Flux reversal in ferromagnetic thin films

Citation

Kryder, Mark Howard (1970) Flux reversal in ferromagnetic thin films. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/AP2X-M430. https://resolver.caltech.edu/CaltechTHESIS:03222012-143256016

Abstract

Flux reversal in Ni-Fe thin films has been studied with the use of 10 nsec exposure time Kerr magneto-optic photographs which depict the dynamic magnetization configuration with 10 µ resolution during the reversal process. The photographs show that at least five mechanisms are involved in the reversal of thin films: (1) domain wall motion, (2) coherent rotation, (3) diffuse boundary propagation, (4) nucleation and subsequent reversal of partially reversed regions, and (5) noncoherent rotation. The present investigation has been concerned mainly with the latter three mechanisms as they have either been previously unreported or poorly understood.

Non-coherent rotation is a complex rotational reversal process wherein the rate of the magnetization rotation varies over the surface of the magnetic film. It occurs only when transverse fields are applied. The Kerr magneto-optic photographs show that initially there is a fast relaxation of the ripple and a coherent rotation of the magnetization during the risetime of the pulse field, followed by a breakup of the configuration after the field exceeds the Stoner-Wohlfarth threshold. With fields just exceeding the Stoner-Wohlfarth threshold, after the stripes form, regions of reversed magnetization nucleate throughout the stripes and complete the reversal process. With larger applied fields, the reversal is completed by the slow (as compared to coherent rotation) rotation of the magnetization which is accompanied by a gradual decrease in amplitude of the stripes. The angle at which the stripes form is dependent on the applied field and varies from film to film, but indicates that the magnetization rotates coherently to angles significantly greater (5° to 20°) than the critical angle for reversal. The previously proposed fast relaxation models of Stein and Harte both predict that the coherent rotation should cease before the critical angle and therefore do not agree with the data. A model, based on ripple theory, has been constructed to show that after the magnetization rotates past the critical angle, an instability should occur in the ripple and cause a striped configuration at the observed angles.

The nucleation and subsequent reversal of partially reversed regions is a reversal process occuring predominantly with zero or small transverse fields. During this process the magnetization reverses in small (0.01 mm^2) regions of the film sequentially in time. With zero transverse field, the nucleation occurs with longitudinal fields exceeding a well defined nucleation threshold (>H_k) to fields greater than 2H_k. Anisotropy dispersion and magnetostatic stray fields are believed to be important in the nucleation process.

Diffuse boundary propagation involves a poorly defined, jagged, and diffuse boundary separating regions of anti-parallel magnetization. With zero transverse field the boundary lies transverse to the easy axis and propagates in the longitudinal direction -- just the opposite of domain wall motion. When a transverse field is applied, the boundary propagates rapidly from diffuse tips pointing in the direction of the stripes which are observed during non-coherent rotation. The boundary velocity varies roughly as the fifth power of the field and ranges from 0.033 cm/µsec at fields near H_c to 1.25 cm/µsec at 1.3 H_k which is one to three order of magnitude faster than domain wall motion. The high velocity is in part attributed to the large width of the boundary which widens from 0.2 mm to 2 mm as the field is increased.

Photographs taken when the drive field is terminated before saturation of the magnetic film show that the magnetization configuration continues to change for more than 200 nsec in the absence of an applied field. The diffuse boundaries become more sharply defined, though still quite jagged, and are found to change structure and propagate more slowly during subsequent pulses. Many of the partially reversed nucleated regions revert to the non-reversed state when the field is terminated. The striped partially rotated magnetization of the noncoherent rotation process, depending on the direction of the magnetization and the magnitude of the magnetostatic fields arising from the stripes, either relaxes to the non-reversed state or continues to reverse. Usually the static state shows no evidence of the stripes, and those which are observed are broken up by nucleated regions and frequently lie at different angles than the dynamic stripes. The large changes which occur after the field is terminated show that it is misleading to try to infer the dynamic state from the final static state as has been previously done. The long relaxation time is attributed to the sequential nature of the relaxation process.

Thesis Files