Numerical studies of the magnetism of FeNiMn alloys in the Invar region

Abstract

By means of self-consistent semi-empirical LCAO calculations we study the itinerant magnetism of (Fe 0.65 Ni 0.35 ) 1−y Mn y alloys for y between 0 and 0.22 at T = 0 K, neglecting only the transverse spin components. We find that the magnetic behaviour is quite complicated on a local scale. In addition to ferromagnetic behaviour, also metastable spin-glass-like configurations are found. In the same approach, using a direct numerical calculation by the Kubo-Formalism without any fit parameters, we also calculate the electrical conductance in the magnetic state and find that the y-dependence observed in the experiments is well reproduced by our calculations, except of an overall factor of ≈ 5, by which our resistivities are too large. * based on the Diploma thesis of A. Paintner, Regensburg 1994

Ab-initio calculations of the GMR-effect in Fe/V multilayers

“…2.) This is supported by the observation that the numerical results for the resistivities in the magnetic state of disordered Fe/Ni/Mn alloys in our former paper [11] have also come out too large just by a constant factor ∼ 5, in spite of the fact that with increasing Mn concentration the conductance decreased considerably. However again, apart from the constant factor, the concentration dependence of the experimental values was well-reproduced in [11] with our formalism.…”

“…But even for these small systems, such results can be obtained, as seen below. This has also been exemplified in former calculations of the resistivity in the magnetic state of disordered systems, [11,12].…”

“…We use our realistic self-consistent semi-empirical LCAO approach already described in [15,11], i.e. with 9 orthogonalized orbitals per atom (five 3d-, three 4p-and one 4s-orbital) and a collinear magnetic state described by the equations…”

“…Thus it is only necessary to modify the hopping matrices of Papaconstantopoulos according to the modified positions, i.e. for the two-center integrals one uses relations as I dd (r) ∝ r −5 , see [15,11]; additionally, if site l is occupied by a Fe atom, but site m by V, we assume as usual H F e,V lα,mβ = (H F e,F e lα,mβ •H V,V lα,mβ ) 1/2 . Finally, to get agreement with ab initio calculations, see below, we have used a common shift of ∆E = 0.5 eV for the Fe d-levels with respect to the values in [16], whereas for V no shift was assumed, and for the Hubbard-energies we have taken U lα = 5.8 and 2.58 eV for the Fe and V d-orbitals, respectively, U l,α = 0 otherwise.…”

“…We have applied a similar approach already earlier for almost-ab-initiocalculations of the magnetic and transport properties of strongly disordered or even amorphous systems. Both the electronic structure, [9,10,11], and the transport properties of the systems, [11,12], have been calculated, the transport properties directly from the Kubo formula, and separated into contributions from the up-spin and down-spin carriers. We stress that the Kubo approach is rigorous in principle, and does not invoke the usual Boltzmann approximations.…”