of 16
Cell Reports Physical Science, Volume
3
Supplemental information
Addressing solar photochemistry durability
with an amorphous nickel antimonate photoanode
Lan Zhou, Elizabeth A. Peterson, Karun K. Rao, Yubing Lu, Xiang Li, Yungchieh Lai, Sage
R. Bauers, Matthias H. Richter, Kevin Kan, Yu Wang, Paul F. Newhouse, Junko
Yano, Jeffrey B. Neaton, Michal Bajdich, and John M. Gregoire
Supplemental Experimental Procedures
Tables S1
-
S3 provide experimental details for photoelectrochemistry experiments, including the
calibration data required to calculate EQE as follows:
퐸푥푡푒푟푛푎푙
푄푢푎푛푡푢푚
퐸푓푓푖푐푖푒푛푐푦
(
퐸푄퐸
)
=
1240
×
표푡표푐푢푟푟푒푛푡
[
]
퐼푛푐푖푑푒푛푡
푙푖푔
푤푎푣푒푙푒푛푔푡
[
푛푚
]
×
푃표푤푒푟
[
]
EQE is also known as incident photon
-
to
-
current conversion efficiency (IPCE).
Table S1
. S
ummary of electrolytes used in PEC measurements.
pH
Abbreviation
Electrolyte + (sacrificial hole acceptor)
1
MET1
0.1 M sulfuric acid + 0.25 M sodium sulfate +
(0.1 M methanol)
7
SLF7
0.05 M potassium phosphate monobasic + 0.05 M potassium phosphate
dibasic + 0.25 M sodium sulfate + (0.01 M sodium sulfite)
10
OER10
0.1 M boric acid + 0.085 M potassium hydroxide + 0.25 M sodium sulfate
10
SLF10
0.1 M boric acid + 0.085 M potassium hydroxide + 0.25 M sodium sulfate +
(0.01 M sodium sulfite)
Table S2
. Irradiance for SDC experiments in SLF7 and SLF10 in Figure 1 (measurement area = 2 mm
2
).
LED (eV) (Doric LEDC4)
Power (mW)
Irradiance (mW cm
-
2
)
3.20 ± 0.05
3
150
2.72 ± 0.08
2.18
109
2.41 ± 0.08
0.92
46
2.06 ± 0.03
0.42
21
Table S3
.
Irradiance for SDC experiments in MET1 in Figure 1 (measurement area = 2 mm
2
).
LED (eV) (Doric LEDC4)
Power (mW)
Irradiance (mW cm
-
2
)
3.20 ± 0.05
2.93
146.5
2.72 ± 0.08
2.48
124
2.41 ± 0.08
1.05
52.5
2.06 ± 0.03
1
50
Supplemental
Notes
Supporting
data include
Pourbaix analysis (Figure S1)
, PEC (Figure S2), XPS (Figures S3
-
S4),
TEM (Figures S5
-
S6), TEM with PEC on an additional sample (Figure S7), XAS (Figures S8
-
S10 and Table S4), optical spectroscopy (Figure S11), DFT (Figures S12
-
S14 and Table S5
),
and comparisons with literature (Table S6).
Figure S1:
Pourbaix analysis of NiSb
2
O
6
.
Computationally predicted Pourbaix diagram of Ni
-
Sb
-
O
using MP data augmented with NiSb
2
O
6
rutile phase at a ratio of Ni:Sb
= 1:2. The total ion concentration
is fixed at 10
-
6
M. The Gibbs free energy of Ni(SbO
3
)
2
decomposition with respect to the Pourbaix stable
phases is superimposed on the diagram and represented by the color bar. The thermodynamic stability
window of Ni(Sb
O
3
)
2
extends from approximately 0.5 V vs RHE and above and across all pH. The two
red dashed lines denote 0 and 1.23 V vs RHE, respectively.
Figure S2.
PEC data in pH 1, 3, and 10.
A series of anodic chopped
-
illumination CV sweeps of am
-
NiSbO
z
thin film
photoanodes under 3.2 eV illumination in 3 electrolytes: a) SLF10, b) SLF7, and c)
MET1, respectively, followed by a CA at the applied potential of 1.23 V vs RHE. The photocurrent
degrades on 10
2
s scale in pH 1, 10
3
s scale in pH 7, and >>10
3
s scale in p
H 10.
Figure S3.
XPS analysis.
XPS
spectra for the am
-
NiSbO
z
sample before (as
-
synth.) and after (post
-
eche)
30 min photoelectrochemical stability measurement (Fig. 4b) in pH 10 at applied potential of 1.23 V vs
RHE under 3.2 eV LED illumination. (a
) Ni 2p and (b) Sb 3d/O 1s spectra used for the quantification of
Ni:Sb near
-
surface composition. For the as
-
synthesized thin film, the Ni surface composition was
x
= 0.5
the same to the bulk composition determined from XRF,
and was
x
= 0.46 Sb
-
rich after
PEC operation.
Figure S4.
XPS fitting for Ni.
The high
-
resolution core
-
level Ni2p
3/2
XPS spectra
of Ni
x
Sb
1
-
x
O
z
thin film
photoanodes: (a)
x
= 0.5, as
-
synthesized am
-
NiSbO
z
,
and (b)
x
= 0.
33
, as
-
synthesized NiSb
2
O
6
.
Deconvoluted Ni 2p3/2 XPS using
Gaussian
-
Lorentzian function represents 2 peaks with binding
energies of 856.0 and 857.2 eV, which are related to Ni
2+
(OH)
2
and Ni
3+
OOH, respectively, according to
the literature values. [
http://www.xpsfitting.com/2012/01/nickel.html
]. The Ni
2+
/Ni
3+
ratio
was directly
calculated from the area under each spectrum peak.
Figure S
5
:
TEM characterization.
The bright
-
field cross
-
section TEM images and the corresponding
energy dispersive x
-
ray spectroscopy (EDX) elemental mapping and depth profiles of
Ni
x
Sb
1
-
x
O
z
thin film
photoanodes: (a)
x
= 0.33, as
-
synthesized NiSb
2
O
6
, (b)
x
= 0.5, as
-
synthesized am
-
NiSbO
z
, and (c)
x
=
0.5, am
-
NiSbO
z
after 30 mins operation under
the
applied potential of 1.23 V vs RHE in
pH
10
electrolyte.
Figure S
6
:
HRTEM images.
HRTEM images of Ni
x
Sb
1
-
x
O
z
(
x
= 0.5) thin film after 30 mins operation
under
the
applied p
otential of 1.23 V vs RHE in pH
10 electrolyte reveal no lattice fringes. This
observation combined with XRD and SAED analysis together demonstrates the disordered at
omic
arrangement and the formation of amorphous Ni
-
Sb oxide (denoted as am
-
NiSbO
z
). There is no
discernible passivation layer.
Figure S
7
.
TEM and PEC on lower
-
surface
-
area sample.
Characterization of am
-
NiSbO
z
thin film
photoanode deposited on Pt/Ti/Si
O
2
/Si substrate and subsequently annealed at 610 °C in air. (a) The
bright
-
field cross
-
section TEM images and the corresponding energy dispersive x
-
ray spectroscopy
(EDX) elemental mapping; (b) two selected area (indicated by red circle 1 and 2) electron dif
fraction
(SAED) images showing diffuse rings that indicate the presence of amorphous materials; (c) the depth
profile of Ni, Sb, and O elemental concentrations cross the film thickness indicating the value of
z
is
about 3.55; (d) Chopped
-
illumination CA me
asured under 4 illumination sources in the SLF10 electrolyte
under
the
applied potential of 1.23 V vs RHE, followed by a CV sweep under 3.2 eV illumination at a rate
of 0.02 V s
-
1
. The anodic sweep is shown in (e). These results demonstrate that the nanost
ructure and
lateral composition inhomogeneities of Figure 5 and S4 arise from deposition on a rough substrate and
that photoactivity is not reliant on either of the morphological aspects of the photoanode. Conversely, the
photocurrent is approximately 4
l
ower for this film on Pt compared to Figure 1, likely due to the
relatively high surface area of the film deposited on F
-
doped SnO
2
.
Figure S
8
.
EXAFS analysis for Ni:Sb=1:1.
EXAFS single
-
shell analysis of
as
-
synthesized Ni
x
Sb
1
-
x
O
z
photoanodes at Ni concentration of
x
=
0.5 (am
-
NiSbO
z
) at
Ni K
-
edge. Total fit signal (red line)
superimposed on the experimental signal (blue dot).
Figure S
9
.
EXAFS analysis for Ni:Sb=1:2
.
EXAFS single
-
shell analysis of
as
-
synthesized Ni
x
Sb
1
-
x
O
z
photoanodes at Ni concentration of
x
= 0.33 (NiSb
2
O
6
) at
Ni K
-
edge. Total fit signal (red line)
superimposed on the experimental signal (blue dot).
Figure S
10
. EXAFS multiple
-
shell analysis
.
Multiple
-
shell fitting
of
the a) real and b) imaginary
components on the Ni K
-
edge EXAFS signal for
as
-
synthesized Ni
x
Sb
1
-
x
O
z
photoanodes at Ni
concentration of
x
= 0.33 (NiSb
2
O
6
)
. Curves from top to bottom are backscattering signals χ
3
of different
paths, and total fit signal
(red line) superimposed on the experimental signal (blue dot).
Table S4.
EXAFS fitting results.
EXAFS fits of
as
-
synthesized Ni
x
Sb
1
-
x
O
z
photoanodes at Ni
concentration of
x
= 0.33 (NiSb
2
O
6
) and 0.5 (am
-
NiSbO
z
). Fit #1: Single
-
shell fit of
x
= 0.5 (am
-
NiS
bO
z
);
Fit #2: single
-
shell fit of
x
=
0.33 (NiSb
2
O
6
); Fit #3: multiple
-
shell fit of
x
=
0.33 (NiSb
2
O
6
).
Fit
#
Sample
Path
R (Å)
CN
σ
2
×
10
3
2
)
E
0
(eV)
R
factor
1
Ni
x
Sb
1
-
x
O
z
(
x
= 0.5)
Ni
-
O
2.02 ± 0.01
5.02 ± 0.72
8 ± 1
-
0. 03 ± 1.55
0.012
2
Ni
x
Sb
1
-
x
O
z
(
x
= 0.
33
)
Ni
-
O
2.04 ± 0.01
5.53 ± 0.30
6 ± 1
0.55 ± 0.65
0.005
3
Ni
x
Sb
1
-
x
O
z
(
x
= 0.
33
)
Ni
-
O
2.03 ± 0.01
5.
73
±
1.27
6 ± 2
-
0.28 ± 1.64
0.07
Ni
-
Sb
3.06
± 0.02
2
8 ± 2
Ni
-
Sb
3.66
± 0.02
8
12 ± 2
For the multiple
-
shell fitting of
Ni
x
Sb
1
-
x
O
z
(
x
=
0.33) sample, FEFF paths were generated
using
a NiSb
2
O
6
crystal structure with mp
-
505271
. Due to the complexity of the 2+ shell on the NiSb
2
O
6
model (multiple
single
-
scattering and multiple
-
scattering paths), besides the first
-
shell Ni
O path, we chose the two major
Ni
Sb paths during the fit and fixed the Ni
Sb coordination numbers (2 and 8, respectively) to match
with the
computational
NiSb
2
O
6
structure (mp
-
505271) and fit the coordination distance and test whether
the NiSb
2
O
6
model would match. The first shell Ni
O coordination number is fitted as 5.02 ± 0.72, and
distance is 2.03 ± 0.01
Å, which is comparable to the
Ni
O coordination number o
f 6 and distance of
2.07
Å for
the computational
NiSb
2
O
6
structure
(mp
-
505271). The coordination distance of two Ni
Sb paths
are fitted as
3.06 and 3.66
Å, respectively, which is comparable to the 3.11 and 3.69 Å of the
NiSb
2
O
6
structure (mp
-
505271). The o
verall well
-
matched fitting shows that the
Ni
x
Sb
1
-
x
O
z
(
x
=
0.33) is well
aligned with the NiSb
2
O
6
model (mp
-
505271).
Figure S1
1
. Optical characterization of am
-
NiSbO
z
and NiSb
2
O
6
.
(a) The normalized transmittance T,
normalized total reflectance R, and absorption coefficient α are shown where for each spectrum the solid
and dashed lines correspond to am
-
NiSbO
z
and NiSb
2
O
6
, respectively. (b) Spectral EQE at the applied
potential of 1.23 V vs RHE in SLF10 electrolyte and Tauc signals, where the EQE (left) and the direct
allowed (DA) and indirect allowed (IA) Tauc signals each have a dedicated vertical axis. To best
approxima
te the intrinsic spectral absorption, these calculations were performed using thicknesses of 270
and 310 nm for am
-
NiSbO
z
and NiSb
2
O
6
, respectively, which were calculated from the XRF measurement
of molar content of Ni and Sb using weighted molar densities
of NiO and Sb
2
O
5
. This model for thickness
corresponds to the thickness of a perfectly dense film. The TEM images of Figure 5 indicate the thickness
of the film, which contains a substantial void density, is up to 2× larger. Using this thickness would lea
d
to a down
-
scaling of the absorption coefficient and Tauc signals but would not alter the conclusions
drawn from this optical spectroscopy data.