Band magnetism in the spin‐density‐functional formalism

1978 ◽  
Vol 49 (3) ◽  
pp. 1399-1404 ◽  
Author(s):  
O. Gunnarsson
1981 ◽  
Vol 59 (4) ◽  
pp. 500-505 ◽  
Author(s):  
A. H. MacDonald ◽  
K. L. Liu ◽  
S. H. Vosko ◽  
L. Wilk

Two suggested nonlocal approximations for the spin–spin exchange-correlation interaction functional of the spin-density functional formalism have been applied to the calculation of the Pauli susceptibility, χp, of the alkali metals. The nonlocal approximations were found to imply values of the density-functional Stoner parameter, I, typically ~ 3% lower than values implied by the more usual local approximation. This qualitative trend was found to be supported by comparison of available experimental values of χp with new more accurate theoretical values for the local approximation to χp.


2009 ◽  
Vol 9 (S9) ◽  
pp. 83-93 ◽  
Author(s):  
O. Gunnarsson ◽  
P. Johansson ◽  
S. Lundqvist ◽  
B. I. Lundqvist

1981 ◽  
Vol 59 (7) ◽  
pp. 888-896 ◽  
Author(s):  
L. Wilk ◽  
S. H. Vosko

The augmented plane wave method, in the muffin-tin approximation, was used to perform self-consistent spin-polarized calculations of the electron number densities, n(r), and spin magnetic moment densities, m(r), within the framework of the spin density functional formalism. For the exchange-correlation energy functional we used the new improved results of Vosko and coworkers in the local spin density approximation. The calculations were carried out for the range of volumes 0.9 ≤ V/V0 < 1. The contributions to the total Fermi contact term, m(0), relative to the Fermi surface contribution, were approximately −24, −2, and −2% from the core electron polarization and −0.7, −5, and −6% from the polarization of the valence electrons below the Fermi surface, for Li, Na, and K respectively. In Li the volume dependence of m(0) was strongly affected by the volume dependence of the core contribution, which differed markedly from that of the Fermi surface contribution. In both Li and K the calculated volume dependence for small volume changes was in good agreement with recent experiments, while in Na the trend was correct but the slope of the curve was too large. In all cases the results were very sensitive to achieving a high degree of self-consistency. The absolute values of m(0) are in excellent agreement with the measured Knight shifts.


1981 ◽  
Vol 59 (4) ◽  
pp. 585-595 ◽  
Author(s):  
L. Wilk ◽  
M. Nusair ◽  
S. H. Vosko

The linear augmented plane wave method in the muffin-tin approximation was used to perform self-consistent spin-polarized calculations of the electron number density n(r) and spin (magnetic moment) density m(r) in metallic Be, within the framework of the spin density functional formalism. For the exchange-correlation functional we used the recent accurate results of Vosko et al. in the local spin density approximation. The Fermi contact contribution to the Knight shift is proportional to the sum of three spin densities (evaluated at the nucleus) arising from (i) the valence electrons at the Fermi surface, (ii) the core electrons, and (iii) the valence electrons below the Fermi surface. We find a 90% cancellation between (i) and (ii) which greatly magnifies the significance of the relatively small effect (iii). Although our contact term is still positive in sign, its magnitude is nearly one-fourth of the previous smallest first principles result and thus requires a smaller orbital diamagnetic contribution than previously invoked to explain the negative experimental value of the Knight shift.


Sign in / Sign up

Export Citation Format

Share Document