Abstract
A calculation of the first-order effect of spin-orbit coupling on a (3d)5 unperturbed S-state ion in a cubic crystal field has led to a new contribution σ1(r) to the ionic spin density. The integral over space of σ1(r) is zero and σ1(r) is perpendicular to the unperturbed spin density σ0(r); while σ0(r) is spherically symmetric about the nucleus, σ1(r) is highly aspherical. In α-Fe2 O3 at room temperature, the spin density consists of the large "antiferromagnetic" component σ0, plus the new term σ1, plus the weak ferromagnetic or Dzialoshinsky term σD, which is spherical. In neutron scattering, it is found that σ1 contributes to the same "ferromagnetic" Bragg peaks as does σD, and in the same order of magnitude. Hence the new term is probably important for understanding the surprising, highly aspherical ferromagnetic spin-density distribution recently observed in α-Fe2 O3 by Pickart, Nathans, and Halperin. In general, σ1(r) will be nonzero under much less restrictive symmetry requirements than those needed for the nonvanishing of σD. In the course of discussion, it is pointed out that the Dzialoshinsky-Moriya theory of the weak ferromagnetism in α-Fe2 O3, for example, implies that the application of a spatially uniform magnetic field should influence the distribution of domains of antiferromagnetic spin components. This effect was observed by Pickart et al.