No abstract

We present a first-principles theory of the variation of magnetic anisotropy, K, with temperature, T, in metallic ferromagnets. It is based on relativistic electronic structure theory and calculation of magnetic torque. Thermally induced local moment magnetic fluctuations are described within the relativistic generalization of the disordered local moment theory from which the T dependence of the magnetization, m, is found. We apply the theory to a uniaxial magnetic material with tetragonal crystal symmetry, L1 0-ordered FePd, and find its uniaxial K consistent with a magnetic easy axis perpendicular to the Fe/ Pd layers for all m and proportional to m 2 for a broad range of values of m. This is the same trend that we have previously found in L1 0-ordered FePt and which agrees with experiment. We also study a magnetically soft cubic magnet, the Fe 50 Pt 50 solid solution, and find that its small magnetic anisotropy constant K 1 rapidly diminishes from 8 eV to zero. K 1 evolves from being proportional to m 7 at low T to m 4 near the Curie temperature. The accounts of both the tetragonal and cubic itinerant electron magnets differ from those extracted from single ion anisotropy models and instead receive clear interpretations in terms of two ion anisotropic exchange.

The heavy rare earth elements crystallize into hexagonally close packed (h.c.p.) structures and share a common outer electronic configuration, differing only in the number of 4f electrons they have. These chemically inert 4f electrons set up localized magnetic moments, which are coupled via an indirect exchange interaction involving the conduction electrons. This leads to the formation of a wide variety of magnetic structures, the periodicities of which are often incommensurate with the underlying crystal lattice. Such incommensurate ordering is associated with a 'webbed' topology of the momentum space surface separating the occupied and unoccupied electron states (the Fermi surface). The shape of this surface-and hence the magnetic structure-for the heavy rare earth elements is known to depend on the ratio of the interplanar spacing c and the interatomic, intraplanar spacing a of the h.c.p. lattice. A theoretical understanding of this problem is, however, far from complete. Here, using gadolinium as a prototype for all the heavy rare earth elements, we generate a unified magnetic phase diagram, which unequivocally links the magnetic structures of the heavy rare earths to their lattice parameters. In addition to verifying the importance of the c/a ratio, we find that the atomic unit cell volume plays a separate, distinct role in determining the magnetic properties: we show that the trend from ferromagnetism to incommensurate ordering as atomic number increases is connected to the concomitant decrease in unit cell volume. This volume decrease occurs because of the so-called lanthanide contraction, where the addition of electrons to the poorly shielding 4f orbitals leads to an increase in effective nuclear charge and, correspondingly, a decrease in ionic radii.

Using a first-principles, relativistic electronic structure theory of finite temperature metallic magnetism, we investigate the variation of magnetic anisotropy K with magnetization M in metallic ferromagnets. We apply the theory to the high uniaxial K material, L1(0)-ordered FePt, and find its magnetic easy axis perpendicular to the Fe/Pt layers for all M and K to be proportional to M2 for a broad range of values of M. For small M, near the Curie temperature, the calculations pick out the easy axis for the onset of magnetic order. Our ab initio results for this important magnetic material agree well with recent experimental measurements, whereas the single-ion anisotropy model fails to give the correct qualitative behavior.