1
Supplementary Information
for:
A Comparison of
the
Chemical, Optical
and Electrocatalytic Properties
of Water-
Oxidation Catalysts
for
Use
in Integrated
Solar-Fuels Generators
Ke Sun
1
, Ivan A. Moreno-Hernandez
1
, William C. Schmidt, Jr.
1
, Xinghao Zhou
2
, J.
Chance Crompton
1
, Rui Liu
3
, Fadl H. Saadi
2
, Yikai Chen
3
, Kimberly
M.
Papadantonakis
1
,
Nathan
S. Lewis
1,4,5*
1
Division of Chemistry and Chemical
Engineering,
2
Department of Applied Physics and Materials Science,
3
Joint Center
for Artificial Photosynthesis,
4
Beckman Institute and Molecular Materials Research Center,
5
Kavli Nanoscience Institute,
California Institute of Technology, Pasadena, CA
91125, USA
*Corresponding author: nslewis@caltech.edu
I.
EXPERIMENTAL
DETAILS
A.
Solvents
and Prepared Solutions
All materials
were
used as received,
except where otherwise noted. H
2
O
with a resistivity
of 18.2
MΩ-cm
was
obtained from a Barnsted
Nanopure
system. 1.0 M KOH(aq) was prepared
using potassium hydroxide
pellets
(Macron
Chemicals, ACS 88%). A 1.0 M
potassium borate
solution (K-Bi) was
prepared using a 1.0 M KOH(aq) solution made from potassium
hydroxide
pellets
and 2.0 M aqueous boric
acid (H
3
BO
3
, Sigma-Aldrich, BioReagent
≥99.5%).
1.0 M
Electronic
Supplementary
Material
(ESI)
for
Energy
&
Environmental
Science.
This
journal
is
©
The
Royal
Society
of
Chemistry
2017
2
H
2
SO
4
(aq) was prepared from sulfuric acid
(J. T. Baker, ACS reagent,
95%-98%).
0.10 M
potassium phosphate buffer (K-Pi) was prepared
using a mixture of 0.10 M
monobasic
potassium
phosphate (HK
2
PO
4
, Fisher Scientific,
ACS >99%)
and 0.10 M
dibasic
potassium phosphate
solution (H
2
KPO
4
, Fisher Scientific, ACS >99%) with a
volume ratio of 61.5:
38.5.
B.
Catalyst Deposition and
Electrode Preparation
1.
Substrates
Glass slides coated with
fluorine-doped tin oxide (FTO, TEC-15, Hartford) were soaked
for 1 min in 1.0
M H
2
SO
4
(aq),
thoroughly
rinsed with
H
2
O,
and then sequentially sonicated
in
acetone (CH
3
COCH
3
, EMD
Millpore
Corporation, ACS
grade,
99.5%) and
isopropanol (Wards
Science, lab grade,
≥99%)
for
5 min each. The
FTO-coated
glass substrates
were rinsed with
H
2
O following each
sonication step. The FTO-glass
substrates
were then dried under a stream
of N
2
(g)
and diced
into pieces
having an approximate size
of 1×1 cm.
2.
Sputtered
thin films
Thin films of
Pt
and Ni-Mo were deposited from
Pt
(Alfa, 99.99%) and Ni-Mo (Alfa,
99.99%) targets,
respectively,
onto FTO-glass substrates using a DC magnetron
sputtering
system (AJA
International
Inc.)
at a
chamber
pressure of 5 mTorr and under
a 20 sccm flow of
Ar.
The deposition power was 100 W for
Pt
and
150
W for Ni-Mo, respectively,
and no stage
heating was
used. Pt deposition
times of 23 sec, 1 min,
and
13 min, respectively produced film
thicknesses of
2 nm, 5 nm
and
60 nm, respectively. The deposition times
for
Ni
x
Mo
y
were 1 min
(~2 nm), 5 min (~10 nm) and 10 min (~20
nm).
Thin films of
NiO
x
and
IrO
x
were
deposited
from Ni (Kurt Lesker,
2”
diameter × 0.125”
thickness, 99.95%) or Ir (99.9% from
ACI Alloys,
Inc.) targets onto FTO-glass substrates via
3
reactive RF
sputtering using a
high-vacuum
magnetron sputtering system
(AJA International
Inc.). The maximum base
pressure
was 8×10
-8
Torr, and the working pressure was held at 5
mTorr under
a 20 sccm flow of Ar. For NiO
x
films, the
deposition rate was
maintained at 0.2 Å
s
-1
by adjusting the
sputtering power on the Ni target. The
NiO
x
films were 7 nm, 35 nm, and 70
nm thick, respectively,
as
measured by a calibrated stylus profilometer
(DektakXT). The IrO
x
catalyst layer
was
sputtered onto the FTO-coated glass substrates
from an RF source at 200 W
at
300 °C under a
constant
flow of 3.0/3.0 sccm Ar/O
2
, while maintaining an
overall pressure
of 5
mTorr. Control of the
deposition time yielded IrO
x
film
thicknesses
of 10 nm (2 min), ~60 nm
(10 min) and >100 nm (30 min), respectively.
3.
Preparation of Electrodes
Silver paste (SPI supplies)
was
used to attach
a Cu wire to the
front
side of each of the
FTO-coated glass
substrates. The
Ag paste was allowed to
dry
in
air for at least 1 h and then the
Cu wire was
threaded through a glass tube. The
substrate was
attached and sealed
to
the end of
the glass tube using gray
epoxy. All of the substrates
were centered on the glass tubing and were
oriented in a side-facing manner. The epoxy
was allowed to
cure in air
for > 12 h. The exposed
area of each electrode,
as defined by the edge
of the applied epoxy, was
imaged with
a high-
resolution optical
scanner and measured using ImageJ software. Exposed
areas for the FTO-
coated glass substrates were 0.3–0.4 cm
2
, unless specified
otherwise.
4.
Electrodeposition of
Thin
Films
CoO
x
films,
CoO
x,ed
, were deposited anodically onto FTO-coated glass electrodes
followed by performing a
pretreatment consisting of 10
voltammetric cycles in
1.0 M KOH(aq)
at potentials between 0.0 and 0.8 V vs a Hg/HgO/1.0
M KOH(aq) reference electrode (CH
4
Instruments 152). The FTO-glass electrodes were then immersed in an
aqueous solution that
contained 10
mM cobalt nitrate hexahydrate
(Co(NO
3
)
2
·6H
2
O
Alfa Aesar, ACS grade,
≥98%)
and 0.10 M
sodium acetate anhydrate
(NaOOCCH
3
Sigma Aldrich, >99%
).
1
The pH
of the
solution was 7.5. The deposition was
conducted
using a three-electrode
configuration with a
fritted Pt mesh (Aldrich) counter electrode
and
a saturated calomel
electrode,
SCE (CH
Instruments 150), as a reference electrode.
The current density was
maintained at 76 μA cm
-2
,
and no bubbles
were observed during deposition. The deposition
was
performed until
the total
charge density passed was between 5 mC cm
-2
and
120 mC
cm
-2
, which required 1–25 min,
depending on the
desired
film
thickness.
Cobalt phosphate films, Co-Pi, were deposited anodically onto FTO-glass electrodes
from a fresh
aqueous precursor solution
that
consisted of 0.10 M potassium phosphate,
K-Pi, and
0.50 mM
cobalt
nitrate (Co(NO
3
)
2
Alfa
Aesar,
ACS grade,
≥98%
).
2
The
K-Pi solution was
prepared by mixing an
aqueous 0.10 M solution of dibasic potassium phosphate
(H
2
KPO
4
, Fisher
Scientific, ACS >99%) with an aqueous 0.10 M solution
of monobasic
potassium phosphate
(HK
2
PO
4
, Fisher Scientific,
ACS
>99%) such that the K-Pi
solution was
0.062 M
K
2
HPO
4
and
0.038 M
KH
2
PO
4
. Electrodeposition was performed at room
temperature
under galvanostatic
control in a two-compartment cell,
with a current density
of 0.10 mA cm
-2
. The solution was
neither stirred
nor agitated during the deposition. A fritted Pt mesh
served as the counter
electrode and
a SCE
(calibrated to 0.654 V versus a reversible hydrogen electrode, RHE) served
as the reference electrode. The
deposition was continued until
5–120 mC cm
-2
had been passed,
depending on the
desired
thickness for the film. Samples were then rinsed with H
2
O
and dried
under a
stream
of
N
2
(g).
5
Ni(OH)
2
and Ni
x
Fe
1-x
(OH)
y
films were electrodeposited cathodically
onto
FTO-glass
electrodes. Prior to deposition,
the substrates were immersed in 1.0 M KOH(aq) and cycled ten
times through
the
potential range
of 0.0 to 0.8 V versus
a Hg/HgO reference electrode
(CH
Instruments) with a
fritted Pt
mesh counter electrode.
Films of Ni(OH)
2
were deposited from
an
unstirred aqueous solution
of 0.10 M Ni(NO
3
)
2
·6H
2
O
(Sigma Aldrich, 99.999% trace-metals
basis).
Films
of
Ni
x
Fe
1-x
(OH)
y
were deposited from
unstirred solutions with
a total
metal
content
of 0.10
M,
with Fe accounting for
25% of the
metal content, 5 mM Ni(NO
3
)
2
·6H
2
O
(Sigma
Aldrich, 99.999%
trace-metals basis)
and 5 mM FeSO
4
·7H
2
O
(Sigma Aldrich, ACS reagent,
≥99.0%).
To
prevent precipitation, the Ni solution was
purged
with Ar(g) for at least 30 s prior
to addition of
the
Fe salt. Depositions
were
conducted
for 2-10 s using a current density of
−10
mA cm
−2
, resulting in nominal film thicknesses of 7-40 nm.
5.
Atomic-layer
Deposition
CoO
x
films,
CoO
x,ALD
, were deposited onto FTO-glass substrates
via atomic-layer
deposition (ALD) using a
Cambridge Nanotech S200
ALD
system. The cobaltocene
(bis(cyclopentadienyl)cobalt(II), Cp
2
Co, Strem, 98%) precursor was heated to 80 °C and
maintained at that
temperature during the deposition, which
was
conducted
using a
substrate
temperature of 150 °C. Each ALD cycle consisted of
a 2 s
pulse
of the
cobaltocene precursor, a
10 s purge under a
20 cm
3
min
-1
flow of
research-grade N
2
(g), a 5 s ozone
pulse,
and a
further
10
s N
2
(g)
purge.
60-1000
ALD
cycles were
performed
to produce film thicknesses of 2-50 nm.
C.
Scanning
Electron Microscopy
Images of
electrocatalyst
films
were
collected
using a calibrated Nova NanoSEM 450
(FEI) SEM with an
accelerating
voltage
of 5 kV.
6
D.
Electrochemical Measurements
A three-necked flask
was used for all electrochemical measurements. Cyclic
voltammetry, current interruption
for
resistance
correction, and
in
situ electrochromism data
were obtained
using a
Biologic
MPG-2-44 potentiostat (Bio-Logic
Science Instruments). The
cyclic voltammetric data were recorded at a scan
rate
of 10 mV s
-1
. Current interruption at a
current density
of 10 mA cm
-2
was used to determine, and
compensate for, losses
due
to
ohmic
resistance. 85% iR
corrections were
performed when measuring the kinetic overpotential of the
electrocatalysts, and
were critical when large-area
FTO-coated
glass substrates were used in
the
optical transmittance measurements.
For electrochemical
measurements in 1.0 M
KOH(aq), including photoelectrochemical
measurements, spectral response
behavior,
and electrochemical impedance spectroscopy, a
mercury/mercury oxide
(Hg/HgO
in 1.0 M KOH(aq), CH
instruments,
CH152)
reference
electrode was
used, and a
carbon cloth in a
fritted glass tube (gas dispersion tube Pro-D,
Aceglass, Inc.) was
used as
the counter
electrode. The Hg/HgO reference electrode had
a
potential of
0.926 V versus
the reversible hydrogen electrode, RHE.
For electrochemical
measurements in 1.0 M
H
2
SO
4
(aq) or 1.0 M K-Bi(aq), a SCE was
used as the reference
electrode and
a fritted Pt mesh (0.5
mm diameter,
99.99% trace metal basis,
Alfa, Aesar)
was used as the counter
electrode.
For HER catalysts, the solutions
were purged
with high purity H
2
(g) for
30
min prior to measurements of electrocatalytic activity for the HER,
and the H
2
(g)
flow was maintained during such measurements. For measurements of
electrocatalytic activity
for the OER, solutions were pre-saturated using high
purity O
2
(g), and
the O
2
(g)
flow was
maintained during such measurements.
7
For Ni-based OER catalysts, 10 CVs were
performed
in the range 0.7 – 1.91 V versus
RHE to activate the
catalyst prior
to characterization.
E.
In situ
Measurements of Optical Transmittance
In situ measurements of the optical
transmittance
of OER-catalyst-coated FTO-glass
substrates were performed
using a custom
Teflon cell. Monochromatic light
was
incident upon a
quartz
window
at an
angle
normal to the surface, as well as upon a calibrated Si photodiode
(Thorlab Fs-100cal)
that
was
isolated
from the corrosive electrolyte by another quartz
window.
Samples were
placed
between the two quartz windows
with
the counter and
reference electrodes
out of the
optical
path. The
exposed
area of the
Si
photodiode was 0.03 cm
-2
. The exposed areas
of the sample
electrodes (0.3-0.4
cm
-2
) and the
spot
size of the
incident
light
were both larger
than the exposed area of the Si
photodiode.
The electrochromism of the sample was
determined
by monitoring the
transmittance at a fixed
illumination wavelength (550 nm) while scanning the
applied potential in the range
of 0.0–1.0 V versus Hg/HgO,
or while galvanostatically holding
the current density at 10 mA cm
-2
. A peristaltic pump
was
used to
circulate the electrolyte at a
high circulation
rate over the sample surface,
to minimize the accumulation of bubbles and thus
to minimize scattering of
light
by bubbles formed
on
the
electrode
surface. The transmittance
was defined
by
the
ratio between the transmission of catalyst-coated FTO substrates and the
transmission of the
bare FTO substrate.
HER catalysts are typically metallic
and
do not
exhibit electrochromism under reductive
conditions, so
the
optical transmittances of the HER catalysts were measured ex situ without
applied cathodic
bias.
The total optical transmittance
was
calculated
using equation (1):
8
(1)
퐽
=
1100
∫
휆
=
350
푞
ℎ푐
푇
(
휆
)
∙
푃
(
휆
)
∙
휆
∙
푑휆
1100
∫
휆
=
350
푞
ℎ푐
푃
(
휆
)
∙
휆
∙
푑휆
where
h
is Planck’s constant (6.63×10
-34
J s),
c
is the speed of light (2.998×10
8
m
s
-1
),
q
is
unsigned elementary charge (1.602×10
-19
C),
T
is the
transmittance
at each wavelength,
P
is the
irradiance in W m
-2
nm
-1
,
and
λ
is the light wavelength in nm.
Figure S1 shows
the in situ
transmittance-measurement setup.
9
Figure S1.
a) An optical image of
a typical
electrode
for
in
situ
optical
transmittance
measurements. b) A schematic of the experimental
setup for the in situ optical transmittance
measurements. Measurements were
conducted at a
constant current density of |10 mA cm
-2
| with
anodic current for
OER catalysts
and cathodic current for HER catalysts. Circulation was used
to remove
bubbles from electrode
surface. A calibrated Si photodiode
was used to
measure
the
optical attenuation by the
sample under working conditions.
c)
an optical image of an electrode
in the measurement
cell showing also the Si photodiode at the back.
Other
components are not
shown in the
image.
10
Figure S2.
SEM images of electrocatalysts studied in this
work
including Co-Pi, CoO
x
and
Ni
x
Fe
1-x
(OH)
y
at
different loadings.
11
Figure S3.
Solar irradiance spectra filtered by a water layer with
different
thicknesses
showing increasing
long-wavelength losses with increasing
water-layer thickness.
The active
surface
areas of CoO
x,ed
were
determined
using the cathodic charge density
(
Q
c
) measured
in the
pseudocapacitive region
(Figure S3),
which increased from
0.12 ± 0.02 mC
cm
-2
, to
0.58 ±
0.06, 10.06 ± 0.06 and 23.06 ± 1.48 mC
cm
-2
. A 20-fold increase in
the active
surface area
thus reduced
η
10 mA cm-2
by ~67 mV.
As
Q
d
for the Co-Pi films was increased from 10 mC cm
-2
to
60 or 120 mC cm
-2
,
respectively,
Q
c
increased from 1.18 ± 0.04 mC cm
-2
to
2.46 ± 0.26 and 4.15 ± 0.16 mC cm
-2
,
respectively. The ratio
between
Q
c
and
Q
d
is a
measure of the deposition
efficiency. This
efficiency remained small (~0.03)
and constant as
Q
d
increased for Co-Pi films. In contrast,
for
the CoO
x,ed
films,
Q
d
continuously increased from
0.10 to
0.18 during deposition.
As the number
12
of ALD cycles increased from 60 (~3 nm) to 1000 cycles (~50 nm),
Q
c
(Figure 1c
inset)
remained constant at 0.082 ± 0.002 mC cm
-2
. The distinctive redox peaks
observed at ~1.1 V
versus
RHE (highlighted by red
arrows
in Figure S3 on CoO
x,ed
and Co-Pi films) were not
observed for CoO
x,ALD
films
in contact with 1.0 M KOH(aq).
13
Figure S4.
Dependence of redox peak area (
Q
c
) on method of preparation
and
material
loading. a)
CoO
x,ed
, b)
Co-Pi, and c)
CoO
x,ALD
.
14
Figure S5.
Tafel plots
of
CoO
x,ed
at
different loadings in 1.0 M KOH(aq) and 1.0 M K-
Bi(aq).
15
Figure S6.
Tafel plots
of
Co-Pi at different
loadings in 1.0 M
KOH(aq) and 1.0 M K-
Bi(aq).
16
Figure S7.
The
J-E
behavior showing a decrease of
Q
c
on Co-Pi at a
loading of 120 mC
cm
-2
in
1.0 M
KOH(aq)
during 10 cycles of CV scans.
17
Figure S8.
Tafel plot of CoO
x,ALD
at
different loadings in
1.0 M KOH(aq).
18
Figure S9.
Tafel plots
of
NiO
x,sp
at
different
loadings in
1.0 M KOH(aq)
and
1.0 M
K-
Bi(aq).
19
Figure S10
.
J-E
behaviors of NiO
x,sp
at a loading of
70 nm
(a) and CoO
x,ed
at a loading
of
127.9 mC
cm
-2
(b)
in 1.0 M
KOH(aq) over 10 cycles
of CV
scans showing
continuous
improvement potentially caused by Fe contamination. Arrows
in the figure indicate
the direction
of the shift
of
the
J-E
data over
CV scans.
20
Figure S11.
Tafel plots
of
Ni
x
Fe
1-x
(OH)
y
at
different loading in
1.0 M KOH(aq)
and
1.0
M K-Bi(aq).
21
Figure S12.
J-E
behaviors of Ni
x
Fe
1-x
(OH)
y,ed
at
a loading of 10 s
in
1.0 M KOH(aq)
over 10
cycles of CV scans showing slight degradation.
The arrow in
the figure indicates the
direction of
the shift of the
J-E
data
with additional
CV scans.
22
Figure S13.
Comparison of polarization
plots
of
Ni(OH)
2,ed
and
Ni
x
Fe
1-x
(OH)
y
at the
highest loadings in 1.0 M KOH(aq).
23
Figure S14.
The
J-E
behavior showing a decrease of
current
density
on NiO
x,sp
at a
loading of
70
nm in 1.0 M
K-bicarbonate(aq)
during 10 cycles of CV scans. Arrow shows
the
shift of
the
J
-
E
data over CV scans.
Ir and Ru based oxides are
the only reported dimensionally stable and
active catalysts
for
electrochemical water oxidation in acidic media, both
of which
are
scarce and expensive.
We
also note
that there are
recent
reported
efforts directed toward the development of acid-stable
OER catalysts based on earth-abundant elements.
In
fact, Ir and
Ru by themselves have
shown
corrosion under operational conditions and poor
kinetics.
Some metal oxides
(Ta,
Sn, Ti, Si, Ce,
Ni, Co,
Nb,
and
etc)
have been used together with Ir or
Ru
to
form metal
oxide alloys to
improve
the catalytic activity
and
electrochemical stability.
24
Figure S15.
a) Representative chronoamperometric (black curve) and
chronopotentiometric (blue curves) behaviors observed during
deposition
of Co-Pi on freshly
HF-etched p
+
-Si and
FTO-glass
substrates,
and b)
J-E
behavior of Ni
x
Fe
1-x
(OH)
y,ed
coated p
+
-Si
substrates showing
degradation over
100 cycles of cyclic voltammetric scans.
25
Figure S16.
Potential-dependent
dynamic
transmittance at a fixed illumination
wavelength of 550 nm on CoO
x,ed
, Co-Pi, and CoO
x,ALD
at
different
loadings. a) CoO
x,ed
showed
increased electrochromism and reduced
transmittance at the bleached
state
with increasing
catalyst loading.
b) Co-Pi
showed increased
electrochromism, but negligible change in
transmittance at
the
bleached
state
with increased catalyst loading. b) CoO
x,ALD
showed reduced
transmittance at
the
bleached
state
and negligible
electrochromism
with increasing catalyst
loading.
26
Figure S17.
A)
Reflectance
spectra
on FTO
glass with
and
without a NiO
x
coating
showing negligible effects
from the direction
of the incident, and b) calculated NiO
x
thickness-
dependent transmittance spectra. The
total reflectances of the samples versus wavelength, with
different NiO
x
thickness
varied from
0-70 nm with water
as incident media and illuminated
under various
incidence angles,
were
calculated based
on the propagation matrix of plane waves
at dielectric interfaces. The Matlab function
(multidiel)
developed
by Sophocles
J. Orfanidis
from Rutgers
University,
was
used to calculate the reflection responses of the isotropic non-lossy
multilayer dielectric structures
3
. Optical constants for water,
FTO, glass and NiO
x
were adopted
from peviously
reported
values
4
.
27
Figure S18.
Tafel plots
of
sputtered
Pt films in various electrolytes showing no loading
effect within the
thickness range
studied.
28
Figure S19.
Tafel plots
of
sputtered
Ni
x
Mo
y
films showing
the loading
effect in various
electrolytes.
29
Figure S20.
J-E
behaviors of CoO
x,ed
at a loading
of 127.9 mC
cm
-2
(a), CoO
x,ALD
at
a
loading of
50
nm (b),
and Ni
x
Fe
1-x
(OH)
y,ed
at a loading of 2 s
(c) in 1.0 M
K-Bi(aq)
during 10
cycles
of CV
scans.
30
Supporting table:
Table
S1
. Efficiency and
band-gap combinations for
solar fuels devices
based on a
tandem light
absorber with a photocathode
top cell
and Si photoanode bottom
cell, in
contact
with 1.0 M
H
2
SO
4
(aq) as
a function of the loading
of
a Pt catalyst film.
Efficiencies
are
calculated based on a
tandem cell
with a
fixed band-gap
combination
of 1.70 eV/1.12
eV (entries
above the double horizontal
line),
and based
on tandem cells
with the band
gap for the
photoanode optimized for
the properties of the Pt catalyst film
(entries below the double
horizontal line). Ru was used as an
OER catalyst
in acidic condition with
an
overpotential loss
of
350 mV.
Pt loading
(nm)
Top junction
(1.70 eV)
current
density (mA
cm
-2
)
Bottom
junction
(1.12 eV)
current
density (mA
cm
-2
)
Total
overpotential
loss
(
η
OER+HER
/mV)
Integrated
transmittance
(%)
STH
Efficiency
(%)
2
18.6
14.3
397
83
17.6
5
10.6
7.2
398
45
10.6
70
<0.1
<0.1
390
<0.4
<0.1
Pt loading
(nm)
Top junction
band gap
Bottom
junction band
Total
overpotential
Integrated
transmittance
Efficiency
(%)