physical chemistry chemical physics this paper is ......25 alloy nanoparticles annealed at both low...
TRANSCRIPT
Physical Chemistry Chemical Physics
This paper is published as part of a PCCP Themed Issue on:
Electrocatalysis: theory and experiment at the interface
Guest Editor: Andrea Russell
Editorial
Electrocatalysis: theory and experiment at the interface Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b808799g
Papers
Structure of the water/platinum interface––a first principles simulation under bias potential Minoru Otani, Ikutaro Hamada, Osamu Sugino, Yoshitada Morikawa, Yasuharu Okamoto and Tamio Ikeshoji, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b803541e
A first-principles study of molecular oxygen dissociation at an electrode surface: a comparison of potential variation and coadsorption effects Sally A. Wasileski and Michael J. Janik, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b803157f
Electrochemical and FTIRS characterisation of NO adlayers on cyanide-modified Pt(111) electrodes: the mechanism of nitric oxide electroreduction on Pt Angel Cuesta and María Escudero, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b717396b
Electrodeposited noble metal particles in polyelectrolyte multilayer matrix as electrocatalyst for oxygen reduction studied using SECM Yan Shen, Markus Träuble and Gunther Wittstock, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802688b
CoPt nanoparticles and their catalytic properties in electrooxidation of CO and CH3OH studied by in situ FTIRS Qing-Song Chen, Shi-Gang Sun, Zhi-You Zhou, Yan-Xin Chen and Shi-Bin Deng, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802047g
Kinetic studies of adsorbed CO electrochemical oxidation on Pt(335) at full and sub-saturation coverages Prachak Inkaew and Carol Korzeniewski, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b804507k
Formic acid electrooxidation on Pd in acidic solutions studied by surface-enhanced infrared absorption spectroscopy Hiroto Miyake, Tatsuhiro Okada, Gabor Samjeské and Masatoshi Osawa, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b805955a
Voltammetric surface dealloying of Pt bimetallic nanoparticles: an experimental and DFT computational analysis Peter Strasser, Shirlaine Koh and Jeff Greeley, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b803717e
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online / Journal Homepage / Table of Contents for this issue
Unique activity of Pd monomers: hydrogen evolution at AuPd(111) surface alloys Y. Pluntke, L. A. Kibler and D. M. Kolb, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802915f
Shape-dependent electrocatalysis: methanol and formic acid electrooxidation on preferentially oriented Pt nanoparticles J. Solla-Gullón, F. J. Vidal-Iglesias, A. López-Cudero, E. Garnier, J. M. Feliu and A. Aldaz, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802703j
The role of adsorbed hydroxyl species in the electrocatalytic carbon monoxide oxidation reaction on platinum Anthony R. Kucernak and Gregory J. Offer, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802816h
An unexpected enhancement in methanol electro-oxidation on an ensemble of Pt(111) nanofacets: a case of nanoscale single crystal ensemble electrocatalysis Ceren Susut, George B. Chapman, Gabor Samjeské, Masatoshi Osawa and YuYe Tong, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802708k
Surface Pourbaix diagrams and oxygen reduction activity of Pt, Ag and Ni(111) surfaces studied by DFT Heine A. Hansen, Jan Rossmeisl and Jens K. Nørskov, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b803956a
Electrocatalytic mechanism and kinetics of SOMs oxidation on ordered PtPb and PtBi intermetallic compounds: DEMS and FTIRS study Hongsen Wang, Laif Alden, F. J. DiSalvo and Héctor D. Abruña, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b801473f
Combinatorial screening of PtTiMe ternary alloys for oxygen electroreduction Ting He and Eric Kreidler, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802818b
Electrochemical study on the adsorption of carbon oxides and oxidation of their adsorption products on platinum group metals and alloys Hanna Siwek, Mariusz Lukaszewski and Andrzej Czerwinski, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b718286b
Ethanol electrooxidation onto stepped surfaces modified by Ru deposition: electrochemical and spectroscopic studies V. Del Colle, A. Berná, G. Tremiliosi-Filho, E. Herrero and J. M. Feliu, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802683a
The effects of the specific adsorption of anion on the reactivity of the Ru(0001) surface towards CO adsorption and oxidation: in situ FTIRS studies J. M. Jin, W. F. Lin and P. A. Christensen, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802701c
Experimental and numerical model study of the limiting current in a channel flow cell with a circular electrode J. Fuhrmann, H. Zhao, E. Holzbecher, H. Langmach, M. Chojak, R. Halseid, Z. Jusys and J. Behm, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802812p
Using layer-by-layer assembly of polyaniline fibers in the fast preparation of high performance fuel cell nanostructured membrane electrodes Marc Michel, Frank Ettingshausen, Frieder Scheiba, André Wolz and Christina Roth, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802813n
Stripping voltammetry of carbon monoxide oxidation on stepped platinum single-crystal electrodes in alkaline solution Gonzalo García and Marc T. M. Koper, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b803503m
PtxRu1-x/Ru(0001) surface alloys—formation and atom distribution H. E. Hoster, A. Bergbreiter, P. M. Erne, T. Hager, H. Rauscher and R. J. Behm, Phys. Chem. Chem. Phys., 2008 DOI: 10.1039/b802169d
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
Voltammetric surface dealloying of Pt bimetallic nanoparticles:
an experimental and DFT computational analysis
Peter Strasser,w*ac Shirlaine Koha and Jeff Greeleyb
Received 4th March 2008, Accepted 15th May 2008
First published as an Advance Article on the web 27th May 2008
DOI: 10.1039/b803717e
Voltammetric dealloying of bimetallic platinum–copper (Pt–Cu) alloys has been shown to be an
effective strategy to modify the surface electrocatalytic reactivity of Pt bimetallic nanoparticles
(S. Koh and P. Strasser, J. Am. Chem. Soc., 2007, 129, 12624). Using cyclic voltammetry and
structural XRD studies, we systematically characterize the Pt–Cu precursor compounds as well as
the early stages of the selective Cu surface dissolution (dealloying) process for Pt25Cu75, Pt50Cu50,
and Pt75Cu25 alloy nanoparticles annealed at both low and high temperature. We also assess the
impact of the synthesis conditions on the electrocatalytic reactivity for the oxygen reduction
reaction (ORR). To gain atomistic insight into the observed voltammetric profiles, we compare
our experimental results with periodic DFT calculations of trends in the thermodynamics of
surface Cu dissolution potentials from highly stepped and kinked Pt(854) single crystal surfaces.
The modeling suggests a dependence of the electrochemical Cu dissolution potentials on the
detailed atomic environment (coordination number, nature of coordinating atoms) of the
bimetallic Pt–Cu surfaces. The DFT-predicted shifts in electrochemical Cu dissolution potentials
are shown to qualitatively account for the observed voltammetric profiles during Cu dealloying.
Our study suggests that metal-specific energetics have to be taken into account to explain the
detailed dealloying behavior of bimetallic surfaces.
1. Introduction
Over 2000 years ago, pre-Columbian goldsmiths of the high
cultures of Andean South America developed simple surface
processing technologies in order to change the properties of
metal alloys.2 They were interested in changing the appearance
of Au-containing metal alloys to make them appear as if they
consisted of pure gold. To achieve this, they invented two
techniques: surface metal galvanic displacement and surface
metal depletion gilding.2,3
Surface metal galvanic displacement consisted of treating
thin polished Cu metal sheets in neutralized heated baths of
Au salts, which were prepared from metallic Au by dissolution
in corrosive mineral mixtures.2 The Cu sheets were uniformly
coated with a thin layer of Au and appear as if they consisted
of pure Au: an affordable strategy for impoverished royal
families to dress in gold, which, in the then cultural attitude,
represented descent from the sun.
The Surface metal depletion technology started with bime-
tallic ingots of a less noble (Cu) and a more noble metal (Au),
known as ‘tumbaga’, which were repeatedly subjected to cold-
treatments, annealing, and pickling cycles in order to enrich
the surface with copper and then deplete surface copper by
chemical dissolution. In this process, the surface of the bi-
metallic ingot became more and more enriched in Au until it
displayed a golden surface.
Both ancient metallurgic strategies for altering surface
properties have found their way into modern electrochemical
metal surface processing.3–7 While visual appearance was the
target property of interest then, the surface electronic struc-
ture, and with it, the surface catalytic properties, are the focus
of the modifications today.
In order to change the surface electrocatalytic properties of
Pt metal alloys, a modern galvanic displacement method has
been developed and utilized over the past decade,8–12 which
came to be known as the Pt Monolayer Catalyst9 concept.
Following the ancient goldsmiths’ strategy, Pt monolayers
were deposited onto non-Pt substrates using galvanic displace-
ment of layers of surface Cu atoms (Fig. 1a). The resulting
catalysts have shown improved intrinsic surface electrocataly-
tic activity for the electroreduction of oxygen at much reduced
Pt content; they represent one of the most promising new
catalyst classes for fuel cell cathodes. Vukmiovic et al.8 re-
viewed the Pt monolayer concept in detail and pointed out its
structural and catalytic characteristics.
Herein, we report an experimental and DFT computational
study of a modern voltammetric depletion gilding method
designed to deliberately change the electrocatalytic surface
reactivity of Pt alloy catalysts for the electroreduction of
oxygen (ORR).1,13–16 We selectively dissolve Cu atoms (vol-
tammetric dealloying) from the surface of Pt–Cu bimetallic
compounds until a multi-layer Pt-rich shell region has formed
(Fig. 1b). While the dealloying of a less noble component from
aDepartment of Chemical and Biomolecular Engineering, Universityof Houston, Houston, Texas 77204, USA. E-mail: [email protected]
bCenter for Nanoscale Materials, Argonne National Laboratory,Argonne, IL 60439, USA
c Institut fur Chemie, Technische Universitat Berlin, 10624 Berlin,Germany. E-mail: [email protected] Secondary contact address: , , ,
3670 | Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 This journal is �c the Owner Societies 2008
PAPER www.rsc.org/pccp | Physical Chemistry Chemical Physics
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
bimetallic alloys has been investigated on flat surfaces, such as
alloy single crystals, we focus on the initial stages of Cu
dealloying from nanometer-sized alloy particle ensembles.
Metal dealloying from alloy nanoparticles is a poorly under-
stood research area, yet it has tremendous importance for the
understanding of high-surface area alloy catalyst degradation
in electrocatalysis.
We found Cu dealloying from Pt–Cu bimetallic alloy nano-
particle precursors to result in very active ORR catalysts.1,17
We investigate the voltammetric profiles of Cu dissolution
during the early stages of the dealloying process and correlate
the voltammetry with structural alloy phase characteristics at
various Cu contents and synthesis conditions. We combine our
experimental results with computational DFT predictions of
trends in the shift in thermodynamic Cu dissolution potentials
in Pt–Cu bimetallic surfaces. In our calculations, we consider
metal-specific energetics and assess their impact on the Cu
dissolution potentials.
2. Experimental methods
Preparation of the alloy catalysts
Carbon supported Pt–Cu alloy nanoparticle electrocatalysts of
stoichiometries Pt25Cu75, Pt50Cu50, and Pt75Cu25 were synthe-
sized via a liquid salt precursor impregnation method reported
earlier. Measured amounts of a Cu nitrate (Cu(NO3)2 3H2O,
Aldrich) were dissolved in de-ionized water (418.2 MO, Milli-
Qs gradient system, Millipore Inc.). The precursor solutions
were then added to a weighted amount of commercial carbon-
supported Pt nanoparticle electrocatalyst (Tanaka Kikinzoku
Inc.) with a Pt weight loading of about 30 wt%. The mixtures
were then ultrasonicated with a sonifier horn (Branson 150,
level 2) for 1 min and then frozen in liquid nitrogen for 5 min.
Subsequently, the frozen samples were freeze-dried (Labcon-
co) for 24 h and annealed in flowing hydrogen/Argon mixtures
in a flow furnace (Lindberg Blue) to a temperature of 600 and
950 1C. The annealed powders were stored in a Nalgenes
desiccator, with continuous nitrogen gas flow, for future use.
The resulting Pt loadings of the Pt–Cu catalysts ranged from
22 wt% to about 30 wt%.
Electrochemical dealloying and ORR testing
A custom-made, three-compartment electrochemical cell was
employed for all voltammetric measurements. The working
electrode was a commercial glassy carbon rotating disk elec-
trode of 5 mm diameter (PINE Instruments). A catalyst ink
was prepared by mixing 20 mg catalyst powder with 16 ml
water (Millipore Gradient), 4 ml isopropanol (IPA) and 80 mlof a 5 wt% Nafions solution (Sigma, #274704) , followed by
sonication for 15 min at level 2 (Branson 150). A 10 ml aliquotwas dispensed onto the 5 mm working electrode disk resulting
in a typical geometric Pt loading of about 10 mg Pt/cm2. The
applied absolute amount of Pt of the catalyst thin films on the
RDE varies with stoichiometry of the electrocatalysts and
decreased in the order Pt 4 Pt75Cu25 4 Pt50Cu50 4 Pt25Cu75.
The reference electrode was a mercury–mercury sulfate
electrode. All electrode potentials were subsequently con-
verted into and are reported with respect to the reversible
hydrogen electrode (RHE) scale. The counter electrode was a
piece of platinum gauze to ensure large surface area. A
commercial rotator from Pine Instrument was used to conduct
the rotating disk experiment. The electrolyte used was 0.1 M
HClO4, prepared by diluting 70% redistilled HClO4 (Sigma
#311421) with de-ionized water. The disk potential was con-
trolled with a potentiostat, BiStat (Princeton Applied Re-
search, Ametek). All measurements were conducted at room
temperature. At the beginning of electrochemical measure-
ments, electrocatalysts were immersed into the electrolyte
under potential control and held at 0.05 V/RHE until the
measurements commenced. Dealloying was conducted in a de-
aerated N2 atmosphere electrolyte. The initial three cyclic
voltammetric cycles (CVs) were recorded at a scan rate of
100 mV s�1 between 0.05 V and 1.2 V. Thereafter, dealloying
was continued for 200 CV cycles over the same potential range
at a scan rate of 500 mV s�1, until a stable CV was obtained.
Subsequently, another slow CV was recorded at 100 mV s�1.
The platinum electrochemical surface area (Pt-ECSA) of the
catalyst was determined from that latter CV via the mean
integral charge of the hydrogen adsorption and desorption
areas after double-layer correction, using 210 mC cm2Pt as the
conversion factor. ORR measurements were conducted by
linear sweep voltammetry (LSV) from 0.05 V to the open
circuit potential (around 1.0 V) at the scan rate of 5 mV s�1. Pt
surface area normalized activities were established at 900
mV/[RHE], at room temperature by correcting the measured
current for mass-transport effects18–20 and dividing the result-
ing current by the Pt-ECSA. The electrochemical behavior
Fig. 1 Two distinct ancient pre-Columbian metallurgical techniques
for modifying the optical surface properties of noble metal/non-noble
metal bimetallics serve as the basis of two of the most promising
synthetic strategies towards more active oxygen reduction reaction
(ORR) electrocatalysts. (a) The ancient method of ‘Galvanic displace-
ment’ has been used to create Pt monolayer nanoparticles. (b) ancient
‘depletion gilding’ is the basis of a voltammetric dealloying technique
to modify the surface catalytic properties of bimetallic nanoparticles.
This journal is �c the Owner Societies 2008 Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 | 3671
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
(CV and LSV) of the Pt–Cu catalysts was compared to a
30 wt% reference platinum electrocatalyst supported on a high
surface area carbon support.
X-Ray diffraction measurements
Laboratory-source XRD (Siemens D5000 y/2y Diffracto-
meter) was used to characterize the electrocatalysts structu-
rally. The diffractrometer is equipped with a Braun Position
Sensitive Detector (PSD) with an angular range of 81. The Cu
Ka source operates at a potential of 35 kV and current of
30 mA. 2y diffraction angles ranged from 20–701, scanned with
step size of 0.021 per step and holding time of 10–30 s per step.
Advanced X-ray Solution (X-ray commander, Bruker AXS)
software was used to control the diffractometer from a desk-
top computer.
The XRD sample holder was a custom-made 3 � 3 cm
plexiglass with a 1 cm width � 2 cm length � 1 mm depth well
in the center that hold the powder catalyst samples. The
catalyst powder was poured into the well and carefully flat-
tened to form a smooth surface, flushed with the surface of the
plexiglass.
All XRD diffraction patterns were analyzed using Jade
(MDI). Each peak profile of each reflection was obtained by
a non-linear least square fit of the Ka2 and background
corrected data. Instrumental broadening was determined using
an alumina standard under identical measurement conditions.
Particle sizes were estimated using the observed line broad-
ening and Scherrer equation at several fundamental reflec-
tions. Compositions of Pt–Cu alloys obeyed Vegard’s law21,22
which states that the composition of an alloy is directly
proportional to the lattice constant of the alloy with respect
to those of its components.
3. Theoretical and DFT computational models and
methods
The DACAPO code23 was used for all total energy calcula-
tions in this study. The Pt(854) surface was modeled with a
40-atom unit cell periodically repeated in a super cell geometry
with at least 17 A of vacuum between any two successive metal
slabs; the density functional theory-optimized Pt lattice con-
stant of 4.02 A is in reasonable agreement with the experi-
mental value of 3.91 A.24 The unit cell geometry is such that
either three or four metal layers (depending upon the location
on the slab surface and the number of surface atoms that have
dissolved) perpendicular to the (111) terraces are employed;
the top layer of Pt atoms is substituted with Cu, and Cu atoms
are sequentially removed to simulate Cu dissolution (see also
section 4.4). The calculations are fully converged with respect
to the number of metal layers included; removal of the bottom
layer of Pt atoms, and subsequent recalculation of the differ-
ential binding energies of surface Cu atoms at high coverages
(40.69 ML), shows average differential BE changes of less
than 0.02 eV (the actual differential BE changes calculated
were 0.011, �0.006, 0.024, and 0.037 eV for dissolution from
slabs with 1, 0.92, 0.85, and 0.77 ML of adsorbed Cu atoms,
respectively); these values are well within the accuracy range
associated with standard DFT calculations. The top (Cu) layer
of the slabs is relaxed until the total force on all atoms is less
than 0.04 eV A�1 in any Cartesian direction. Interslab dipole
corrections are not included in the reported results.13,25 Ionic
cores were described by ultrasoft pseudopotentials,25 and the
Kohn–Sham one-electron valence states were expanded in a
basis of plane waves with kinetic energy below 340 eV; a
density cutoff of 500 eV was used. The surface Brillouin zone
was sampled with a Chadi-Cohen 6(1�1) k point grid. The
convergence of the total energy with respect to the cutoff
energies and the k point set was confirmed. The exchange–
correlation energy and potential were described by the general-
ized gradient approximation (GGA-RPBE).23 The self-consis-
tent RPBE density was determined by iterative
diagonalization of the Kohn–Sham Hamiltonian, Fermi
population of the Kohn–Sham states (kBT ¼ 0.1 eV), and
Pulay mixing of the resulting electronic density.26 All total
energies were extrapolated to kBT ¼ 0 eV.
The approach for modeling the thermodynamic aspects of
metal dissolution on surfaces at standard ion concentrations
has been described previously;27 briefly, the raw, DFT-derived
total energies are corrected for electrode potential effects; the
latter effects are estimated by referencing the potential to the
standard hydrogen electrode. The dissolution potential of bulk
copper,24 U1, is taken as an empirical reference point, and all
shifts in dissolution potential due to changes in the surface
chemical environment are referenced to this value. The result-
ing expression for the total free energy of the Cu atoms on the
Pt(854) substrate, neglecting surface entropy effects, is:
GðNCu;UÞ ¼ Eslab;NCu� Eclean � ðEbulk;Cu þ 2ðU��UÞÞNCu;
ð1Þ
where NCu is the total number of Cu atoms per unit cell on the
Pt(854) surface, Eslab;NCuis the total, DFT-derived energy of
the slab with adsorbed Cu atoms, Eclean is the Pt(854) energy
with no adsorbed Cu, and Ebulk,Cu is the cohesive energy of
bulk Cu. Implicit in this expression is the assignment of a value
of 0 to the free energy of the clean Pt slab (with no adsorbed
Cu atoms). In all cases, Eslab;NCurefers to the most stable
arrangement of Cu atoms at the indicated value of NCu; test
calculations demonstrate that these most stable arrangements
are associated with removal of Cu atoms from the Cu kinks
(removal of Cu atoms from terraces or step edges is less
energetically favorable). We note that no corrections for
potential-induced electric field effects are included in the
calculated free energies; such effects have been shown to have
a negligible impact on the thermodynamics in computational
studies of the oxygen reduction reaction.28 Finally, in eqn (1),
it is assumed that two electrons are transferred to the substrate
for every copper ion that dissolves; this choice of this charge
state is motivated by the fact that Cu21 ions dissolve at lower
potentials than Cu1 ions.24,29
4. Results
4.1 ORR electrocatalysis of dealloyed Pt–Cu bimetallic alloy
nanoparticles
Six Pt–Cu bimetallic precursor alloy particle catalysts—three
different stoichiometries (Pt25Cu75, Pt50Cu50, Pt75Cu25) at
two different annealing temperatures (950 and 600 1C)
3672 | Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 This journal is �c the Owner Societies 2008
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
(see Table 1)—were electrochemically dealloyed and subse-
quently subjected to slow anodic potential sweeps in oxygen
saturated electrolyte in order to characterize their intrinsic
oxygen reduction reaction (ORR) activity. Fig. 2 shows the
anodic potential sweeps of the dealloyed catalysts in oxygen-
saturated electrolyte. The flat portions of the sweep curves for
potentials below 0.8 V/RHE indicate a diffusion-limited re-
gime. For increasing Cu content of the precursor, the steep
portions of the curves are significantly shifted to higher
electrode potentials compared to the pure Pt standard catalyst
(dashed curve), indicative of an enhanced electrocatalytic
reaction rate for ORR (reduced overpotential) on dealloyed
Pt–Cu surfaces. Combined with the Pt amounts of the alloy
catalysts in the RDE film, a kinetic analysis resulted in a 4–5
fold increase in the Pt-mass normalized ORR reactivity for the
Cu-richest low temperature catalyst compared to the estab-
lished 0.10 A per mgPt of pure Pt at 0.9 V/RHE30 (see
Table 1). This dramatic increase is in accord with previously
reported RDE measurements of dealloyed Cu-rich Pt–Cu
nanoparticle catalysts.1 With increasing Pt content of the
precursor material, the observed Pt mass activity dropped to
values reported earlier for Pt rich alloys such as Pt75Ni25 or
Pt75Co25.31–35 Catalysts prepared at high temperatures gener-
ally showed reduced Pt mass based ORR activity.
Table 1 and the inset of Fig. 2 also report the measured
values of the Pt-ECSA and the Pt-ECSA normalized ORR
activities of the three dealloyed catalysts at 0.9 V/RHE. The
Pt25Cu75 catalyst prepared at 950 1C exhibited an almost 4 fold
intrinsic activity increase over pure Pt, in line with earlier
reports of this catalyst class.1,15,17,36 Low temperature
Pt25Cu75 and Pt50Cu50 catalysts showed slightly smaller activ-
ities, while both Pt75Cu25 and the high temperature Pt50Cu50only showed a 1–2� increase in specific activity, similar to
other Pt rich catalysts.30–32,37–49 The electrochemical surface
areas of the dealloyed catalysts indicate that none of the
catalysts showed a significantly increased surface area com-
pared to a standard carbon-supported Pt with about 77 m2).
To the contrary, the higher annealing temperature led to a
serious decrease of Pt-ECSA due to particle sintering.
4.2 Structural characteristics of Cu rich Pt–Cu bimetallic
alloy nanoparticles
To gain insight into the compositional and structural char-
acteristics of the Pt–Cu alloy precursor nanoparticle electro-
catalysts, we performed X-ray diffraction studies of all six
alloys prior to dealloying.
Pt25Cu75. Fig. 3a and b show the measured profiles of the
Pt25Cu75 particles over the 2y range of the fundamental (111)
and (200) peaks of Pt and Cu annealed at low and high
temperatures, respectively. The broad peaks around 2y ¼41.91 and 2y ¼ 48.51 envelop the (111) and (200) reflections
of fcc Pt–Cu bimetallic alloy phases.50,51 The absence of
superlattice peaks at lower angles confirmed the presence of
disordered fcc Pt–Cu alloys. Given the alloy nanoparticle
ensemble, the broad alloy phase peaks indicate the presence
of multiple Pt–Cu bimetallic fcc phases. In fact, as shown in
Fig. 3a, the broad (111) peak was fit using two distinct peaks at
2y ¼ 42.31 and 2y ¼ 40.91, representing two distinct Pt–Cu
alloy phases. Furthermore, the low temperature sample shows
sharp peaks around 2y ¼ 43.31 and 2y ¼ 50.51, representing
the (111) and (200) reflections of a face-centered cubic (fcc) Cu
phase. Vegard’s law, which establishes a linear relationship
between the unit cell parameter of a bimetallic alloy and its
composition, provides an excellent model to predict trends in
the composition of crystalline Pt–Cu bimetallic alloy phases
based on their peak position.21,22,52 The XRD peak analysis of
Table 1 Comparison of nominal and final compositions, Pt mass based catalytic ORR activities (at 900 mV/RHE), surface area normalized ORRactivities (at 900 mV/RHE), and surface areas of Pt–Cu alloy nanoparticle electrocatalysts
Norminalcomposition (at.%)
Annealilngtemperature
Specificactivity
Pt massactivity
Surface area Composition afterelectrochemical dealloying (at.%)
Catalyst Pt Cu 1C mA cm�2Pt A mg�1Pt m2 g�1Pt Pt Cu
1 25 75 600 643.77 0.53 88.51 79 212 25 75 950 787.68 0.35 45.16 80 203 50 50 600 595.96 0.51 85.83 — —4 50 50 950 427.61 0.11 26.80 — —5 75 25 600 359.71 0.27 74.77 86 146 75 25 950 233.85 0.03 13.01 81 19
Fig. 2 Sweep voltammetry of dealloyed carbon-supported Pt–Cu
nanoparticle electrocatalysts of various Pt to Cu ratios compared to
a standard Pt catalysts. Conditions: 5 mV s�1, 0.1 m HClO4, 25 C,
1600 rpm. Stoichiometries in the legend indicate the compositions of
the precursor alloys. Inset: surface area normalized current density
(specific activity) of de-alloyed Pt–Cu alloy nanoparticle catalysts at
900 mV/RHE. Dealloyed Cu rich Pt–Cu catalysts exhibit a clear
activity advantage over pure Pt catalysts (see also Table 1). Cu richer
precursors result in more active dealloyed catalysts.
This journal is �c the Owner Societies 2008 Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 | 3673
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
the Pt25Cu75 particles thus revealed that low-temperature
annealing (600 1C) resulted in incomplete alloying of Pt and
Cu atoms, with excess Cu atoms forming large pure Cu
particles. Coexisting with the pure Cu phase are multiple
Pt–Cu alloy phases with varying compositions, some of which
must consist of a Pt content of 4 25 atomic % in order to
meet the overall stoichiometry of Pt25Cu75. The high-tempera-
ture catalyst in Fig. 3b, however, shows no sign of excess Cu.
The fundamental fcc (111) reflection exhibits a much smaller
full peak width at half maximum (FWHM). A peak fit of the
(111) reflection still required at least two different individual
peaks (Fig. 3b), suggesting the presence of more than a single
alloy phase. Yet, overall, annealing at 950 1C clearly leads to
an improved alloying between Pt and Cu atoms.
Pt50Cu50. Fig. 3c and d show the structural XRD profiles of
the Pt50Cu50 alloys annealed at 600 1C and 950 1C, respec-
tively. As with Pt25Cu75, broad reflections for the low tem-
perature catalyst indicated small particles and/or multiple
alloy phases. The low temperature profile was successfully fit
with two distinct (111) and (200) peak couples, suggesting the
presence of at least two bimetallic alloy phases. The high
temperature material profile, on the other hand, suggested
the presence of a rhombohedral Pt50Cu50 Hongshiite
phase53–56 (PDF file 00-042-1326, space group: R�3m, unit cell
parameters a, b ¼ 1.070 nm, c ¼ 1.319 nm). Consistent with
our measurements, this phase exhibits an intense (006) reflec-
tion at 2y ¼ 41.011 and a (404) reflection at 2y ¼ 47.91. Within
our probed angle range, the phase is uniquely identified by its
characteristic (205) and (241) reflections at 2y ¼ 39.21 and
2y ¼ 52.61, respectively. These latter two peaks are located at
angles outside the range between the fundamental (111) and
(200) reflections of pure Pt and pure Cu phases; hence, they
cannot be assigned to any disordered Pt–Cu bimetallic alloy
phase which would show its (111) and (200) reflection in
between those of the two pure metals.
Pt75Cu25. Finally, Fig. 3e and f depict the diffraction profile
of the Pt75Cu25 alloy nanoparticle electrocatalyst annealed at
low and high temperatures, respectively. As expected, the low
temperature profile exhibits broader peaks associated with
smaller alloy nanoparticles.50 However, both profiles can be
fit very well with one single Voigt-type peak, suggesting a
relatively homogeneous alloying in the Pt-rich composition
range. No sign of excess pure Cu was seen in either the low or
the high temperature alloy samples.
4.3 Cyclic voltammetric behavior during Pt–Cu dealloying
This section focuses on both the early stage and the final stage
of the catalyst dealloying process. Catalyst dealloying is
typically performed prior to ORR testing in deaerated electro-
lyte by voltammetric treatment of the bimetallic precursors.
Fig. 4a–f reports the initial cyclic voltammetric cycles of all six
Pt–Cu catalysts of interest.
Pt25Cu75. The first cyclic voltammograms (CVs) of the
Pt25Cu75/600 1C catalyst (Fig. 4a) show no hydrogen adsorp-
tion between 0.05 and 0.3 V, suggesting a very Cu-rich
precursor surface. A strong anodic current onset is observed
at around 0.25 V, forming a very broad anodic multi-peak
band stretching from 0.25 to 0.85 V with three individual peak
maxima at 0.4, 0.58, and 0.7 V. This broad anodic current
peak represents surface Cu dissolution from the as-prepared
precursor bimetallic surface. On the cathodic portion of the
first scan, a number of reduction peaks were observed at 0.6,
0.4 and 0.25 V, the latter two of which are likely associated
with redeposition of some Cu-ions from solution onto the
interface. On the anodic portion of the second and third CV,
the broad peak band gave way to a two-peak profile at 0.35
and 0.7 V. On the cathodic direction, the reduction peak
currents gradually decreased, and the initial three-peak profile
turned into a single reduction peak around 0.65 V. The third
CV shows some oxidation of underpotentially deposited hy-
drogen in the 0.1–0.3 V range, indicating the onset of the
exposure of Pt surface atoms, even at these very low poten-
tials. The voltammetric cycling was then continued until a
time-stable CV was obtained. The final recorded cyclic vol-
tammogram of the Pt25Cu75 is shown in Fig. 4a as well (solid
line). The final profile strongly resembles that of a pure Pt
catalyst surface with a well-developed, underpotentially-
deposited (upd) hydrogen ad/desorption region and the
formation of oxygenated surface species at anodic potentials
Fig. 3 X-Ray diffraction profiles of carbon supported Pt–Cu alloy nanoparticle precursor catalysts of varying stoichiometries and annealing
temperatures. All annealing time were 7 h. (a) Pt25Cu75 600 1C; (b) Pt25Cu75 950 1C; (c) Pt50Cu50 600 1C; (d) Pt50Cu50 950 1C; (e) Pt75Cu25 600 1C;
(f) Pt75Cu25 950 1C.
3674 | Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 This journal is �c the Owner Societies 2008
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
40.80 V. The final CV suggests that dealloying led to an
almost pure Pt catalyst surface by gradual removal of Cu
atoms from the alloy particles.
The first three CVs of the Pt25Cu75/950 1C (Fig. 4b) ex-
hibited a distinctly different peak profile upon dealloying.
While the onset potential of the Cu dissolution again occurred
around 0.25 V with a current peak maximum at 0.4 V, there
were no Cu intermediate dissolution current peaks in the
0.4–0.8 V range. Another current peak maximum was obser-
vable at 0.8 V on the first scan. The first cathodic sweep was
similar to that of the low temperature material with a pro-
nounced peak around 0.65 V and smaller ones at 0.4 and
0.25 V. On the second and third cycles, the Cu dissolution
decreased in peak current and shifted slightly more cathodi-
cally compared to the first cycle. The final CV again resembled
a pure Pt catalyst with a small residual peak at around 0.7 V,
likely associated with continued Cu dissolution (see Discus-
sion). A clear, yet smaller amount of underpotentially depos-
ited hydrogen was observed to ad- and desorb in the 0.05–
0.4 V range.
Pt50Cu50. The first 3 CVs of the low and high temperature
Pt50Cu50 precursors are shown in Fig. 4c and d, respectively.
The behavior and profiles of the two catalysts are very similar
to each other and share most of their features. The anodic
sweep of the first scan immediately exhibits upd-hydrogen
oxidation followed by a relatively small Cu dissolution peak at
0.4 V. Similar to later stages of the Pt25Cu75 behavior, a
second asymmetric Cu dissolution peak appears at 0.7 V
during the first CV. Already on the second CV, the dissolution
peak near 0.4 V disappeared completely, indicating a limited
amount of Cu in the surface region. During the second and
third CV, the broad peak at 0.7 V was replaced by a sigmoid-
ally shaped voltammetric feature which gradually decreased in
intensity and finally disappeared completely. No significant
redeposition peaks were discernible on the cathodic portion of
any of the initial CVs. The final CV of Pt50Cu50 again strongly
resembled that of pure Pt, suggesting no residual Cu atoms in
the surface. A stark difference between the low and high
temperature material consisted in the magnitude of the H
upd integrals of the initial and the final time-stable CV. The
upd hydrogen integral, typically associated with the surface
area of Pt surfaces, was much larger for the Pt50Cu50/600 1C
material than for the Pt50Cu50/950 1C catalyst, in agreement
with the notion that high annealing temperature resulted in
larger alloy particles with reduced surface area.
Pt75Cu25. Fig. 4e and f report the initial 3 CVs of the
Pt75Cu25 precursor. Unlike the other catalysts, no Cu dissolu-
tion peak around 0.4 V was observable. Slight Cu dissolution
was indicated for the low temperature material by a peak at
around 0.7 V, which rapidly dropped in intensity during the
second and third cycles. As in the other two cases, the final CV
was in large part identical to that of a pure Pt surface, with the
high temperature profile showing a drastically reduced H upd
integral, suggesting large dealloyed nanoparticles.
4.4 Trends in the thermodynamics of Cu dissolution on Pt–Cu
alloy surfaces
The working surface of the Pt–Cu alloy nanoparticles is clearly
complex, involving both strong compositional profiles near the
surface and a variety of effects resulting from undercoordi-
nated surface sites. To gain insight into the detailed, atomic-
level energetic factors influencing the dissolution of Cu in these
bimetallic systems, we perform periodic density functional
theory calculations. We consider the dissolution of a Cu
monolayer from a kinked Pt(854) surface; this model provides
atomic-level insight into the effect of defects and other rough
surface features on the thermodynamics of Cu dissolution
(a detailed discussion of the relationship between the model
and the experimental systems is presented in section 5.2).
Table 2 gives the calculated binding energies of Cu atoms to
the Pt(854) surface as a function of Cu coverage. These data,
corrected for potential effects, are used to plot the surface free
energy of different numbers of Cu atoms on the Pt(854)
substrate at different potentials (Fig. 5(a)–(c)—the potential
scale, U, is referenced to the experimentally-measured equili-
brium potential for Cu21 dissolution from bulk Cu, 0.34 V vs.
Fig. 4 Initial three cyclic voltammetric profiles during the dealloying process of Pt–Cu catalysts from Fig. 3. (a) Pt25Cu75/600 1C; (b) Pt25Cu75/
950 1C; (c) Pt50Cu50/600 1C; (d) Pt50Cu50/950 1C; (e) Pt75Cu25/600 1C; (f) Pt75Cu25/950 1C. The solid line displays the cyclic voltammogram at the
end of the dealloying process. Conditions: 10 mV s�1, deaerated 0.1 M HClO4, 25 C.
This journal is �c the Owner Societies 2008 Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 | 3675
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
SHE). The maximum number of Cu atoms considered, 13, is
equivalent to a full monolayer of Cu. If kinetic effects are
absent, then at a given value of the potential, Cu dissolution
will proceed until further dissolution would require an increase
in the free energy of the surface. Hence, at a value of UE 0.34
V, at most a single Cu atom is thermodynamically favored to
dissolve. This corresponds to a situation wherein the most
easily removed Cu atoms (those at the Cu-rich kink
sites—Fig. 5d) are primarily in contact with other Cu atoms;
hence, the dissolution potential is little shifted from the
dissolution potential of bulk Cu. At higher potentials, more
Cu atoms begin to dissolve. At B0.45 V, a large plateau in the
free energy vs. NCu plot develops; this plateau implies that a
significant number of Cu atoms will dissolve at or near a single
potential. This single dissolution potential, corresponding to
dissolution of Cu atoms from the regions between kinks
(Fig. 5e), is analogous to what might be observed in dissolu-
tion from flat regions of the Pt–Cu nanoparticles. Finally, at
potentials greater than 0.5 V, dissolution of Cu atoms that are
strongly bound to Pt step and kink edges (Fig. 5f) becomes
thermodynamically favorable. These results are summarized in
Fig. 5g, an isotherm of Cu dissolution from the Pt(854)
surface. As in earlier subpanels of Fig. 5, the number of Cu
atoms remaining on the surface at given potentials is deter-
mined by the criterion that further dissolution requires an
increase in free energy. We note that, although this isotherm is
determined on a bulk fcc Pt crystal, the qualitative trends
should remain valid for fcc Pt–Cu bulk alloys of other
compositions. We further note that, while this isotherm does
not include the effects of surface entropy on the dissolution
Table 2 Differential binding energy (B.E.) of Cu atoms to the Pt(854)surface. For a given value of NCu, the differential B.E. is defined asENCu
� ENCu�1 � Ebulk;Cu. Computational details are provided in theMethods section in the text (section 3)
NCu Differential Cu binding energy/eV
0 0.011 �0.142 �0.153 �0.134 �0.205 �0.216 �0.207 �0.228 �0.179 �0.3310 �0.4611 �0.4512 �0.6213 —
Fig. 5 Density functional theory analysis of Cu dissolution from the surface of a Pt(854) substrate. (a)–(c) Top views of the Pt(854) surface with
13, 6, and 3 Cu atoms per unit cell, respectively, adsorbed on the surface. The yellow lines in (a) denote the surface unit cell. Cu atoms are shaded
ochre, and Pt atoms are shaded silver-gray. (d)–(f) Free energy per unit cell (eV) for different Cu coverages at potentials of 0.34, 0.45, and 0.52 V vs.
SHE, respectively; for all potential values, the equilibrium potential for bulk Cu dissolution is taken from standard experimental tables. (g)
Isotherm of the most thermodynamically stable Cu coverage as a function of the electrode potential.
3676 | Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 This journal is �c the Owner Societies 2008
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
potentials, it directly accounts for site-specific changes in
adsorption energy due to both changes in Cu coverage and
the structural heterogeneity of the Pt(854) substrate.
5. Discussion
The selective dissolution of a less noble component from a
bimetallic alloy nanoparticle is commonly referred to as
‘‘metal dealloying’’;4,57–60 at the atomic level, it continues to
remain a poorly understood process. The onset potential of
selective dissolution from the alloy is conventionally referred
to as the ‘‘critical dissolution potential’’. Dealloying is often
associated with the formation of porous structures provided
that the initial alloy composition was below the so-called
parting limit, that is the maximum noble metal content for
which continuous dissolution (dealloying) occurs.59 However,
the underlying physics controlling the dissolution potential as
function of alloy characteristics, such as composition, surface
roughness and others, is unclear. Unlike previous reports on
dealloying on flat surfaces, our experimental results focus on
the Cu dissolution in its early stages from Pt–Cu alloy
nanoparticle surfaces, that is, strongly curved and very rough
(stepped, kinked) surfaces. Our computational results address
trends in the thermodynamics of the Cu dissolution potential
from first principles, introducing quantum mechanical DFT
calculations of metal- and site-specific binding energies into
the discussion of metal dealloying. The importance of metal
specific energetics for deeper insight in the physics of the
critical dissolution potential on stepped and kinked surfaces
will now be discussed, and compared and contrasted to our
experimental results.
5.1 Synthesis, structure, and ORR activity
Synthesis. We have used an impregnation/freeze drying/
annealing synthesis method for the preparation of carbon-
supported Pt–Cu bimetallic alloy nanoparticle electrocata-
lysts. The nature of this synthesis accounts for the some of
the unusual multi-phase structural features discussed below.
As described in an earlier report,1,14 the synthesis does not
involve any liquid-phase reduction, precipitation, filtration, or
other steps where incomplete reaction or materials transfer
could occur. As a result, it yields very controlled (average)
metal stoichiometries for the final catalysts. The preparation
employs carbon-supported Pt nanoparticle precursors which
are impregnated with Cu salts.61 Thereby, the pre-existing
metallic Pt particles serve as centers for alloy formation; their
high dispersion (small mean diameter) keeps the resulting alloy
mean particle size small compared to co-impregnation techni-
ques in the absence of structure-directing agents. During the
annealing process, Cu ions are reduced to the metallic state
and diffuse into the preexisting Pt metal particle lattices,
gradually forming uniformly alloyed Pt–Cu alloy lattices.
Depending on the desired average Pt to Cu ratio, different
amounts of Cu atoms have to diffuse into the preexisting Pt
particles. This is where the pronounced sensitivity of this
method to annealing conditions arises and accounts for the
observed multiphase structure for Cu-rich stoichiometries.
Preparation of Pt25Cu75 requires about 10 times as many Cu
atoms (compared to preparation of Pt3Cu) to inter diffuse with
Pt particles in order to form a uniformly alloyed material.
The synthesis method used here differs from all-liquid-
precursor impregnation/liquid-phase reduction methods. In
these, often assisted by surfactants,62,63 metal ions are pre-
mixed to achieve atomic proximity and then impregnated onto
high surface area supports, followed by reduction and alloy-
ing. While the surfactants may provide a means to control the
resulting alloy particle size, they are generally difficult to
remove afterwards without compromising the particle disper-
sion. Co-impregnation/ reduction of two or more metal salts in
the absence of surfactants generally results in large alloy
particles.
Alloy uniformity and phase structure. Alloy nanoparticle
ensembles are generally not characterized by discrete proper-
ties but exhibit distributions of particle size and composi-
tion.64–69 For the Pt25Cu75 catalyst, low temperature annealing
results in small, yet incompletely alloyed, Pt–Cu particles, as
observed in Fig. 3a. Incomplete alloying was obvious from the
presence of excess Cu. The broad (111) reflection of the alloy
phases (Fig. 3a) also suggested the presence of multiple Pt–Cu
phases with different Pt-to-Cu atomic ratios. Annealing at
higher temperatures (Fig. 3b) resulted in more uniform Pt–Cu
alloys, yet also larger average particles.
In the Pt50Cu50 catalyst (Fig. 3c and d), no Cu fcc reflections
are observed, indicating that all Cu has diffused into Pt. While
the low temperature phase structure still suggests the presence
of multiple Pt–Cu alloy phases, the high temperature profile
shows the rarely reported single rhombohedral Hongshiite
Pt–Cu phase.53,55,56 The small peak at 2y ¼ 39.21 is a uniquely
characteristic reflection of this ordered Pt–Cu alloy phase. In
contrast to all other alloy phase reflections in Fig. 3, this peak
cannot arise from a disordered Pt–Cu fcc phase, because
insertion of Cu atoms into Pt fcc lattices invariably leads to
lattice-contracted Pt–Cu alloy unit cells with (111) reflections
at larger 2y compared to the pure Pt (111), at 2y ¼ 39.71.
In the Pt75Cu25 catalyst (Fig. 3e and f), even at an annealing
temperature of 600 1C, the Cu atoms form uniformly alloyed
disordered fcc Pt–Cu alloy phases. Again, with increasing
annealing temperature, the mean alloy particle size increases,
as seen by a comparison of the FWHM of the (111) reflections
in Fig. 3e and f.
ORR activity of Pt–Cu alloys. In earlier reports,1,36,70,71 we
described the significant increase in the ORR activity achieved
using dealloyed Pt25Cu75 bimetallic nanoparticle electrocata-
lysts. Prepared at 600 and 800 1C, the dealloyed Pt25Cu75catalyst precursors were found to exhibit a 4–6 fold activity
advantage over a standard carbon supported 45 wt% Pt
nanoparticle catalyst in terms of Pt mass-based (0.4–0.6 A
mg�1Pt) and Pt surface area-based activity (700–900 mAcm�2Pt).
1 Similar activity increases were recently also realized
in single MEAs under realistic fuel cell conditions.17
The current results provide insight into the relationship
between the ORR reactivity and the Cu content of the
electrocatalyst precursor alloy. The compositional data given
in the last column of Table 1 evidences that Cu dealloying does
occur during voltammetric treatment of the Cu-rich Pt–Cu
This journal is �c the Owner Societies 2008 Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 | 3677
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
precursors (catalysts 1 and 2). The Pt75Cu25 catalyst precursor
showed better corrosion stability and suffered less relative
Cu loss.
Annealed at 600 1C, the dealloyed active catalysts seem to
maintain their high Pt-mass and specific activity, at least up to
a Pt atomic ratio of 50% (see Table 1, Catalysts 1 and 3).
Annealed at 950 1C, alloy particle growth generally reduced
the ESCA which, in turn, reduced the Pt mass activity. The
alloy phase uniformity, however, improved and the individual
phase compositions approached that of the overall stoichio-
metry. The Pt25Cu75/950 1C precursor characterized by the
largest particle size and by uniform alloying exhibited the
highest intrinsic activity after Cu dealloying; this is in agree-
ment with recent MEA measurements of dealloyed Pt25Cu75catalysts.17 This suggests that a highly dispersed, uniformly
alloyed Pt25Cu75 precursor might result in a very favorable
dealloyed catalyst from both a Pt-mass and an intrinsic
activity point of view.
The opposite relation between annealing temperature and
intrinsic activity was observed for Pt50Cu50. This analysis
shows that the alloy phase structure characteristics (ordered,
disordered, multi phase or single phase, etc.) significantly
impact the catalytic activity of the resulting dealloyed
catalysts.
The Pt-rich Pt75Cu25 precursors showed a familiar activity
increase of about 2–2.5 � over pure Pt at low annealing
temperatures, in line with many previous reports on Pt alloy
ORR catalysts with about 75 at.% Pt.35,38,45,72,73 Pt-rich alloys
annealed at high temperature (large particles) did not show
any activity advantage at all.
To put the observed ECSA-normalized activity values of
dealloyed Pt–Cu bimetallic catalysts in perspective with the
corresponding activities of extended flat surfaces, we point out
that flat Pt and Pt alloy surface exhibit much larger ESCA
normalized (specific) ORR activity.30,31,33,74,75 Pt-mass based
ORR activities are meaningless for extended surfaces. Flat
pure Pt shows a specific ORR activity of about 1500 mAcm�2Pt,
30,74 while extended Pt alloy surfaces show the familiar
2–2.5 fold enhancement (2000–4000 mA cm�2Pt).30,74
Selective metal dissolution from flat Pt alloy surfaces for the
purpose of improved electrocatalysis was reported as early as
1988 by Paffett et al.76 The authors investigated the impact of
Cr surface dissolution from flat Cr-rich Pt–Cr precursors on
the observed ORR activity. They observed significant surface
roughening due to Cr dissolution, which led to increases in the
absolute (geometric area-normalized) ORR currents. When
normalized to the real surface area, however, the ORR
activities of Pt–Cr was comparable to that of pure Pt. The
authors concluded that the addition of Cr to a Pt surface, and
the associated surface roughening, does not provide a sub-
stantial benefit for ORR activity. Other reports where de-
alloyed flat metal surfaces were proposed for use as high
surface area interfaces for catalysis and sensing include the
work by Ding et al.57 as well as Liu et al.77 Both studies
emphasize the advantage of nanoporosity derived from flat
surfaces for achieving improved catalyst dispersion.
As outlined earlier1,17 and confirmed in the present study,
the mechanistic origin of the ORR activity improvements of
carbon-supported dealloyed Pt–Cu nanoparticle catalysts
compared to carbon-supported Pt catalysts is not based on
an increased surface area. We believe that the increased
reactivity of dealloyed Cu-rich Pt–Cu nanoparticle catalysts
lies primarily in the modified geometric and electronic proper-
ties of the Pt-enriched nanoparticle shell. This is in analogy to
other core–shell type catalyst concepts, such as the ‘‘Pt mono-
layer’’ particle catalysts reported by Adzic et al.9 or the
‘‘Pt-skin’’ catalyst concept introduced by Markovic and Sta-
menkovic.74,75,78 The ‘‘Pt monolayer’’ and ‘‘Pt skin’’ catalysts
are characterized by an immediate neighborhood of Pt surface
atoms and non-Pt atoms in the second layer, resulting in
changes in the Pt–Pt distance on the surface (geometric effects)
as well as changes in the electronic structure due to charge
transfer (ligand effects). Both effects are convoluted in these
catalyst structures. In contrast, tuning the thickness of the Pt
enriched shell of dealloyed Pt–Cu particles should, in princi-
ple, allow a separation between geometric and ligand effects by
isolating geometric effects for thicker shells.
Recently, the hypothesized core-shell structure of our active
catalysts was confirmed using anomalous small angle X-ray
scattering experiments.50 Compressive strain in the Pt-en-
riched particle shell appears to play a critical role in the
activity enhancement.
5.2 Experimental and theoretical Cu dissolution potentials
Pt25Cu75. The absence of H upd peaks during the first CV in
Fig. 4a and b for the two Pt25Cu75 precursors indicates an
essentially pure Cu surface and is in agreement with the
tendency of Cu to segregate to the surface of Pt–Cu alloys
under thermal treatment.79 The onset potential of the broad
anodic faradic process in Fig. 4a and b in the 0.30–0.35 V
range is associated with pure Cu dissolution according to
Cu(s) - Cu21 þ 2e� E1 ¼ þ0.34 V/NHE. (2)
Formation of Cu(I) species is not favored under the chosen
conditions.29 We suspect that the anodic process represents the
dissolution of Cu surface atoms embedded in a Cu environ-
ment (Cu bulk atoms) and denote these Cu surface atoms as
Cu@Cu. Considering the alloy phase structure discussion
above, multi layers of pure Cu are indeed present in the
Pt25Cu75/600 1C sample, both in the form of pure Cu particles
as well as in the form of surface-segregated Cu layers on top of
the Pt–Cu alloy particles for both the Pt25Cu75/600 1C and the
Pt25Cu75/950 1C precursor. The behavior of the Cu@Cu
atoms is qualitatively similar to that of the first one or two
Cu atoms to dissolve from the monolayer on the Pt(854)
surface in our theoretical calculations (see discussion below);
as noted above, those atoms are primarily in contact with
other Cu atoms and are calculated to dissolve at essentially the
bulk Cu dissolution potential.
The broad voltammetric Cu dissolution feature stretching
from 0.4–0.8 V, with peaks at 0.6 and 0.75 V for the
Pt25Cu75/600 1C catalyst, indicates the presence of other Cu
surface atoms with distinctly more anodic dissolution poten-
tials. We suspect that, depending on their chemical environ-
ment at the atomic scale, their individual corrosion potentials
vary over a certain range. Indeed, a direct correlation between
the Cu surface atomic environment (local composition,
3678 | Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 This journal is �c the Owner Societies 2008
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
coordination number, and nature of the coordinated neigh-
boring atoms) of each individual alloy particle and its corre-
sponding local Cu corrosion potential is hypothesized and
further supported by our DFT results in Fig. 5.
For the Pt25Cu75/950 1C material, only Cu@Cu bulk
dissolution around 0.35 V and the Cu dissolution around
0.75 V are present. This is consistent with the notion of a
more uniformly alloyed nanoparticle ensemble, which shows a
narrower particle composition distribution and therefore ex-
hibits more defined corrosion potentials. This is also in
excellent agreement with anodic polarization behavior re-
ported by Pugh et al.7,80 who monitored the dealloying of
arc-melted, flat Pt25Cu75 surfaces with very well controlled
stoichiometries. Pugh’s measurements (Fig. 2 in ref. 7) showed
Cu@Cu bulk dissolution around 0.35 V, followed by reduced
currents in the 0.4–0.6 V potential range (termed ‘‘Passiva-
tion’’ by Pugh); he identified the peak at 0.75 V as the actual
Cu dissolution peak from the Pt–Cu alloy. He did not,
however, provide atomic insight into the Cu dissolution
processes.
A classic single-crystal study by Markovic and Ross of the
faradic Cu dissolution and deposition processes on Pt(111)
and Pt(100)81–83 provides valuable experimental insight into
the atomistic processes underlying our measured Cu dissolu-
tion behavior. Remarkably, the observed qualitative voltam-
metric profiles on Pt(111) and Pt(100), as well as the Cu-
related peak potentials, are nearly identical to the measured
profiles on the Pt–Cu nanoparticles. The surprising correspon-
dence between the structurally complex, compositionally var-
ied nanoparticle surfaces and the structurally well-defined
Cu@Pt(111) and Cu@Pt(100) surfaces suggests that reason-
ably flat regions may exist on our nanoparticles; it may also be
true that occasional steps or kinks are present in the single
crystal samples. The correspondence also suggests that quali-
tatively useful insights into the nanoparticle dealloying process
can be obtained from our simple Cu@Pt(854) single
crystal model.
Markovic and Ross’s combined voltammetric and LEED
studies83 related the anodic peak around 0.35 V to Cu
multilayer (Cu@Cu bulk) dissolution, while asymmetric
anodic peaks around 0.75 V were unambiguously associated
with Cu monolayer upd and Cu monolayer stripping.
It follows that Cu atoms sitting on top of pure Pt subsurface
layers (denoted here as Cu@Pt) exhibit a significantly
more anodic dissolution potential than those sitting on pure
Cu surfaces. This observation is completely consistent
with our theoretical calculations (showing an increased
dissolution potential of Cu atoms on terrace-like Pt(111)
regions compared to bulk Cu—Fig. 5e and g) and lends
support to our hypothesis that the chemical environment has
a significant impact on the detailed metal dissolution profiles.
Markovic and Ross further observed that while the Cu@Pt
dissolution occurred in a fairly narrow potential range, its
formation on the cathodic scan occurs over a broad potential
range with multiple peaks at 0.6 and 0.4 V, in line with the
observed peak potentials in Fig. 4. Bulk Cu deposition started
at around 0.2 V on the single crystals (Fig. 4 and 5 in ref. 83),
also in close agreement with the observed cathodic peaks in
this study.
Pt50Cu50, Pt75Cu25. The small Cu@Cu bulk dissolution
peak present in the Pt50Cu50 alloys (Fig. 4c and d) indicates
that some bulk-like Cu surface atoms must be present, given
the absence of unalloyed pure Cu particles. However, the
presence of H upd between 0.05–0.4 V also suggests the
presence of Pt surface atoms. Segregated Cu therefore appears
to be present as islands unable to cover the entire particle
surface. Finally, the Cu@Pt dissolution peak around 0.7–
0.75 V becomes weaker with increasing Pt content and sharply
drops in intensity on the second potential scan, suggesting a
rapid depletion of Cu near the surface. As with the 75 at.% Cu
materials, continued cycling finally resulted in a pure Pt
profile. Finally, the Pt75Cu25 compositions (Fig. 4e and f)
seem to have very little surface Cu initially and show a well-
developed H upd peak, even on the first CV scan.
The critical Cu dissolution potential. Fig. 6a plots the anodic
polarization portions of the initial 3 CVs of the Pt25Cu75/
600 1C alloy particles in a log I–E plot in order to highlight the
quasi-constant Tafel Slope of the Cu dissolution onset be-
tween 0.2 and 0.4 V. We attempt to provide some insight into
the observed dissolution behavior of Pt–Cu alloy nanoparticle
Fig. 6 (a) A semi-logarithmic plot of the anodic sweeps of Fig. 4a,
Pt25Cu75/600 1C. A sharp onset of Cu dissolution currents are
discernible at a critical dealloying potential Ec. (b) Diagrammatic
illustration of how the critical dissolution potential of a Cu monolayer
depends on the composition of its subsurface layer. Cu monolayers on
Cu substrates behave bulk-like with dissolution potential E0. Cu
monolayers on Pt–Cu alloy layers exhibit a significantly more positive
dissolution (dealloying) potential Ecritical,2.82,83 Cu atoms on top of a
Pt–Cu alloy exhibit intermediate dissolution potentials Ecritical,1. The
critical potential of pure Pt is very high (also denoted as E0 here).
This journal is �c the Owner Societies 2008 Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 | 3679
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
ensembles, such as that described in Fig. 6a, using simple,
schematic, anodic log I–E polarization schemes (Fig. 6b). Such
schemes were first discussed by Pickering5 and later by
Sieradzki84 in order to illustrate the observed polarization
behavior of flat bimetallic alloy surfaces.
The scheme in Fig. 6b shows the Butler Volmer (BV)
dissolution characteristics of bulk Cu and bulk Pt as the left-
most (solid) and right-most (dotted) diagonal line, respec-
tively, with their characteristic thermodynamic dissolution
onset potential E1 given at the bottom of the graph. Also
shown are cartoons of the surface atomic arrangements asso-
ciated with the two distinct dissolution potentials of pure Pt
and pure Cu. The cartoons are to emphasize the impact of the
nature of the coordinating atoms surrounding a surface atom
to be dissolved; they do not do justice to the effects of
coordination number (discussed in our theoretical treatment
below), which also impact the dissolution behavior. Cu atoms
on Cu atoms (Cu@Cu) result in the Cu bulk dissolution, while
Pt surface atoms on top of Pt atoms (Pt@Pt) lead to the Pt
dissolution at 1.19 V.24
In between the two pure metal log I–E curves, two addi-
tional Cu dissolution characteristics are depicted (dashed and
solid curves) in Fig. 6b for two different Pt–Cu alloy surfaces
with distinct stoichiometry. From dissolution studies on flat
surfaces,4,58–60,84 it is known that the bimetallic alloy stoichio-
metry below the parting limit correlates with a characteristic
critical dissolution potential (Ecitical,1 and Ecritical,2 in Fig. 6b)
above which major dissolution of the less noble metal compo-
nent commences (steep portion of curve). Cathodic of the
critical dissolution potentials no significant Cu dissolution
occurs (flat portion of polarization curve).
A huge body of literature has been dedicated to clarifying
the origin and the controlling factors of the critical dissolution
potential Ecritical and the associated dealloying pro-
cesses.3–6,57–60,84–93 There is even some controversy as to
how to best determine the true critical dissolution potential
of a bimetallic alloy.94 Experimental and theoretical methods
used to characterize and describe the processes around the
critical potential include polarization curves, cyclic voltamme-
try, in situ XRD, electron microscopy, Monte Carlo simula-
tions, as well as empirical free energy models.
Pickering5 argued that the critical potential arises from the
competition between the surface dissolution rate of the less
noble element and its volume diffusion rate to the surface. Past
the critical potential, he argues, pitting corrosion sets in and
inflicts large-scale damage to the alloy surface.
Sieradzki84 reproduced essential features of bimetallic
dealloying curves using kinetic Monte Carlo simulations in
2D and 3D. Based on percolation theoretical arguments, he
stressed the competition between the dissolution kinetics of the
less noble metal and the surface diffusion of the more noble
metal as the controlling factor for the critical dissolution
potential; it represents the balance point between roughening
and smoothing.
Quasi-atomistic approaches4,60 to model the critical dissolu-
tion potential were reported by Erlebacher and coworkers;
these analyses considered the thermodynamic impact of the
number of nearest neighbors on the activation energy and rate
of metal dissolution and diffusion. The kinetic models were
able to capture the evolution of nanoporosity length scales
during the bulk dealloying above the critical potential.
Other modeling approaches have considered the thermo-
dynamics of curved surfaces85 by taking into consideration the
Gibbs–Thomson relation or have identified the interfacial
energy of two metals as a lumped empirical thermodynamic
parameter to model the critical dissolution potential.
We stress that the modeling approaches reviewed above do
not consider alloy-specific energetics of intermetallic bond
making and bond breaking. Thus, while the models can
provide useful insight into the dealloying process and in the
evolution of nanoporosity,4 their general applicability to any
given bimetallic alloy system remains unclear. Indeed, Pugh
et al.7 pointed out in his discussion of dealloying in the Pt–Cu
system that the inclusion of metal- and site-specific binding
energies, that is, the thermodynamics on the atomic scale,
might be necessary in order to accurately capture the experi-
mental voltammetric dealloying behavior; below, we discuss
the first steps in such a metal- and site-specific theoretical
treatment.
Experimental in situ XRD investigations of the dealloying
processes at the atomic scale to the left and right of the critical
dissolution potential was realized by a number of excellent
studies by Renner et al.89–91 In a study on Cu dissolution from
Cu75Au25 single crystal surfaces, the authors were able to show
the formation of a three-monolayer thick Au-enriched passi-
vation layer with inverted stacking at potentials cathodic of
the dissolution potentials. Anodic of these potentials, the
authors found 12 monolayer thick Au-rich islands, which
inherited the inverted stacking of the passivation layer.91
While these studies clarified the dealloying mechanism from
Cu3Au, it is unclear whether these results are equally valid for
other bimetallic systems, such as Pt–Cu. A similar study
performed with Cu83Pd17 remained inconclusive whether or
not a similar Pd rich passivation layer was formed cathodic of
the Cu corrosion potential.89
The impact of metal specific binding energies. Our first-
principles DFT calculations explicitly incorporate metal-spe-
cific binding energies into the treatment of dissolution, thereby
permitting us to establish relationships between the detailed
local chemical environment (the coordination number and the
nature of the coordinating atoms) of the Cu surface atoms and
their thermodynamic dissolution potential. As discussed
above, although the structural model does not explicitly treat
the effects of bulk alloying on dissolution dynamics, it does
provide insight into the effect of undercoordinated surface
features on the dealloying process, and it is able to qualita-
tively reproduce certain features of the cyclic voltammograms
in Fig. 4.
Our analysis provides a direct link between the binding
energy of Cu atoms to the Pt(854) substrate and the dissolu-
tion potential. From eqn (1) above, it may be shown that the
potential at which surfaces with Ncu � 1 and Ncu copper
atoms, respectively, have equal free energies (i.e., the potential
at which at transition between these two states—a dissolution
event—will become thermodynamically favorable) is
U�U� ¼ � 1
2ðENCu
� ENCu�1 � Ebulk;CuÞ ð3Þ
3680 | Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 This journal is �c the Owner Societies 2008
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
where the term in parentheses may be interpreted as the
binding energy of a Cu atom to the Pt substrate and can vary
with both the coverage of Cu and the structure of the
underlying Pt surface; U1 is the experimentally determined
dissolution potential of bulk Cu.24 This relationship immedi-
ately establishes an atomistic interpretation of the thermody-
namic dissolution potential; namely, for Cu atoms that
are strongly bound to the Pt substrate and to the neighboring
Cu atoms (implying that the term in parenthesis is
very negative), the dissolution potential will be shifted to more
positive values.
Eqn (3) can be used to interpret the simple isotherm
presented in Fig. 5g. Cu atoms that dissolve at potentials near
the bulk Cu dissolution potential are bound primarily to other
Cu atoms (Fig. 5a). In contrast, the significant number of Cu
atoms that dissolve at B0.45 V (corresponding to dissolution
from Cu on terrace-like Pt(111) regions of the (854)
surface—Fig. 5b) are bound to more Pt atoms than to Cu
atoms; the higher dissolution potential implies that Cu–Pt
bonds are stronger than Cu–Cu bonds (eqn (3)). We note
that, although this shift is qualitatively consistent with our
experimental CVs (Fig. 4), the quantitative value of the
dissolution potential, 0.45 V, is lower than the high-potential
peak shown in many of the CVs; this modest quantitative
discrepancy may be due to kinetic barriers to Cu atom
dissolution (not accounted for in this analysis) associated with
rearrangement of water atoms or other related processes,
analogous to the modest barriers determined for proton
discharge on Pt(111) by Skullason et al.95 Alternatively, as
suggested by Leiva et al.,96,97 it is possible that small quantities
of oxide or other species contaminate the experimental Cu
surface in this potential range, thereby rendering dissolution
more difficult—although it is important to note that no
voltammetric evidence of such species has been found in this
or previous work.82,83 Finally, the relatively small number of
Cu atoms that are bound primarily to Pt steps and kinks show
the highest dissolution potentials (Fig. 5c) and, by eqn (3), the
strongest binding to the surface.
The theoretical isotherm in Fig. 5g can also be related to the
experimental dissolution curves presented in Fig. 4. Each point
in Fig. 5g represents the thermodynamic dissolution potential
of a chemically distinct type of Cu site on the Pt(854) surface.
If, as a first approximation, we neglect interactions between
the various sites, and if we consider only the thermodynamic
limit of dissolution, then the Cu atoms at each type of site
should follow Langmuir dissolution isotherms. Each site
would then contribute an approximately Gaussian-shaped
peak to the dissolution current, with the peaks centered at
the equilibrium dissolution potential of the corresponding site.
The net superposition of such peaks is shown in Fig. 7.
Although this theoretical curve is clearly a simple approxima-
tion to the measured voltammograms, and although, as dis-
cussed above, many complex structural and electronic features
are likely responsible for these curves and for the critical
potential, the curve nonetheless provides insight into some
of the voltammetric features, including the multilayer peak at
0.34 V and the shifted underpotential deposition peak at
higher potentials. This simple analysis therefore suggests that
some features of the curves may be related to simple, atomic-
scale changes in the binding energy of Cu atoms to the Pt/alloy
surfaces as a function of Cu atom coverage.
The simple thermodynamic results presented in Fig. 5 and 7
can also provide qualitative insight into certain features of the
Tafel-style dissolution plots in Fig. 6. The second and third
curves of that figure represent dissolution of Cu atoms in
different chemical environments, analogous to the different
sites shown in Fig. 5. In Fig. 7, we have treated the current
produced by dissolution from such sites by assuming a finite
number of Cu atoms at each site; as these atoms are depleted,
the current decreases, leading to the Gaussian-type peak
shapes. However, if the surface Cu atoms are continuously
replaced by Cu atoms from the bulk, as suggested by Pickering
in his analysis,5 there will be no decrease, and continuous
current will be produced from each type of site after the
thermodynamic potential is reached, as in Fig. 6.
6. Conclusions
(1) We have presented a strategy to modify the surface
catalytic properties of Pt bimetallics. The strategy is in prin-
ciple based on the ancient method of ‘surface depletion gilding’
for changing surface compositions of bimetallic compounds,
which dates back over 2000 years to pre-Columbian gold-
smiths in Mesoamerica. Our modern flavor of the ancient
method uses electrochemical voltammetric sweeps to selec-
tively dissolve less noble alloy components (dealloying), in our
case Cu, while the more noble components, herein Pt, remain
in the alloy.
(2) We have established detailed qualitative relationships
between the synthesis conditions of the Pt–Cu alloy nanopar-
ticle precursors, their structural characteristics and their vol-
tammetric profiles during the early stages of Cu dissolution.
(3) We have shown that dealloyed Pt–Cu alloy nanoparticle
eletrocatalysts prepared by partial Cu dissolution from Cu rich
Pt25Cu75 precursor alloys exhibited a 4–5 fold ORR activity
enhancement on a Pt-mass basis compared to state-of-the-art
Fig. 7 Theoretical plot of Cu dissolution current versus potential.
Langmuir isotherms are assumed for each distinct Cu site, and ion
concentrations are taken at their standard state values. The potential
scale is referenced to the experimentally-tabulated dissolution poten-
tial of bulk Cu, and the current is normalized by the highest calculated
value.
This journal is �c the Owner Societies 2008 Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 | 3681
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
pure Pt nanoparticle catalysts. The most active catalyst for the
ORR on a ECSA-normalized metric (specific ORR activity)
was achieved after surface Cu dissolution from a uniformly
alloyed (950 1C annealing) Pt25Cu75 nanoparticle precursor
alloy in accordance with recent MEA studies.17
(4) DFT computations of the energetics of Cu adatom
dissolution from a (854) model surface yielded a metal- and
site-specific correlation between the local atomic environment
of Cu surface atoms and their detailed dissolution potential.
Cu surface atoms with multiple Cu coordinations dissolved
near the thermodynamic bulk Cu potential, whereas under-
coordinated Cu surface atoms or Cu surface atoms with
primarily Pt neighbors showed higher dissolution potentials.
(4) The combination of experimental data and DFT com-
putational results has corroborated our hypothesis that, in
addition to the previously suggested balance between metal
dissolution and surface diffusion kinetics, metal-specific inter-
actions on the atomic scale are important controlling factors in
the understanding of the onset potentials of Cu dissolution for
Pt–Cu bimetallics (critical dissolution potentials). First prin-
ciples computation of binding energies has revealed that the
local chemical environment of surface Cu atoms is a control-
ling factor for the dissolution onset potentials.
Acknowledgements
This project was supported by the Department of Energy,
Office of Basic Energy Sciences (BES), under grant LAB04-20
via a subcontract with the X-ray Laboratory for Advanced
Materials (XLAM) and the Stanford Synchrotron Radiation
Laboratory (SSRL). Further support was provided by the
National Science Foundation (NSF) under the award
#0729722. Acknowledgment is made to the Donors of the
American Chemical Society Petroleum Research Fund for
partial support of this research (grant #44165). Support by
the State of Texas through the Advanced Research Program
(ARP) during the 2008–2010 funding period is gratefully
acknowledged. Also, partial financial support from Houston
Area Research Center (HARC) is gratefully acknowledged.
Use of the Center for Nanoscale Materials was supported by
the US Department of Energy, Office of Science, Office of
Basic Energy Sciences, under contract No. DE-AC02-
06CH11357. We acknowledge computer time at the Labora-
tory Computing Resource Center (LCRC) at Argonne
National Laboratory, The National Energy Research Scien-
tific Computing Center (NERSC), and the Molecular Science
Computing Facility (MCSF) at Pacific Northwest National
Laboratory.
References
1. S. Koh and P. Strasser, J. Am. Chem. Soc., 2007, 129,12624–12625.
2. H. Lechtman, Sci. Am., 1984, 250, 56–63.3. A. J. Forty, Nature, 1979, 282, 597–598.4. J. Erlebacher, M. J. Aziz, A. Karma, N. Dimitrov and
K. Sieradzki, Nature, 2001, 410, 450–453.5. H. W. Pickering, Corrosion Sci., 1983, 23, 1107–1120.6. H. W. Pickering and C. Wagner, J. Electrochem. Soc., 1967, 114,
698.
7. D. V. Pugh, A. Dursun and S. G. Corcoran, J. Electrochem. Soc.,2005, 152, B455–B459.
8. M. B. Vukmiovic, J. Zhang, K. Sasaki, A. U. Nilekar, F. Uribe,M. Mavrikakis and R. R. Adzic, Electrochim. Acta, 2007, 52,2257–2263.
9. R. R. Adzic, J. Zhang, K. Sasaki, M. B. Vukmirovic, M. Shao,J. X. Wang, A. U. Nilekar, M. Mavrikakis, J. A. Valerio andF. Uribe, Top. Catal., 2007, 46, 249–262.
10. J. Zhang, M. B. Vukmiovic, Y. Xu, M. Mavrikakis andR. R. Adzic, Angew. Chem., Int. Ed., 2005, 44, 2132.
11. J. Zhang, Y. Mo, M. Vukmirovic, R. Klie, K. Sasaki andR. Adzic, J. Phys. Chem. B, 2004, 108, 10955–10964.
12. S. R. Brankovic, J. X. Wang and R. R. Adzic, Electrochem. Solid-State Lett., 2001, 4, A217–A220.
13. L. Bengtsson, Phy. Rev. B, 1999, 59, 12301–12304.14. R. Srivastava, P. Mani, N. Hahn and P. Strasser, Angew. Chem.,
Int. Ed., 2007, 46, 8988–8991.15. P. Mani, R. Srivastava and P. Strasser, ECS Trans., 2007, 11,
933–940.16. N. Hahn, S. Koh and P. Strasser, ECS Trans., 2006, 3, 139–140.17. P. Mani, S. Srivastava and P. Strasser, J. Phys. Chem. C, 2008,
112, 2770–2778.18. T. J. Schmidt and H. A. Gasteiger, Rotating thin-film method for
supported catalysts, in Handbook of Fuel Cells—Fundamentals,Technology, and Applications, ed. W Vielstich, A Lamm andH Gasteiger, Wiley, Chichester, 2003, ch. 22, vol. 2, p. 316.
19. T. J. Schmidt, H. A. Gasteiger and R. J. Behm, J. Electrochem.Soc., 1999, 146, 1296–1304.
20. T. J. Schmidt, H. A. Gasteiger, G. D. Stab, P. M. Urban,D. M. Kolb and R. J. Behm, J. Electrochem. Soc., 1998, 145,2354–2358.
21. J. Friedel, Philos. Mag., 1955, 46, 514.22. L. Vegard, Z. Phys., 1921, 5, 2–26.23. B. Hammer, L. B. Hansen and J. K. Nørskov, Phys. Rev. B:
Condens. Matter Mater. Phys., 1999, 59, 7413–7421.24. CRC Handbook of Chemistry and Physics, CRC Press, New York,
76th edn, 1996.25. D. Vanderbilt, Phys. Rev. B: Condens. Matter Mater. Phys., 1990,
41, 7892–7895.26. G. Kresse and J. Furthmuller, Comput. Mater. Sci., 1996, 6,
15–50.27. J. Greeley and J. K. Nørskov, Electrochim. Acta, 2007, 52,
5829–5836.28. J. K. Norskov, J. Rossmeisl, A. Logadottir, L. Lindqvist,
J. R. Kitchin, T. Bligaard and H. Jonsson, J. Phys. Chem. B,2004, 108, 17886–17892.
29. M. Pourbaix, Atlas of Electrochemical Equilibria in AqueousSolutions, Pergamon Press, Oxford, New York, 1966.
30. H. A. Gasteiger, S. S. Kocha, B. Sompalli and F. T. Wagner, Appl.Catal., B, 2005, 56, 9–35.
31. U. A. Paulus, A. Wokaun, G. G. Scherer, T. J. Schmidt,V. Stamenkovic, N. M. Markovic and P. N. Ross, Electrochim.Acta, 2002, 47, 3787–3798.
32. U. A. Paulus, A. Wokaun, G. G. Scherer, T. J. Schmidt,V. Stamenkovic, V. Radmilovic, N. M. Markovic andP. N. Ross, J. Phys. Chem. B, 2002, 106, 4181–4191.
33. V. Stamenkovic, T. J. Schmidt, P. N. Ross and N. M. Markovic,J. Phys. Chem. B, 2002, 106, 11970–11979.
34. V. Stamenkovic, T. J. Schmidt, P. N. Ross and N. M. Markovic,J. Electroanal. Chem., 2003, 554, 191–199.
35. S. Koh, M. F. Toney and P. Strasser, Electrochim. Acta, 2007, 52,2765–2774.
36. P. Strasser, S. Koh and C. Yu, ECS Trans., 2007, 11, 167–180.37. S. Mukerjee, S. Srinivasan, M. P. Soriaga and J. McBreen,
J. Phys. Chem., 1995, 99, 4577.38. S. Mukerjee, S. Srinivasan, M. P. Soriaga and J. McBreen,
J. Electrochem. Soc., 1995, 142, 1409–1422.39. S. Mukerjee and S. Srinivasan, J. Electroanal. Chem., 1993, 357,
201.40. N. M. Markovic and P. N. Ross, Surf. Sci. Rep., 2002, 45, 117.41. E. Antolini, J. e. R. C. Salgado and E. R. Gonzalez, J. Power
Sources, 2006, 155, 161–166.42. F. H. B. Lima, W. H. E. Lizcano-Valbuena, E. Teixeira-Neto,
F. C. Nart, E. R. Gonzalez and E. A. Ticianelli, Electrochim. Acta,2006, 52, 385–393.
3682 | Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 This journal is �c the Owner Societies 2008
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online
43. N. Travitsky, T. Ripenbein, D. Golodnitsky, Y. Rosenberg,L. Burshtein and E. Peled, J. Power Sources, 2006, 161, 782.
44. J. N. Soderberg, A. H. C. Sirk, S. A. Camphell and V. I. Birss,J. Electrochem. Soc., 2005, 152, A2017–A2022.
45. L. Xiong and A. Manthiram, J. Electrochem. Soc., 2005, 152,A697–A703.
46. D. Thompsett, Pt alloys as oxygen reduction catalysts, in Hand-book of Fuel Cells—Fundamentals, Technology and Applications,ed. W Vielstich, A Lamm and H. A. Gasteiger, Wiley, Chichester,2003, ch. 37, vol. 3, p. 467.
47. E. Antolini, R. R. Passos and E. A. Ticianelli, Electrochim. Acta,2002, 48, 263–270.
48. M.-k. Min, J. Cho, K. Cho and H. Kim, Electrochim. Acta, 2000,45, 4211–4217.
49. G. Tamizhmani and G. A. Capuano, J. Electrochem. Soc., 1994,41, 968.
50. C. Yu, S. Koh, J. Leisch, M. T. Toney and P. Strasser, FaradayDicuss., 2008, 140, in print.
51. C. W. B. Bezerra, L. Zhang, H. Liu, K. Lee, A. e. L. B. Marques,E. P. Marques, H. Wang and J. Zhang, J. Power Sources, 2007,173, 891–908.
52. A. Cebollada, R. F. C. Farrow and M. F. Toney, Structure andmagnetic properties of chemically ordered magnetic binary alloysin thin film form, in Magnetic Nanostructures, ed. H. S. Nalwa,American Scientific, Stevenson Ranch, 2002, pp. 93–123.
53. ICDD, Power Diffraction File (PDF) 2—data CD. In Interna-tional Center for Diffraction Data—Power Diffraction File (PDF)2 data CD http://www.icdd.com/: 2002.
54. Z. Peng, C. Chang and L. Ximen, Acta Geol. Sinica, 1978, 4,326–336.
55. Z. Yu, Bull. Inst. Geol., Chin. Acad. Sci., 1982, 4, 75–81.56. K. Ding, Scientia Geol. Sinica, 1980, 2, 168–171.57. Y. Ding, M. Chen and J. Erlebacher, J. Am. Chem. Soc., 2004,
126, 6876–6877.58. K. Sieradzki, N. Dimitrov, D. Movrin, C. McChall, N.
Vasijevic and J. Erlebacher, J. Electrochem. Soc., 2002, 149,B370–B377.
59. J. Erlebacher, J. Electrochem. Soc., 2004, 151, C614–C626.60. J. Rugolo, J. Erlebacher and K. Sieradzki, Nature, 2006, 5,
946–949.61. G. Ertl, H. Knozinger and J. Weitkamp, Handbook of Hetero-
geneous Catalysis, Wiley-VCH, Weinheim, 1997, vol. 1, p. 426.62. Z. Liu, S. Koh, C. Yu and P. Strasser, J. Electrochem. Soc., 2007,
154, B1192–B1199.63. Z. Liu, C. Yu, I. Rusakova, D. Huang and P. Strasser, Top.
Catal., 2007, DOI: 10.1007/s11244-008-9083-2, in press.64. J. Xie, D. L. Wood, K. L. More, P. Atanassov and R. L. Borup,
J. Electrochem. Soc., 2005, 152, A1011–A1020.65. Y. Shao-Horn, W. C. Sheng, S. Chen, P. J. Ferreira, E. F. Holby
and D. Morgan, Top. Catal., 2007, 46, 285–305.66. P. J. Ferreira, G. J. laO’, Y. Shao-Horn, D. Morgan,
R. Makharia, S. Kocha and H. Gasteiger, J. Electrochem. Soc.,2005, 152, A2256–2271.
67. D. L. Fedlheim and C. A. Foss, Metal Nanoparticles: SynthesisCharacterization and Applications, CRC – Taylor Francis Group,London, 2001.
68. R. G. Finke, Transition-Metal Nanoclusters: Solution-PhaseSynthesis, Characterization and Mechanism of Formation, inMetal Nanoparticles: Synthesis, Characterization, and Applica-
tions, ed. D. L. Feldheim and C. A. Foss, Jr, Marcel Dekker,Inc., New York, 2002, pp. 17–54.
69. G. Schmid, Nanoparticles: From Theory to Applications, Wiley-VCH, New York, 2004.
70. S. Koh, C. Yu and P. Strasser, ECS Trans., 2007, 11, 205–215.71. S. Koh, M. F. Toney and P. Strasser, Lattice-strained Pt shell
Nanoparticle Catalysts for the Electroreduction of Oxygen atPEMFC cathodes, in American Institute of Chemical Engineers(AIChE) annual meeting abstracts, San Francisco, 2006.
72. S. Mukerjee, S. Srinivasan, M. P. Soriaga and J. McBreen,J. Phys. Chem., 1995, 99, 4577–4589.
73. D. Thompsett, Catalysts for the Proton Exchange Membrane FuelCell, in Fuel Cell Technology Handbook, CRC PressLLC, London,New York, 2003, ch. 6.
74. V. Stamenkovic, B. S. Mun, M. Arenz, K. J. J. Mayerhofer,C. A. Lucas, G. Wang, P. N. Ross and N. Markovic, Nat. Mater.,2007, 6, 241.
75. V. R. Stamenkovic, B. Fowler, B. S. Mun, G. Wang, P. N. Ross,C. A. Lucas and N. M. Markovic, Science, 2007, 315, 493.
76. M. T. Paffet, J. G. Beery and S. Gottesfeld, J. Electrochem. Soc.,1988, 135, 1431–1436.
77. H. Liu, P. He, Z. Li and J. Li, Nanotechnology, 2006, 17.78. V. R. Stamenkovic, B. S. Mun, K. J. J. Mayrhofer, P. N. Ross and
N. M. Markovic, J. Am. Chem. Soc., 2006, 128, 8702–8988.79. A. V. Ruban, H. L. Skriver and J. K. Nørskov, Phys. Rev. B:
Condens. Matter Mater. Phys., 1999, 59, 15990–16000.80. D. V. Pugh, A. Dursun and S. G. Corcoran, J. Mater. Res., 2003,
18, 216.81. V. Stamenkovic and N. Markovic, Langmuir, 2001, 17,
2388–2394.82. N. Markovic, H. A. Gasteiger and P. N. Ross, Langmuir, 1995, 11,
4098–4108.83. N. Markovic and P. N. Ross, Langmuir, 1993, 9, 580–590.84. K. Sieradzki, R. R. Corderman, K. Shukla and R. C. Newman,
Philos. Mag. A, 1989, 59, 713–746.85. K. Sieradzki, J. Electrochem. Soc., 1993, 140, 2868–2872.86. R. Li and K. Sieradzki, Phys. Rev. Lett., 1992, 68, 1168–1172.87. K. Wagner, S. R. Brankovic, N. Dimitrov and K. Sieradzki,
J. Electrochem. Soc., 1997, 144, 3545–3555.88. R. C. Newman and K. Sieradzki, Science, 1994, 263, 1708–1709.89. F. U. Renner, Y. Grunder, P. F. Lyman and J. Zegenhagen, Thin
Solid Films, 2007, 515, 5574.90. F. U. Renner, A. Stierle, H. Dosch, D. M. Kolb and
J. Zegenhagen, Electrochem. Commun., 2007, 9, 1639.91. F. U. Renner, A. Stierle, H. Dosch, D. M. Kolb, T. L. Lee and
J. Zegenhagen, Nature, 2006, 430, 707–710.92. S. G. Corcoran, The Morphology of Alloy Corrosion, in Critical
Factors in Localized Corrosion III, eds. R. G. Kelley,G. S. Frankel, P. M. Natishan and R. C. Newman, Electroche-mical Society, Pennington, NJ, 2000, vol. 98-17, pp. 500–507.
93. J. R. Davis, Corrosion—Understanding the basics, ASM Interna-tional, Materials Park, Ohio, 2000.
94. A. Dursun, D. V. Pugh and A. G. Corcoran, J. Electrochem. Soc.,2005, 152, B65–B72.
95. E. Skulason, G. S. Karlberg, J. Rossmeisl, T. Bligaard, J. Greeley,H. Jonsson and J. K. Nørskov, Phys. Chem. Chem. Phys., 2007, 9,3241–3250.
96. E. Leiva, Electrochim. Acta, 1996, 41, 2185–2206.97. C. G. Sanchez and E. P. M. Leiva, Langmuir, 2001, 17, 2219–2227.
This journal is �c the Owner Societies 2008 Phys. Chem. Chem. Phys., 2008, 10, 3670–3683 | 3683
Publ
ishe
d on
27
May
200
8. D
ownl
oade
d by
TU
Ber
lin -
Uni
vers
itaet
sbib
l on
13/0
4/20
15 1
3:24
:20.
View Article Online