Atomistic modelling of bimetallic surfaces: from many-body potentials to lattice models

1. Atomistic modelling of bimetallic surfaces:from many-body potentials to lattice models Bernard Legrand(1) and Guy Tréglia(2) (1) Service de Recherche en Métallurgie Physique (SRMP-CEA), Saclay, France (2)Centre de Recherche en Matière Condenséeet Nanosciences (CRMCN),Marseille, France … based on works by : Isabelle Meunier, Hazar Guesmi, Laurent Lapéna, Pierre Müller : CRMCN, Marseille Isabelle Braems, Fabienne Berthier, Robert Tétot, Jérôme Creuze : LEMHE Orsay François Ducastelle : ONERA, Châtillon

2. Chemical configuration (K = A, B) : with Atomistic approach (tight-binding) SMA TBIM Consistency TBIM-SMA Chemical rearrangements Atomic rearrangements

3. How to study chemistry on a rigid lattice … including atomic relaxation ! Segregation energy at the surface of alloyAcB1-c ΔHseg = ΔHsite + ΔHsize+ ΔHalloy+ ΔHinter interaction impurity SMA* SMA SMA Quenched Molecular Dynamics relaxation VR B(B) B(B*) B(A)

4. p = 0 p = N Choice of a reference system : Cu(Ag) • Silver concentrates on the surface: • higher surface energy of Cu: Dg=.14eV/at • larger atomic radius of Ag: rAg = 1.13rCu • miscibility gap in the bulk: Vb<0 Expected surface behaviour Cu(Ag) (100) c(10x2) (1x1) chemical transition + structural transitions … also (111), (110)

5. Influence of relaxation on TBIM parameters • size effect • failure of elasticity (asymetry) •  Anisotropy (relaxation) • alloying effect: VR Relaxation (bulk) phase separation tendency   long range interactions Surface (001) ordering tendency  anisotropy

6. schematic surface phase diagram 3-dim (Z = 0) T 2-dim (Z = 2) 2-dim (Z = 3) Miscibility gap order tendency c Cu Ag Relaxation  surface ordering : Cu1-cAgc • surface effective pair interactions Bond breakingphase separation tendency  (ordering )  ordering reversal depending of orientation and concentration Cu(Ag) (100) : surface ordering whatever Ag surface concentration ?

7. Superstructure  surface ordering : Cu1-cAgc(100), c0 Structuraldependence of VR ? • STM Ag/Cu(100) • Sprunger et al. PRB 1996 • q < 0.1 : mixed Cu-Ag phase … with (1x1) structure • q  1.0 : pure Ag island … with c(10x2) structure T = 300 K Nad=1/10*Nlayer 1x1 c(10x2)

8. Superstructure  surface ordering : Cu1-cAgc(100), c0 c(10x2) structure (1x1) structure V1100(1x1) > 0 V1100(10x2) < 0 O.K. with STM experiment Ag/Cu(100) Sprunger et al. 1996 q < 0.1 : mixed Cu-Ag phase with (1x1) structure q  1.0 : pure Ag islands with c(10x2) structure

9. Lattice analysis validity : Cu1-cAgc(100), c0 DHseg mixed Monte Carlo exchange + displacements Beginning of expulsion V > 0 V < 0, Tc=300 K V1x1 Vc10x2 c0

10. equilibrium Experimental validation : Sb/Si(111) Mass Spectrometry Thermodesorption Bragg-Williams : P = P0. /(1-). exp (DHads/kT) DHads = VSb-Si +  VSb-Sb Langmuir-like behaviour … but not exactly Two regimes: q < 0.7 : V = 0 q > 0.7 : V < 0, Tc = 870 K Adsorption isotherms

11. Sb/Si(111): adsorption on top 7.68A° 2.44A° Sb/Si(111): adsorption H3 2.79A° 7.68A° 2.9A° 3.06A° Sb/Si(111): adsorption T4 3.84A° Adsorption Mode of Sb/Si(111) : FP-LAPW study • nature of interactions dépend on site: • Ternary : positive slope  Sb-Sb repulsive • On top : negative slope Sb-Sb attractive • Adsorption site transition (q0.7) •  change in the sign of V

12. Conclusion … when ab initio meets classical simulations ? Decomposition into main driving forces  Tight Binding –Ising Model  Ab Initio-Ising Model … to be used for long scale (time, space) modeling of phenomena at complex material interfaces