Supplementary Figure 1 Nano-beam electron diffraction Nano-beam electron diffraction

Transcription

1 Supplementary Figure 1 Nano-beam electron diffraction Nano-beam electron diffraction (NBED) patterns of different Pd-W nanoparticles on OMCs (Pd-W/OMCs), all exhibiting a body-centered cubic (bcc) microstructure.

2 Supplementary Figure 2 Atomic resolution Z-contrast images Atomic resolution Z- contrast image of a Pd-W nanoparticle on OMCs (Pd-W/OMCs), with face-centered cubic (fcc) structure. Scale bar is 1 nm.

3 Supplementary Figure 3 Atomic resolution Z contrast Atomic resolution Z contrast image of the Pd-W nanoparticle on OMCs (Pd-W/OMCs). The circled area shows a twin boundary, with atomic simulation in the middle. Fast fourier transformations (FFTs) of the top and bottom part of the twin boundary are shown at left. Scale bar in right figure is 2 nm.

4 Supplementary Figure 4 Atomic resolution Z-contrast images Atomic resolution Z- contrast image of different Pd-W nanoparticles on OMCs (Pd-W/OMCs). The bright spots are Pd-W atom columns. All atom columns are not resolved due to the polycrystallinity of the particles. Scale bars in all images are 1 nm.

5 Supplementary Figure 5 Cohesive energy calculations 2-D super cell used to calculate the cohesive energy of surfaces in Figure 2g in main manuscript. (a-b) Alloy configurations, where Pd atoms are randomly distributed along the structure. (c-d) Segregated configuration, where Pd atoms are collocated as an island at the surface. Pd atoms are in blue, tungsten in gray.

6 Supplementary Figure 6 X-ray diffraction Powder x-ray diffraction of ordered mesoporous carbon (OMCs) (black curve), tungsten decorated OMCs (red curve), and Pd-W decorated OMCs (blue curve).

7 Supplementary Figure 7 Particle size distribution Particle size distribution before and after the stability test. (a) 60% Pt/Vulcan. (b) Palladium-tungsten/ordered mesoporous carbon (Pd- W/OMCs). Size distribution was obtained by measuring at least 400 nanoparticles for each sample using TEM micrographs.

8 Intensity (Arb. Units) Pd-W 18 months old material Pd-W fresh material 0.0 Pure Pd foil Energy (ev) Supplementary Figure 8 Extended x-ray absorption fine structure spectra Experimental Pd K-edge Extended x-ray absorption fine structure spectra.

9 Supplementary Figure 9 Coordination number simulation Different configurations of Pd islands with their respective average coordination number (N) and the total number of atoms (At). The number next to each atom indicates the number of first neighbors. Islands depicted in (a-c) exhibit a single atomic Pd layer, while islands in (d-e) are composed of two Pd layers, resulting in larger values of N.

10 Supplementary Figure 10 Electrochemical characterization (a) Cyclic voltammetry curves of commercial Pt/Vulcan (black line), tungsten/ordered mesoporous carbon (W/OMCs) (red line) and Palladium-tungsten/ordered mesoporous carbon (Pd-W/OMCs) (blue line) in (a) acidic and (b) neutral solution.

11 Current (ma/cm 2 ) 0.00 Pd-W/OMC catalyst Time (hours) Supplementary Figure 11 Electrochemical stability test Cyclic voltammetry I-t curve of Palladium-tungsten/ordered mesoporous carbon (Pd-W/OMCs) in alkaline solution. The sample was prepared by casting on a glassy carbon electrode 25 micrograms of catalyst. A three-electrode system was used, where a working electrode potential of V was applied in 0.1 M of KOH, an oxygen bubbling rate of 15 ml/min was kept during the whole experiment. The material was tested continuously for s (~54 hrs).

12 Supplementary Figure 12 Molecular dynamics (MD) of palladium-tungsten (Pd-W) Energy change during the MD simulation, where the energy reach a steady state around 400 fs (a) Energy change during the ab initio molecular dynamics. (b) Final snapshot at 1.6 ps of the Pd 24 W 151 nanoparticle with a Pd 4 islands. During the simulation time there is no substantial change in the geometry of the nanoparticle.

13 Supplementary Figure 13 Fuel cell stability tests Stability fuel cell tests with the cathode electrode modified by various materials, as specified in the figure.

14 Supplementary Figure 14 Molecular models of Pd-W clusters Pd-W nanoparticles (175 atoms) showing different catalytic sites used to calculate the oxygen reduction reaction (ORR) overpotential, and thus the catalyst activity in Figure 5d in main manuscript. The catalytic sites are marked in red. The nanoparticles were first geometrically optimized before any adsorption event. Also, during O* and HO* adsorption all atoms were allowed to move. Clusters in (a-d) are the same shown in Figure 5a in the main manuscript, with one, two, three and four-pd atomic island in each facet. Clusters in (e-j) are 3 different W clusters where diverse catalytic sites are selected.

15 Supplementary Figure 15 Gibbs free energy of Pd-W nanoparticles Change in Gibbs free energy (ΔG ) during the oxygen reduction reaction (ORR) performed at the equilibrium potential (U 0 ) of 1.23 V on Pd 24 W 151 nanoparticles with different catalytic sites. See also Supplementary note 1 for detailed information on the calculation procedure. (a) W 3 -island, (b) PdW 2 -island, (c) Pd 2 W-island and (d) Pd 3 -island. The reduction of ΔG is easily observed in all ORR steps. This leads to a reduced overpotential, specially for the Pd 3 -island, which in some configurations has a better ORR activity than a Pd(111) surface.

16 Supplementary Figure 16 Effect of oxygen adsorption Adsorption energy of oxygen (E O ) at two different oxygen coverage on a Pd 4 island on a W surface. In (a-b) the adsorption event is performed on a free oxygen surface (Ө O = 0). For (c-d), the O* adsorption is performed on a pre-oxidized W surface (Ө O = 1/2). The change in adsorption energy (ΔE O ) is ~0.14 ev weaker at higher O-coverage, this information do not significantly change the results reported in Figure 4 in main manuscript. The supercell is marked with a black line. Color code: Pd (blue), W (gray), and O (red).

17 Supplementary Figure 17 Charge in adsorption energy of intermediates (O* and HO*) due the presence of an external electric field The electric field was applied perpendicularly to the adsorption site in the range of ±0.4 V/Å, which is the range of the interest. The numbers indicates the resulting overpotential (η ORR ) considering the effects of the electric field. As observed, the adsorption energy is slightly modified by the presence of the electric field, less than 0.15 ev in each direction, which results in a similar variation of the η ORR in ~0.15 V. Our results are in agreement with those reported by G. S. Karlberg, et al [1], indicating that the influence of the electrolyte is just important for materials near the top of the volcano.

18 Supplementary Table 1. Summary of EXAFS analysis Sample N R (±ΔR) σ 2 R factor Pd-W sample 4.0(3) (0.0031) ± Pd foil 10.2(6) (0.0019) ±

19 Supplementary Note 1. Theoretical analysis of ORR on Pd-W nanoparticles. We analyzed the catalytic properties for ORR of seven different Pd-W clusters, where a total of ten different catalytic sites were selected. Optimized structures of the dodecahedral nanoparticles, with a bcc structure, are shown in Fig. S14. In this case, the catalytic sites are composed of three atoms (enough to perform the reactions of interest) marked in red, where some W 3 -, Pd 1 W 2 -, Pd 2 W-, Pd 3 -, and Pd 4 -islands can be identified. We then used the dissociative mechanism for oxygen reduction and the computational hydrogen-electrode [2]. The elementary reactions studied are indicated in Eq. 1-3, where * denotes adsorbed species. The reaction path is initiated by the adsorption of an atomically oxygen (Eq. 1) followed by two proton-electron pairs (H + + e - ) transfer producing one water molecule (Eq. 2-3). Eq. 1 Eq. 2 Eq. 3 We then calculated the change in Gibbs free energy (ΔG) during the chemical reactions shown in Eq. 1-3 at the equilibrium potential (U 0 ) for ORR equal to 1.23 V. The results are plotted in Fig. 5c in main manuscript, and Fig. S15. We find that the adsorption of atomic oxygen is exergonic and depends strongly on the atoms located at the reaction site. From Fig. S15 we observe that when the amount of W atoms in the catalytic island decreases, the oxygen

20 adsorption energy decreases substantially, which will affect the whole ORR performance. The subsequent two proton-electron (H + + e - ) transfers (steps 2-3) are uphill in free energy which indicates that the reactions in Eq. 2-3 will not proceed at U 0. As proposed by Norskov, et al [3] the largest of these steps is directly related to the ORR overpotential, and as observed from Fig. S15 and Fig. 5c in main manuscript, the ΔG of Eq. 2 and 3 is significantly reduced when the catalytic island contains 3 and 4 Pd atoms, resulting in a drastically boost of the ORR activity. These results are shown in main manuscript. Supplementary references 1. Karlberg, G. S., Rossmeisl, J. & Norskov, J. K. Estimations of electric field effects on the oxygen reduction reaction based on the density functional theory. Physical Chemistry Chemical Physics 9, , (2007) 2. J. Greeley, J.K. Nørskov, Electrochemical dissolution of surfaces alloys in acids: thermodynamic trends from first-principles calculations, Electrochem Acta 52, (2007) 3. J. K. Nørskov, J. Rossmeisl, A. Logadottir, and L. Lindqvist, J. R. Kitchin, T. Bligaard, H. Jonsson, Origin of the Overpotential for Oxygen Reduction at a Fuel-Cell Cathode, J. Phys. Chem. B, 108, (2004)