arXiv subjects
Publications and source records attributed to Evaggelos Kritsikis.
In this paper, we rigorously study an order 2 scheme that was previously proposed by some of the authors. A slight modification is proposed that enables us to prove the convergence of the scheme while simplifying in the same time the inner iteration.
Dimensionality cross-over is a classical topic in physics. Surprisingly it has not been searched in micromagnetism, which deals with objects such as domain walls (2D) and vortices (1D). We predict by simulation a second-order transition between these two objects, with the wall length as the Landau parameter. This was conrmed experimentally based on micron-sized ux-closure dots.
We report a detailed magnetic study of a new type of self-organized nanowires disclosed briefly previously [B. Borca et al., Appl. Phys. Lett. 90, 142507 (2007)]. The templates, prepared on sapphire wafers in a kinetically-limited regime, consist of uniaxially-grooved W(110) surfaces, with a lateral period here tuned to 15nm. Fe deposition leads to the formation of (110) 7 nm-wide wires located at the bottom of the grooves. The effect of capping layers (Mo, Pd, Au, Al) and underlayers (Mo, W) on the magnetic anisotropy of the wires was studied. Significant discrepancies with figures known for thin flat films are evidenced and discussed in terms of step anisotropy and strain-dependent surface anisotropy. Demagnetizing coeffcients of cylinders with a triangular isosceles cross-section have also been calculated, to estimate the contribution of dipolar anisotropy. Finally, the dependence of magnetic anisotropy with the interface element was used to tune the blocking temperature of the wires, here from 50K to 200 K.
The bottleneck of micromagnetic simulations is the computation of the long-ranged magnetostatic fields. This can be tackled on regular N-node grids with Fast Fourier Transforms in time N logN, whereas the geometrically more versatile finite element methods (FEM) are bounded to N^4/3 in the best case. We report the implementation of a Non-uniform Fast Fourier Transform algorithm which brings a N logN convergence to FEM, with no loss of accuracy in the results.
A non-conventional finite element formalism is proposed to solve the dynamic Landau-Lifshitz-Gilbert micromagnetic equations. Two bidimensional test problems are treated to estimate the validity and the accuracy of this finite element approach