Particle size effects in the catalytic electroreduction of CO2 on Cu nanoparticles

Article abstract of DOI:10.1021/ja500328k

A study of particle size effects during the catalytic CO₂ electroreduction on size-controlled Cu nanoparticles (NPs) is presented. Cu NP catalysts in the 2-15 nm mean size range were prepared, and their catalytic activity and selectivity during CO₂ electroreduction were analyzed and compared to a bulk Cu electrode. A dramatic increase in the catalytic activity and selectivity for H₂ and CO was observed with decreasing Cu particle size, in particular, for NPs below 5 nm. Hydrocarbon (methane and ethylene) selectivity was increasingly suppressed for nanoscale Cu surfaces. The size dependence of the surface atomic coordination of model spherical Cu particles was used to rationalize the experimental results. Changes in the population of low-coordinated surface sites and their stronger chemisorption were linked to surging H₂ and CO selectivities, higher catalytic activity, and smaller hydrocarbon selectivity. The presented activity-selectivity-size relations provide novel insights in the CO₂ electroreduction reaction on nanoscale surfaces. Our smallest nanoparticles (～2 nm) enter the ab initio computationally accessible size regime, and therefore, the results obtained lend themselves well to density functional theory (DFT) evaluation and reaction mechanism verification.

Full text of DOI:10.1021/ja500328k

Article
pubs.acs.org/JACS
Particle Size Effects in the Catalytic Electroreduction of CO on Cu
2
Nanoparticles
Rulle Reske,†,∥ Hemma Mistry,‡,∥ Farzad Behafarid, Beatriz Roldan Cuenya,* and Peter Strasser*
‡
,§
,†
†Department of Chemistry, Chemical Engineering Division, Technical University Berlin, 10623 Berlin, Germany
‡Department of Physics, University of Central Florida, Orlando, Florida 32816, United States
§Department of Physics, Ruhr-University Bochum, 44780 Bochum, Germany
*
S Supporting Information
ABSTRACT: A study of particle size effects during the catalytic CO2
electroreduction on size-controlled Cu nanoparticles (NPs) is presented. Cu NP
catalysts in the 2−15 nm mean size range were prepared, and their catalytic activity
and selectivity during CO electroreduction were analyzed and compared to a bulk
2
Cu electrode. A dramatic increase in the catalytic activity and selectivity for H and
2
CO was observed with decreasing Cu particle size, in particular, for NPs below 5
nm. Hydrocarbon (methane and ethylene) selectivity was increasingly suppressed
for nanoscale Cu surfaces. The size dependence of the surface atomic coordination
of model spherical Cu particles was used to rationalize the experimental results.
Changes in the population of low-coordinated surface sites and their stronger
chemisorption were linked to surging H and CO selectivities, higher catalytic
2
activity, and smaller hydrocarbon selectivity. The presented activity−selectivity−
size relations provide novel insights in the CO electroreduction reaction on nanoscale surfaces. Our smallest nanoparticles (∼2
2
nm) enter the ab initio computationally accessible size regime, and therefore, the results obtained lend themselves well to density
functional theory (DFT) evaluation and reaction mechanism verification.
1
. INTRODUCTION
Although the standard potentials of CO reduction to CH
2 4
0
or C H are small and positive (E = +0.169 V vs the
Carbon dioxide (CO ) is a greenhouse gas which is produced
2
4
CH4
2
0
standard hydrogen electrode (SHE) and E
SHE), the over potential of these reactions are in a range of
= +0.079 V vs
C2H4
by the burning of fossil fuels for energy generation and
transportation, and through industrial processes such as
concrete production. Due to the world’s growing energy
demands, rising CO emissions are a serious environmental
threat. To mitigate this, closing the anthropogenic carbon cycle
through a chemical conversion of CO into usable chemicals
would be an attractive solution. Such methods would involve
the utilization of waste CO as a chemical feedstock to make
hydrocarbons, alcohols, or CO-rich feeds of high interest for
5
,6
6
−10
−
800 to −900 mV,
which makes these processes
energetically very inefficient. Furthermore, the catalytic
selectivities of hydrocarbons are generally relatively small
compared to the formation of molecular hydrogen according
2
2
−
−
0
H2
1
to 2H O + 2 e → H + 2OH (E
= 0 V versus the
2
2
2
reversible hydrogen electrode, RHE).
In the past few years, several studies have investigated the
2
9
,11−14
mechanism of the CO reduction reaction.
It is generally
industry. Of all chemical ways to convert CO into useful
2
2
accepted that carbon monoxide (CO) is an intermediate in the
products, the electrocatalytic route appears as the most
sustainable as it occurs at ambient pressures and temperatures,
in neutral pH conditions, catalyzed by non-noble metals, and
may involve renewable electricity from wind, solar, or hydro
power plants.
reduction of CO . Density functional theory (DFT)
2
calculations put forward by the Norskov group proposed that
the protonation of the adsorbed CO to CHO* is the potential
determining step (PDS) of methane formation on (211) and
9
,14
3,4
(
100) surfaces. On (111) surfaces, the initial protonation of
It is well-known
that CO2 can be electrochemically
CO was predicted to dictate the potential requirements. By the
reduced on copper surfaces in aqueous solutions, producing
formate, CO, and hydrocarbons like methane (CH ) and
ethylene (C H ) according to
2
same authors, ethylene was predicted to form through a
4
chemical dimerization step of CHO or similar adsorbates.
2
4
1
2,13
Schouten et al.
demonstrated experimentally that the
−
−
−
CO + H O + 2e → HCOO + OH
(1)
(2)
(3)
(4)
2
2
Cu(100) surface offers a uniquely active reaction pathway for
−
−
the formation of C H via an electron-mediated dimerization
2
4
CO + H O + 2e → CO + 2OH
2
2
reaction of two adsorbed CO molecules. Dimerization of CO
−
−
CO + 6H O + 8e → CH + 8OH
2
2
4
Received: January 12, 2014
Published: April 18, 2014
−
−
2
CO + 8H O + 12e → C H + 12OH
2
2
2
4
©
2014 American Chemical Society
6978
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
a
Table 1. Parameters Used in the Synthesis of Size-Controlled Micellar Cu NPs
sample name
polymer
loading
no. of coats
particle size h (nm)
real Cu surface area Acu,norm (cm2 /cm2geo)
p
Cu
S1
S2
S3
S4
S5
S6
PS(26000)-P2VP(4800)
PS(27700)-P2VP(4300)
PS(33000)-P2VP(46000)
PS(33000)-P2VP(46000)
PS(130000)-P2VP(135000)
PS(130000)-P2VP(135000)
0.2
0.4
0.4
0.4
0.1
0.4
4
4
3
5
5
5
1.9 ± 0.7
2.3 ± 0.8
4.8 ± 1.4
6.7 ± 1.8
13.1 ± 3.5
15.1 ± 5.2
0.018 ± 0.001
0.037 ± 0.002
0.110 ± 0.005
0.153 ± 0.008
0.06 ± 0.01
0.07 ± 0.02
a
Molecular weight of the PS-P2VP polymers, metal-salt to polymer-head ratio (loading), AFM heights with error bars (standard deviation) measured
on SiO /Si(111) supported NPs, and normalized Cu surface areas calculated from the AFM heights as surface area of NPs/geometric surface area of
2
support, assuming spherical NPs.
on Cu(111) indeed was unfavorable compared to the
effects have been explored include the oxygen reduction
9
,11,14
11
23,49−54
protonation of CO to COH.
More recently, Nie et al.
reaction on noble metals,
the electro-evolution of
55−58
proposed that methane and ethylene formation on Cu(111)
molecular oxygen,
and electro-oxidation of small organic
59
60−62
occurs via a common hydroxyl−methylidyne intermediate,
molecules, as well as electrocatalytic CO oxidation.
The
*
C−OH, which is subsequently hydrogenated to a common
2
oxygen electroreduction reaction (ORR) on Pt, for instance,
exhibits a pronounced particle size effect: Plotted against the
particle diameter, the Pt mass-based catalytic activity shows a
nonmonotonic dependence with a maximum activity for Pt
CH * intermediate. That latter adsorbate may undergo a
nonelectrochemical dimerization and C−C bond formation
11
toward ethylene or else hydrogenate further to methane.
4
9,52,63,64
Controlling the selectivity of the CO reduction reaction and
NPs between 3 and 5 nm in size.
To date, reports on
2
reducing the over potential for the formation of methane and
ethylene are currently major scientific challenges. The
geometry, morphology, and roughness of copper single crystal
and polycrystalline surfaces have shown a dramatic influence on
NP size effects involving Cu NPs have been limited to
heterogeneously catalyzed gas-phase reactions, such as the
heterogeneous oxidation of CO in an excess of hydrogen and
65,66
the catalytic degradation of gaseous carbon tetrachloride.
The present study is the first exploration of catalytic particle
size effects in CO electroreduction on Cu NPs.
Here, we investigate the influence of the size of Cu NPs on
the CO electroreduction activity and, in particular, on product
the catalytic activity and product selectivity during CO
2
7
,10,15
electroreduction.
Enhanced ethylene and CO formation
2
10
was reported on roughened Cu foils. The altered selectivity
was rationalized based on differences in the chemisorption
characteristics of (111), (100), and (211) surfaces. DFT
calculations confirmed that key intermediates, such as *CO and
2
selectivity. The copper NPs ranged from 2 to 15 nm in size and
gave rise to unexpected selectivity and activity variations. Aside
from providing a practical strategy for deliberately altering
activity and product selectivity, this size effect study provides
new clues and hypotheses how chemical bonding controls the
*
CHO were stabilized on stepped surfaces owing to more
negative chemisorption energies. Vollmer et al. used temper-
ature-programmed desorption (TPD) to study CO adsorption
on (i) low index surfaces (Cu(111), Cu(100), and Cu(110)),
selectivity in CO electro conversion. Experimental materials,
2
reaction conditions, observables, and their interpretation lend
themselves easily to theoretical verification and thus will add
important aspects to a more complete mechanistic under-
standing of such an important energy storage reaction.
(
ii) stepped and kinked surfaces (Cu(211), Cu(221), and
Cu(532)), and on (iii) sputtered and polycrystalline Cu films,
concluding that CO binds more strongly on Cu step edges and
kinks than on terrace sites.
Chemisorption and catalytic selectivities can also be
controlled by lattice strain in thin metal overlayers.
was shown in the electroreduction of CO2 on thin Cu
overlayers on Pt studied using online electrochemical mass
spectrometry (OLEMS). Results showed that by decreasing
the copper overlayer thickness, the selectivity of C H versus
CH4 increased, while the activity decreased. This was
explained by the alteration of the chemisorption of reactive
intermediates due to geometric lattice effects.
Another well-known morphology-based strategy to tune
surface chemisorption and resulting catalytic activities and
selectivities is varying the size and/or shape of the catalytically
active species, for instance, metal nanoparticles (NPs).
Utilized in practice for centuries, this dispersion dependence is
generally known as the “catalytic particle size effect”.
very small NP diameters below around 2 nm, where quantum
effects become noticeable,
to as the “catalytic finite-size effect”. Size effects have
frequently been reported for gas-phase catalytic reactions on
metal NPs, such as hydrogenations,
alcohol decomposition reactions, or partial oxidation
16
17−20
This
2. EXPERIMENTAL METHODS
2.1. Sample Preparation. Size-selected Cu NPs were prepared via
inverse micelle encapsulation. Poly(styrene)-block-poly(2-vinylpyri-
dine) (PS-P2VP) diblock copolymers (Polymer Source, Inc.) with
polar head (PVP) and nonpolar tail (PS) blocks were dissolved in
nonpolar toluene to form inverse micelles. The micelles were then
21
2
4
21
loaded with CuCl and stirred for 2 days to form monodispersed Cu
2
NPs. The size of the NPs is controlled either by varying the molecular
weight of the PVP head or by varying the metal salt/PVP ratio. By
varying these two parameters, five different solutions were prepared,
see Table 1. Glassy carbon plates were dip-coated into the solution,
resulting in a monolayer of NPs deposited on the support. The
molecular weight of the polymer tail can be varied to control
interparticle spacing. Next, the encapsulating polymers were removed
2
2,23
24−28
For
by O plasma etching for 20 min (20 W power). To increase the NP
2
density, the dip-coating and plasma etching (ligand removal) steps
28−30
were successively repeated 3−4 times on each individual sample. NPs
the effect is often also referred
28
were simultaneously prepared on flat SiO (4 nm)/Si(111) supports
2
and underwent the same treatment as the glassy carbon supported
NPs. The complete removal of polymeric ligands from Cu/
3
1−33
34
ammonia synthesis,
SiO Si(111) samples was demonstrated based on the lack of any C-
3
5
2/
1
s signal in high resolution XPS measurements (Supporting
3
6−46
reactions,
prominent example.
and CO oxidation on Au NPs is the most
Information Figure S1). However, such measurements are not possible
for Cu NPs supported on glassy carbon due to the very strong carbon
4
7,48
Electrocatalytic reactions where size
6
979
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
signal from the support. Nevertheless, since similar micelle prepared
electrode interface. All potentials are reported with respect to the
reversible hydrogen electrode (RHE) and were corrected by the
experimental voltage loss (IR-drop) caused by the uncompensated
resistances of the electrolyte and external electrical contacts and
connections. ICP-OS measurements of Pt ion concentration before
and after the electrochemical experiments showed no detectable
evidence of dissolved Pt ions, confirming the lack of any significant
dissolution of the Pt anode (counter electrode) during the
experiments. In addition, high-resolution XPS measurements con-
firmed the absence of Pt on the Cu foil or Cu NPs after
electrocatalysis, ruling out any possibility of Pt ion cross contam-
ination.
Cu surface-area normalized catalytic activity and selectivity values
were calculated from raw current data subtracting coverage-corrected
current and selectivity values of bare glassy carbon supports, followed
by subsequent division by the real Cu surface area, ACu,norm × Ageo, of
the Cu NPs according to Table 1.
NPs (same NP solution) were deposited on both SiO /Si(111) and
2
glassy carbon supports, and since the same plasma treatment
procedure was carried out on both samples, the complete removal
of the polymeric ligands is guaranteed from our XPS measurements on
SiO /Si(111).
2
Detailed information on the parameters used for the synthesis of the
Cu NPs is listed in Table 1.
2.2. Sample Characterization. Atomic force microscopy (AFM)
images acquired in tapping-mode (Digital Instruments Nanoscope III)
were used to characterize the NP morphology on SiO (4 nm)/
2
Si(111). Average AFM particle heights h (size) and support area-
p
normalized NP surface areas, ACu,norm (in units of cm /cm2geo), were
2
Cu
calculated assuming spherical particles, Table 1. Histograms of the NP
heights obtained by AFM for all samples are shown in Supporting
Information Figure S2. The assumption of a spherical NP shape on
weakly interacting supports such as SiO or C is based on previous
2
cross sectional transmission electron microscopy studies on similarly
2.4. Numerical Calculation of Model Nanoparticles. To
generate spherical Cu NP model shapes, a large FCC structure with
a Cu lattice unit (L) of 3.61 Å was constructed using a MATLAB code.
When the desired diameter was varied, all the atoms outside the
defined sphere were removed from the bulk FCC structure in order to
construct each model shape. For the database to be comprehensive,
different symmetry centers (SC) were considered with (i) the SC
located on one of the Cu atoms and (ii) the SC shifted 0.5L along one
of the Cartesian axes (e.g., x-axis). The largest spherical NP considered
had a diameter of 18 nm and about 261000 atoms. The relation
between the number of Cu atoms and the NP diameter is shown in
Supporting Information Figure S5a.
In the next step, for each model shape in the database, the
coordination number (CN) of all individual atoms was calculated.
Atoms with a CN value of 12 were considered as bulk atoms and those
with CN below 12 were considered as surface atoms. Supporting
Information Figure S5b shows the percentage of surface and bulk
atoms as a function of the NP diameter. The surface atoms were also
further distinguished based on their coordination number as will be
discussed later.
67
synthesized NPs. The surface area of a single particle Ap was
2
estimated according to A = 4π(h /2) . Using the surface particle
p
p
density on the flat support, ACu,norm was evaluated.
To corroborate the accuracy of the NP sizes (height) obtained via
AFM for our samples containing small NPs, an equivalent of sample
S1 was also measured via transmission electron microscopy (TEM)
with a FEI Tecnai 20 microscope operated at 200 kV. For the TEM
study, Cu NPs analogous to those in sample S1 (same NP solution)
were drop-coated on a 40 nm-thick SiO2 TEM grid, and the
encapsulating ligands were removed via O -plasma exposure. The
2
TEM image of this sample and the corresponding size histogram are
shown in Supporting Information Figure S3. The average diameter
obtained from TEM is 2.4 ± 0.6 nm, in relatively good agreement with
the average size (NP height) obtained via AFM (1.9 ± 0.7 nm).
2
.3. Electrochemical Measurements of the CO Reduction.
2
An airtight electrochemical cell with a three-compartment, three-
electrode design was used for the electrochemical CO reduction. The
2
glassware was cleaned in a “nochromix” bath and afterward in
concentrated HNO for 1 h, respectively, rinsed with ultra pure water
3
several times, and dried at T = 60 °C in a drying cabinet. A platinum
mesh 100 (Sigma-Aldrich 99.9%) formed to a cylinder was used as
counter electrode (CE) and a leak-free Ag/AgCl electrode as reference
electrode (Hugo Sachs Elektronik Harvard apparatus GmbH).
The glassy carbon plate, on which the Cu NPs were supported, was
contacted with a gold clamp and used as working electrode (WE). The
3
. RESULTS
3.1. Synthesis and Structural Characterization of Cu
NP Catalysts. A well-established micelle-based nanoparticle
preparation method
48,68−70
was utilized to prepare six metallic
2
geometric surface area, Ageo, of the particle-covered support was 3 cm .
Cu NP catalysts with spherical shape and different, well-
controlled particle height (“size”), henceforth referred to as S1
to S6 in order of increasing size, Table 1. This synthesis method
has been known to produce highly reproducible particle sizes,
and multiple duplicates of all six NP catalysts were prepared
and measured independently. Figure 1 shows AFM images of
the spherical Cu NPs after ligand removal.
Twenty-five milliliters of 0.1 M KHCO (Roth >99.5%) was used as
3
electrolyte and purged with CO (Air liquid 4.5) (30 mL/min) from
2
the bottom of the cell under the WE until a final stable pH of 6.8 was
3
reached. Before and during the electrochemical measurements, the
CO saturation of the gas atmosphere was controlled with an in situ
2
mass spectrometer (OmniStar GSD 301c, Pfeiffer). Hydrogen (H2),
carbon monoxide (CO), methane (CH ), and ethylene (C H ) were
4
2
4
analyzed using a gas chromatography (GC) system protocol detailed
in Figure S4 of the Supporting Information. The gas chromatograph
used was a Shimadzu GC 2016 equipped with a thermal conductivity
detector (TCD) and a flame ionization detector (FID). Argon (Air
liquid 5.0) was employed as carrier gas. The gaseous products H , N ,
Analysis of the AFM particle heights revealed that six distinct
average NP sizes between 1.9 and 15.1 nm were obtained (see
Table 1). Particle size histograms, shown in Supporting
Information Figure S2, evidenced largely near-symmetric,
narrow size distributions. Increasing the molecular weight of
the micelle head or increasing the metal loading inside the
micelles generally resulted in larger NP sizes. The interparticle
distance generally decreased with smaller NP size due to the
successive multiple dip-coatings employed in the preparation of
such samples in order to ensure sufficient active Cu NP surface
and coverage on the carbon support.
2
2
O , CH and CO were separated in a molecular sieve column (Alltech,
2
4
part no. 57732, 1.65 m × 1/8 in., molecular sieve 13X, 60/80 mesh)
and for C −C hydrocarbons and CO in a HayeSep column (Alltech,
2
3
2
part no. 14487, 3.5 m × 1/8 in., HayeSep D, 80/100 mesh).
The working electrodes were always immersed into the electrolyte
under potential control at E = +0.22 V/RHE. A linear voltammetric
sweep was performed with a scan rate of −5 mV/s between E = +0.22
V/RHE and E = −1.1 V/RHE followed by a chronoamperometric
measurement at E = −1.1 V/RHE for t = 10 min. At the end of the
3
.2. Electrocatalytic Activity Measurements. The Cu
NP catalysts S1−S6 were supported on flat glassy carbon
electrodes and were subsequently deployed as working
electrode in a three-electrode electrochemical setup designed
amperometric step, product gas samples were analyzed. CO gas was
2
bubbling through the electrochemical cell throughout the electro-
chemical measurements to ensure sufficient convective mass transport
to keep the Nernst diffusion layer thicknesses to a minimum and
minimize mass transport limitations and local pH changes at the
for the electroreduction of CO in a 0.1 M bicarbonate solution
2
3
at pH 6.8. The activity and chemical selectivity of catalysts
6
980
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
Figure 3. Particle size effect during catalytic CO electroreduction.
2
The faradaic current densities at −1.1 and −1.0 V/RHE are plotted
against the size of the Cu NP catalysts. The current densities have
been normalized by the Cu particle surface area after subtraction of the
glassy carbon background signal. Error bars indicate scatter in particle
size distributions. Lines are guides to the eye. Data of a Cu foil
electrode (“foil”) are inserted as reference. Conditions: 0.1 M KHCO3,
pH = 6.8, 25 °C.
each Cu NP catalyst in comparison to the LSV of a Cu foil
electrode (”foil”), which was considered an extended macro-
scopic Cu reference catalyst. To correct for differences in the
electrochemically active real surface area of each Cu NP
catalyst, the geometric current density was converted into the
Cu surface area-specific current density after the subtraction of
the glassy carbon (support) background.
Figure 2 reveals a dramatic particle size effect on the overall
catalytic CO electroreduction activity of Cu NPs. The LSV of
2
the Cu foil showed the characteristic shoulder between −0.8
and −1.0 V, previously associated with the formation of
carbonous surface species. The catalytic LSV currents of the Cu
bulk reference in Figure 2 are in excellent quantitative
Figure 1. Tapping-mode AFM images of micellar Cu NPs: (a) S1, (b)
S2, (c) S3, (d) S4, (e) S5, and (f) S6.
agreement with previous reports of CO reduction on Cu foil
2
3
−5,71
S1− S6 for CO electroreduction were monitored during a slow
electrodes in buffered bicarbonate solutions.
The total
2
2
cathodic potential scan as well as at a constant electrode
potential of −1.1 V/RHE. Reproducibility of individual catalytic
activity was confirmed and sample averages are provided.
Figure 2 displays the Linear Sweep Voltammograms (LSV) of
current of the Cu foil at −1.1 V/RHE was −23 mA/cm , while
4
that reported by Hori under similar conditions was −25 mA/
2
cm .
In stark contrast to the Cu foil electrode, the Cu NPs
exhibited significantly higher catalytic activity (larger negative
current density J) as the Cu NP size decreased. The smallest Cu
NPs (S1 and S2, ∼2 nm) showed almost twice the catalytic
activity of extended bulk Cu, while catalysts S3−S6 exhibited
similar catalytic activities over a wide potential window, ranging
between those of bulk Cu and of S2. Electrochemical Tafel lines
of the NP electrocatalysts are plotted in Supporting
Information Figure S6 evidencing similar slopes and curvatures
in the −0.4 to −0.8 V range, yet a clearly more pronounced
activity plateau in the −0.8 to −1.0 V range.
Figure 3 highlights the dependence of the overall catalytic
CO reduction activity on particle size at two different electrode
2
potentials, E = −1.1 V/RHE and E = −1.0 V/RHE (detailed
activity values are given in Supporting Information Table S1).
The data suggest comparable overall activity between bulk Cu
and Cu NPs down to about 5 nm, below which an explosive
trend toward higher catalytic activity with smaller Cu NP size
becomes evident. The smallest Cu NP catalyst (1.9 nm)
Figure 2. Linear sweep voltammetry of the CO electroreduction on
2
displayed a 100% increase in faradaic current density (−48 mA/
Cu NP catalysts, S1−S6, in 0.1 M KHCO3 acquired at room
temperature with −5 mV/s scan rate. Current densities are normalized
by the Cu particle surface area after subtraction of the glassy carbon
background signal. A Cu foil electrode (“foil”) is included as reference.
2
cm , see Supporting Information Table S1), while the 2.3 nm
Cu
NP catalyst S2 still showed a 50% increase in activity compared
to Cu foil electrode. Together, the observed trends evidence a
6
981
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
sudden increase in H2 and CO formation relative to
hydrocarbons. It is evident that small Cu NPs, in stark contrast
to bulk Cu surfaces, evolve no significant amounts of the highly
desired reaction product ethylene.
Taking the number of transferred electrons of each species
into consideration, the volumetric ratios were converted into
faradaic selectivities, see Figure 4b. We note that the faradaic
selectivities of the Cu bulk electrode are in good agreement
4
with earlier reports by Hori. Methane showed the highest
selectivity of 57% (55% reported by Hori), ethylene and
hydrogen showing about 20% (25% and 10% reported by
Hori), and CO exhibiting less than 5% selectivity (3% reported
by Hori).
The data in Figure 4b evidence a drastic dependence of the
product selectivities on the size of the spherical Cu NP catalysts
at a given overpotential. In contrast to the bulk Cu electrode,
the selectivity profiles of Cu NPs are largely dominated by
hydrogen evolution (60−70% compared to 20% on Cu bulk)
over the entire size range considered, a pattern that typically
occurs for extended Cu bulk surfaces at much less negative
overpotentials. The selectivities of methane and ethylene
dropped drastically from their bulk values to much reduced
levels (10−15% for methane and 0−10% for ethylene) in the
5
−15 nm particle size regime, while CO selectivities followed
the trend of hydrogen and increased from 5% to 20−25%. The
anticorrelation between methane and ethylene, on the one
hand, and H and CO, on the other, is quite obvious in Figure
Figure 4. Particle size dependence of (a) the composition of gaseous
2
reaction products (balance is CO ) during catalytic CO2 electro-
reduction over Cu NPs, (b) the faradic selectivities of reaction
products during the CO2 electroreduction on Cu NPs. Lines are
4b. At particle sizes below 5 nm, hydrocarbon selectivities
dropped again with ethylene formation ceasing on 1.9 nm Cu
particles. At the same time, CO selectivity increased to almost
30%.
2
guides to the eye. Conditions: 0.1 M KHCO , E = −1.1 V/RHE, 25
3
°
C. For (a), volume concentrations were normalized using the Cu
3
.3. Modeling Surface Atomic Coordination of
surface areas. Values for H , CH , and C H are scaled by factors of
2
4
2
4
Spherical Cu NPs. To gain insight into the origin of the
observed particle size effects, model spherical Cu NPs between
0.33, 3, and 5, respectively.
1
and 18 nm diameter were considered with a face-centered
dramatic activity-enhancing size effect of nanoscale spherical Cu
cubic structure and a unit cell parameter of a = 3.610 Å. Figure
5a illustrates the surface structure of 2.2 nm (left) and 6.9 nm
(right) NPs corresponding to catalysts S2 and S4. The
percentage ratio of the surface atoms with specific nearest-
neighbor coordination numbers (CN) between 5 and 11 were
considered and binned into atoms with CN < 8, CN = 8, CN =
9, and CN > 9. Figure 5b breaks down the size-dependent
populations of atoms with these four CN groups.
surfaces during CO reduction.
2
3.3. Electrochemical Selectivity Measurements. To
obtain insight into the relevant catalytic reaction pathways,
faradaic selectivity measurements were performed under
stationary reaction conditions at −1.1 V/RHE. Supporting
Information Table S1 and Figure 4a report the background-
corrected, NP surface area-normalized product composition of
the gas stream leaving the working electrode compartment.
Each data point represents the average of at least three
Atoms with coordination 8 (blue), 9 (red), and >9 (green)
accounted together for about 70% of the surface atoms of
spherical particles between 6 and 18 nm. The contribution of
low coordinated atoms (steps and kinks) (gray) increased
slightly between 18 and 6 nm, then started to become
significant between 6 and 2 nm, and finally drastically
augmented for NPs below 2 nm. Below 4 nm, the scatter in
the individual populations increased considerably, until, below
2 nm, low-coordinated steps prevailed on the surface of the Cu
NPs. The color coding in Figure 5a reflects the location of
atoms with individual CN providing an intuitive picture of how
the individual CNs are distributed across the surface of the
particles. Atoms in Cu(111) and Cu(100) surfaces have CN
values of 9 (red) and 8 (blue), respectively. Also the second
layer of atoms in Cu(110) and Cu(211) surfaces have CN of 11
and 10, respectively (green). However, there are a small
number of atoms with the mentioned CN values that do not
belong to the assigned Cu surface orientation (e.g., 6 blue-
colored kink atoms with CN = 8 at the corner of each 111
facet). Nevertheless, the color coding illustrates the complexity
independent measurements. The excess CO flow keeps the
2
combined volumetric ratios of gas reaction products below 1
vol % for all NP sizes (note that the values of three products are
scaled with the factor given in the plot). We note that the
absolute volume % values (gas molar ratios) of the reaction
products are sensitive to variations in the volumetric CO flow,
2
a shortcoming which is eliminated when relative faradaic
selectivities are considered.
It is evident from Figure 4a that major reaction products
were hydrogen, CO, methane, and ethylene. While a minute
amount of liquid formic acid was detectable for the Cu foil
(
<5% consistent with ref 4), no detectable quantities of formic
acid were formed for the nanoparticle samples.
The size of the Cu NP catalysts had a significant effect on the
formation rate of individual products and, hence, their catalytic
selectivity. Larger particles in the 5−15 nm size regime clearly
showed much less hydrocarbon formation, and instead yielded
more H and CO. Smaller particles below 5 nm displayed a
2
6
982
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
and selective for ethylene and proposed a decoupled proton−
electron transfer during a dimerization of CO to explain the pH
independence of the reaction. More recently, surface
morphology effects in CO electroreduction were explored by
2
10
Tang et al., who found that steps on roughened Cu surfaces
show increased ethylene, but very low methane selectivity.
Unlike Hori, however, Tang et al. also reported increased CO
selectivities on roughened polycrystalline Cu surfaces. Similar
9
to others, the authors corroborated their experimental results
using thermochemical DFT calculations that suggested stronger
chemisorption of CO and CHO intermediates on (211)
stepped surfaces as compared to planar (100) or (111) surfaces.
These predictions are consistent with earlier experimental
studies of CO chemisorption on polycrystalline and stepped
1
6,73,74
16
single crystals of Cu.
CO TPD peaks to higher temperatures as they moved from
111) to (211) and polycrystalline Cu surfaces, indicating
Vollmer et al. reported shifts of
(
stronger CO binding on stepped surface sites. While stepped
planar-type surfaces do provide qualitative insight in the effect
of steps on reactivity and selectivity, they typically fall short of
capturing the more detailed behavior of highly curved NP
surfaces.
Another recent computational thermochemical DFT analysis
14
of the CO electroreduction on the stepped Cu(211) surface
2
explored the effect of CO chemisorption on the limiting
Figure 5. (a) Ball models of spherical Cu NPs with 2.2 and 6.9 nm
diameters. Surface atoms are color-coded according to their first
neighbor coordination number (CN), CN < 8 (gray), CN = 8 (blue),
CN = 9 (red), CN > 9 (green). (b) Population (relative ratio) of
surface atoms with a specific CN as a function of particle diameter.
potential U . The limiting potential of an elementary
L
electrochemical reaction step represents the electrochemical
overpotential at which the elementary step becomes exergonic
(or downhill in free energy), and can be regarded as a simple
measure of the potential where the rate of this step becomes
of the surface atomic coordination on such spherical particles
compared to faceted Wulff-type models (truncated cuboctahe-
dron shape).
appreciable. The elementary step with the largest U
the overpotential of the overall reaction. Stronger CO binding
was predicted to cause a moderate increase in the U
potential-limiting protonation of COad according to
L
dictates
of the
L
4
. DISCUSSION
+
−
CO + H + e → CHO
ad
ad
Structure sensitivity is very common in homogeneous,
heterogeneous, electro-, and biocatalysts. However, the depend-
ence of the catalytic activity and selectivity on the catalyst
dispersion is an idiosyncrasy of heterogeneous catalysts and
heterogeneously catalyzed reactions.
This study has explored the effect of the size of spherical Cu
NPs on the catalytic activity and selectivity during the
mechanistically fairly complex electrocatalytic reduction of
CO . It is the first of its kind for Cu and CO electrocatalysis,
and has revealed a distinctly different behavior of nanoscale Cu
surfaces compared to their extended bulk counterparts. A
significant increase in the overall faradaic activity was found
with shrinking NP size. This was coupled to an increased
selectivity for hydrogen and CO, but decreased selectivity for
hydrocarbon production. Entering the smallest size domains of
particles below 3 nm, hydrocarbon formation drops drastically,
while CO is the preferred carbonaceous product.
This should translate into lower CO protonation rates and
lower hydrocarbon selectivities at any constant electrode
potential. While this prediction is qualitatively consistent with
the present selectivity trends, the dependence of U on the CO
binding energy appears too weak to account for the sudden
selectivity changes in Figure 4.
L
14
4.2. Nanoparticle Models. There is a consensus that NP
size effects largely originate from size-dependent surface metal
atomic coordination and its corresponding surface electronic
2
2
2
4,26,28,30,33,62,75,76
structure.
This is the reason three-dimen-
sional particle models and their surface coordination structure
are better suited to help understand particle size effects. A
meaningful analysis of surface atomic coordination using
particle models requires the selection of structural models
that are relevant to the experimental system with respect to size
and shape. In the present study, the micelle-assisted, plasma-
cleaned NP synthesis initially yielded surface-oxidized Cu NPs,
which were subsequently in situ electrochemically reduced to
metallic Cu NPs, as confirmed by XPS. Since neither thermal
treatments nor repeated potential cycling was applied, the
present Cu NPs are rather spherical in shape with only limited
surface faceting. Unlike faceted models, spherical particle
models also include energetically less favorable, higher-index
facets and their characteristic coordination. Clearly, this
consideration renders highly faceted Wulff-type particle models
inappropriate for the present study, and this is the reason why
spherical models with limited faceting were explored in terms of
4.1. Stepped Extended Surfaces. Stepped extended
surfaces have often been considered as models for the behavior
of particle surfaces. The effect of steps and kinks on well-
defined planar Cu surfaces on the catalytic activity and
selectivity of CO electroreduction has first been addressed
2
3
,72
by Hori and co-workers. They reported a high selectivity for
ethylene on Cu(100), while methane formation was clearly
preferred on Cu(111) facets. Steps in (100) terraces, regardless
of their orientation, increased the ethylene formation over
methane. Hori’s observations were later confirmed by Koper
1
2,13
and co-workers,
who found Cu(100) to be highly active
6
983
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
their surface atomic coordination. Coordination numbers (CN)
of surface atoms ranged from 5 to 11, where CN 8 atoms could
be largely associated with (100) facets, while CN 9 atoms
represented (111) facets, see Figure 5a. The present models
also exhibit (110) facets and atoms with CNs of up to 11
detected may be explained based on a different reaction
mechanism, as for example, an Eley−Rideal-type one involving
the earlier desorption of hydrogen as compared to CO, its
diffusion in the liquid phase, readsorption on the Cu surface
near a strongly bound CO atom, and subsequent reaction and
desorption in the form of a hydrocarbon. For the large NPs,
due to the expected weaker CO and H bonding, their diffusivity
on the particle surface might be greater, favoring reaction and
hydrocarbon formation without having to involve an additional
desorption/liquid phase diffusion/readsorption/reaction path-
way.
(
green in Figure 5a).
To understand the observed size effects in Figure 4 in more
detail, the experimental trends in catalytic activity and
selectivity were correlated with the size-dependent surface
structure and coordination in Figure 5.
We start our discussion with the smallest size regime. The
smallest Cu NPs S1 and S2 with sizes around 2 nm showed a
dramatic increase in overall catalytic activity, which is attributed
to the high ratio of low-coordinated surface atoms (CN < 8)
starting as low as CN = 5 for NP diameters below 4 nm, see
Figure 5b. In this size regime, particle size effects are often
referred to as “catalytic finite size effects”, and small variations
in size induce drastic changes in the NP’s electronic structure,
It should also be mentioned here that we do not have any
indication of a CO-poisoning effect for the small NP samples. If
that were the case, we would have a decrease in the surface area
available for water dissociation, which should lead to a decrease
in the yield of H , which opposes the present observation.
2
8
Li and Kanan recently reported interesting CO electro-
2
reduction experiments on Cu oxide electrodes. They achieved
significantly increased CO selectivities on electrochemically
reduced micrometer-thick Cu oxide films and foils. Following
the authors, Cu oxide electroreduction creates a uniquely
selective Cu surface. However, for thin Cu oxide layers, this
effect vanishes. This is the reason we believe we can exclude the
formation of uniquely CO-selective Cu metal phases from Cu
oxide particles as origin for the current observations.
28,29,76
while quantum effects may become non-negligible.
Following our arguments on stepped surfaces above, step and
kink atoms of Cu NPs with such low CNs are expected to
exhibit stronger chemisorption of CO , CO, atomic H, and
2
CO H as compared to larger particles or bulk Cu surfaces.
y
x
Since the reductive adsorption of protons (the Volmer
reaction) on free Cu surface sites (H + + e + * → H ) is
−
ad
79
the rate-determining step of the overall electrocatalytic
In a recent report by Gupta et al., the authors highlighted
7
7,78
hydrogen evolution reaction (HER),
the observed increase
the role of local pH changes during the CO reduction reaction.
2
in hydrogen production, in hydrogen selectivity, and overall
faradaic activity for small NPs appears plausible. A similar
Under the present experimental conditions, the authors
predicted a local pH increase near the electrode surface of
about 2 pH units at a Nernst layer thickness of 100 μm. This
effect would result in a downshift of the zero point of the RHE
scale relative to the NHE scale (and along with it a downshift in
the thermodynamic equilibrium potentials of coupled proton/
electron half-cell reactions). Given our constant applied
electrode potential, this would lead to a decrease in the
electrochemical overpotential of proton-coupled electron
transfer reactions, leading to a smaller hydrogen evolution
rate. This is in obvious conflict with our experimental
observations, in which increased hydrogen selectivities and
increased current densities were measured for the smallest
nanoparticles. We therefore conclude that a strong local pH
change is not the origin of the increased faradaic selectivities of
hydrogen obtained for our small nanoparticles.
argument can be made for those Cu surface orientations where
+
the initial reductive CO adsorption to COOH (CO + H
+
2
2
−
9,14
e
+ * → COOH ) is rate-determining.
Stronger CO2
ad
adsorption would increase the overall rate of this initial reaction
step toward CO.
For Cu particles in the intermediate size range (5−18 nm),
the populations of surface atoms with CN = 8 and 9 amount to
about 12 and 20%, respectively (see Figure 5b), and remain
surprisingly constant over the entire intermediate size range.
These population numbers are significantly smaller compared
to those found on low-index smooth extended surfaces, such as
the (100), (111), or (211) surfaces. With methane and ethylene
forming preferentially on (111) (CN = 9) and (100) (CN = 8)
3,12,13
sites, respectively,
changes in the relative populations of
80
(
111) and (100) facets provide a plausible explanation for both
Finally, in recent years, Ono et al. and subsequently Eckle
81−83
the sudden steep drop in faradaic selectivities between Cu bulk
et al.
advanced a particle density based mechanism to
surfaces and Cu NPs, as well as the selectivity plateau between
account for stability and selectivity changes in NP ensembles.
The role of the interparticle distance on the activity and
stability of Au NP catalysts during CO oxidation was
investigated by Ono et al., and NP samples with larger
distances were found to result in lack of reactive-coarsening for
these catalysts, while more closely spaced catalysts experienced
drastic morphological changes under reaction conditions.
Taking the methanation of CO as a model reaction, Eckle et
al. argued that low catalyst particle densities favor desorption of
intermediates rather than their subsequent reaction by repeated
collision with the catalytic surface. We cannot rule out that this
effect might also contribute to the present observations. While
2
and 15 nm discernible in Figure 5.
The distinct selectivity observed when comparing small (2
nm) and large (15 nm) NPs can be explained based on the
increase in the strength of the binding of products and
intermediate reaction species on the smaller NPs. Strong CO
binding on low coordinated sites (small NPs) favors high
activity, high CO and H selectivities, and reduced hydrocarbon
2
formation. The latter is understood if the reduced surface
mobility of the more strongly bound CO and H on the NP
ad
ad
surface is considered, leading to a lower probability of surface
reaction to form hydrocarbons. Instead, it is likely that on small
NPs COad and Had desorb before reacting on the Cu surface,
in our study the particle density of the most H - and CO-
2
which explains the higher yields of CO and H products on
selective particle ensembles (smaller size Cu NPs, S1 and S2) is
larger than that of the larger Cu NPs (S6) which were found to
be less CO-selective, the samples with the smallest NPs also
have the smallest total Cu surface coverage (see Table 1).
Therefore, readsorption and reaction should also be less
2
these samples. The lower mobility of adsorbed intermediate
species on the surface of our small spherical NPs can be
assigned to their higher curvature and stepped and roughened
surfaces. Thus, on the small NPs, the small hydrocarbon yields
6
984
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
favorable on our smallest NP samples. This is an interesting
aspect that requires further attention and will be addressed in a
separate study where we are systematically changing the
interparticle distance while keeping the particle size constant.
In summary, a spherical particle model provided valuable
insight into experimental trends in activity and selectivity of
CO2 electroreduction as a function of particle size. An
increasing number of surface atoms with CN below 8 are
responsible for enhanced hydrogen and CO evolution on NPs
around 2 nm and below. Characteristic changes in the
populations of low coordinated sites provide plausible
explanations for experimental activity and selectivity variations
for NPs between 5 and 15 nm. The smallest Cu NPs at and
below 2 nm displayed the most drastic dependence of product
selectivities on particle size consistent with a dramatic increase
in the content of low-coordinated surface atoms.
However, where syngas is the preferred reaction product, for
instance, to serve as feed for gas-to-liquid reaction technologies,
the size of the Cu particles could serve as a convenient
parameter to deliberately control the resulting H to CO
2
product ratio.
ASSOCIATED CONTENT
Supporting Information
■
*
S
XPS data from the C-1s region, TEM data and the
corresponding histogram obtained from sample S1; histograms
of nanoparticle heights obtained from AFM; an overview of
catalytic activity and product distribution, schematic illustration
of the used GC method, number of atoms and surface
population as a function of NP diameter for spherical FCC Cu
model shapes, and the Tafel characteristics for all samples. This
material is available free of charge via the Internet at http://
pubs.acs.org.
5. CONCLUSION
The particle size effect has been explored for the catalytic
AUTHOR INFORMATION
■
electroreduction of CO on Cu NPs in the size range of 2−15
2
Corresponding Authors
pstrasser@tu-berlin.de
Beatriz.Roldan@rub.de
nm and compared to bulk Cu. Given that this reaction is
emerging as one of the most important processes for chemical
storage of electricity, knowledge about the reactivity and
selectivity of this reaction on nanoscale catalysts is of utmost
importance.
Author Contributions
These authors contributed equally.
∥
Nanometer-sized Cu NPs showed a dramatic increase in
overall catalytic activity (faradaic current). Selectivity analysis
Notes
The authors declare no competing financial interest.
has revealed that enhanced formation of CO and H accounts
2
for the increase in the faradaic activity observed on the Cu NPs.
The selectivity of hydrocarbons first changed to a reduced
constant plateau, until it virtually vanished for NP sizes at and
below 2 nm.
A model of nonfaceted, spherical, and hence experimentally
relevant, Cu particles was used to estimate the population of
surface atoms with specific coordination numbers. Below 2 nm,
a drastic increase in under-coordinated atoms with CN < 8 is
observed. These strongly binding sites accelerate the hydrogen
ACKNOWLEDGMENTS
■
The authors thank Dr. Lin Gan from TU Berlin for the
acquisition of TEM data. We thank Dr. Ana Varela for helpful
discussions. H.M., F.B., and B.R.C. (UCF and RUB) acknowl-
edge funding from the U.S. Department of Energy DOE-BES
DE-FG02-08ER15995). B.R.C. acknowledges financial sup-
port by the cluster of excellence RESOLV (DFG, EXC-1069) at
(
the Ruhr University Bochum. We thank the “Zentraleinrichtung
̈
fur Elektronenmikroskopie (Zelmi)” of the Technical Uni-
versity Berlin for their support with TEM techniques. Financial
support by the German Research Foundation (DFG) through
Grant STR 596/3-1 under the Priority Program 1613
Regeneratively formed fuels by water splitting” is gratefully
acknowledged.
evolution and the CO reduction to CO. However, they seem
2
to be unfavorable for the subsequent hydrogenation of CO,
which lowers hydrocarbon selectivity at small particles. A
plausible explanation for the observed trend is the reduced
mobility of intermediate reaction species (CO and H) on the
small NPs due to their stronger bonding, which decreases the
chance of further recombination and reaction to form
hydrocarbons on the Cu surface.
At intermediate particle sizes, the spherical particle model
predicts low and constant populations of (100) and (111)
facets, which is consistent with the reduced, yet constant
hydrocarbon selectivities observed for Cu NPs between 5 and
“
REFERENCES
■
(
(
1) Schwartz, S. E. Energy Environ. Sci. 2008, 1, 430.
2) Aresta, M. Carbon Dioxide as Chemical Feedstock; Wiley:
Weinheim, 2010.
3) Hori, Y. In Modern Aspects of Electrochemistry; Vayenas, C. G.,
Ed.; Springer: New York, 2008; p 89.
4) Hori, Y.; Murata, A.; Takahashi, R. J. Chem. Soc., Faraday Trans. 1
989, 85, 2309.
(5) Gattrell, M.; Gupta, N.; Co, A. J. Electroanal. Chem. 2006, 594, 1.
(6) Hori, Y.; Takahashi, R.; Yoshinami, Y.; Murata, A. J. Phys. Chem.
B 1997, 101, 7075.
7) Durand, W. J.; Peterson, A. A.; Studt, F.; Abild-Pedersen, F.;
Norskov, J. K. Surf. Sci. 2011, 605, 1354.
8) Li, C. W.; Kanan, M. W. J. Am. Chem. Soc. 2012, 134, 7231.
9) Peterson, A. A.; Abild-Pedersen, F.; Studt, F.; Rossmeisl, J.;
(
1
5 nm compared to Cu bulk surfaces. For these larger NPs,
(
1
weaker binding of CO and H is expected, favoring hydrocarbon
formation.
Overall, the presented size-dependent selectivity conclusions
help obtain a more complete mechanistic understanding of the
CO2 electroreduction process. Catalyst materials, their
geometry, and the reaction conditions lend themselves easily
to DFT theoretical verification and refined mechanistic
interpretation. In particular, this study considers particles
below 2 nm, which have become computationally tractable by
theorists using ab initio codes. This opens the possibility for
further in-depth DFT work on the present system.
(
(
(
Norskov, J. K. Energy Environ. Sci. 2010, 3, 1311.
10) Tang, W.; Peterson, A. A.; Varela, A. S.; Jovanov, Z. P.; Bech, L.;
Durand, W. J.; Dahl, S.; Norskov, J. K.; Chorkendorff, I. Phys. Chem.
(
Chem. Phys. 2012, 14, 76.
Clearly, where hydrocarbon yields are of the essence, very
small (<3 nm) nanoscale Cu catalysts should be avoided.
(11) Nie, X.; Esopi, M. R.; Janik, M. J.; Asthagiri, A. Angew. Chem.,
Int. Ed. 2013, 52, 2459.
6
985
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986

Journal of the American Chemical Society
Article
(
12) Schouten, K. J. P.; Kwon, Y.; van der Ham, C. J. M.; Qin, Z.;
Koper, M. T. M. Chem. Sci. 2011, 2, 1902.
13) Schouten, K. J. P.; Qin, Z.; Gallent, E. P.; Koper, M. T. M. J. Am.
Chem. Soc. 2012, 134, 9864.
14) Peterson, A. A.; Norskov, J. K. J. Phys. Chem. Lett. 2012, 3, 251.
15) Christophe, J.; Doneux, T.; Buess-Herman, C. Electrocatalysis
(48) Ono, L. K.; Sudfeld, D.; Roldan Cuenya, B. Surf. Sci. 2006, 600,
5041.
(
(49) Kinoshita, K. J. Electrochem. Soc. 1990, 137, 845.
(50) Mayrhofer, K. J. J.; Blizanac, B. B.; Arenz, M.; Stamenkovic, V.
R.; Ross, P. N.; Markovic, N. M. J. Phys. Chem. B 2005, 109, 14433.
(51) Shao, M. H.; Peles, A.; Shoemaker, K. Nano Lett. 2011, 11,
3714.
(52) Nesselberger, M.; Ashton, S.; Meier, J. C.; Katsounaros, I.;
Mayrhofer, K. J. J.; Arenz, M. J. Am. Chem. Soc. 2011, 133, 17428.
(53) Mukejee, S.; McBreen, J. J. Electroanal. Chem. 1998, 448, 163.
(54) Corradini, P. G.; Pires, F. I.; Paganin, V. A.; Perez, J.; Antolini, E.
J. Nanopart. Res. 2012, 14, No. 1080.
(
(
2
(
(
1
012, 3, 139.
16) Vollmer, S.; Witte, G.; Wo
17) Mavrikakis, M.; Hammer, B.; Norskov, J. K. Phys. Rev. Lett.
998, 81, 2819.
18) Molodkina, E. B.; Ehrenburg, M. R.; Polukarov, Y. M.; Danilov,
A. I.; Souza-Garcia, J.; Feliu, J. M. Electrochim. Acta 2010, 56, 154.
19) Oezaslan, M.; Hasche, F.; Strasser, P. J. Electrochem. Soc. 2012,
59, B444.
20) Strasser, P.; Koh, S.; Anniyev, T.; Greeley, J.; More, K.; Yu, C.;
̈
ll, C. Catal. Lett. 2001, 77, 97.
(
(
1
(
(55) Jirkovsky,
2006, 8, 1417.
(56) Jirkovsky,
Electrochem. Soc. 2006, 153, E111.
́
J.; Makarova, M.; Krtil, P. Electrochem. Commun.
́
́ ́
J.; Hoffmannova, H.; Klementova, M.; Krtil, P. J.
Liu, Z.; Kaya, S.; Nordlund, D.; Ogasawara, H.; Toney, M. F.; Nilsson,
A. Nat. Chem. 2010, 2, 454.
(57) Kwon, G.; Ferguson, G. A.; Heard, C. J.; Tyo, E. C.; Yin, C. R.;
DeBartolo, J.; Seifert, S.; Winans, R. E.; Kropf, A. J.; Greeley, J.;
Johnston, R. L.; Curtiss, L. A.; Pellin, M. J.; Vajda, S. ACS Nano 2013,
(
21) Reske, R.; Duca, M.; Oezaslan, M.; Schouten, K. J. P.; Koper, M.
T. M.; Strasser, P. J. Phys. Chem. Lett. 2013, 4, 2410.
22) Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl.
Catal., B 2005, 56, 9.
7
, 5808.
(
(
(
(
58) Reier, T.; Oezaslan, M.; Strasser, P. ACS Catal. 2012, 2, 1765.
59) Park, S.; Xie, Y.; Weaver, M. J. Langmuir 2002, 18, 5792.
60) Diao, P.; Zhang, D. F.; Guo, M.; Zhang, Q. J. Catal. 2007, 250,
(
(
(
(
(
23) Mukerjee, S. J. Appl. Electrochem. 1990, 20, 537.
24) Arai, M. J. Chem. Eng. Jpn. 1997, 30, 1123.
25) Coq, B.; Figueras, F. Coord. Chem. Rev. 1998, 178, 1753.
26) Bond, G. C. Surf. Sci. 1985, 156, 966.
27) Ono, L. K.; Croy, J. R.; Heinrich, H.; Roldan Cuenya, B. J. Phys.
2
(
47.
61) Jaramillo, T. F.; Baeck, S.-H.; Cuenya, B. R.; McFarland, E. W. J.
Am. Chem. Soc. 2003, 125, 7148.
62) Roldan Cuenya, B.; Baeck, S.-H.; Jaramillo, T. F.; McFarland, E.
W. J. Am. Chem. Soc. 2003, 125, 12928.
63) Watanabe, M.; Sei, H.; Stonehart, P. J. Electroanal. Chem. 1989,
61, 375.
64) Merte, L. R.; Behafarid, F.; Miller, D. J.; Friebel, D.; Cho, S.;
(
Chem. C 2011, 115, 16856.
28) Li, L.; Larsen, A. H.; Romero, N. A.; Morozov, V. A.; Glinsvad,
(
(
2
(
C.; Abild-Pedersen, F.; Greeley, J.; Jacobsen, K. W.; Norskov, J. K. J.
Phys. Chem. Lett. 2013, 4, 222.
(
29) Kleis, J.; Greeley, J.; Romero, N. A.; Morozov, V. A.; Falsig, H.;
Mbuga, F.; Sokaras, D.; Alonso-Mori, R.; Weng, T.-C.; Nordlund, D.
ACS Catal. 2012, 2, 2371.
Larsen, A. H.; Lu, J.; Mortensen, J. J.; Dulak, M.; Thygesen, K. S.;
Norskov, J. K.; Jacobsen, K. W. Catal. Lett. 2011, 141, 1067.
(65) Firsova, A. A.; Khomenko, T. I.; Il’ichev, A. N.; Korchak, V. N.
(
2
(
30) Tritsaris, G.; Greeley, J.; Rossmeisl, J.; Nørskov, J. Catal. Lett.
011, 141, 909.
31) Bezemer, G. L.; Bitter, J. H.; Kuipers, H. P. C. E.; Oosterbeek,
Kinet. Catal. 2011, 52, 434.
(
(
66) Liou, Y. H.; Lo, S. L.; Lin, C. J. Water Res. 2007, 41, 1705.
67) Ono, L.; Yuan, B.; Heinrich, H.; Cuenya, B. R. J. Phys. Chem. C
H.; Holewijn, J. E.; Xu, X. D.; Kapteijn, F.; van Dillen, A. J.; de Jong, K.
P. J. Am. Chem. Soc. 2006, 128, 3956.
2
(
010, 114, 22119.
68) Croy, J. R.; Mostafa, S.; Heinrich, H.; Roldan Cuenya, B. Catal.
Lett. 2009, 131, 21.
69) Mostafa, S.; Behafarid, F.; Croy, J. R.; Ono, L. K.; Li, L.; Yang, J.
C.; Frenkel, A. I.; Roldan Cuenya, B. J. Am. Chem. Soc. 2010, 132,
5714.
(
32) den Breejen, J. P.; Radstake, P. B.; Bezemer, G. L.; Bitter, J. H.;
Froseth, V.; Holmen, A.; de Jong, K. P. J. Am. Chem. Soc. 2009, 131,
197.
33) Argo, A. M.; Odzak, J. F.; Gates, B. C. J. Am. Chem. Soc. 2003,
25, 7107.
34) Jacobsen, C. J. H.; Dahl, S.; Hansen, P. L.; Tornqvist, E.; Jensen,
(
7
(
1
(
1
(70) Roldan Cuenya, B. Thin Solid Films 2010, 518, 3127.
(71) Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F. Energy
L.; Topsoe, H.; Prip, D. V.; Moenshaug, P. B.; Chorkendorff, I. J. Mol.
Catal. A: Chem. 2000, 163, 19.
Environ. Sci. 2012, 5, 7050.
72) Hori, Y.; Takahashi, I.; Koga, O.; Hoshi, N. J. Mol. Catal. 2003,
99, 39.
73) Xu, X. P.; Vesecky, S. M.; He, J. W.; Goodman, D. W. J. Vac. Sci.
Technol., A 1993, 11, 1930.
(
1
(
(
35) Croy, J. R.; Mostafa, S.; Liu, J.; Sohn, Y.-h.; Cuenya, B. R. Catal.
Lett. 2007, 118, 1.
36) Subramanian, V.; Wolf, E. E.; Kamat, P. V. J. Am. Chem. Soc.
004, 126, 4943.
(
2
(
(
74) Xu, X. P.; He, J. W.; Goodman, D. W. Surf. Sci. 1993, 284, 103.
75) Behafarid, F.; Ono, L.; Mostafa, S.; Croy, J.; Shafai, G.; Hong, S.;
(
37) Guczi, L.; Beck, A.; Paszti, Z. Catal. Today 2012, 181, 26.
(
38) Ishida, T.; Kinoshita, N.; Okatsu, H.; Akita, T.; Takei, T.;
Rahman, T.; Bare, S. R.; Cuenya, B. R. Phys. Chem. Chem. Phys. 2012,
4, 11766.
76) Greeley, J.; Rossmeisl, J.; Hellman, A.; Norskov, J. K. Z. Phys.
Chem. 2007, 221, 1209.
77) Norskov, J. K.; Bligaard, T.; Logadottir, A.; Kitchin, J. R.; Chen,
J. G.; Pandelov, S.; Stimming, U. J. Electrochem. Soc. 2005, 152, J23.
78) Greeley, J.; Kibler, L.; El-Aziz, A. M.; Kolb, D. M.; Nørskov, J. K.
ChemPhysChem 2006, 7, 1032.
79) Gupta, N.; Gattrell, M.; MacDougall, B. J. Appl. Electrochem.
006, 36, 161.
Haruta, M. Angew. Chem., Int. Ed. 2008, 47, 9265.
39) Haruta, M. Relevance of Metal Nanoclusters Size Control in
Gold(0) Catalytic Chemistry; Elsevier Science Bv: Amsterdam, 2008.
40) Haruta, M. Size- and Support-Dependency in the Catalysis of Noble
Metals: Gold as a Typical Example; Imperial College Press: London,
001; Vol. 117.
41) Cunningham, D. A. H.; Vogel, W.; Kageyama, H.; Tsubota, S.;
Haruta, M. J. Catal. 1998, 177, 1.
1
(
(
(
(
2
(
(
(
2
(
(
42) Kumar, S.; Zou, S. Z. Langmuir 2009, 25, 574.
43) Overbury, S. H.; Schwartz, V.; Mullim, D. R.; Yan, W. F.; Dai, S.
(
(
(
80) Ono, L. K.; Roldan Cuenya, B. Catal. Lett. 2007, 113, 86.
81) Eckle, S.; Denkwitz, Y.; Behm, R. J. J. Catal. 2010, 269, 255.
82) Eckle, S.; Augustin, M.; Anfang, H. G.; Behm, R. J. Catal. Today
J. Catal. 2006, 241, 56.
44) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet,
M. J.; Delmon, B. J. Catal. 1993, 144, 175.
45) Heiz, U.; Sanchez, A.; Abbet, S.; Schneider, W. D. J. Am. Chem.
Soc. 1999, 121, 3214.
(
2
(
3
012, 181, 40.
83) Eckle, S.; Anfang, H. G.; Behm, R. J. Appl. Catal., A 2011, 391,
25.
(
(
46) Chin, Y. H.; Iglesia, E. J. Phys. Chem. C 2011, 115, 17845.
(
47) Bond, G. C. Molecules 2012, 17, 1716.
6
986
dx.doi.org/10.1021/ja500328k | J. Am. Chem. Soc. 2014, 136, 6978−6986