High selectivity for ethylene from carbon dioxide reduction over copper nanocube electrocatalysts

Article abstract of DOI:10.1002/anie.201412214

Nanostructured surfaces have been shown to greatly enhance the activity and selectivity of many different catalysts. Here we report a nanostructured copper surface that gives high selectivity for ethylene formation from electrocatalytic CO₂ reduction. The nanostructured copper is easily formed in situ during the CO₂ reduction reaction, and scanning electron microscopy (SEM) shows the surface to be dominated by cubic structures. Using online electrochemical mass spectrometry (OLEMS), the onset potentials and relative selectivity toward the volatile products (ethylene and methane) were measured for several different copper surfaces and single crystals, relating the cubic shape of the copper surface to the greatly enhanced ethylene selectivity. The ability of the cubic nanostructure to so strongly favor multicarbon product formation from CO₂ reduction, and in particular ethylene over methane, is unique to this surface and is an important step toward developing a catalyst that has exclusive selectivity for multicarbon products. Cubic nanostructures formed on a polycrystalline copper surface give high selectivity for ethylene formation from carbon dioxide electroreduction. The nanocubes are easily synthesized in situ, and online electrochemical mass spectrometry is used to compare the reactivity to other copper single-crystal surfaces.

Full text of DOI:10.1002/anie.201412214

Angewandte
Chemie
DOI: 10.1002/anie.201412214
Electrocatalysis
High Selectivity for Ethylene from Carbon Dioxide Reduction over
Copper Nanocube Electrocatalysts**
F. Sloan Roberts, Kendra P. Kuhl, and Anders Nilsson*
Abstract: Nanostructured surfaces have been shown to greatly
enhance the activity and selectivity of many different catalysts.
Here we report a nanostructured copper surface that gives high
selectivity for ethylene formation from electrocatalytic CO2
reduction. The nanostructured copper is easily formed in situ
during the CO2 reduction reaction, and scanning electron
microscopy (SEM) shows the surface to be dominated by cubic
structures. Using online electrochemical mass spectrometry
(OLEMS), the onset potentials and relative selectivity toward
the volatile products (ethylene and methane) were measured
for several different copper surfaces and single crystals, relating
the cubic shape of the copper surface to the greatly enhanced
ethylene selectivity. The ability of the cubic nanostructure to so
strongly favor multicarbon product formation from CO2
reduction, and in particular ethylene over methane, is unique
to this surface and is an important step toward developing
a catalyst that has exclusive selectivity for multicarbon
products.
compared the activity of single-crystal faces and found that
although most surfaces have similar activity for methane and
ethylene, the Cu(100) surface favors ethylene formation at
a lower overpotential compared to methane.[4] In an effort to
produce higher surface area catalysts with a large amount of
exposed (100) facets, there has been an interest in copper
nanocubes as catalysts for CO2RR with higher selectivity for
ethylene and better energy efficiency than unstructured
polycrystalline copper.[5]
Increasing the selectivity of copper for ethylene over
methane is desirable as ethylene is a popular chemical
commodity used in industrial applications, and there is
abundant and cheap access to methane via natural gas.
Ethylene selectivity also has broader implications in the field
À
of CO2RR catalysis, because it provides insight into the C C
coupling reaction step that must occur for the formation of
multicarbon products. Understanding and controlling this
step is of paramount importance to guide the design of
catalysts with high selectivity for the desired product. Herein,
we report a simple in situ synthesis of a nanocube-covered
copper (CuCube) surface with high selectivity and low
overpotential for ethylene formation, adding to evidence
A
promising path to reducing carbon dioxide emissions into
the atmosphere is recycling carbon dioxide into fuels and
commodity chemicals through an electrochemical process.[1]
Electrocatalyst development for the carbon dioxide reduction
reaction (CO2RR) is key to enabling the widespread adoption
of this technology. Many studies have focused on copper
metal as a catalyst, because it is known to produce a mixture
of methane and ethylene and other minor products
(Table S1).[2] However, improvements in energy efficiency,
achieved by lowering the overpotential of the CO2RR, and in
selectivity, by increasing the yield of the desired product, are
needed.
À
that (100) sites are highly active for C C coupling and can
lead to developing strategies to target multicarbon products.
Successive oxidative–reductive cycles in the presence of
KCl changes the surface structure of the starting polycrystal-
line copper into the CuCube surface (Figure 1A), composed
of monolithic structures made up of predominantly cubic
faces, in contrast to the copper surface cycled without chloride
ions (Figure 1B), which has few distinct features. The addition
of KCl likely promotes the formation of the CuCube structure
through the initial formation of CuCl under oxidizing
conditions.[6] CuCl is known to form cubes when it precip-
itates in solutions above pH 4.[7] In the presence of water,
CuCl can convert to Cu2O through an equilibrium reaction
which favors copper oxide formation in neutral and basic
solutions with low chloride concentrations,[6,7] as present in
this experiment. XPS of CuCube surfaces removed from
solution at oxidative potentials show that only trace amounts
of Cl are present (Figure S5), suggesting that CuCl formed at
these oxidative potentials has been converted to Cu2O. XRD
(Figure S4), which probes the bulk rather than the surface,
shows that only copper and Cu2O phases are detected in the
CuCube sample. These results support the mechanism of
CuCube formation described above, in which CuCl formation
and precipitation is responsible for the cubic shape, but is then
converted to Cu2O. It is worth mentioning that the Cu2O
surface is subsequently reduced before the CO2RR begins,
leaving metallic copper (now in a cubic structure) as the
active catalytic surface.
Recent work has shown that modifying the structure of
the copper surface is a path to better efficiency and selectivity.
Oxidizing and then reducing copper metal leads to a highly
nanostructured surface, with much lower overpotential
(higher energy efficiency) for the CO2RR.[3] Other work has
[*] Dr. F. S. Roberts,[+] Dr. K. P. Kuhl,[+] Prof. A. Nilsson
Department of Chemistry, Stanford University
Stanford, CA 94305 (USA)
E-mail: nilsson@slac.stanford.edu
[+] These authors contributed equally to this work.
[**] This work is supported by Air Force Office of Scientific Research
through the MURI program under AFOSR Award No. FA9550-10-1-
0572 and the Global Climate Energy Project at Stanford University.
The XPS, XRD, and SEM studies were performed at the Stanford
Nano Center (SNC)/Stanford Nanocharacterization Laboratory
(SNL) part of the Stanford Nano Shared Facilities.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201412214.
Angew. Chem. Int. Ed. 2015, 54, 1 – 5
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
respectively, and at the negative end of the CV scan, nearly
twice as much methane is detected than ethylene. Figure 1D
shows that polycrystalline copper and CuCube have similar
onsets for hydrogen formation, but the CuCube sample
produces more hydrogen overall, which is consistent with the
fact that also more overall current is drawn on the CuCube
sample. The onset for methane production on the two samples
is also very similar (albeit the intensity for methane produc-
tion on CuCubes is almost negligible), but their onset for and
selectivity toward ethylene is very different. CuCube has an
onset of À0.6 for ethylene formation (vertical dashed line),
which is more positive than the polycrystalline sample, and
also produces a relatively large amount of ethylene across the
voltage range measured. In fact, with nearly complete
suppression of the methane signal, the selectivity for the
CuCube sample toward ethylene has increased by more than
two orders of magnitude compared to polycrystalline copper.
To better understand the dramatic increase in the ethylene
to methane selectivity of the CuCube sample, the CO2RR
activity of three different copper single-crystal surfaces,
Cu(111), Cu(211), and Cu(100), was studied in the same
setup. The cubic structure present on the CuCube sample
suggests the surface is dominated by the (100) facet of copper
and (100) step sites. By comparing the activity of the single
crystals to the CuCube surface, we can begin to understand
what leads to its high activity.
Figure 1. SEM images of A) the CuCube surface formed by cycling
a polycrystalline electrode between À1.15 and 0.9 V three times with
4 mm KCl in 0.1m KHCO3 and B) a polycrystalline copper electrode
after cycling in the same voltage region three times without KCl in
0.1m KHCO3. C) Cyclic voltammograms of CuCube (red) and polycrys-
talline (black) copper surfaces and corresponding hydrogen formation
measured by OLEMS shown in (D). Methane (dashed) and ethylene
(solid) formation measured by OLEMS on E) the CuCube surface and
F) polycrystalline copper, showing a dramatic shift in selectivity toward
ethylene and away from methane.
Figure 2 shows the results for the single-crystal experi-
ments with Figure 2A displaying the CVs for the three
Figure 1C shows the CVs of a polycrystalline copper
electrode in aqueous 0.1m KHCO3 without (black) and with
(red) the addition of 4 mm KCl. The addition of KCl leads to
significant changes in the CV. At oxidative potentials (0.5 to
0.9 V vs. RHE) orders of magnitude more current is drawn
when chloride ions are present and the oxidized copper that is
formed at high potentials is reduced in the large negative peak
between À0.2 and À0.5 V. Going to voltages lower than
À0.5 V, the CuCube sample (red), formed by successive
oxidative–reductive scans in the presence of KCl, draws more
current than polycrystalline copper without the addition of
KCl. The higher current density drawn by the CuCube sample
negative of À0.5 V is likely due to the higher surface
roughness of the nanostructured surface compared to the
electropolished polycrystalline Cu surface (see the Support-
ing Information, SI).
Measurements of the products formed from the CO2RR
(and competing water reduction to hydrogen) were made
using an OLEMS system (see the Experimental Section and
SI for a detailed description). The data for polycrystalline
copper (black) are consistent with reported results.[2a] Poly-
crystalline copper has an early onset for hydrogen production
compared to any CO2RR products. Methane and ethylene,
CO2RR products, have onsets around À0.90 V and À0.75 V,
Figure 2. A) Comparison of single-crystal CVs in 0.1m KHCO3. Meth-
ane and ethylene formation on B) Cu(211), C) Cu(100), and D) Cu-
(111).
surfaces and Figure 2B–D showing the ethylene and methane
products of the (211), (100), and (111) orientations, respec-
tively. Both the forward and reverse scans are plotted for the
ethylene and methane product channels and show very little
hysteresis. All three surfaces make both ethylene and
methane, and a qualitative interpretation of the results
shows the Cu(111) surface being the least active (it also
pulls the least amount of current) whereas the Cu(100) and
Cu(211) appear fairly similar in activity. Upon closer exami-
2
www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
nation it becomes clear that the Cu(100) has a higher
selectivity for ethylene at lower overpotential (vertical
dashed line in Figure 2) and there is in fact a range of
potentials on the Cu(100) surface at which a significant
amount of ethylene is produced and methane is not (À0.8 to
À1.0 V).
recent reports estimate that at typical current densities and
electrolyte concentrations, the local pH in a neutral bulk
solution of KHCO3 can be as high as 10–11.[9] Given these
results for an electrode roughness factor of 16, and the fact
that we estimate the CuCubes roughness factor to be about
20, it is reasonable to assume that the local pH during the
CO2RR over CuCubes is also somewhere around 10.5 which
could shift, somewhat, the selectivity toward ethylene. This
difference in local pH versus bulk pH is only thought to be
present on rough surfaces, and absent from single-crystal
surfaces, possibly contributing to the disconnect we see
between the CuCube and Cu(100) selectivities.
Table 1 summarizes the ethylene onset potentials for the
different surfaces investigated and they are arranged in order
Table 1: Onset potentials (V vs. RHE) for ethylene and methane
production from reducing CO2 in 0.1m KHCO3. The copper surfaces are
listed in order of ethylene onset.
The CuCube electrocatalyst presented here is one of the
most selective for ethylene over methane, implying that it
Surface
C2H4 onset potential
CH4 onset potential
Cu(Cube)
Cu(100)
Cu(Poly)
Cu(211)
Cu(111)
À0.60 V
À0.73 V
À0.74 V
À0.79 V
À0.96 V
À0.93 V
À0.90 V
À0.95 V
À0.94 V
À0.99 V
À
favors C C coupling to make multicarbon products instead of
completely reducing a single carbon to methane. The ability
to catalyze ethylene formation without significant methane
production indicates that these two products are most likely
formed through two distinct, competing pathways (see SI for
À
more discussion). Finding a surface that favors the C C
coupling pathway (such as the CuCube surface reported here)
is important for ultimately controlling the selectivity of the
CO2RR for multicarbon product formation.
of best to worst in terms of ethylene selectivity with CuCube
being far and away the best surface investigated. When
comparing the single crystals, Cu(100) has the most favorable
selectivity for ethylene production and the earliest onset
potential. Cu(211), which has (100) step sites, is the next best
in terms of ethylene selectivity, with Cu(111) being the worst
surface studied. The single-crystal study suggests that this
cubic surface structure is preferable for ethylene selectivity
compared to the close-packed (111) surface or the highly
stepped (211) surface.
Experimental Section
Polycrystalline copper disks with dimensions of 8 mm diameter,
2.5 mm high were machined from OHFC copper of 99.9% purity.
Copper single crystals of the same dimensions were purchased from
Princeton Scientific and are polished to an accuracy of < 0.1 degree
with a roughness < 0.01 micron (99.9999% pure). Copper working
electrodes were prepared by electropolishing in 85% phosphoric acid
at 2 V for 60 seconds (polycrystalline electrodes) or 20 seconds
(single-crystal electrodes). The electrodes were then directly sub-
merged in an aqueous electrolyte solution of 0.1m KHCO3 purged
with CO2. The resulting electrolyte pH was 6.8. Cyclic voltammo-
grams (CVs) were measured using an Ag/AgCl reference electrode
What the single-crystal studies fail to explain, however, is
the complete lack of methane formation on the CuCube
sample. Whereas the (100) surface is the most comparable to
the CuCube surface in terms of ethylene onset potential, the
(100) surface still makes a significant amount of methane. The
(211) surface, which has two-atom-wide terraces and (100)
steps, also produces significant amounts of methane and is less
selective for ethylene. Work performed by Hori et al.[8] on
single-crystal surfaces suggest there is an ideal terrace length
that gives the maximum ethylene to methane ratio with the
(100), largely devoid of steps, and the (211), having the
highest step density, the two extremes that give the smallest
ethylene to methane ratio. Hori found the Cu(711) surface to
be the most selective for ethylene, giving an ethylene to
methane ratio of 10:1. Due to the high selectivity of the
CuCubes, it is reasonable to conclude that the CuCube
surface either has a very high density of ethylene-selective
active sites compared to other CO2RR catalysts studied, or it
presents a new active site not found on other surfaces.
Another possibility that cannot be ruled out at this point is
that due to the rough nature of the CuCube surface, the local
pH during reaction is much higher than the bulk pH, and it is
this rise in local pH that improves the selectivity of the
CuCube surface. Previous work has demonstrated the impor-
tance of pH (bulk) on CO reduction over copper electrodes,
especially showing an earlier onset potential for ethylene at
higher pH.[4b,d] Similarly, local pH changes (not just bulk
changes) can affect the selectivity of copper electrodes, and
(Accumet),
a boron doped diamond counter electrode (CCL
Diamond), and a Biologic VSP 200 potentiostat scanning at a rate
of 5 mVsÀ1. The solution resistance was determined by electro-
chemical impedance spectroscopy measured after each CV at 10 kHz
and the potentiostat was set to compensate for 85% of the measured
IR drop. CuCube electrodes were prepared in situ by adding 4 mm
KCl to the standard 0.1m KHCO3 electrolyte and cycling the potential
of an electropolished polycrystalline copper disk between À1.15 V
and 0.9 V at 5 mVsÀ1 for three cycles until a stable CV was achieved.
Copper electrode surfaces were characterized by X-ray photoelectron
spectroscopy (XPS, PHI Versaprobe), X-ray diffraction (XRD,
Panalytic X’Pert 1), and scanning electron microscopy (SEM, FEI
XL30 Sirion). Capacitive scans (Figure S6) were used to estimate the
roughness factor of the CuCube electrode to around 20.
Online electrochemical mass spectrometry (OLEMS) was used to
detect the CO2RR products formed during each CV. The setup was
adapted from Koper et al.[10] and consisted of a mass spectrometer
(SRS CIS 300) measuring electrochemical reaction products that
enter the system through a porous Teflon frit (Porex 15–25 mm)
placed near the surface of the copper working electrode. Details of
the setup are shown in Figure S2. The principal improvement over the
previously reported system, was the use of custom copper disk holder.
This allowed the face of the working electrode to be placed vertically,
so that bubbles formed at high current density could escape from the
surface. Trapped bubbles can interfere with the mass transport of CO2
Angew. Chem. Int. Ed. 2015, 54, 1 – 5
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
to the surface and contribute to noise in the potentiostat and mass
spectrometer readings.
[3] C. W. Li, M. W. Kanan, J. Am. Chem. Soc. 2012, 134, 7231 – 7234.
[4] a) K. J. P. Schouten, Y. Kwon, C. J. M. van der Ham, Z. Qin,
M. T. M. Koper, Chem. Sci. 2011, 2, 1902 – 1909; b) K. J. P.
Schouten, E. Pꢀrez Gallent, M. T. M. Koper, J. Electroanal.
Chem. 2014, 716, 53 – 57; c) K. J. P. Schouten, E. Pꢀrez Gallent,
M. T. M. Koper, ACS Catal. 2013, 3, 1292 – 1295; d) K. J.
Schouten, Z. Qin, E. Perez Gallent, M. T. Koper, J. Am. Chem.
Soc. 2012, 134, 9864 – 9867.
[5] a) C. S. Chen, A. D. Handoko, J. H. Wan, L. Ma, D. Ren, B. S.
Yeo, Catal. Sci. Technol. 2015, 5, 161 – 168; b) J. Bugayong, G. L.
Griffin, ECS Trans. 2013, 58, 81 – 89.
[6] G. Bianchi, P. Longhi, Corros. Sci. 1973, 13, 853 – 864.
[7] H. Liu, Y. Zhou, S. A. Kulinich, J.-J. Li, L.-L. Han, S.-Z. Qiao, X.-
W. Du, J. Mater. Chem. A 2013, 1, 302.
[8] Y. Hori, I. Takahashi, O. Koga, N. Hoshi, J. Mol. Catal. A 2003,
199, 39 – 47.
[9] R. Kas, R. Kortlever, A. Milbrat, M. T. Koper, G. Mul, J.
Baltrusaitis, Phys. Chem. Chem. Phys. 2014, 16, 12194 – 12201.
[10] A. H. Wonders, T. H. M. Housmans, V. Rosca, M. T. M. Koper, J.
Appl. Electrochem. 2006, 36, 1215 – 1221.
Received: December 19, 2014
Revised: February 2, 2015
Published online: && &&, &&&&
Keywords: carbon dioxide · electrocatalysis ·
.
electrochemical reduction · nanomaterials
[1] a) G. A. Olah, G. K. S. Prakash, A. Goeppert, J. Am. Chem. Soc.
2011, 133, 12881 – 12898; b) N. S. Lewis, D. G. Nocera, Proc.
Natl. Acad. Sci. USA 2006, 103, 15729 – 15735.
[2] a) Y. Hori , Handbook of Fuel Cells: Fundamentals, Technology
and Application, Vol. 2 (Eds.: A. L. W. Vielstich, H. A. Gas-
teiger), John Wiley & Sons, Weinheim, 2003, pp. 720 – 733;
b) A. A. Peterson, F. Abild-Pedersen, F. Studt, J. Rossmeisl, J. K.
Norskov, Energy Environ. Sci. 2010, 3, 1311 – 1315; c) K. P. Kuhl,
E. R. Cave, D. N. Abram, T. F. Jaramillo, Energy Environ. Sci.
2012, 5, 7050 – 7059; d) K. Manthiram, B. J. Beberwyck, A. P.
Alivisatos, J. Am. Chem. Soc. 2014, 136, 13319 – 13325.
4
www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
Communications
Electrocatalysis
Cubic nanostructures formed on a poly-
crystalline copper surface give high
selectivity for ethylene formation from
carbon dioxide electroreduction. The
nanocubes are easily synthesized in situ,
and online electrochemical mass spec-
trometry is used to compare the reactivity
to other copper single-crystal surfaces.
F. S. Roberts, K. P. Kuhl,
A. Nilsson*
&&&& — &&&&
High Selectivity for Ethylene from Carbon
Dioxide Reduction over Copper
Nanocube Electrocatalysts
Angew. Chem. Int. Ed. 2015, 54, 1 – 5
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!