SI Appendix for:
Title (max 135 characters including spaces, comprehensible to broad audience):
Stable Solar
-Driven Oxidation of Water by Semiconducting Photoanodes Protected by
Transparent Catalytic Nickel Oxide Films
Short Title (for mobile devices and
RSS feeds, < 50 characters including spaces):
Stable Water Oxidation by NiO
x
-Coated Photoanodes
Authors:
Ke Sun
a,b
, Fadl H. Saadi
b,c
, Michael F. Lichterman
a,b
, William G. Hale
b,d
, Hsin
-ping Wang
e
,
Xinghao Zhou
b,c
, Noah
T. Plymale
a
, Stefan
T. Omelchenko
b,c
, Jr
-Hau He
e
, Kimberley M.
Papadantonakis
a,b
, Bruce S. Brunschwig
b,f
, Nathan S. Lewis
a,b,f,g,1
Author Affiliations:
a
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena,
CA 91125, USA
b
Joint Center for Artificial
Photosynthesis, California Institute of Technology, Pasadena, CA
91125, USA
c
Division of Engineering and Applied Sciences, California Institute of Technology, Pasadena, CA
91125, USA
d
Department of Chemistry, University of Southampton, Southampton, Hampshi
re, SO17 1BJ UK
e
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division, King
Abdullah University of Science & Technology (KAUST), Thuwal 23955-
6900, Saudi Arabia
f
Beckman Institute Molecular Materials Research Center, California I
nstitute of Technology,
Pasadena, CA 91125, USA
1
g
Kavli Nanoscience Institute, California Institute of Technology, Pasadena, CA 91125, USA
SI Appendix
Text S1
:
Preparation of Samples:
Chemicals:
Except where otherwise noted, all materials were used as rece
ived, including sulfuric acid
(H
2
SO
4
, J. T. Baker, ACS reagent, 95%
-98%), concentrated hydrochloric acid (HCl, Sigma
Aldrich, ACS Reagent 37%), hydrogen peroxide (H
2
O
2
, Macron Chemicals, ACS grade 30%),
concentrated ammonium hydroxide (NH
4
OH, Sigma Aldric
h, ACS reagent 28%
-30%), buffered
HF improved (
Transene Company Inc.), potassium hydroxide pellets (KOH, Macron Chemicals,
ACS 88%), potassium ferrocyanide trihydrate (K
4
Fe(CN)
6
• 3H
2
O, Acros, >99%), potassium
ferricyanide (K
3
Fe(CN)
6
, Fisher Chemicals, certified ACS 99.4%), potassium chloride (KCl,
Macron Chemicals, Granular ACS 99.6%), monobasic potassium phosphate (
HK
2
PO
4
, Fisher
Scientific, ACS >99%
), dibasic potassium phosphate
(H
2
KPO
4
, Fisher Scientific, ACS >99%
),
cobalt(II) nitrate (
Co(NO
3
)
2
, Acro
s, ACS
>99%
), Br
2
(Sigma Aldrich, 99.999%), and CH
3
OH
(EMD Millipore, >99.9%). Water with a resistivity of 18.2 MΩ∙cm was obtained from a Millipore
deionized water system.
Preparation of substrates:
Boron degenerately doped (p
+
-type, resistivity <0.005
Ω∙
cm
, single
-side polished, (111)
-
oriented) Si wafers were purchased from Addison Engineering. The Si wafers were cleaned in
buffered oxide etchant, typically for 30-
45 s, until the surface was hydrophobic. The samples
2
were then immersed
for 30 min
in mixtur
e of 1:1:5 (v/v) of H
2
O
2
:HCl:H
2
O at 70 °C, and were
subsequently thoroughly rinsed using deionized H
2
O and dried using N
2
(g). Before loading into
the sputtering chamber, samples were etched with buffered HF(aq) to remove native oxides,
rinsed with deionize
d H
2
O, and dried using N
2
(g).
As
-cut Czochralski n
-type monocrystalline Si (100-
oriented, double
-side polished, with a
dopant density of 10
17
cm
-3
) wafers with thickness of 200 μm were used for fabrication of the
HTJ
-Si cells. First, Si micropyramids were fabricated via an anisotropic etching process
at 85 °C
for 20 min using an aqueous solution of KOH and isopropyl alcohol in water with
a volume ratio
of 1:1:17
. Then, plasma
-enhanced chemical
-vapor deposition (PECVD) of an intrinsic
a
- Si:H
buffer lay
er (5 nm) with a p
-type
a
-Si:H layer (6 nm) on top as emitter layers was performed at
150 ̊C, and an intrinsic
a
-
Si:H buffer layer (5 nm) with an n-
type
a
-Si:H layer (10 nm) was
deposited on the back, to serve as surface
-field layers, under a typical growt
h rate of 1.5 Å
s
-1
.
Samples were stored in N
2
(g) before etching in buffered HF(aq) to remove the surface native
oxide, rinsed with deionized water, and dried using N
2
(g). Samples were then loaded into the
sputtering chamber immediately for deposition of N
iO
x
.
CdTe pieces were cut from n
-CdTe (111-
oriented, double
-side polished, with a dopant
density of 5.5
×
10
17
cm
-3
) wafers. Pieces of the CdTe were etched for 30 s in a freshly prepared
solution of Br
2
dissolved in CH
3
OH (0.0
4% v/v). The samples were then rinsed vigorously with
CH
3
OH, and immersed in CH
3
OH until being dried using N
2
(g) before being placed in the
sputtering chamber for deposition of NiO
x
.
The
a
-
Si:H films were prepared by RF decomposition of silane, and were all prepared at
HT Products by H. Wiesmann. The In
-doped SnO
2
(ITO) substrate was maintained at 220 ̊C with
a deposition rate of 1.5 Å s
-1
. A 200 Å thick n-
doped layer was deposited from silane with 1%
phosphine, and was then followed by a nominally intrinsic 0.5-
0.6 μm layer of
a
-
Si:H(1, 2)
.
a
-
Si:H samples were cleaned in buffered HF(aq) until the surface was hydrophobic, typically
3
requiring ~
30-
45 s. The samples were then immersed in a mixture of 1:1:5 (v/v)
H
2
O
2
:HCl:H
2
O
for 30 min at 70 °C. The samples were then
thoroughly rinsed with H
2
O and dried using N
2
(g)
before being transferred
into the sputtering chamber for NiO
x
deposition.
GaAs pieces were cut from n
+
-GaAs (100
-oriented, carrier concentration 10
18
cm
-3
,
single
-side polished) wafers obtained from AXT, Inc
. The pieces were etched for 30 s in a freshly
prepared solution of 0.04%
(v/v)
Br
2
/CH
3
OH solution and were
thoroughly rinsed with CH
3
OH,
then etched in 1.0 M NaOH(aq) for 15 s, rinsed with deionized water, and dried using
N
2
(g)
before being placed in a sp
uttering chamber for deposition of NiO
x
.
NiO
x
sputtering deposition:
Reactive RF sputtering using an AJA high
-vacuum magnetron sputtering system (AJA
International Inc.) was conducted in a chamber with a maximum base pressure of 8×10
-8
Torr.
The O
2
concentration
was varied from 0 to 0.33 with a constant Ar flow of 20 standard cubic
centimeter per minute (
sccm
) while the working pressure was held at 5 mTorr. The substrates
were maintained at 300 °C. The deposition rate was maintained at 0.2 Å s
-1
by tuning the
sputtering power on the Ni target (Kurt Lesker, 2” diameter × 0.125” thickness, 99.95%). To
deposit metallic Ni, a constant RF power of 150 W was used without substrates
heating or
O
2
gas
supply.
Text S2:
Characterization of Samples
X-ray p
hotoelectron spectroscopy (XPS)
4
XPS data were obtained using an AXIS Ultra DLD instrument (Kratos Analytical,
Manchester, UK) at a background pressure of 1 × 10
-9
Torr. High
-intensity excitation was
provided by monochromatic Al Kα X
-rays, 1486.6 eV in ener
gy and 0.2-
eV resolution at full
width at half maximum. For the Fe
2p high resolution scans, Mg Ka X
-rays of 1253.7 eV were
used instead in order to minimize the overlapping of Ni Auger peaks with the Fe 2p peaks.
Photoelectrons were collected at 0º from t
he surface normal at a retarding (pass) energy of 80 eV
for the survey scans, whereas a pass energy of 20 eV was used for the high
-resolution scans. The
peak energies were calibrated against the binding energy
, E
B
, of the adventitious C 1s peak (taken
to be 284.8 eV).
X-ray diffraction spectroscopy
XRD analysis was conducted using a Bruker D8 Discover system equipped with a 2-
dimen
sional Vantec
-500 detector. Cu-
K
α
radiation (1.54 Å) was generated at a tube voltage of 1
kV and a tube current of 50 mA. The i
ncident beam was focused using a mono
-capillary
collimator. A laser beam mark
ed the focal spot on the specimen
that was fixed on a
n xyz stage.
The scattered diffraction was registered by a 2
-di men
sional detector with the
angular resolution
of the detector smaller than 0.04
°
, and enable
d the simultaneous detection of the diffraction data
in a 2
θ
range of 20
°
. The detected radiation was counted for 2000
s to obtain an appropriate XRD
profile.
The data were analyzed using Bruker EVA software.
UV
-vis reflectance and transmittance measurement:
The optical reflectance and transmittance of the NiO
x
- or Ni metal
-coated Si and HTJ
-Si
substrates were determined by using an integrating sphere at normal incidence (Agilent Cary
5000 UV
-Vis spectrometer). The absorptance
(
A
) of NiO
x
-coated Si was calculated from the
5
measured total reflectance (
R
) and transmittance (
T
) of the electrode as
A
= 1 –
R
–
T
. All optical
measurements were done in air on fresh samples without chemical/electrochemical treatment.
Prepar
ation of electrodes:
Ohmic contacts were formed by scribing an In
-Ga eutectic alloy (Alfa Aesar, 99.
99%) to
the back side of the HTJ
-Si cell and to the p
+
-Si substrates. To form an ohmic contact to n-
CdTe
and n
+
-GaAs, In was soldered onto the back side of
the sample. Ohmic contact for
a
- Si was made
by scribing In
-Ga across the
a
- Si:H to the bottom In
-doped Sn oxide
(ITO)
substrate. High
-purity
Ag paint (SPI supplies) was then used to mechanically attach the ohmic contact to a coiled, tin
-
plated Cu wire (Mc
Master
-Carr) which was then threaded through a glass tube (Corning
Incorporated, Pyrex tubing, 7740 glass). The sample was encapsulated and sealed to the glass
tube using a mixture of 2:1 grey epoxy (Hysol 9460F) and white epoxy (Hysol 1C). The epoxy
was
allowed to dry under ambient conditions for at least 12 h. A high-
resolution optical scanner
(Epson Perfection V370 with a resolution of 2400 psi) was used to image the exposed surface
area of each electrode, and the geometric areas were determined by an
alyzing the images using
ImageJ software. All of the electrodes in this study were 0.1-
0.2 cm
2
in area, unless specified
otherwise.
Prepara
tion of HTJ
-Si|ITO|Co
-Pi photoanode
and FTO|Co
-Pi dark electrode
:
Indium doped tin oxide (ITO)
with a thickness of
90 nm w
as deposited on
the front and
back of the HTJ
-Si using sputtering. Front and back contacts were made by using Ag paste to
attach a Cu wire to the front and back ITO layers, respectively. Wire contacts and epoxy were
made on the edges on the front co
ntacts, while the entire back contacts were isolated by epoxy.
The Co
-phosphate precursor solution was 0.1 M potassium phosphate (
K-Pi, a mixture of 61.5%
6
0.1 M dibasic K
2
HPO
4
and 38.5%
0.1 M monobasic KH
2
PO
4
) and 0.5 mM cobalt nitrate.
Electrodeposition w
as pe
rformed in a two
-compartment cell with the front ITO on the HTJ
-Si as
the working electrode and a fritted Pt mesh
as a counter electrode, with
a saturated calomel
electrode (SCE,
0.654 V vs. RHE)
as the reference electrode. Deposition of the Co cataly
st
precursor in various thicknesses was performed with agitation under a constant current density of
1 mA cm
-2
by passing various amounts of charge
(30 to 1800 mC cm
-2
). Samples were then rinsed
with DI
-H
2
O and dried under a stream of N
2
(g). The transmittance
of
Co
-Pi coated FTO
substrates w
as measured
prior to the electrochemical measurement
. The samples
were then
transferred to
0.1 M K
-Pi(aq) and then
to
1.0 M KOH(aq) for electrochemical
measurement
s. The
photoelectrochemical measurement of an
HTJ
-Si|ITO
|CoO
x
electrode was conducted through the
back contact
under simulated 1-
Sun illumination
in 1.0 M KOH(aq)
. Detailed electrochemical
measurement setup can be found in the section below
. The stability data for this system were
obtained using the s
ame protocol as for all of the NiO
x
-protected photoanodes evaluated herein
described in the section below
.
Electrochemical measurements:
A Mercury/Mercury oxide (Hg/HgO in 1.0 M KOH(aq
), CH Instruments, CH152)
electrode was used as the reference electrode, and a carbon cloth placed within a fritted glass tube
(gas dispersion tube Pro
-D, Aceglass, Inc.) was used as the counter electrode for all
electrochemical measurements performed in 1
.0 M KOH(aq) electrolyte, including
photoelectrochemical, spectral response, and faradaic efficiency measurements. The Hg/HgO
reference electrode was calibrated versus the reversible hydrogen electrode and the Hg/HgO
electrode potential was determined to b
e 0.926 V vs. RHE. The equilibrium potential for the
oxygen
-evolution reaction was therefore 0.304 V vs the Hg/HgO reference. A custom
electrochemical cell with a flat glass (Pyrex) bottom was used for all of the electrochemical
7
measurements. During measur
ements, the electrolyte was vigorously agitated with a magnetic stir
bar driven by a model
-train motor (Pittman) with a Railpower 1370 speed controller (Model
Rectifier Corporation). The data presented for photoelectrochemical measurements in aqueous
solut
ions do not include compensation for the series resistance of the solution.
To evaluate the
catalytic activity, correction for the solution resistance was applied. ELH
-type (Sylvania/Osram)
and ENH
-type (EIKO) tungsten-
halogen lamps with a custom housing a
nd with a transformer
(Staco Energy Products Co.) were used for long
-term photoelectrochemical stability
measurements. A Xe arc lamp (Newport 67005 and 69911) equipped with an IR filter (Newport
61945) and with an AM 1.5 filter (Newport 81094 and 71260) wa
s used as the light source for
J
-
E
measurements and for the spectral response measurements. The illumination intensity at the
position of the working electrode in the electrochemical cell was determined by placing a
calibrated Si photodiode (FDS100-
Cal, Th
or Labs) into the cell at the same position occupied by
the exposed area of the photoelectrode. To illuminate bottom
-facing photoelectrodes, a quartz
diffuser (Newport 15Diff
-Vis) together with a broadband reflection mirror (Newport dielectric
mirror, 10Q20PR
-HR) was used to direct the uniform light beam from the horizontal to the
vertical direction.
Cyclic voltammetry as well as quantum yield data were
obtained using a Biologic SP
-200
potentiostat (Bio
-Logic Science Instrument). The cyclic voltammetric data were recorded at a
constant scan rate of 40 mV s
-1
with a scan range that varied depending on the photovoltage of the
sample. The external quantum yield was collected using the potentiostat connected to a lock
-in
amplifier with the light chopped at
20 Hz.
Photoanode water
-oxidation s
tability
-measurement
protocols
:
Stability measurements were performed in 1.0 M KOH(aq) under simulated 1-
Sun
illumination using a
n ENH-
type tungsten halogen lamp with a dichroic rear reflector. The
8
electrochemical cell w
as continually cooled by flowing compressed air on the sidewall of the cell.
The electrolyte in the cell,
as well as in the Hg/HgO reference electrode, was changed every 24
-
48 h to prevent concentrating species within
the cell solution and to prevent syste
m over
-loading
caused by the presence of gas
bubbles in the reference electrodes, as well as to maintain a stable
reference potential. The reference electrode was calibrated during the course of the measurement
using the method described above. The light intensity at the sample position was also calibrated
whenever the electrolyte was replaced. A transformer was used to lower the voltage supply on the
bulb to
40-
50 V
ac
to improve the lifetime of the bulb and to reduce the generation of heat.
The
light bulb
was changed every 200-
300 h to prevent unexpected bulb burn-
out.
The stability of the
light source was also monitored by placing a Si reference photodiode at a fixed position relative
to the lamp. A
n automatic data-
collection protocol was implemented usin
g the Biologic
potentiostat such that 10 cyclic voltammetric scans were collected after every 10 h of
chronoamperometric stability measurements. To minimize the change in conditions of the
electrode when switching between chronomperometric and cyclic volta
mmetric measurements,
the cyclic voltammteric measurement was set to use an initial and a final potential of 1.73 V vs.
RHE, which was the potential used for
chronoamperometry
. The potential was then scanned
between
E
low
and
E
high
. where E
high
was fixed at 1.93 V vs. RHE and
E
low
was set depending on
the photovoltage of the photoelectrodes, with
E
low
= 0.
63 V vs. RHE for HTJ
-Si|NiO
x
and
E
low
=
0.83 V vs. RHE for CdTe|NiO
x
and a
-Si:H|NiO
x
photoelectrodes.
Measurement of Faradaic efficiency:
A Neofox
fluorescence probe (Foxy probe, Ocean Optics) was used to monitor the
concentration of O
2
throughout the experiment. The fluorescence response was calibrated against
the standard concentration of oxygen in water (7700 μg L
-1
or 2.4×10
-4
M) under a standar
d
atmosphere that contained 20.9% (by volume) O
2
(g). The fluorescence probe, the Hg/HgO/1.0 M
9
KOH reference electrode, a fritted Pt mesh counter electrode (Alfa-
Aesar, 100 mesh, 99.9% trace
metal basis), and the NiO
x
-coated semiconductor working electrodes
with a geometric surface
area of 0.84-
0.87 cm
2
(HTJ
-Si|NiO
x
), 0.12
-0.22 cm
2
(
a
-Si:H|NiO
x
), and 0.07 cm
2
(n
-CdTe|NiO
x
)
were loaded into an airtight glass cell that had a volume of 43.6 mL with no headspace, and that
was equipped with four ports and a side
-facing quartz window. The cell and the 1.0 M KOH(aq)
electrolyte in the cell were purged with a stream of ultra-
high purity Ar(g) for ~1 h prior to the
measurement of the O
2
produced by the electrode maintained in the water bath to minimize the
temperature
fluctuation during illumination. The current density was maintained at 1 mA cm
-2
or
lower, to prevent bubble formation on the electrode surfaces. The fluorescence probe was kept
in
the dark during the measurement. To compare the charge
-versus-
time data fr
om the potentiostat
with the amount of oxygen generated versus time for a system operating at 100% faradaic
efficiency, the charge passed (in mA∙h) was multiplied by 3.6 to convert the data into coulombs,
and the result was then multiplied by 83 (the facto
r for conversion of 1 C of electrons to 1 μg of
O
2
) to convert the value into micrograms of O
2
.
Spectroscopic ellipsometry and data fitting:
Thin films of Ni oxide backed by Si were investigated by use of a variable-
angle
spectroscopic ellipsometer with
a rotating analyzer and equipped with an autoretarder (J.A.
Woollam Co., Inc.). Measurements were performed with 0.05 eV steps in the 0.5-
4.25 eV spectral
range and for incidence angles between 65° and 85°.
The thickness of the film was modeled by fitting the data using a Cauchy layer on top of
a native oxide coated Si surface in a non
-absorbing region, typically at a photon energy > 1.1 eV.
A point
-by-
point fit was then used for initial determination of the optical properties of the film.
Once a good fit w
as obtained (low MSE value), the Cauchy model was then converted to a
general oscillator (Genosc) model to ensure the correctness of the fit using physical oscillators
10
and to
enforce the Kramers
–Kronig relation, which connects the real and imaginary parts of the
index of refraction. In the Genosc model, a series of the Lorentz and Drude oscillators is used to
model the NiO
x
layer. The Lorenz model incorporates interband absorptions with band tails, and
the Drude
model is especially useful in modeling the intraband absorption of the free carriers near
the infrared region. The fitted refractive indices were physically Kramers-
Kronig (K
-K)
-
consistent
(3)
.
Index grading, anisotropy, and thickness non-
uniformity w
ere added to the Genosc
model. Grading in both the refractive index and extinction coefficient was implemented by
mo
deling the layer assumed to have a grading consisting of multiple thin layers of the same
material, each with optical properties that varied, to mimic grading due to the preparation method,
where nucleation is typically involved. To improve the fitting ac
curacy, surface roughness was
modeled by adding an effective medium approximation (EMA) layer of the material with 50%
voids.
The mean
-squared error (MSE) between the model and the experimental data was low (<
6) for multiple samples with different thickne
sses of NiO
x
, and consideration of intermixing of
SiO
x
and NiO
x
at the interfaces was therefore not required to obtain good fits to the data. Surface
-
adsorbed H
2
O or OH groups as suggested by the Ni(III) peaks in the XPS spectra (Figure S2a)
typically resu
lt in a lower refractive index of
the existing surface layer than that of bulk NiO
x
(4)
.
This complicated
thin surface layer was not needed explicitly in the modeling to obtain good fits
to the data.
Total reflectance calculation
s:
The total reflectances of the semiconductor|NiO
x
samples versus wavelength, with
different NiO
x
thickness varied from 0
-300 nm with air or 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.
11
Orfanidis from Rutgers University, was used to calculate the reflection responses of the isotropic
non-
lossy
multilayer dielectric structures (5)
. Optical constants for water, crystalline Si,
crystalline CdTe, and amorphous Si were adopted from previously reported values. The refractive
indices for sputtered NiO
x
were measured usin
g spectroscopic ellipsometry described in the
section above
. The optical constant from the middle layer of the NiO
x
was used for the
calculation, averaging the graded index in the z
-direction.
Surface recombination velocity:
Surface recombination
velocity (
S
) measurements were made using a contactless
microwave
-conductivity apparatus that has been described previously (6, 7)
. A 20
-ns pulsed-
diode laser with a wavelength of 905 nm (OSRAM
laser diode with an ETX
-10A
-93 driver) was
used to generate electron
-hole pairs in HTJ
-Si. The lifetime of the photogenerated charge carriers
was monitored using a PIN diode that detected the microwave radiation reflected from the
sample. The HTJ
-Si cell before and after NiO
x
coating was measured. Samples were stored in N
2
glove box between measurements. Charge
-carrier lifetimes and surface recombination velocities
were calculated based on previous reports
(6, 7)
.
12
Supporting figures:
Figure S1.
Tafel plot
for
samples prepared from different oxygen concentration
s without
substrate heating
after resistance correction
.
Figure S2c shows the crystallinity of the sputter
-deposited film. When
the oxygen
concentration was 1%, the NiO
x
film showed a comparable peak intensity in the (111), (200) and
(220) directions, while films prepared with an oxygen concentration
≥
4.8% showed a dominant
peak in the (111) direction. Further increases in the tem
perature during sputter deposition did not
change the preferred orientation, but produced a higher crystallinity in the resulting films. When
the oxygen concentration was 1%, no metallic Ni peaks were observable at 2
θ
angles highlighted
by the gray markers
in Figure S
2c, suggesting the formation of completely oxidized films under
such conditions, consistent with the XPS observations (Figure S2a).
13
Figure S2
.
XPS data on p
+
-Si|NiO
x
samples prepared from different oxygen concentrations
during sputter deposi
tion (a), as well as on fresh and activated p
+
-Si|NiO
x
samples prepared with
4.8% O
2
concentration with and without substrate heating (b). For comparison, spectra are also
displayed for 2 nm of sputtered Ni metal. Spectral regions containing the Ni 2p
3/2
peaks (with
Ni(II) at 853.86 eV (green area), Ni(III) at 855.69 eV (orange area), and metallic Ni at 852.55 eV
(blue area)), and the O 1s peaks, are shown for NiO
x
films deposited with oxygen concentrations
that were varied between 0% (black curve) and 33.3% (magenta curve). (c) XRD data for NiO
x
-
coated p
+
-Si (111) substrates prepared with different oxygen concentrations and substrate heating
temperatures. The gray arrows indicate the diffraction
-angle positions of crystalline Ni metal.
14
Figure S
3.
XPS survey scan and high resolution scan at Fe 2p region on Ni metal film coated
non-
photoactive p
+
-Si before and after KOH treatment.
The refractive index in the region of 350–800 nm is 3.7–5.7 for Si, 4.1–5.2 for
a
-
Si, and
3.0–3.4 for CdTe (Figure S4a)
. The minima of the total reflectance as well as the total reflection
on these three substrates showed a trend of CdTe|NiO
x
> c
-Si|NiO
x
>
a
-Si|NiO
x
, consistent with
the refractive index trend of CdTe < c
-Si <
a
- Si for wavelengths > 400 nm. Tuning the physic
al
thickness (
t
), and thus the optical thickness (
nt
), of the film affects the interference of light at
different wavelengths (
λ
) and therefore affects the reflectance of the sample. Thin films (~
3.5
nm) of NiO
x
produced almost no change in the reflectanc
e of polished crystalline Si (Figure S
4d).
However, increases up to 75 nm in the thickness of the NiO
x
coating produced a pronounced
reduction in the reflectance of the structure. Light interference became significant when the
thickness of the NiO
x
film was increased to 150 nm, with the reflection minima shifted to longer
wavelengths.
The reflectance of the aforementioned bare and NiO
x
-coated semiconductors should be
reduced in water compared to air, because the optimum
n
l
needed for minimization of
reflection
15
at a wavelength corresponding to its optical thickness (4
nt
) increases from 2 to 2.28 when the
medium is switched from air (n=1) to water (n=1.3) (
,where
n
l
,
n
i
and
n
s
are
refractive indices of the antireflection layer, the incident media, and
the substrate, respectively).
This increases the optimum
n
l
and gives better index matching in water (Figure S
4e shows
reflectance in air and water using Si as an example). The dependence of the reflectance at the
air/semiconductor interface on the angl
e of incidence of the illumination was also broadly
suppressed by 75-
nm thick NiO
x
coatings (Figure S
4f), especially in the “reflectionless” region
(~600 nm)
.
In the present study changes in the refractive index and extinction coefficient were not
monitore
d in situ under operating conditions
(8)
. However, based on the complex refractive
indices reported for Ni(OH)
2
(1.38-
0.06i) and NiOOH (1.41-
0.07i)
(4)
, and assuming that Ni
atoms only within the top 10 nm of NiO
x
were converted to catalytically active sites with a 50%
volume expansion
(9)
, the reflectance in the shorter
-wavelength (<
600 nm) region will be
suppressed (Figure S4g).
16
Figure S
4.
Calculated total reflectance on semiconductors with and without NiO
x
films in
different incident media. (a) Refractive indices of semiconductors including crystalline Si (black),
crystalline CdTe (red), and
a
- Si (blue). (b) Reflectance from air|semiconductor interfaces based
on the refractive index data of (a). (c) Reflect
ance from air|NiO
x
|semiconductor interfaces. (d)
Reflectance from air|NiO
x
|Si interfaces with varied thickness of NiO
x
. (e) Reflectance showing
the effect of the incident media (water versus air) from media|NiO
x
|Si
and media|Si
interfaces. (f)
Reflectance
showing the incident
-angle effect at air|Si and air|NiO
x
|Si interfaces. (g) Reflectance
data
showing the effect of a layer of Ni(OH)
2
/NiOOH electrochemically converted from NiO
x
.
17
The r
eflectance in panels
b- e, g were obtained at normal incidence.
The thic
knesses of NiO
x
layer
in panels c, and e
-g were
75 nm
.
Figure S
5.
(a)
J
-
E
data for electrocatalytic water oxidation in 1.0 M KOH(aq), and (b)
J
-
E
data in
0.35 M Fe(CN)
6
3-
/0.05
0 M Fe(CN)
6
4-
redox couple in 1.0 M KCl(aq), as a function of the
thickness (10
-150 nm) of the NiO
x
film on p
+
-Si substrates. (c)
J
-
E
data of the electrocatalytic
activity of the NiO
x
films in 1.0 M KOH(aq), before and after running at a constant current
density of 30 mA cm
-2
for 100 h, for films with thicknesses of 10 nm and 75 nm.
18
Figure S
6.
(a) Absor
ptance and (b)
J
-
E
behavior of a
Ni metal coated HTJ
-Si photoelectrode
.
Figure S
7.
Representative time
-resolved microwave conductivity
(TRMC)
decay signals for bare
HTJ
-Si and NiO
x
coated HTJ
-Si substrates.
19
Figure S
8.
J
-E
data for HTJ
-Si (a), CdTe (b) and
a
- Si:H (c) photoelectrodes without a NiO
x
coating. (d) Chronoamperometry at 1.73 V vs. RHE of HTJ
-Si, CdTe and
a
-
Si:H photoelectrodes
without a NiO
x
coating
. All of the
measurement
s were performed in 1.0 M KOH(aq) under 100
mW cm
-2
of simulated solar illumination.
Co
-Pi (phosphate
-containing CoO
x
) has been used as a heterogeneous OER catalyst on
various semiconductors
(10
-16)
. On FTO|Co
-Pi (60 mC cm
-2
) electrodes in 0.1 M K
-Pi(aq), an
overpotential of ~465 mV was observed at a current density of 1 mA cm
-2
for the OER in the
dark, consistent with the reported values (17)
. In alkaline media, the Co
-Pi converted to the more
active, but porous, known CoO
x
OER electrocatalyst (Figure S
9f). Consistently, the overpotential
20
at 1 mA cm
-2
and 10 mA cm
-2
decreased from 465 and 835 mV to 330 and 420 mV, respectively,
when the 0.1 M phosphate buffered electrolyte, K
-Pi(aq), was replaced by 1.0 M KOH(aq)
(Figure S
9a). In both electrolytes the observed overpotentials at 10 mA cm
-2
were larger than the
~330 mV value reported herein for the NiO
x
films in 1.0 M KOH(aq). At either pH, increases in
the loading of the Co-
Pi to obtain higher catalytic activity and thus improved fill factors in
integrated photoelectrode structures resulted in an increase in light at
tenuation (Figure S
9c and d),
a significant reduction in the photocurrent
(15)
, as well as deleterious increases in the resistivity
(transport), and electrochromic coloration of the catalyst film. Moreover, Co
-Pi coatings cannot
be directly incorporated onto Si surfaces due t
o severe oxidation of the Si at the Co-
Pi/Si interface
as a result of the anodic catalyst
-deposition process
(16)
.
Figure S
9.
(a)
J
-E
behavior
of FTO|Co
-Pi in 0.1 M K
-Pi(aq) and FTO|CoO
x
in 1 M KOH(aq)
at a
scan rate of 10 mV s
-1
. The
Co
-Pi film was deposited by passing 60 mC cm
-2
of charge.
(b)
Optical transmittance
spectra of FTO glasses coated with different Co
-Pi loadings (30-
1800 mC
cm
-2
). (c) Optical images of Co
-Pi deposited on FTO substrates by passing different amount of
charges.
(d) The stability of a HTJ
-Si|ITO|Co
O
x
(60 mC cm
-2
) photoelectrode that
showed the
21
minimum optical loss
, measured by contact
through the back in the presence of
1- Sun of front
-
side illumination in 1.0 M
KOH(aq). (
e)
J
-
E
behavior
showing the
continual
loss of photovoltage
and a decrease of photocurrent
for operat
ion of a HTJ
-Si|ITO|Co
O
x
(60 mC cm
-2
) photoanode
under these conditions.
(f) XPS data on an
as-
deposited Co-
Pi (black curves), on
a Co
-Pi film
immersed
in 1.0 M KOH (aq) for 1 h (red curves)
, and
on a Co
-Pi film
operated at
an anodic
current density of 5 m
A cm
-2
for 10 min in 1.0 M KOH(aq)
(blue curves)
. All three
Co
-Pi films
were deposited on FTO glass by passing 60 mC cm
-2
of charge
, and were then independently
processed and
investigated
spectroscopically
. Spectral regions containing the P 2p peaks, Co 2p
peaks and O 1s peaks are shown.
Figure S
10.
SEM images of (a) Si(100)|NiO
x
immersed in 10.0
M KOH(aq) for 240 h at open
circuit. (b)
J-E
data for a
n n
+
-GaAs|NiO
x
electrode showing an increasing current due to
corrosion vs time. The inset depicts an SEM
image obtained after anodic operation in 1.0 M
KOH(aq) for 5 h, showing surface damage from the corrosion/dissolution of GaAs and peeling of
the NiO
x
layer.
The thicknesses of the NiO
x
layer in both cases were 75 nm.
22
Table S1.
Analysis of the catalytic a
ctivity of NiO
x
films prepared from different oxygen
concentration
s without substrate heating
during sputtering. The data were corrected for solution
resistance.
Oxygen concentration
Tafel slope (1
-
10 mA cm
-
2
)
Overpotential (mV) at 10 mA cm
-
2
0%
50
315
1%
40
335
4.8%
41
335
20%
171
517
33%
192
563
References:
1.
Kumar A & Lewis NS (1990) Short
-wavelength spectral response properties of
semiconductor/liquid junctions.
J. Phys. Chem.
94(15):6002-
6009.
2.
Gronet CM
, et al.
(1984) Correlation of the Photoelectrochemistry of the Amorphous
Hydrogenated Silicon/Methanol Interface with Bulk Semiconductor Properties.
J.
Electrochem. Soc.
131(12):2873-
2880.
3.
Tompkins HG & McGahan WA (1999)
Spectroscopic Ellipsometry and Reflectometry: A
User's Guide
(Wiley).
4.
Wronkowska AA (1989) An ellipsometric investigation of nickel oxide films in alkaline
solution: The multilayer film approach.
Surf. Sci.
214(3):507-
522.
5.
Orfanidis SJ (2014)
Electromagnetic waves and antennas
.
23
6.
Plass K
E, Liu X, Brunschwig BS, & Lewis NS (2008) Passivation and Secondary
Functionalization of Allyl
-Terminated Si(111) Surfaces.
Chemistry of Materials
20(6):2228-
2233.
7.
O’Leary LE
, et al.
(2013) Heck Coupling of Olefins to Mixed Methyl/Thienyl
Monolayers on
Si(111) Surfaces.
J. Am. Chem. Soc.
135(27):10081-
10090.
8.
Gottesfeld S & Srinivasan S (1978) Electrochemical and optical studies of thick oxide
layers on iridium and their electrocatalytic activities for the oxygen evolution reaction.
J.
Electroanal. Ch
em. Interfacial Electrochem.
86(1):89-
104.
9.
Sun K
, et al.
(2015) Stable Solar
-Driven Water Oxidation to O
2
(g) by Ni
-oxide Coated
Silicon Photoanodes
J. Phys. Chem. Lett.
6:592
-598.
10.
Cong Y
, et al.
(2012) Synthesis of Ta
3
N
5
Nanotube Arrays Modified with
Electrocatalysts for Photoelectrochemical Water Oxidation.
J. Phys. Chem. C
116(27):14541-
14550.
11.
Jeon TH, Choi W, & Park H (2011) Cobalt
-phosphate complexes catalyze the
photoelectrochemical water oxidation of BiVO
4
elect
rodes.
PCCP
13(48):21392-
21401.
12.
Barroso M
, et al.
(2011) The Role of Cobalt Phosphate in Enhancing the Photocatalytic
Activity of α
-Fe
2
O
3
toward Water Oxidation.
J. Am. Chem. Soc.
133(38):14868-
14871.
13.
Steinmiller EMP & Choi K
-S (2009) Photochemical
deposition of cobalt
-based oxygen
evolving catalyst on a semiconductor photoanode for solar oxygen production.
Proc.
Natl. Acad. Sci. USA
106(49):20633-
20636.
14.
Pijpers JJH, Winkler MT, Surendranath Y, Buonassisi T, & Nocera DG (2011) Light
-
induced wate
r oxidation at silicon electrodes functionalized with a cobalt oxygen
-
evolving catalyst.
Proc. Natl. Acad. Sci. USA
108(25):10056-
10061.
15.
Reece SY
, et al.
(2011) Wireless Solar Water Splitting Using Silicon
-Based
Semiconductors and Earth-
Abundant Cataly
sts.
Science
334(6056):645-
648.
24