Ab initio study of canted magnetism of finite atomic chains at surfaces

By using ab initio methods on different levels we study the magnetic ground state of (finite) atomic wires deposited on metallic surfaces. A phenomenological model based on symmetry arguments suggests that the magnetization of a ferromagnetic wire is aligned either normal to the wire and, generally, tilted with respect to the surface normal or parallel to the wire. From a first principles point of view, this simple model can be best related to the so– called magnetic force theorem calculations being often used to explore magnetic anisotropy energies of bulk and surface systems. The second theoretical approach we use to search for the canted magnetic ground state is first principles adiabatic spin dynamics extended to the case of fully relativistic electron scattering. First, for the case of two adjacent Fe atoms an a Cu(111) surface we demonstrate that the reduction of the surface symmetry can indeed lead to canted magnetism. The anisotropy constants and consequently the ground state magnetization direction are very sensitive to the position of the dimer with respect to the surface. We also performed calculations for a seven–atom Co chain placed along a step edge of a Pt(111) surface. As far as the ground state spin orientation is concerned we obtain excellent agreement with experiment. Moreover, the magnetic ground state turns out to be slightly noncollinear. PACS numbers: 75.10.Lp, 75.30.Gw, 75.75.+a