Abstract
A theory of planar fluctuations of the direction of magnetization in a thin film which takes into consideration the nonlinear longitudinal magnetostatic force is outlined, and the resultant ripple is compared with that predicted by linear theories. The longitudinal magnetostatic force arises from the change along the direction of mean magnetization m0 of the component of magnetization M(r) parallel to m0; it leads to a torque that varies as the cube of the ripple amplitude. As a result of this torque, the linear theory breaks down for dispersion δ≳δ1, where δ is the rms angular deviation of M from m0 and δ1 is typically only 1° or 2° for a 1000 Å Permalloy film in zero field and varies as L−½, where 2L is the film thickness. With the inclusion of this force the range of validity of the theory has been extended to δ≲20°. A number of interesting effects have been found, including (1) an effective field in the direction of m0 (and proportional to δ4 for δ>δ1) acting on ripple components with wavelengths greater than 4πL; (2) an effective field 4πm0δ2, acting on shorter wavelength components; (3) as a result of (1), a decrease in the longitudinal and transverse ripple coherence lengths rl and rt (or, depending on viewpoint, a decrease in the longitudinal and transverse cutoff wavelengths); (4) a dispersion δ∝ (Krc)2/5, where K=local anisotropy energy and rc=crystallite size (in the linear theory δ∝ Krc); (5) almost no change in the mean ripple wall spacing as observed by Lorentz microscopy, ∼πrc, provided rc<rl; and (6) a spin‐wave reaction torque coefficient in high‐speed switching R∝δ10/3 (in the linear theory R∝δ2).