PCCP

View Article Online

PERSPECTIVE View Journal

Bimetallic alloys in action: dynamic atomistic

Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

motifs for electrochemistry and catalysis

Cite this: DOI: 10.1039/c4cp01591f
Jonathan E. Mueller,a Petr Krtil,b Ludwig A. Kiblera and Timo Jacob*a

Bimetallic alloys show great promise for applications in a wide range of technologies related to electrochemistry
and heterogeneous catalysis. The alloyed nature of these materials supports the existence of surface
phenomena and structural motifs not present in single-component materials. These novel features result
in electrochemical and catalytic behaviors, requiring entirely new categories of explanations. In this
perspective concrete examples are used to illustrate several of these chemical and structural features,
Received 12th April 2014, which are unique to multi-component metal surfaces. The influence of the surface’s structure and
Accepted 12th June 2014 surroundings (e.g. adsorbates) on each other provides a common thread, with the emergence of dynamic
DOI: 10.1039/c4cp01591f surfaces as its terminus. In considering three model systems (PtRu, PtNi and AuPd), we discuss not only
a selection of surface phenomena relevant to each, but also the implications of these alloy-related
www.rsc.org/pccp behaviors for the electrochemical and catalytic properties of each surface.

1. Introduction depends not only on the nature of the surface, but also on the
nature of the adsorbate. Nevertheless, if the varying reactivities
Metal surfaces are indispensable in the overlapping fields of different adsorbates are factored out, then one arrives at
of electrochemistry and heterogeneous catalysis. Both fields a general reactivity for a surface, which is not just related to
contain numerous examples of systems in which the atomic any one particular adsorbate, but characterizes the surface’s
surface structure of a metal critically determines its behavior as contribution to the adsorption of any adsorbate. This surface
an electrode or catalyst. Thus, the possible presence of surface reactivity is often expressed in terms of quantifiable electronic
defects is often an essential complication in satisfactorily properties, such as the d-band center.
accounting for a surface’s electrochemical and catalytic properties. Catalytic activity can only be specified in terms of a parti-
The introduction of a second metallic element to form a bimetallic cular reaction, and characterizes the rate at which that reaction
system greatly increases the variety of surface structures possible, is catalyzed on the surface. Unlike its reactivity, a surface’s
with important ramifications for the study and application of the catalytic activity cannot be naively abstracted away from particular
surfaces of multi-component systems. One reason for this is that catalytic reactions. Nevertheless, surface catalysis and reactivity are
the greater variation in electrochemical and catalytic properties, related, at least insofar as surface catalysis requires the adsorption
corresponding to the larger number of possible structures, allows of reactants (favored by high surface reactivity) and desorption
for greater optimization in the design of functional materials. of products (favored by low surface reactivity). By assuming that
However, an even more important reason is that structural one of these two processes (i.e. reactant adsorption or product
motifs and surface phenomena, stemming from the chemical desorption) is rate-controlling, one is able to conclude that a
heterogeneity of the alloy, can lead to novel electrochemical surface with a reactivity that balances the rates of reactant
and catalytic behaviors, with drastically improved performance adsorption and product desorption will display optimal (i.e. the
for a given application. highest attainable) catalytic activity. Indeed, similar lines of
In the context of heterogeneous catalysis and electrochemistry reasoning are commonly used to explain experimentally obtained
it is essential to distinguish between the reactivity of a surface and volcano plots.1–7 Thus, on one hand, it is essential to distinguish
its catalytic activity. The reactivity of a surface directly reflects between the reactivity and catalytic activity of a metal surface.
the strength of adsorption on it. Adsorption strength, however, However, on the other hand, the relationship between the two
can be taken advantage of in efforts to simply and efficiently
predict catalytic behavior based on adsorption properties,
a
Institut für Elektrochemie, Universität Ulm, Albert-Einstein-Allee 47, 89081 Ulm,
albeit with caution.
Germany. E-mail: [email protected]; Fax: +49 7315 025409;
Tel: +49 7315 025401
In light of the chemical heterogeneity of bimetallic surfaces, at
b
J. Heyrovský Institute of Physical Chemistry, Academy of Sciences of the Czech least three different classes of phenomena have been distinguished
Republic, Dolejškova 3, Prague 18223, Czech Republic in efforts to explain their unique characteristics as electrodes

This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys.
 View Article Online

Perspective PCCP

and catalysts, i.e. their reactivity and catalytic activity.8 These of this paper is to highlight those unique structural features of
categories correspond to explanations which are derived from bimetallic surfaces, which set them apart from single metal
either (i) bifunctional mechanisms, (ii) delocalized electronic surfaces, and then to consider the influence of these structural
properties of the catalyst, or (iii) the local atomic configuration features on the electrochemical and catalytic properties of bimetallic
of the catalyst surface. surfaces. To facilitate this, three examples of bimetallic model
In a bifunctional mechanism each component of the bimetallic electrodes/catalysts have been chosen as typical instances of pheno-
alloy is responsible for either catalyzing a different reaction step mena pertinent to understanding the functionality of bimetallic
or adsorbing a different intermediate. This synergy between the surfaces under electrochemical conditions: PtRu, NiPt and AuPd.
two components results in increased catalytic activity. A classic Indeed, the study of model systems is an indispensable strategy
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

example of bifunctional mechanisms at work is the electrooxida- in both heterogeneous catalysis and electrochemistry, because it
tion of carbon monoxide, methanol and other small organics on enables one to consider relevant parameters in isolation, before
PtRu alloys.9–15 attempting to put together a picture of the whole. In considering
The delocalized electronic band structure of a bimetallic these three model systems, we begin by looking at the role played
alloy varies from that of either pure metal and will have its own by average, delocalized electronic properties in PtRu alloys before
unique adsorption and catalytic properties. Thus, some aspect turning to the role of local structures in NiPt and AuPd alloys.
of the band structure of the surface as a whole is appealed to in The influence of the catalytic environment on the catalyst’s
order to explain the surface’s reactivity towards adsorbates and surface structure and composition, leading to a dynamic rather
catalytic activity. A prime example of this approach is the model than static catalyst surface, is given special attention here, as a
developed by Hammer and Nørskov for adsorption on transition vital, but often overlooked, aspect of multi-component transi-
metal overlayer structures, from which they conclude that there tion metal catalysts.
is a direct relationship between the chemisorption energies of a
given adsorbate on a transition metal heteroepitaxial monolayer
on various transition metal substrates and the d-band center of 2. ORR on PtRu: disentangling strain
the substrate metal.16,17 This approach can be further extended and ligand effects
to explain the catalytic activities of transition metal surfaces.
2.1 Introduction to PtRu alloys
Thus, plotting the experimental exchange currents as a function
of the d-band center results in a so called volcano plot, whose The excellent catalytic properties of bimetallic PtRu alloys in
apex corresponds to an optimal value for the d-band center, the electro-oxidation of small organic molecules have long been
which can then be aimed at in the development of improved recognized.10,13,28,29 In particular, the low sensitivity of PtRu alloys
catalysts.18 However, caution is required, as additional pheno- to poisoning combined with the highly reduced overpotentials
mena (e.g. surface reconstruction, restructuring, surface segregation) required for the oxidation of key intermediates such as CO30–35
can result in surface properties significantly different than have led to extensive studies of their structure and catalytic proper-
those predicted by a volcano plot. ties.36–44 A number of different effects have been appealed to in
Lastly, active sites corresponding to particular local atomic explaining the catalytic activity of PtRu alloys in various reactions,
structures are sometimes implicated in the increased reactivity including: (i) bifunctional mechanisms, as in the case of oxidation
and catalytic activity often realized in bimetallic alloy catalysts. reaction involving CO,10 (ii) the ligand effect, resulting from the
Here, instead of delocalized perturbations in the electronic influence of neighboring atoms on the electronic states of surface
structure of the catalyst being responsible for increased catalytic atoms,34,45 (iii) the strain effect resulting from altered nearest-
activity, local arrangements of atoms (corresponding to local neighbor distances in a pseudomorphically grown surface layer,46,47
perturbations in the electronic structure) provide the key adsorp- and (iv) the ensemble effect in which particular structural
tion or critical catalytic active sites.8,19,20 Metal surfaces have motifs of surface atoms serve as active sites.15,48
often been considered static under reaction or catalytic condi-
tions; however, there is a growing body of evidence pointing 2.2 Role of substrate vs. role of overlayer
toward dynamic surface structures for a number of metals under The different reactivities of Ru and Pt can be taken advantage of to
typical reaction conditions.21,22 Such changes in a metal’s surface form segregated, pseudomorphic surface layers of either pure Pt or
structure naturally result in changes in its reactivity and catalytic pure Ru on a PtRu low-index surface. Thus surface heating
activity, especially when particular local active sites are key players, followed by cooling under an inert or reductive atmosphere results
since their presence or absence can result in drastic alterations in in the segregation of the less oxophilic Pt to the surface.49–51
the overall chemical behavior of the surface. It is thus essential to Similarly, surface heating followed by cooling under an oxidative
take the actual structure of the surface under reaction or catalytic (i.e. oxygen-containing) atmosphere draws Ru to the surface.
conditions into account, in explaining its behavior. Indeed, cyclic voltammetric measurements lead to the conclu-
A number of excellent review articles already provide com- sion that the surface segregation in both cases is complete so
prehensive overviews of the literature on bimetallic electrodes that the surface layer is pure Pt in the former case and pure Ru
and catalyst surfaces, including their synthesis, characteriza- in the latter.51 In the case of the Pt-enriched surface, the base
tion, atomic and electronic structures, physical, chemical and voltammogram in 0.1 M H2SO4 (see Fig. 1a) strongly resembles
catalytic properties, and applications.23–27 In contrast, the aim the characteristic base voltammogram for Pt(111) in the same

Phys. Chem. Chem. Phys. This journal is © the Owner Societies 2014
 View Article Online

PCCP Perspective

less reactive Pt in the substrate increases the reactivity of a

Ru(0001)/Ru(111) surface (i.e. increases the strength of adsop-
tion to it). Thus placing an overlayer on a modified substrate is
a useful strategy for modifying its catalytic properties in a
controlled manner.52

2.3 Disentangling the strain and ligand effects

2.3.1 Strategies for disentanglement. Of the effects which
may contribute to the unique catalytic properties of bimetallic
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

catalysts only the so called ligand and strain effects can be

applied to epitaxial catalysts in which the surface is composed
of atoms of only one component. In the first case the electronic
influence of the substrate alters the adsorption and catalytic
properties of the surface, while in the second case the alteration
in the chemical properties of the surface results from its taking
on the lattice constant of the substrate. In practice these two
effects are typically encountered simultaneously, and are thus
difficult to distinguish. For example, a Pt(111) epitaxial layer on
a PtRu alloy experiences both strain and ligand effects due to
the PtRu substrate. Furthermore, any attempt to vary one of
these effects by varying the composition of the alloy will cause
a simultaneous (and likely proportional) effect in the other
effect, since both effects are expected to be proportional to
the fraction of Ru in the substrate. Nevertheless clever experi-
mental or theoretical approaches are available for differentiat-
ing these effects.
One of the first approaches to disentangling the strain and
Fig. 1 Current–potential curves for: (a) a Pt-enriched surface of PtRu(111)
ligand effects made use of the varying length scales over which
(blue curve), and Pt(111) (black curve), (b) a Ru-enriched surface of
PtRu(111) (red curve), and Ru(0001) (black curve). All scans were performed they are expected to exert influence.53 The ligand effect is
in 0.1 M H2SO4 at a scan rate of 50 mV s1. These measurements were expected to die out very quickly once the epitaxial layer is more
carried out with the same electrode, and so one should note that the than a monolayer thick. In contrast, in this particular case the
changes seen here as a function of sample preparation might also occur strain on epitaxial layers due to lattice mismatch does not
in situ or in operando for real systems. Adapted from ref. 51.
relax until the epitaxial layer is several monolayers thick. Thus,
while an epitaxial monolayer exhibits both effects, an epitaxial
layer composed of several monolayers exhibits primarily the
electrolytic solution, with the cation (e.g. H+) adsorption peaks strain effect. Based on this assumption Schlapka et al. have
shifted to more negative and anion (e.g. HSO4) peaks shifted employed a combination of experiment and theory to disen-
to more positive potentials, due to decreasing free energies of tangle strain and the ligand effect in the adsorption of CO on
adsorption. In the case of the Ru-enriched surface, the base Pt(111) epitaxial layers on Ru(0001) under ultra high vacuum
voltammogram in 0.1 M H2SO4 (Fig. 1b) shows the same (UHV) conditions. The application to electrochemical systems
characteristic peaks as the base voltammogram for Ru(0001) was made shortly thereafter by Zhang et al., who considered the
(the equivalent surface for Ru, an hcp metal) in the same ORR on Pt overlayers on transition metal substrates with
electrolytic solution, with the surface oxidation and reduction varying strains and ligand effects.47 More detailed insight was
and hydrogen evolution peaks shifted by 200 mV to more obtained by Hoster et al.54 by considering the electrochemical
negative potentials. adsorption of OHad and Had on Pt(111) overlayers of varying
These two sets of voltammograms illustrate well the respec- thicknesses on a Ru(0001) substrate. To probe the influence of
tive roles of the substrate and the surface in epitaxial catalysts. various degrees of coupled strain and ligand effects on the
On one hand, the chemical nature of the surface determines electrochemical properties of a Pt(111) monolayer, cyclic voltam-
the overall form of the cyclic voltammogram, i.e. the overall metric measurements have been carried out on Pt hetero-
electrodic properties of the surface. On the other hand, the epitaxial overlayers on PtRu substrates of various compositions.
substrate alters the shape and potential placement of the As can be seen in Fig. 2, the extent of the deviation of the
peaks, i.e. influences the exact electrode potential at which CV from that of pure Pt is directly related to the composition of
individual reactions and adsorption events take place. In this the substrate, with more Ru resulting in greater deviation.
particular example, the presence of more reactive Ru in the Because there is an approximately linear relationship between
substrate decreases the reactivity of the Pt(111) surface (i.e. the composition and the lattice parameter of the substrate,
decreases the strength of adsoption to it), while the presence of both the ligand effect and the strain effect are present in this

This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys.
 View Article Online

Perspective PCCP
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

Fig. 2 Current–potential curves for Pt-rich surfaces of PtRu(111) electro-

des of various alloy composition. All scans were performed in 0.1 M H2SO4
at a scan rate of 50 mV s1.55

 pffiffiffi
Fig. 3 Energies of formation (eV) per p 2  3 unit cell for 1/3 ML
of oxygen on a Pt(111) overlayer on PtRu substrate slabs with
system in similar proportions, making it very difficult, if not
various PtRu compositions (corresponding to the ligand effect; these are
impossible, to disentangle them experimentally. reported as % Pt, and correspond to compositions for the 5 substrate
2.3.2 Surface adsorption. To gain more detailed insight layers—i.e. excluding the Pt6 overlayer—of Ru30, Pt10Ru20, Pt20Ru10 & Pt30)
into the respective roles of the strain and ligand effects on the and unit cell parameters (corresponding to the strain effect; these
catalytic characteristics of Pt(111) epitaxial layers on PtRu alloys, are reported as the corresponding fcc cubic unit cell dimension in
Angstroms). Reference states are H2 gas, H2O gas and the corresponding
we performed a systematic DFT study of the adsorption energies
bare PtRu–Pt(111) slab. The same 28 energies are plotted twice: (a)
of key intermediates and the reaction energies of critical steps in as a function of the unit cell parameter and (b) as a function of substrate
the oxygen reduction reaction (ORR) on Pt(111) as functions of the composition. Red atoms are O, purple atoms are Ru and silver atoms
lattice constant (i.e. the strain effect) and the substrate composi- are Pt.
tion (i.e. the ligand effect), which were varied independently.
Along similar lines, related approaches have already been applied
to Pd(111) overlayers on transition metal substrates.56,57 Fig. 4a shows the adsorption energies for Had on a Pt(111)
Fig. 3 shows the adsorption energies for Oad on a Pt(111) epitaxial layer on PtRu substrates of various compositions
epitaxial layer on PtRu substrates of various compositions plotted plotted as a function of the unit cell parameter, while Fig. 4b
as a function of the unit cell parameter in Fig. 3a, then again as a shows the same set of adsorption energies plotted as a function
function of the substrate composition in Fig. 3b. The dependence of the substrate composition. The dependence of the adsorp-
of the adsorption energy on the unit cell parameter is strongly tion energy on the unit cell parameter is strongly linear for the
linear for the range of values checked (3.80–3.95 Å), which range of values checked (3.80–3.95 Å), as is the dependence on
correspond to the region between and immediately outside of the substrate composition for the compositions checked. In
the equilibrium unit cell parameters for pure Ru (B3.825 Å) and contrast to the case of O adsorption, H adsorption shows very
pure Pt (B3.925 Å) substrates. The slope shows a pronounced little dependence on the unit cell parameter. Thus almost the
dependence on the composition of the substrate, as it nearly entire 0.67 eV weakening of H adsorption upon replacement of
doubles from 2.51 eV Å1 to 4.93 eV Å1 as a pure Ru substrate is a pure Pt substrate at equilibrium with a pure Ru substrate at
replaced with pure Pt. The dependence on the substrate compo- equilibrium can be attributed to the ligand effect.
sition is not linear, but is more pronounced for higher fractions Analogous results for OHad reveal a case in which the two
of Pt, regardless of unit cell parameter. Because of this, the effects are equal: compression (from 3.925 to 3.825 Å) and
fractions change in adsorption energy attributable to the strain substrate replacement (from pure Pt to pure Ru) each decrease
and ligand effects for a change in substrate will vary depending the binding energy by 0.11 eV, regardless of the order in which
on which path is taken between them. Thus exchanging a pure Pt they are carried out, for a net decrease of 0.22 eV. Thus, we see
substrate at equilibrium (unit cell parameter = 3.925 Å) for a pure the relative contributions of the strain and ligands effect
Ru substrate at equilibrium (unit cell parameter = 3.825 Å) results ranging from contributing equally in the case of OH adsorp-
in O adsorption which is 1.02 eV weaker, 0.50 eV of which would tion, to the ligand effect dominating the changes observed in
be attributed to the ligand effect if the substrate were to be the case of H adsorption, with the case of O adsorption lying in
compressed before replacing the Pt with Ru, which would then between. However, in all cases considered, both the strain and
result in another 0.52 eV weakening in the adsorption. However if ligand effects resulting from replacing a Pt substrate with Ru
the Pt is replaced with Ru before compression, then the ligand substrate work together in weakening adsorption.
affect accounts for 0.77 eV of the 1.02 eV total adsorption energy 2.3.3 Surface catalysis. Having considered the influence of
change, and the strain effect for only 0.24 eV. the strain and ligand effects on several adsorbates we now turn

Phys. Chem. Chem. Phys. This journal is © the Owner Societies 2014
 View Article Online

PCCP Perspective
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

 pffiffiffi  pffiffiffi
Fig. 4 Energies of formation (eV) per p 2  3 unit cell for 1/3 ML of Fig. 5 Reaction energies (eV) per p 2  3 unit cell for the formation
hydrogen on a Pt(111) overlayer on PtRu substrate slabs with various PtRu of 1/3 ML of OH from 1/3 ML each of preadsorbed O and H on a
compositions (corresponding to the ligand effect; these are reported as % Pt(111) overlayer on PtRu substrate slabs with various PtRu compositions
Pt, and correspond to compositions for the 5 substrate layers—i.e. exclud- (corresponding to the ligand effect; these are reported as % Pt, and
ing the Pt6 overlayer—of Ru30, Pt10Ru20, Pt20Ru10 & Pt30) and unit cell correspond to compositions for the 5 substrate layers—i.e. excluding the
parameters (corresponding to the strain effect; these are reported as the Pt6 overlayer—of Ru30, Pt10Ru20, Pt20Ru10 & Pt30) and unit cell parameters
corresponding fcc cubic unit cell dimension in Angstroms). Reference (corresponding to the strain effect; these are reported as the corresponding
states are H2 gas and the corresponding bare PtRu–Pt(111) slab. The same fcc cubic unit cell dimension in Angstroms). The same 28 resulting reaction
28 energies are plotted twice: (a) as a function of the unit cell parameter energies are plotted twice: (a) as a function of the unit cell parameter and (b)
and (b) as a function of substrate composition. White atoms are H, purple as a function of substrate composition. Red atoms are O, white atoms are H,
atoms are Ru and silver atoms are Pt. purple atoms are Ru and silver atoms are Pt.

our attention to model surface reactions, where we consider dissociation than in the case of OHad formation. However, we
first the formation of OHad from Oad and Had, followed by the should note that for the first time we observe dependence on
dissociation of O2,ad to form Oad. The reaction energies for the subsurface composition which is non-linear. This step-like
the first of these reactions, the formation of OHad, on Pt(111) dependence in Fig. 6b reflects a similar non-linear dependence
overlayers on our 28 test substrates are presented in Fig. 5. The of the O2,ad adsorption energy on composition, which has not
dependence of the reaction energy on the strain is approxi- yet been satisfactorily explained.
mately linear for fixed substrate compositions. Similarly, the Teasing out the concrete implications of the strain and
dependence of the reaction energy on the substrate composi- ligand effects for a process, such as the oxygen reduction
tion is approximately linear for fixed values of the lattice reaction (ORR), requires making assumptions regarding the
parameter. The overall result of replacing Pt (at equilibrium) rate limiting process and how it is affected by the strain and
with Ru (at equilibrium) is a decrease in the exothermic ligand effects respectively. We have already seen that the
reaction energy by 0.74 eV, making the reaction energy more reaction energies for O2,ad dissociation and OHad formation
negative, i.e. more exothermic. If the overlayer is compressed show opposite trends in response to both the strain and ligand
before the composition of the substrate is changed, the strain effects. Thus, one might suppose (as is often done) that a
effect contributes 0.21 eV; however, if the slab is compressed compromise between these two (or two other analogous) reac-
after the Pt substrate has been replaced with Ru, then the tions is required, and that the best attainable catalyst will
compression only results in a change of 0.07 eV. The replace- support each reaction equally. A more rigorous approach would
ment of Pt with Ru in the substrate results in changes of 0.53 be to calculate the barriers for the individual steps in the ORR,
and 0.67 eV to the reaction energy respectively. as has already been done for Pt(111),58,59 as a function of
In the case of O2,ad dissociation (Fig. 6), replacing a pure Pt substrate composition/lattice constant. These barriers could
substrate at equilibrium with a pure Ru substrate at equili- then be used in a kinetic model for the ORR. The optimal
brium decreases the exothermicity of the reaction by 0.67 eV, catalyst would then be the catalyst with the fastest overall rate
from 1.22 eV to 0.55 eV. If the substrate composition is in the kinetic model. In fact, the laterally compressed Pt(111)
changed before the slab is compressed, then the ligand effect surfaces of Pt0.7Ru0.3(111) and Pt0.9Ru0.1(111) exhibit enhanced
accounts for a strong majority (0.52 eV out of 0.67 eV) of this ORR activities compared with pure Pt(111) when studied in
change, however if the order is reversed then compression and 0.1 M HClO4 and 0.1 M H2SO4, respectively. An enhancement in
substitution yield equal contributions of 0.34 eV. Thus, the ORR activity has also been observed for PtRu surface alloys on a
strain effect plays a more prominent role in the case of O2,ad Ru(0001) substrate.60

This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys.
 View Article Online

Perspective PCCP

Fig. 7 (a) Diagram of high-symmetry adsorption sites on Pt3Ni(111). T = top 

Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

m1, B = bridge  m2, F = fcc  m3, H = hcp  m30 ; ‘‘Pt’’ denotes a site adjacent to
only Pt atoms; ‘‘Ni’’ denotes a site with a neighboring Ni atom. Blue atoms are Ni
and silver atoms are Pt. (b) Potential energy surface for hydrogen adsorption on
Pt3Ni(111). Adsorption energies (Ead in eV) are referenced to gas phase H2.

3.2 Site selectivity for hydrogen adsorption

To illustrate site selectivity in bimetallic catalysts, we first
consider the case of hydrogen adsorption on a Pt3Ni(111) surface,
 pffiffiffi
Fig. 6 Reaction energies (eV) per p 2  3 unit cell for the dissocia- in the absence of surface segregation. The case of oxygen adsorption
tion of 1/6 ML of O2 to form 1/3 ML each of adsorbed O on a Pt(111) (along with hydrogen adsorption) has been considered by Jacob
overlayer on PtRu substrate slabs with various PtRu compositions
et al.,73,74 while the influence of only subsurface Ni on ORR reaction
(corresponding to the ligand effect; these are reported as % Pt, and
correspond to compositions for the 5 substrate layers—i.e. excluding energies and barriers has been computed by Sha et al.70
the Pt6 overlayer—of Ru30, Pt10Ru20, Pt20Ru10 & Pt30) and unit cell If we consider a Pt3Ni(111) surface, in which the Ni atoms
parameters (corresponding to the strain effect; these are reported are evenly spaced and isolated, then each type of high symme-
as the corresponding fcc cubic unit cell dimension in Angstroms). try site (T = top  m1, B = bridge  m2, F = fcc  m3, H = hcp  m3 0 )
The same 28 resulting reaction energies are plotted twice: (a) as a
can be further divided into sites with an adjacent Ni atom
function of the unit cell parameter and (b) as a function of substrate
composition. Red atoms are O, purple atoms are Ru and silver atoms (TNi, BNi, FNi & HNi) and sites having only Pt nearest neighbors
are Pt. (TPt, BPt, FPt & HPt), as illustrated in Fig. 7. In Table 1, we compare
hydrogen adsorption energies to these two different categories of
sites on Pt3Ni(111) along with Pt(111). The presence of Ni increases
site selectivity of hydrogen adsorption on the surface. In particular,
3. Hydrogen adsorption on Pt3Ni hydrogen now selectively adsorbs at sites where it comes into
alloys: surface site selectivity and direct contact with a Ni surface atom. As a result the potential
energy landscape of the surface with respect to hydrogen adsorp-
reversible surface segregation tion becomes substantially rougher, with adsorption-potential-
3.1 Introduction to PtNi alloys energy wells centered on surface Ni atoms. Thus, instead of being
As one of the most promising catalysts for the ORR in acidic relatively free to move across the surface—as on Pt(111) where
media, PtNi alloys have received a great deal of attention over the adsorption energies to the various binding sites fall within a
past decade in the search for efficient oxygen reduction catalysts B0.1 eV range—adsorbed hydrogen will tend to get trapped
for cathode materials in polymer electrolyte membrane fuel cells within the immediate vicinity of Ni surface atoms.
(PEMFC).61–70 The outstanding performance of Pt3Ni catalysts
has been attributed by some groups to a d-band shift,18,71 while 3.3 Adsorbate-induced surface segregation
others have argued that it is due to a decrease in the lattice Motivated by this theoretically predicted restricted mobility of the
parameter that results from the alloying72 (in analogy to the adsorbed hydrogen, the local structure of the nanoparticulate
strain effect), and still others have sought to understand the
catalytic behavior in terms of local atomic configurations at
Table 1 Adsorption energies (eV) for hydrogen at high symmetry sites on
the surface.70,73,74 Consideration of the combination of a stable
Pt(111) and Pt3Ni(111) referenced to gas phase H2
noble metal (Pt) with a reactive transition metal (Ni), which is
unstable at operating conditions (i.e. at positive potentials in Substrate Adsorption site Adsorption energy (eV)
acid media), stresses the importance of the catalysts stability and Pt(111) fcc 0.41
also of the dynamics of the alloy catalyst during operation. hcp 0.38
Another important consideration for Pt3Ni catalysts is the selec- top 0.49
tive surface segregation observed in experiments,75–79 whose Pt3Ni(111) fcc-Ni 0.46
influence on their catalytic properties has been considered fcc-Pt 0.40
theoretically.80 Here, we use Pt3Ni to consider, first, the role of hcp-Ni 0.43
hcp-Pt 0.36
local surface configurations followed by the influence of adsor- top-Ni 0.01
bates on surface segregation. top-Pt +0.38

Phys. Chem. Chem. Phys. This journal is © the Owner Societies 2014
 View Article Online

PCCP Perspective

of each constituent between catalyst surface and bulk sites,

along with the local chemical composition and bond distances
of neighboring atoms.
The results of the in situ XAS paint a dynamic picture of the
Pt3Ni catalyst’s structure. The combination of the coordination
number (CN) and local chemical composition (expressed as a
Pt/Ni ratio) indicates that Ni is primarily confined to sub-surface
sites at potentials outside the hydrogen adsorption region. The
decrease in the Ni coordination number from 11.5 to 9.5 (see
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

Fig. 8) following the polarization of the Pt3Ni catalyst to the

hydrogen adsorption region suggests a gradual transfer of Ni
from the bulk to surface sites, as a result of hydrogen adsorption.
The chemical composition of the Ni environment shows a
complementary effect when the local Ni content relaxes from
slightly over-stoichiometric value in the double layer region
(when the Ni is confined to the bulk) to that approaching the
Fig. 8 (a) Average coordination number (CN) and (b) local chemical average chemical composition once the presence of the Ni in
composition in the vicinity of Ni for a nafion-stabilized nanoparticulate the surface becomes possible.
Pt3Ni electrode, polarized at various potentials in the HUPD region. Local The transfer and subsequent accumulation of Ni in the
chemical composition was calculated from independently refined Pt and
surface also manifests itself in the voltammetric behavior of
Ni occupancy in the first coordination shell. Adapted from ref. 81.
the Pt3Ni catalysts cyclically polarized in the HUPD region. The
charge corresponding to the hydrogen adsorption gradually
Pt3Ni (d = 5 nm) carbon supported catalyst in 0.1 HClO4 decreases with decreasing polarization rate as well as with
was assessed using X-ray absorption spectroscopy. Here, the Ni K the total time the catalysts resides in the HUPD potential region
edge and Pt L3 edge X-ray absorption spectra (XAS) were recorded (see Fig. 9). A direct link between hydrogen adsorption and Ni
under conditions corresponding to hydrogen underpotential accumulation in the surface confirming the in situ XAS experi-
deposition (HUPD). When subject to a full profile refinement, ments gives the clear stripping signal of the accumulated Ni if
the extracted extended X-ray absorption fine structure (EXAFS) the Pt3Ni catalyst is polarized to sufficiently positive potentials
function yields average coordination numbers for each alloy’s following the extensive polarization of the Pt3Ni catalysts in the
constituent. These coordination numbers reflect the distribution HUPD region. The observed behavior also stresses the essential

Fig. 9 (a) Voltammetric curves and (b) integrated charge of the underpotential deposition of hydrogen on a Pt3Ni nanoparticulate electrode in 0.05 M
H2SO4 recorded consecutively at various polarization rates (see figure legend). (c) Comparison of the last voltammogram recorded before polarization in
HUPD region (blue curve) with the first two cycles (green and red curves respectively) recorded after cycling the electrode in the HUPD region for 2 hours.
All voltammograms in (c) were recorded at the polarization rate of 20 mV s1. Adapted from ref. 81.

This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys.
 View Article Online

Perspective PCCP
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

Fig. 11 Schematic of adsorbate-induced surface segregation, in which

the presence of adsorbed hydrogen induces a change in the surface
composition within an experimentally relevant timeframe. White atoms
are H, blue atoms are Ni and silver atoms are Pt. Adapted from ref. 81.

surface segregation, but reversible surface segregation, in which the

catalyst’s structure (and thus catalytic nature) is altered only tem-
porarily during catalysis due to the adsorption of intermediates.

Fig. 10 (a) Top views of slabs illustrating the most stable adsorbate 4. HER on AuPd alloys: the ensemble
configuration and metal surface composition at each coverage consid-
ered. The coverages correspond to the legend in (b) directly below them.
effect and adsorbate-induced
(b) Surface segregation energies (eV) per 2  2 unit cell for Pt3Ni slabs with ensemble formation
no adsorbates, 1/2 ML H, 1 ML H, 1 1/2 ML H, and 1/4 ML O. Surface
compositions are (left to right) 0%, 25%, 50%, 75% and 100% Ni. H atoms
4.1 Introduction to AuPd alloys
are white, Ni atoms are blue, O atoms are red, and Pt atoms are gray. The adsorption and catalytic properties of AuPd systems have
(c) Side views of slabs along the bottom illustrate Ni–Pt layer composition
been investigated in the context of a variety of chemical reactions
for various surface segregation states and correspond to the compositions
along the x-axis in (b) directly above them. Red atoms are O, white atoms
for more than two decades.82 Because of the strong dependence
are H, blue atoms are Ni and silver atoms are Pt. Adapted from ref. 81. of their catalytic activity on the atomic structure and composition
of the catalytic surface, they have provided an ideal model system
for investigating a wide range of (electro)catalytic reactions.83
instability of the Pt3Ni catalysts at potentials characteristic of Indeed, a plethora of diverse structures have been fabricated,84
the ORR process, hence kinetic stabilization of Pt3Ni catalysts is including heteroepitaxial layers ranging from submonolayers85,86
prerequisite for the application of these catalysts in the cathode to multilayers,82 decorated defects (e.g. steps87), adsorbed nano-
reaction of fuel cells. structures,88,89 and alloyed nanoparticles.90 This rich variety of
To confirm the responsibility of adsorbate hydrogen for alloy structures is met with an equally rich selection of (electro)-
reversing the surface segregation of the Pt3Ni nanoparticles, DFT chemical reactions which have been characterized on AuPd
calculations were performed examining the effect of hydrogen (electro)catalysts.82 Among these, the hydrogen evolution reac-
adsorbates on surface segregation for a Pt3Ni(111) slab. These tion (HER) has attracted particular attention.82,83,86,91
surface segregation energies are presented in Fig. 10, where the
influence of 1/4 ML of atomic oxygen is also shown for the sake of 4.2 Ensemble effect
comparison. As can be seen in the figure, the presence of adsorbed One of the recurring themes in studies of AuPd surface alloys is
hydrogen stabilizes Ni atoms in the surface due to the strength the ensemble effect, in which specific geometric groupings of
of Ni–H bonds in comparison to Pt–H bonds, as was shown in atoms at the surface function as adsorption or catalytic active
the previous section. Thus, the DFT calculations confirm that sites.8,91–93 Temperature programmed desorption (TPD) studies
adsorbed hydrogen is capable of inducing a change in the of preadsorbed CO on AuPd alloyed surfaces provide an illustrative
surface composition of the Pt3Ni nanoparticles by stabilizing example of this effect. Because CO does not adsorb on a pure
Ni atoms in the particle surface. The experiments show that this Au(111) surface, its adsorption on AuPd is attributed to the
change in surface composition takes place on the order of presence of Pd atoms. A single Pd atom (i.e. a Pd surface monomer)
seconds and is reversible as depicted in Fig. 11. Thus, Pt3Ni suffices to support CO adsorption; however, stronger adsorption
alloys provide us not only with an example of adsorbate-induced takes place when coordination to two adjacent Pd atoms is

Phys. Chem. Chem. Phys. This journal is © the Owner Societies 2014
 View Article Online

PCCP Perspective

Fig. 12 Examples of Pd ensembles in a Au(111) surface: (a) monomer,

(b) dimer, (c) monomer pair and (d) trimer. Blue atoms are Pd and gold
atoms are Au.
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

made possible via a Pd surface dimer. Adsorption at a Pd trimer

is even stronger. (See Fig. 12 for examples of these and other Pd
surface ensembles.) These three different types of adsorption
sites are reflected in three different TPD peaks, which can be
used to determine the populations of Pd monomers, dimers
and trimers.94,95
Ethylene dehydrogenation on AuPd is another example of a
catalytic reaction governed by an ensemble effect. Here pairs of
adjacent Pd atoms (i.e. Pd dimers) serve as the active site,
enabling the C atoms to transform their C–C p-bond into a pair Fig. 14 k3-normalized EXAFS functions extracted from X-ray absorption
of s-bonds, one with each of the Pd atoms in the dimer.94 In a spectra collected at Au L3 for controlled potentials: (a) 0.15 V, (b) 0.20 V
third example, an experimental study of the synthesis of vinyl and (c) 0.25 V, and at the Pd K-edge for the same controlled potentials:
acetate on AuPd alloys with various surface compositions (d) 0.15 V, (e) 0.20 V and (f) 0.25 V. Solid lines show the NLLS fits of the
local structure refinement. The coordination numbers (CN) from the NLLS
suggested that a pair of noncontiguous Pd surface monomers is
fits are given above the plots. Adapted from ref. 97.
the critical ensemble for catalytically supporting this reaction.92
As a final example, the reactivity of AuPd surface structure with
hydrogen has been studied extensively.82,83,91,93,96,97 While Pd
surface dimers are believed to be the critical (i.e. minimal in this local environment of Pd, where the Pd EXAFS functions prove to
case) ensemble for the underpotential deposition of hydrogen,8 be highly sensitive to hydrogen adsorption and thus reflect a
monomers appear to furnish the active sites in the hydrogen change in the surface composition triggered by hydrogen adsorp-
evolution reaction.91 Fig. 13 illustrates how DFT calculations have tion (see Fig. 14). In particular, when the applied electrode
been used to evaluate the increase in hydrogen adsorption energy potential is set to values corresponding to hydrogen adsorp-
as the number of Pd atoms directly coordinated with the hydrogen tion/evolution, the average coordination number of Pd decreases
adsorbate increases from zero to three.93 with decreasing electrode potential from ca. 8.6 (0.15 V vs. SCE)
to ca. 4.5 (0.25 V vs. SCE). Furthermore the absorbing Pd atoms
appear to be coordinated an average of one hydrogen (with the
4.3 Adsorbate-induced ensemble formation
bonding distance of 1.82 Å) and have two types of metal
To investigate the dynamic nature of AuPd systems in situ EXAFS neighbors, whose distances change from 2.77 Å and 2.98 Å (at
data were collected and analyzed in a manner similar to the data 0.15 V vs. SCE) to 2.81 and 3.46 Å (at 0.25 V vs. SCE). These
for the Pt3Ni system. Along these lines in situ X-ray absorption local arrangements differ significantly from known palladium
spectroscopy was recently used to track surface segregation in hydride phases (thus ruling out the formation of a palladium
Au4Pd nanoparticles during potential driven hydrogen adsorp- hydride phase) and reflect an extensive transfer of Pd to the
tion.97 Due to the overall chemical composition of the nano- surface, presumably to maximize the number of Pd–H interac-
particles, their dynamic nature is reflected most clearly in the tions. The experimentally obtained average Pd coordination
number of 4.5 is significantly lower than expected for atoms
confined to low index surfaces (i.e. 8–9). Thus, the coordination
numbers strongly suggests the complete confinement of the Pd
atoms to the surface, forming a stable core–shell like arrange-
ment, at intermediate potentials (near 0.15 V vs. SCE), followed
by the confinement of Pd to even lower coordination sites (e.g. steps
or edges) and possible formation of Pd surface structures with
ultra-low coordination numbers (e.g. adatoms) as the potential is
further decreased (E o 0.25 V vs. SCE).
Fig. 13 Hydrogen adsorption on a Au(111) surface with various Pd ensembles:
(a) no Pd, (b) Pd monomer, (c) Pd dimer and (d) Pd trimer. Adsorption energies
In order to provide independent evidence for this interpreta-
are from DFT and referenced to gas phase H2. Adapted from ref. 93. White tion of the experimental results and gain additional insight into
atoms are H, blue atoms are Pd and gold atoms are Au. the detailed atomistic mechanisms involved, a series of DFT

This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys.
 View Article Online

Perspective PCCP
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

Fig. 15 Schematic with energies for hydrogen adsorption and Pd surface

segregation involving a Pd monomer at or near the Au(111) surface. White
atoms are H, blue atoms are Pd and gold atoms are Au.

Fig. 17 Scheme of reversible surface trimer formation on AuPd catalyst

(111) during the hydrogen evolution reaction when the hydrogen coverage
is equal to the Pd concentration at or near the surface. State A is the
equilibrium state in the absence of adsorbates. State B is the metastable
state immediately following hydrogen adsorption, but before the catalyst
has had a chance to rearrange itself. State C is the equilibrium state of the
catalyst with adsorbed hydrogen. State D is the metastable state of the
catalyst following the evolution of H2 gas from the surface. The conversion
of hydrogen atoms to H2 gas results in an energy gain of 2.27 eV per
hydrogen atom (the energy gain per cycle). All energies (eV) are computed
per hydrogen atom. White atoms are H, blue atoms are Pd and gold atoms
are Au. Adapted from ref. 97.

the favorability of subsurface Pd to DE = 0.06 eV, but bringing

the Pd monomer to the surface is still unfavorable. The
adsorption of an additional hydrogen atom is able to reverse
the trend so that a pair of hydrogen atoms stabilizes surface Pd
Fig. 16 Scheme of reversible surface monomer formation on a AuPd by 0.11 eV (see Fig. 16).
catalyst (111) during the hydrogen evolution reaction when the hydrogen Alternatively, three hydrogen atoms are able to cooperate
coverage is double the number of Pd atoms at or near the surface. State A is and draw three Pd atoms to the surface to form a surface trimer
the equilibrium state in the absence of adsorbates. State B is the metastable
with an energy gain of DE = 0.18 eV per hydrogen atom (see
state immediately following hydrogen adsorption, but before the catalyst
has rearranged itself. State C is the equilibrium state of the catalyst with Fig. 17). Thus, adsorbed hydrogen atoms are not only respon-
adsorbed hydrogen. State D is the metastable state of the catalyst following sible for bringing Pd to the surface, but also for the formation
the departure of H2 gas from the surface. The conversion of hydrogen of Pd trimers, which in the absence of the adsorbed hydrogen
atoms to H2 gas results in an energy gain of 2.27 eV per hydrogen atom (the are DE = 0.06 eV less stable than three non-interacting Pd
energy gain per cycle). All energies (eV) are computed per hydrogen atom.
monomers. If the hydrogen leaves the surface, then the surface
White atoms are H, blue atoms are Pd and gold atoms are Au.
Pd (as either monomers or trimers) becomes thermodynami-
cally unstable so that migration of Pd into the subsurface to
calculations were performed to investigate hydrogen-adsorbate- form subsurface monomers is exothermic (DE = 0.37 eV and
induced surface segregation in AuPd alloys.97 A Au(111) surface DE = 0.18 eV per previously involved hydrogen for the two
described using a (p3  3) unit cell was chosen as a first cases respectively). Thus, on the assumption that segregation is
approximation of the nanoparticle surface, which is expected not kinetically hindered (which would appear to be the case
to involve other faces along with edges, corners and possibly based on experimental observations of nanoparticles), then Pd
other defects. In the absence of adsorbed hydrogen a Pd moves between the surface and subsurface layers as hydrogen
monomer is found to prefer occupying a subsurface site by is adsorbed and desorbed/evolved.
DE = 0.37 eV over a surface site, as shown in Fig. 15. The However, there is another possibility which we ought to
adsorption of one hydrogen atom per Pd monomer decreases consider: namely, that the attraction between hydrogen and Pd

Phys. Chem. Chem. Phys. This journal is © the Owner Societies 2014
 View Article Online

PCCP Perspective

low Pd–M coordination numbers obtained from the analysis of

the EXAFS data.

5. Concluding remarks
5.1 Indispensability of rational design
Bimetallic alloys exhibit a spectrum of electrochemical and
catalytic characteristics that go far beyond those of unalloyed
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

metals. The numerous combinations and nearly endless array

of conceivable compositions and structures (see ref. 98 for an
overview of surface morphologies of bimetallic alloys as well as the
discussion of additional complicating factors) for each composition
might tantalize one with the thought that for many an application a
better bimetallic material (i.e. a new composition and/or structure)
is out there, just waiting to be synthesized or discovered. Yet the
same overabundance of variation that would seem to guarantee the
existence of a better bimetallic surface for a given application turns
the search for the optimal materials into a game of finding the
needle in a haystack. Strategies for the rational design of electrode
and catalyst materials are needed in order to quickly focus on
promising candidate materials and improve their performance in
Fig. 18 Energies for the formation of subsurface palladium hydride from
adsorbed hydrogen and subsurface Pd for various concentrations of Pd at an informed rather than haphazard manner.
or near the surface and various coverages of adsorbed hydrogen: (a) 1 : 3 The proper foundation for rational design is a fundamental
H : Pd ratio with 1/9 ML adsorbed hydrogen, (b) 1 : 1 H : Pd ratio with 1/9 ML understanding of the functionality of candidate materials. This
adsorbed hydrogen, (c) 1 : 1 H : Pd ratio with 1/3 ML adsorbed hydrogen, (d) entails a detailed understanding of the atomistic mechanisms for
3 : 1 H : Pd ratio with 1/3 ML adsorbed hydrogen. All energies (eV) are
all relevant processes. The study of model systems is indispen-
computed per hydrogen atom. White atoms are H, blue atoms are Pd and
gold atoms are Au. sable for laying this foundation, because it allows for the isolation
and subsequent elucidation of the effects of individual structural
and compositional parameters. In other words, model systems
could result in subsurface hydrogen, rather than surface Pd. As can help isolate the structures and behaviors of a material which
illustrated in Fig. 18, this was only found to be the case for low are essential for its functionality. In discussing three particular
hydrogen to Pd ratios. Thus, three Pd subsurface monomers are model systems one of the central aims of this perspective has
able to cooperate to draw a single hydrogen atom into a subsurface been to highlight behaviors that are essential to the functionality
interstitial position between them. Interestingly, this means that of bimetallic systems in general, and thus need to be considered
the Pd atoms lose their status as monomers as they now form a in efforts to understand other systems as well. Therefore,
subsurface trimer, which is then ready to be drawn to the surface by insofar as these behaviors are critically involved in determining
the appearance of additional hydrogen adsorbates. the functionality of bimetallic materials, the successful rational
Combining the EXAFS and DFT results we arrive at the design of catalyst and electrode materials cannot proceed
following, most probable atomistic description of adsorbate- without taking them into account.
induced ensemble formation. The most stable surface of the
Au4Pd particle is pure Au prior to the adsorption of hydrogen, 5.2 Considering structural stability
with subsurface Pd atoms avoiding each others, and thus In light of the very real danger of catalyst degradation, the impor-
forming subsurface monomers. The adsorption of small quantities tance of structural stability has long been a central concern,
of hydrogen, leads to hydrogen absorption into the subsurface, particularly in the fuel cell community. In this context, it is not
where a stabilizing Pd trimer is gathered around each subsurface just the macroscopic material, but the catalytically active surface
hydrogen atom. As the number of ad- and absorbed hydrogen structure itself, which must be stable under catalytic conditions.
atoms approaches the number of subsurface Pd atoms, these Pd Adsorbed intermediates are an intrinsic part of heterogeneous
subsurface trimers are drawn to the surface. However, the adsorp- (electro)catalysis, and so the possibility of adsorbate-induced
tion of additional hydrogen encourages the breakup of the trimers surface segregation or ensemble formation is unavoidable.
into monomers, because monomers enable a greater number of A first implication of this possibility is that mechanistic models
hydrogen atoms to bond with Pd (up to three hydrogen adsorbates and studies must identify the surface structure relevant to
per Pd monomer instead of just one or two). The migration of catalytic conditions and then investigate the reaction on this surface,
Pd to even lower-coordination sites (e.g. edges, steps, corners, since it is the only available surface for catalysis. A second
adatoms) is needed to allow for more than three hydrogen implication is that surface stability in the presence of adsorbed
atoms to coordinate with each Pd and is supported by the very intermediates (i.e. under the conditions under which catalysis

This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys.
 View Article Online

Perspective PCCP

takes place) must be taken into account during rational catalyst secrets of bimetallic electrodes and catalysts, and is likely
design, because only surface structures that are realizable to play an important role in the further development of
under reaction conditions are viable candidates. On one hand, nanostructured electrodes and catalysts for all sorts of diverse
this might mean that the ‘‘ideal’’ surface structure is unstable applications.
and thus unrealizable under reaction conditions. On the other
hand, the ‘‘ideal’’ surface might be formed in operando so
that only the proper precursor system needs synthesized or Acknowledgements
prepared beforehand.
We gratefully acknowledge financial support from the Deutsche
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

5.3 Synthesizing surface structures Forschungsgemeinschaft (DFG) through FOR-1376 (Elementary

reaction steps in Electrochemistry: Theory meets Experiment).
The close relationship between the atomic structure and func-
JEM gratefully acknowledges the Alexander von Humboldt
tional performance of a surface stresses the fact that successful
Foundation for financial support. PK gratefully acknowledges
surface synthesis is an indispensable aspect of the development
the support of the Deutsches Elektronen Snychrotronen (project
of improved electrodes and catalysts. One would expect that
11734) and of Brookhaven National Laboratory (project 17851).
obtaining such a high degree of synthetic control (e.g. to produce
Support from the European Union through the ERC-Starting
surface ensembles) would be likely to pose a significant challenges.
Grant THEOFUN (Grant-Agreement No. 259608) is also gratefully
While there will certainly be challenges along these lines, there
acknowledged.
are two considerations which could work to one’s benefit in the
task of synthesizing newly designed surface structures. First, if
rational design limits itself to surface structures, which are Notes and references
stable under operating conditions, then fruitless attempts to
synthesize unstable surfaces will be avoided. Second, it may not 1 H. Gerischer, Bull. Soc. Chim. Belg., 1958, 67, 506.
be necessary to directly synthesize the desired surface. Instead, 2 R. Parsons, Trans. Faraday Soc., 1958, 54, 1053.
the synthesis of a precursor, which automatically transforms 3 S. Trasatti, J. Electroanal. Chem., 1972, 39, 163.
in operando into the desired surface, may suffice. 4 J. K. Nørskov, T. Bligaard, A. Logardottir, J. R. Kitchin,
J. G. Chen, S. Pandelov and U. Stimming, J. Electrochem.
5.4 Delving into the dynamics Soc., 2005, 152, J23.
An important theme that has already been touched on several 5 W. Schmickler and S. Trasatti, J. Electrochem. Soc., 2006,
times is the dynamic, as opposed to static, nature of bimetallic 153, L31.
surfaces. Indeed the presence of adsorbed intermediates or 6 R. Parsons, Volcano Curves in Electrochemistry, in Catalysis
other adsorbates is capable of exerting a profound influence on in Electrochemistry, ed. E. Santos and W. Schmickler,
the surface structure of a bimetallic alloy. Possible results Wiley-VCH, 2011, p. 1–16.
include a change in the atomic composition at the surface 7 A. Jackson, V. Viswanathan, A. J. Forman, A. H. Larsen, J. K.
(i.e. adsorbate-induced surface segregation), the formation Nørskov and T. F. Jaramillo, ChemElectroChem, 2014, 1, 67–71.
of a particular ensemble (i.e. adsorbate-induced ensemble 8 F. Maroun, F. Ozanom, O. M. Magnussen and R. J. Behm,
formation), a change in the relative stabilities of the different Science, 2001, 293, 1811–1814.
surfaces faces, and the stabilization of defects or new surface 9 M. Watanabe and S. Motoo, J. Electroanal. Chem., 1975, 60,
structures (e.g. steps, adatoms or surface buckling). Thermo- 259–266.
dynamic favorability requires that a decrease in the surface free 10 M. Watanabe and S. Motoo, J. Electroanal. Chem., 1975, 60,
energy upon the adsorption of intermediates drive transforma- 267–274.
tion of the surface into its active structure, while a similar 11 M. Watanabe and S. Motoo, J. Electroanal. Chem., 1975, 60,
decrease in free energy drives the vacated surface to revert to its 275–283.
original structure. Kinetic feasibility requires that the reaction 12 H. A. Gasteiger, N. Markovic, P. N. Ross and E. J. Cairns,
energy barriers for these surface transformations are low enough J. Phys. Chem., 1993, 97, 12020–12029.
for them to take place within the timeframe of the surface’s 13 H. A. Gasteiger, N. Markovic, P. N. Ross and E. J. Cairns,
exposure to the reactive environment. Obviously, a metastable J. Phys. Chem., 1994, 98, 617–625.
structure could be either advantageous (e.g. if it is more catalyti- 14 H. A. Gasteiger, N. Markovic, P. N. Ross and E. J. Cairns,
cally active than the kinetically hindered global structure) or Electrochim. Acta, 1994, 39, 1825–1832.
disadvantageous if the structure corresponding to the global 15 N. M. Markovic, H. A. Gasteiger, P. N. Ross, X. Jiang, I. Villegas
minimum is desired. In either case, a full understanding of the and M. J. Weaver, Electrochim. Acta, 1995, 40, 91–98.
functionality of a bimetallic surface subject to some form of 16 B. Hammer and J. K. Nørskov, Surf. Sci., 1995, 343, 211–220.
adsorbate-induced surface segregation requires not only a knowl- 17 J. Greeley, J. K. Nørskov and M. Mavrikakis, Annu. Rev. Phys.
edge of the thermodynamics of catalyst dynamics, but also an Chem., 2002, 53, 319–348.
understanding of the kinetics relevant to these changes. Delving 18 V. R. Stamenkovic, B. S. Mun, K. J. J. Mayrhofer, P. N. Ross,
into this kinetic aspect of surface dynamics is one of the most N. M. Marković, J. Rossmeisl, J. Greeley and J. K. Nørskov,
exciting challenges facing researchers seeking to unlock the Angew. Chem., Int. Ed., 2006, 45, 2897–2901.

Phys. Chem. Chem. Phys. This journal is © the Owner Societies 2014
 View Article Online

PCCP Perspective

19 G. Kyriakou, M. B. Boucher, A. D. Jewell, E. A. Lewis, T. J. Lawton, 46 L. A. Kibler, A. M. El-Aziz, R. Hoyer and D. M. Kolb, Angew.
A. E. Baber, H. L. Tierney, M. Flytzani-Stephanopoulos and Chem., 2005, 117, 2116–2120; L. A. Kibler, A. M. El-Aziz,
E. C. H. Sykes, Science, 2012, 335, 1209–1212. R. Hoyer and D. M. Kolb, Angew. Chem., Int. Ed., 2005, 44,
20 J. M. Thomas, Phys. Chem. Chem. Phys., 2014, 16, 7647–7661. 2080–2084.
21 F. Tao, M. E. Grass, Y. Zhang, D. R. Butcher, J. R. Renzas, 47 J. L. Zhang, M. B. Vukmirovic, Y. Xu, M. Mavrikakis and
Z. Liu, J. Y. Chung, B. S. Mun, M. Salmeron and R. R. Adzic, Angew. Chem., 2005, 117, 2170–2173; J. L. Zhang,
G. A. Somorjai, Science, 2008, 322, 932–934. M. B. Vukmirovic, Y. Xu, M. Mavrikakis and R. R. Adzic,
22 C. Cui, M. Ahmadi, R. Behafarid, L. Gan, M. Neumann, Angew. Chem., Int. Ed., 2005, 44, 2132–2135.
M. Heggen, B. R. Cuenya and P. Strasser, Faraday Discuss., 48 T. Diemant, T. Hager, H. E. Hoster, H. Rauscher and
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

2013, 162, 91–112. R. J. Behm, Surf. Sci., 2003, 541, 137–146.

23 J. A. Rodriguez, Surf. Sci. Rep., 1996, 24, 223–287. 49 B. D. McNicol and R. T. Short, J. Electroanal. Chem., 1977,
24 M. A. Vasiliev, J. Phys. D: Appl. Phys., 1997, 30, 3037–3070. 81, 249–260.
25 R. Ferrando, J. Jellinek and R. L. Johnston, Chem. Rev., 2008, 50 H. A. Gasteiger, P. N. Ross and E. J. Cairns, Surf. Sci., 1993,
108, 845–910. 293, 67–80.
26 H.-L. Jiang and Q. Xu, J. Mater. Chem., 2011, 21, 13705–13725. 51 A. M. El-Aziz, R. Hoyer and L. A. Kibler, ChemPhysChem,
27 D. Wang and Y. Li, Adv. Mater., 2011, 23, 1044–1060. 2010, 11, 2906–2911.
28 J. Bockris and H. Wroblowa, J. Electroanal. Chem., 1964, 7, 52 A. Groß, Top. Catal., 2006, 37, 29–39.
428–451. 53 A. Schlapka, M. Lischka, A. Groß, U. Käsberger and P. Jakob,
29 O. A. Petry, B. I. Podlovchenko, A. N. Frumkin and H. Lal, Phys. Rev. B: Condens. Matter Mater. Phys., 2003, 91, 016101.
J. Electroanal. Chem., 1965, 10, 253–269. 54 H. E. Hoster, O. B. Alves and M. T. M. Koper, ChemPhysChem,
30 B. R. Rauhe, F. R. McLarnon and E. J. Cairns, J. Electrochem. 2010, 11, 1518–1524.
Soc., 1995, 142, 1073–1084. 55 E. A. Heider, Diploma thesis, University of Ulm, 2011.
31 T. J. Schmidt, M. Noeske, H. A. Gasteiger, R. J. Behm, 56 E. Santos, P. Quaino and W. Schmickler, Electrochim. Acta,
P. Britz and H. Bçnnemann, J. Electrochem. Soc., 1998, 145, 2010, 55, 4346–4352.
925–931. 57 T. A. Maark and A. A. Peterson, J. Phys. Chem. C, 2004, 118,
32 N. M. Markovic, T. J. Schmidt, B. N. Grgur, H. A. Gasteiger, 4275–4281.
R. J. Behm and P. N. Ross, J. Phys. Chem. B, 1999, 103, 58 J. A. Keith and T. Jacob, Angew. Chem., Int. Ed., 2010, 49,
8568–8577. 9521–9525.
33 A. Hamnett, in Interfacial Electrochemistry: Theory, Experi- 59 J. A. Keith, G. Jerkiewicz and T. Jacob, ChemPhysChem, 2010,
ment, and Applications, ed. A. Wieckowski, Marcel Dekker, 11, 2779–2794.
New York, 1999, pp. 843–883. 60 S. Brimaud, A. K. Engstfeld, O. B. Alves, H. E. Hoster and
34 S. R. Brankovic, N. S. Marinkovic, J. X. Wang and R. R. Adzic, R. J. Behm, Top. Catal., 2014, 57, 222–235.
J. Electroanal. Chem., 2002, 532, 57–66. 61 A. Appleby and F. Foulkes, Fuel Cell Handbook, Van Nostrand
35 P. Waszczuk, G. Q. Lu, A. Wieckowski, C. Lu, C. Rice and Reinhold, New York, 1989.
R. I. Masel, Electrochim. Acta, 2002, 47, 3637–3652. 62 K. Kordesch and G. Simader, Fuel Cells and Their Applica-
36 K. A. Friedrich, K. P. Geyzers, U. Linke, U. Stimming and tions, VCH, New York, 1996.
J. Stumper, J. Electroanal. Chem., 1996, 402(2), 123–128. 63 V. R. Stamenković, T. J. Schmidt, P. N. Ross and N. M. Marković,
37 W. F. Lin, M. S. Zei, M. Eiswirth, G. Ertl, T. Iwasita and J. Phys. Chem. B, 2002, 106, 11970–11979.
W. Vielstich, J. Phys. Chem. B, 1999, 103, 6968–6977. 64 V. R. Stamenković, B. Fowler, B. S. Mun, G. F. Wang,
38 K. A. Friedrich, K. P. Geyzers, A. Marmann, U. Stimming and P. N. Ross, C. A. Lucas and N. M. Marković, Science, 2007,
R. Vogel, Z. Phys. Chem., 1999, 208, 137–150. 315, 493–497.
39 H. Massong, H. S. Wang, G. Samjeske and H. Baltruschat, 65 N. P. Brandon, S. Skinner and B. C. H. Steele, Annu. Rev.
Electrochim. Acta, 2000, 46, 701–707. Mater. Res., 2003, 33, 183–213.
40 T. Iwasita, H. Hoster, A. John-Anacker, W. F. Lin and 66 V. Mehta and J. S. J. Cooper, Power Sources, 2003, 114, 32–53.
W. Vielstich, Langmuir, 2000, 16, 522–529. 67 J. Greeley, I. E. L. Stephens, A. S. Bondarenko, T. Johansson,
41 H. Hoster, T. Iwasita, H. Baumgärtner and W. Vielstich, H. A. Hansen, T. F. Jaramillo, J. Rossmeisl, I. Chorkendorff
Phys. Chem. Chem. Phys., 2001, 3, 337–346. and J. K. Nørskov, Nat. Chem., 2009, 1, 552–556.
42 Y. Tong, H. S. Kim, P. K. Babu, P. Waszczuk, A. Wieckowski 68 J. K. Nørskov, T. Bligaard, J. Rossmeisl and
and E. Oldfield, J. Am. Chem. Soc., 2002, 124, 468–473. C. H. Christensen, Nat. Chem., 2009, 1, 37–46.
43 G.-Q. Lu, P. Waszczuk and A. Wieckowski, J. Electroanal. 69 T. Wadayama, N. Todoroki, Y. Yamada, T. Sugawara,
Chem., 2002, 532, 49–55. K. Miyamoto and Y. Iijama, Electrochem. Commun., 2010,
44 N. P. Lebedeva, M. T. M. Koper, J. M. Feliu and R. A. van 12, 1112.
Santen, J. Electroanal. Chem., 2002, 524–525, 242–251. 70 Y. Sha, T. H. Yu, B. V. Merinov, P. Shirvanian and
45 C. Roth, N. Benker, T. Buhrmester, M. Mazurek, M. Loster, W. A. Goddard, J. Phys. Chem. C, 2012, 116, 21334.
H. Fuess, D. C. Koningsberger and D. E. Ramaker, J. Am. 71 J. Greeley and J. K. Nørskov, J. Phys. Chem. C, 2009, 113,
Chem. Soc., 2005, 127, 14607–14615. 4932–4939.

This journal is © the Owner Societies 2014 Phys. Chem. Chem. Phys.
 View Article Online

Perspective PCCP

72 H. Yang, W. Vogel, C. Lamy and N. Alonso-Vante, J. Phys. 85 J. Tang, M. Petri, L. A. Kibler and D. M. Kolb, Electrochim.
Chem. B, 2004, 108, 11024–11034. Acta, 2005, 51, 125–132.
73 T. Jacob and W. A. Goddard III, J. Phys. Chem. B, 2004, 108, 86 L. A. Kibler, ChemPhysChem, 2006, 7, 985–991.
8311–8323. 87 J. Steidtner, F. Hernandez and H. Baltruschat, J. Phys. Chem.
74 T. Jacob, B. V. Merinov and W. A. Goddard III, Chem. Phys. C, 2007, 111, 12320–12327.
Lett., 2004, 385, 374–377. 88 J. Tang, M. Petri, L. A. Kibler and D. M. Kolb, Acta
75 Y. Gauthier, Y. Joly, R. Baudoing and J. Rundgren, Phys. Rev. Phys.-Chim. Sin., 2005, 21, 1303–1306.
B: Condens. Matter Mater. Phys., 1985, 31, 6216–6218. 89 D. M. Kolb and F. C. Simeone, Electrochim. Acta, 2005, 50,
76 Y. Gauthier, Surf. Rev. Lett., 1996, 3, 1663–1689. 2989–2996.
Published on 20 June 2014. Downloaded by North Dakota State University on 21/06/2014 02:39:20.

77 U. Bardi, A. Atrei, E. Zanazzi, G. Rovida and P. N. Ross, 90 D. Wang, A. Villa, F. Porta, D. Sua and L. Prati, Chem.
Vacuum, 1990, 41, 437–440. Commun., 2006, 1956.
78 U. Bardi, B. C. Beard and P. N. Ross, J. Catal., 1990, 124, 91 Y. Pluntke, L. A. Kibler and D. M. Kolb, Phys. Chem. Chem.
22–29. Phys., 2008, 10, 3684–3688.
79 S. Deckers, F. H. P. M. Habraken, W. F. Vanderweg, 92 M. Chen, D. Kumar, C.-W. Yi and D. W. Goodman, Science,
A. W. D. Vandergon, B. Pluis, J. F. Vanderveen and 2005, 310, 291–293.
R. Baudoing, Phys. Rev. B: Condens. Matter Mater. Phys., 93 S. Venkatachalam and T. Jacob, Phys. Chem. Chem. Phys.,
1990, 42, 3253–3259. 2009, 11, 3263–3270.
80 T. H. Yu, Y. Sha, B. V. Merinov and W. A. Goddard, J. Phys. 94 K. Luo, T. Wei, C.-W. Yi, S. Axnanda and D. W. Goodman,
Chem. C, 2010, 114, 11527–11533. J. Phys. Chem. B, 2005, 109, 23517–23522.
81 H. Hoffmannová, M. Okube, V. Petrykin, P. Krtil, 95 M. Ruff, N. Takehiro, P. Liu, J. K. Nørskov and R. J. Behm,
J. E. Mueller and T. Jacob, Langmuir, 2013, 29, 9046–9050. ChemPhysChem, 2007, 8, 2068–2071.
82 L. A. Kibler, Electrochim. Acta, 2008, 53, 6824–6828. 96 P. Quaino, E. Santos, H. Wolfschmidt, M. A. Montero and
83 M. E. Björketun, G. S. Karlberg, J. Rossmeisl, I. chorkendorff, U. Stimming, Catal. Today, 2011, 177, 55–63.
H. Wolfschidt, U. Stimming and J. K. Nørskov, Phys. Rev. B: 97 M. Okube, V. Petrykin, J. E. Mueller, D. Fantauzzi, P. Krtil
Condens. Matter Mater. Phys., 2011, 84, 045407. and T. Jacob, ChemElectroChem, 2014, 1, 207–212.
84 P. J. Schäfer and L. A. Kibler, Phys. Chem. Chem. Phys., 2010, 98 F. Calle-Vallejo, M. T. M. Koper and A. S. Bandarenka, Chem.
12, 15225–15230. Soc. Rev., 2013, 42, 5210–5230.

Phys. Chem. Chem. Phys. This journal is © the Owner Societies 2014