Origin of Oxygen Reduction Reaction Activity on “Pt3Co

Dec 24, 2008 - Multi-Component Fe–Ni Hydroxide Nanocatalyst for Oxygen Evolution and Methanol Oxidation Reactions under Alkaline Conditions...
1 downloads 3 Views 4MB Size
J. Phys. Chem. C 2009, 113, 1109–1125

1109

Origin of Oxygen Reduction Reaction Activity on “Pt3Co” Nanoparticles: Atomically Resolved Chemical Compositions and Structures Shuo Chen,† Wenchao Sheng,‡ Naoaki Yabuuchi,† Paulo J. Ferreira,§ Lawrence F. Allard,| and Yang Shao-Horn*,†,⊥ Department of Mechanical Engineering, Department of Chemistry, and Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, Materials Science and Engineering Program, UniVersity of Texas at Austin, Austin, Texas 78712, and High Temperature Materials Laboratory, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831 ReceiVed: August 11, 2008; ReVised Manuscript ReceiVed: October 27, 2008

Rotating disk electrode measurements of acid-treated “Pt3Co” nanoparticles showed specific oxygen reduction reaction (ORR) activity (∼0.7 mA/cmPt2 at 0.9 V vs RHE in 0.1 M HClO4 at room temperature), twice that of Pt nanoparticles. Upon annealing at 1000 K in vacuum, the ORR activity at 0.9 V was increased to ∼1.4 mA/cmPt2 (four times that of Pt nanoparticles). High-resolution transmission electron microscopy and aberrationcorrected high-angle annular dark-field in the scanning transmission electron microscope was used to reveal surface atomic structure and chemical composition variations of “Pt3Co” nanoparticles on the atomic scale. Such information was then correlated to averaged Pt-Pt distance obtained from synchrotron X-ray powder diffraction data, surface coverage of oxygenated species from cyclic voltammograms, and synchrotron X-ray absorption spectroscopy. It is proposed that ORR activity enhancement of acid-leached “Pt3Co” relative to Pt nanoparticles is attributed to the formation of a percolated structure with Pt-rich and Pt-poor regions within individual particles, while the increase in the specific ORR activity of annealed “Pt3Co” nanoparticles relative to Pt can be attributed to the presence of surface Pt segregation. 1. Introduction Fuel cells that operate on hydrogen produced from solar energy are promising technologies to provide energy storage for renewable energy supply and energy demands.1 Oxygen reduction reaction (ORR) that typically results in 300-400 mV loss in the cell voltage2 limits fuel cell efficiency. It has been of great interest to discover nanoscale catalysts with ORR activity superior to that of Pt. Significant experimental and computational research efforts have been focused on understanding surface chemistry and atomic and electronic structures of Pt alloy model surfaces that exhibit ORR activity higher than Pt,3-8 from which strategies can be developed to discover highly active catalysts on the nanoscale. Two strategies that are shown to enrich Pt on extended Pt alloy surfaces are (1) surface segregation of Pt by high-temperature annealing9,10 and (2) acid removal of less noble alloying elements than Pt from alloy surfaces.11 In both cases, ORR activity has shown to increase by 2-10 times relative to that of Pt.3,7,10-15 Stamenkovic et al.12,15 have shown that annealed polycrystalline Pt3M (M ) Co, Fe, and Ni) surfaces consist of segregation of pure Pt in the top surface layer, which is associated with the depletion of Pt in the second layer and composition oscillation in the subsequent layers. The top surface layer of pure Pt is referred to as “Pt-skin”11 while the composition oscillation in the first three to four layers away * Corresponding author. E-mail: [email protected]. † Department of Mechanical Engineering, Massachusetts Institute of Technology. ‡ Department of Chemistry, Massachusetts Institute of Technology. § University of Texas at Austin. | Oak Ridge National Laboratory. ⊥ Department of Materials Science and Engineering, Massachusetts Institute of Technology.

from the surface has been reported as the “sandwich-segregation” structure.9 Density functional theory (DFT) studies of Norskov et al.6,16 and Mavrikakis et al.17 have shown that the enhanced ORR activity of Pt3M(111) (M ) Co, Fe, Ni) surfaces with the Pt-skin can be attributed to weakening metal-oxygen bond strength relative to Pt(111). This reduced surface reactivity of the Pt-skin structure results from lowered Pt valence band center relative to the Fermi level,18 which is induced both by shortened surface Pt-Pt bond distance constrained by transition metals in the second layer beneath the surface and by ligand effect.17,19,20 Pure Pt on the outermost surface layer can be also created on Pt alloys by acid removal of transition metals.7,11,21 Little is known about the atomic structure and chemical compositions of near-surface regions after acid treatments, and the origin of the ORR enhancement of acid-treated Pt alloys is not clearly understood. Toda et al.7 has used the Pt-skin phrase to describe the acid-treated Pt alloy surface regions where Pt is enriched in the top three to four atomic layers, which can have different structures from that of the sandwich-segregation structure.9 The activity enhancement relative to Pt has been attributed to increased d electron vacancy in the Pt-enriched surface region induced by underlying transition metals, and thus increased oxygen adsorption and weakened O-O bonds.7,22 On the other hand, it has been proposed recently that removal of transition metals from Pt alloy in acid renders a near-surface region with a “Pt-skeleton” structure,11,12 which consists of Pt atoms and vacancies left behind by the transition metal in the top few atomic layers. The enhanced ORR activity of the Pt-skeleton structure12 relative to Pt has been correlated recently to reduced surface reactivity toward surface-oxygenated species as a result of lowered d band center energy normalized to the Fermi level11 of Pt alloy surfaces before acid treatments. One issue worth

10.1021/jp807143e CCC: $40.75  2009 American Chemical Society Published on Web 12/24/2008

1110 J. Phys. Chem. C, Vol. 113, No. 3, 2009 mentioning is that the Pt d band center energy of the Pt-skeleton surface of Pt alloys after acid removal of transition metal, which is strongly dependent on near-surface compositions, may deviate from those obtained in UHV.12 It is essential to examine if ORR mechanisms established for Pt bulk surfaces are applicable to nanoparticles, and to bridge fundamental understanding of ORR activity enhancement established on extended Pt alloy model surfaces to design Pt alloy nanoparticles with high ORR activity for practical applications. It is well-known that Pt alloy nanoparticles exhibit enhanced ORR activity in A/cmPt2 relative to Pt nanoparticles.14,23-28 The origin of the enhancement mechanism remains elusive. The enhanced ORR activity of Pt alloy nanoparticles has been attributed to many factors such as Pt-Pt distance25,29,30 and surface roughness (the nature of facets and steps)31 and surface electronic effects (the ligand effect) such as Pt 5d valence band vacancy.25,28 Although X-ray powder diffraction and synchrotron X-ray absorption spectroscopy techniques have allowed the correlation of Pt-Pt interatomic distance and Pt d band vacancy averaged among nanoparticle surface and interior atoms to ORR activity,25,28 little is known about nanoparticle surface atomic structures and near-surface chemical compositions that govern activity. Although Pt segregation on the {111} surfaces has been predicted on truncated octahedral “Pt3Co” nanoparticles of 2.5 nm,12 it remains unclear whether and to what extent Pt segregation occurs on Pt alloy nanoparticles surfaces and how it may influence ORR activity. It is of particular interest to determine if and to what extent depletion of transition metals in the surface layers and Pt surface segregation can occur in acid-treated and high-temperatureannealed Pt alloy nanoparticles, respectively. In this study, we use conventional and aberration-corrected high-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) to reveal the composition inhomogeneity of “Pt3Co” nanoparticles on the atomic scale, from which size-dependent Pt/Co ratios, percolated and core-shell nanoparticle structures, and surface Pt segregation are revealed. In addition, highresolution transmission electron microscopy (HRTEM) is employed to determine differences in the surface atomic structure between acid-treated “Pt3Co” and Pt nanoparticles. Such information is then correlated to surface coverage of oxygenated species from synchrotron X-ray absorption spectroscopy and cyclic voltammetry, and specific ORR activity (based on Pt surface area) of Pt and “Pt3Co” nanoparticles in weakly adsorbing HClO4 acid, from which the origin of ORR activity enhancement of Pt alloy nanoparticles is discussed. 2. Experimental Methods 2.1. Sample Preparation. Two samples of “Pt3Co” nanoparticles supported on carbon (Pt 46 wt %), with average atomic ratios of Pt/Co close to 3, were examined. One was prepared by acid treatments from “PtCo” alloy nanoparticles with an average atomic ratio of Pt/Co close to 1, which was supplied by Tanaka Kikinzoku (TKK) International, Inc. This sample had an average Co atomic percentage of ∼22 at %, which is referred to as AT-“Pt3Co” in this paper. The other was obtained by annealing AT-“Pt3Co” at 1000 K for 3 h at a low pressure of 3 × 10-2 torr and then cooling in the furnace to 373 K (average heating, 30 K/min; average cooling, 5 K/min). This sample is referred to as HT-AT-“Pt3Co”. In addition, a sample of Pt nanoparticles was used for comparison in this study. In order to minimize the effect of particle sizes on specific ORR activity,2,32-34 this sample was prepared from heat-treating a catalyst sample (Pt nanoparticles of 2 nm supported on high-

Chen et al. surface-area carbon, Pt 46 wt %, TKK) at 1173 K for 1 min in argon (∼30 K/min) and cooling to ambient (∼5 K/min) to obtain a comparable number-averaged particle size of ∼4 nm for “Pt3Co” nanoparticles. This Pt sample is referred to as Pt-4nm. 2.2. Electrochemical Measurements. 2.2.1. Preparation of Pt Catalyst Electrodes. Thin catalyst layers were prepared by using methods reported previously.14,35,36 Suspensions of Pt catalysts of 0.2∼0.5 mg/mL were obtained by dispersing the catalysts in deionized water (18.2 MΩ · cm, Millipore) using water bath sonication, and ultrasonication was performed when needed. Twenty microliters of the Pt catalyst suspension (in water) was deposited on glassy carbon electrodes (GCE) (5 mm in diameter, Pine Instruments), which were prepolished to 0.05 µm of alumina. After drying in air at room temperature, 20 µL of 0.025 wt % of Nafion water solution (diluted from 5 wt % of Nafion, Ion Power, Inc.) was added and dried in air to immobilize the catalysts. Pt loadings of ∼10 µg/cm2 on GCE were used for CV and ORR measurements. 2.2.2. Cyclic Voltammetry. As-prepared electrodes were then mounted to a rotator (Pine Instruments) and immersed into 0.1 M HClO4 that was diluted from 1.0 M HClO4 (Sigma Aldrich) with deionized water (18.2 MΩ · cm, Millipore). A spiral Pt wire was employed as the counter electrode, and a saturated calomel electrode (SCE, Analytical Sensor, Inc.) was used as the reference electrode. The potential of SCE with respect to the reversible hydrogen electrode (RHE) was calibrated from rotating disk electrode measurements of hydrogen oxidation. All the potential values reported in this paper refer to that of the RHE. After the electrolyte was bubbled with nitrogen (N2) for half an hour, the working electrodes were scanned between 0.02 and 1.21 V vs RHE at a sweep rate of 200 mV/s for 60 cycles to remove contaminants from Nafion solution. Steadystate cyclic voltammograms were then recorded at 50 mV/s in the same potential range at room temperature. The electrochemical surface area (ESA) of supported Pt and “Pt3Co” nanoparticles was determined from the hydrogen underpotential deposition region (0.05 V to the onset of the double-layer region), i.e., the desorption pseudocapacitance (210 µC/cmPt2 for Pt nanoparticles, 200 µC/cmPt2 for AT-“Pt3Co”, and 180 µC/ cmPt2 for HT-AT-“Pt3Co” as reported previously).11 For the specific surface area calculation of “Pt3Co” nanoparticles, it is assumed that protons adsorb on Pt atoms only but not on Co atoms in the hydrogen underpotential region. 2.2.3. Rotating Disk Electrode Measurements of ORR ActiWity. After the electrolyte was purged with pure oxygen (O2) for at least half an hour, polarization curves were recorded between 0.02 and 1.21 V vs RHE under a voltage sweep rate of 10 mV/s at room temperature. Current-voltage data were recorded at rotating speeds of 100, 400, 900, 1600, and 2500 rpm. All of the GCEs with thin catalyst layers reached a welldefined limiting current starting from 0.7 V. The Koutecky-Levich plots based on 1/j ) 1/jk + 1/jD ) 1/jk + 1/(BC0ω1/2) at 0.35, 0.65, and 0.75 V revealed an excellent linear behavior between 1/j and 1/ω1/2 (see Figures 1a-c, Supporting Information). The slope, 1/(BC0), reflects the number of electrons involved in the reaction. The calculated BC0 values are 0.157, 0.149, and 0.159 mA cm2 ω-1/2 for Pt-4nm, AT-“Pt3Co”, and HT-AT-“Pt3Co”, respectively. They are in good agreement with the predicted value (0.1505 mA cm2 ω-1/2) from the relationship jD ) 0.2nFCO2 DO2/32 V-1/6 ω1/2,37 where n, the apparent number of electrons transferred in the reaction, is equal to 4, F is the Faraday constant, CO2 is the O2 concentration in 0.1 M HClO4 (1.26 × 10-3 mol L-1), DO2 is the diffusivity of O2 (1.93 × 10-5 cm2 s-1) in dilute electrolyte solutions, and ν is the

ORR Activity on “Pt3Co” Nanoparticles kinematic viscosity of the electrolyte (1.009 × 10-2 cm2 s-1).38 Knowing ORR on both Pt and “Pt3Co” nanoparticles occurs via a four-electron pathway, the specific activity was calculated from jk at 0.9 V vs RHE and was compared among the catalysts samples examined in this study. 2.3. Characterization of Nanoparticles: Atomic Structure, Electronic Structure, and Size-Dependent Composition. The average Pt-Pt distance in the “Pt3Co” samples was obtained from synchrotron X-ray diffraction analysis and synchrotron extended X-ray adsorption fine structure (EXAFS). The particle size distribution and the surface atomic structure of “Pt3Co” and Pt nanoparticles were determined by high-resolution transmission electron microscopy (HRTEM). The compositions of individual “Pt3Co” nanoparticles were determined by X-ray energy dispersive spectroscopy (EDS) in a conventional STEM mode. Surface atomic structure and chemical compositions of “Pt3Co” nanoparticles were revealed by aberration-corrected HAADF STEM imaging. D band vacancy values of “Pt3Co” and Pt nanoparticles were obtained from synchrotron X-ray absorption near-edge structure (XANES) data. 2.3.1. Synchrotron X-ray Powder Diffraction. Synchrotron radiation of BL02B2 at Spring-8 (Sayo-gun, Hyogo, Japan), equipped with a large Debye-Scherrer camera,39 was used to collect X-ray diffraction data of Pt and “Pt3Co” nanoparticle samples. The incident beam was adjusted to a wavelength of 0.5 Å by a Si(111) monochromator to minimize the absorption by the samples. The wavelength was calibrated to be 0.5027 Å by using CeO2 standard (a ) 5.4111(1) Å). The data were collected in the range of 0-75° in 2θ. A few milligrams of each sample in the powder form was placed in a quartz capillary (0.5 mm diameter and approximately 1.5 cm height) during the measurement. X-ray diffraction data were recorded on the imaging plate for 30 min. Structural analysis was performed using FullProf40 in the range of 5-60° in 2θ. 2.3.2. HRTEM Imaging. The size and distributions of Pt and “Pt3Co” nanoparticles were examined in a JEOL 2010F TEM operated at 200 kV with a point-to-point resolution of 0.19 nm. Nanoparticles were first immersed in ethanol and subsequently dispersed ultrasonically for 5 min. The suspension was then deposited on a lacey carbon grid and dried in air for TEM observations. Two hundred randomly selected nanoparticles from HRTEM images were used to produce particle size distributions of Pt and “Pt3Co” samples. For each distribution, the number-averaged diameter dn was determined by dn ) (∑i 41 n ) 1 di)/n while the volume-surface-area-averaged diameter dv/a n 3 n 2 was calculated by dv/a ) (∑i ) 1 di )/(∑i ) 1 di ), where di is the diameter of individual particles. The specific surface area of nanoparticles based on the volume-surface-area-averaged diameter from TEM measurements was determined by 6(1000/ (FPtdv/a)). Recent studies have shown that steps on the surfaces of Pt nanoparticles can be identified by HRTEM imaging.42 The indexes of surface facets of Pt and “Pt3Co” nanoparticles were determined from the HRTEM images and fast Fourier transform (FFT) analysis. The FFTs of HRTEM images obtained from Digital Micrograph (Version 3.11.2, Gatan Inc., Pleasanton, CA) were then compared with simulated electron diffraction patterns of Pt (space group Fm3j m), ordered “Pt3Co” (space group Pm3jm), and ordered PtCo (space group P4/mmm). Each diffracted spot in the FFT images represents a set of equal spacing planes whose normal direction coincides with the straight line that connects this diffraction spot with the transmitted spot. The length of this straight line is proportional to the reciprocal of the interplanar distance. With the aid of Digital

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1111 Micrograph, the interplanar distance and the angle between atomic planes were obtained from FFT images, which allowed indexing of lattice fringes and surface facets in the HRTEM images. In this study, high-index planes were separated into three groups according to the notations used in previous stepped single-crystal studies:43-46 (1) when the {111} planes are present as terraces and single {100} as steps, the notation of n{111} × {100} is used; (2) when the {100} planes are present as terraces and single {111} as steps, the n{100} × {111} notation is used; (3) the {111} terraces separated by single {111} steps is denoted as n{111} × {111}. The n in the above notation corresponds to the number of atoms along the terrace viewed from the zone axis. 2.3.3. ConWentional STEM Imaging and EDS Analysis. The chemical compositions of “Pt3Co” samples were determined using EDS in a VG HB603 STEM at room temperature using a beam voltage of 250 kV and INCA control software (Version 4.08, Oxford Instruments Analytical Limited). Pt LR (∼9.442 keV) and Co KR (∼6.930 keV) signals were used for composition quantification. The average chemical composition of each “Pt3Co” sample was determined from Pt and Co signals collected for 600 s from an area of few micrometers squared with a scanning beam of 2 nm in diameter. In addition, the chemical compositions of individual “Pt3Co” nanoparticles were determined from signals collected for 20-30 s with a scanning beam of 2 nm in diameter, which provided sufficient signal-noiseratios for quantification of Pt and Co atomic fractions. Moreover, in order to probe the variation in the Pt and Co atomic fractions across a given particle, Pt and Co signals were collected from the particle center and edge of 20-40 “Pt3Co” nanoparticles using a static beam in the spot capture mode. A collection time of 30 s was used, and the drifting corrector was employed to ensure the correct beam position within the nanoparticle. Errors in the Pt and Co atomic fractions in the analysis of each spectrum were generated by INCA, which were shown to support the significance of compositional variation across different “Pt3Co” nanoparticles (size-dependent) and within some individual particles. 2.3.4. Aberration-Corrected STEM Imaging. The atomic structures and compositions of “Pt3Co” nanoparticles were examined in an aberration-corrected JEOL 2200FS-AC STEM located at Oak Ridge National Laboratory and operating at 200 kV. The aberration-corrected STEM is equipped with a Schottky FEG, a CEOS GmbH hexapole aberration corrector, an incolumn Omega filter, and a high-angle angular dark-field detector. The intensities of different elements in the HAADF images are approximately proportional to Z2, where Z is the atomic number of a certain element. The aberration corrector combined with the HAADF detector allows the formation of Z-contrast images with resolutions below 0.1 nm. This capability is ideal for imaging regions within each nanoparticle, where different compositions may exist. 2.3.5. Synchrotron X-ray Adsorption Measurements. Synchrotron XANES and EXAFS data were collected at the bending magnet beam line 20-BM-PNC at the Advanced Photo Source at the Argonne National Laboratory (ANL) at ambient conditions. X-rays were monochromatized by using a Si(111) doublecrystal monochromator, which was calibrated by defining the inflection point (first-derivative maxima) with Pt and Co foils, having a thickness of 4 and 7.5 µm, respectively. The incident and transmitted X-ray intensities were detected with two ion chambers which were continuously purged with 100% N2 (I0) and 50% N2 + 50% Ar (I1) for Pt, and 50% He + 50% N2 (I0) and 20% Ar + 80% N2 (I1) for Co. XAFS samples were

1112 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Chen et al.

prepared by pelletizing the mixture of the samples and finely ground boron nitride powders, and their thickness was determined to obtain an appropriate absorption jump. The detailed information on the beam line operation and APS storage ring can be found elsewhere.47,48 The spectra were processed and analyzed by using a program IFEFFIT,49 by which the preedge region was subtracted based on the energy range of (-130)-(-30) eV relative to the E0 for the Pt L2,3 edge and (-150)-(-30) eV for the Co K edge, and then normalized based on the energy range above 150 eV relative to the E0. EXAFS analysis was performed by using the Pt L3 edge only. The phase shift and backscattering amplitude was calculated theoretically using FEFF 8.4 code.50 3. Results 3.1. Cyclic Voltammogram and RDE Measurements of ORR Activity of “Pt3Co” Nanoparticles. Figure 1 shows cyclic voltammograms of Pt-4nm, AT-“Pt3Co”, and HT-AT-“Pt3Co” in an N2-saturated 0.1 M HClO4 solution. All samples clearly exhibit features associated with hydrogen adsorption and desorption on Pt{110} (a peak between ∼0.1 and ∼0.2 V vs RHE) and Pt{100} facets (a broad peak at ∼0.3 V vs RHE) in the hydrogen underpotential deposition region.51,52 The electrochemically active area of nanoparticles (Pt surface area) was determined from the charge associated with hydrogen desorption, which is listed in Table 1. The onset potential (∼0.8 V vs RHE) of adsorption of oxygenated species53,54 was shifted positive for HT-AT-“Pt3Co” relative to Pt and AT-“Pt3Co”. The charge density (normalized to Pt surface area) associated with water activation and adsorption of oxygenated species on Pt in the positive-going sweep and reduction of oxygenated species on Pt on the negative-going sweep are shown as a function of potential in Figures 1b and 1c, respectively. It is interesting to note that the charge for all three samples up to 1.0 V in the positive-going scan was found to be comparable, which suggests a similar degree of surface oxidation for Pt and “Pt3Co” nanoparticles at 1.0 V. This observation is in agreement with recent CV results of Pt and acid-treated Pt-Fe thin films.22 In contrast, a considerable difference in the amount of charge in the negative-going scan was found. The coverage of oxygenated species53 such as OHad was normalized to a monolayer charge density of 210 µC/cmPt2 for polycrystalline Pt (Pt-4nm), 200 µC/cmPt2 for the Pt-skeleton surface (AT-“Pt3Co”), and 180 µC/ cmPt2 for the Pt-skin surface (HT-AT-“Pt3Co”),11 as shown in Figure 1c. It was found that surface coverage of oxygenated species of AT-“Pt3Co” was higher than that of Pt while that of HT-AT-“Pt3Co” is the lowest. The difference will be discussed in the context of surface atomic structure in later sections. Specific ORR activity values of “Pt3Co” nanoparticles in O2saturated 0.1 M HClO4 were compared with that of Pt-4nm nanoparticles at 0.9 V vs RHE. Potentiodynamic polarization data of Pt-4nm, AT-“Pt3Co”, and HT-AT-“Pt3Co” nanoparticles in the positive-going sweep under different rotating speeds are shown in Figure 2b, c, respectively, where the negative-going sweep at 1600 rpm is also included for each sample. Tafel plots of specific ORR activity in µA/cmPt2 of these three samples are shown in Figure 2d. At 0.9 V, the specific activity of AT“Pt3Co” nanoparticles is ∼0.7 mA/cmPt2 at 0.9 V vs RHE (Table 1 and Figure 2d), which is comparable to those of acid-leached Pt alloy nanoparticles reported previously.14,55,56 On the other hand, the activity at 0.9 V was increased to ∼1.4 mA/cmPt2 after annealing, which is among the highest reported activity values of Pt alloy nanoparticles2 and is equal to ∼10% of the highest specific ORR activity reported for extended Pt alloy

Figure 1. (a) Cyclic voltammograms of Pt-4nm (solid black), AT“Pt3Co” (dashed red), and HT-AT-“Pt3Co” (dotted green) in N2saturated 0.1 M HClO4 with a sweeping rate of 50 mV/s at room temperature. (b) Background-corrected charge density associated with Pt-OH formation in the positive-going scans. (c) Background-corrected charge density associated with Pt-OH reduction in the negative-going scans. The OH coverage for each sample was calculated based on the charge density on different surfaces: 210 µC/cmPt2 for Pt (Pt-4nm), 200 µC/cmPt2 for the Pt-skeleton surface (AT-“Pt3Co”), and 180 µC/ cmPt2 for the Pt-skin (HT-AT-“Pt3Co”).11

surfaces.3 Mass activities of AT-“Pt3Co” (0.35 A/mgPt) and HTAT-“Pt3Co” (0.31 A/mgPt) were found similar, which is comparable to or slightly higher than those reported previously for Pt alloy nanoparticles.57,58 The Tafel slopes of Pt-4nm (75 mV/dec), AT-“Pt3Co” (73 mV/dec), and HT-AT-“Pt3Co” (88 mV/dec) as determined by [∂E/∂(log is)] agree well with those of extended Pt and “Pt3Co” polycrystals.14 3.2. Average Pt-Pt Bond Length of Pt and “Pt3Co” Nanoparticles. Synchrotron X-ray powder diffraction data of Pt and “Pt3Co” nanoparticle samples are shown in Figure 3a, where all the major peaks can be indexed to a face-centered cubic (FCC) structure with space group Fm3jm. The diffraction data of these three samples were first modeled using a single-

ORR Activity on “Pt3Co” Nanoparticles

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1113

TABLE 1: Average Particle Size Diameters, Surface Area, Mass, and Specific and ORR Activities of Pt-4nm, AT-“Pt3Co”, and HT-AT-“Pt3Co”a

sample description Pt-4nm AT-“Pt3Co” HT-AT-“Pt3Co”

specific activity is (mA cmPt-2) number-averaged volume-surface-areasurface area electrochemical surface mass activity electrochemical surface dnb (TEM) (nm) averaged dv/ac (TEM) (nm) (TEM) (m2/gPt) area (m2/gPt) im (A mgPt-1) area 3.6 (200 particles) 4.1 (199 particles) 4.9 (214 particles)

4.8 (200 particles) 5.4 (199 particles) 9.2 (214 particles)

75 (200 particles) 51.8 (199 particles) 30.3 (214 particles)

47 47 22

0.17 0.35 0.31

0.36 0.74 1.39

a The ORR activities were obtained in O2-saturated 0.1 M HClO4 at room temperature at a sweeping rate of 10 mV/s. ORR activities are compared at 0.9 V vs RHE at 1600 rpm. b dn ) (∑in) 1 di)/n. c dn/a ) (∑in) 1 di3)/(∑in) 1 di2).

Figure 2. Polarization curves of ORR collected from Pt-4nm (a), AT-“Pt3Co” (b), and HT-AT-“Pt3Co” (c) in an O2-saturated 0.1 M HClO4 solution with a sweeping rate of 10 mV/s at room temperature. (d) Tafel plots of ORR specific activity obtained from the polarization curves in the positive-going scans at 1600 rpm, which was normalized to Pt electrochemical surface area (Figure 1). The loadings of these samples on glassy carbon electrodes were 9, 11, and 12 µgPt/cmGCE2 for Pt-4nm, AT-“Pt3Co”, and HT-AT-“Pt3Co”, respectively.

phase model (the calculated and experimental data of select peaks are shown in Figure S2), from which the average lattice parameter of the FCC structure was determined. The average Pt-Pt bond length of Pt-4nm nanoparticles is slightly shorter than that of bulk Pt [2.7740(1) Å] found in this study and previous studies,25,28,59 which has been attributed to the relaxation of nanoparticle surface atoms with low coordination sites.60-62 The average Pt-Pt bond length of AT-“Pt3Co” and HT-AT“Pt3Co” nanoparticles was 2.712(3) and 2.724(2) Å, respectively, which are in agreement with those reported previously for “Pt3Co” nanoparticles (∼7 nm, 2.725 Å)25,28 and for bulk ordered and disordered “Pt3Co”.63-66 It is interesting to note that the average Pt-Pt bond length of “Pt3Co” nanoparticles found in this study is in agreement with a linear trend in the nearest atomic distance with increasing Co atomic concentration between Pt and PtCo, as shown in Figure 3b. Moreover, the volume-averaged particle size of each sample was evaluated from the extent of observed peak broadening. The particle sizes of Pt-4nm, AT-“Pt3Co”, and HT-AT-“Pt3Co” nanoparticles were found to be 4, 5, and 5 nm from full peak width at halfmaximum of the {111} reflection using the Scherrer formula.67 The volume-averaged particle sizes of these samples will be

compared and discussed in the context of particle size histograms obtained from HRTEM measurements. Weak but visible superlattice diffraction peaks such as {110}, which are forbidden in the parent FCC structure (Fm3jm), were noted in the HT-AT-“Pt3Co” (Figure 3a and Supporting InformationFigure S3a). The refinement of X-ray powder diffraction data of HT-AT-“Pt3Co” was improved considerably by employing a two-phase model relative to one FCC structure: (1) a major phase (∼90% by volume and average Pt-Pt interatomic distance of 2.730 Å), which has a smaller volume-averaged particle size of ∼5 nm, and (2) a minor phase (∼10% by volume and average Pt-Pt interatomic distance of 2.695 Å), which has larger volume-averaged particle sizes of 10-15 nm. Comparison between the experimental and the calculated spectra using the two-phase model can be found in the Supporting Information in Figure S3b. It is hypothesized that annealing of AT-“Pt3Co” nanoparticles leads to the appearance of Pt-rich major and Ptpoor minor phases upon high-temperature annealing, which will be further discussed in the context of HRTEM data in the later sections. As the specific ORR activity of “Pt3Co” nanoparticles scales with the surface area of nanoparticles, the influence of the minor phase (having a much larger volume-averaged particle

1114 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Figure 3. (a) Synchrotron X-ray diffraction patterns of Pt-4nm, AT“Pt3Co”, and HT-AT-“Pt3Co” from which a volume-averaged particle size was obtained for each sample. (b) Average interatomic distance as a function of Pt atomic percentage. Open large circles show the interatomic distances obtained by synchrotron X-ray diffraction data of bulk Pt, Pt-4nm, AT-“Pt3Co”, and HT-AT-“Pt3Co” in this study. The interatomic distances for bulk Pt and PtCo alloys63-66 are plotted as hatched symbols for comparison.

size than the major phase, less than 5% total surface area) on the specific ORR activity is considered minimum. 3.3. Sizes and Size-Dependent Compositions of “Pt3Co” Nanoparticles. Histograms of ∼200 randomly selected nanoparticles and representative HRTEM images of AT-“Pt3Co” and HT-AT-“Pt3Co” samples are shown in Figures 4a and 4c, respectively. Although most particles were found to have diameters of ∼4 nm in both “Pt3Co” samples, a few crystals larger than 10 nm were noted, which resulted in a tail in the particle size distribution toward the large particle size end. Unlike “Pt3Co” nanoparticles, Pt-4nm nanoparticles were shown to have relatively a narrow size distribution centered at ∼4 nm (shown in the Supporting Information Figure S4). As shown in Table 1, the number-averaged diameters of for Pt-4nm, AT“Pt3Co”, and HT-AT-“Pt3Co” were 3.6, 4.1, and 4.9 nm, respectively. These values are consistent with the volumeaveraged particle diameters from synchrotron X-ray diffraction. In addition, volume/area averaged diameters,41 and Pt-specific surface area based on TEM data, are listed in Table 1 for comparison, which is in agreement with electrochemical surface area of these three samples. Both “Pt3Co” samples were found to have an average atomic Pt/Co ratio of 78/22, which was determined from STEM EDS analysis of Pt LR and Co KR signals collected from regions, typically 3 × 3 µm, that consisted of numerous nanoparticles at low magnifications (∼20 000 times). However, Pt/Co ratios

Chen et al. of individual “Pt3Co” nanoparticles were found to greatly deviate from the average and to decrease as a function of particle size. Co atomic fractions at the center of AT-“Pt3Co” and HT-AT“Pt3Co” nanoparticles using a static 2 nm electron beam are plotted as a function of particle size, as shown as open circles in Figures 4b and 4d, respectively, which were determined from averaging raw data over a bin size of 2 nm. In the AT-“Pt3Co” sample, the Co atomic fraction at the particle center was found to change linearly from ∼15 to ∼40 at % from ∼2 to ∼15 nm. Similar size-dependent Co compositions have been reported previously in “PtCo” nanoparticles after corrosion tests in H3PO4 at 210 °C.23 HT-AT-“Pt3Co” was found to have the Co atomic fractions (at the particle center) vary considerably less as a function of particle size relative to AT-“Pt3Co”. Co atomic fraction increased from ∼15 at % at ∼5 nm and gradually approached ∼25 at % at ∼20 nm, as shown in Figure 4d. Average Co atomic fractions of individual AT-“Pt3Co” and HT-AT-“Pt3Co” nanoparticles were determined from spectra obtained from scanning a 2 nm electron beam over the entire particle, which are plotted as filled circles in Figures 4b and 4d, respectively. It is interesting to note that average Co atomic fractions (filled circles) of individual AT-“Pt3Co” nanoparticles were shown to considerably deviate from those found at the particle center (open circles) while average Co atomic fractions of individual HT-AT-“Pt3Co” particles were found to be in reasonably good agreement with those at the particle center in Figure 4d. This observation suggests that the distribution and composition of Co is not uniform within individual AT-“Pt3Co” particles while the composition is becoming more uniform (averaged over an area with 2 nm diameter) upon annealing. Previous in situ TEM work68 has shown that coalescence of 5 nm Pt nanoparticles on carbon support can occur at temperatures as low as 428 K. Coalescence of AT-“Pt3Co” nanoparticles with different compositions during annealing, where small Pt-rich particles join large Pt-poor particles, would reduce sizedependent composition variations and lead to more uniform compositions among individual particles in the HT-AT-“Pt3Co” sample. As such analysis is insensitive to chemical composition variations in the near-surface regions, we then examine the chemical composition from multiple locations across individual nanoparticles using a 2 nm electron beam. 3.4. Direct Evidence of Pt Enrichment in the Near-Surface Region of “Pt3Co” Nanoparticles. 3.4.1. ConWentional STEM EDS Results. Using a static beam of 2 nm and a collection time of 30 s (to minimize particle drift during data acquisition), X-ray spectra collected from different locations within a given nanoparticle can reveal Pt surface enrichment on the nanometer scale. A large number of AT-“Pt3Co” nanoparticles greater than 10 nm (∼50% among 38 particles analyzed) were found to exhibit Pt enrichment. One example is shown in Figure 5a, where four locations marked 1-4 on a ∼10 nm particle were analyzed. The individual spectra from these locations are shown in Figure 5b, where X-ray intensities detected near the particle edge are weaker relative to those near the center. It is noted that the Co atomic fractions near the center are considerably higher (∼37%) than those found near the edge (∼27%). For AT-“Pt3Co” nanoparticles smaller than 10 nm, Pt enrichment on the surface was found in fewer nanoparticles (∼18% among 11 particles analyzed) by this technique with experimental certainty. Similar analyses were performed on HT-AT-“Pt3Co” nanoparticles, where Pt enrichment on the surface was detected on only 1 particle among 46 particles by STEM EDS. This result and the data in Figure 4d suggest that the distributions of Pt and Co become more uniform (over an area of 2 nm in diameter)

ORR Activity on “Pt3Co” Nanoparticles

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1115

Figure 4. (a) Particle size histogram and (b) size-dependent Co atomic percentages of AT-“Pt3Co”. (c) Particle size histogram and (d) sizedependent Co atomic percentages of HT-AT-“Pt3Co”. Percentages collected from particle center (solid circle in Figure 3c, inset) from STEM EDS analysis, which was then averaged over a bin size of 2 nm, are shown in open circles. Data in solid circles were obtained from the whole particle (dashed circle in Figure 3c, inset).

Figure 5. (a) Representative HAADF STEM image of AT-“Pt3Co” nanoparticles. EDS analysis in the spot capture mode was performed in four different positions (shown as white dots labeled 1-4) of the particle using an electron beam of 2 nm diameter for 30 s. (b) EDS spectra and Pt/Co atomic ratios of positions 1-4, where Co KR and Pt LR peaks were used for quantification. The peaks at ∼8 keV correspond to Cu KR from the TEM copper grid.

within individual and among different “Pt3Co” particles during annealing of AT-“Pt3Co” at 1000 K. As this analysis has a spatial resolution of 2 nm, it is not sensitive to compositional variations on the atomic scale. To examine compositional variations from one atomic column to the next and composition oscillations in the top two to three surface layers, we have used aberration-corrected STEM imaging to study “Pt3Co” particles, and we discuss the results below. 3.4.2. Aberration-Corrected STEM Imaging. As the HAADF detector captures electrons that are inelastic scattered to higher angles, the contrast in the HAADF images is proportional to the product of the square of average atomic number and

thickness and is not affected by elastically scattered electrons (no diffraction contrast). As shown in Figure 6a, regions where nanoparticles overlap are clearly brighter in dark-field mode due to an increase in thickness, whereas in bright-field mode the contrast is significantly reduced (Figure 6b). In addition, it should be noted that carbon support particles exhibit very weak intensities in dark-field mode, to a point where single Pt atoms and very small Pt clusters can be observed (Figure 6c). In contrast, the diffraction contrast of carbon support structure in bright-field mode prevents individual Pt atoms and clusters from being seen (Figure 6d). Single Pt atoms and clusters were found only in AT-“Pt3Co” but not in HT-AT-“Pt3Co”. It is postulated

1116 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Chen et al.

Figure 6. Typical aberration-corrected HAADF and BF images at a low magnification (a) and (b) and at a high magnification (c) and (d) of AT-“Pt3Co” particles supported on high-surface-area carbon. The images were acquired on a JEOL 2200 FS aberration-corrected STEM. The dash line in (c) marks the edge of high-surface-area carbon support particle, where individual Pt atoms are visible in (c), as indicated by white arrows.

that Pt atoms and clusters were highly mobile during the annealing step and they migrated, coalesced, and joined large Pt nanoparticles. Aberration-corrected STEM HAADF images not only can reveal individual columns of atoms projected in the beam direction but also can provide direct evidence for variation in the chemical composition within individual particles of AT“Pt3Co”. FFTs of HAADF images showed that most AT-“Pt3Co” particles were disordered (in solid solution, where Pt and Co atoms are randomly distributed in the FCC structure), which is consistent with synchrotron powder X-ray diffraction results in Figure 3a. Two examples are shown in Figure 7a, b. In darkfield mode, notable intensity variations across individual particles were found for many particles with particle sizes greater than 10 nm. Normalized HAADF intensities of the particle along the dashed line in the inset of Figure 7b show a change of 2.4 times between the highest and lowest intensity spots. Such intensity variations suggest that thickness changes (porosity development) up to 2.4 times for a given composition, and/or changes occur in the chemical compositions at the atomic scale up to a compositional change from Pt to Pt0.4Co0.6 for a given thickness. Considering Co dissolution from a PtCo alloy that is comparable to that of the parent sample of AT-“Pt3Co” before the acid treatment, we here discuss two limiting scenarios of acid removal of Co from two atomic columns with Pt50Co50

(having an averaged atomic number of Z0 ) 52.5) as the starting composition (Figure S5). In order to predict maximum intensity contrast between these two columns in the STEM images, it is assumed that the Co atoms in column 1 dissolved completely but column 2 is unaffected by the acid treatment. In one case, vacancy sites are left after the removal of Co in column 1 and the thickness of column 1 is the same as the thickness of column 2 (Figure S5), where the average atomic number of column 1 is half of that of Pt (Z1 ) ZPt/2 ) 39), and this leads to higher intensity for column 2 than column 1 in the HAADF image, with I2/I1 ) 1.8. In the other case, Pt atoms form a dense atomic column (formation of porosity at the atomic scale) with no vacancy in the structure when Co atoms are removed from column 1, where having Z1 ) ZPt ) 78 and Z2 ) Z0 ) 52.5 for a given thickness if one neglects the size difference between Pt and Co leads to higher intensity in column 1 relative to column 2 with I1/I2 ) 1.1. This analysis indicates that dissolution alone is not able to explain the observed intensity variations in Figure 7b. Therefore, in addition to Co removal, Pt diffusion from one column to the other is needed to create intensity ratios comparable to those observed in the HAADF images. It is proposed that the dissolution of Co is not confined near the surface, and these AT-“Pt3Co” nanoparticles of 10 nm and greater have percolated Pt-rich (possibly thicker) and Pt-poor (possibly thinner) regions with sizes in the range of 1-2 nm.

ORR Activity on “Pt3Co” Nanoparticles

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1117

Figure 7. Aberration-corrected HAADF images of two AT-“Pt3Co” particles, where large intensity variations are noted. (b) Inset: intensity variation of an AT-“Pt3Co” particle along the dashed line. The intensities at position 1 (near the center) and 2 (near the edge) are compared, where the intensity ratio of I2 and I1 is 2.4.

Such nanostructure development has been reported by a number of previous studies,69-71 which have shown that dealloying of a less noble element from an alloy in acid can be accompanied by clustering of more noble atoms. Although the HAADF intensities across the particle were found to be more uniform for particles with sizes of 5 nm and smaller than that observed for larger particles, percolated Ptrich and Pt-poor regions were also found within individual particles. Several particles show intensity variations that cannot be explained by thickness variations associated with particle shape. HAADF images of two AT-“Pt3Co” particles of ∼4 nm are shown in Figure 8a, b, where FFTs of the images are included in the inset. Normalized HAADF intensity variation of these particles along the dashed line in Figure 8a, b, which was measured with Digital Micrograph, is compared with the normalized thickness variation of a truncated octahedron projected in the corresponding orientation in Figure 8c, d, respectively. AutoCAD 2009 (Version C.56.0, AutoDesk, Inc., San Rafael, CA) was used to create a truncated octahedron (Figure 8c, d insets) in the same orientation as the STEM image, where thickness changes of the truncated octahedron projected along the dashed line in the beam direction (in or out of paper) were determined. As the HAADF intensity is proportional to the product of thickness and square of atomic number averaged over the column of atoms along the beam direction, the normalized atomic number along the dashed line was obtained from the normalized thickness and HAADF intensity data in Figure 8c, d, as shown in Figure 8e, f, respectively. It is evident that some regions close to the particle center have lower average atomic numbers than do the surface regions. It should be noted that estimated atomic numbers for the outermost atomic columns are not meaningful due to uncertainty relative to those of other columns in this analysis, which are shaded in gray in Figure 8e, f. It is reasonable to suggest that the regions with high atomic numbers are Pt-rich relative to the regions with lower atomic numbers, which may consist of higher amounts of Co atoms and vacancies. Therefore, normalized atomic number profiles in Figure 8e, f provide direct evidence for the formation of a percolated structure of Pt-enriched and Pt-poor regions in the AT-“Pt3Co” nanoparticles. The regions of Co removal upon acid leaching are not confined near the surface, and on average, near-

surface regions are Pt-rich relative to the particle interior, as shown in Figure 8g. With no direct evidence, several previous studies23,26 have proposed a core-shell structure (unleached regions in the core, and Co-removed, Pt-enriched surface regions for the shell) for acid-treated Pt alloy nanoparticles, which is supported by a very recent study of anomalous small-angle X-ray scattering.72 The percolated structure proposed for AT“Pt3Co” nanoparticles in this study is different from the previously proposed core-shell structure in that Pt-enriched regions can extend from surface regions into the particle core. It should be mentioned that these acid-treated Pt alloy nanoparticles23,26 were obtained under very different treatment conditions and from very different Pt alloy compositions, where different leached structures might form. Further aberrationcorrected STEM studies of acid-treated Pt alloy nanoparticles from Pt alloys of different compositions and different treatment conditions are needed to provide direct evidence for the leached structures and address the influence of alloy compositions and leaching conditions on the resulting leached particle structures. Some evidence of the core-shell morphology (Pt-enriched shell with higher Z than particle core) proposed in previous studies23,26 was noted for ordered “Pt3Co” nanoparticles found in the acid-treated sample. STEM HAADF imaging found very few AT-“Pt3Co” nanoparticles exhibiting ordering which could be indexed consistently to space group Pm3jm from the FFTs of STEM images. One example is shown in Figure 9. The HAADF intensities of the first seven layers along the dashed line in Figure 9a are plotted in Figure 9b, where the intensities in the first three layers in Figure 9a exhibit a linear increase. Starting from the fourth layer in Figure 9a, the intensity oscillates, which is associated with the ordering of Pt-rich (bright lines) and Co-rich planes (dark lines) of the ordered structure (space group Pm3jm). Although this particle could not be modeled by a truncated octahedron, it is reasonable to suggest that the linear intensity change can be attributed primarily to thickness variations associated with the particle morphology, and the particle consists of a Pt-enriched shell in the first two surface layers relative to particle interior. The results of Figures 8 and 9 suggest that acid leaching of ordered and disordered Pt alloy nanoparticles may lead to different nanoparticle structures (percolated for disordered vs core-shell for ordered). Watanabe

1118 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Chen et al.

Figure 8. (a) and (b) Aberration-corrected HAADF images of two AT-“Pt3Co” particles, where FFTs of the HAADF images are shown in the insets. (c) and (d) Solid curve: normalized intensity variations of columns of atoms (along the beam direction) in (a) and (b) along the dotted lines, respectively. Dashed curve: normalized thickness variations of the truncated octahedron shown in the insets (c) and (d), respectively. The image intensity was measured with Digital Micrograph, and thickness variation of the truncated octahedron was measured with AutoCAD. (e) and (f) Normalized atomic number averaged over each column of atoms in the beam direction, which is defined as the square root of the ratio of normalized integrated intensity in (a) and (b) and normalized thickness in (c) and (d), respectively. The normalized atomic numbers near the particle edges, the first and last columns in (e) and (f), contain large uncertainties in the analysis, which are shaded in gray. (g) Scheme that shows acid removal of Co from a uniform PtCo nanoparticle rendering a percolated nanostructure with Pt-rich and Pt-poor regions.

ORR Activity on “Pt3Co” Nanoparticles

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1119

Figure 9. (a) Aberration-corrected HAADF image of one ordered particle with Pm3jm symmetry in the AT-“Pt3Co” sample. FFT of the image in the inset shows superlattice reflections marked by arrows. (b) Normalized integrated intensity of the first seven atomic columns of the particle along the dashed line in (a). The observed linear intensity variation near the surface is more probably attributed to a thickness change, which suggests that the surface regions (the first three layers) are Pt-rich relative to the particle interior.

et al.23 have previously suggested that a thicker layer of Ptenriched surface regions in the notation of this study for ordered relative to disordered PtCo nanoparticles. Further studies are needed to examine in detail the difference in the evolution of nanoparticle structure of ordered and disordered Pt alloy nanoparticles during acid treatments. Aberration-corrected HAADF images of some HT-AT“Pt3Co” nanoparticles revealed the existence of Pt segregation in the top one to three surface layers (bright line) and Co enrichment (dark line) in the subsurface layers. The particle in the bottom right corner of Figure 10a shows that the outermost {100} plane is considerably brighter than the second plane beneath the surface. The normalized intensities for the first seventh atomic layer along the dashed line in Figure 10a and the corresponding normalized thickness changes of a truncated octahedron are compared in Figure 10b, from which the normalized atomic numbers can be obtained. As shown in Figure 10c, it is evident that the top {100} surface is Pt-rich and the second layer beneath is Co-rich. Although Pt segregation is wellknown on bulk Pt polycrystalline and single-crystal alloy surfaces upon high-temperature annealing,3,9-13 we show, for the first time, that Pt segregation can occur on the {100} surface of ordered “Pt3Co” nanoparticles of ∼5 nm upon annealing at 1000 K. In addition to surface segregation, FFT of the particle near the bottom right of Figure 10a can be indexed consistently to ordered “Pt3Co” with space group Pm3jm along the [1j30] zone axis. In fact, a large number of particles in HT-AT-“Pt3Co” were found to exhibit superlattice ordering with the Pm3jm symmetry according to both synchrotron X-ray diffraction analysis (Figure S3a) and FFT analyses of STEM images. Moreover, the {100} surfaces become more prevalent in nanoparticles of ∼5 nm after annealing, which is in good agreement with the cyclic voltammogram data having pronounced adsorption and desorption peaks at 0.3 V vs RHE characteristic of the {100} surfaces.73 The ordering and Pt segregation on the {100} surface were found to be more evident in particles shown in Figure 11a, b, where normalized intensities for the first seventh atomic layer along the dashed line in are show in Figure 11c, d, respectively. Although these particles could not be modeled by a truncated octahedron or a cuboid, it is reasonable to suggest that the linear intensity change in the top two surface layers can be attributed primarily to thickness

variations associated with the particle morphology, and it is most likely that Pt segregation or Pt enrichment occurs in the top two {100} surface layers. Similar Pt segregation profiles have been reported for the {100} surface of PtNi by Gauthier et al.9 Pt segregation on the {111} surfaces predicted on cuboctahedral Pt3Ni nanoparticles74 was not observed in this study. This difference may be attributed to the facts that the simulations74 were performed on disordered FCC Pt3Ni nanoparticles and surface adsorbents on “Pt3Co” nanoparticles during hightemperature annealing can influence the formation of nanoparticle facets.75 3.5. Surface Atomic Structure of Pt and “Pt3Co” Nanoparticles. Indexed surface atomic planes of representative particles from Pt-4nm and AT-“Pt3Co” are shown in Figure 12a, b, respectively. Not only were low-index surfaces such as {111} and {100} observed, but also some high-index surfaces such as {221} and {331} were noted. Surface atomic planes from at least 12 representative particles from each sample were analyzed, from which the average fractions of particle perimeter corresponding to low-index and high-index surface facets projected along the direction were calculated, as shown in Table 2. After high-temperature annealing, a considerable fraction (25% by number from HRTEM images) of HT-AT-“Pt3Co” nanoparticles was found to exhibit pronounced {100} surface facets (Figures 10 and 11), and the fractions of high-index surface facets were reduced significantly. As the particle shape could not be modeled by the truncated octahedron, the quantification in the area fraction of different surface facets for HTAT-“Pt3Co” nanoparticles is not straightforward and thus not included in Table 2. The surface area fractions of individual facets on a truncated octahedron are comparable to the corresponding length fractions of particle perimeter projected along the direction (shown in Figure S6). Provided that Pt-4nm and AT-“Pt3Co” nanoparticles have particle shape that resembles a truncated octahedron, the fractions of particle perimeter observed for lowindex and high-index surfaces can be used to quantitatively compare the area fractions of different nanoparticle facets. The fractions of particle perimeter for the low-index surfaces such as the {111} and {100}, and high-index surfaces including the n{111} × {111} and n{111} × {100} types, which were averaged over a number of Pt-4nm and AT-“Pt3Co” nanopar-

1120 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Figure 10. (a) Aberration-corrected HAADF image of annealed “Pt3Co” nanoparticles. FFT of the particle outlined by the dotted box is shown in the inset, where superlattice reflections marked by the arrows can be indexed consistently with the Pm3jm symmetry. (b) Normalized intensities (solid line) of the first seven atomic layers in (a) along the white dashed line and normalized thickness (dashed line) estimated from an ideal truncated octahedron model in the same orientation, as shown in the inset of (b). (c) Normalized atomic number of each atomic column along the dashed line in (a) in the beam direction, which was calculated by the square root of the ratio of normalized intensity and thickness.

Chen et al. ticles, are shown in Figure 13a, b, respectively. The difference in the surface atomic structure of these two samples is subtle. The {111} surfaces are the most commonly present, and the fractions of the {110} and {113} surfaces are comparable in both samples. AT-“Pt3Co” nanoparticles have a lower fraction of the {111} but a higher fraction of the {100} surfaces. This is in good agreement with the CV data, where the hydrogen adsorption and desorption peak at 0.3 V vs RHE, which is characteristic of the {100} surfaces,73 was found to be larger for AT-“Pt3Co” than for Pt-4nm. In addition, higher fractions of the {111}-type and {100}-type steps on the {111} terrace relative to Pt-4nm were noted. The influence of surface atomic structure on surface reactivity and the specific activity of ORR will be discussed in later sections. 3.6. Adsorbed Oxygenated Species on the Surface of “Pt3Co” Nanoparticles at Ambient. Normalized Pt L2,3-edge XANES spectra of AT-“Pt3Co” and HT-AT-“Pt3Co” nanoparticles are compared with those of bulk Pt in Figures 14a and 14b. It is interesting to note that the white line intensity of AT“Pt3Co” is higher than those of bulk Pt and HT-AT-“Pt3Co” while that of HT-AT-“Pt3Co” is comparable to the bulk Pt white line. The increase in the Pt L2,3 white line intensity for AT“Pt3Co” relative to bulk Pt indicates greater probability of the transition from 2p3/2 (fully filled) to 5d5/2 and 6s (partially filled) orbitals, which can be associated primarily with reduced filling of electrons at the 5d orbital near the Fermi level if a broad distribution of 6s orbital (partially filled) were assumed. Previous studies of supported Pt nanoparticles have shown that chemical adsorption of oxygenated species such as OH and O25,28,76-79 on Pt can lead to the appearance of unoccupied antibonding states above the Fermi level, and thus an increase in the L2,3 X-ray absorption edge for AT-“Pt3Co” nanoparticles can be attributed to higher coverage of adsorbed oxygenated species on nanoparticle surfaces relative to bulk Pt surface.79 This is also in good agreement with the presence of a small but visible peak at ∼1.5 Å that can be assigned to Pt-O in the radial structure function in Figure 1579,80 and Table 3. On the other hand, lower white line intensity of HT-AT-“Pt3Co” nanoparticles relative to bulk Pt in Figure 14a suggests that the HT-AT“Pt3Co” nanoparticle surface is less reactive toward adsorbed oxygenated species relative to AT-“Pt3Co” and comparable to bulk Pt79,80 at ambient. To further understand the effect of adsorbed oxygenated species on the surface, ∆µ spectra of Pt L2,3 edges of “Pt3Co” nanoparticles normalized relative to bulk Pt, where adsorption of oxygenated species is considered negligible compared to AT-“Pt3Co”, are shown in the insets of Figure 14a, b, respectively. These ∆µ spectra for the L3 edge are similar to those of “Pt3Co” nanoparticles obtained in situ at 0.84 V vs RHE (normalized relative to 0.54 V vs RHE) reported previously.79 The ∆µ spectrum of AT-“Pt3Co” nanoparticles exhibited a large peak centered around ∼5 eV above the L3 edge, which further suggests the presence of adsorbed oxygenated species.79 Moreover, the d band vacancy was found to increase from 0.30 for bulk Pt to 0.37 for AT-“Pt3Co” nanoparticles in Table 3. It is interesting to note that the d band vacancy calculated for HT-AT-“Pt3Co” (0.32) is smaller than that reported for Pt and Pt alloy nanoparticles (0.330.41)25,28,78 measured at 0.80 and 0.54 V vs RHE, and it is comparable to bulk Pt, where surface oxygenate adsorbates can be considered minimal. This is in good qualitative agreement with observed surface atomic structures of HT-AT-“Pt3Co” nanoparticles (Figures 10 and 11), where the fractions of highindex surface facets are much smaller than those of AT-“Pt3Co” nanoparticles (Figures 12 and 13).

ORR Activity on “Pt3Co” Nanoparticles

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1121

Figure 11. (a) and (b) Aberration-corrected HAADF images of two particles of HT-AT-“Pt3Co”, where corresponding FFT images are shown as insets with superlattice reflections of space group Pm3jm marked by white arrows. (c) and (d) Normalized integrated intensities of the first seven atomic columns of the particle along the dashed line in (a) and (b) in the beam direction, respectively. The surface region in the first three atomic layers is believed to be Pt-rich as linear intensity variations are most likely attributed to the thickness changes associated with particle shape rather than variations in the atomic number.

Analysis of EXAFS data at the Pt L3 edge allows the determination of the average Pt-Pt bond length of “Pt3Co” samples, which were compared with that of bulk Pt. The radial structure functions of bulk Pt and AT-“Pt3Co” and HT-AT“Pt3Co” nanoparticles is shown in Figure 15, where the peak at ∼2.7 Å corresponds to the first-nearest-neighbor Pt-Pt coordination. The integration range for the Fourier transform, the window in the r-space used for the inverse Fourier transform, the Pt-Pt coordination number, and the Debye-Waller factor are shown in Table 3. Both “Pt3Co” samples have average Pt-Pt bond length considerably smaller than bulk Pt, and the average Pt-Pt bond length of HT-AT-“Pt3Co” was found to be slightly larger than that of AT-“Pt3Co”, which is in agreement with synchrotron X-ray powder diffraction data in Figure 3. 4. Discussion 4.1. Average Nanoparticle Pt-Pt Bond Length and Enhanced ORR Activity. Jalan and Taylor30 have first proposed that the shortening of average Pt-Pt interatomic distances in alloy nanoparticles supported on carbon black leads to ORR activity enhancement. Mukerjee and Srinivasan28 have subsequently supported this view and have shown a volcano relation-

ship between the Pt-Pt bond length or d band vacancy and the ORR activity. However, X-ray powder diffraction and EXAFS analyses used in these previous studies reveal the Pt-Pt bond length averaged over entire nanoparticles, which may considerably deviate from the Pt-Pt bond length in the first two to three surface layers of Pt alloy nanoparticles. For example, Toda et al.7 have suggested that the Pt-Pt bond distance on the surface of acid-treated alloys may be similar to that of Pt. In this study, despite the fact that AT-“Pt3Co” and HT-AT-“Pt3Co” nanoparticles have very similar average Pt-Pt interatomic distances in the range of 2.72-2.74 Å, which falls in the region for maximum ORR activity,28 these two samples exhibit different levels of enhancement in specific ORR activity relative to Pt nanoparticles. The difference can be attributed to dissimilar surface atomic structures and compositional distributions near the particle surface, which will be discussed in detail below. 4.2. Enhanced ORR Activity of Acid-Treated Pt Alloy Nanoparticles Relative to Pt. 4.2.1. Surface Atomic Structure Effects. It is of great interest to understand the influence of surface atomic structure on the surface reactivity and specific ORR activity of “Pt3Co” nanoparticles relative to Pt. The equilibrium shape of a Pt crystal at zero temperature is a

1122 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Figure 12. Representative HRTEM images and surface structures of (a) Pt-4nm and (b) AT-“Pt3Co” nanoparticles, where corresponding FFT images are shown in the insets. The surface atoms are marked by white dots, and the corresponding planes are indexed.

truncated octahedron consisting predominantly of the {111} and {100} facets by Frenken et al.81 Pt and AT-“Pt3Co” nanoparticles can be described generally as a truncated octahedron.32,82 The increased d band vacancy and surface oxygen coverage of AT“Pt3Co” nanoparticles relative to bulk Pt (Figure 13 and Table 3) can be attributed to the fact that increasing the number of undercoordinated atoms of nanoparticles (such as edges and corner)32 can increase the reactivity toward oxygenated species such as O.83,84 In addition, AT-“Pt3Co” nanoparticles have relatively higher fractions of the {111} and {100} steps on the {111} terrace relative to Pt-4nm nanoparticles (Figure 13). This observation suggests that the surface of “Pt3Co” nanoparticles is qualitatively rougher than Pt-4nm, which is in general agreement with previous work31 and with the observation that AT-“Pt3Co” nanoparticles exhibit considerably higher coverage of oxygenated species than Pt-4nm at voltages greater than 0.8 V vs RHE in the negative-going scan (Figure 1c). It is not straightforward to relate the enhancement in the ORR activity of “Pt3Co” nanoparticles to the changes in the surface atomic structure relative to Pt. We here examine the influence of surface atomic facets on the specific ORR activity of AT-“Pt3Co”

Chen et al. nanoparticles in some detail. On one hand, experimental studies of bulk Pt surfaces have shown that the ORR activity of Pt{100} in weakly adsorbing HClO43,73,85 is smaller than that of Pt{111}, and increasing the {100} surface fraction on AT-“Pt3Co” nanoparticles would decrease ORR activity relative to Pt-4nm nanoparticles. On the other hand, increasing the {100} and {111} step densities on the {111} and {100} terraces has shown to improve ORR activity in 0.1 M HClO4.45,46 It should be noted that these experimental findings are not in agreement with recent DFT results,86 which show that lowering the coordination number of surface atoms from low-index to high-index surfaces leads to stronger binding of oxygenated species and thus yields lower ORR activity. For example, high-index {211} surface with {100} steps on the {111} terrace is predicted to have lower ORR activity than the {111} surface.86 If one directly maps the activity values of bulk stepped surfaces45,46 onto the nanoparticle surface (Table S1), the specific ORR activities of Pt-4nm and AT-“Pt3Co” nanoparticles were found to be comparable (Table 2). The increasing fraction of the {100} surfaces relative to the {111} surfaces in decreasing ORR activity counterbalances increasing fractions of high-index surfaces with higher ORR activity. This analysis suggests that the specific ORR activity enhancement of AT-“Pt3Co” relative to Pt-4nm cannot be attributed to differences in the atomic structure alone. Further studies are required to clarify the discrepancies that exist on the effect of the surface atomic structure (such as step density and the nature of steps) on the ORR activity.45,46,86 4.2.3. Percolated Nanoparticle Structure. In this study, aberration-corrected STEM images have provided direct experimental evidence in the surface compositional variations within individual nanoparticles on the nanometer scale. In particular, analysis of observed HAADF intensity variations (Figures 8 and 9) indicates the formation of percolated Pt-rich surface and Ptpoor core regions within individual “Pt3Co” nanoparticles, where on average near-surface regions can be Pt-rich relative to particle core. As a result, the interface between surface Pt-rich and core Pt-poor regions can result in compressive strains and shortened surface Pt-Pt bond distance compared to Pt nanoparticles of similar particle sizes. Fourier filtering of Figure 8a clearly shows distortion along the {111} planes, which is indicative of strain within the particle (Figure S7). Further analysis is needed to determine strain distributions with individual particles. Such a structure can be analogous to the skeleton structure proposed previously for polycrystalline Pt alloy surfaces.11 Compressive strains and ligand effects associated with such percolated nanostructures can lead to d band broadening and lowered valence band center relative to the Fermi level.17,19,87 It is postulated that the change in the electronic structure induced by the percolated nanoparticle structure can reduce the binding energy of oxygenated species,16 thus lowering the activation barrier for the rate-limiting step and increasing the activity for ORR6 relative to Pt nanoparticles. It should be mentioned that although Pt-4nm and AT-“Pt3Co” nanoparticles exhibit comparable coverage of oxygenated species on the positive-going scan (Figure 1b), recent DFT studies have shown that direct correlation between ORR activity and OH coverage from water activation is not straightforward.88 Therefore, the enhanced activity of AT-“Pt3Co” nanoparticles can be attributed primarily to the ligand effect and compressive strains associated with a percolated nanoparticle structure with Pt-rich and Pt-poor regions (Figure 8g). Further ultraviolet photoelectron spectroscopy of acid-treated Pt alloy nanoparticles is needed to understand the changes in the surface electronic structure relative to Pt and Pt alloy nanoparticles before acid treatments. It is interesting to note that the particle core-shell development proposed previously7,22,23,26 upon acid removal of transition metal

ORR Activity on “Pt3Co” Nanoparticles

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1123

TABLE 2: Observed Surface Facets of Pt-4nm and AT-“Pt3Co” Nanoparticlesa surface atomic plane index {111} {100} {110} {113} {112} {335} {223} {115} {117} {331} {221} {553} jb at 0.9 V (vs RHE) in 0.1

surface atomic plane notation

percentage of particle perimeter Pt-4nm

percentage of particle perimeter AT-“Pt3Co”

j at 0.9 V vs RHE (mA/cm2) in 0.1 M HCIO4

2{111} × {111} 2{111} × {100} 3{111} × {100} 4{111} × {100} 5{111} × {100} 3{100} × {111} 4{100} × {111} 3{111} × {111} 4{111} × {111} 5{111} × {111} M HCIO4 (mA/cm2)

60.6 17.9 5.3 6.3 3.5 0 0.7 0.7 0.9 3.1 0.6 0.5 8.2

50.0 21.3 7.2 4.4 3.5 2.5 0.7 2.7 0 4.8 2.5 0.5 9.2

8.2 1.7 7.7 2.1 25.2 17.4 15.6 1.5 1.9 35.5 31.2 25.3

The normalized projected length of each facet to the particle perimeter along the direction was calculated by (1) projecting i high-index planes into stepped {111} and {100} surfaces and (2) summing the projected length of the {111} and {100} planes. b j0.9V ) Σfij0.9V , i the ORR activity of a given plane from stepped single-crystal studies reported by Feliu with fi the surface area fraction of each plane and j0.9V et al.45,46 (listed in Table S1). a

TABLE 3: Structural and Electronic Structural Parameters Obtained by EXAFS Analysis, Ranges of the k3-Weighted Forward and Inverse Fourier Transform at the Pt L3 Edge for Bulk Pt, AT-“Pt3Co”, and HT-AT-“Pt3Co”, and the Pt d Band Vacancy (hTs)a bulk Pt AT-“Pt3Co” HT-AT-“Pt3Co” a

R (Å)

σ2 × 10-3(Å2)

k-range (Å-1)

R-range (Å)

R-factor

hTs

11.0(3)

2.766(1)

5.1

2.1-18.1

1.0-3.1

0.0016

0.30c

10.7(16)b 1.3(7) 11.7(15)b

2.718(7) 2.00(2) 2.739(6)

8.3 4.7 6.3

1.9-16.0

1.0-3.05

0.0511

0.37

1.9-15.4

1.0-3.05

0.0419

0.32

coordination shell

C. N.

Pt-Pt Pt-O Pt-Pt and Pt-Co Pt-O Pt-Pt and Pt-Co Pt-O

Obtained from XANES data of the Pt L2,3 edge. b Co/Pt ratio has been fixed to be 1/3. c hTs has been calculated based on 0.3 for bulk Pt.

Figure 13. Average fractions of projected length for low-index and high-index surface planes relative to the particle perimeter, which were obtained from averaging data over at least 12 particles in the Pt-4 nm and AT-“Pt3Co” samples. Schematics of three high-index planes are also shown: (113) with the (111) terrace (n ) 2) and one (100) step, (221) with the (111) terrace (n ) 4) and one (111) step, and (112) with the (111) terrace (n ) 3) and one (100) step.

was only observed on a few ordered “Pt3Co” nanoparticles in the AT-“Pt3Co” sample (Figure 9). The Pt-rich shell was found to have a thickness of three to four atomic layers (∼1 nm), where surface Pt atoms may experience smaller strains than the percolated structures with Pt-rich and Pt-poor regions. In addition, the ligand effect from the Pt-poor core regions can be reduced considerably by the Pt-rich shell as the influence of underlying elements decays rapidly with Pt layer thickness and becomes negligible beyond four Pt monolayers.89 Therefore, the Pt-rich shell developed on the ordered nanoparticles might not be as active for ORR as the

structure with percolated Pt-rich and Pt-poor regions, which can allow greater contraction of Pt-Pt bond distance in the near-surface Pt-rich regions. This argument is supported by previous findings23 that ordered “PtCo” nanoparticles become less active than disordered “Pt3Co” nanoparticles after acid treatments. Further aberration-corrected STEM studies of ordered Pt alloy nanoparticles upon acid treatments are needed to verify this hypothesis. 4.3. Enhanced ORR Activity of Annealed Pt Alloy Nanoparticles Relative to Pt: Surface Pt Segregation. Similar to polycrystalline and single-crystal Pt alloy surfaces,3,9,11,12,17 Pt segregation in the first one to three atomic layers on the {100} surface have been found on ordered “Pt3Co” nanoparticles of ∼5 nm in the HT-AT-“Pt3Co” sample. Co-rich layers beneath Pt segregated surface layer(s) not only can compress Pt-Pt bond distance in the adjacent layers but also can induce the ligand effect due to Pt-Co bonds. This is in good agreement with previous DFT findings10 and the Pt-skin structure reported by Markovic et al.,3,11,12 where Pt segregation and antisegregation of transition metals has been found. Combined effects lead to broadening of the d band. To maintain a constant d band filling, the valence band center relative to the Fermi level17,19,87 is lowered,20,90 which reduces the binding of surface Pt atoms toward oxygenated adsorbates. Decreased surface reactivity relative to Pt is supported by the positive shift in the onset of adsorption of oxygenated species in the voltammogram data of HT-AT-“Pt3Co” (Figure 1) and decreased d-band vacancy of HT-AT-“Pt3Co” observed from the Pt L2, 3 edge (Figures 14 and 15) relative to AT-“Pt3Co” (Table 3) and Pt nanoparticles.25,28 This may decrease the activation barrier for protonation reactions of ORR and reduce the coverage of surface-blocking species to allow more O2 to reach active sites. Therefore, enhanced specific ORR activity (roughly four times that

1124 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Chen et al. to bridge fundamental understanding in the ORR activity enhancement mechanisms established for bulk Pt alloy surfaces to the development of highly active catalysts on the nanometer scale. In this work, we show that nanoparticle near-surface chemical compositions can greatly influence the ORR activity of “Pt3Co” nanoparticles, from which the origin in the ORR activity enhancement of Pt alloy relative to Pt nanoparticles is postulated. The enhanced activity (roughly two times based on Pt surface area) of acid-treated “Pt3Co” nanoparticles is attributed to the formation of percolated Pt-rich and Pt-poor regions within individual nanoparticles. In addition, mapping specific activity values of different bulk Pt surfaces reported by Feliu et al.45,46 onto nanoparticle surfaces shows that differences in the surface atomic structure of acid-treated “Pt3Co” and Pt nanoparticles may play a minor role in the specific activity enhancement. Moreover, we show that high-temperature annealing of acid-treated “Pt3Co” promotes ordering of Pt and Co and induces Pt segregation on the (100) surface in the first two to three layers of ordered “Pt3Co” nanoparticles, which can play an important role in the specific ORR activity enhancement of roughly four times relative to Pt. Controlling surface chemical compositions of nanoparticle facets such as inducing Pt segregation by annealing and monolayer synthesis4,91 represents a promising route to develop highly active and low-cost catalysts for ORR in low-temperature fuel cells.

Figure 14. Normalized Pt L3-edge (a) and (b) L2-edge XANES spectra for bulk Pt (solid line), AT-“Pt3Co” (dashed line), and HT-AT-“Pt3Co” (dotted line). The difference spectra relative to bulk Pt are shown in the insets.

Figure 15. k3-weighted radial structural function of the Pt L3 edge for bulk Pt, AT-“Pt3Co”, and HT-AT-“Pt3Co”, which were obtained from the forward Fourier transform of EXAFS spectra. The peaks marked by the dotted line, at approximately 1.6 and 2.6 Å, are assigned to Pt-O (Pt-OH) and Pt-Pt coordination, respectively.

of Pt) on the annealed “Pt3Co” nanoparticles is attributed to the presence of surface Pt segregation. 5. Conclusions The search for ORR nanocatalysts more active than Pt is key to the development of efficient low-temperature fuel cells. It is critical

Acknowledgment. This work was supported in part by the DOE Hydrogen Initiative program under award number DE-FG0205ER15728, an Air Products Faculty Excellence grant, and the Toyota Motor Company. This research made use of the Shared Experimental Facilities supported by the MRSEC Program of the National Science Foundation under award number DMR 02-13282. Synchrotron X-ray adsorption measurements of Pt bulk, supported Pt, and Pt3Co nanoparticles were performed at Advanced Photo Source at the Argonne National Laboratory. The authors gratefully acknowledge T. Toda from TKK Co. Ltd. and A. Mansour from Naval Surface Warefare Center for stimulating discussion, and Y. T. Kim, M. Balasubramanian, and H. You for their assistance in the XANES and EXAFS measurements and analyses. The authors would like to thank Dr. Anthony J. Garratt-Reed for his help with EDS analysis at VG HB603 STEM. The authors would like to thank Ethan Crumlin for his assistance with CAD modeling and Y. T. Kim for helpful comments on the manuscript and assistance in the synchrotron X-ray diffraction measurements at Spring-8. Research was sponsored by the Asst. Secretary for Energy Efficiency and Renewable Energy, Office of FreedomCAR and Vehicle Technologies, as part of the High Temperature Materials Laboratory User Program, Oak Ridge National Laboratory, managed by UT-Battelle LLC for the U. S. DOE under contract number DE-AC05-00OR22725. Supporting Information Available: Koutecky-Levich plots of RDE data, phase analysis of synchrotron X-ray diffraction data of “Pt3Co” nanoparticles, particle size histogram of Pt-4nm nanoparticles, the relationship between the surface fraction of particles with shapes close to a truncated octahedron and the length fraction of surfaces projected on the particle perimeter along the 〈110〉 zone axis, and the {111} FFT filtering of the particle in Figure 8a. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Dresselhaus, M. S.; Thomas, I. L. Nature 2001, 414, 332. (2) Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Catal., B 2005, 56, 9. (3) Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G. F.; Ross, P. N.; Lucas, C. A.; Markovic, N. M. Science 2007, 315, 493.

ORR Activity on “Pt3Co” Nanoparticles (4) Zhang, J. L.; Vukmirovic, M. B.; Xu, Y.; Mavrikakis, M.; Adzic, R. R. Angew. Chem., Int. Ed. 2005, 44, 2132. (5) Zhang, J.; Mo, Y.; Vukmirovic, M. B.; Klie, R.; Sasaki, K.; Adzic, R. R. J. Phys. Chem. B 2004, 108, 10955. (6) Norskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jonsson, H. J. Phys. Chem. B 2004, 108, 17886. (7) Toda, T.; Igarashi, H.; Uchida, H.; Watanabe, M. J. Electrochem. Soc. 1999, 146, 3750. (8) Paffett, M. T.; Beery, J. G.; Gottesfeld, S. J. Electrochem. Soc. 1988, 135, 1431. (9) Gauthier, Y. Surf. ReV. Lett. 1996, 3, 1663. (10) Ruban, A. V.; Skriver, H. L.; Norskov, J. K. Phys. ReV. B 1999, 59, 15990. (11) Stamenkovic, V. R.; Mun, B. S.; Mayrhofer, K. J. J.; Ross, P. N.; Markovic, N. M. J. Am. Chem. Soc. 2006, 128, 8813. (12) Stamenkovic, V. R.; Mun, B. S.; Arenz, M.; Mayrhofer, K. J. J.; Lucas, C. A.; Wang, G. F.; Ross, P. N.; Markovic, N. M. Nat. Mater. 2007, 6, 241. (13) Stamenkovic, V.; Schmidt, T. J.; Ross, P. N.; Markovic, N. M. J. Electroanal. Chem. 2003, 554, 191. (14) Paulus, U. A.; Wokaun, A.; Scherer, G. G.; Schmidt, T. J.; Stamenkovic, V.; Radmilovic, V.; Markovic, N. M.; Ross, P. N. J. Phys. Chem. B 2002, 106, 4181. (15) Stamenkovic, V.; Schmidt, T. J.; Ross, P. N.; Markovic, N. M. J. Phys. Chem. B 2002, 106, 11970. (16) Stamenkovic, V.; Mun, B. S.; Mayrhofer, K. J. J.; Ross, P. N.; Markovic, N. M.; Rossmeisl, J.; Greeley, J.; Norskov, J. K. Angew. Chem., Int. Ed. 2006, 45, 2897. (17) Xu, Y.; Ruban, A. V.; Mavrikakis, M. J. Am. Chem. Soc. 2004, 126, 4717. (18) Mun, B. S.; Watanabe, M.; Rossi, M.; Stamenkovic, V.; Markovic, N. M.; Ross, P. N. J. Chem. Phys. 2005, 123, 204717. (19) Mavrikakis, M.; Hammer, B.; Norskov, J. K. Phys. ReV. Lett. 1998, 81, 2819. (20) Kitchin, J. R.; Norskov, J. K.; Barteau, M. A.; Chen, J. G. Phys. ReV. Lett. 2004, 93, 156801. (21) Greeley, J.; Norskov, J. K. Electrochim. Acta 2007, 52, 5829. (22) Wakisaka, M.; Suzuki, H.; Mitsui, S.; Uchida, H.; Watanabe, M. J. Phys. Chem. C 2008, 112, 2750. (23) Watanabe, M.; Tsurumi, K.; Mizukami, T.; Nakamura, T.; Stonehart, P. J. Electrochem. Soc. 1994, 141, 2659. (24) Mukerjee, S.; Srinivasan, S. J. Electroanal. Chem. 1993, 357, 201. (25) Mukerjee, S.; Srinivasan, S.; Soriaga, M. P.; Mcbreen, J. J. Phys. Chem. 1995, 99, 4577. (26) Koh, S.; Strasser, P. J. Am. Chem. Soc. 2007, 129, 12624. (27) Strasser, P.; Koha, S.; Greeley, J. Phys. Chem. Chem. Phys. 2008, 10, 3670. (28) Mukerjee, S.; Srinivasan, S.; Soriaga, M. P.; Mcbreen, J. J. Electrochem. Soc. 1995, 142, 1409. (29) Zhang, J.; Lima, F. H. B.; Shao, M. H.; Sasaki, K.; Wang, J. X.; Hanson, J.; Adzic, R. R. J. Phys. Chem. B 2005, 109, 22701. (30) Jalan, V.; Taylor, E. J. J. Electrochem. Soc. 1983, 130, 2299. (31) Beard, B. C.; Ross, P. N. J. Electrochem. Soc. 1990, 137, 3368. (32) Kinoshita, K. J. Electrochem. Soc. 1990, 137, 845. (33) Peuckert, M.; Yoneda, T.; Betta, R. A. D.; Boudart, M. J. Electrochem. Soc. 1986, 133, 944. (34) Mayrhofer, K. J. J.; Blizanac, B. B.; Arenz, M.; Stamenkovic, V. R.; Ross, P. N.; Markovic, N. M. J. Phys. Chem. B 2005, 109, 14433. (35) Schmidt, T. J.; Gasteiger, H. A.; Stab, G. D.; Urban, P. M.; Kolb, D. M.; Behm, R. J. J. Electrochem. Soc. 1998, 145, 2354. (36) Paulus, U. A.; Schmidt, T. J.; Gasteiger, H. A.; Behm, R. J. J. Electroanal. Chem. 2001, 495, 134. (37) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications;John Wiley & Sons: New York, 1980; p 288. (38) Markovic, N. M.; Gasteiger, H. A.; Ross, P. N. J. Phys. Chem. 1995, 99, 3411. (39) Nishibori, E.; Takata, M.; Kato, K.; Sakata, M.; Kubota, Y.; Aoyagi, S.; Kuroiwa, Y.; Yamakata, M.; Ikeda, N. Nucl. Instrum. Methods Phys. Res., Sect. A 2001, 467, 1045. (40) Rodriguezcarvajal, J. Physica B 1993, 192, 55. (41) Ferreira, P. J.; la O′, G. J.; Shao-Horn, Y.; Morgan, D.; Makharia, R.; Kocha, S.; Gasteiger, H. A. J. Electrochem. Soc. 2005, 152, A2256. (42) Gontard, L. C.; Chang, L. Y.; Hetherington, C. J. D.; Kirkland, A. I.; Ozkaya, D.; Dunin-Borkowski, R. E. Angew. Chem., Int. Ed. 2007, 46, 3683. (43) Markovic, N. M.; Marinkovic, N. S.; Adzic, R. R. J. Electroanal. Chem. 1988, 241, 309. (44) Markovic, N. M.; Marinkovic, N. S.; Adzic, R. R. J. Electroanal. Chem. 1991, 314, 289. (45) Macia, M. D.; Campina, J. M.; Herrero, E.; Feliu, J. M. J. Electroanal. Chem. 2004, 564, 141.

J. Phys. Chem. C, Vol. 113, No. 3, 2009 1125 (46) Kuzume, A.; Herrero, E.; Feliu, J. M. J. Electroanal. Chem. 2007, 599, 333. (47) Heald, S. M.; Brewe, D. L.; Stern, E. A.; Kim, K. H.; Brown, F. C.; Jiang, D. T.; Crozier, E. D.; Gordon, R. A. J. Synchrotron Rad. 1999, 6, 347. (48) Heald, S.; Stern, E.; Brewe, D.; Gordon, R.; Crozier, D.; Jiang, D. T.; Cross, J. J. Synchrotron Rad. 2001, 8, 342. (49) Newville, M. J. Synchrotron Rad. 2001, 8, 322. (50) Rehr, J. J.; Albers, R. C. ReV. Mod. Phys. 2000, 72, 621. (51) Markovic, N. M.; Ross, P. N. Surf. Sci. Rep. 2002, 45, 117. (52) Gomez, R.; Orts, J. M.; Alvarez-Ruiz, B.; Feliu, J. M. J. Phys. Chem. B 2004, 108, 228. (53) Wang, J. X.; Markovic, N. M.; Adzic, R. R. J. Phys. Chem. B 2004, 108, 4127. (54) Murthi, V. S.; Urian, R. C.; Mukerjee, S. J. Phys. Chem. B 2004, 108, 11011. (55) Koh, S.; Leisch, J.; Toney, M. F.; Strasser, P. J. Phys. Chem. C 2007, 111, 3744. (56) Koh, S.; Yu, C.; Mani, P.; Srivastava, R.; Strasser, P. J. Power Sources 2007, 172, 50. (57) Paulus, U. A.; Wokaun, A.; Scherer, G. G.; Schmidt, T. J.; Stamenkovic, V.; Markovic, N. M.; Ross, P. N. Electrochim. Acta 2002, 47, 3787. (58) Markovi, N. M.; Stamenkovi, T. J. S. V.; Ross, P. N. Fuel Cells 2001, 1, 105. (59) Samant, M. G.; Boudart, M. J. Phys. Chem. 1991, 95, 4070. (60) Sun, C. Q. Prog. Solid State Chem. 2007, 35, 1. (61) Jiang, Q.; Liang, L. H.; Zhao, D. S. J. Phys. Chem. B 2001, 105, 6275. (62) Wasserman, H. J.; Vermaak, J. S. Surf. Sci. 1972, 32, 168. (63) Buschow, K. H. J.; Vanengen, P. G.; Jongebreur, R. J. Magn. Magn. Mater. 1983, 38, 1. (64) Laar, B. V. J. Phys. 1964, 25, 600. (65) Berg, H.; Cohen, J. B. Metall. Mater. Trans. 1972, 3, 1797. (66) Jen, S. U. J. Alloys Compd. 1996, 234, 231. (67) Fultz, B.; Howe, J. M. Transmission Electron Microscopy and Diffractometry of Materials; Springer: Berlin, 2008; p 426. (68) Suzuki, H.; Shintaku, M.; Sato, T.; Tamano, M.; Matsuura, T.; Hori, M.; Kaito, C. Jpn. J. Appl. Phys., Part 2 2005, 44, L610. (69) Oppenheim, I. C.; Trevor, D. J.; Chidsey, C. E. D.; Trevor, P. L.; Sieradzki, K. Science 1991, 254, 687. (70) Erlebacher, J.; Aziz, M. J.; Karma, A.; Dimitrov, N.; Sieradzki, K. Nature 2001, 410, 450. (71) Erlebacher, J. J. Electrochem. Soc. 2004, 151, C614. (72) Yu, C.; Koh, S.; Leisch, J. E.; Toney, M. F.; Strasser, P. Faraday Discuss. 2008, 140, 185. (73) Markovic, N.; Gasteiger, H.; Ross, P. N. J. Electrochem. Soc. 1997, 144, 1591. (74) Wang, G. F.; Van Hove, M. A.; Ross, P. N.; Baskes, M. I. J. Chem. Phys. 2005, 122, 024706. (75) Wang, T.; Lee, C.; Schmidt, L. D. Surf. Sci. 1985, 163, 181. (76) Lytle, F. W.; Wei, P. S. P.; Greegor, R. B.; Via, G. H.; Sinfelt, J. H. J. Chem. Phys. 1979, 70, 4849. (77) Mukerjee, S.; McBreen, J. J. Electrochem. Soc. 1996, 143, 2285. (78) Mukerjee, S.; McBreen, J. J. Electroanal. Chem. 1998, 448, 163. (79) Teliska, A.; O’Grady, W. E.; Ramaker, D. E. J. Phys. Chem. B 2005, 109, 8076. (80) Teliska, M.; Murthi, V. S.; Mukerjee, S.; Ramaker, D. E. J. Electrochem. Soc. 2005, 152, A2159. (81) Frenken, J. W. M.; Stoltze, P. Phys. ReV. Lett. 1999, 82, 3500. (82) Ferreira, P. J.; Shao-Horn, Y. Electrochem. Solid-State Lett. 2007, 10, B60. (83) Gland, J. L.; Korchak, V. N. Surf. Sci. 1978, 75, 733. (84) Hammer, B.; Nielsen, O. H.; Norskov, J. K. Catal. Lett. 1997, 46, 31. (85) Markovic, N. M.; Adzic, R. R.; Cahan, B. D.; Yeager, E. B. J. Electroanal. Chem. 1994, 377, 249. (86) Greeley, J.; Rossmeisl, J.; Hellman, A.; Norskov, J. K. Z. Phys. Chem. 2007, 221, 1209. (87) Grabow, L.; Xu, Y.; Mavrikakis, M. Phys. Chem. Chem. Phys. 2006, 8, 3369. (88) Rossmeisl, J.; Karlberg, G. S.; Jaramillo, T.; Norskov, J. K. Faraday Discuss. 2008, 140, 337. (89) Schlapka, A.; Lischka, M.; Gross, A.; Kasberger, U.; Jakob, P. Phys. ReV. Lett. 2003, 91, 016101. (90) Ruban, A.; Hammer, B.; Stoltze, P.; Skriver, H. L.; Norskov, J. K. J. Mol. Catal. A: Chem. 1997, 115, 421. (91) Nilekar, A.; Xu, Y.; Zhang, J.; Vukmirovic, M.; Sasaki, K.; Adzic, R.; Mavrikakis, M. Top. Catal. 2007, 46, 276.

JP807143E