
SO
4
(aq)
at 24 h intervals
by
interrupting a
continuous galvanostatic hold
of Mn
0.63
Sb
0.37
O
x
at
J
= 10 mA cm
-
2
.
(B)
OER overpotential to
obtain
J
=
10 mA cm
-
2
derived f
ro
m the
cyclic
voltammetric
data in
(A)
.
T
he IR
-
compensation
was
~
14 mV.
Larger
overpotentials were observed in
the first
voltammogram
acquired in each interval relative to the second and third
voltammograms
, and this
effect
increased
at later
time
intervals
of the experiment
(Figure S2B)
.
Figure S2
S
8
Figure S
3
.
Replicate of Mn
0.63
Sb
0.37
O
x
dissolution measurements
during
an
OER durability test.
(A)
The
time
-
average
d
OER overpotential was
η = 706 ± 21
mV at
J
=
10 mA cm
-
2
over
176 h in
1
.0
M H
2
SO
4
(aq)
. ICP
-
MS
measurements of the
corrosion products in the electrolyte during the
analogous 176
h durability test in relative (B) and absolute terms (C, D). The corresponding bar
graphs in panels (E
-
G) compare metal dissolution rates in various regimes:
0
-
168 h, 0
-
48 h, and
48
-
168 h.
Error bars represent the standard error of the dissolution rate given by linear regression
over the specified time regime.
Figure S3
Bubble in counter
electrode
interrupted circuit
S
9
Figure S
4
. (A) Catalyst composition as determined by EDX and ICP
-
MS. Representative SEMs
of Mn
0.63
Sb
0.37
O
x
(B) before and (C) after OER
at
J
= 10 mA cm
-
2
for 168 h in 1
.0
M H
2
SO
4
(aq)
.
Catalyst composition as determined by EDX after OER was
Mn
0.
49
Sb
0.
51
O
x
.
Figure S
5
.
(A) Impedance
data
and (B) roughness factor
for
Mn
0.63
Sb
0.37
O
x
taken at 24
h intervals
during OER at
J
= 10 mA cm
-
2
for 168 h in 1
.0
M H
2
SO
4
(aq).
The
TEC 8 FTO substrate roughness factor was assumed to be
equal to
that of
ATO (
RF
= 1.32).
1
According to preestablished methods, 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
).
1
C)
EDX
EDX
ICP
-
MS
B)
500 nm
C
)
R
s
R
c
t
CPE
S
10
Figure S
6
.
XP spectra of Mn
0.63
Sb
0.37
O
x
(A
-
C)
before and
(D
-
F)
after OER at
J
= 10 mA cm
-
2
for
168 h in 1
.0
M H
2
SO
4
(aq).
(A
, D
) Mn 2p spectra; (B
, E
) Mn 3p spectra; (C
, F
) Sb 3d, O 1s spectra.
Raw data (black dots)
were fit by peak models as described in experimental methods. The sum of
all fit peaks (e
nvelope) is denoted by a red line in all panels.
S
11
Figure S
7
.
O
xidation state
derived from analysis of
XP spectra
of
Mn
0.63
Sb
0.37
O
x
before and after
OER at
J
= 10 mA cm
-
2
for 168 h in 1
.0
M H
2
SO
4
. (A) Mn oxidation states
;
(B)
derived
Mn
oxidation states from
the
3p and 2p peaks. (C) Sb oxidation state
s
derived from the
3d
peak.
Mn 2p
Mn 3p
Sb 3d
Before OER
2.9
2.8
3.2
After
OER
3.2
3.4
5.0
Table S
3
.
Oxidation
state of
Mn
0.63
Sb
0.37
O
x
derived from analysis of
XP spectra
before and after
OER at
J
= 10 mA cm
-
2
for 168 h in 1
.0
M H
2
SO
4
(aq)
.
S
12
Figure S
8
.
Electrochemical activity and stability of Mn
0.63
Sb
0.37
O
x
during OER at
J =
100 mA
cm
-
2
for 9 h in 1
.0
M H
2
SO
4
(aq)
.
Figure
S
8
presents a magnified view of the same data
depicted
in
Figure 3.
(A) Chronopotentiometric response. (B) Cyclic voltammograms collected both
initially and
after
8 h
of
the
galvanostatic
hold. (C)
C
orrosion products
in the anolyte
, as
determined by
ICP
-
MS
.
Figure S8