Probing the Active Surface Sites for CO Reduction on Oxide-Derived Copper Electrocatalysts

Article abstract of DOI:10.1021/jacs.5b06227

CO electroreduction activity on oxide-derived Cu (OD-Cu) was found to correlate with metastable surface features that bind CO strongly. OD-Cu electrodes prepared by H₂ reduction of Cu₂O precursors reduce CO to acetate and ethanol with nearly 50% Faradaic efficiency at moderate overpotential. Temperature-programmed desorption of CO on OD-Cu revealed the presence of surface sites with strong CO binding that are distinct from the terraces and stepped sites found on polycrystalline Cu foil. After annealing at 350 °C, the surface-area corrected current density for CO reduction is 44-fold lower and the Faradaic efficiency is less than 5%. These changes are accompanied by a reduction in the proportion of strong CO binding sites. We propose that the active sites for CO reduction on OD-Cu surfaces are strong CO binding sites that are supported by grain boundaries. Uncovering these sites is a first step toward understanding the surface chemistry necessary for efficient CO electroreduction.

Full text of DOI:10.1021/jacs.5b06227

Page 1 of 6
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Probing the active surface sites for CO reduction on oxide-
derived copper electrocatalysts
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Arnau Verdaguer-Casadevall,1‡ Christina W. Li,2‡ Tobias P. Johansson,1 Soren B. Scott,1 Joseph T.
McKeown,3 Mukul Kumar,4 Ifan E. L. Stephens,1 Matthew W. Kanan,2* and Ib Chorkendorff1*
1
Center for Individual Nanoparticle Functionality, Department of Physics, Building 312, Technical University of Denmark
(DTU), DK-2800, Kongens Lyngby, Denmark
2 Department of Chemistry, Stanford University, 337 Campus Drive, Stanford, CA 94305, United States
3
Materials Science Division, Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, CA 94550, United
States
4
Materials Engineering Division, Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, CA 94550,
United States
Supporting Information Placeholder
show enhanced CO reduction catalytic activity. Based on these
data, we hypothesized that grain boundaries were responsible for
creating the active sites for CO reduction. Insight into the surface
chemistry of OD-Cu is vital for extracting design principles that
may lead to improved catalysts.
ABSTRACT: CO electroreduction activity on oxide-derived Cu
(OD-Cu) was found to correlate with metastable surface features
that bind CO strongly. OD-Cu electrodes prepared by H2 reduc-
tion of Cu2O precursors reduce CO to acetate and ethanol with
nearly 50 % Faradaic efficiency at moderate overpotential. Tem-
perature programmed desorption of CO on OD-Cu revealed the
presence of surface sites with strong CO binding that are distinct
from the terraces and stepped sites found on polycrystalline Cu
foil. After annealing at 350 °C, the specific current density for CO
reduction is 44-fold lower and the Faradaic efficiency for ethanol
and acetate is less than 5%. This loss of activity is accompanied
by a reduction in the proportion of strong CO binding sites. We
propose that the active sites for CO reduction on OD-Cu surfaces
are strong CO binding sites that are supported by grain bounda-
ries. Uncovering these sites is a first step towards understanding
the surface chemistry necessary for efficient CO electroreduction.
The use of renewable energy to convert CO2 into chemicals and
fuels is an attractive means of mitigating emissions and increasing
energy security.1,2 Electrochemical CO2 conversion could be piv-
otal to this goal; electrolyzers allow for fast startup and require
little infrastructure, which would enable deployment of decentral-
ized CO2 recycling units scaled to local renewable energy sources.
One strategy for efficient electrochemical CO2 conversion is to
separate the process into two steps: CO2 reduction to CO, fol-
lowed by CO reduction to oxygenates and hydrocarbons. Numer-
ous CO2 reduction catalysts have been reported that may enable
practical CO2 to CO conversion at low temperature.3-8 Copper is
the only catalyst that has shown significant CO reduction activity
in aqueous electrolyte.9,10 However, the overpotential required to
drive CO reduction on bulk polycrystalline Cu is prohibitively
high, which indicates that the low-index terrace and stepped facets
that dominate the bulk Cu surface are inefficient catalysts. We
recently showed that nanocrystalline Cu prepared via reduction of
a Cu2O precursor, “oxide-derived Cu” (OD-Cu), has unprecedent-
ed activity for the reduction of CO to acetate, ethanol, and propa-
nol.11 Transmission electron microscopy (TEM) revealed that
OD-Cu consists of an interconnected network of nanocrystallites
with randomly-oriented grain boundaries at the crystallite inter-
faces. Dispersed nanoparticles of comparable dimensions did not
Figure 1. SEM (a-d), image quality (IQ) maps constructed from electron
diffraction data (e-h) of oxide-derived Cu electrodes. (a, e) OD-Cu oxi-
dized at 500 °C and reduced in H2 (OD-Cu–500), (b, f) OD-Cu–500 after
200 °C anneal in N2, (c, g) after 350 °C anneal in N2 (d, h) OD-Cu oxi-
dized at 300 °C and reduced in H2 (OD-Cu–300).
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 6
This study examines the impact of OD-Cu microstructure and
ple, providing information for nanostructures analogous to that
1
2
3
4
5
6
7
8
adsorbate binding thermodynamics on electrocatalytic CO reduc-
tion activity. By comparing OD-Cu as-prepared to thermally an-
nealed derivatives, we show that both the nanograined structure
and CO reduction activity of OD-Cu are metastable. Temperature-
programmed desorption (TPD) of CO is then used to probe the
binding strength of CO on the surfaces of these electrodes.12 We
show that CO reduction activity correlates with the abundance of
strong binding sites that are not found on terraces or stepped sur-
faces. TPD has been used to study high-temperature bulk catalysts
such as Cu/ZnO/Al2O3 for methanol synthesis12,13 as well as mod-
el surfaces for low-temperature electrocatalysis,14-16 but to the best
of our knowledge this is the first time TPD has been used on ac-
tive nanocrystalline electrocatalysts.
acquired using electron backscattering diffraction mapping in an
SEM.17,18 The nanodiffraction patterns are indexed to generate
grain orientation and grain boundary maps from which grain
boundary density and misorientation angles can be extracted.
From this analysis, the grain boundary densities of OD-Cu–500,
OD-Cu–500 annealed at 200 °C, and OD-Cu–300 are similar: 15
± 2 µm–1, 13 ± 1 µm–1, and 21 ± 2 µm–1, respectively (Figure S9,
S10, S11, S13). Moreover, the distribution of grain boundaries is
similar in these three electrodes, with ~37% low-energy twin
boundaries and the rest randomly oriented and non-twin coinci-
dent site lattice boundaries (Figure S9–S14). By contrast, the OD-
Cu–500 electrode annealed at 350 °C has a much lower grain
boundary density of 1.7 ± 0.1 µm–1 with ~53% low-energy twins
(Figure S9, S12).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
OD-Cu can be prepared by either electrochemical or H2 reduc-
tion of Cu oxide precursors. For this study, H2 reduction was used
to facilitate TPD experiments. OD-Cu samples were synthesized
via air oxidation at 500 °C followed by H2 reduction at 130 °C
(herein referred to as OD-Cu–500).11 Scanning electron micros-
copy (SEM) showed aggregated particles in a porous morphology.
Transmission electron microscopy nanodiffraction indicated that
the material is composed of irregularly shaped, 100–300 nm
grains that are joined by grain boundaries in a dense polycrystal-
line network (Fig. 1a, e). In order to modulate its structure and
catalytic activity, OD-Cu–500 electrodes were annealed under N2
at 200 °C and 350 °C. Annealing at 200 °C had little effect on the
grain size, while annealing at 350 °C resulted in a smoother mor-
phology and significantly increased grain sizes of ~1 µm (Fig. 1b,
c, f, g). For comparison, an OD-Cu electrode was prepared by
growing the oxide in air at 300 °C for 30 min and reducing in H2
at 130 °C (OD-Cu–300). The grain sizes in OD-Cu–300 are simi-
lar to those of OD-Cu–500 and its 200 °C–annealed derivative,
but OD-Cu–300 has a much smoother morphology (Fig 1d, h).
Cross-sectional SEM revealed that OD-Cu–300 is essentially non-
porous (Figure S6).
X-ray diffraction and X-ray photoelectron spectroscopy (XPS)
of all four samples indicate that the only bulk crystalline phase
present is Cu0, and the surface comprises primarily Cu0 with some
Cu1+ due to native oxide formation during transfer of the sample
to the XPS chamber (Figure S1–S4). The electrochemically active
surface area of each sample was assessed using an electrochemi-
cal capacitance measurement. OD-Cu–500 has a roughness factor
of 51 (where electropolished Cu foil is 1) arising from the high
surface area porous structure. The roughness factor drops to 24
and 14 after annealing at 200 °C and 350 °C, respectively. OD-
Cu–300 has a roughness factor of 6, lower than that of both ther-
mally annealed electrodes.
Electrochemical activity for CO reduction was assessed over a
range of potentials via constant potential electrolysis in 0.1 M
KOH electrolyte, continuously purged with 1 atm of CO. The CO
reduction activity of OD-Cu–500 has been previously reported.11
OD-Cu–500 exhibits peak Faradaic efficiency of 48% for CO
reduction at –0.4 V vs. RHE and generates C2 and C3 products
including acetate, ethanol, n-propanol, ethylene, and ethane (Fig
2a), measured by gas chromatography and nuclear magnetic reso-
nance (NMR). The surface-area normalized partial current density
for CO reduction (jCORedn) increases with increasing overpotential,
but rapidly reaches a plateau corresponding to the mass-transport
limit of ~1 mM dissolved CO in water (Fig 2c). 19
To further characterize the grain boundaries, orientation maps
were collected on all OD-Cu samples. This TEM-based technique
collects nanodiffraction patterns over scanned regions of a sam-
Upon annealing, the intrinsic CO reduction activity of the elec-
trode is significantly reduced. The surface-area corrected jCORedn is
attenuated 2.5-fold on the 200 °C-annealed sample and 44-fold on
the 350 °C-annealed sample relative to as-prepared OD-Cu–500
(Fig. 2c). The loss of intrinsic CO reduction activity is reflected in
the Faradaic efficiencies for CO reduction at –0.4 V vs. RHE (Fig
2a). The ratios of oxygenate to hydrocarbon products remain simi-
lar, but there is a drop in total Faradaic efficiency for CO reduc-
tion and a corresponding increase in Faradaic efficiency for H2
evolution. While the 200 °C-annealed sample still shows 30%
total Faradaic efficiency for CO reduction, the 350 °C-annealed
electrode has less than 5% efficiency, which is comparable to
what was seen previously with commercial Cu nanoparticles. The
loss of CO reduction activity upon annealing suggests that the
catalytically active sites are tied to metastable structural features
that are kinetically trapped during the low-temperature H2 reduc-
tion of the Cu2O precursor. In contrast, annealing has little effect
on the surface area–corrected partial current density for H2 evolu-
tion (jH2) (Fig 2d).
The catalytic activity of OD-Cu depends on the thickness of the
Cu2O precursor. OD-Cu–300 is prepared from a thinner oxide and
is intrinsically less active for CO reduction than OD-Cu–500. In
fact, OD-Cu–300 has essentially the same surface area–corrected
jCORedn and CO reduction Faradaic efficiency as OD-Cu–500 after
annealing at 200 °C. Since OD-Cu–300 has 4-fold lower rough-
ness factor than the 200 °C annealed electrode, this result indi-
cates that neither electrode porosity nor microscopic surface area
Figure 2. (a) Faradaic efficiency for CO reduction at –0.4 V vs. RHE in
CO-saturated 0.1 M KOH. The remainder of the charge passed is H2
evoltuion. (b) Electrochemical roughness factors (RF), measured from the
capacitance in a non-Faradaic potential regime, with Cu foil defined as 1.
(c) Specific current density for CO reduction vs. potential. (d) Specific
current density for H2 evolution vs. potential. The error bars represent one
standard deviation from the mean based on 2–5 measurements.
ACS Paragon Plus Environment

Page 3 of 6
Journal of the American Chemical Society
are significant determinants of CO reduction catalytic activity.
This conclusion is further supported by the comparison between
1
OD-Cu–500 annealed at 350 °C and OD-Cu–300. OD-Cu–300
has lower surface area but much higher CO reduction activity than
the 350 °C annealed sample.
2
3
4
In order to connect catalytic activity to the thermodynamics of
adsorbate binding on the surface, CO-TPD was used to probe the
surface chemistry of OD-Cu. To prepare OD-Cu samples for
TPD, Cu foil was oxidized externally and mounted in the UHV
chamber. Adventitious contamination from exposure to air was
removed by mildly sputtering the surface and the samples were
subsequently reduced at 130 ºC under a flow of 1 bar of H2 (see
the Supporting Information). XPS at this stage showed a metallic
Cu surface (Fig. S17) with no trace of nickel (a common contami-
nant in UHV in the presence of CO due to the formation of nickel
carbonyl in the gas lines20) or other metallic impurities. The sam-
ples were cooled down to 77 K, and CO was dosed into the cham-
ber. The temperature was then ramped up at 2 K s–1 and a mass
spectrometer monitored the signal from CO at mass 28. The an-
nealed OD-Cu–500 samples were prepared in an identical manner
to OD-Cu–500 and annealed in UHV at 200 °C or 350 °C prior to
CO dosing. The experiments were carried out with a constant
dosage of 200 L, which results in variable CO coverage given the
different surfaces areas of the samples (Fig. S15).
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3a shows a typical TPD profile for electropolished poly-
crystalline Cu foil. Based on literature data from single crystalline
Cu,12,21,22 the TPD features can be assigned to different facet ter-
minations. The lowest temperature desorption feature, from 100 K
to 200 K, is due to low index facets such as Cu(111) and Cu(100),
which bind CO weakly. The higher temperature feature between
200 K and 250 K is characteristic of stepped sites such as
Cu(211). Figure 3b shows the CO-TPD profile of OD-Cu–500.
The OD-Cu–500 profile shows a high-temperature feature cen-
tered at 275 K that is clearly absent from the profiles for polycrys-
talline Cu and single crystalline Cu but is present in different pro-
portions in the profiles for the other OD-Cu samples (Fig. 3b-e,
). Annealing OD-Cu–500 to 200 °C and 350 °C reduces the area
of the high-temperature feature (Figures 3c-d). The profile of OD-
Cu–300 exhibits broader and less-defined peaks than OD-Cu–500
and its annealed derivatives because its low roughness factor re-
sults in a higher CO coverage. Nonetheless, the high-temperature
feature is clearly still present on OD-Cu–300 (Fig. 3e).
Figure 3. TPD profiles for various Cu samples. a-e) Experimental data
(solid black lines) and resulting fit from microkinetic simulations (dashed
red lines). The dotted green lines indicate the central position of the low-
index facets and stepped sites in polycrystalline Cu. a) Polycrystalline
Cu. Features arising from low-index facets are highlighted in light green,
undercoordinated sites are highlighted in dark green; b) OD-Cu–500; c)
OD-Cu annealed at 200 °C; d) OD-Cu annealed at 350 °C; e) OD-Cu–
300; f) overlay of TPD profiles. Measurements were performed at a ramp
rate of 2 K s–1 with a CO dose of 200 L. g) Experimental CO reduction
activity at –0.4 V as a function of the percentage of strong binding sites,
estimated from simulations of the TPD profiles. The solid black line is a
linear fit through all points. The error bars are a means of showing uncer-
tainty in the simulations and have been obtained by varying pore depth in
the model by ±5 µm or 50 %, whichever is lowest.
binding to Cu surfaces with different binding sites and porosities
(Fig. 3a-e,
).
To determine the contribution of readsorption to the OD-Cu
profiles, simulations were performed for porous polycrystalline
Cu with two binding sites that have adsorption energies of ~52
and ~60 kJ/mol, which correspond to low-index and stepped fac-
ets of Cu (described in the Supporting Information). As shown in
Figure S21, increasing the depth of the porous layer causes a shift
of the TPD profile to higher temperature. While simulated porous
polycrystalline Cu has a major peak at 260 K, the overall profile
clearly does not fit that of OD-Cu. It is only possible to model the
experimental TPD profile for OD-Cu by including a strong bind-
ing site with adsorption energy of ~67 kJ/mol.
Two possible factors could account for the high-temperature
feature in the OD-Cu profiles: (i) CO readsorption in the catalyst
layer, which causes peak broadening and a shift of the peak max-
imum to higher temperature or (ii) surface sites with enhanced
binding to CO. Both of these effects have been observed in TPD
studies on industrial Cu/ZnO/Al2O3 and Cu/SiO2 catalysts.23,24
OD-Cu–500 has a large roughness factor and a significant fraction
of the sample surface may reside in mesopores where readsorption
could occur. However, OD-Cu–300 has a roughness factor of only
6, which is the lowest of the OD-Cu samples and 2.5-fold lower
than the 350 °C-annealed sample. Despite its lower roughness
factor, OD-Cu–300 presents a higher proportion of the high-
temperature feature than the 350 °C annealed sample, indicating
that readsorption alone cannot account for the high-temperature
feature on OD-Cu.
The microkinetic model also enabled quantification of the rela-
tive proportions of each surface site on OD-Cu. OD-Cu–500 has
the largest proportion of strong-binding sites with 11% surface
coverage. The proportion drops to 5% after annealing at 200 °C
and 1.5% after annealing at 350 °C. While these percentages have
large uncertainty due to assumptions made about the electrode
pore structure, the proportion of strong-binding sites is clearly
reduced by thermal annealing. The proportion of strong-binding
sites on OD-Cu–300 is 5%. When plotted against electrochemical
activity, a linear correlation is obtained between surface area–
corrected jCORedn and the percentage of strong binding sites at the
electrode surface (Fig 3f). On the basis of this correlation, we
propose that the active sites for CO reduction at low overpotential
on Cu are strong-binding sites. Strong CO binding may increase
Direct, quantitative interpretation of the TPD profiles shown
here is not possible because the roughness factor and morphology
of each sample is different. The same CO dosage gives variations
in surface coverage and differing degrees of readsorption on each
sample, both of which affect the relative proportions of the low-
index, stepped, and strong-binding TPD features. To decouple
these effects, the TPD profiles of polycrystalline Cu and the OD-
Cu samples were simulated using microkinetic modeling of CO
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 6
the rate of CO reduction at a surface site by increasing its steady-
ACKNOWLEDGMENT
1
2
3
4
5
6
7
8
state CO coverage. In addition, a high density of strong binding
sites at a surface region created by a bulk defect may be necessary
if the rate-limiting step involves reductive CO coupling.25,26 How-
ever, these strong-binding sites may not be sufficient for CO re-
duction. The OD-Cu–500 electrode annealed at 350 °C retains
1.5% of the strong binding site despite having jCORedn that is 4-fold
lower than that of polycrystalline Cu foil, which has only terrace
and step sites (Figure S8, Table S4). The broad high-temperature
feature observed in TPD does not reflect a single surface atomic
structure; rather it most likely consists of a range of strained or
defective structures, all of which bind CO more strongly within a
~10 kJ/mol range. As such, it is possible that only a small subset
of these defective structures is capable of catalyzing the CO re-
duction reaction. In addition, CO is simply a probe molecule in
these experiments. In electrolyte solution with an applied poten-
tial, the relevant catalytic transition state structures and binding
energies may not scale with the CO binding energy in UHV.
The Danish National Research Foundation’s Center for Individual
Nanoparticle Functionality is supported by the Danish National
Research Foundation (DNRF54). We thank the Global Climate
and Energy Project (106765) and the AFOSR (FA9550-14-1-
0132). C.W.L. gratefully acknowledges a Stanford Graduate Fel-
lowship. Sample characterization was performed at the Stanford
Nano Shared Facilities. Lawrence Livermore National Laboratory
is under the auspices of the U.S. Department of Energy, Contract
No. DE-AC52-07NA27344. J. T. M. and M. K. were supported by
the Office of Basic Energy Sciences, Division of Materials Sci-
ence and Engineering under FWP #SCW0939.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
REFERENCES
(1) Hori, Y. In Modern Aspects of Electrochemistry; Vayenas, C. G.,
White, R. E., Gamboa-Aldeco, M. E., Ed.; Sringer: New York, 2008; Vol.
42, p 89.
(2) Appel, A. M.; Bercaw, J. E.; Bocarsly, A. B.; Dobbek, H.; DuBois, D.
L.; Dupuis, M.; Ferry, J. G.; Fujita, E.; Hille, R.; Kenis, P. J. A.; Kerfeld,
C. A.; Morris, R. H.; Peden, C. H. F.; Portis, A. R.; Ragsdale, S. W.;
Rauchfuss, T. B.; Reek, J. N. H.; Seefeldt, L. C.; Thauer, R. K.; Waldrop,
G. L. Chem Rev 2013, 113, 6621.
(3) Chen, Y. H.; Li, C. W.; Kanan, M. W. J Am Chem Soc 2012, 134,
19969.
(4) DiMeglio, J. L.; Rosenthal, J. J Am Chem Soc 2013, 135, 8798.
(5) Zhu, W. L.; Michalsky, R.; Metin, O.; Lv, H. F.; Guo, S. J.; Wright, C.
J.; Sun, X. L.; Peterson, A. A.; Sun, S. H. J Am Chem Soc 2013, 135,
16833.
(6) Mistry, H.; Reske, R.; Zeng, Z. H.; Zhao, Z. J.; Greeley, J.; Strasser,
P.; Cuenya, B. R. J Am Chem Soc 2014, 136, 16473.
(7) Lu, Q.; Rosen, J.; Zhou, Y.; Hutchings, G. S.; Kimmel, Y. C.; Chen, J.
G. G.; Jiao, F. Nat Commun 2014, 5.
(8) Asadi, M.; Kumar, B.; Behranginia, A.; Rosen, B. A.; Baskin, A.;
Repnin, N.; Pisasale, D.; Phillips, P.; Zhu, W.; Haasch, R.; Klie, R. F.;
Kral, P.; Abiade, J.; Salehi-Khojin, A. Nat Commun 2014, 5, 4470.
(9) Hori, Y.; Takahashi, R.; Yoshinami, Y.; Murata, A. J Phys Chem B
1997, 101, 7075.
Grain boundaries are bulk phenomena, but they can affect the
surface of the sample probed in TPD experiments. The reduction
in the proportion of strong binding sites upon thermal annealing
qualitatively correlates to the microstructural change observed in
TEM. The 7-fold drop in strong binding sites percentage is similar
to the 9-fold drop in grain boundary density observed between
OD-Cu-500 and the electrode after 350 °C annealing. Notably, the
proportion of strong binding sites estimated by microkinetic mod-
eling of the TPD profiles is significantly larger than the propor-
tion of the surface corresponding to grain boundary surface termi-
nations. Based on the grain boundary orientation maps and assum-
ing a 1 nm boundary width, we estimate that OD-Cu-500 has 1-
2% coverage of grain boundary surface termination area while the
TPD modeling indicates 7–15% coverage of strong binding sites.
If the grain boundaries are responsible for creating all of the
strong binding sites, these results indicate that the surface termi-
nations affect the structure and binding properties of a region that
encompasses a much larger width of 4–10 nm. Previous simula-
tions of nanocrystalline fcc metals have shown that grain bounda-
ries strain the surfaces up to several nm away 27. Annealing of the
samples up to a certain temperature would leave grain boundaries
unaffected, but relax out metastable strong binding sites.
(10) Schouten, K. J. P.; Qin, Z. S.; Gallent, E. P.; Koper, M. T. M. J Am
Chem Soc 2012, 134, 9864.
(11) Li, C. W.; Ciston, J.; Kanan, M. W. Nature 2014, 508, 504.
(12) Fu, S. S.; Somorjai, G. A. Surf Sci 1992, 262, 68.
(13) Sandoval, M. J.; Bell, A. T. J Catal 1993, 144, 227.
(14) Perez-Alonso, F. J.; McCarthy, D. N.; Nierhoff, A.; Hernandez-
Fernandez, P.; Strebel, C.; Stephens, I. E. L.; Nielsen, J. H.; Chorkendorff,
I. Angew Chem Int Edit 2012, 51, 4641.
(15) van der Niet, M. J. T. C.; den Dunnen, A.; Juurlink, L. B. F.; Koper,
M. T. M. Angew Chem Int Edit 2010, 49, 6572.
(16) Johansson, T. P.; Ulrikkeholm, E. T.; Hernandez-Fernandez, P.;
Escudero-Escribano, M.; Malacrida, P.; Stephens, I. E. L.; Chorkendorff,
I. Phys Chem Chem Phys 2014, 16, 13718.
(17) LaGrange, T.; Reed, B. W.; Wall, M.; Mason, J.; Barbee, T.; Kumar,
M. Appl Phys Lett 2013, 102.
(18) Rauch, E. F.; Veron, M.; Portillo, J.; Bultreys, D.; Maniette, Y.;
Nicolopoulos, S.; 2008, S5. Microscopy and Analysis 2008, 22, S5.
(19) In IUPAC Solubility Data Series; Cargill, R. W., Ed.; Pergamon
Press: Oxford, 1990; Vol. 43.
(20) Nerlov, J.; Sckerl, S.; Wambach, J.; Chorkendorff, I. Appl Catal A-
Gen 2000, 191, 97.
(21) Vollmer, S.; Witte, G.; Woll, C. Catal Lett 2001, 77, 97.
(22) Makino, T.; Okada, M. Surf Sci 2014, 628, 36.
(23) Strunk, J.; d'Alnoncourt, R. N.; Bergmann, M.; Litvinov, S.; Xia, X.;
Hinrichsen, O.; Muhler, M. Phys Chem Chem Phys 2006, 8, 1556.
(24) d'Alnoncourt, R. N.; Bergmann, M.; Strunk, J.; Loffler, E.; Hinrich-
sen, O.; Muhler, M. Thermochim Acta 2005, 434, 132.
(25) Montoya, J. H.; Shi, C.; Chan, K.; Nørskov, J. K. J. Phys. Chem. Lett.
2015, 6, 2032.
(26) Calle-Vallejo, F.; Koper, M. T. M. Angew Chem Int Ed 2013, 52,
7282.
(27) Stukowski, A.; Markmann, J.; Weissmuller, J.; Albe, K. Acta Mater
2009, 57, 1648.
In summary, we present evidence that the high CO reduction
activity on OD-Cu is correlated to surface sites that bind CO more
strongly than low-index and stepped Cu facets. These metastable
sites may arise from the disordered surfaces at grain boundary and
defect terminations, which are stabilized by the interconnected
nanocrystalline network. Further structural and mechanistic eluci-
dation of the surface chemistry of OD-Cu, in particular regarding
the exact nature of the active site, will enable the design and syn-
thesis of more efficient catalytic materials for CO reduction.
ASSOCIATED CONTENT
Supporting Information
Experimental procedures and additional data. This material is
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Authors
*mkanan@stanford.edu, ibchork@fysik.dtu.dk
Author Contributions
‡These authors contributed equally to this work.
ACS Paragon Plus Environment

Page 5 of 6
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
95x34mm (300 x 300 DPI)
ACS Paragon Plus Environment