Simulation of atomic migration in CoPt ordered alloys |
Tounsia Bouzar ^{2}, Christine Goyhenex ^{1}, Romaric V. Montsouka ^{1}, Hamid Bouzar ^{2}, Véronique Pierron-Bohnes ^{1} |
1. Institut de Physique et Chimie des Materiaux de Strasbourg, UMR7504, CNRS - ULP, 23, rue du Loess, BP 43, Strasbourg CEDEX 2 67034, France |
Abstract |
The L1_{0} systems are extensively studied as good candidates for high density magnetic storage media due to their high magnetic anisotropy, which is related to their chemical order anisotropy. In recent experiments, epitaxied thin bilayers NiPt/FePt/MgO(001) grown at 700K were annealed at 800K and 900K. At 800K, the L1_{0} long range order increases without measurable interdiffusion. Surprisingly further annealing at 900K leads to interdiffusion that takes place without destroying the L1_{0} long range order [1]. The same phenomenon has been observed by Rennhofer et al. [2] in the case of L1_{0} FePt multilayers suggesting that interdiffusion can occur through a series of atomic mechanisms while keeping the L1_{0} structure. As this point is important for the growth of high quality layers of alloys, we have investigated the possible atomic processes through numerical simulations for such diffusion without order change. We have used Molecular Dynamics in the second moment approximation of the tight binding method taking as reference CoPt that orders in the L1_{0} tetragonal structure like FePt and NiPt. In a first stage, molecular statics calculations were performed in order to study the vacancy migration and to determine in particular if some jump cycles are feasible with an energy barrier low enough to make possible a diffusion without order change. For instance we find that a 6 jump cycle mechanism is favourable from an energy point of view relatively to a second nearest neighbour jump that has a much higher energy barrier. We have also checked mechanisms involving two or three vacancies. The static results are supported by a statistical study for which large series of constant energy simulations are performed at the temperatures of interest (900-1000 K) in order to check the actual occurrence of each elementary step of the previously identified jump cycles. [1] R. V. Montsouka et al., Phys. Rev. B 74, 144409 (2006) |