Journal of Applied Electrochemistry

https://doi.org/10.1007/s10800-018-1166-6

RESEARCH ARTICLE

Electrochemical reduction of ­CO2 at CuAu nanoparticles: size and alloy

effects
Evan Andrews1 · Yuxin Fang1 · John Flake1

Received: 13 October 2017 / Accepted: 5 February 2018

© Springer Science+Business Media B.V., part of Springer Nature 2018

Abstract
Reduction of C ­ O2 at Cu or Au electrodes typically yields methane or CO, respectively. Cu and Au nanoparticles and their
alloys offer unique advantages over foil electrodes in terms of reduced overpotentials and product selectivities. In this work,
we explore the electrochemical reduction of ­CO2 in aqueous electrolytes using alloys of Cu and Au including 2 and 6 nm
nanoparticles along with polycrystalline foils. These results show the CuAu alloys primarily produce CO; however, yields
are dramatically increased relative to Au. CuAu electrodes in the form of planar foils produce up to 3.4 times more CO
yields relative to Au foil. Most remarkably, nanoparticle electrodes provide up to 12.5-fold CO yield increases relative to
polycrystalline alloy foils and 175-fold CO yield increases relative to bulk Au foils. Voltammetry shows that onset potentials
for ­CO2 reduction are shifted anodically with smaller nanoparticle sizes and with greater Au content. The dramatic increase
in CO yields with nanoparticle alloys is attributed to the improved ­CO2 deoxygenation associated with Cu interfaces and
the relatively facile desorption of CO from low-coordination Au sites.
Graphical Abstract

Foil 2nm 6nm

O=C=O CO
O=

C C
Cu Au

Keywords  Copper · Gold · Alloy · CO2 reduction · Electrocatalysis · Carbon dioxide · Nanoparticle

1 Introduction

Electronic supplementary material  The online version of this
article (https​://doi.org/10.1007/s1080​0-018-1166-6) contains
supplementary material, which is available to authorized users.
* John Flake
johnflake@lsu.edu
1
Cain Department of Chemical Engineering, Louisiana State
University, Baton Rouge, LA 70803, USA
Transition metals such as Cu, Ni, Sn, and Fe have long been
considered as potential catalysts for ­CO2 reduction [1–4].
Of the transition metals, Cu [5–7] has been the most heav-
ily investigated as it produces hydrocarbons at relatively
high current densities (~ 5 mA cm−2) and Faradaic efficien-
cies > 60% [6, 8]. Likewise, other noble metals such as Au
and Ag are known to produce CO at relatively high current
densities and Faradaic efficiencies exceeding 90% [9, 10].

13
Vol.:(0123456789)

In recent years, there has been significant progress in the
synthesis of well-defined nanoparticles using wet proce-
dures such as Brust Schiffrin, Perrault [11], Martin [12], or
Turkevich methods. These particles are particularly inter-
esting in their application as electrocatalysts as they offer
unique advantages over bulk analogs including remarkably
greater activities. In terms of wet synthesis methods, the
most significant advances have been centered on Au nano-
particles including “magic number” nanoparticles that are
stabilized using ligands such as glutathione, dodecanethiol,
triphenylphosphine, or phenylethanethiol [13]. These Au
nanoparticles are relatively stable and do not agglomerate for
several months of low-temperature storage [13, 14]. The wet
synthesis of Cu nanoparticles is somewhat more challenging
since these particles are relatively more susceptible to oxi-
dation and agglomeration; however, there have been several
reports showing the synthesis of Cu nanoparticles [15–17].
In terms of the particle size effect on the electrochemical
behavior of C ­ O2 reduction at Au and Cu electrodes, Zhu
et al. and Kauffman et al. have shown that Au nanoparti-
cles are more active than bulk Au, producing at up to 97%
CO Faradaic efficiency at 8 nm sizes, and 100% using A ­ u25
[10, 18, 19]. Likewise, Reske et al. and Baturina et al. have
shown that smaller Cu particles are nearly twice as active as
foil electrodes [15, 20].
Previous reports with CuAu alloys have shown that sin-
gle crystal CuAu electrodes result in decreased methane,
ethylene, and ethanol yields compared to bulk Cu, but show
increases in CO yield, and at lower overpotentials [21, 22].
Increasing the molar ratio of Au relative to Cu also shows
increased CO production. For example, at a C ­ u50Au50 ratio,
the CO partial current was ~ 1.5 times greater than bulk
Au [22]. Likewise, 10-nm CuAu nanoclusters have been
reported to produce CO at up to 65% Faradaic efficiency,
with ­Cu1Au3 alloys outperforming pure Au nanoparticles
in both yield (~ 10% greater) and Faradaic efficiency [21].
Density functional theory (DFT) studies have shown that
alloy electrodes offer a potential means to create catalysts
that outperform pure metal catalysts due to low coordina-
tion. DFT has been used to determine the binding energies
of intermediates, allowing for the prediction of reaction
pathways based on thermodynamics and kinetics [23–25].
Hansen et al. show that the binding energies of the *CO and
*COOH intermediates adsorbed on transition metal surfaces
follow linear scaling relationships [26]. On all transition
metals, the adsorption energies of the *CO, *COOH, and
*CHO intermediates are proportionally in such a way that
the reaction energetics and pathways are not altered by the
electronic properties of the catalyst [27]. This limitation sug-
gests that the optimal pathways for ­CO2 reduction to hydro-
carbons cannot be obtained simply by using pure transition
metal electrodes, although Cu is nearest to the optimal spot
[26]. Other DFT simulations show that Cu alloys do not
13
Journal of Applied Electrochemistry
follow the same linear relationship as pure metals [28]; so
these may offer an opportunity to break the linear scaling in
binding energies and alter product selectivities.
Cu and Au are prime candidates for alloying, as Au has a
high CO activity and Cu is known to be active in the hydro-
genation of CO. Alloying Cu and Au causes the adsorption
energies of *CO to shift by < 0.2 eV, while the energies of
*OCOH shift by up to 0.5 eV [29]. A DFT simulation of a
­Cu3Au1 surface by Hirunsit et al. suggests that the decrease
in hydrocarbons observed with CuAu alloys relative to Cu is
due to a weakened ability to bond CO [29]. Another mecha-
nism to compensate for the weakened CO bonding is via
the use of nanoscale catalysts, which also has the benefit of
increasing mass activity. CO preferentially adsorbs at edge
and corner sites, which are more numerous with nanoscale
catalysts. The preference of CO for edge and corners is also
apparent on high disorder single crystal catalysts [5]. Here,
we explore the electrocatalytic behavior of CuAu alloys as
a function of alloy composition and particle size.
2 Experimental
The 2 nm nanoclusters were synthesized using the following
procedure modified from Hostetler et al. [30] and Yin et al.
[31] ­HAuCl4 ­xH2O and ­CuCl2 were dissolved in DI water
along with KBr. Separate solutions with Cu/Au molar ratios
of 1:1, 3:1, and 9:1 were prepared. Tetraoctylammonium
bromide (TOAB) was dissolved in toluene. The solutions
were stirred together until the aqueous solution was clear,
indicating the phase transfer of the ions. Dodecanethiol
(600 µL) was added and the solution was stirred for 1 h or
until the toluene phase became clear. 0.5 g of ­NaBH4 in DI
water was slowly added to the solution, resulting in a brown
color. The solution was stirred for 3 h and the aqueous phase
removed by pipette. The toluene solvent was evaporated
under vacuum and the nanoparticles washed and filtered
with ethanol.
The 6-nm nanoclusters were synthesized using a modified
procedure based on the literature [31, 32]. 2-nm nanoparti-
cles were synthesized following the previous procedure, but
after removing the aqueous phase, the toluene was evapo-
rated to give a nanoparticle solution of ~ 15× concentra-
tion. The resulting solution was sealed in a glass reactor
and heated at 150 °C for 2 h. The remaining toluene was
evaporated under vacuum and the nanoparticles washed and
filtered with ethanol.
Bulk alloy foils were obtained commercially from ESPI
metals. The alloys were 50, 25, and 10% Au by weight, cor-
responding to molar ratios of C­ u75.6Au24.4, ­Cu90.3Au9.7, and
­Cu96.6Au3.4, respectively. Several 1 cm2 squares were cut
from the foils and used as working electrodes for electrolysis.
Journal of Applied Electrochemistry
Transmission electron microscopy (TEM) samples were
prepared by dispersing dilute samples in toluene onto Cu
mesh grids (Fig. 1). The diameters were measured and aver-
aged. (Table S1).
X-ray photoelectron spectroscopy (XPS) was used to
determine the Cu:Au ratios of the nanoparticles after syn-
thesis (Figures S1, S2).
Electrodes were prepared by mixing the nanoparticles
with carbon black inks using Nafion™ as a conductive poly-
mer binder. 100-mg carbon black was mixed with 50 mg of
Nafion™ resin and 10 mg of nanoparticles in 2 mL acetone,
then sonicated for 30 min to ensure even dispersion. Ink was
applied to a glassy carbon electrode in a Teflon™ holder by
paintbrush, giving a nanoparticle loading of 0.06 mg cm−2.
Electrochemical experiments were performed in a sealed
two-compartment cell with a Nafion™ membrane separat-
ing the two compartments. One compartment contained the
Ag/AgCl reference electrode and working electrode, and the
other compartment contained the Pt counter electrode. 0.1M
­KHCO3 was used as the electrolyte. The catholyte in the
working compartment was continuously bubbled with ­CO2
and effluent gas vented to a gas chromatograph injection
port (Shimadzu GC-2014). The pH of carbonate electrolyte
was measured at 6.8 when saturated with C ­ O2, and 8.9 when
Fig. 1  Cu68Au32 2 nm (top) and ­Cu38Au62 6 nm (bottom) TEM
purged with ­N2, and used for converting to the RHE scale.
Voltammetry was conducted at a scan rate of 10 mV s−1.
Onset potentials were calculated by plotting the natural log
of current against voltage and determining where the curve
departed from linear behavior (Figure S3). Reaction yields
were determined using potentiostatic experiments using gas
chromatography (TCD and FID).
The yield data were normalized using the surface area of
the nanoparticles in order to compare nano and bulk scale
catalysts. Surface area normalization was also necessary
to compare nanoparticles of the same size, as the density
of the alloy changes with composition. The nanoparticle
weight loading was estimated from the ink composition and
the number and geometric surface area of the nanoparticles
was calculated using the average sizes from the TEM images
(Equation S1). 50% of the surface area was assumed inac-
cessible due to the carbon black substrate and the Nafion™
polymer.
3 Results and discussion
XPS characterization of the alloys showed the 2 nm nan-
oparticles’ compositions to be ­Cu25Au75, ­Cu62Au38, and
­Cu70Au30 while the 6 nm nanoparticles’ compositions were
­Cu9Au91, ­Cu38Au62, and ­Cu59Au41. The noticeable difference
in 2 and 6 nm compositions is likely due to migration of Au
towards the surface during heat treatment due to its lower
surface energy [33]. While there is some level of surface
segregation, our XPS and that of literature results using the
same synthesis techniques show that an alloy is present on
the surface despite segregation [30, 31, 34].
Bulk foils were evaluated as catalysts in order to set a
baseline for comparison with the nanoparticles. Figure 2
shows the CO yields of the bulk alloy foils as well as indi-
vidual Cu and Au foils. All alloy foils show higher CO yields
compared to individual Cu and Au foils, with the ­Cu75Au25
foil yielding the most. At − 1.0 V vs. RHE, ­Cu75Au25 foils
yield 3.4 times more CO than bulk Au foil and 22.3 times
more than bulk Cu foil. However, Au foil showed a higher
Faradaic efficiency than ­Cu90Au10 and ­Cu96Au4 alloys at
lower overpotentials, likely due to the increased hydrogen
affinity of Cu (Fig. 3). This is consistent with the results
published previously by Christophe et al. [22].
The 2 nm alloy nanoparticles’ CO yields and Faradaic
efficiencies are shown in Figs. 4 and 5, respectively. The
­C u 25Au 75 alloy gave the highest CO yield, peaking at
1.47 mmol cm−2 h−1 and producing ~ 9 times the CO yield
as 2 nm ­Cu70Au30 at − 1.1 V vs. RHE. ­Cu25Au75 also had the
greatest peak CO Faradaic efficiency of 37%, while ­Cu70Au30
possessed only 22% Faradaic efficiency towards CO.
­Cu68Au32 showed a maximum yield of 0.65 mmol cm−2 h−1
and a Faradaic efficiency of 28%.
13

Fig. 2  CO yield on bulk CuAu, pure Au, and pure Cu foils. Yields are
normalized to foil surface area
Fig. 3  Faradaic efficiency of CO on bulk CuAu, Au, and Cu foils
The 6 nm alloy nanoparticles’ CO yields and Faradaic effi-
ciencies are shown in Figs. 6 and 7, respectively. ­Cu38Au62
gave the highest CO yield, peaking at 8.8 mmol cm−2 h−1
with a Faradaic efficiency of 50%. Despite having a lower
relative yield of 5.7 mmol cm−2 h−1, ­Cu9Au91 possessed the
greatest Faradaic efficiency towards CO with a maximum
of 73%. ­Cu59Au41 showed a maximum yield and Faradaic
efficiency of 1.25 mmol cm−2 h−1 and 18%, respectively.
When compared with foils of similar composition, the
CuAu nanoparticles exhibit increased CO yields. At − 1.1 V
vs. RHE, 6 nm C ­ u59Au41 produces CO at ~ 10 times the
yield of bulk C­ u75Au25. When compared with the pure Cu
and Au foils, the difference is even more distinct, with 6 nm
­Cu38Au62 showing ~ 175 times the CO yield of bulk Au
13
Journal of Applied Electrochemistry
Fig. 4  CO yield on 2 nm CuAu nanoparticle catalysts. Yields are nor-
malized to the surface areas of the nanoparticles
Fig. 5  Faradaic efficiency of 2 nm CuAu nanoparticle catalysts
and 2 nm ­Cu25Au75 showing ~ 29× the CO yield of bulk
Au. At their maximums, the 6-nm nanoparticles (Fig. 6)
yield 8.8 mmol cm−2 h−1 of CO; the 2-nm nanoparticles
(Fig. 4) yield 1.47 mmol cm−2 h−1, and the foils (Fig. 2)
yield only 0.175 mmol cm−2 h−1. Previous reports involv-
ing ­CO2 reduction on 1.9 and 4.8 nm Cu nanoparticles [15]
and 2 and 6 nm Au nanoparticles [10] have described single
order of magnitude increases (×9 and ×40 for Au 6 and 2 nm
nanoparticles respectively at − 1.2 V vs. RHE).
The ­CO2 reduction onset potentials (Table 1) are indic-
ative of a size effect being responsible, as the size of the
catalyst has a significant effect on the onset potential. The
6-nm nanoparticle electrodes show at least a 200 mV anodic
shift compared to bulk alloys, and the 2-nm nanoparticles
Journal of Applied Electrochemistry

Table 2  HER formation onset potentials in N

­ 2 purged 0.1 M ­KHCO3
HER formation onset potentials/V vs. RHE
Bulk 2 nm 6 nm

Au − 0.36 – – Cu9Au91 − 0.41

Cu75Au25 − 0.38 Cu25Au75 − 0.45 Cu38Au62 − 0.47
Cu90Au10 − 0.37 Cu62Au38 − 0.48 Cu59Au41 − 0.54
Cu96Au4 − 0.38 Cu70Au30 − 0.50 – –

show at least a 150 mV anodic shift. The onset potential is

not heavily dependent on alloy composition, with catalysts
of the same size having potentials within 50 mV windows.
The onset potentials for the hydrogen evolution reaction
are given in Table 2. While there was some cathodic shift
in hydrogen evolution reaction onset, it was not significant
relative to that observed with C ­ O2 reduction, indicating that
Fig. 6  CO yield on 6 nm CuAu nanoparticle catalysts. Yields are nor-
malized to the surface areas of the nanoparticles
the increases in CO yield are not simply from changes in
surface area.
The relatively high activity of the 6-nm nanoparticles
compared to the 2 nm nanoparticles is somewhat surpris-
ing. The difference may be due to the size effects observed
on small nanoparticles where there are optimal particle
sizes for reaction activity [35, 36]. The electrode morphol-
ogy and binder may also play a role in the activity of small
particle electrodes; smaller nanoparticles are more likely to
be trapped in internal pores, limiting their access to C ­ O2
[37]. However, the most likely explanation is that the ­CO2
reduction active sites on a 6 nm are denser than on a 2-nm
nanoparticle. The edge site to corner site ratio is higher on
a larger diameter spherical nanoparticle [38], and as shown
by Zhu et al., metal catalysts with high edge to corner ratios
such as Au nanowires are very active for ­CO2 reduction [9].
Mistry et al. show that Au nanoparticles in the 2–8 nm size
range have significant size effects, where hydrogen selectiv-
ity increases as the particle size decreases, a result attributed
to the higher ratio of corner sites where hydrogen evolution
Fig. 7  Faradaic efficiency of CO on 6  nm CuAu nanoparticle cata-
lysts is preferred [39]. The relatively low CO Faradaic efficiency
of the 2-nm CuAu nanoparticles compared to CuAu foils
matches this observation. Despite the improved yields, the
Table 1  CO2 reduction onset potentials in C
­ O2 saturated 0.1  M overall selectivity shifts towards hydrogen.
­KHCO3 Although lesser than the size effects, alloy composition
CO2 reduction onset potentials/V vs. RHE effects were also readily observed in both the foil and the
2 and 6 nm nanoparticle electrodes, with a larger Au ratio
Bulk 2 nm 6 nm
in the alloy generally correlating to a greater CO yield. At
Au − 0.42 – – Cu9Au91 − 0.14 − 1.0 V vs. RHE, the ­Cu75Au25 alloy foil yields ~ 1.2 times
Cu75Au25 − 0.44 Cu25Au75 − 0.22 Cu38Au62 − 0.12 more CO than the ­Cu90Au10 and ­Cu96Au4 foils. Similarly,
Cu90Au10 − 0.43 Cu62Au38 − 0.25 Cu59Au41 − 0.13 at − 1.1 V vs. RHE, the 2 nm C ­ u25Au75 produces ~ 9 times
Cu96Au4 − 0.37 Cu70Au30 − 0.26 – – the CO yield as the 2 nm C ­ u70Au30 and the 6 nm C ­ u38Au62
produces ~ 11 times the CO yield as the ­Cu59Au41. How-
ever, the 6-nm ­C u 9Au 91 nanoparticles yield ~ 50% less
CO than the 6-nm C ­ u38Au62 nanoparticles, despite having
a Faradaic efficiency of 77% towards CO. An excess of

13

Au content may result in higher Faradaic efficiency, up
to the > 95% seen from the pure Au nanoparticles from
Kauffman et al. [10], but decrease the net catalytic activity.
There may be an optimum CuAu alloy ratio for CO yield
between the ­Cu38Au62 and ­Cu9Au91 compositions.
The increased CO yields have been attributed to
improved CO desorption on alloys due to weakened CO
bond energy on Cu [22]. Alloying the metals results in
a change in electronegativity, showing a shift in elec-
tron charge of ~ 0.2 eV towards Au [40]. This results in
improved Au binding sites for CO, which is an electron
acceptor, primarily bonding to the surface via pi orbitals,
and prefers to bind at metal sites with electron rich d orbit-
als [41]. Likewise, *COOH is expected to be stabilized on
CuAu alloy corner sites [42]. As such, *CO and *COOH
bonded to CuAu will be less stable relative to Cu surfaces,
but more stable than Au.
There are several explanations for the underlying nature
of the dramatic improvements in CO yield observed with
nanoparticle CuAu alloys. Christophe et al. proposed that
electronic effects can alter the bond strength of CO, causing
an increase in kinetics [22]. Hirunsit et al. have used DFT
calculations to reach a similar conclusion, showing that the
adsorption energy of *CO is 0.19 eV weaker on C ­ u75Au25
than on Cu and that the change in bond energy plays a
role on several alloys [28, 29]. The most significant yield
increases are associated with nanoparticle alloys. Kim et al.
proposed that the presence of Cu adjacent to Au surface
atoms also increases the stability of the *COOH intermedi-
ate by allowing a single *COOH to bond to both Au and Cu
atoms [21]. The Cu atom is the more oxyphilic of the two
metals, so the C atom bonds to the Au, and an O to the Cu.
As the alloys increase in Au content, the number of Cu sites
with neighboring Au atoms increases, which contributes
towards the increase in CO yield. The aforementioned reac-
tions are most likely to occur at low-coordinated sites and
thus the alloy effect is more pronounced at the nanoscale.
Other groups utilizing alloys such as CuIn [43], ­Cu3Ni [44],
and AuCd [45] have also proposed bimetallic active sites as
the key part of the electrocatalysts; however, DFT calcula-
tions of CuAu alloy nanoclusters performed by Lysgaard
et al. suggest no O–Cu bonds were formed by *COOH in
their simulation, although Au edge and apex sites adjacent to
Cu were still preferred for reactivity [42]. Other works pro-
pose the metal with lower surface energy (Au in this case)
is observed to migrate to the surface, causing the alloy to act
similarly to the segregated surface metal [33, 46]. Likewise,
works considering intentionally segregated electrocatalysts
such as core–shell nanoparticles [47] and thin overlayers
have noted the lattice strain as another factor affecting the
reaction [33, 48–50]. In all cases, edge sites [9] and other
low-coordination sites such as grain boundaries [51] are
identified as the key active sites for ­CO2 reduction.
13
Journal of Applied Electrochemistry
The results from this work suggest that the dramatic yield
increases originate from both improved deoxygenation of
­CO2 at Cu and the reduced adsorption energy of CO at Au.
Likewise, the remarkable increases in yield associated with
nanoscale CuAu alloys suggest that the relative impact of
these effects is significantly greater at low-coordinated
Cu–Au interfaces, as the 6-nm nanoparticles which have
the highest ratio of edge sites give the greatest CO yield
[38]. It was also found that the nanoparticles with higher
Au ratios were more favorable for CO formation which is
in agreement with DFT by Lysgaard et al. [42]. The DFT
suggests that Au heavy surfaces with Cu cores are the most
stable structurally and have similar adsorption strength to
pure Au nanoparticles, though CuAu bimetallic sites are the
most preferred for CO adsorption.
4 Conclusion
CuAu alloys significantly increase CO yield per area com-
pared to pure Au or Cu electrocatalysts, and as nanoscale
catalysts enhance the CO yield even further. The use of alloy
foils leads to a ~ 3× increase in CO yields compared to pure
metals, while the use of CuAu nanoparticles leads to an
increase in CO yield by > 175×. In addition to the increased
yields, voltammetry shows reduction C ­ O2 reduction onset
potentials are shifted anodically by > 150 mV for 2 nm
alloys and > 200 mV for 6 nm alloy electrodes. Likewise, the
increased yields and anodically shifted onset potentials are
associated with an increase in edge sites and the synergistic
electronic effects of CuAu alloys on the nanoparticle alloy
electrodes. These results suggest that CuAu alloys, particu-
larly low-coordinated interfaces thereof, improve both the
deoxygenation ­CO2 and desorption of CO.
Acknowledgements  This work was supported by the National Science
Foundation under Grant CBET-1438385.
References
1. Hori Y, Kikuchi K, Suzuki S (1985) Chem Lett 14(11):1695–1698
2. Hori Y (2008) Modern aspects of electrochemistry. Springer, New
York, pp 89–189
3. Hori Y, Wakebe H, Tsukamoto T, Koga O (1994) Electrochim
Acta 39(11):1833–1839
4. Jitaru M, Lowy DA, Toma M, Toma BC, Oniciu L (1997) J Appl
Electrochem 27(8):875–889
5. Hori Y, Takahashi I, Koga O, Hoshi N (2002) J Phys Chem B
106(1):15–17
6. Hori Y, Kikuchi K, Murata A, Suzuki S (1986) Chem Lett
15(6):897–898
7. DeWulf DW, Jin T, Bard AJ (1989) J Electrochem Soc
136(6):1686–1691
8. Kuhl KP, Cave ER, Abram DN, Jaramillo TF (2012) Energy Envi-
ron Sci 5(5):7050–7059
Journal of Applied Electrochemistry
9. Zhu W, Zhang Y-J, Zhang H, Lv H, Li Q, Michalsky R, Peterson
AA, Sun S (2014) J Am Chem Soc 136(46):16132–16135
10. Kauffman DR, Alfonso D, Matranga C, Qian H, Jin R (2012) J
Am Chem Soc 134(24):10237–10243
11. Per rault SD, Chan WCW (2009) J Am Chem Soc
131(47):17042–17043
12. Martin MN, Basham JI, Chando P, Eah S-K (2010) Langmuir
26(10):7410–7417
13. Negishi Y, Nobusada K, Tsukuda T (2005) J Am Chem Soc
127(14):5261–5270
14. Boyen HG, Kästle G, Weigl F, Koslowski B, Dietrich C, Ziemann
P, Spatz JP, Riethmüller S, Hartmann C, Möller M (2002) Science
297(5586):1533–1536
15. Reske R, Mistry H, Behafarid F, Roldan B, Cuenya, Strasser P
(2014) J Am Chem Soc 136(19):6978–6986
16. Wu S-H, Chen D-H (2004) J Colloid Interface Sci 273(1):165–169
17. Lisiecki I, Billoudet F, Pileni MP (1996) J Phys Chem
100(10):4160–4166
18. Zhu W, Michalsky R, Metin Ö, Lv H, Guo S, Wright CJ, Sun X,
Peterson AA, Sun S (2013) J Am Chem Soc 135(45):16833–16836
19. Chen Y, Li CW, Kanan MW (2012) J Am Chem Soc
134(49):19969–19972
20. Baturina OA, Lu Q, Padilla MA, Xin L, Li W, Serov A, Artyush-
kova K, Atanassov P, Xu F, Epshteyn A, Brintlinger T, Schuette
M, Collins GE (2014) ACS Catal 4(10):3682–3695
21. Kim D, Resasco J, Yu Y, Asiri AM, Yang P (2014) Nat Commun
5:4948
22. Christophe J, Doneux T, Buess-Herman C (2012) Electrocatalysis
3(2):139–146
23. Hansen HA, Montoya JH, Zhang Y-J, Shi C, Peterson AA, Nør-
skov JK (2013) Catal Lett 143(7):631–635
24. Nie X, Esopi MR, Janik MJ, Asthagiri A (2013) Angew Chem Int
Ed 52(9):2459–2462
25. Schouten KJP, Kwon Y, van der Ham CJM, Qin Z, Koper MTM
(2011) Chem Sci 2(10):1902–1909
26. Hansen HA, Varley JB, Peterson AA, Nørskov JK (2013) J Phys
Chem Lett 4(3):388–392
27. Peterson AA, Nørskov JK (2012) J Phys Chem Lett 3(2):251–258
28. Hirunsit P, Soodsawang W, Limtrakul J (2015) J Phys Chem C
119(15):8238–8249
29. Hirunsit P (2013) J Phys Chem C 117(16):8262–8268
30. Hostetler MJ, Zhong C-J, Yen BKH, Anderegg J, Gross SM,
Evans ND, Porter M, Murray RW (1998) J Am Chem Soc
120(36):9396–9397
31. Yin J, Shan S, Yang L, Mott D, Malis O, Petkov V, Cai F, Shan
Ng M, Luo J, Chen BH (2012) Chem Mater 24(24):4662–4674
32. Maye MM, Zheng W, Leibowitz FL, Ly NK, Zhong C-J (2000)
Langmuir 16(2):490–497
33. Friebel D, Mbuga F, Rajasekaran S, Miller DJ, Ogasawara H,
Alonso-Mori R, Sokaras D, Nordlund D, Weng T-C, Nilsson A
(2014) J Phys Chem C 118(15):7954–7961
34. Llorca J, Domínguez M, Ledesma C, Chimentão RJ, Medina
F, Sueiras J, Angurell I, Seco M, Rossell O (2008) J Catal
258(1):187–198
35. Kinoshita K (1990) J Electrochem Soc 137(3):845–848
36. Shao M, Peles A, Shoemaker K (2011) Nano Lett
11(9):3714–3719
37. Uchida M, Park Y-C, Kakinuma K, Yano H, Tryk DA, Kamino
T, Uchida H, Watanabe M (2013) Phys Chem Chem Phys
15(27):11236–11247
38. Back S, Yeom MS, Jung Y (2015) ACS Catal 5(9):5089–5096
39. Mistry H, Reske R, Zeng Z, Zhao Z-J, Greeley J, Strasser P,
Cuenya BR (2014) J Am Chem Soc 136(47):16473–16476
40. Sham TK, Hiraya A, Watanabe M (1997) Phys Rev B 55(12):7585
41. Bagus PS, Nelin CJ, Bauschlicher CW (1984) J Vac Sci Technol
A 2(2):905–909
42. Lysgaard S, Myrdal JSG, Hansen HA, Vegge T (2015) Phys Chem
Chem Phys 17(42):28270–28276
43. Rasul S, Anjum DH, Jedidi A, Minenkov Y, Cavallo L, Takanabe
K (2015) Angew Chem 127(7):2174–2178
44. Adit Maark T, Nanda BRK (2016) J Phys Chem C
120(16):8781–8789
45. Jovanov ZP, Hansen HA, Varela AS, Malacrida P, Peterson
AA, Nørskov JK, Stephens IEL, Chorkendorff I (2016) J Catal
343:215–231
46. Choi J, Kim MJ, Ahn SH, Choi I, Jang JH, Ham YS, Kim JJ, Kim
S-K (2016) Chem Eng J 299:37–44
47. Monzo J, Malewski Y, Kortlever R, Vidal-Iglesias FJ, Solla-
Gullon J, Koper MTM, Rodriguez P (2015) J Mater Chem A
3(47):23690–23698
48. Reske R, Duca M, Oezaslan M, Schouten KJP, Koper MTM,
Strasser P (2013) J Phys Chem Lett 4(15):2410–2413
49. Todoroki N, Yokota N, Nakahata S, Nakamura H, Wadayama T
(2016) Electrocatalysis 7(1):97–103
50. Varela AS, Schlaup C, Jovanov ZP, Malacrida P, Horch
S, Stephens IEL, Chorkendorff I (2013) J Phys Chem C
117(40):20500–20508
51. Feng X, Jiang K, Fan S, Kanan MW (2015) J Am Chem Soc
137(14):4606–4609
13