1
Supporting Information for:
SolarDriven Reduction of 1 atm CO
2
to
Formate at 10% EnergyConversion Efficiency by
Use of a TiO
2
Protected IIIV Tandem Photoanode
in Conjunction with Bipolar Membrane and a Pd/C
Cathode Electrocatalyst
Xinghao Zhou
1,2
, Rui Liu
2
, Ke Sun
2,3
, Yikai Chen
2
, Erik Verlage
1,2
, Sonja A. Francis
2-4
, Nathan S.
Lewis
2,3,5,6*
, Chengxiang Xiang
2,3*
1
Division of Engineering and Applied Science, Depar
tment of Applied Physics and Materials Science,
California Institute of Technology, Pasadena, CA 91
125, USA
2
Joint Center for Artificial Photosynthesis, Califo
rnia Institute of Technology, Pasadena, CA 91125,
USA
3
Division of Chemistry and Chemical Engineering, Ca
lifornia Institute of Technology, Pasadena, CA
91125, USA
4
Resnick Sustainability Institute, California Insti
tute of Technology, Pasadena, CA 91125, USA
2
5
Beckman Institute and Molecular Materials Research
Center, California Institute of Technology,
Pasadena, CA 91125, USA
6
Kavli Nanoscience Institute, California Institute
of Technology, Pasadena, CA 91125, USA
*To whom correspondence should be addressed: nslewi
s@caltech.edu, cxx@caltech.edu
3
Chemicals and materials
All materials were used as received unless noted ot
herwise: potassium hydroxide pellets
(KOH, Macron Chemicals, ACS 88%), 10 wt.% Pd on Vul
can XC72 carbon black (Pd/C,
Premetek Co.), titanium wire (Ti wire, Alfa Aesar,
99.7%), CO
2
gas (ALPHAGAZ 1), and 10 wt.%
Nafion (SigmaAldrich). Water with a resistivity >
18 MGcm was obtained from a Barnsted
Nanopure deionized (DI) water system. The KHCO
3
solution was prepared by vigorously
bubbling CO
2
(g) through K
2
CO
3
solutions until the pH did not change. The bipola
r membrane
(fumasep® FBM) was obtained from FuMATech GmbH (St
. Ingbert, Germany) and was stored
in 1.0 M NaCl(aq) at room temperature. The bipolar
membrane (BPM) was cut into 1.5×1.5 cm
pieces and rinsed with H
2
O before use. Nafion® PFSA 117 membrane (Chemours)
with a
thickness of 183 μm was obtained from Dupont. The
Nafion was soaked in H
2
O for > 4 h and
rinsed with H
2
O before use.
Growth of IIIV dual junction photoabsorbers
The tandemjunction IIIV device was grow
n commercially (Sumika Electronic Materials,
Inc.) according to specifications determined by 1D
numerical simulation using Helmholtz
Zentrum Berlin’s AFORSHET software. Planar IIIV
layers were grown epitaxially by metal
organic chemicalvapor deposition (MOCVD) on an n
+
GaAs wafer that had a (100)oriented
polished surface (Sidoped, acceptor concentration
of 1×10
19
cm
3
, 6” diameter). Detailed
information on the cell stack, including the thickn
ess and the dopants for the IIIV layers, has
been provided previously.
1
Atomiclayer deposition of TiO
2
layer
Atomiclayer deposition (ALD) of TiO
2
was conducted at 150 °C using
tetrakis(dimethylamido)titanium (TDMAT, SigmaAldri
ch, 99.999%, used as received) and H
2
O
(>18.2 MGcm resistivity, Millipore) in an Ultratec
h Fiji 200 plasma atomiclayer deposition
system. Prior to ALD, the epitaxial surface was im
mersed in 1.0 M KOH(aq) (aqueous solution
4
of potassium hydroxide pellets, semiconductor grade
, 99.99% trace metal basis, SigmaAldrich)
for 30 s, rinsed with copious amounts of H
2
O, dried using a stream of N
2
(g), and loaded
immediately into the ALD chamber. An ALD cycle con
sisted of a 0.06 s pulse of H
2
O, a 15 s
purge under a constant 0.13 L min
1
flow of researchgrade Ar(g), a 0.25 s pulse of TD
MAT, and
another 15 s Ar(g) purge. ALDTiO
2
films with thickness of 62.5 nm were used to prote
ct IIIV
surfaces.
Ohmic contacts and deposition of Ni
Ohmic contact to the n
+
GaAs wafer was formed using a GeAu eutectic (20 n
m Ge/ 30
nm Au/ 15 nm Ni/ 100 nm Au) deposited by radiofreq
uency (RF) sputtering. The contact layers
were annealed under N
2
(g) at 400 °C for 30 s using rapid thermal annealin
g (RTA) with a ramp
up rate of 40 °C s
1
. For all photoanode surfaces, an optically transp
arent nominally 2 nm thick
Ni film was deposited on TiO
2
via RF sputtering from a Ni target (Kurt Lesker, 2
’’ diameter ×
0.125’’ thickness, 99.95%) using an AJA highvacuum
magnetron sputtering system. The Ar
flow was kept at 10 sccm, while the working pressur
e was held at 5 mTorr. A deposition rate of
~0.1 Å s
1
was maintained by adjusting the sputtering power o
n the Ni target.
ALDdeposited amorphous TiO
2
coatings have been used on various single crystall
ine
substrates, including Si, IIIV and IIVI compound
semiconductors previously, and have showed
excellent ohmic behavior due to similar growth chem
istry. Degenerately doped Si substrates
with an acceptor concentration > 10
19
cm
3
p
+
Si were used in this study as the dark electrode
control sample, to be consistent with the actual sa
mple that had a window layer consisting of a
30nm p
+
In
0.48
Al
0.57
P and a 7nm thick p
+
GaAs with acceptor concentrations > 10
19
cm
3
. The
tandem IIIV ntype photoelectrode was rectifying i
n the dark and therefore could not be used as
a reliable reference for the behavior of the photoe
lectrode under illumination.
2
The JE behavior
of the p
+
Si/TiO
2
/Ni dark electrode effecting the OER in 1.0 M KOH(a
q) thus provided a
measure of the OER overpotential properties of the
Nicontaining catalyst on a relevant substrate.
5
Preparation of Pd/C electrodes
The preparation of electrodes comprising
Pd/C on a Ti mesh was similar to that reported
previously for Pd/C on a Ti foil electrode.
3
A Ti mesh was first cut into 0.04 cm
2
pieces. Five
pieces of Ti mesh were stacked and stringed togethe
r with a Ti wire. The stacked Ti mesh was
etched in boiling 10% oxalic acid, rinsed thoroughl
y with H
2
O, and dried overnight 75 °C in an
oven. 4.8 mg of Pd/C powder was then mixed with 2.
4 ml isopropanol and 40 μL 10 wt.%
Nafion and the mixture was sonicated for > 30 min.
The resulting solution was dropdried onto
the Ti mesh, with a Pd mass loading of 250 μg cm
2
. The Pd mass loading of Pd/C coated Ti foil
reported previously was 50 μg cm
2
. As a comparison, a Pd/C nanoparticlecoated Ti f
oil
electrode was made using the same conditions as on
the Ti mesh except that the Pd mass loading
was 50 μg cm
2
.
Preparation of IIIV dual junction photoanodes
The ohmically contacted tandemjunction wa
fers were cleaved into samples that were ~0.1
cm
2
area. Ag paste was then used to attach the ohmic
contact to a coiled, tinplated Cu wire
(McMasterCarr) which was then threaded through a g
lass tube (Corning Inc., Pyrex tubing,
7740 glass). The samples were encapsulated and sea
led to the glass tube using grey epoxy
(Hysol 9460F). The epoxy was allowed to dry under a
mbient pressure for > 12 h. The exposed
electrode surfaces were imaged with a highresoluti
on optical scanner (Epson perfection V370
with a resolution of 2400 dpi) and the areas were m
easured using ImageJ software. Electrodes
were ~ 0.03 cm
2
in photoactive area unless specified otherwise.
3electrode measurements
For electrochemical measurements in 2.8 M
KHCO
3
(aq), a Ag/AgCl/1.0 mol/kg KCl
electrode was used as the reference electrode, and
a Pt mesh was used as the counter electrode.
The Ag/AgCl/1.0 mol kg
1
KCl electrode had a potential of 0.235 V vs the no
rmal hydrogen
electrode (NHE). The pH of the CO
2
saturated 2.8 M KHCO
3
(aq) solution was 8.0, as measured
6
by a VWR SympHony SB70P Digital, Benchmodel pH Met
er. The equilibrium potential for the
CO
2
/HCO
2
redox couple at pH 8 is 0.02 V versus the reversib
le hydrogen electrode (RHE).
4
Therefore, the equilibrium potential for CO
2
reduction to formate in CO
2
saturated 2.8 M
KHCO
3
(aq) solution was calculated to be 0.687 versus Ag
/AgCl/1.0 mol kg
1
KCl.
Figure S2A shows the configuration of 3
electrode electrochemical measurement. Two
custommade threenecked cells were used in the 3e
lectrode measurement. The BPM was used
for prevention of formate product crossover, and t
o prevent the formate from being oxidized at
the counter electrode. The electrolyte was vigorou
sly agitated with a magnetic stir bar driven by
a modeltrain motor (Pittman) with a Railpower 1370
speed controller (Model Rectifier
Corporation). The data presented for electrochemic
al measurements in aqueous solutions do not
include compensation for the series resistance of t
he solution. CO
2
(g) was bubbled into the
KHCO
3
solution during electrochemical measurements, to p
revent oxygen dissolution. Cyclic
voltammetric data (10 mV s
1
scan rate) were obtained with a Biologic MPG244
potentiostat
(BioLogic Science Instrument).
2electrode measurements
The configuration for 2electrode measure
ments was similar to that used to obtain 3
electrode measurements (Figure S2B). In addition t
o the stirring described above, a peristaltic
pumping system (Simply Pumps PM300F) with a minimum
flow rate of 500 mL min
1
, as
controlled by a tunable power supply, was used to f
acilitate the removal of CO
2
bubbles at the
BPM surface, and to minimize the voltage loss at th
e BPM caused by bubbles. The BPM in this
configuration was replaced as required by a Nafion
membrane for additional 2electrode
measurements. For photoelectrochemical experiments
, a Xe arc lamp (Newport 67005 and
69911) equipped with an IR filter (Newport 61945) a
nd with an AM 1.5 filter (Newport 81094
and 71260) was used as the light source for
J
E
measurements. The intensity in the solution
containing cell was then calibrated by placing a Si
photodiode (FDS100Cal, Thorlabs) with a
similar area as that of the photoanode (0.03 cm
2
) in the custommade threenecked beaker with
7
flat quartz windows, with the Si located at nominal
ly the same position as that occupied by the
exposed area of the photoelectrode. The Si photodi
ode had been previously calibrated by
measurement of the shortcircuit current density un
der 100 mW cm
2
of AM 1.5 simulated
sunlight. The voltage scan range was confined to b
e more positive than the onset of photoanodic
current, to ensure a constant direction of ionic cu
rrent flow in the membrane. The areas for the
photoanode, BPM, Nafion membrane and Pd/C cathode w
ere 0.03 cm
2
, 0.03 cm
2
, 0.03 cm
2
, 0.04
cm
2
, respectively. A 0.2 cm diameter hole was punched
at the middle of a piece of rubber using
a hole making tool. An Oring with 0.2 cm diameter
was glued with epoxy on the rubber with
the hole aligned. The BPM was right between two pi
eces of rubber, and the hard rubber/O
ring/BPM/Oring/hard rubber was tightly clamped bet
ween the two cells as shown in Figure S2B.
Fourpoint measurement system for the bipolar membr
ane
The fourpoint measurement setup was simi
lar to that used for the 2electrode
measurement (Figure S2C). Two Luggin capillaries w
ith Ag/AgCl/1.0 mol kg
1
KCl reference
electrodes were used to measure the potential near
the membrane. A peristaltic pumping system
was also used in these experiments to facilitate th
e removal of CO
2
bubbles from the BPM
surface, and to minimize the voltage loss at the BP
M caused by bubbles and concentration
gradients. The BPM in this configuration could be
replaced by Nafion membrane as appropriate
for fourpoint measurements of the Nafion membrane.
The voltage loss at 10 mA cm
2
was
lower than previously measured,
5
due to the use of Luggin capillaries as well as an
efficient
pumping system.
ICPMS potassium ion crossover measurements
To measure the potassium ion crossover fr
om the anolyte to the catholyte through the
BPM, the KHCO
3
catholyte was replaced by CsHCO
3
(SigmaAldrich, 99.9% trace metals basis).
Two Pt mesh electrodes were used as the source and
drain electrodes during the potassium ion
crossover tests. The potassium ion concentrations
in the catholyte in a 2electrode BPM
8
configuration were determined as a function of time
by Inductively Coupled Plasma – Mass
Spectrometry using an Agilent 8800 spectrometer. T
he sampleintroduction system consisted of
a MicroMist nebulizer with a Scotttype spray chamb
er. A fixedinjector quartz torch was used
with a guard electrode, and the plasma was operated
at 1500 W. In addition to use of single
quad (MS) mode with no gas, elements were determine
d by MSMS modes in which different
collision or reactive gases were present in the gas
cell located between the quadrupoles. These
modes were nogas, He, H. External standards were u
sed to quantify the analytes determined.
TIC/TOC carbonate ion crossover measurements
To measure the carbonate ion crossover from
the catholyte to the anolyte through the BPM,
the anode chamber was sealed and purged with O
2
(g) (ALPHAGAZ 1) to prevent sorption of
CO
2
(g) from ambient air. Samples were analyzed for to
tal inorganic carbon (TIC) using an OI
Analytical Aurora Total Organic Carbon Analyzer. B
riefly, a sample aliquot was combined with
phosphoric acid and heated in a closed vessel. At
the end of the reaction time, carbon dioxide
was swept out of the vessel by a purge of N
2
(g). The gas stream was dried and passed through a
nondispersive infrared (NDIR) detector. The instr
ument was calibrated against potassium
hydrogen phthalate that had been oxidized to form c
arbon dioxide by heating persulfate, after
removal of TIC as described above. For these exper
iments, the instrument was calibrated and
showed a linear response between 5 and 50 ppm.
Faradaic efficiency for formate generation
1
H NMR spectroscopy was performed on a Bruker 400 MH
z Spectrometer to quantify the
amount of formate generation. Standards of 40 to 5
00 μM sodium formate solutions were
prepared by serial dilution, and were used to calib
rate the instrument (Figure S7). In general,
100 μL of 0.0800 vol.% 1,1’dimethylformamide, DMF
(internal standard), in H
2
O were added
to a 2 mL aliquot of the standard solution. 0.5 mL
of this solution was then transferred to a
NMR tube that contained 200 μL of D
2
O. A water suppression method was used to suppress
the
9
signal of the water in the electrolyte and to allow
visualization of the formate peaks. The same
procedure was used to quantify the formate product,
with 100 μL of 0.0800 vol.% DMF solution
added to 2 mL of the electrolyte after electrolysis
, and 0.5 mL of the solution added in a NMR
tube with 200 μL of D
2
O.
Discussion of nickel catalyst
The catalysts are denoted as Ni because Ni(OH)
2
was the catalyst precursor species,
however high catalytic activity, such as that obser
ved herein, generally originates from
incorporation of trace amounts of Fe into the Ni(OH
)
2
films to form a FeNi oxyhydroxide OER
electrocatalyst.
6, 7
Discussion of time dependence of the Faradaic effic
iency and potential of the Pd/C coated
Ti mesh cathode
The slight decay of the Faradic efficienc
y and potential of the Pd/C coated Ti mesh
cathode was due to CO poisoning of the catalyst, wh
ich can be regenerated by brief exposure to
air,
3
but the pH of the electrolyte was maintained at th
e initial values throughput the experiments
by the use of the bipolar membrane.
Calculation of standard error/standard error bar
The standard error was estimated by divid
ing the standard deviation of data obtained, for
three electrodes run under nominally the same condi
tions, by the square root of sample size. An
overpotential of 310 mV was measured at a current d
ensity of 8.5 mA cm
2
. As shown in Figure
1C, at 10 mA cm
2
, the p
+
Si/TiO
2
/Ni anode exhibited an overpotential of 320 mV for
the OER.
The
J
E
behavior of two additional p
+
Si/TiO
2
/Ni electrodes was measured, and the overpotential
and standard errors were determined to be: 330 ± 10
mV for 10 mA cm
2
and 320 ± 7 mV for 8.5
mA cm
2
. The Ni overpotential is consistent with prior wo
rk that has reported an overpotential
of 330 ± 7 mV at 10 mA cm
2
under the same Ni fabrication and testing conditio
ns as those used
10
herein.
8
Performance characteristics of larger area cathodes
and bipolar membranes
The Faradaic efficiency of formate produc
tion as a function of time, for 70 mV and 120
mV overpotentials, using a 1 cm
2
Pd/C nanoparticlecoated Ti mesh in CO
2
saturated 2.8 M
KHCO
3
(aq) was similar to that observed for cathode that
had a geometric area of 0.04 cm
2
(Figure S5A). The bipolar membrane voltage loss ve
rsus current density, normalized to the
0.785 cm
2
bipolar membrane area, was also similar to that ob
served for a bipolar membrane
having a geometric area of 0.03 cm
2
(Figure S5B).
Calculation and discussion of solarto formate conv
ersion efficiency (η
STF
)
The equilibrium potential for the CO
2
/HCO
2
couple was 0.02 V versus the reversible
hydrogen electrode (RHE).
4
Hence, the voltage required for the full chemical
reaction, 2OH
+
2CO
2
= 2HCOO
+ O
2
, was 1.21 V. The solarto fuels efficiency is def
ined by:
2, 9
=
A cm
×
× !! "# $! !% &' ($%)$"
*
+ ,
where
J
op
is operating current density;
E
0
is the voltage required for the full chemical reac
tion;
and
I
solar
is the solar power density.
In this example:
-./
=
1.21 V ×
45
,6 ,
× !! "# $! !$,)
100 ,+ ,
During the stability test, the overpotential for Pd
/C coated Ti mesh cathode was between 40 mV
and 100 mV, therefore, as shown in Figure 1B, the
Faradaic efficiency of CO
2
reduction to
formate was ~100%, 98%, 95%, 94% after 30 min, 1 h,
2 h, 3 h, respectively. The
corresponding solarto formate conversion efficienc
y was thus 10.5%, 10.3%, 10.0%, and 9.9%,
respectively.
Calculation of the area required to collect enough
solar photon flux for CO
2
reduction
11
Consider a coal plant operating 24 hours a
day with a CO
2
emission rate of 1400 lb/MWh.
Given that the photocurrent density is 8 mA cm
2
, and CO
2
is reduced to formate (2electron
process) with an average photocurrent density of 1.
6 mA cm
2
(20% capacity factor), then for a
100 MW power plant:
The CO
2
emission rate:
89:;/=>?×.9@ABCD/:;
AB /E
× 100 F+ = 17.64 kg/s
Photocurrent:
8M.B9 CD N
OP
×8 D CD
OP
99 Q R
OP
× 96500 U ,$&
8
× 2 = 7.74 × 10
M
6
Area:
M.M9×8
V
W
.8B W 4R
OX
∗ 10
9
,
,
= 5 × 10
B
,
Therefore, the required solar photon capture area i
s on the order of 10
7
m
2
.
Tables:
Table S1. Comparison of voltage losses for three c
ell configurations at
J
electrode/membrane
= 10 mA
cm
2
Components
(Cathode) 2.8 M
KHCO
3
/BPM/1.0 M
KOH (Anode)
(Cathode) 2.8 M
KHCO
3
/BPM/2.8 M
KHCO
3
(Anode)
(Cathode) 2.8 M
KHCO
3
/Nafion
Membrane/2.8 M
KHCO
3
(Anode)
Membrane
0.507 ± 0.038 V ~0.880 V (Figure S8) 0.214
± 0.015 V
Ni OER overpotential
0.33 ± 0.01 V
0.793 ± 0.026 V
0.793 ± 0.026 V
Pd/C coated Ti mesh
CO
2
R to formate
overpotential
0.057 ± 0.008 V
0.057 ± 0.008 V
0.057 ± 0.008 V
Total voltage loss
~0.89 V
~1.73 V
~1.06 V
Table S2. Comparison of voltage losses for three c
ell configurations at
J
electrode/membrane
= 8.5 mA
cm
2
Components
(Cathode) 2.8 M
(Cathode) 2.8 M
(Cathode
) 2.8 M
12
KHCO
3
/BPM/1.0 M
KOH (Anode)
KHCO
3
/BPM/2.8 M
KHCO
3
(Anode)
KHCO
3
/Nafion
Membrane/2.8 M
KHCO
3
(Anode)
Membrane
~0.48 V
~0.82 V (Figure S8)
~ 0.18 V
Ni OER overpotential
0.320 ± 0.007 V
0.783 ± 0.026 V
0.783 ± 0.026 V
Pd/C coated Ti mesh
CO
2
R to formate
overpotential
0.052 ± 0.008 V
0.052 ± 0.008 V
0.052 ± 0.008 V
Total voltage loss
~0.85 V*
~1.66 V
~1.01 V*
*The total voltage loss difference between (Cathode
) 2.8 M KHCO
3
/BPM/1.0 M KOH (Anode)
and (Cathode) 2.8 M KHCO
3
/Nafion Membrane/2.8 M KHCO
3
(Anode) configurations was
close to 180 mV experimental value.
Figures:
Figure S1.
Cyclic voltammetry of GaAs/InGaP/TiO
2
/Ni photoanode in 1.0 M KOH(aq) under
100 mW cm
2
of simulated AM1.5 illumination.
1
13
Figure S2.
Schematic drawings showing the configuration for
3electrode (A), 2electrode (B)
electrochemical measurements and the 4point measur
ement system for BPM measurements (C).
Figure S3
. The bipolar membrane voltage as a function of ti
me when the current density
through the bipolar membrane was maintained at 8.5
mA cm
2
.
14
Figure S4
.
1
HNMR spectrum of the solutions in the cathode (A)
and anode (B) compartments
in a 2electrode electrochemical configuration. Si
gnals at 7.92, 3.01 and 2.98 ppm are the
internal standard DMF, while the singlet at 8.44 pp
m is formate. The concentration of formate in
catholyte was ~ 1 mM, and no formate was detected in
the anolyte by
1
HNMR spectroscopy.
The membrane area was ~0.03 cm
2
, and the membrane current density during operation
was 8.5
mA cm
2
. The volumes of the catholyte and anolyte were 50
ml and 25 ml, respectively. The
operation time was 3 hours. When the concentration
of formate in catholyte increased to 0.1 M,
and the area of the BPM increased to 0.12 cm
2
(other conditions unchanged), small amount of
formate (~ 16 μM) in the anolyte was detected by
1
HNMR spectrum (C) after 3 hours of
operation. The current density for formate crossov
er from catholyte to anolyte was ~30 μA cm
2
.
Figure S5.
(A) Faradaic efficiency of formate production as
a function of time, for 70 mV
and 120 mV overpotentials, using 1 cm
2
Pd/C nanoparticlecoated Ti mesh in CO
2
saturated 2.8
15
M KHCO
3
(aq). (B) Membrane voltage loss (left axis) and me
asured total membrane voltage
(right axis) as a function of the current density n
ormalized to the 0.785 cm
2
BPM area. The cell
configuration was KHCO
3
(aq) (pH=8.0)/BPM/KOH (aq, pH=13.7).
Figure S6.
(A) The overpotential for a Pd/C cathode during sol
ardriven CO
2
R using
GaAs/InGaP/TiO
2
/Ni as a photoanode in a 2electrode electrochemica
l configuration under
simulated AM1.5 1Sun illumination. (see p. S9 for
discussion about the time dependence of
the potential for the Pd/Ccoated Ti mesh cathode.)
(B) 2electrode
J
V
behavior with a
GaAs/InGaP/TiO
2
/Ni photoanode and a Pd/C on Ti mesh cathode in a B
PM configuration under
simulated AM1.5 1Sun illumination at 0 and 3 h, re
spectively, of the stability test.
16
Figure S7.
Formate concentration calibration curve: the rati
o of the formate 8.44 ppm peak area
and the DMF 7.92 ppm peak area as a function of for
mate concentration. (R
2
= 0.9996)
Figure S8.
Membrane voltage loss (left axis) and measured to
tal membrane voltage (right axis)
as a function of the current density normalized to
the 0.030 cm
2
BPM area. The cell
configuration was KHCO
3
(aq) (pH=8.0)/BPM/KHCO
3
(aq) (pH=8.0).
17
References:
(1) Verlage, E.; Hu, S.; Liu, R.; Jones, R. J. R.;
Sun, K.; Xiang, C.; Lewis, N. S.; Atwater, H. A.
A monolithically integrated, intrinsically safe, 10
% efficient, solar-driven water-
splitting system based on active, stable earth-abun
dant electrocatalysts in
conjunction with tandem III-V light absorbers prote
cted by amorphous TiO
2
films.
Energy Environ. Sci.
2015
,
8
, 3166-3172.
(2) Coridan, R. H.; Nielander, A. C.; Francis, S. A
.; McDowell, M. T.; Dix, V.; Chatman, S. M.;
Lewis, N. S. Methods for comparing the performance
of energy-conversion systems
for use in solar fuels and solar electricity genera
tion.
Energy Environ. Sci.
2015
,
8
,
2886-2901.
(3) Min, X.; Kanan, M. W. Pd-Catalyzed Electrohydro
genation of Carbon Dioxide to
Formate: High Mass Activity at Low Overpotential an
d Identification of the
Deactivation Pathway.
J. Am. Chem. Soc.
2015
,
137
, 4701-4708.
(4) Kortlever, R.; Peters, I.; Koper, S.; Koper, M.
T. M. Electrochemical CO
2
Reduction to
Formic Acid at Low Overpotential and with High Fara
daic Efficiency on Carbon-
Supported Bimetallic Pd–Pt Nanoparticles.
ACS Catal.
2015
,
5
, 3916-3923.
(5) Sun, K.; Liu, R.; Chen, Y.; Verlage, E.; Lewis,
N. S.; Xiang, C. A Stabilized, Intrinsically
Safe, 10% Efficient, Solar-Driven Water-Splitting C
ell Incorporating Earth-Abundant
Electrocatalysts with Steady-State pH Gradients and
Product Separation Enabled by
a Bipolar Membrane.
Adv. Energy Mater.
2016
,
6
, 1600379.
(6) Trotochaud, L.; Young, S. L.; Ranney, J. K.; Bo
ettcher, S. W. Nickel–Iron Oxyhydroxide
Oxygen-Evolution Electrocatalysts: The Role of Inte
ntional and Incidental Iron
Incorporation.
J. Am. Chem. Soc.
2014
,
136
, 6744-6753.
(7) Corrigan, D. A. The Catalysis of the Oxygen Evo
lution Reaction by Iron Impurities in
Thin Film Nickel Oxide Electrodes.
J. Electrochem. Soc.
1987
,
134
, 377-384.
(8) Sun, K.; Saadi, F. H.; Lichterman, M. F.; Hale,
W. G.; Wang, H.-P.; Zhou, X.; Plymale, N.
T.; Omelchenko, S. T.; He, J.-H.; Papadantonakis, K
. M.; et al. Stable solar-driven
oxidation of water by semiconducting photoanodes pr
otected by transparent
catalytic nickel oxide films.
Proc. Natl. Acad. Sci. U.S.A.
2015
,
112
, 3612-3617.