Construction of modified embedded atom method potentials for Cu, Pt and Cu-Pt and modelling surface segregation in CU3Pt alloys

In this work, surface segregation to Cu3Pt surfaces is studied with the modified embedded atom method (MEAM). This work is triggered by the catalytic importance of Cu-Pt alloys, together with the contradictory experimental results for the surface segregation in Cu3Pt(1 1 1) alloys based on low energy ion scattering (LEIS) [Y.G. Shen, D.J. O'Connor, K. Wandelt, R.J. MacDonald, Surf. Sci. 328 (1995) 21] and low energy electron diffraction (LEED) [Y. Gauthier, A. Senhaji, B. Legrand, G. Treglia, C. Becker, K. Wandelt, Surf. Sci. 527 (2003) 71]. In order to accurately describe the segregation behaviour in the Cu3Pt system, a reliable potential, that is also applicable to surface phenomena, is indispensable. Therefore, first, new MEAM parameters are derived, consistently based on ab initio density functional theory (DFT) calculations, according to a method that is a modification of previous work [P. van Beurden, G.J. Krarner. Phys. Rev. B 63 (2001) 165106]. Upon testing, these parameters prove to reproduce very well various surface properties of this system. Next, Monte Carlo (MC) simulations combined with the newly derived MEAM potentials are set up to investigate surface segregation to low index single crystal surfaces. For the Cu3Pt(1 1 1) surface, our MC/MEAM simulations agree completely with the available LEIS evidence and contradict the unusual depth profile based on LEED. However, the slight Pt enrichment observed in the LEED experiments can be reproduced by assuming a slight Pt excess in the bulk of the sample. The simulated composition depth profile, on the other hand, does not agree with the LEED evidence. Also, for the Cu3Pt(1 0 0) surface, the MC/MEAM results agree completely with LEIS experiments. For the Cu3Pt(1 1 0) surface, finally, the MC/MEAM simulations show a somewhat deviating behaviour with respect to the experimental LEIS evidence. The possibility of a missing-row reconstruction is evaluated, but cannot explain the discrepancy for the Cu3Pt(1 1 0) system. In order to further investigate the deviation from the experiments, additional DFT and MEAM calculations are performed in search of the preferred surface termination for Cu3Pt(1 1 0). Both DFT and MEAM calculations agree on the pure Cu layer as the most stable surface termination. Although the experiment was extensively tested for reproducibility, it possibly reflects a metastable state. Finally, in view of the importance of small and less orderly particles in catalysis, the newly derived MEAM parameters are used in order to study the segregation to Cu3Pt vicinal surfaces with {1 1 1} terraces, for which no experimental information is available yet. (C) 2007 Elsevier B.V. All rights reserved.