of 12
1
Supporting information for:
Probing the Catalytically Active Region in a Nanoporous Gold Gas Diffusion
Electrode for Highly Selective Carbon Dioxide Reduction
Aidan Fenwick,
1,2
* Alex J. Welch,
2,3
* Xueqian Li,
2,3
Ian Sullivan,
1,3
Joseph S. DuChene,
4
Chengxiang
Xiang,
1,3
and Harry A. Atwater
2,3
1
Department of Chemistry, California Institute of Technology, Pasadena, CA 91125, USA
2
Liquid Sunlight Alliance, California Institute of Technology, Pasadena, CA 91125, USA
3
Department of Applied Physics and Material Science, California Institute of Technology, Pasadena, CA 91125, USA
4
Department of Chemistry, University of Massachusetts Amherst, MA 01003, USA
*These authors contributed equally to this work
Corresponding Author:
†E-mail:
haa@caltech.edu
Experimental:
Materials:
Potassium carbonate (99.995% trace metal basis, Sigma Aldrich), Nitric Acid (70%, purified by
redistillation, >99.999% trace metals basis, Sigma Aldrich) Si [p-type, 0-10
cm, (100) orientation, 620
± 25
m thick, University Wafers], Pt gauze (99.9%, 52 mesh woven from 0.01 mm wire, Alfa Aesar)
and Copper(II) sulfate
(≥99%,
Sigma Aldrich) were used without modification unless otherwise noted.
The carbon paper substrate (Sigracet 38 BC, FuelCell Store) was non-woven with a microporous layer
that was treated with 5% PTFE. The materials for electron beam deposition were ordered from
Plasmaterials. The Au target was 99.99% pure with 3-6 mm random size pieces, Ag was 99.99% pure
with 3-6 mm random size pieces, and Ti was 99.995% pure in 0.25 diameter pellets. All water used for
experiments was deionized and filtered through a 0.22
m Millipak Express 40, serial number 0826.
2
F
abrication of Au
x
Ag
1-x
alloy on carbon paper or silicon:
All electrodes were fabricated with an AMOD dual electron beam deposition system (System
02520, Angstrom Engineering). Carbon paper substrates were lightly dusted under an N
2
stream before
insertion into the electron beam deposition chamber. Si substrates were cleaned by sonicating
sequentially in acetone, isopropanol, and deionized water for two minutes each, stored in deionized
water and dried under a stream of N
2
prior to use. For the planar gold electrodes on silicon, 2 nm of Ti
was deposited at a rate of 1 Å/s, then 300 nm of Au was deposited at a rate of 1 Å/s. For the nanoporous
gold samples of varying atomic percent gold, Au and Ag were co-deposited at varying rates/thicknesses
to yield the desired Au
x
Ag
1-x
alloy. For example, a 25% Au and 75% Ag alloy was fabricated by
depositing 75 nm of Au at a rate of 1 Å/s and 225 nm of Ag at a rate of 3 Å/s, respectively. Over the
course of the deposition the partial pressure of the chamber was maintained between ~10
-7
torr to ~10
-6
torr.
Fabrication of nanoporous gold (np-Au) electrodes by chemical dissolution of Ag:
The Au
x
Ag
1-x
alloy coated carbon paper or silicon substrates were placed in a covered glass petri
dish of room temperature concentrated nitric acid (70 weight/volume) for 15 minutes. A notable color
transition from white gold to burnt umber occurs as the silver in the base alloy is removed to yield the
desired np-Au morphology. Samples were then removed from the nitric acid and rinsed under a copious
flow of water and then soaked in a water bath for an hour. Finally, the substrates were dried under a
stream of N
2
and subsequently dried in vacuo for 24 hour.
Substrate characterization by Scanning Electron Microscopy (SEM):
Substrates were imaged in a Nova200 Nanolab Dualbeam FIB/SEM with an acceleration voltage
of 10 keV and spot size of 3. To prepare the a sample for cross sectional imaging, the np-Au coated
carbon paper substrate was soaked in liquid N
2
, cracked in half, and placed on a holder that allowed a
90
orientation relative to the electron beam.
3
A
u
/
A
g
a
l
l
o
y
5
0
0
μ
m
5
0
0
n
m
3
0
0
n
m
B
a
r
e
S
i
g
r
a
c
e
t
3
8
B
C
5
0
0
μ
m
5
0
0
n
m
3
0
0
n
m
A
B
C
D
E
F
5
0
0
n
m
5
0
0
μ
m
A
f
t
e
r
H
N
O
3
e
t
c
h
3
0
0
n
m
G
H
I
P
o
s
t
e
l
e
c
t
r
o
c
h
e
m
i
s
t
r
y
J
K
L
3
0
0
n
m
5
0
0
n
m
5
0
0
n
m
Figure S1:
Shows SEM images of the electrode at different phases in the fabrication process. (a)-(c)
shows images of the bare carbon paper, Sigracet 38BC. (d)-(f) show images of the gold silver alloy on
the carbon paper. (g)-(i) show images of the nanoporous gold morphology from a 35% Au alloy that
forms after the nitric acid etch. (j)-(k) show images of the electrodes after electrolysis.
4
4
5
a
t
o
m
i
c
%
A
u
1
5
0
n
m
6
0
0
n
m
3
μ
m
1
5
a
t
o
m
i
c
%
A
u
A
6
0
0
n
m
1
5
0
n
m
B
2
5
a
t
o
m
i
c
%
A
u
1
5
0
n
m
D
6
0
0
n
m
E
3
5
a
t
o
m
i
c
%
A
u
1
5
0
n
m
G
6
0
0
n
m
H
J
K
3
μ
m
3
μ
m
3
μ
m
C
F
I
L
Figure S2:
SEM characterization of nanoporous gold (np-Au) electrodes with a varying gold atomic
percent (%
Au
)
of 15%
Au
(a-c), 25%
Au
(d-f), 35%
Au
(g-i), and 45%
Au
(j-l).
5
2
0
0
n
m
2
0
a
t
o
m
i
c
%
A
u
2
5
a
t
o
m
i
c
%
A
u
3
0
a
t
o
m
i
c
%
A
u
1
μ
m
A
B
H
E
D
G
2
0
0
n
m
1
μ
m
2
0
0
n
m
1
μ
m
1
0
μ
m
1
0
μ
m
1
0
μ
m
C
I
F
Figure S3:
SEM characterization of nanoporous gold (np-Au) electrodes with a varying gold atomic
percent (%
Au
)
on silicon. 20%
Au
(a-c), 25%
Au
(d-f), and 30%
Au
(g-i).
6
1
μ
m
Figure S4
: Cross sectional SEM of a 300 nm thick 35% gold nanoporous gold electrode on carbon
paper. It was observed that while the carbon paper substrate has a high rugosity and is quite uneven, the
nanoporous gold had a homogenous pore structure through the entire thickness.
7
Electrochemical carbon dioxide reduction (CO
2
R) experimental set up:
Electrochemical CO
2
R was performed in a custom flow cell configuration that consisted of a
two-compartment cell made of polyether ether ketone (PEEK) and separated with a laser cut Viton
rubber gaskets. The tapered catholyte chamber was 3D printed (Creality3D CR-10 3D printer) out of
acrylonitrile butadiene styrene (ABS). This chamber design was found to reduce bubble buildup in the
catholyte chamber. The carbon paper electrodes were taped with 3M double sided Cu tape onto the
PEEK back plate with a serpentine channel carved into it. The carbon paper GDE served as the working
electrode with an Ag/AgCl leakless reference electrode (Innovative Instruments LF-2, 2mm OD) and a
platinum mesh counter electrode. The cathode and anode chambers were separated by an anion
exchange membrane (AGC, Selemion AMV). A 20 mL volume of 1 M KHCO
3
was separately added to
a catholyte and anolyte reservoir and independently recirculated through each respective chamber at
flow rate of 15 mL/min (two Masterflex 77120-62 pumps). The 1 M KHCO
3
electrolyte was bubbled
with CO
2
(Research grade from Airgas) for 30 minutes minimum prior to use. CO
2
was humidified by
a glass fritted water bubbler prior to entrance into the electrolysis cell to prevent evaporation of the
electrolyte. CO
2
was flowed across the back of the carbon paper cathode through the serpentine channel
at a flow rate of 50 standard cubic centimeters per minute (SCCM) via an Alicat M-Gas mass flow
controller and monitored downstream by an Alicat M-Gas mass flow meter. The cathode outlet gas was
sent back to the catholyte reservoir to catch any breakthrough of electrolyte through the GDE. From
there the outflowing gas was sent through a10 mL water trap to prevent any water from entering the gas
chromatograph (SRI-8610) equipped a Haysep D and Molsieve 5A columns and with a thermonal
conductivity detector (TCD) and flame ionization detector (FID). Product gas was passed through a
methanizer prior to the FID. The electrolyte of the anode and cathode were sampled to measure liquid
producs by high performance liquid chromatography (HPLC) however, no liquid products were ever
observed.
Electrochemical experiments were performed at room temperature using a potentiostat (Biologic
VSP-300). All potentials were converted to the reversible hydrogen electrode (RHE) scale using the
following equation: V
RHE
= V
Ag/AgCl
+ 0.197 + 0.059*(pH of the medium). Prior to each experiment,
potentiostatic impedance spectroscopy (PEIS) was carried out to determine the solution resistance of the
cell (typically between 5-10
Ω).
The applied potential was compensated 85% by the potentiostat and
remainder determined by the resistance form PEIS. When an operating cell potential was applied, the
current was stabilized for 200 seconds minimum before any GC analysis was carried out. The current
density was averaged for the 150 seconds prior to the GC injection. After an electrochemical run, the
cell was disassembled, sonicated in 15% nitric acid for 10 minutes, rinsed with water and sonicated in
water for 10 minutes prior to reuse at least 2 times. The cell was stored in a 15% nitric acid solution at
the end of each day. Each electrode was disposed of after a single run. Carbon paper controls in 1 M
KHCO
3
were carried out. At -1.48 V
Ag/AgCl
we observed a 0.63 mA/cm
2
current density that went only
towards HER. At -1.68 V
Ag/AgCl
we observed a 1.2 mA/cm
2
current density that consisted entirely of
HER. No CO
2
R products were detected by GC and thus we conclude that all CO observed in our
experiments arises from genuine CO
2
R carried out by our np-Au catalysts.
8
A
B
Figure S5
: Photographs of electrochemical cell and experimental set up. (A) shows the cell set up for
operation with leads attached. (B) shows each individual component of the disassembled cell. From left
to right there is the serpentine channel, 3D printed catholyte chamber (with np-Au electrode on top),
anode chamber and anode back plate. Below are the laser cut Viton gaskets with the gaskets. The left
gasket has the Pt mesh counter taped behind it with an Al lead attached.
9
Figure S6
: Contact angle measurements of nanoporous gold electrodes with varying atomic gold
percentages. 0 atomic percent Au indicates that there is no catalyst layer and 100 atomic percent Au
indicates that there was a solid gold film deposited. All samples were deposited on Sigracet 38BC
unless otherwise noted.
10
Cu underpotential deposition (Cu UPD) measurements:
Cu UPD measurements were utilized to determine the surface area of an electrode in contact with the
electrolyte. A CuSO
4
solution (0.1 M) in 0.5 M H
2
SO
4
was used as the deposition bath for all Cu UPD
experiments. The solution was sparged with N
2
(research grade form Airgas) for 30 minutes to remove
dissolved O
2
prior to any experiment. The working electrode was a planar Au or np-Au film of various
thickness on carbon paper or a flat silicon wafer with a Pt mesh counter electrode and an Ag/Ag/Cl
reference electrode. Flooded measurements were carried out in a single chamber compression cell and
in situ GDE measurements were carried out in a single chamber compression cell in which N
2
was
flowed at various flow rates through a serpentine channel behind the electrode. CVs from 450 mV to 50
mV vs Ag/AgCl at a scan rate of 5 mV/sec were acquired until the traces converged. A total of three
electrodes were measured at each thickness and flow rate. The anodic stripping peak was integrated to
determine charge passed. The electrochemical surface area enhancement was obtained by taking the
average surface area of the each electrode relative to a planar Au film deposited onto a flat Si wafer.
Secondary Ion Mass Spectroscopy (SIMS) sample preparation and measurements:
The depth profiles of the samples were collected with a Cameca ims 7f-GEO secondary ion mass
spectrometer (SIMS) at the Caltech Microanalysis Center. A rastering (100 um x 100 um) O2+ primary
beam (+13 keV, 15nA) was used to sputter the sample surface. Positive secondary ions of 7 keV were
extracted and collected in peak-jumping mode with either an electron multiplier (EM, for 12C+; 65Cu+;
107Ag+; and 197Au+) or a Faraday Cup (FC, for 65Cu+). A field aperture was used to avoid edge effect
from beam sputtering and to limit collected secondary ions only from the center area of 35 um in
diameter. In each data collection cycle, the collection time was 1 sec for each mass with appropriate
magnet settling time in between. Because there were no significant interferences for the masses being
measured, the mass resolving power (MRP) of the mass spectrometer was set at 1200. The np-Au
electrode analyzed by SIMS were prepared by putting each electrode into the Cu UPD deposition bath in
the desired cell configuration and electroplating Cu onto the electrodes by holding a constant potential
of -0.08 V vs Ag/AgCl for 45 seconds for each distinct electrode. The residual Ag in the fully etched np-
Au electrodes was determined by SIMS as well. The silver content was calibrated with a 15%
Au
, and
35%
Au
white gold alloy and the residual Ag in an etched 35%
Au
np-Au electrode was determined to be
1.3%.
a
)
b
)
c
)
11
Figure S7
: Secondary ion mass spectroscopy raw data for aqueous CO
2
fed system and vapor CO
2
fed
system. (a) shows the gold counts, (b) the copper counts, and (c) the carbon counts. In conjunction with
figure 5 of the main text, we hypothesize the Cu:Au counts decrease as a function of depth in both
samples as consequence of the rough and bumpy nature of the underlying carbon paper substrate. We
hypothesize the Cu:Au ratio is enhanced at the surface of the catalyst layer because at this location (and
time) we are only sputter profiling through pure Cu and Au. As we penetrate deeper into the catalyst
layer, we are now also sputter depth profiling through some of the underlying carbon. The copper is not
going to plate onto the unflooded carbon, so it appears that there is less copper deeper in the substrate.
This is supported by figure S7 C, where after 100 seconds the carbon counts steadily increase.
a
)
b
)
Figure S8
: Secondary ion mass spectroscopy raw data for 300 nm planar Au on a silicon wafer (a)
shows the Au counts, and (b) shows the Si counts.
12
Figure S9
: Long term chronoamperometry of a 25%
Au
electrode run at 50 SCCM of CO
2
in a 1M
KHCO
3
electrolyte. The x axis is time in seconds and the y axis is current density in mA/cm
2
. Injections
were taken at every 15 minutes starting at 5 minutes in. The first three injections have an averaged FE
CO
of 76%. The electrode is quite stable for the first hour of operation (the large spikes in current density
are bubbles forming and coming off the electrode surface). After the first ~50 minutes of operation the
carbon paper substrate begins to flood and HER increases at the expense of HER. The fourth injection
has FE
CO
of 71%. At 2 hours of operation the electrolyte was observed to break through the carbon
paper substrate and into the serpentine channel gas line. Carbon paper is well known to degrade after an
hour and this is not unexpected. HER continues to increase to over 50% of the current density
throughout the duration of the experiment. We note that all the experiments presented in the main text
of this paper were taken during the first 45 minutes of operation of any electrode.