Nanoporous Gold as a Highly Selective and Active Carbon Dioxide
Reduction Catalyst
Alex J. Welch,
†
,
‡
Joseph S. DuChene,
†
,
‡
Giulia Tagliabue,
†
,
‡
Artur Davoyan,
†
,
‡
Wen-Hui Cheng,
†
,
‡
and Harry A. Atwater
*
,
†
,
‡
†
Department of Applied Physics and Material Science, California Institute of Technology, Pasadena, California 91125, United States
‡
Joint Center for Arti
fi
cial Photosynthesis, California Institute of Technology, Pasadena, California 91125, United States
*
S
Supporting Information
ABSTRACT:
Electrochemical conversion of CO
2
into useful chemicals is a promising approach for transforming CO
2
into
sustainably produced fuels and/or chemical feedstocks for industrial synthesis. We report that nanoporous gold (np-Au)
fi
lms,
with pore sizes ranging from 10 to 30 nm, represent promising electrocatalytic architectures for the CO
2
reduction reaction
(CO
2
RR) due to their large electrochemically active surface area, relative abundance of grain boundaries, and ability to support
pH gradients inside the nanoporous network. Electrochemical studies show that np-Au
fi
lms support partial current densities for
the conversion of CO
2
to CO in excess of 6 mA cm
−
2
at a Faradaic e
ffi
ciency of
∼
99% in aqueous electrolytes (50 mM K
2
CO
3
saturated with CO
2
). Moreover, np-Au
fi
lms are able to maintain Faradaic e
ffi
ciency greater than 80% for CO production over
prolonged periods of continuous operation (110 h). Electrocatalytic experiments at di
ff
erent electrolyte concentrations
demonstrate that the pore diameter of nanoporous cathodes represents a critical parameter for creating and controlling local pH
gradients inside the porous network of metal ligaments. These results demonstrate the merits of nanoporous metal
fi
lms for the
CO
2
RR and o
ff
er an interesting architecture for highly selective electrocatalysis capable of sustaining high catalytic currents over
prolonged periods.
KEYWORDS:
CO
2
reduction, nanoporous cathode, pH gradient, grain boundaries, electrocatalysis
T
he ability to reduce CO
2
into useful chemicals or fuels
will not only enable clean technology but also close the
carbon cycle by recycling CO
2
and preventing its further
addition to the atmosphere.
1
The CO
2
reduction products can
either be liquid fuels such as ethanol or gaseous products such
as syngas (H
2
and CO), which are feedstocks for
thermocatalytic transformations via the Fischer
−
Tropsch
process.
2
−
5
To date, CO
2
reduction is not a widespread
technology because of low energy e
ffi
ciency associated with
high overpotentials, a lack of electrocatalytic stability, and poor
selectivity for the CO
2
reduction reaction (CO
2
RR) over the
H
2
evolution reaction (HER), which results in low partial
current densities for the product of interest.
6
Various approaches have been explored to improve the
activity and selectivity of Au-based electrocatalysts for the
CO
2
RR, from controlling nanocrystal size to tailoring of the
exposed crystal facets, and even surface functionalization with
molecular coatings.
7
−
12
Recently, nanoporous catalytic archi-
tectures have shown promise for electrochemical CO
2
reduction due to their large internal surface area and
prevalence of stepped sites and grain boundaries inherent in
their complex structure of highly curved metal ligaments.
13
−
19
This propensity for under-coordinated atomic sites has been
suggested to play a pivotal role in improving the selectivity of
CO
2
reduction in nanoporous silver (np-Ag) cathodes by
stabilizing CO
2
−
intermediates involved in the electrochemical
conversion of CO
2
to CO.
13
Similar mechanisms have been
invoked to explain the electrocatalytic performance of
nanoporous gold (np-Au)
fi
lms.
18
,
19
While the highly irregular
surface atomic structure of np-Au is well-known, relatively less
Received:
September 17, 2018
Accepted:
December 26, 2018
Published:
December 26, 2018
Letter
www.acsaem.org
Cite This:
ACSAppl.EnergyMater.
2019, 2, 164
−
170
© 2018 American Chemical Society
164
DOI:
10.1021/acsaem.8b01570
ACSAppl.EnergyMater.
2019, 2, 164
−
170
This article is made available for a limited time sponsored by ACS under the ACS Free to
Read License, which permits copying and redistribution of the article for non-commercial
scholarly purposes.
Downloaded via CALIFORNIA INST OF TECHNOLOGY on January 29, 2019 at 16:30:02 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
attention has been devoted to exploring how molecular
transport into and out of such a tightly con
fi
ned catalytic
system may also a
ff
ect the selectivity for CO
2
reduction within
the porous network. Mesoporous Au
fi
lms with controlled pore
sizes around 200 nm in diameter have previously been shown
to exhibit increased selectivity for CO
2
reduction with
increasing
fi
lm thickness from 0.5 to 2.7
μ
m.
20
The improved
selectivity is attributed to the formation of a pH gradient
within the porous network as protons are consumed during
electrolysis faster than they can be replenished by the
supporting electrolyte; this e
ff
ect is increased with increasing
thickness of the mesoporous metal cathode. Although this
study only adjusted the overall
fi
lm thickness, these
observations strongly suggest that
fi
ne-tuning the metal
porosity by controlling the pore size could o
ff
er a simple
route to further improving the selectivity of porous cathodes
for electrochemical CO
2
reduction in aqueous electrolytes.
Here, we use np-Au
fi
lms with pore diameters on the order
of tens of nanometers to explore the in
fl
uence of metal
porosity on the selectivity for CO
2
reduction in aqueous
electrolytes. Due to their small pore diameters, the porous
network of metal ligaments is able to sustain pH gradients
within np-Au
fi
lms that are half as thick (
∼
800 nm) as those
previously reported in mesoporous Au
fi
lms (
∼
2
μ
m).
20
This
e
ff
ect becomes more prominent upon further decreasing the
pore diameter from
∼
30 to
∼
10 nm, as evidenced by
electrochemical studies. We
fi
nd that np-Au
fi
lms are highly
selective for the conversion of CO
2
to CO with high Faradaic
e
ffi
ciency (FE
∼
99%) at modest overpotentials (
η
= 0.40 V),
while at the same time delivering large partial current densities
for CO (
J
CO
= 6.2 mA cm
−
2
). Finally, we demonstrate that
these np-Au
fi
lms exhibit excellent electrochemical durability
and maintain Faradaic e
ffi
ciency of
∼
80% for CO production
over 4.5 days of continuous electrolysis at an applied potential
of
E
=
−
0.5 V vs the reversible hydrogen electrode (RHE).
The np-Au
fi
lms were fabricated by electron-beam
deposition of Ag and Au alloys onto clean silicon (Si)
substrates, followed by selectively etching Ag from the Ag/Au
alloy with nitric acid (see Experimental Methods in the
Supporting Information
). Brie
fl
y, 2 nm of titanium (Ti) was
initially deposited onto Si as an adhesion layer, followed by 50
nm of Au as a planar, nonporous base layer to support the np-
Au structure and to ensure that the Ti is not exposed to the
electrolyte. The Au/Ag alloy (25/75 (vol %)) was then co-
deposited in the electron beam, ranging in thickness from 100
nm to 2
μ
m. The np-Au morphology was then obtained by
etching the Au/Ag alloy
fi
lms in 70 wt % nitric acid for 10 min
at room temperature (
∼
22
°
C; denoted RT np-Au) or at low
temperature (
−
20
°
C; denoted LT np-Au) in a freezer.
Secondary ion mass spectrometry (SIMS) indicates that
approximately 1.3 at. % of residual Ag remains in the structure
after etching, consistent with prior reports.
21
Figure 1
shows
helium focused ion beam (He FIB) images of np-Au samples
that were etched at room temperature (
Figure 1
a) and at low
temperature (
Figure 1
b), displaying average ligament thick-
nesses of 28
±
8 and 10
±
2 nm, respectively. Chemical
etching at low temperatures restricts the surface mobility of the
Au atoms during etching and ensures that the ligament
diameter is decreased.
22
A scanning electron microscope
(SEM) cross-section image of the
∼
800 nm thick RT-etched
np-Au
fi
lm shows that the entirety of the
fi
lm is porous down
to the planar Au base layer (
Figure 1
c). As shown in this cross-
sectional image, we routinely observed that the fully etched np-
Au
fi
lms were approximately 20% thinner (
∼
800 nm) than the
initial thickness of the Au/Ag alloy (
∼
1
μ
m). To characterize
the electrochemical surface area of the np-Au
fi
lms, we
performed Cu underpotential deposition (UPD) experiments
(see Experimental Methods in the
Supporting Information
)
and obtained a maximum roughness factor
∼
57 for the thickest
fi
lms. It is also important to note that the surface area increases
linearly with
fi
lm thickness, indicating that the entire surface
area of the np-Au
fi
lm is electrochemically accessible (
Figure
1
d).
To estimate the average grain size of the np-Au
fi
lms, we
performed X-ray di
ff
raction (XRD) on a RT-etched sample of
∼
800 nm thickness (
Supporting Information
Figure S1a). For
comparison, we also examined a 50 nm thick planar Au base
layer and a commercial Au foil (Alfa Aesar, 99.9975%;
Figure
S1b,c
). These data show that the average full width at half-
maximum of the di
ff
raction peaks from np-Au are larger than
the Au foil (
Figure S1d
). According to the Scherrer equation,
23
the np-Au
fi
lm and Au foil have average grain sizes of 20
±
4
nm and 77
±
23 nm, respectively. These calculations assume a
shape factor of unity and do not take into account the
possibility of microstrain.
24
We also performed transmission
electron microscopy (TEM) to directly visualize the
distribution of grain boundaries within individual ligaments
of the np-Au
fi
lm. Consistent with prior reports,
25
−
27
we
observed many grain boundaries along the surface of the
curved Au ligaments (
Figure S2
). Dark-
fi
eld TEM images were
also collected to estimate the average grain size within the np-
Figure 1.
Helium FIB images of (a) top-down view of a nanoporous
Au (np-Au)
fi
lm that was etched at room temperature (RT) and (b)
top-down view of a np-Au
fi
lm that was etched at low temperature
(LT). (c) SEM cross-section image of a RT-etched np-Au
fi
lm. All
scale bars represent 100 nm. (d) Electrochemical surface area
enhancement as a function of
fi
lm thickness for RT-etched np-Au
fi
lms as determined by Cu underpotential deposition (UPD)
experiments.
ACS Applied Energy Materials
Letter
DOI:
10.1021/acsaem.8b01570
ACSAppl.EnergyMater.
2019, 2, 164
−
170
165
Au
fi
lm (
Figure S3
and
Figure S4
). The average grain size that
we observed in TEM (17
±
4 nm) is very similar to the
estimate obtained through analysis of the XRD data. It is
known that grain boundaries and under-coordinated sites in
Au electrocatalysts serv
e as the active sites for CO
2
reduction,
28
,
29
suggesting that these np-Au
fi
lms should exhibit
substantial activity for CO
2
reduction.
A two-compartment electrochemical cell was used to
evaluate the electrocatalytic properties of the np-Au
fi
lms for
the CO
2
RR (see Experimental Methods in the
Supporting
Information
). An anion exchange membrane separated the Pt
foil anode from the np-Au cathode in 50 mM K
2
CO
3
electrolyte saturated with dissolved CO
2
. It is known that
higher supporting electrolyte concentrations provide higher
current densities during electrolysis,
30
−
32
but we found this
concentration su
ffi
cient to enable reliable evaluation of our
electrodes. The reference electrode was a leakless Ag/AgCl
electrode; all potentials are reported relative to RHE to aid
comparison with literature. The cathode compartment was
bubbled with CO
2
at a
fl
ow rate of 5 cm
3
(STP) min
−
1
through
a glass dispersion frit to maximize the delivery of CO
2
into
solution during controlled potential electrolysis. The e
ffl
uent
gas was sent directly into a gas chromatograph (SRI
instruments) to analyze the chemical products.
Figure 2
a shows the FE of a RT np-Au
fi
lm (661
±
10 nm
thick) for both CO (
fi
lled blue bars) or H
2
(white bars) as a
function of the applied potential (
E
) from
−
0.3 to
−
1.1 V
RHE
(V vs RHE). Each data point shown in
Figure 2
represents the
average FE for CO or H
2
obtained over 2
−
3 h of continuous
electrolysis at the indicated potential with
iR
compensation. All
data were obtained from the same electrode along the potential
sweep. The RT np-Au
fi
lm exhibits a maximum FE for CO of
90% at
E
=
−
0.5 V
RHE
with the remainder of the current
producing H
2
. We note that no liquid products were detected
for any of the Au electrodes studied. Notably, the LT np-Au
fi
lm (664
±
5 nm thick) obtains a maximum FE for CO (
fi
lled
gray bars) of 99% at
E
=
−
0.5 V
RHE
and maintains at least 80%
FE for CO from
−
0.3 to
−
0.7 V
RHE
before the HER (white
bars) begins to account for a larger portion of the products at
more negative applied potentials (
Figure 2
b). To examine the
in
fl
uence of the np-Au morphology on CO
2
reduction
selectivity, we tested the activity of a 50 nm thick planar Au
fi
lm, which is the base Au layer of the np-Au electrodes. As
shown in
Figure 2
c, the planar Au
fi
lm primarily produces H
2
(white bars) across the entire potential window; the FE for CO
production (green bars) only reaches
∼
40% at
−
0.5 V
RHE
.We
also evaluated the activity of a commercial Au foil (Alfa Aeasar,
99.9975%) to con
fi
rm that our experimental conditions and
cell con
fi
guration are capable of adequately reproducing
commonly observed activity trends for Au
fi
lms.
33
As shown
in
Figure 2
d, the Au foil obtained a maximum FE for CO of
92% at
−
0.5 V
RHE
(
fi
lled red bars), consistent with prior
reports.
33
A signi
fi
cant advantage of the np-Au morphology over the
planar Au electrodes is illustrated by the high partial current
density for CO (
J
CO
) relative to H
2
(
Figure 2
e
−
h). At an
applied potential of
−
0.7 V
RHE
, the LT np-Au
fi
lm exhibits a
peak
J
CO
of 8.1 mA cm
−
2
(
Figure 2
f), which is four times
higher than the Au foil (
Figure 2
h) and eight times higher than
the planar Au
fi
lm (
Figure 2
g). At the optimum applied
potential for CO production (
−
0.5 V
RHE
), the LT np-Au
fi
lm
displays
J
CO
of nearly 6.2 mA cm
−
2
, while the RT np-Au
fi
lm
J
CO
is around 4.5 mA cm
−
2
. Interestingly, the LT np-Au
fi
lm
shows only a slight increase in
J
CO
as compared to that of RT
np-Au despite the
∼
3 times increase in surface area between
the LT and RT np-Au
fi
lms (
Figure S5a
). This lower than
expected
J
CO
from LT np-Au
fi
lms likely arises due to mass
transport limitations, whereby the geometry of the electro-
chemical cell does not allow for su
ffi
cient delivery of CO
2
throughout the porous electrode to keep up with the
Figure 2.
Electrochemical performance of Au cathodes. Faradaic e
ffi
ciency (FE) for CO (
fi
lled bars) and H
2
(open bars) as a function of applied
potential (
E
) with (a) 661
±
10 nm thick RT-etched np-Au
fi
lm, (b) 664
±
5 nm thick LT-etched np-Au
fi
lm, (c) planar Au
fi
lm, and (d)
commercial Au foil. Partial current density (
J
) for CO (
fi
lled circles) and H
2
(open circles) as a function of applied potential for (e) 661
±
10 nm
thick RT np-Au
fi
lm, (f) 664
±
5 nm thick LT np-Au
fi
lm, (g) planar Au
fi
lm, and (h) commercial Au foil. Each data point represents the average
FE for CO or H
2
obtained over 2
−
3 h of continuous electrolysis at the indicated potential with
iR
compensation. The partial current densities also
represent the average value observed over the same time period. All data were obtained from the same electrode along the potential sweep.
ACS Applied Energy Materials
Letter
DOI:
10.1021/acsaem.8b01570
ACSAppl.EnergyMater.
2019, 2, 164
−
170
166
electrochemically active surface area of the entire
fi
lm. This
hypothesis is con
fi
rmed by comparing the FE and
J
CO
for RT-
etched samples that were
∼
800 and
∼
150 nm thick (
Figure
S6
). Despite the drastic di
ff
erence in electrochemical surface
area as determined by Cu UPD (
Figure 1
d), the thicker
fi
lm
only showed a
∼
30% increase in
J
CO
under our experimental
conditions (
Figure S6c
). Note that the linear relationship
between the surface area of the np-Au
fi
lm and alloy thickness
(
Figure 1
d) implies that the entire network is accessible to the
electrolyte, suggesting that a large fraction of dissolved CO
2
does not penetrate the entire depth of the
fi
lm at the current
densities studied. These results suggest that to better use the
full electrocatalytic surface area of np-Au for CO
2
reduction
requires that the geometry of the cell be modi
fi
ed to
fl
ow the
CO
2
directly through the np-Au
fi
lm so that CO
2
is e
ffi
ciently
delivered to the catalyst, as opposed to simply
fl
owing the CO
2
past the electrode surface. Indeed, it has recently been shown
that such a tactic is highly bene
fi
cial for improving the rate of
electrochemical CO
2
reduction.
34
,
35
Both the RT and LT np-Au
fi
lms exhibit superior FE for CO
(
Figure 2
a,b) relative to the planar Au
fi
lm or Au foil (
Figure
2
c,d) across the entire potential window studied. It has
previously been shown that the residual Ag in the np-Au
fi
lm is
not the source of the high FE for CO.
18
We therefore attribute
such signi
fi
cant improvements in catalytic selectivity to the
prevalence of grain boundaries that exist within the np-Au
structure relative to the planar Au
fi
lm. Another factor that
likely contributes to such marked improvements in selectivity
is the ability of the np-Au
fi
lm to support locally alkaline pH
conditions within the porous network as protons are consumed
during electrolysis. Such an e
ff
ect has previously been observed
in mesoporous Au electrodes, which serves to suppress the rate
ofHERwhiletherateofCO
2
reduction is relatively
una
ff
ected.
20
To examine the in
fl
uence of local pH gradients within the
nanoporous network on the selectivity of np-Au
fi
lms, we
examined the electrochemical activity of a RT and a LT np-Au
fi
lm (
Figure 3
d,e) at two di
ff
erent electrolyte concentrations
(50 mM K
2
CO
3
and 200 mM K
2
CO
3
both fully saturated
with CO
2
). Increasing the electrolyte concentration increases
the bu
ff
ering capacity of the solution,
20
,
32
which reduces any
swings in local pH that are anticipated to form within the pores
of the np-Au
fi
lms as protons are consumed during electrolysis.
It was therefore anticipated that the np-Au
fi
lms would exhibit
reduced FE for CO in 200 mM K
2
CO
3
electrolyte if an
increased solution pH within the porous network was
responsible for the high selectivity observed on the np-Au
fi
lms. As shown in
Figure 3
a,b, the selectivity on both RT np-
Au and LT np-Au is essentially unchanged at low applied
potentials (
−
0.3 V
RHE
), but as the current density increases
with increased applied bias (
Figure 3
d,e), any pH gradient that
may form within the np-Au
fi
lm in the 50 mM K
2
CO
3
electrolyte (
Figure 3
a,b (circles)) is diminished by the
improved bu
ff
ering capacity of the 200 mM K
2
CO
3
electrolyte
(
Figure 3
a,b (diamonds)). In contrast, no change in FE for CO
is observed on a planar Au
fi
lm at any applied potential (
Figure
3
c), con
fi
rming that the change in selectivity observed on the
np-Au electrodes is not simply a consequence of the increased
Figure 3.
In
fl
uence of electrolyte concentration on CO
2
reduction selectivity with Au cathodes. (a
−
c) Faradaic e
ffi
ciency for CO as a function of
applied potential (
E
) obtained at two di
ff
erent electrolyte concentrations (both saturated with CO
2
) for (a) 809
±
15 nm thick RT-etched np-Au
fi
lm, (b) 821
±
22 nm thick LT-etched np-Au
fi
lm, and (c) 50 nm thick planar Au
fi
lm. (d
−
f) Corresponding average current density (
J
) obtained
at
E
observed at two di
ff
erent electrolyte concentrations for (d) RT np-Au, (e) LT np-Au, and (f) planar Au
fi
lm. (g
−
i) Predicted solution pH at
the surface of the electrode for (g) RT np-Au, (h) LT np-Au, and (i) planar Au
fi
lm. A planar electrode geometry is assumed for the simulations.
ACS Applied Energy Materials
Letter
DOI:
10.1021/acsaem.8b01570
ACSAppl.EnergyMater.
2019, 2, 164
−
170
167
electrolyte concentration (
Figure 3
c,f). These results strongly
suggest that a pH gradient is forming within the porous
network of the Au electrocatalyst and serves an important role
in achieving high selectivity of CO
2
reduction in aqueous
electrolytes.
Interestingly, the LT np-Au
fi
lm shows a larger reduction in
FE for CO (
Figure 3
b) than the RT np-Au
fi
lm (
Figure 3
a).
Analysis of the pore sizes between these two samples reveals
that the LT np-Au
fi
lm has pores with an average diameter of
10
±
2 nm, while the RT np-Au sample has pores with an
average diameter of 28
±
8nm(
Figure S5b
). We therefore
attribute the improved selectivity of the LT np-Au
fi
lm relative
to the RT np-Au
fi
lm to the smaller pore size of the former
(
∼
10 nm) relative to the latter (
∼
30 nm), which more
e
ff
ectively supports a high local pH within the porous network
that improves the selectivity for CO
2
RR relative to HER
(
Figure 2
a,b). Previous work has shown that increasing the
thickness of a mesoporous Au catalyst to
∼
2
μ
m helps to
achieve a similar e
ff
ect within
∼
200 nm pores.
20
Our
observations suggest that further reducing the pore volume
below 10 nm may enable realizing such an e
ff
ect within even
thinner nanoporous metal
fi
lms than those studied here.
To further explore whether the pH gradient is developed
within the porous Au network or occurs largely in the
boundary layer, we simulated the pH pro
fi
le as a function of
distance away from the electrode surface using the model
previously reported by Gupta et al.
30
Brie
fl
y, the model
assumes a planar electrode geometry, which is a valid
assumption for calculating the pH at the outer surface of the
electrode because the
fl
ux of reactants and products must be
the same for either a porous or planar electrocatalyst at this
location. The assumption of a planar electrode is clearly
incapable of accounting for changes in the transport of
reactants and products into and out of the porous
fi
lm itself,
and we therefore interpret any experimental deviations from
the model to originate from changes occurring within the
porous network of metal ligaments. The inputs into the model
are the electrolyte concentration, the total current density, and
the Faradaic e
ffi
ciency for CO and H
2
.A70
μ
m thick
boundary layer was assumed based on the experimental
fl
ow
rate of CO
2
of 5 cm
3
(STP) min
−
1
through the catholyte.
31
As
shown in
Figure 3
g
−
i, these calculations predict very little
change in local pH at the electrode surface between the two
electrolyte concentrations, albeit small deviations from the
bulk electrolyte are predicted for the RT np-Au
fi
lm and the
planar Au electrode (
Figure 3
g,i). While signi
fi
cant reductions
in FE for CO were observed on both the RT and LT np-Au
fi
lms (
Figure 3
a,b, respectively), no change in FE was observed
experimentally on the planar Au
fi
lm (
Figure 3
c). This obvious
contradiction between the results of experiment with those
from the model indicates that the local pH changes must be
occurring within the porous network itself. Otherwise, we
would have observed a similar reduction in FE for CO with the
planar electrode at the higher electrolyte concentration. We
note that these experimental observations are consistent with a
previous report on mesoporous Au
fi
lms,
20
yet are achieved
with much thinner
fi
lms. Taken together, these results indicate
that the pore diameter of porous metal electrocatalysts is a
critical parameter for optimizing their selectivity and suggest
that control over the pore size on the nanometer length scale
may o
ff
er further improvements in electrochemical selectivity.
We further evaluated the electrocatalytic stability of these Au
fi
lms for the CO
2
RR at an applied potential of
E
=
−
0.5 V
RHE
(with
iR
compensation). Signi
fi
cantly, we observed that the np-
Au
fi
lm maintained a high FE for CO (nearly 90%) over the
course of 24 h of continuous electrocatalytic testing (
Figure
4
a). In stark contrast, the Au foil and planar Au
fi
lms exhibit
drastic reductions in FE for CO over just 1 day of testing at the
same applied potential (
Figure 4
b,c). Continued testing of a
di
ff
erent RT np-Au
fi
lm for 4.5 days (110 h) showed continued
catalytic stability (
Figure S7
). Comparison of the SEM images
before and after testing show no signi
fi
cant changes in
morphology except that the np-Au ligaments appear to coarsen
slightly (
Figure S8
). Analysis of these
fi
lms by XRD indicates
no signi
fi
cant changes in peak width before and after testing,
but all
fi
lms showed a decrease in the overall signal magnitude
from di
ff
raction peaks associated with high-index re
fl
ections
(
Figure S9
). We note that the activity of the Au foil can be
recovered if the
fl
ame annealing treatment is repeated, but
such a process is undesirable as it hinders long-term continual
operation under CO
2
RR conditions. These observations serve
to highlight the bene
fi
t of using the nanoporous metal
structure to perform CO
2
reduction: the prevalence of grain
boundaries o
ff
ers numerous active sites on the metal ligaments,
while the porous network is able to support a locally alkaline
Figure 4.
Extended electrochemical stability data for Au cathodes.
The Faradaic e
ffi
ciency for CO (
fi
lled circles) and H
2
(open circles)
was measured every 15 min via gas chromatography over the course of
24 h at an applied potential of
E
=
−
0.5 V
RHE
with
iR
compensation
for (a) RT-etched np-Au, (b) planar Au
fi
lm, and (c) Au foil.
ACS Applied Energy Materials
Letter
DOI:
10.1021/acsaem.8b01570
ACSAppl.EnergyMater.
2019, 2, 164
−
170
168
pH within the
fi
lm that helps improve electrocatalytic
selectivity for the CO
2
RR over the HER.
In conclusion, we have demonstrated that np-Au
fi
lms
constitute a promising electrocatalytic architecture for CO
2
reduction to yield CO in aqueous electrolytes. The np-Au
fi
lms
exhibit a maximum Faradaic e
ffi
ciency for CO of 99% at
−
0.5
V
RHE
while operating at a partial current density for CO in
excess of 6 mA cm
−
2
. We attribute the catalytic performance of
np-Au to its high electrochemical surface area possessing a
large number of grain boundaries and its ability to support a
local depletion of protons within the porous network.
Signi
fi
cantly, these np-Au
fi
lms maintain a Faradaic e
ffi
ciency
of greater than 80% over the course of 110 h of continuous
electrolysis at
−
0.5 V
RHE
, while the activity and selectivity of
both planar Au
fi
lms and Au foils diminishes signi
fi
cantly over
much shorter periods of operation (
∼
4 h). These studies
highlight the bene
fi
ts of nanoporous metal cathodes for CO
2
reduction and indicate that the pore size is an important
parameter to control for improving selectivity in these
promising electrocatalytic architectures.
■
ASSOCIATED CONTENT
*
S
Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website
at DOI:
10.1021/acsaem.8b01570
.
Materials; fabrication of nanoporous Au
fi
lms; scanning
electron microscopy characterization; helium focused
ion beam characterization; X-ray di
ff
raction character-
ization; Cu underpotential deposition on Au
fi
lms;
electrocatalytic reduction reaction experiments; analysis
of chemical products (
PDF
)
■
AUTHOR INFORMATION
Corresponding Author
*
E-mail:
haa@caltech.edu
;.
ORCID
Joseph S. DuChene:
0000-0002-7145-323X
Wen-Hui Cheng:
0000-0003-3233-4606
Harry A. Atwater:
0000-0001-9435-0201
Author Contributions
A.J.W., J.S.D., G.T., and H.A.A. conceived of the experimental
study. A.J.W. and J.S.D. executed all electrochemical experi-
ments and performed the data analysis. W.-H.C. assisted with
gas chromatography and high-pressure liquid chromatography.
A.J.W., J.S.D., and H.A.A. wrote the paper, and all authors
commented on the manuscript.
Notes
The authors declare no competing
fi
nancial interest.
■
ACKNOWLEDGMENTS
This work is done within the Joint Center for Arti
fi
cial
Photosynthesis, a Department of Energy (DOE) Energy
Innovation Hub, supported through the O
ffi
ce of Science of
the U.S. Department of energy under Award Number De-
SC0004993. A.J.W. acknowledges support from the National
Science Foundation (NSF) Graduate Research Fellowship
Program under Base Award No. 1745301. G.T. acknowledges
support from the Swiss National Science Foundation through
the Advanced Mobility Fellowship, grant n. P300P2_171417.
We gratefully acknowledge critical support and infrastructure
provided for this work by the Kavli Nanoscience Institute at
Caltech. We thank Matthew S. Hunt of the Kavli Nanoscience
Institute at Caltech for assistance with SEM, He FIB, and TEM
imaging of nanoporous Au
fi
lms. Any opinions,
fi
ndings, and
conclusions expressed in this material are those of the authors
and do not necessary re
fl
ect those of DOE or NSF.
■
REFERENCES
(1) Jones, J.-P.; Prakash, G. K. S.; Olah, G. A. Electrochemical CO
2
Reduction: Recent Advances and Current Trends.
Isr. J. Chem.
2014
,
54
, 1451
−
1466.
(2) White, J. L.; Baruch, M. F.; Pander, J. E.; Hu, Y.; Fortmeyer, I.
C.; Park, J. E.; Zhang, T.; Liao, K.; Gu, J.; Yan, Y.; Shaw, T. W.;
Abelev, E.; Bocarsly, A. B. Light-Driven Heterogeneous Reduction of
Carbon Dioxide: Photocatalysts and Photoelectrodes.
Chem. Rev.
2015
,
115
, 12888
−
12935.
(3) Whipple, D.; Kenis, P. Prospects of CO
2
Utilization via Direct
Heterogeneous Electrochemical Reduction.
J. Phys. Chem. Lett.
2010
,
1
, 3451
−
3458.
(4) Kuhl, K. P.; Hatsukade, T.; Cave, E. R.; Abram, D. N.;
Kibsgaard, J.; Jaramillo, T. F. Electrocatalytic Conversion of Carbon
Dioxide to Methane and Methanol on Transition Metal Surfaces.
J.
Am. Chem. Soc.
2014
,
136
, 14107
−
14113.
(5) Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F. New
Insights into the Electrochemical Reduction of Carbon Dioxide on
Metallic Copper Surfaces.
Energy Environ. Sci.
2012
,
5
, 7050
−
7059.
(6) Hori, Y. Electrochemical CO
2
Reduction on Metal Electrodes. In
Modern Aspects of Electrochemistry
; Vayenas, C. G., White, R. E.,
Gamboa-Aldeco, M. E., Eds.; Springer: New York, NY, USA, 2008; pp
89
−
189.
(7) Cao, Z.; Kim, D.; Hong, D.; Yu, Y.; Xu, J.; Lin, S.; Wen, X.;
Nichols, E. M.; Jeong, K.; Reimer, J. A.; Yang, P.; Chang, C. J. A
Molecular Surface Functionalization Approach to Tuning Nano-
particle Electrocatalysts for Carbon Dioxide Reduction.
J. Am. Chem.
Soc.
2016
,
138
, 8120
−
8125.
(8) Fang, Y.; Flake, J. C. Electrochemical Reduction of CO
2
at
Functionalized Au Electrodes.
J. Am. Chem. Soc.
2017
,
139
, 3399
−
3405.
(9) Zhu, W.; Michalsky, R.; Metin, O
̈
.; Lv, H.; Guo, S.; Wright, C. J.;
Sun, X.; Peterson, A. A.; Sun, S. Monodisperse Au Nanoparticles for
Selective Electrocatalytic Reduction of CO
2
to CO.
J. Am. Chem. Soc.
2013
,
135
, 16833
−
16836.
(10) Zhu, W.; Zhang, Y.-J.; Zhang, H.; Lv, H.; Li, Q.; Michalsky, R.;
Peterson, A. A.; Sun, S. Active and Selective Conversion of CO
2
to
CO on Ultrathin Au Nanowires.
J. Am. Chem. Soc.
2014
,
136
, 16132
−
16135.
(11) Chen, Y.; Li, C. W.; Kanan, M. W. Aqueous CO
2
Reduction at
Very Low Overpotential on Oxide-Derived Au Nanoparticles.
J. Am.
Chem. Soc.
2012
,
134
, 19969
−
19972.
(12) Liu, M.; Pang, Y.; Zhang, B.; De Luna, P.; Voznyy, O.; Xu, J.;
Zheng, X.; Dinh, C. T.; Fan, F.; Cao, C.; de Arquer, F. P. G.; Safaei, T.
S.; Mepham, A.; Klinkova, A.; Kumacheva, E.; Filleter, T.; Sinton, D.;
Kelley, S. O.; Sargent, E. H. Enhanced Electrocatalytic CO
2
Reduction via Field-Induced Reagent Concentration.
Nature
2016
,
537
, 382.
(13) Lu, Q.; Rosen, J.; Zhou, Y.; Hutchings, G. S.; Kimmel, Y. C.;
Chen, J. G.; Jiao, F. A Selective and Efficient Electrocatalyst for
Carbon Dioxide Reduction.
Nat. Commun.
2014
,
5
, 3242.
(14) Dutta, A.; Morstein, C. E.; Rahaman, M.; Ceden
̃
oLo
́
pez, A.;
Broekmann, P. Beyond Copper in CO
2
Electrolysis: Effective
Hydrocarbon Production on Silver-Nanofoam Catalysts.
ACS Catal.
2018
,
8
, 8357
−
8368.
(15) Sen, S.; Liu, D.; Palmore, G. T. R. Electrochemical Reduction
of CO
2
at Copper Nanofoams.
ACS Catal.
2014
,
4
, 3091
−
3095.
(16) Ma, M.; Trzes
́
niewski, B. J.; Xie, J.; Smith, W. A. Selective and
Efficient Reduction of Carbon Dioxide to Carbon Monoxide on
Oxide-Derived Nanostructured Silver Electrocatalysts.
Angew. Chem.,
Int. Ed.
2016
,
55
, 9748
−
9752.
ACS Applied Energy Materials
Letter
DOI:
10.1021/acsaem.8b01570
ACSAppl.EnergyMater.
2019, 2, 164
−
170
169