Search

We study spin-diffusion effects within a continuously variable magnetization distribution, integrating with micromagnetics the diffusive model of Zhang and Li [Phys. Rev. Lett. 93, 127204 (2004)]. Current-driven wall motion is, in the steady velocity regime, shown to be adequately described by an effective nonlocal nonadiabatic parameter. This parameter is found to be 20% larger than its local counterpart for a vortex wall in a NiFe nanostrip and hardly modified for a transverse wall. This may account for the yet unexplained experimental evidence that vortex walls move more easily under current when compared with transverse walls. It is shown that this effective parameter can be derived from the domain wall structure at rest.

Experimental measurements of domain wall propagation are typically interpreted by comparison to reduced models that ignore both the effects of disorder and the internal dynamics of the domain wall structure. Using micromagnetic simulations, we study vortex wall propagation in magnetic nanowires induced by fields or currents in the presence of disorder. We show that the disorder leads to increases and decreases in the domain wall velocity depending on the conditions. These results can be understood in terms of an effective damping that increases as disorder increases. As a domain wall moves through disorder, internal degrees of freedom get excited, increasing the energy dissipation rate.

In some magnetic devices that have been proposed, the information is transmitted along a magnetic wire of submicrometre width by domain wall (DW) motion. The speed of the device is obviously linked to the DW velocity, and measured values up to 1 km x s(-1) have been reported in moderate fields. Although such velocities were already reached in orthoferrite crystal films with a high anisotropy, the surprise came from their observation in the low-anisotropy permalloy. We have studied, by numerical simulation, the DW propagation in such samples, and observed a very counter-intuitive behaviour. For perfect samples (no edge roughness), the calculated velocity increased with field up to a threshold, beyond which it abruptly decreased--a well-known phenomenon. However, for rough strip edges, the velocity breakdown was found to be suppressed. We explain this phenomenon, and propose that roughness should rather be engineered than avoided when fabricating nanostructures for DW propagation.

We evidence multiple coherent precessional magnetization reversal in microscopic spin valves. Stable, reversible, and highly efficient magnetization switching is triggered by transverse field pulses as short as 140 ps with energies down to 15 pJ. At high fields a phase coherent reversal is found revealing periodic transitions from switching to nonswitching under variation of pulse parameters. At the low field limit the existence of a relaxation dominated regime is established allowing switching by pulse amplitudes below the quasistatic switching threshold.

We demonstrate a quasiballistic switching of the magnetization in a microscopic magnetoresistive memory cell. By means of time resolved magnetotransport, we follow the large angle precession of the free layer magnetization of a spin valve cell upon application of transverse magnetic field pulses. Stopping the field pulse after a 180 degrees precession rotation leads to magnetization reversal with reversal times as short as 165 ps. This switching mode represents the fundamental ultrafast limit of field induced magnetization reversal.

As magnetic devices become smaller and faster, it is becoming increasingly important to understand the dynamics of magnetic domains on small spatial and temporal scales. In their Perspective, Miltat and Thiaville highlight the report by Acremann et al., whose magnetooptical microscope combines state of the art space and time resolution, thus allowing the precessional motion of magnetic domains to be studied directly. The dynamics they observe are complex, indicating that in both the space and time domains, neither the magnetization nor the applied field should be viewed as uniform.