A qualitative change can occur, in a certain particle size, if the magnetization reverses by the curling mode, as demonstrated by superparamagnetic spheres, and by the initial susceptibility of ideally-soft spheres and disks. The theory of exchange resonance modes in small ferromagnetic spheres is extended to the case of a sphere whose centre is not ferromagnetic, as is the case when the magnetic particle is grown around a non-magnetic nucleus, such as the fine particles of the alloys of Fe, Co, and Ni that are made around a nucleus of a noble metal. A first-order calculation leads to a negligible modification of the eigenvalues for the inner radius used in the experiments. For larger holes, the change can be measurably large for some modes.
IntroductionThe theory of micromagnetics has never been able [1] to account for the magnetization processes in bulk ferromagnets. It is commonly believed that these processes are mostly governed by crystalline imperfections in hard materials, and by surface roughness and internal voids in soft materials. Both types help the nucleation of reversed domains on the one hand, and hinder the motion of the walls on the other hand. In both cases, however, the mechanisms are not sufficiently clear, and there is no satisfactory theory for them. There are many numerical simulations, but the results always look as if something is still missing there [2], so that they do not even seem promising yet, and it is not really known how to proceed.The situation is much better with fine ferromagnetic particles, for which it is energetically unfavorable to subdivide into domains, thus eliminating the most difficult theoretical problem of nucleation and growth of reversed domains. Theoretical nucleation fields are [3] close to the experimental coercivities of nearly-perfect whiskers for small whisker radius, but are off by several orders of magnitude for larger radii, thus indicating that the effect of body and surface imperfections is less important in fine particles than it is in larger ones. The computer resources necessary for numerical study of particles, by subdividing them into a sufficiently fine mesh, is also manageable when the particles are not too large. Moreover, there are now several experimental studies of individual particles, with which it should be easy to compare the theoretical results, without the problem of interactions between particles that complicated the analysis of older experiments. And there are some analytic results for ideal particles that can be used as a start and a guidance for the study of less perfect particles.It should thus seem natural for numerical micromagnetics to concentrate on extending the analytic results, by studying the effects of imperfections, and of other phenomena neglected in the old theory, such as surface anisotropy, magnetostriction, etc. For some historical reasons, however, numerical computations have been [2] mostly search-