The computational framework of this study is based on the local-spin-density approximation with firstprinciples full-potential linear muffin-tin orbital calculations including orbital polarization ͑OP͒ correction. We have studied the magnetic anisotropy for a series of bilayer CuAu͑I͒-type materials such as FeX, MnX (X ϭNi,Pd,Pt), CoPt, NiPt, MnHg, and MnRh in a ferromagnetic state using experimental structural parameters to understand the microscopic origin of magnetic-anisotropy energy ͑MAE͒ in magnetic multilayers. Except for MnRh and MnHg, all these phases show perpendicular magnetization. We have analyzed our results in terms of angular momentum-, spin-and site-projected density of states, magnetic-angular-momentumprojected density of states, orbital-moment density of states, and total density of states. The orbital-moment number of states and the orbital-moment anisotropy for FeX (XϭNi,Pd,Pt) are calculated as a function of band filling to study its effect on MAE. The total and site-projected spin and orbital moments for all these systems are calculated with and without OP when the magnetization is along or perpendicular to the plane. The results are compared with available experimental as well as theoretical results. Our calculations show that OP always enhances the orbital moment in these phases and brings them closer to experimental values. The changes in MAE are analyzed in terms of exchange splitting, spin-orbit splitting, and tetragonal distortion/crystal-field splitting. The calculated MAE is found to be in good agreement with experimental values when the OP correction is included. Some of the materials considered here show large magnetic anisotropy of the order of meV. In particular we found that MnPt will have a very large MAE if it could be stabilized in a ferromagnetic configuration. Our analysis indicates that apart from large spin-orbit interaction and exchange interaction from at least one of the constituents, a large crystal-field splitting originating from the tetragonal distortion is also a necessary condition for having large magnetic anisotropy in these materials. Our calculation predicts large orbital moment in the hard axis in the case of FePt, MnRh, and MnHg against expectation.

The magneto-optic ͑MO͒ Kerr and Faraday spectra for manganese pnictides are calculated using the all electron, relativistic, full-potential linear muffin-tin orbital method. The amplitude of our calculated spectra are found to be in good agreement with corresponding experimental spectra. Although the MO property is a rather complicated function of the diagonal and off-diagonal elements of the optical conductivity tensor, present theory nevertheless provides very practical insight about its origin in these compounds. The largest Kerr effect observed in MnBi can be understood as a combined effect of maximal exchange splitting of Mn 3d states and the nearly maximal spin-orbit ͑s-o͒ coupling of Bi. The frequency-dependent optical properties, namely reflectivity, absorption coefficient, electron-energy-loss spectra, refractive index, extinction coefficient are given. From our calculations ͑including spin-orbit coupling and orbital polarization͒ the site-projected spin and orbital moments are also obtained and compared to the available experimental values and a good agreement is found. The magnetic anisotropy energy is calculated with a minimal number of approximations for the three systems. A disagreement between theory and experiment is found. Using the generalized gradient corrected fullpotential linear augmented plane-wave method we have calculated the unscreened plasma frequencies and the hyperfine parameters such as electric-field gradient as well as the hyperfine field. ͓S0163-1829͑99͒01419-8͔

No abstract

The matrix describing a trigonal distortion [Eq.(2)] is erroneous. It should read:where d is chosen to conserve the unit-cell volume. The calculations, however, were done with the correct matrix, so this change has no effect on the results or conclusions.

It is demonstrated that the coupling between spin and orbital moments in magnetic systems may-for certain materials-be reversed from antiparallel to parallel, via the influence of ligand states. This is exemplified by first-principles calculations for an intermetallic compound VAu 4 , but the effect may be found also in other classes of materials. From a computational analysis of the influence of the ligand states, and from an expression based on perturbation theory, we show that ligand states, that traditionally are known only to quench the orbital moment, may produce anomalous orbital moments that violate Hunds third rule.