Mechanism for Oxygen Reduction Reaction on Pt 3 Ni Alloy Fuel Cell Cathode Yao Sha, † Ted H. Yu, † Boris V. Merinov,* ,† Pezhman Shirvanian, ‡ and William A. Goddard, III* ,† † Materials and Process Simulation Center, MC 139-74, California Institute of Technology, Pasadena, California 91125, United States ‡ Ford Motor Co., Research & Advanced Engineering, 2101 Village Road, Dearborn, Michigan 48104, United States ABSTRACT: We use quantum mechanics, density functional theory at the PBE level, to predict the binding-site preferences and reaction barriers for all intermediates involved in the oxygen reduction reaction (ORR) on the low energy surface of Pt 3 Ni alloy. Here we calculate that the surface layer is Ni depleted (100% Pt) while the second layer is Ni enriched (50% Pt) as shown by experiment. Even though the top layer is pure Pt, we find that the sublayer Ni imposes strong preferences in binding sites for most intermediates, which in turn strongly influences the reaction barriers. This strong preference leads to a strong site dependence of the barriers. Considering water as the solvent, we predict that, at low coverage of O ad and OH ad , the barrier for the rate-determining step is 0.81 eV, whereas, at high coverage, this barrier decreases to 0.43 eV. It can be compared to a barrier of 0.50 eV for pure Pt, explaining the improved ORR rate for the Pt 3 Ni alloy. We report the results both for gas phase and for aqueous phase environments. 1. INTRODUCTION The efficiency of the oxygen reduction reaction (ORR), 4H + + 4e- +O 2 → 2H 2 O, at the cathode of a polymer electrolyte membrane fuel cell (PEMFC) is a critical issue for commercial application of this type of fuel cells. 1-4 The best current catalysts are Pt and Pt-based binary alloys, such as Pt 3 Ni. 5,6 The origin of the superior performance of the Pt 3 Ni alloy has not been clearly understood yet. Some researchers believe it is due to the shift of the d-band center to the desired region that occurs due to alloying Pt with Ni. 7,8 Others came to the conclusion that alloying makes OH ad removal favorable, increasing the surface area available for O 2 binding. 5 It has also been argued that alloying Pt with Ni or Co decreases the surface lattice parameters to values optimal for ORR. 9 A prominent property of Pt 3 Ni and Pt 3 Co is the strong surface segregation observed in experiments. 10-12 Our quantum mechanics (QM) calculations, density functional theory (DFT) at the Perdew-Becke-Ernzerhof (PBE) level, using a two- dimensional slab model find the similar segregation effect for the Pt 3 Ni alloy which results in the surface structure with 100% Pt in the first layer, 50% Pt in the second layer, and 75% Pt in deeper layers. 13 This strong segregation to form a pure Pt surface layer (similar to core-shell systems where the surface is also pure Pt) is supposed to be important in ensuring the improved ORR activity of these alloy catalysts. A recent study of Matanovic et al. 14 argues that the sublayer concentration directly influences the over- potential. However, to our knowledge, no papers have been published that explore the influence of the atomic level configuration for alloying atoms to the reaction mechanism and barriers simultaneously by taking into account solvent effects. In our study, we used QM calculations to study the unique binding-site preferences due to the placement of sublayer alloying atoms for all intermediates involved in the ORR on the segregated surface of Pt 3 Ni and the consequent changes to the reaction barriers and mechanisms. We propose a new detailed atomistic level chemical mechanism explaining the increased reactivity of the Pt 3 Ni alloy. In particular, we show that subsurface Ni has a strong influence on the binding energies and induces a coverage dependence for the preferred ORR mechanism. 2. METHODOLOGY The Pt 3 Ni(111) alloy surface was modeled as a two-dimensionally infinite periodic slab with four atoms per cell and six layers of atoms. We consider the atomic Pt composition as 100-50-75- 75-75-75% Pt, as observed experimentally 10 and calculated theoretically. 13,15,16 All calculations used the PBE functional of DFT. We applied small core norm-conserving angular momentum projected pseudopotentials 17-20 in which the 3p, 3d, and 4s electrons of Ni and the 5p, 5d, and 6s electrons on the Pt are treated explicitly with 16 electrons for neutral Ni and Pt. The SeqQuest software 21 with optimized double-ζ plus polarization quality Gaussian-type orbitals on Pt and Ni was employed for our calculations. The periodic cell parameter of the slab corresponds to that of the optimized Pt 3 Ni bulk structure and 0.08% smaller than the experimental value. 22 All charges came from the Mulliken population analysis of the DFT wave function. Received: April 24, 2012 Revised: August 30, 2012 Published: September 4, 2012 Article pubs.acs.org/JPCC © 2012 American Chemical Society 21334 dx.doi.org/10.1021/jp303966u | J. Phys. Chem. C 2012, 116, 21334-21342