Oxygen Reduction Reaction on

REVIEW
Pt and Pt Bimetallic Surfaces:
A Selective Review
N. M. MarkovicÂ*, T. J. Schmidt, V. Stamenkovic and P. N. Ross
Materials Sciences Division, Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720 (US)

Received 28.03.2001, revised 20.06.2001

Abstract
In this review we selectively summarize recent progress, primarily from our laboratory, in the development of the oxygen
reduction reaction (ORR) catalysis on well-defined surfaces. The focus is on two type of metallic surfaces: platinum single
crystals and bimetallic surfaces based on platinum. The single crystal results provide insight into the effects of the platinum
structure on the kinetics of the ORR, and create a fundamental link between the specific activity of Pt (rate per unit area)
and particle size (for various particle shapes). The results show that the structure sensitive kinetics of the ORR arise primar-
ily due to structure sensitive adsorption of anions. In the absence of specific adsorption, such as in Nafion polymer electro-
lyte, no particle size effect is expected. The knowledge of the electrocatalysis of the ORR on model bimetallic surfaces on Pt-
Ni and Pt-Co bulk alloys was used to resolve the enhanced ORR kinetics on supported Pt-Ni and Pt-Co catalysts. Finally, we
show that the ORR on platinum modified with pseudomorphic Pd metal film in alkaline solution is the best catalysts ever
used in O2 reduction. For both bimetallic systems, we demonstrated that the ability to make a controlled and well character-
ized arrangement of two elements in the electrode surface region presage a new era of advances in the ORR electrocatalysis.

1 Introduction a given level of activity. The catalytic factors include the mod-
ification of the intrinsic activity of platinum, usually by mak-
In spite of many attempts in the last decade by researchers ing bimetallic surfaces. The enhancement of catalytic proper-
and fuel cell developers to create a non-Pt catalyst for low ties by the second component may occur through the change
temperature (< 200 C) air cathodes, Pt remains the catalyst of of the local bonding geometry (structure effects), the distribu-
choice at least for acid-based fuel cells. Therefore, much of tion of active sites (ensemble effects) or directly by modifying
the art and science of catalysts development for the ORR rely the reactivity of platinum surface atoms (electronic effects). In
on both the fundamental understanding of the reaction at the real systems, all of these factors may, in general, operate si-
platinum-electrolyte interface [1-4] and the optimization of multaneously, thus, separating these effects and assessing
the catalytic properties of the platinum surface [5,6]. Our ap- their relative importance in the catalytic activity and reaction
proach is to start with model systems, such as single crystal mechanism is a very challenging problem.
electrodes, in order to create a link between the microscopic In this paper, we summarize recent progress, primarily
level of understanding of surface processes and the macro- from our laboratory, in the development of the ORR catalysis
scopic measurement of the kinetic rates of electrochemical re- on well-defined platinum surfaces. The focus will be on two
actions. This knowledge can be used to create tailor-made type of metallic surfaces: platinum single crystals and bime-
surfaces with improved catalytic activity. The optimization of tallic surfaces based on platinum. The single crystal results
the most promising catalysts must incorporate both non-cata- provide insight into the effects of the platinum structure on
lytic and catalytic factors. The non-catalytic factors include the kinetics of the ORR, and create a fundamental link be-
either partial replacement of platinum catalysts with less no- tween the specific activity of Pt (rate per unit area) and parti-
ble metals, or the maximization of the catalyst surface area
while exposing the most active microstructures. This, in turn, ±
will minimize the amount (and cost) of catalysts required for [*] Corresponding author, [email protected]

FUEL CELLS 2001, 1, No. 2 1615-6846/01/0207-105 $ 17.50+.50/0 105

 Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

cle size (for various particle shapes). If the rate of the ORR on to ref. [19,20]. We investigated a commercially available Pt
REVIEW

Pt is independent of particle size, then in practice one wants catalyst (20 wt% platinum supported on Vulcan XC72), two
to achieve and maintain in use the highest possible disper- Pt-Co-alloy catalysts (20 wt% PtCo/Vulcan XC72, a/o = 1/1,
sion, where all Pt atoms are surface atoms. There are a variety and 20 wt% Pt3Co/Vulcan XC72, a/o = 3/1), and a Pt3Ni/
of physical forms that could be used to achieve such disper- Vulcan XC72 (20 wt%, a/o = 3/1) catalyst (E-Tek). To avoid
sion, but the thermodynamically favored shape is a facetted any misunderstanding we will refer to these catalysts as to
pseudosphere on the order of 1 nm in diameter. On the other Pt/Vulcan, PtCo/Vulcan, Pt3Co/Vulcan, and Pt3Ni/Vulcan
hand, if the rate of the ORR varies with particle size and/or respectively. All potentials in this manuscript refer to that of
shape, then there are not only interesting fundamental ques- the reversible hydrogen electrode. For experimental details
tions raised as to why such a variation would occur, i.e., the about our measurements, we refer to ref. [15,17-24].
ªcrystallite size effectº [6], but a practical issue of ªbuildingº
and maintaining in use the optimum particle size and/or
shape. In practice, finding the optimum form of the Pt cata- 3 ORR at the Pt(hkl) surfaces
lyst becomes a much more complicated task because of this
Interest in the role of the local symmetry of platinum sur-
extra variable. In the next two sections we focus on the inhibi-
face atoms in electrocatalysis has risen steadily over the last
tive effects of trace levels of Clad and Cu2+ on the kinetics of
three decades. Ever since the pioneering work of Will [25],
the ORR. Considering that high-surface area catalysts are of-
one of the major themes in electrochemical research has been
ten synthesized from halide-containing educts [7-12] and that
the study of the relationship between the electrochemical re-
trace levels of chloride may also be present in the water con-
activity and surface structure, a functionality generally
tained in the feed-stream, the effect of Cl± is of particular
termed ªstructural sensitivityº. The recent advances in the in-
practical importance. Trace levels of copper may also be pres-
situ methods of surface x-ray scattering (SXS) [26,27] have
ent in the real system, mostly as a corrosion product, thus the
made it possible to establish the true relationship between the
effects of underpotentially deposited copper (Cuupd) on
kinetics of electrochemical reactions and surface structure of
Pt(hkl) will serve as a model system to understand why a
platinum single crystal surfaces [4]. As we demonstrate be-
small amount of copper has a devastatingly inhibiting effect
low, the crystallite size effect for the ORR reported for sup-
on the ORR. The last two sections are devoted to the en-
ported Pt catalysts in sulfuric acid solution may be explained
hanced kinetics of the ORR on Pt-Ni and Pt-Co bulk alloys
by applying the single crystal results to a classical model of
and on platinum single crystals modified by a thin Pd metal
the variation in particle shape with size. The presentation
film. We demonstrate that the ability to make a controlled
here will be restricted to a relatively brief overview of the
and well characterized arrangement of two elements in the
science; further details, including experimental procedures,
electrode surface region presage a new era of advances in the
can be found in the references cited.
ORR electrocatalysis.

3.1 Reaction pathway

2 Experimental Oxygen reduction reaction is a multielectron reaction that
All measurements were carried out by making use of the may include a number of elementary steps involving differ-
rotating ring-disk electrode (RRDE) technique. The Pt single ent reaction intermediates. Of various reaction schemes pro-
crystals and polycrystalline Pt3M alloy (M = Ni, Co) elec- posed for the ORR [6], a modified Wroblowa et al., [28]
trodes were prepared either by the flame annealing technique scheme appears to be the most effective one to describe the
[13,14] or in ultra-high vacuum by sputtering/annealing cy- complicated reaction pathway by which O2 is reduced at met-
cles, respectively. After preparation, the electrodes were al surfaces:
transferred into the disk position of an insertable ring-disk ar- k1
bor as described in ref. [15]. Before the transfer, the Pt3M al- k2 k3
loys were characterized in UHV by Auger electron spectros- O2 O2,ad H2O2,ad H2O
copy (AES) and Low Energy Ion Scattering (LEIS). Details k4 k5
about the UHV preparation and characterization of Pt alloy H2O2
crystals can be found in ref. [16,17]. LEIS revealed that the
surface composition of the Pt3M alloy electrodes equaled the
bulk composition, i.e., the surface consists of 75 % Pt and Based on this reaction scheme, O2 can be electrochemically
25 % of the alloying component [17]. Details about the prep- reduced either directly to water with the rate constant k1
aration of the Pd-modified Pt(hkl) electrodes can be found in without intermediate formation of H2O2,ad (so-called ªdirectº
ref. [18]. 4e± reduction) or to H2O2,ad with the rate constant k2 (ªseriesº
For measurements with the high-surface area catalysts, we 2e± reduction). The adsorbed peroxide can be electrochemi-
used the thin-film RDE method [19]. For details about the cally reduced to water with the rate constant k3 (ªseriesº 4e±
preparation of the thin-film electrode, the reader is referred pathway), catalytically (chemically) decomposed on the elec-

106 FUEL CELLS 2001, 1, No. 2

Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

trode surface (k4) or desorbed into the bulk of the solution becomes a constant factor. As a consequence, the structure sen-

REVIEW
(k5). Although a number of important problems pertaining to sitivity in the solution containing bisulfate and halide anions
the interpretation of the reaction pathway for the ORR on is completely determined by the availability of free Pt sites, the
Pt(hkl) have yet to be resolved, recent studies presented from (1±Had) coverage dependent term.
our laboratory [21,29,30] suggest that a ªseriesº pathway via For the purpose of demonstrating the importance of the
an (H2O2)ad intermediate may be operative on Pt and Pt bi- (1 ± Had) term in the kinetics of the ORR on Pt(hkl), two
metallic catalysts. This can be considered as a special case of representative sets of polarization curves are shown in
the general mechanism where k1 is essentially zero, i.e., there Figure 1. Figure 1a shows that the ORR in H2SO4 increases in
is no splitting of the O-O bond before a peroxide species is the sequence Pt(111) < Pt(100) < Pt(110) [37]. An exception-
formed. Peroxide, on the other hand may (k5=0) or may not ally large deactivation is observed at the Pt(111) surface. This
(k56ˆ0) be further reduced to water. In either case, the rate de- is probably due the strong adsorption of the (bi)sulfate anion
termining step appears to be the addition of the first electron from the symmetry match between the fcc (111) face (see in-
to O2,ad. The rate expression is then [2,30-32], sert of Figure 1) and the C3v geometry of the oxygen atoms of
the sulfate anion [38]. The fact that the activity of all three
i = nFKcO2(1±Had)xexp(±bFE/RT)exp(±crHad)/RT (1) low-index platinum planes is significantly higher in HClO4
[4] does, however, suggest that adsorption of (bi)sulfate an-
where n is the number of electrons, K is the chemical rate ions effects the kinetics of the ORR on all three surfaces, but
constant, cO2 is the concentration of O2 in the solution, Had is particularly strong on Pt(111). It should be noted that al-
the total surface coverage by all adsorbed species, x is either 1 though bisulfate adsorption onto Pt(hkl) surfaces inhibits the
or 2 depending on the site requirements of the adsorbates, i is reduction of molecular O2, most likely by blocking the initial
the observed current, E is the applied potential, b and c are adsorption of O2, it does not affect the pathway of the reac-
the symmetry factors (assumed to be ‘), and rHad is param- tion, since no H2O2 is detected on the ring electrode for any
eter characterizing the rate of change of the apparent stan-
dard free energy of adsorption with the surface coverage by
100
adsorbing species. Assuming that the coverage of ORR inter-
I [µA]

mediates is small under reaction conditions [31], for most 50

cases, only two adsorbed species need to be considered, OHad
and specifically adsorbing anions like (bi)sulfate or halides. 0
In the following sections, we will use this reaction pathway (a) 0.05 M H2SO4, 50 mV/s, 900 rpm
iD [mAcm ]
-2

and rate expression to analyze the effects of various factors -2

on the kinetics of the ORR on Pt(hkl) surfaces.

-4
3.2 Structure sensitivity of the ORR and implications for crystallite
0.0 0.2 0.4 0.6 0.8 1.0
size effects E [VRHE]
It is now well established that the reaction rates of the ORR
on Pt(hkl) surfaces are a structure sensitive [33,34] which arise
due to structure sensitive adsorption of spectator species, such
as Hupd [15,34], OHad [35], HSO4(ad) [15], Clad [21], and Brad [30].
Pt(100) Pt(110)
Within the limited scope of this report, it will not be possible to Pt(111)
review all of these results. Rather we will show, using represen- 15
tative examples, the kind of information that can be used to im-
I [µA]

10
prove our understandings of the role of the local symmetry of
5
platinum surface atoms in the O2 reduction electrocatalysis.
0
There are two general observations concerning the structure
iD [mAcm-2]

sensitivity of the ORR: (i) the same activation energy in both acid (b) 0.1 M KOH, 50 mV/s, 900 rpm
(at the reversible potential ca. 42 kJ/mol [29]) and alkaline solu- -2
tion (at 0.9 V ca. 40 kJ/mol [36]) has been found for all three
Pt(hkl) surfaces; (ii) the structural sensitivity is most pro-
-4
nounced in electrolytes in which there is strong adsorption of
0.0 0.2 0.4 0.6 0.8 1.0
anions. At the Pt(111)-HSO4(ad) [15], Pt(hkl)-Clad [21], and E [VRHE]
Pt(hkl)-Brad [30] interfaces a single Tafel slope of ca. 120 mV/
dec is deduced from the kinetic analyses of the results for ORR Fig. 1 (a) Oxygen Reduction on Pt(hkl) in 0.05 M H2SO4 (50 mV/s,
900 rpm, 298 K), lower part, disk currents; upper part, ring currents. (b)
on Pt(hkl). This appears to be consistent support that the stan- Oxygen Reduction on Pt(hkl) in 0.1 M KOH (50 mV/s, 900 rpm, 298 K),
dard free energies of adsorption of the reaction intermediates lower part, disk currents; upper part, ring currents. Insert: Surface struc-
are not affected by the adjacent anions and thus the term rHad ture of Pt(111), Pt(100), and Pt(110), respectively.

FUEL CELLS 2001, 1, No. 2 107

 Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

of the surfaces in the kinetically controlled potential region figuration they are used in electro catalytic cells. Neverthe-
REVIEW

(Fig. 1a). less, to begin pursuing any of highly structured nanoparticule

The first correlation between the kinetics of the ORR and concepts, well-characterized surfaces should be used as the
the surface coverage of chemisorbed oxygen-containing spe- most adequate model for real catalysts. Furthermore, the ob-
cies, historically termed ªoxideº, was established on a poly- servations of structure sensitivity of the ORR are not as useful
crystalline platinum electrode. In this study Tarasevich [31] at present as they might be because the experimental condi-
proposed that both transition in the Tafel slope and the ad- tions, e.g., temperature and electrolyte, used in most ORR ki-
sorption of O2 is controlled by the surface coverage of chemi- netic measurements with supported catalysts are not those
sorbed oxygen-containing species. Very recently, by eliminat- employed in the RRDE measurements with single crystals.
ing the surface heterogeneity of a polycrystalline Pt electrode There is only one out of the many particle size effect studies
with single crystal electrodes, Markovic et al., [4] gained new that employed exactly the same conditions as those used here
insight into the relationships between the surface coverage of for single crystals: the study by Peukert et al. [39], who re-
platinum ªoxidesº and the kinetics of the ORR. We note that ported results in dilute sulfuric acid at ambient temperature.
single crystal work clearly showed that two different types of Fortunately, this was a very carefully done study using well-
chemisorbed oxygen-containing species occur on platinum, a characterized supported Pt catalysts with (average) crystallite
reversible form (denoted hereafter as OHad) and irreversible sizes ranging from 11.7 nm down to 0.78 nm. As shown by
form, which we shall call ªoxideº hereafter. The most direct Kinoshita [6], the crystallite size dependence in the data of
probe of the effects of OHad on the ORR kinetics can be ob- Peukart et al., are essentially perfectly fit by a model which as-
tained in alkaline solution, since no other anions are co-ad- sumes: i) that only the (100) face is active for the ORR; and
sorbed with hydroxyl species. Figure 1b shows that in the po- ii) that the Pt particles are perfect cubooctohedra. Below
tential range where O2 reduction is under combined kinetic- about 6 nm, the fraction of the surface having (100) facets
diffusion control (E > 0.75 V), the order of activity of Pt(hkl) drops rapidly, and actually goes to zero below 1.8 nm; parti-
in 0.1M KOH increase in the sequence (100) < (110) < (111) cles below 2 nm are formed almost entirely from (111) facets
[35]. As already discussed in [35], the structure-sensitivity of and their intersections (edges and corners) [6]. Our ORR data
the ORR in this potential region arise due to the structure sen- from single crystal Pt in sulfuric acid fully support this mod-
sitive adsorption of the hydroxyl species. It is interesting that el, since we have observed that the (100) face is more active
although the reversible adsorption of hydroxyl ions (OHrv) by two orders of magnitude versus the (111) face (in the mod-
on Pt(hkl) suppresses the kinetics of the ORR, it does not af- el calculations, a factor of 100 between the two faces is suffi-
fect the pathway of the reaction since peroxide is not detected cient to fit the data [6]). By inference then, the crystallite size
on the ring electrode for any of the surfaces in the potential effect for the ORR on supported Pt in sulfuric acid at ambient
range where the reversible form of OHad is present on the temperature would be due to the structure sensitive adsorp-
surface. In contrast, irreversibly adsorbed ªoxideº does tion of (bi)sulfate anions, this adsorption effectively eliminat-
change the reaction pathway, i.e., peroxide is detected at the ing the contribution of the (111) facets to the overall rate.
ring when there is a significant amount of this form on the Kinoshita has also shown that ORR data for supported cat-
surface, as seen in the negative going sweeps on (100) and alysts in hot, concentrated H3PO4 (180 C, 97±98 % acid) re-
(110) [35]. Note that the ORR is more strongly inhibited by ported in three different studies were also fit by this model
OHad which is adsorbed on the Pt(100) surface than on two [6]. Since the physical basis for the crystallite size effect in sul-
other single crystal surfaces, Figure 1b. This is possibly re- furic acid is anion adsorption, it would be a reasonable to
lated to a high affinity of the (100) sites (insert of Figure 1) for suggest that the same physical basis applies to this size effect,
the hydroxyl adsorption and the lack of active centers for the i.e., structure sensitive anion adsorption. There are, nonethe-
O2 adsorption on highly covered Pt(100) with OHad. Al- less, indications that this is the case. Anion adsorption in di-
though at the same overpotentials the apparent surface cover- lute phosphoric has very similar structure sensitivity as sul-
age by adsorbed OHad on Pt(110) is similar [35], if not slightly fate adsorption [33]. Sattler and Ross [40] reported a similar
higher, than on Pt(100), the activity of Pt(110) is higher than crystallite size dependence of the ORR on supported Pt in di-
the activity of Pt(100). If on the fcc (110) surface (insert of Fig- lute phosphoric acid at ambient temperature as that found in
ure 1) the adsorption of hydroxyl ion is predominantly in the hot, concentrated acid with the same catalysts. But it is un-
ªtroughº positions, then the top atoms may serve as active clear whether similar adsorption chemistry would exist in the
centers for the ORR even when the Pt(110) surface is a ªfullyº extreme conditions of hot, concentrated phosphoric acid.
covered with OHad [35]. With differences in activity between crystal faces of only a
The structure sensitivity of the ORR described above may factor of five or less for oxygen reduction in perchloric acid
provide some new insight into the particle size effects re- [41], the particle models of Kinoshita indicate that there
ported for the ORR using supported Pt catalysts. We note would little or no crystallite size effect for the ORR in a non-
however, that there is no any single system that will model adsorbing acid, such as, e.g., trifluoromethane sulfonic acid or
all the aspects of real catalysts; particularly in the exact con- Nafion.

108 FUEL CELLS 2001, 1, No. 2

Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

3.3 Effects of Clad ring currents were a small fraction of the disk currents, im-

REVIEW
plying that on the surface highly covered with Clad oxygen
3.3.1 Pt(hkl)
reduction proceeds almost entirely through the 4e± reduction
The family of polarization curves for the ORR on Pt(hkl) in pathway. The appearance of peroxide oxidation currents on
0.05 M H2SO4 + 10±3 M Cl± are shown in Figure 2. The polar- the ring electrode begins at potentials negative of 0.25 V, and
ization curves for the ORR in pure 0.05 M H2SO4 solutions is parallels with the adsorption of hydrogen on Pt(111).
shown for one rotation rate as a reference. A full description In contrast to Pt(111), the ORR on Pt(110) (Figure 2b) and
of the polarization curves for the ORR on Pt(111) in pure sul- Pt(100) (Figure 2c) in Cl-containing solution is accompanied,
furic acid solution is given in our previous publications [15]. but not quantitatively mirrored, with peroxide formation, im-
In the presence of Cl±, the variation in activity increases in the plying that within the same potential limits H2O2 is formed
order (100) < (110) < (111) [21]. Figure 2a shows that the ORR as an intermediate on the Pt(110) and Pt(100) electrodes cov-
on the Pt(111) surface in 0.05 M H2SO4 +10±3 M Cl± is strongly ered with Clad. On Pt(100) the fraction of H2O2 produced be-
inhibited, with an activity in a solution containing Cl± being tween 0.25 < E < 0.6V during the ORR, calculated from Fig-
several orders of magnitude lower than in 0.05 M H2SO4 ure 2c using equation 2 [42],
[21,33]. Figure 2a also shows that between 0.3 < E < 0.9 V, the
2Ir=M
xH2 O2 ˆ (2)
Id ‡Ir=M
a)
is almost constant and is close to 25% (note in this relation, N
50
denotes the collection efficiency of the ring-disk configura-
IR /µA

tion, Id and Ir are the disk and ring currents, respectively). At

0
potentials more negative than E = 0.3 V the surface coverage
with Clad is reduced substantially, but the O2 reduction cur-
-2
rents sharply decrease to diffusion limiting currents for ca.
iD /mAcm-2

-4 2e± per O2 molecule at the negative potential limit. In the

Pt(111) same potential region the O2 reduction currents on the disk
-6 electrode are accompanied quantitatively by the H2O2 oxida-
tion currents on the ring electrode, reaching maximum perox-
ide production of ca. 90% at 0.075 V. In Cl±-free solution,
however, a 3e± reduction is observed at the same potential
IR /µA; H2O2 /%

b)
2 limit [15]. This may suggest that Clad is always present on
0 Pt(100) covered by Hupd, thus the availability of Pt atoms re-
quired for the breaking of the O-O bond are significantly re-
-2 duced and the ORR proceeds entirely through the 2e± perox-
ide pathway on the Pt(100)-Hupd-Clad surface. The amount of
iD /mAcm-2

-4
peroxide formed on the Pt(110) surface is substantially smal-
Pt(110) ler, Figure 2b, ca. 2.5 %, and no increase in peroxide formation
-6
was observed in the Hupd region.
As we have demonstrated recently [21], the key to resol-
0.05M H2SO4 + 10-3M Cl-
ving the structural sensitivity of the ORR on the Pt(hkl) sur-
IR /µA; H2O2 /%

0.05M H2SO4 c) faces in Cl± containing solution is to be found in understand-

50
ing the relationships between the strength of the Pt-Clad
interaction, the specific type of sites for Clad, and the effect of
0 Clad at these sites on the rate of O-O bond breaking and/or
bond making. In particular, due to the strong interaction of
-2
Clad with Pt(100) the ORR between 0.3 V and 0.75 V is inhib-
iD /mAcm-2

-4 ited and the number of platinum sites required for adsorption

Pt(100) of O2 and the breaking of O-O bond are reduced, which is in
-6
accord with the peroxide formation in the same potential re-
gion. In the case of Pt(110), although the interaction of Clad
0.2 0.4 0.6 0.8
with Pt(110) is as strong as with (100) sites, Clad is most likely
E [VRHE]
adsorbed in the ªtroughsº of the surface (insert of Figure 1),
Fig. 2 Ring and Disk currents for the ORR in 0.05 M H2SO4 + 10±3 M Cl± thus leaving the top sites available for O2 adsorption and the
on (a) Pt(111) (b) Pt(110), and (c) Pt(100) at different rotation rates (full cleavage of the O-O bond. Apparently, the Pt(111)-Clad sys-
lines). The respective polarization curves on Pt(hkl) in 0.05 H2SO4 for
2500 rpm are shown as a reference (dotted line). The gray circles rep-
tem has the same effect on the ORR as bisulfate anions, i.e., al-
resents the fraction of peroxide formed during the ORR according to though Clad inhibits the initial adsorption of O2 molecules, it
eq. 2. 50 mV/s, 293 K. does not affect the pathway of the reaction, Figure 2a.

FUEL CELLS 2001, 1, No. 2 109

 Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

3.3.2 Supported catalysts crystal faces and edges and corner sites with probable (110)
REVIEW

± symmetry. At the (111) facets Clad acts exclusively as a site-

In developing the basis for discussing the effects of Cl in
blocking species without enhancing the formation of H2O2. In
real catalysts systems, we start by summarizing the Tafel
addition to blocking the adsorption of O2, the (100) facets and
plots for the ORR on a Pt/carbon fuel cell catalyst [24]. Figure
(110) corner/edge sites are effecting the breaking of the O-O
3a shows the current versus potential curve for the ORR on
bond and, hence, the production of peroxide on a supported Pt
Pt/Vulcan in 0.5 M HClO4 with and without chloride. In ac-
catalysts, ca. 15%, is close to the average value of H2O2 produced
cord with the ORR on Pt(hkl), the onset for the ORR is shifted
on Pt(100) (ca. 25%) and Pt(110) (ca. 2.5%) single crystal surfaces.
to higher overpotentials with increasing chloride concentra-
tion in the solution. Compared to the ORR in the absence of
3.3.3 Implications for PEM fuel cells
Cl± the activity at constant potential decreases roughly one or-
der of magnitude for each order of magnitude increase in As shown in the aforementioned sections, even trace
chloride anion concentration. In addition, considerable H2O2 amounts of chloride (i.e., 10±4 M Cl± or ca. 4ppm) drastically
currents on the ring electrode (Figure 3b) are observed in solu- change both the activity and the reaction pathway of the ORR
tions containing Cl±, e.g., with increasing Cl± concentration, the on Pt catalysts. Although these results were obtained in liq-
fraction of peroxide is increasing up to 10 % at cCl± = 10-3 M and uid electrolyte, similar effects can be expected on electrodes
14 % at cCl± = 10±2 M around 0.6 V, respectively. As for the ki- in a solid-polymer electrolyte as present in a PEMFC. In this
netics of the ORR on Pt(hkl) in the preceding section, Clad has case, chloride impurities on the level of a couple of ppm,
two effects in the kinetics of the ORR on a supported platinum either due to incorporation into the MEA during preparation
catalyst: (i) Clad acts as a site-blocking species which is reducing or due to contamination of the humidified fuel cell feed
the number of active sites for the adsorption of O2 molecules streams, will strongly affect the MEA performance. Based on
and (ii) Clad is affecting the number of sites required for the the data in Figure 3, chloride contamination on the order of
breaking of the O-O bond. Therefore, the kinetic activity and the 4ppm would result in a fuel cell voltage loss of ca. 50 mV,
reaction pathway for the ORR on supported catalysts in the which should equally effect the open circuit cell voltage.
presence of Clad is fully explained by applying the single crystal These results indicate in general the high purity conditions
results to the relationship between particle size and the different required for MEA preparation and the humidified feed-
surface sites (crystal faces, edges and corners) [6]. Based on a streams for PEMFC's. In addition to the kinetic effect of Cl±
HRTEM analysis, the structure of typical Pt particle (mean par- adsorbed on the supported metal catalyst particles, its en-
ticle size ca. 3.7  1 nm [19]) illustrated in Figure 3 show clearly hancement of the H2O2 production will be deleterious to the
that the cubo-octahedral Pt nanoparticles close to the edge of stability of perfluorinated membranes and ionomers in the
the carbon support is facetted, exposing mostly (111) and (100) catalyst layers (e.g., Nafion membranes and ionomers), which
are known to degrade in the presence of
peroxide radicals. This does not only ap-
(111)
ply to the oxygen electrode, but also to
the anode side which is exposed to oxy-
gen either via O2 crossover through thin
(100) state-of-the-art membranes (e.g., GoreSelect
or Nafion 112) or from air-bleed oxygen
used for reformate operation (note when
20 using air-bleed oxygen is present in a
(b) drastically over-stoichiometric concentra-
0.90
16 tion).
0.85
E/V vs. RHE

12
xH O /%

0.80 3.4 Effects of ªAir Bleedº

2

Previous work by Gottesfeld et al., [43]

2

0.75 8
0.70 has shown that bleeding of 0.4-2% oxygen
4 into the CO-contaminated hydrogen can
0.65 completely correct for deleterious effects
(a)
0.60 0 of 5±100 ppm CO with a conventional Pt
0.1 1 10 100 0.0 0.2 0.4 0.6 0.8 catalysts (0.4 mg/cm2) in PEMFC. How-
ik/mAcm-2 E/V vs. RHE ever many important questions regarding
Fig. 3 (a) Tafel plots for the ORR on Pt/Vulcan in 0.5 M HClO4 (white circles) and after addition the air bleeding function are still unan-
of 10±4 M Cl± (black diamonds), 10±3 M Cl± (gray triangles), and 10±2 M Cl± (gray squares). (b) swered. In order to get some new insights
Fraction of peroxide formed during the ORR in the presence of chloride. Notation as in (a). Top pa- about the cleansing the anode catalysts by
nel: High resolution TEM micrograph of a representative Pt particle and the proposed cubooctahe- oxygen bleeding, the ORR was studied on
dral shape.

110 FUEL CELLS 2001, 1, No. 2

Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

were found for our unpublished Pt-Ru-CO and Pt-Sn-CO sys-

REVIEW
b) tems and a Pt/Vulcan high surface area catalyst, it is not pos-
80
sible unambiguously to propose a reaction mechanism for the
Pt / COads-O2 ªcleansingº of CO adsorbed on these metallic surfaces, e.g., it
60 0.05M H2SO4 may involve both chemical and electrochemical reactions.
H2O2 [%]

@ 50 mV

40
4 ORR on bimetallic surfaces
20 In this section, we review recent progress in the funda-
mental science of the electrocatalysis of the ORR on Pt bime-
c) tallic surfaces. Bimetallic electrodes can be prepared in a
0 80
number of ways. In this section, the focus will be on bimetal-
i [mA/cm2]

CO stripping

lic platinum single crystal surfaces that are created by: the un-
after Air Bleed
derpotential deposition (UPD) method, the classical metallur-
iD [mAcm ]

-1
-2

40 gical method and deposition of pseudomorphic metal films.

The UPD adlayer is forming stable bimetallic surfaces only in
the presence of the corresponding soluble cations. For the
-2
0 purpose of this review, the effects of Cuupd on the ORR will
0.2 0.4 0.6 0.8 1.0 1.2 briefly be discussed. The rate of the ORR will also be summar-
-3
E [V/RHE] ized on metallurgically prepared Pt-Co and Pt-Ni alloy sur-
faces. These results will, then, be compared with the kinetics on
a)
the supported Pt-Co catalysts. Because pseudomorphic Pd films
0 100 200 300
supported on platinum single crystal surfaces has recently
τ [s] received considerable attention in surface electrochemistry, the
ORR on the Pt(hkl)-Pd will be summarized as a model system.
Fig. 4 (a) Potentiostatic oxygen reduction on a polycrystalline Pt electrode
covered by 1 ML of COad at 0.05 V. (b) Fraction of peroxide formed dur-
ing the potentiostatic ORR on the CO-covered Pt surface. (0.05 M H2SO4,
4.1 Pt(hkl)- Cuupd system
0.05 V, 1600 rpm, 293 K). (c) Stripping of a COad ML on polycrystalline
Pt (dashed line) and of the remaining COad after O2 exposure in (a) (solid The effects of Cuupd on the ORR at the Pt(100) interface in
line). The Pt base CV is shown as a reference (dotted line).
0.05 M H2SO4 and 0.05 M H2SO4 +10±3 M Cl± are shown in
a CO covered polycrystalline electrode. In these experiments, Figure 5. For comparison, the polarization curves for the ORR
CO is adsorbed by holding the potential at 0.05 V in CO-satu- on Pt(100) in Cu2+-free solutions under the same experimen-
rated solution for 5 min followed by purging the solution tal conditions are shown as dotted curves. Clearly, in a solu-
with Ar to completely remove CO from solution. The transi- tion containing Cu2+ ions the ORR is strongly inhibited in
ent current for the ORR on CO-covered Pt was recorded at both Cl±-free (Figure 5b) and Cl±-containing solutions (Figure
0.05 V, Figure 4a. The amount of peroxide produced on the 5e). Very similar behavior was also observed on a polycrystal-
disk electrode was monitored at the ring electrode (figure 4b). line electrode [44,45] and Pt(111) [17,46] in different acid solu-
Figure 4 shows that on CO-covered surface a significant tions. Given that the kinetics of Cu UPD is a rather slow pro-
amount of peroxide is produced a short time after the intro- cess, the ORR currents at the Cuupd-modified surfaces are
duction of O2 (no CO present in solution!), then the produc- also measured under steady state conditions, see the closed
tion of peroxide drops away to a small but constant rate of ca circles in Figure 5b and e. In these experiments, the ORR is
20 %. We note that under the same experimental condition more inhibited than in the potentiodynamic experiments. As
only ca. 3±5% of peroxide was produced on the CO-free plati- shown in Figure 5, the deactivation is more pronounced in
num surface. This enhanced production of peroxide on the the presence of Cl± ions, presumably due to stronger interac-
Pt-COad surface suggests that O2 is not capable to completely tion of specifically adsorbed chloride (Clad) than bisulfate
oxidize CO, and consequently the remaining CO can effec- (H2SO4(ad)) anions with the Pt-Cuupd surface. Note that at the
tively block the pair of sites required for the peroxide decom- Pt-Cuupd-Clad interfaces the inhibition of the ORR is observed
position. To confirm this, O2 was replaced by Ar and the at relatively small surface coverages by Cuupd, for compari-
stripping voltammetry of CO was recorded, see Figure 4c. son see polarization curves and corresponding potential de-
Unexpectedly, we found that CO was indeed still present on pendent surface coverage by Cuupd (HCu,upd), Fig. 5a and 5d.
the surface since CO stripping peak was clearly resolved at ca Recently, we suggested that the deactivation at Cuupd-modi-
0.8 V. The coverage of CO was ca. 40% of one corresponding fied platinum single crystal surfaces arise due primarily to an
for a surface fully covered by CO but never exposed to O2 (in- enhanced anion adsorption on platinum atoms adjacent to
sert of Figure 4). The fact that COad cannot be completely oxi- Cuupd atoms. This mechanism explains the anomalously large
dized by O2 is a surprising result. Although the same effects inhibition of the ORR and the HOR by a very small amount

FUEL CELLS 2001, 1, No. 2 111

 Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces
REVIEW

1
a) 1.0 d)

Pt(100)
Pt(100)

ΘCu
ΘCu

2
100 mA/cm
-2
50 mAcm

0
0.05M H2SO4 + 8⋅10 Cu 0.05M H2SO4 + 8⋅10 M Cu + 10 M Cl
-5 2+ -5 2+ -3 -

20 -3 -
0.05M H2SO4 0.05M H2SO4+ 10 M Cl
IR[µA]

IR[µA]
20
10
c) f)
0 0
0

0.0
iD[mAcm-2]
iD[mAcm-2]

-2

-4
-0.2
b) e)
-6
0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0
E [VRHE]
E [VRHE]

Fig. 5 (a) CV of Pt(100) in 0.05 M H2SO4 + 8´10±5 M Cu2+ and the Cuupd adsorption isotherm (gray circles) (b) ORR on Pt(100) in
0.05 M H2SO4 + 8´10±5 M Cu2+ at 50 mV/s (solid line) and under potentiostatic conditions (gray circles). The ORR on Pt(100) in pure 0.05 M H2SO4
(dotted line) is shown for reference (c) Ring currents for peroxide detection, same notation as in (b). (d) CV of Pt(100) in 0.05 M H2SO4 + 8´10±5 M
Cu2+ + 10±3 M Cl± and the Cuupd adsorption isotherm (gray circles) (e) ORR on Pt(100) in 0.05 M H2SO4 + 8´10±5 M Cu2+ + 10±3 M Cl± at 50 mV/s
(solid line) and under potentiostatic conditions (gray circles). The ORR on Pt(100) in pure 0.05 M H2SO4 for 2500 rpm (dotted line) is shown for ref-
erence. (f) Ring currents for peroxide detection, same notation as in (e). 50 mV/s, 293 K.

of Cuupd. It is also interesting that under the negative going that five binary Pt-M alloy electrocatalysts (M = Cr, Mn, Co,
sweep direction the ring current is a very small fraction of the and Ni) supported on carbon produced some enhancement in
disk current, i.e., is close to 3%, suggesting that a 4e± reaction the kinetics (factor of 3-5) of the ORR relative to ªstandardº
pathway is still operative at Cuupd-modified platinum single supported Pt catalyst. One of the difficulties in determining
crystals. The physical model that appears to rationalize this the effect of alloying components using supported catalysts is
result is one in which the active sites for adsorption of molec- that the activity of a pure Pt supported catalysts can have a
ular O2 are a small number of platinum islands created each wide range of values depending on its microstructure and/or
time desorption of anions takes place from Cuupd-unmodified method of preparation. The intrinsic activity of Pt for the
platinum sites [22]. ORR depends on both particle shape and size [6,41], i.e., there
is not a single value of the specific activity even when nor-
malized by Pt surface area. Since the alloyed Pt catalysts par-
4.2 Pt-Co and Pt-Ni bimetallic surfaces
ticles may not have either the same particle size or shape as
Several investigations have been carried out to determine the Pt catalysts to which they are compared, a simple compar-
the role of alloying in the electrocatalytic activity of Pt for the ison of activity normalized either by mass or surface area is
ORR (detailed review in ref. [5]). A definitive determination, insufficient to identify a true alloying effect. A more detailed
however, remains elusive. Luczak et al., for example, found discussion of this point, specifically in the case of Pt-Co cata-
an increase in mass activity of a factor 1.5 to 2.5 when using lysts, can be found in ref. [52]. These complexities of support-
PtCoCr or PtCr instead of pure Pt [9,47,48] whereas Beard et ed catalysts reinforced the need for using well-characterized
al. [49] and Glass et al. [50] found no change for PtCo and materials to identify the fundamental mechanisms at work in
PtCr, respectively. Furthermore, Mukerjee et al., [51] showed electrocatalysis. As we demonstrate below, significant im-

112 FUEL CELLS 2001, 1, No. 2

Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

provement of the ORR catalyses on Pt based alloy systems lyses implies that the reaction mechanism on Pt3Ni and Pt3Co

REVIEW
will require the inhibition of Pt-OH formation beyond 0.8 V. alloy surfaces is the same as one proposed for pure Pt, i.e., a
ªseriesº 4e± reduction. Because the (chemical) rate constant K
and the two exponential terms are essentially identical for all
4.2.1 Bulk alloy surfaces
three systems the difference in the ORR kinetics must lie in
Most recently, the intrinsic catalytic activity of Pt3Ni and the pre-exponential term, the (1 ± Had) term in equation 1. As
Pt3Co alloy catalysts was studied in our laboratory with mod- demonstrated for Pt(hkl) [4], in perchloric acid electrolyte this
el bulk alloys characterized in UHV [17,53]. Figure 6 sum- term is primarily determined with the potential dependent
marizes results for the ORR on UHV prepared and character- surface coverage of the OHad species. As one can see from
ized bulk Pt3Ni and Pt3Co alloy catalysts in 0.1 M HClO4 Figure 6, the catalytic effect of this term is ca.1.5-2; consistent
(note, on both samples the surface composition equals the with the shift of ca. 50 mV in oxidation potential. It is not
bulk composition, as deduced by LEIS [17]). Pure polycrystal- clear why Ni and Co in the Pt surface would cause the shit in
line Pt is also shown as a reference. As shown in Figure 6, the the oxidation potential of the Pt surface. One plausible expla-
activity at 333 K increases in the order Pt3Co > Pt3Ni > Pt, nation is that the adsorption of OHad on the Pt sites is modi-
the catalytic activity of Pt3Co alloy being ca. 25mV improved fied by oxygenated species that are adsorbed on the alloying
with respect to the ORR on pure Pt. It should also be noted components. As for Pt-O interaction in UHV, the isosteric
that similar small amounts of peroxide are detected on the heat of adsorption of OHad on Pt sites (the surface coverage)
ring electrode on all three surfaces during the ORR in per- surrounded by ªoxideº covered Ni and Co atoms may signifi-
chloric acid solution. Interestingly, the Tafel slope (90 ± cantly be reduced primarily from lateral repulsive interac-
110 mV/dec) and the activation energy (20 to 25 kJ/mol) dis- tions, and to much lesser extent the occupation of different
cerned for pure Pt is almost the same with values obtained adsorption sites. Therefore, being oxidized at much lower po-
for Pt3Ni and Pt3Co bimetallic surfaces. The fact that essen- tentials then Pt, it appears that Ni and Co atoms may serve as
tially the same kinetic parameters are assessed from the ana- ªsacrificialº elements in Pt-Ni and Pt-Co alloys. Theoretical
aspects of the relationships between the electronic structure
and reactivity are, however, required in order to understand
the trends in atomic/molecular chemisorption energies of
oxygen containing species on Pt-Ni and Pt-Co bimetallic sur-
Pt3Co 0.1M HClO4 faces.
IR [µA]

Pt3Ni 293 K; 1600rpm

20
Pt-poly
4.2.2 Supported catalysts
In this section, we review some of experimental results for
the ORR obtained on supported platinum and Pt alloy cata-
0
lysts. The presentation here is restricted to a relatively brief
overview, further details, including experimental procedure,
0.92 can be found in references cited. As recently shown by Paulus
et al., [23], base voltammetry (Figure 7a) and complementary
0.88 TEM analysis allow the determination of the particle size dis-
E [VRHE]

Pt3Co
-2
Pt-poly
tribution and the estimation of the surface composition. These
analyses can provide an adequate, although not necessarily
iD [mAcm ]

0.84
-2

complete characterization of the microstructure of the catalyst

that we could use to correlate the surface activity of Pt-Co
0.80 b)
Pt3Ni nanoparticles with the model bulk alloy surfaces shown in
-4
1 10
the preceding section. A low magnification of TEM image of
ik [mAcm-2]

Tab. 1 Characterization of Pt/Vulcan and PtM/Vulcan catalysts.

Catalyst atomic de- surface Charge Active (Pt) Estimated
a) nsity per atomic Hupd- surface Surface
-6 electrode density region atoms composi-
[nmol/cm2] [nmol/cm2] [mC/cm2] [nmol/cm2] tion
0.2 0.4 0.6 0.8 1.0
Pt/Vulcan 72.4 21.7 1.97 20.4 (100%)
E [VRHE] PtCo/ 111.4 27.8 0.72 7.5 27%
Vulcan
Fig. 6: (a) ORR on polycrystalline Pt (dotted line), Pt3Ni (dashed line), and
Pt3Co (solid line) in 0.1 M HClO4 at 293 K (20 mV/s). upper panel: Ring Pt3Co/ 87.8 22.0 1.40 14.5 66%
Vulcan
currents for peroxide detection. (b) Mass-transport corrected Tafel plots
deduced from the polarization curves in (a) for Pt (black circles), Pt3Ni Pt3Ni/ 87.8 22.0 1.52 15.8 70%
(white circles) and Pt3Co (gray circles). Vulcan

FUEL CELLS 2001, 1, No. 2 113

 Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces
REVIEW

0.4
35 PtCo
20% PtCo/Vulcan (a) 0.4 0.9 V PtCo 0.85 V 2.5
4.4 ± 1.6 nm
30
0.2 2.0
(c) 0.3

i [Amg-1Pt]

i [Amg-1Pt]
25

distribution [%]
Pt3Ni Pt3Co
i [mAcm-2]

0.0 20 Pt3Ni Pt3Co 1.5

0.2 Pt
15 1.0
-0.2
Pt
298 K, 50 mV/s 10
0.1 M HClO4 0.1
0.5
-0.4 5

0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 12
0 5 10
E [VRHE] 0.8 V PtCo
particle size [nm] 10

i [Amg-1Pt]
30 nm Pt3Co
Pt3Ni
(d) 6

4 Pt

2
3 nm 0

Fig. 8: Comparison of the mass-transport corrected mass-specific current

densities at 0.9, 0.8%, and 0.8 V obtained on Pt/Vulcan, Pt3Ni/Vulcan,
(b) Pt3Co/Vulcan, and PtCo/Vulcan.

Fig. 7: (a) Base CV of 20 %PtCo/Vulcan (1:1) in 0.1 M HClO4 (50 mV/s, catalysts is the same as one proposed for bulk alloys, e.g., a
298 K). (b) Low Resolution TEM micrograph for the PtCo/Vulcan catalyst.
ªseriesº 4e±- reduction pathway. Further comparison of the
(c) Particle size distribution for PtCo/Vulcan. (d) High resolution TEM mi-
crograph of a single PtCo particle. kinetic parameters of the bulk and high surface area catalysts
provide a rational for the enhanced catalytic activity of sup-
Pt-Co catalyst, shown in Figure 7b, illustrates that the distri- ported PtNi and PtCo catalysts. Because both the activation
bution of metal particles on the carbon support is rather uni- energy (ca. 20±25 kJ/mol) and the Tafel slopes (ca. 90±110
form. The histogram for the particle size distribution (Fig- mV/dec) for high-surface-area catalysts are essentially identi-
ure 7c), which included analyses of several different regions cal to that observed on the bulk electrodes in Figure 6, one
of the catalysts, revealed that the average particle size is ca. can propose that the differences in the kinetics of the ORR is
4.4 nm. Finally, the x-ray emission energy filtering imaging
technique enabled us to determine that the catalysts dis-
persed on the support are completely alloyed. This was
IR [µA]

100
achieved by comparing the micrographs and the cobalt map
images (not shown). For details the reader is referred to ref- 0
erence [23,54]. The cyclic voltammogram for PtCo/Vulcan
and Pt/Vulcan is presented in Figure 7a. Figure 7 shows that -2
iD [mAcm-2]

0.1 M HClO4
the voltammetry of PtCo/Vulcan has a suppressed pseudoca- -4 2500 rpm
pacitance in the Hupd-region with respect to Pt/Vulcan elec-
trode, clearly indicating that the charge associated with Hupd -6 a)
depends on the availability of Pt surface atoms. As described
in ref. [20,23], the electrochemical active surface areas for the Pt(111)-Pd
catalysts were determined by integrating the area in the
IR [µA]

20
Pt(111)
Hupd-region. The Hupd-charges are listed in table 1 together 293 K
with the resultant number of Pt-surface atoms for all investi- 0
gated high surface area catalysts.
iD [mAcm-2]

The best form of representation of catalytic activity on sup- -2 0.1 M KOH

ported catalysts is by comparing the kinetic currents for the 1600 rpm

ORR at three different potentials in the form of histograms, -4 b)

Figure 8. These graphs clearly show that the kinetics of the
ORR on supported Pt and platinum alloy catalysts is depen-
dent on the nature of the alloying component, e.g., 0.0 0.2 0.4 0.6 0.8 1.0
Pt < PtNi < PtCo, and the surface composition Pt3Co < PtCo. E [VRHE]
As for bulk alloy surfaces in Figure 6, a small amount of per-
Figure 9: Disk and Ring currents for the ORR on Pt(111)-1 ML Pd (solid
oxide was detected on the ring electrode on all alloy surfaces, line) in (a) 0.1 M HClO4 and (b) 0.1 M KOH. The dotted line represents
implying that the reaction pathway for the ORR on supported unmodified Pt(111) under the same conditions. 50 mV/s, 293 K.

114 FUEL CELLS 2001, 1, No. 2

Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

determined with the (1±Had) term in equation 1, e.g., the re- bimetallic surfaces based on platinum. The single crystal re-

REVIEW
duced adsorption of OHad on Pt sites surrounded by ªoxideº- sults provided insight into the effects of the platinum struc-
covered Ni and Co atoms. To conclude this section, we em- ture on the kinetics of the ORR, and created a fundamental
phasize that the PtCo alloy is a promising catalyst for the link between the specific activity of Pt (rate per unit area) and
ORR for two reasons: (i) the amount of Pt in air cathode fuel particle size (for various particle shapes). The results show
cells may substantially be reduced (by ca. 25-50 %), and the that the structure-sensitive kinetics of the ORR arise primarily
activity is significantly improved with respect to the ORR on due to structure sensitive adsorption of anions of supporting
pure Pt. electrolyte. The ORR on Pt(hkl)-Clad and Pt(hkl)-Cuupd elec-
trodes is strongly inhibited. The single crystal results showed
that Clad has two effects in the kinetics of the: (i) Clad acts as a
4.3 Pt(111)-Pd systems
site-blocking species which is reducing the number of active
Figure 9 summarizes a family of polarization curves for sites for the adsorption of O2 molecules and (ii) Clad is affect-
the ORR on the Pt(111)-Pd electrode in O2-saturated 0.1 M ing the number of sites required for the breaking of the O-O
HClO4, and 0.1 M KOH solutions along with representative bond. The same mechanism of action was used to explain the
polarization curves recorded in the same solution but on un- effect of Cl± on the rate of the ORR on supported Pt catalysts.
modified Pt(111). Figure 8a shows that in 0.1 M HClO4 the The physical model that appears to rationalize a strong inhi-
ORR is inhibited at the Pt(111)-Pd surface, probably due to bitive effect of Cuuod is one in which the active sites for ad-
stronger interaction of Clad (which is present as trace impur- sorption of molecular O2 are a small number of platinum is-
ity in perchloric acid [55]) with the Pd film than with an un- lands created each time desorption of anions takes place from
modified Pt(111) electrode [18]. The decrease in the rate of the Cuupd-unmodified platinum sites.
ORR at the disk electrode is, however, not associated with The knowledge of the electrocatalysis of the ORR on model
peroxide detection on the ring electrode, indicating that pairs bimetallic surfaces on Pt-Ni and Pt-Co bulk alloys has been
of Pd sites are always available for the breaking of the O-O used to resolve the enhanced ORR kinetics on supported Pt-
bond. This implies that the direct 4e± path is operative in this Ni and Pt-Co catalysts. We proposed that significant im-
potential region. In contrast to perchloric acid solution, the provement of the ORR catalyses on Pt based alloy systems ap-
Pt(111)-Pd electrode shows a unique catalytic activity for the pears primarily due to the inhibition of Pt-OH formation on
ORR in alkaline solution. A critical assessments of this work Pt surrounded by ªoxideº-covered Ni and Co atoms beyond
shows that the rate of the ORR on Pt(111)-Pd electrode is im- 0.8 V.
proved by factor of two relative to unmodified Pt(111), Figure The kinetics of the ORR on Pt(111) modified with Pd metal
8b. Keeping in mind that the Pt(111) electrode in KOH was films is depending of anions of supporting electrolyte either
the most active catalyst for the ORR, the catalytic improve- reduced (HClO4) or catalyzed (KOH). Keeping in mind that
ment observed on the Pt(111)-Pd surface is an extremely im- the Pt(111) electrode in KOH was the most active catalyst for
portant new observation which may help in the quest for de- the ORR, the catalytic improvement observed on the Pt(111)-
veloping better ORR catalysts. The fact that the activity of Pd surface is an extremely important new observation which
Pt(111)-Pd electrode is significantly higher in KOH then in may help in the quest for developing better ORR catalysts.
HClO4 implies that the major deactivation in acid electrolytes
arise from strong interaction of specifically adsorbing anions
with the (111) surface. Therefore, the strong adsorption of an- Acknowledgments
ions may hinder the adsorption of molecular oxygen and/or
This work was supported by the U.S. Department of
reduce the likelihood of its dissociation. As a result, the ki-
Energy, the Office of Science, Basic Energy Sciences, Materials
netics of the ORR on Pt(111)-Pd relative to clean Pt(111) is
Sciences Division, and the Office of Advanced Transportation
mostly determined by the pre-exponential (1±Had) term in
Technologies, Fuel Cell Systems, under contract DE-AC03-
equation 1. This term is determined by the modification of
76SF00098.
surface electrochemical properties (the work function and
thus the potential of zero charge) caused by charge redistribu-
tion upon forming the Pt-Pd bond. More details about the
References
ORR on the Pt(111)-Pd surface, including the effects of Hupd
on the kinetics, may be found in references [18]. [1] A. Damjanovic, in Modern Aspects of Electrochemistry,
J.O.M. Bockris, B.E. Conway, Editors, p. 369, Plenum
Press, New York (1969).
5 Summary [2] M.R. Tarasevich, A. Sadkowski, E. Yeager, in Compre-
hensive Treatise in Electrochemistry, J.O.M. Bockris, B.E.
In this review we summarize selectively recent progress, Conway, E. Yeager, S.U.M. Khan, R.E. White, Editors,
primarily from our laboratory, in the development of the p. 301, Plenum Press, New York (1983).
ORR catalysis on well-defined surfaces. The focus was on [3] R.R. Adzic, in Electrocatalysis, J. Lipkowski, P.N. Ross,
two types of metallic surfaces: platinum single crystals and Editors, p. 197, Wiley-VCH, Inc., New York (1998).

FUEL CELLS 2001, 1, No. 2 115

 Markovic, Schmidt, Stamenkovic, Ross: Oxygen Reduction Reaction on Pt and Pt Bimetallic Surfaces

[4] N.M. Markovic and P.N. Ross Jr., in Interfacial Electro- [30] N.M. Markovic, H.A. Gasteiger, B.N. Grgur, P.N. Ross,
REVIEW

chemistry ± Theory, Experiments and Applications, A. J. Electroanal. Chem., 467 (1999) 157.
Wieckowski, Editor, p. 821, Marcal Dekker Inc., New [31] M.R. Tarasevich, Electrochimiya, 9 (1973) 578.
York (1999). [32] F. Uribe, M.S. Wilson, T. Springer and S. Gottesfeld, in
[5] A.J. Appleby, Catalysis Reviews, 4 (1970) 221. D. Scherson, D. Tryk, M. Daroux, X. Xing (Eds.), Pro-
[6] K. Kinoshita, Electrochemical Oxygen Technology, John ceedings of the Workshop on Structural Effetcs in Elec-
Wiley & Sons, New Vork 1992. trocatalytsis and Oxygen Electrochemistry, PV 92-11,
[7] H.G. Petrow and R.J. Allen, United States Patent 3, 992, p. 494, The Electrochemical Society, Pennington, NJ,
331, filed 1974. 1992.
[8] K. Kinoshita, K. Routsis, J.A.S. Bett, Thermochimica [33] F. EL Kadiri, R. Faure, R. Durand, J. Electroanal. Chem.,
Acta, 10 (1974) 109. 301 (1991) 177.
[9] F.J. Luczak and D.A. Landsman, United States Patent No. [34] N.M. Markovic, R.R. Adzic, B.D. Cahan, E. Yeager, J.
US 4.447.506, filed 1984. Electroanal. Chem., 377 (1994) 249.
[10] F.J. Luczak and D.A. Landsman, United States Patent No. [35] N.M. Markovic, H.A. Gasteiger, P.N. Ross, J. Phys.
4.711.829, filed 1987. Chem., 100 (1996) 6715.
[11] P. Stonehart, United States Patent No. US 5.593.934, filed [36] T.J. Schmidt, B.N. Grgur, N.M. Markovic, P.N. Ross Jr.,
1997. J. Electroanal. Chem., 500 (2001) 36.
[12] E. Auer, A. Freund, T. Lehmann, K.-A. Starz, R. [37] B.N. Grgur, N.M. Markovic, P.N. Ross Jr., Langmuir, 13
Schwarz, U. Stenke, German Patent No. DE 197 21 437 (1997) 6370.
A1, filed 1998. [38] N.M. Markovic, N.S. Marinkovic, R.R. Adzic, J. Electro-
[13] J. Clavilier, A. Rodes, K. El-Achi, M.A. Zamakhchari, J. anal. Chem., 241 (1988) 309.
Chim. Phys., 88 (1991) 1291. [39] M. Peuckert, T. Yoneda, B. Dalla, M. Boudart, J. Electro-
[14] N.M. Markovic, M. Hanson, G. McDougal, E. Yeager, J. chem. Soc., 13 (1986) 944.
Electroanal. Chem., 241 (1986) 309. [40] M. Sattler and P.N. Ross, Ultramicroscopy, 20 (1986) 21.
[15] N.M. Markovic, H.A. Gasteiger, P.N. Ross, J. Phys. [41] N.M. Markovic, H.A. Gasteiger, P.N. Ross, J. Electro-
Chem., 99 (1995) 3411. chem. Soc., 144 (1997) 1591.
[16] H.A. Gasteiger, P.N. Ross, E.J. Cairns, Surf. Sci., 293 [42] T.J. Schmidt, U.A. Paulus, H.A. Gasteiger, N. Alonso-
(1993) 67. Vante, R.J. Behm, J. Electrochem. Soc., (1999)
[17] V. Stamenkovic, T.J. Schmidt, N.M. Markovic, P.N. Ross [43] S. Gottesfeld, United States Patent No. 4,910,099, filed
Jr., J. Phys. Chem. B, (2001) 1990.
[18] V. Climent, N.M. Markovic, P.N. Ross, J. Phys. Chem. [44] G. Kokkinidis and D. Jannakoudakis, J. Electroanal.
B, 104 (2000) 3116. Chem., 162 (1984) 163.
[19] T.J. Schmidt, H.A. Gasteiger, G.D. Stäb, P.M. Urban, [45] S.A.S. Machado, A.A. Tanaka, E.R. Gonzales, Electro-
D.M. Kolb, R.J. Behm, J. Electrochem. Soc., 145 (1998) chim. Acta, 36 (1991) 1325.
2354. [46] T. Abe, G.M. Swain, K. Sashikata, K. Itaya, J. Electroa-
[20] U.A. Paulus, T.J. Schmidt, H.A. Gasteiger, R.J. Behm, J. nal. Chem., 382 (1995) 73.
Electrochem. Soc., (1999) [47] C.A. Lucas, N.M. Markovic, P.N. Ross, Phys. Rev. B, 55
[21] V. Stamenkovic, N.M. Markovic, P.N. Ross Jr., J. Elec- (1997) 7964.
troanal. Chem., 500 (2000) 44. [48] F.J. Luczak and D.A. Landsman, United States Patent No.
[22] V. Stamenkovic and N.M. Markovic, Langmuir, 179 US 4.677.092, filed 1987.
(2001) 2388. [49] B. Beard and P.N. Ross Jr., J. Electrochem. Soc., 137
[23] U.A. Paulus, G.G. Scherer, A. Wokaun, T.J. Schmidt, V. (1990) 3368.
Stamenkovic, V. Radmilovic, N.M. Markovic, P.N. Ross, [50] J.T. Glass, G.L. Cahen, G.E. Stoner, J. Electrochem. Soc.,
J. Electroanal. Chem., (2001) 134 (1987) 58.
[24] T.J. Schmidt, U.A. Paulus, H.A. Gasteiger, R.J. Behm, J. [51] S. Mukerjee and S. Srinivasan, J. Electroanal. Chem.,
Electroanal. Chem., (2001) (1993) 201.
[25] F. Will, J. Electroanal. Chem., 112 (1965) 451. [52] P.N. Ross Jr., in Electrocatalysis, J. Lipkowski, P.N. Ross
[26] I.M. Tidswell, N.M. Markovic, P.N. Ross, Phys. Rev. Jr. Editors, p. 43, Wiley-VCH, Inc., New York (1998).
Lett., 71 (1993) 1601. [53] P.N. Ross, N.M. Markovic, T.J. Schmidt and V. Stamen-
[27] I.M. Tidswell, N.M. Markovic, P.N. Ross, J. Electroanal. kovic, in Fuel Cell for Transportation FY 2000 Annual
Chem., 376 (1994) 119. Report, US Department of Energy, Washington, DC,
[28] H. Wroblowa, Y.C. Pan, J. Razumney, J. Electroanal. 2000.
Chem., 69 (1976) 195. [54] V. Radmilovic, in preparation, (2001)
[29] B.N. Grgur, N.M. Markovic, P.N. Ross Jr., Can. J. [55] N.M. Markovic and P.N. Ross, J. Electroanal. Chem.,
Chem., 75 (1997) 1465. 330 (1992) 499.

116 FUEL CELLS 2001, 1, No. 2