S
1
Supporting Information
for
Si Microwire
-
Array Photocathodes
Decorated with Cu
Allow
CO
2
Reduction with Minimal Parasitic Absorption of
Sunlight
Paul A. Kempler
1
, Matthias
H.
Richter
1
, Wen
-
Hui Cheng
2
,
Bruce S. Brunschwig
3
,
Nathan S. Lewis
1,3*
1
Division
of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena,
CA 91125
2
Division of Engineering and Applied Science, California Institute of Technology, Pasadena,
CA
91125
3
Beckman Institute, California Institute of Technology, Pasad
ena, CA 91125
*Corresponding Author:
nslewis@caltech.edu
This PDF file includes:
Equation S1
Figures S1 to S
10
Table
s
S1
to S2
S
2
Detailed Materials and Experimental Methods
Materials:
Isopropyl alcohol,
acetone, hydrochloric acid (36.5
-
38.0
%)
,
and n
itric
a
cid
(67
-
70
%, TraceMetal Grade)
were purchased from Millipore
,
and CuSO
4
was purchased from
Flinn Scientific.
Sulfuric Acid (TraceMetal Grade) was purchased from Fisher
. A
mmonium
hydroxide (28
–
30
%) was purchased from JT Baker
.
K
2
SO
4
(99
%)
,
and
potassium bicarbonate
(
99.995
%
)
were purchased from Sigma
-
Aldrich. Methanol was purchased from VWR Chemical
and hydrogen peroxide (30
%) was purchased from Macro
n
. B
uffered oxide etchant (6:1
NH
4
F/HF) was
purchased from Transene.
Boron doped,
P
-
type silicon wafers with a resistivity of 10
-
20
W
×
cm were purchased
from Addison Engineering.
Platinum
foil (
99.
99
%
)
was purchased
from Alfa Aesar
,
and
copper
foil (99.999
%)
was obtained
from Sigma Aldrich.
CO
2
(99.999
%, <1.0
ppm Ar+O
2
+CO,
<1.0
ppm THC, <3.0
ppm H
2
O, <5.0
ppm N
2
) was purchased from Airgas.
Preparation of Si Photocathodes:
Si was degreased with acetone and isopropyl alcohol and
was then
spin coated
at 4000
rpm
for 30
s
with Shipley 1813 photores
ist. A square array of
circles
,
having a
diameter
of
3
μ
m and a pitch of 7
μ
m
,
was defined using UV exposure through
a chrome mask
. T
he pattern was developed with MF
-
319 developer, and the resist was hard
-
baked at 115
°C for 10
min
. Al
2
O
3
masks, 125
nm in thickness, were deposited via
e
-
b
eam
evaporation at 1
Å
×
s
-
1
into the exposed hole array and the resist was removed via sonication in
Remover
-
PG (MicroChem) at 50
°C. Si was structured into
30
μ
m
tall
microwire arrays
via deep
reactive ion etching
(
RI
E
)
in a SF
6
/O
2
plasma controlled by an Oxford
Plasmalab System 100
at
-
130 °C. An
inductively coupled plasma
power of 900
W produced etching rates
of 1
μ
m min
-
1
,
while a low
c
a
pacitively coupled plasma
power of 3
W minimized sidewall damage
and mask
remova
l
. All Si samples were cleaned in a Radio Corporation of America, RCA, SC1 bath (5:1:1
S
3
H
2
O/NH
4
OH/H
2
O
2
, 80 °C for >10 min), buffered oxide etchant (20 °C for 5 min), and an RCA
SC2 bath (6:1:1 H
2
O/HCl/H
2
O
2
, 70
°C for >10 min) to remove SiO
2
, Al
2
O
3
and trace metal
impurities. A solid
-
state n
+
p junction was formed via diffusion doping
from
CeP
2
O
14
doping
wafers (Saint
-
Gobain, PH
-
900 PDS) at 850 °C for 10 min. The P
2
O
5
glass
formed during the
doping procedure
was removed from the Si surface
via
buffer
ed oxide etc
hant
for > 60 s prior to
electrodeposition. Samples were cleaved with a carbide scribe into ~10 mm
2
chips
.
Ga/In
eutectic was scratched into the back of the electrodes
, and
the electrodes
were attached t
o a Sn
-
coated Cu wire
using
Ag paint (Ted
Pella)
. The back contact and sides of the electrodes were
sealed to 6 mm outer diameter glass tubing using insulating epoxy (Loctite 9460)
. T
he epoxy
was cured for at least 12 h.
The area of the electrodes was measured with a commercial scanner.
(Photo)e
lectrodepositions of Catalyst:
Prior to electrodeposition of Cu, electrodes were rinsed
sequentially with acetone, isopropyl alcohol, methanol, and deionized water and then dipped into
buffered oxide etchant for 60 s. Electrodepositions were controlled wit
h a BioLogic SP
-
200
potentiostat. The Cu
-
plating bath was continuously purged with Ar(g) and contained
0.1
0
M
CuSO
4
(aq), 5.0
mM H
2
SO
4
(aq), and 0.1
0
M
K
2
SO
4
(aq), at a pH of ~3. A saturated
calomel electrode (SCE, CH Instruments) was used as a reference and
the counter electrode was
a high
-
purity graphite rod (Alfa Aesar, 5N)
(Figure 1a)
. The illumination source was
an array of
narrowband light
-
emitting diode
s
(
Luxeon Rebel Blue SMD,
FWHM 22
nm)
with a
peak
intensity at
4
65
nm
. T
he illumination wavelength
was
selected to maximize
transmission
of light
through the colored electrolyte. Cu was deposited potentiostatically at 0.0 V vs. SCE until the
desired charge
density had passed, normalized to the projected area of the electrode.
Photoelectrochemical CO
2
R
eduction Testing:
The electrochemical setup was operated in a
continuous flow mode
(Figure S1)
. Carbon dioxide was provided to the electrochemical cell
at a
S
4
flow rate of
5
sccm
as
controlled
by
an Alicat flow controller. The carbon dioxide stream was
suppl
ied as humidified CO
2
with a gas bubbler
placed
between the cell and flow controller. The
exhaust gases went through a liquid trap, then an Alicat flow meter, and finally to a gas
chromatograph (SRI
-
8610) using a Hayesep D column and a Molsieve 5A column w
ith N
2
as the
carrier gas. The gaseous products were detected using a thermal conductivity detector (TCD) and
a flame ionization detector (FID) equipped with a methanizer. Quantitative analysis of gaseous
products was based on calibrations with several gas
standards over many orders of magnitude in
concentration
.
The calibrations were
used to calculate the partial current density,
j
, towards
products of the CO
2
R and hydrogen evolution reaction. To
measure
liquid products, the
electrolyte on the anode and c
athode sides of the cell
was
sampled at the end of the run and
was
analyzed by
high
-
performance liquid chromatography (HPLC, Thermo Scientific Ultimate
3000).
Products were not quantified in Faradaic efficiency calculations because continuous
purging of th
e catholyte with CO
2
expelled accumulated products
.
Moreover,
c
rossover of
products to the anolyte was observed
and
oxidation at the anode
could potentially occur.
An
Oriel Instruments 75
W Solar Simulator supplied 100 mW
×
cm
-
2
of
AM
1.5 illumination. The
light intensity was calibrated using
the measured photocurrent at
a calibrated (350 to 1100 nm,
1
cm
2
) NIST traceable Si photodiode (Thorlabs FDS1010
-
CAL)
mounted within the testing cell
prior to the addition of the electrolyte.
Elect
rochemical measurements of GDE cell
:
A PEEK compression cell
(
Figure S1
)
was used
as the vessel for the measurement with an anode chamber volume of 2
mL and a cathode
chamber volume of 4
mL. The anode, cathode electrode and membrane area were each 1
cm
2
as
constrained by the design of the compression cell.
CO
2
saturated
0.1
0
M
potassium bicarbonate
(KHCO
3
, pH 6.8)
was used as the electrolyte. A Pt foil anode was used behind a Selemion anion
S
5
exchange membrane. A leakless Ag/AgCl electrode was used as a refer
ence. All electrochemical
measurements were performed using a Biologic VSP
-
300 potentiostat. Scan rates were set to
50
mV
×
s
-
1
. Cu foil (99.999
%, Sigma Aldrich) was mechanically polished (Struers LabPol
-
5)
using 0.05
0
μ
m
a
lumina
suspension (MasterPrep
) and
then was
electropolished
for 5
min
in 85%
H
3
PO
4
at +2.1
V vs.
a
carbon counter electrode.
Physical Characterization:
Scanning
-
electron micrographs (SEMs) were obtained with a FEI
Nova NanoSEM 450 at an accelerating voltage of 10.0
kV and a working di
stance of 5.0
mm
using an Everhart
-
Thornley secondary electron detector.
Comparison of Catalyst Loadings:
The mass loadings of catalyst
were
compared assuming a
F
aradaic
efficiency
of ~
100
%
towards metal plating
.
Equation S1 can be used to calculate the
mass loading density,
M
cat
(
mg
×
cm
-
2
)
, from the cathodic charge density,
-
Q
(C
×
cm
-
2
)
and molar
mass of the catalyst
. For Cu
(
m
a
= 63.55 g
×
cm
-
2
),
M
cat
was 0.329 and 0.0487 mg
×
cm
-
2
for
-
Q
= 1
and 0.148 C
×
cm
-
2
, respectively.
푀
"#$
=
&
'
(
)
*+
(S1)
Explanation of Resistance Measurement and
iR
s
Correction:
The resistance (R
s
) was
determined by electrochemical impedance spectroscopy
(EIS)
at
the open
-
circuit potential.
During the
experiment,
iR
s
was corrected
by
85% and the remaining 15
%
was
corrected
for
after
the experiment.
Measurements of Pt Crossover during Stability Testing
The rate of Pt dissolution in 0.1
0
M
KHCO
3
(aq) and crossover through Selemion
were
measured via a galvanostatic experiment at 10
mA using a Pt anode and a graphite ca
thode. The volume
s
of the anolyte and catholyte
were each
S
6
13 mL. ICP
-
MS measurements confirmed the presence of dissolved Pt in both the anolyte and
catholyte (
Table S2
).
Predicted
j
-
E
behavior:
Illuminated
j
-
E
behavior was predicted by shifting the fitted
Tafel
behavior of the polished Cu foil
towards positive potentials
by
V
ph
+
b
log
10
(
R
μ
). This calculation
assumes
a
comparable microstructured area of Cu
islands
and Si
and
similar [H
+
(aq)] and
[CO
2
(aq)] at the surface of the two electrodes
.
Simulated
j
-
E
behaviors
in Figure 4
b
w
ere
produced
by summing the implicit values for
η
and
V
ph
,
calculated form Equations 1 and 2, as a
function of
R
μ
and
b
. Arbitrary values
were
selected
for
J
o
=
1
⨉
10
-
10
A cm
-
2
and
푎
=
b
log
10
(
i
0
)
,
with
i
0
=
1
⨉ 10
-
7
A cm
-
2
;
these parameters do not affect the potential shift resulting
from a change in
R
μ
but affect the total
V
ph
and
η
observed
.
Figure
S1:
Cell configuration composed of
1
Pt anode,
2
Si
μ
W
cathode,
3
anion exchange
membrane,
4
quartz window,
5
reference
electrode,
6
catholyte chamber,
7
anolyte chamber,
8
CO
2
gas inlet, and
9
product/CO
2
outlet. Illumination to the cathode is provide
d
through a hole in
the
left side
of the compression cell, which extends through the an
o
lyte c
h
amber plate
and
th
e
quartz window
(4)
to the
μ
Si
μ
W cathode
.
S
7
Figure S2
: Electrochemical
J
-
E
behavior of polished graphite (dotted lines) in an Ar purged Cu
deposition bath, before (blue) and after (red) passage of a charge density of
-
1
.
00
C
×
cm
-
2
.
Figure S
3
:
Plot
s
of the p
a
rtial current
densitie
s
towards
(
p
hoto)electrochemical CO
2
R products
vs
.
p
otential
for
n
+
p
-
Si
μ
W/Cu
electrodes
(
filled markers
)
compared
to
planar
Cu electrodes
(
open markers
)
.
(a)
j
CO
-
E
behavior
for production of
CO
, (b)
j
CH4
-
E
behavior
for production of
CH
4
, (c)
j
C2H4
-
E
behavior
for production of
C
2
H
4
, and (d)
j
H2
-
E
behavior
for production of
H
2
,
detected by online GC
-
FID/TCD
.
-0.6
-0.4
-0.2
0
Potential / V vs RHE
-20
-15
-10
-5
0
J
/ mA cm
-2
S
8
Figure S
4
:
Faradaic efficiencies for products detected via GC
-
FID/TCD
, as a function of
potential during CO
2
R in 0.1
0
M KHCO
3
(aq) saturated with CO
2
(g) at
:
(a) an electropolished Cu
film
,
and (b) a n
+
p
-
Si
μ
W/Cu photocathode under simulated sunlight.
S
9
Figure S5
: Partial photocurrent density towards CH
4
, C
2
H
4
, CO, and H
2
, respectively,
as
measured via GC
-
FID/TCD, represented as blue, gold, red, and magenta squares, respectively.
Planar n
+
p
-
Si after
(a,b)
-
1
.
00 C cm
-
2
or
(c,d)
-
0.
148 C cm
-
2
of charge was passed towards Cu
deposition. Electrodes were held
under simulated sunlight
at
-
0.69 V vs RHE in 0.1
0
M
KHCO
3
(aq) saturated with CO
2
(g).
S
10
Figure S
6
:
Photoelectrochemical
data
for n
+
p
-
Si Cu electrode
s
during stability
evaluation
at
-
0.58 V vs
.
RHE in CO
2
-
purged 0.
1
0
M
KHCO
3
(aq) under 100 mW
×
cm
-
2
simulated sunlight.
(a
-
b) Time dependent
J
-
E
behavior for a (a) n
+
p
-
Si
μ
W/Cu electrode and (b) planar n
+
p
-
Si/Cu
electrode, loaded with
-
1.00 and
-
0.
148 C cm
-
2
, respectively
, as measured by a linear sweep
voltammogram recorded at a scan rate of 50 mV
×
s
-
1
.
(
c
)
Absolute
photocurrent
density
and
partial
photocurrent density towards H
2
(g), as measured via
GC
-
FID/TCD, represented as black
x’s and a gray line, respectively
for a
n
+
p
-
Si
μ
W/Cu electrode
. (
d
)
Absolute p
artial photocurrent
density towards CO(g), C
2
H
4
(g), and CH
4
(g)
presented
as red, gold, and
blue lines, respectively
for a
n
+
p
-
Si
μ
W/Cu electrode
.
S
11
Figure S
7
:
SEM
of
an
n
+
p
-
Si
-
μ
W/Cu photocathode after 96
h
of continuous
photoelectrochemical
operation
in CO
2
-
purged 0.1
0
M KHCO
3
(aq) under 100 mW
×
cm
-
2
of
simulated sunlight. Primary scale bar
represents 20
μ
m, inset scale bar represents 2
μ
m.
Figure S8:
Plot of
E
oc
vs. the
ln
of the light
-
limited photocurrent density (A
×
cm
-
2
) in a Cu plating
bath for planar and
μ
W n
+
p
-
Si. The illumination source was a narrow
band LED with a peak
intensity
at
630 nm. The illumination wavelength was selected to maximize transmission through
the Cu film. The ideality factor was calculated from the slope of the linear fit in range of
photocurrent densities relevant to operation under 100 mW
×
cm
-
2 of simulated sunli
ght.
-5
-4
-3
-2
ln|
J
ph
|
0.4
0.45
0.5
0.55
0.6
0.65
0.7
E
oc
/ V
W
Planar
S
12
Figure S
9
:
Surface characterization of films of sputtered Cu before
and
after sequential potential
holds for 2 hr at
-
1.2,
-
1.0,
-
0.8,
-
0.6, and
-
0.4 V vs RHE in CO
2
saturated 0.10 M KHCO
3
(aq).
XPS spectra
of (a)
the Cu 2p region, (c) the Cu 3p and
Pt 4f region
s
, (d) the Pt
4d region, and (b)
Auger electron spectra in the Cu LMM region did not detect Pt on the surface but confirmed that
Cu oxide species were reduced during testing.
S
13
Figure S
10
:
(a) X
-
ray photoelectron spectra of the surface o
f a graphite cathode in the Pt
4f
region after 5
days
under
galvanostatic
control
at
-
10 mA in 0.1
0
M KHCO
3
(aq)
. The graphite
cathode
was
separated from a Pt anode by a Selemion membrane. (b) SEM image of the surface
of the graphite surface measured in (a
)
. Small metal nanoparticles, visible as white dots, were not
present prior to the galvanostatic experiment.
Table S1:
Liquid CO
2
R products for
a n
+
p
-
Si
μ
W/Cu electrode
after
a 48
h
stability test
under
100 mW
×
cm
-
2
of
simulated sunlight
at
-
0.58
V vs.
RHE in CO
2
-
purged 0.1
0
M KHCO
3
(aq)
.
Dashes indicate measurements below the limit of quantitation.
Product
C
cathode
(mM)
C
a
n
ode
(mM)
Formate
13.
5
7.9
Acetate
0.2
0.
3
Methanol
-
-
Ethanol
0.3
3.
4
Propanol
0.
1
0.1
Allyl Alcohol
-
-
Ethylene Glycol
-
-
S
14
Table S
2
:
Dissolved Pt concentration in the anolyte and catholyte, measured via inductively
-
coupled plasma mass
-
spectrometry, following a 5
day experiment
with a constant current of
10
mA
applied
between a Pt foil anode and graphite cathode separated
by Selemion AMV
membrane.
Pt [ug/L]
Pt error [ug/L]
Pt [ug]
Pt error [ug]
Anolyte
5
40
53
7.0
0.7
Catholyte
60
20
0.8
0
.3