вход по аккаунту


Changing the Activity of Electrocatalysts for Oxygen Reduction by Tuning the Surface Electronic Structure.

код для вставкиСкачать
Oxygen Reduction Reaction
DOI: 10.1002/ange.200504386
Changing the Activity of Electrocatalysts for
Oxygen Reduction by Tuning the Surface
Electronic Structure**
Vojislav Stamenkovic, Bongjin Simon Mun,
Karl J. J. Mayrhofer, Philip N. Ross,
Nenad M. Markovic, Jan Rossmeisl, Jeff Greeley, and
Jens K. Nørskov*
The fuel cell is a promising alternative to environmentally
unfriendly devices that are currently powered by fossil fuels.
In the polymer electrolyte membrane fuel cell (PEMFC), the
main fuel is hydrogen, which through its reaction with oxygen
produces electricity with water as the only by-product. To
make PEMFCs economically viable, there are several problems that should be solved; the main one is to find more
effective catalysts than Pt for the oxygen reduction reaction
(ORR), 1/2 O2 + 2 H+ + 2 e = H2O. The design of inexpensive, stable, and catalytically active materials for the ORR will
require fundamental breakthroughs, and to this end it is
important to develop a fundamental understanding of the
catalytic process on different materials. Herein, we describe
how variations in the electronic structure determine trends in
the catalytic activity of the ORR across the periodic table. We
show that Pt alloys involving 3d metals are better catalysts
than Pt because the electronic structure of the Pt atoms in the
surface of these alloys has been modified slightly. With this
understanding, we hope that electrocatalysts can begin to be
designed on the basis of fundamental insight.
[*] Dr. J. Rossmeisl, Dr. J. Greeley, Prof. J. K. Nørskov
Center for Atomic-Scale Materials Physics, NanoDTU
Department of Physics
Technical University of Denmark
2800 Lyngby (Denmark)
Fax: (+ 45) 4593-2399
Dr. V. Stamenkovic, Dr. B. S. Mun, Dr. K. J. J. Mayrhofer, Dr. P. N. Ross
Materials Sciences Division
Lawrence Berkeley National Laboratory
University of California
Berkeley, CA 94720 (USA)
Dr. N. M. Markovic
Materials Science Division
Argonne National Laboratory
Argonne, IL 60439 (USA)
[**] This work was supported by the Director, Office of Science, Office of
Basic Energy Sciences, Division of Materials Sciences, US Department of Energy (contract no. DE-AC03–76SF00098), by The Danish
Research Council through the NABIIT program, and by the Danish
Center for Scientific Computing (grant no. HDW-1103-06). J.G.
acknowledges a H. C. Ørsted Postdoctoral Fellowship from the
Technical University of Denmark. N.M.M. acknowledges support
from the US Department of Energy under contract W-31-109-ENG38.
Angew. Chem. 2006, 118, 2963 –2967
2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
We studied polycrystalline alloy films of the type Pt3M
(M = Ni, Co, Fe, and Ti). Although such alloys have already
been considered,[1–4] the role of the 3d metals in electrocatalytic activity of Pt for the ORR remains elusive. In
previous work,[1, 3] one of the difficulties in illuminating the
effect of alloying components was that the kinetics of the
ORR was determined on poorly characterized Pt3M electrodes. However, we have recently studied Pt3M alloys that were
then annealed to 1000 K under ultrahigh-vacuum (UHV)
conditions and subsequently characterized in detail.[5] From
the analysis of low-energy ion-scattering spectra, we found
that the first surface layer of these alloys consists of pure Pt.
The surface enrichment of Pt atoms occurs as a result of a
surface segregation phenomenon, whereby one of the alloy
components (in this case Pt) enriches the surface region.[6]
UHV experimental and theoretical analysis revealed that a
strong enrichment of Pt in Pt3M alloy systems is counterbalanced by the depletion of Pt in the first two or three layers,
resulting in a concentration profile that oscillates around the
bulk value.[7–11]
Subsequent testing of these Pt overlayers in electrochemical environments showed significant differences in the
activity for the ORR. For the purpose of demonstrating
catalytic enhancement by alloying Pt with the 3d elements,
two representative sets of polarization curves for the ORR on
Pt and Pt3Co, along with data on the corresponding production of peroxide, are summarized in Figure 1. Note that after
these samples were removed from the electrochemical cell
they were again characterized in the UHV chamber, and the
Pt overlayer structure was found to be essentially unchanged.
Although the rate of the ORR is significantly enhanced on
Pt3Co (Figure 1 a,b), equally small amounts of peroxide are
detected on the ring electrode on both surfaces, in agreement
with previous studies.[5] The rate of the ORR on Pt alloyed by
the 3d metals is dependent on the nature of the alloying
component (see Figure 1 c). While the Ni, Co, and Fe alloys
show a large improvement in activity compared to pure Pt,
the enhancement is smaller for the earlier 3d metals. We thus
have a series of well-characterized alloys showing an interesting volcano-shaped variation in the electrocatalytic activity. As such, this result is an excellent basis for developing an
understanding of the factors that control the kinetics of the
A model of the ORR developed previously[12] was used as
the starting point for the analysis of the experimentally
observed volcano-shaped dependence as the 3d metal
changes from Ni to Ti. By using density functional theory
(DFT) calculations, the free energies of all intermediates of
the ORR were calculated as a function of the cell potential
(see Experimental Section). On this basis, a kinetic model was
developed that gives the rate of the ORR at a given potential
as a function of a single parameter characterizing the catalyst
surface—the oxygen chemisorption energy, DEO. Figure 2
shows how this model leads to a volcano-shaped dependence
of the rate on DEO. For metals that bind oxygen too strongly
(to the left of the maximum), the rate is limited by the
removal of adsorbed O and OH species; that is, the surface is
oxidized and thus unreactive, as also suggested by Ross and
co-workers.[13, 14] For metal surfaces that bind oxygen too
Figure 1. a) Production of peroxide from the oxygen reduction reaction
(ORR) detected on a ring electrode and b) polarization curves (kinetic
current density vs potential) for the ORR on Pt and Pt3Co alloy
surfaces on the disc electrode in 0.1 m HClO4 recorded at 50 mVs 1 at
333 K (1600 rpm; RHE = reversible hydrogen electrode). Arrows indicate the values of the half-wave potential, which on Pt3Co is shifted
positively by about 50 mV. c) Specific activity of Pt and Pt3M electrodes
expressed as a kinetic current density for the ORR at 0.9 V (0.1 m
HClO4, 333 K).
weakly, the rate is limited by the dissociation of O2, or more
likely, the transfer of electrons and protons to adsorbed O2.
These different rate-limiting steps are associated with different oxygen reduction mechanisms. Two such mechanisms,
involving either O2 dissociation or proton and electron
transfer to molecular O2, were considered in our analysis.
The mechanisms are both characterized by DEO, but they give
rise to different slopes on the right side of the volcano-shaped
plot (Figure 2). The volcano shape thus illustrates very well
the Sabatier principle,[15] with the important new viewpoint
that it is quantitative; thus, we know which bond energy gives
the maximum of the “volcano”.
2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. 2006, 118, 2963 –2967
in the catalytic activity; when appropriate models are
employed, DFT calculations are known to be capable of
reproducing such trends quite well.[17]
As DEO is a good descriptor of the ORR activity of a given
surface, the question arises as to which property of the
catalyst determines it. It is well-established in the literature of
surface science and heterogeneous catalysis that surface bond
energies correlate with the average energy of the d states on
the surface atoms to which the adsorbate binds (the d-band
center).[18–25] This conclusion also holds for bonding of oxygen
to the Pt3M alloys, as illustrated in Figure 3. Similar effects
Figure 2. Model of the activity (A = kB T ln(r), where r is the rate per
surface atom per second) at a cell potential of 0.9 V shown (g, *)
as a function of the adsorption energy of oxygen, DEO. On the left side
of the plot, the rate is limited by removal of adsorbed O and OH
species; immediately to the right of the maximum, the rate is limited
by O2 dissociation; and on the extreme right, the limiting step is
protonation of adsorbed O2. Also shown in red are the measured
activities relative to that of Pt. The activity of the experiment is
A = kB T ln(i/iPt) + APt, where i/iPt is the current density relative to Pt,
and APt is the theoretical value for the activity of Pt. It is assumed that
the number of active sites per surface area is the same for Pt and all
the alloys. A factor of two in the preexponential factor would give rise
to a difference of only 0.02 eV (kB T ln 2) in the activity. Changes in
coverage for the different alloys are not considered. See text for details.
The predictions of the model for Pt3M (M = Ni, Co, Fe, Ti) alloys are
shown relative to the predictions for Pt at the DFT-calculated values of
One prediction of the model is that Pt binds oxygen a little
too strongly, and finding a better electrocatalyst for the ORR
thus amounts to finding a surface that binds oxygen more
weakly than Pt does by about 0.2 eV. To test this prediction,
we calculated DEO for the (111) surfaces of a series of Pt3M
alloys with pure Pt in the first layer and only 50 % Pt in the
second layer to model the enrichment of Pt in the surface
layer and the depletion of Pt in the next layer (the slabs thus
retain bulk Pt3M stoichiometry). The composition of the
second layer can vary modestly from one alloy to the next for
the different metals (see Reference [16] for a discussion of Pt–
Fe, Pt–Co, and Pt–Ni alloy segregation). However, the
calculated binding energies are not highly sensitive to such
changes; changing the composition of M in the second layer
from 50 % to 25 %, for example, changes binding energies by
only approximately 0.05 eV for M = Fe, Co, and Ni. Thus, our
simple stoichiometric model should be accurate for determining the trends in activities. These data, plus the corresponding
activity predictions of the model, are shown in Figure 2.
Clearly, the alloy surfaces form weaker bonds to oxygen than
pure Pt, and the predicted activities (relative to pure Pt) are
compatible with the experimental measurements that are also
included in the figure. Experiment and theory agree almost
quantitatively, but this excellent agreement in the relative
rates is not the most significant point. Rather, we stress that
we have a model of the ORR that accurately predicts trends
Angew. Chem. 2006, 118, 2963 –2967
Figure 3. a) The correlation between the d-band center and the oxygen
adsorption energy. The d-band center was calculated (DFT) as the
average for the Pt atoms in the two top layers. b) sp-broadened
2p orbital for O(g), c) projected p density of states of oxygen atoms on
Pt(111), and d) projected d density of states of Pt(111).
have been observed for Pt(111) with a 3d transition metal in
the second layer, or for Pt(111) overlayers on 3d metals.[26, 27]
The origin of this relationship is as follows: The variation
in the oxygen–metal bond from one transition-metal surface
to the next depends to a large extent on the strength of the
coupling between the oxygen 2p states and the metal d states.
This coupling forms bonding and antibonding states as
illustrated in Figure 3 b–d. The bonding states are filled, and
the filling of the antibonding states, and thus the strength of
the interaction, varies from surface to surface. In a metallic
environment the filling depends on the position of the states
relative to the Fermi level. An upward shift of the d states
relative to the Fermi level must therefore result in an upward
shift of the antibonding states, leading to less filling and thus
to a stronger bond. For Pt overlayers or other structures
involving bonding of oxygen to Pt, in which other factors that
determine the coupling strength are approximately the same,
this effect leads to the relationship shown in Figure 3.
Unlike the oxygen bond energy, which is hard to measure,
the d-band center is accessible experimentally. We measured
it for all of the alloys by using synchrotron-based highresolution photoemission spectroscopy, a methodology pre-
2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
viously described.[28] Measurement of the d-band centers
allows us to directly correlate the variations in the catalytic
activity for the ORR with the variations in the surface
electronic structure. As summarized in Figure 4, the activity
Figure 4. Activity versus the experimentally measured d-band center
relative to platinum. The activity predicted from DFT simulations is
shown in black, and the measured activity is shown in red.
(A) versus d-band center position at 0.9 V exhibits a classical
volcano-shaped dependence, which agrees very well with the
activity predicted from DFT calculations. For catalysts that
bind oxygen too strongly, the rate is limited by the removal of
surface oxide, while for catalysts that bind oxygen too weakly,
the rate is limited by the transfer of electrons and protons to
adsorbed O2.
Note that a shift in the d states can also often be measured
as a core-level shift, as the d states and the core levels shift
together.[20] This effect can explain the correlations between
ORR activity and the surface core-level shifts observed by
Watanabe and co-workers.[29]
In summary, we have established how alloying Pt with
3d transition metals tunes the electronic structure to change
the performance of an alloy for the ORR, and we have thus
developed a model for the electrocatalytic activity of a metal
surface for this process. The activity is given by the strength of
the oxygen–metal bond interaction, which in turn depends on
the position of the metal d states relative to the Fermi level. In
this way we have established a new approach for the screening
of new catalysts for the ORR, by looking for surfaces that
bind oxygen a little weaker than Pt or, specifically for Pt skins,
by looking for surfaces with a down shift of the Pt d states
relative to the Fermi level.
Experimental Section
DACAPO, the total energy calculation code,[30] was used in this study.
For all calculations, a four-layer slab, periodically repeated in a
supercell geometry with six equivalent layers of vacuum between any
two successive metal slabs, was used. A 2 F 2-unit cell was employed,
and the top two layers of the slab were allowed to relax. Adsorption
was allowed on only one of the two surfaces exposed, and the
associated dipole was corrected for electrostatically.[31] Ionic cores
were described by ultrasoft pseudopotentials,[32] and the Kohn–Sham
one-electron valence states were expanded in a basis of plane waves
with kinetic energy below 340 eV; a density cutoff of 500 eV was used.
The surface Brillouin zone of close-packed metal surfaces was
sampled at 18 special Chadi–Cohen k points. In all cases, convergence
of the total energy was confirmed with respect to the k point set and
with respect to the number of metal layers included. The exchangecorrelation energy and potential were described self-consistently
within the generalized gradient approximation (GGA-RPBE).[30] The
self-consistent RPBE density was determined by iterative diagonalization of the Kohn–Sham Hamiltonian, Fermi population of the
Kohn–Sham states (kB T = 0.1 eV), and Pulay mixing of the resulting
electronic density.[33] All total energies were extrapolated to kB T =
0 eV. The d-band centers were calculated with an infinite cutoff
On all metals and alloys (composition Pt3M) investigated in this
study, oxygen was found to adsorb at face-centered cubic (fcc) sites.
Owing to the limited size of the unit cell, only two different fcc sites
were present, one of which is close to one of the M atoms in the
second layer, while the other is close to two M atoms. The latter was
the most stable for all the considered alloys.
Solvent and potential effects were treated with the same model as
used in our previous study.[12] In that work, it was confirmed that a
bilayer of water has no effect on the binding energy of atomic oxygen.
For the other species in the full model (not explicitly involved in the
present calculations), the binding energies were corrected by the
interaction of a water bilayer with the appropriate species on Pt(111).
Potential effects were included by adjusting the free energies of
adsorption (per transferred electron) by eU. Electric-field effects
were not explicitly included, but these effects were found to change
the adsorption energies by less than 0.05 eV, and the effect was
essentially constant from one metal to the next.
Received: December 9, 2005
Revised: February 6, 2006
Published online: April 5, 2006
Keywords: alloys · density functional calculations ·
electrochemistry · oxygen · platinum
[1] A. J. Appleby, Catal. Rev. 1970, 4, 221 – 244.
[2] P. Stonehart, Chem. Ber. 1990, 94, 913 – 921.
[3] K. Kinoshita, Electrochemical Oxygen Technology, Wiley, New
York, 1992.
[4] N. M. Markovic, P. N. Ross, Surf. Sci. Rep. 2002, 45, 121 – 229.
[5] V. Stamenkovic, T. J. Schmidt, P. N. Ross, N. M. Markovic, J.
Phys. Chem. B 2002, 106, 11 970 – 11 979.
[6] A. V. Ruban, H. L. Skriver, J. K. Nørskov, Phys. Rev. B 1999, 59,
15 990 – 16 000.
[7] Y. Gauthier, Y. Joly, R. Baudoing, J. Rundgren, Phys. Rev. B
1985, 31, 6216 – 6218.
[8] U. Bardi, A. Atrei, E. Zanazzi, G. Rovida, P. N. Ross, Vacuum
1990, 41, 437 – 440.
[9] Y. Gauthier, Surf. Rev. Lett. 1996, 3, 1663 – 1689.
[10] M. A. Vasiliev, J. Phys. D 1997, 30, 3037 – 3070.
[11] M. Lundberg, Phys. Rev. B 1987, 36, 4692 – 4699.
[12] J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R.
Kitchin, T. Bligaard, H. Jonsson, J. Phys. Chem. B 2004, 108,
17 886 – 17 892.
[13] N. M. Markovic, H. A. Gasteiger, P. N. Ross, J. Phys. Chem.
1996, 100, 6715 – 6721.
[14] N. M. Markovic, T. A. Schmidt, V. Stamenkovic, P. N. Ross, Fuel
Cells 2001, 1, 105 – 116.
[15] T. Bligaard, J. K. Nørskov, S. Dahl, J. Matthiesen, C. H.
Chistensen, J. Sehested, J. Catal. 2004, 224, 206 – 217.
[16] Y. Gauthier, R. Baduoingsavois, J. M. Bugnard, U. Bardi, A.
Atrei, Surf. Sci. 1992, 276, 1 – 11.
[17] B. Hammer, J. K. Nørskov, Adv. Catal. 2000, 45, 71 – 129.
[18] B. Hammer, J. K. Nørskov, Nature 1995, 376, 238 – 240.
2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. 2006, 118, 2963 –2967
[19] J. Greeley, J. K. Nørskov, M. Mavrikakis, Annu. Rev. Phys.
Chem. 2002, 53, 319 – 348.
[20] B. Hammer, Y. Morikawa, J. K. Nørskov, Phys. Rev. Lett. 1996,
76, 2141 – 2144.
[21] D. R. Alfonso, A. V. Cugini, D. S. Sholl, Surf. Sci. 2003, 546, 12 –
[22] L. A. Kibler, A. M. El-Aziz, R. Hoyer, D. M. Kolb, Angew.
Chem. 2005, 117, 2116 – 2120; Angew. Chem. Int. Ed. 2005, 44,
2080 – 2084.
[23] V. Pallassana, M. Neurock, J. Catal. 2000, 191, 301 – 317.
[24] A. Roudgar, A. Gross, Phys. Rev. B 2003, 67, 033 409.
[25] M. Gajdos, A. Eichler, J. Hafner, J. Phys. Condens. Matter 2004,
16, 1141 – 1164.
[26] J. Zhang, M. B. Vukmirovic, Y. Xu, M. Mavrikakis, R. R. Adzic,
Angew. Chem. 2005, 117, 2170 – 2173; Angew. Chem. Int. Ed.
2005, 44, 2132 – 2135.
[27] J. R. Kitchin, J. K. Nørskov, M. A. Barteau, J. G. Chen, J. Chem.
Phys. 2004, 120, 10 240 – 10 246.
[28] B. S. Mun, C. Lee, V. Stamenkovic, N. M. Markovic, P. N. Ross,
Phys. Rev. B 2005, 71, 115 420.
[29] T. Toda, H. Igarashi, H. Uchida, M. Watanabe, J. Electrochem.
Soc. 1999, 146, 3750 – 3756.
[30] B. Hammer, L. B. Hansen, J. K. Nørskov, Phys. Rev. B 1999, 59,
7413 – 7421.
[31] L. Bengtsson, Phys. Rev. B 1999, 59, 12 301 – 12 304.
[32] D. Vanderbilt, Phys. Rev. B 1990, 41, 7892 – 7895.
[33] G. Kresse, J. Furthmuller, Comput. Mater. Sci. 1996, 6, 15 – 50.
Angew. Chem. 2006, 118, 2963 –2967
2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Без категории
Размер файла
283 Кб
structure, tuning, electronica, reduction, electrocatalysts, surface, activity, oxygen, changing
Пожаловаться на содержимое документа