The operating speed of devices based on magnetoresistive effects is determined by the time it takes to reverse the magnetization direction in one or more ferromagnetic layers in ultrathin films and multilayers. A thorough understanding and control of the magnetic interaction in such samples on short timescales is inevitable for establishing faster switching schemes that take advantage of the magnetization precession ("precessional switching").
To investigate magnetization dynamics in “real time”, the sample is excited by short magnetic field pulses. The response of the sample to the field pulses is stroboscopically investigated by magnetic microscopy using photoelectron emission microscopy (PEEM) with X-ray magnetic circular dichroism (XMCD) providing magnetic contrast. Element-resolved microscopic magnetic domain images reveal the influence of, for example, coupling effects in magnetic multilayer systems or the fast magnetization reversal of one of the layers.
Example 1: FeNi/Al2O3/Co
Panels b–k of the figure show magnetic domain images of the FeNi layer in a FeNi/Al2O3/Co spin-valve structure during the application of bipolar magnetic field pulses of 80 ns length (a). Images c–e represent the domain structure during the positive part, images f–j the domain structure during the negative part of the pulses. The domain structure of the magnetically harder Co layer stays the same at all times.
The comparison of image l with images f or k shows a preferential nucleation of reversed domains in the FeNi layer during the fast magnetization reversal at the position of the domain boundaries of the Co layer. This is an important finding, since it shows that t he local switching speed in devices containing layered magnetic systems is increased by the presence of domain walls in the other layer.
Only a technique like time-resolved XMCD-PEEM, which allows at the same time microscopic spatial resolution, sensitivity to the different magnetic layers, and time resolution, is able to identify such effects.
Micromagnetic simulations support the experimental observations and reveal that the locally enhanced nucleation is due to magnetic stray fields of the domain walls in the Co layer. The magnetostatic interaction between the domain walls in the Co layer and the FeNi layer induces a transverse magnetization component (x direction) in the FeNi layer. The external magnetic field along the y direction exerts then a higher torque on the FeNi magnetization in these regions, leading to a faster reversal.
The effect of domain wall stray fields can be controlled by manipulating the width and position of the domain wall. This may represent a path towards increasing the local speed and reproducibility of magnetic switching.
These experiments have been conducted in collaboration with Jan Vogel, Stefania Pizzini, Fabien Romanens, and Alain Fontaine of Institut Néel, CNRS Grenoble (France), Julio Camarero, Universidad Autónoma de Madrid (Spain), Marlio Bonfim, Universidade do Paraná (Brazil), Frédéric Petroff, Unité Mixte de Physique CNRS/Thales (France), and Jürgen Kirschner, Max-Planck-Institut für Mikrostrukturphysik Halle (Germany). The micromagnetic simulation has been calculated by Riccardo Hertel, FZ Jülich (Germany).
Example 2: FeNi microstructures
The figure shows experimental and simulated XMCD-PEEM images of the magnetization of a 5 µm x 5 µm x 20 nm FeNi microstructure, exposed to magnetic field pulses of 80 ps rise and fall time. Panels c-e represent the response of the magnetization in the microstructure to the magnetic field pulse. On the right hand side the difference images XMCD(t) - XMCD(t0) are shown (color images). The simulation was performed using OOMMF.
The most pronounced effect in the magnetic contrast appears 170 ps after the onset of the magnetic field pulse. This is when the magnetization in the gray magnetic domains, initially horizontally magnetized, displays a maximum change when compared to the reference image, decreasing their brightness (negative change displayed in blue in the difference image). This reproduces the expected behavior, since the magnetization in these domains is perpendicular to the applied magnetic pulse, thus maximizing the magnetic torque caused by the field pulse. At higher delay times the XMCD contrast in the gray domains vanishes and reverses its sign, following a damped quasi-sinusoidal behavior with a main frequency of ~3.3 GHz.
Another precession frequency is observed at the 90° domain walls dressing the black and white domains and at those of the cross-tie domain walls (panel d). Compared to the magnetization precession in the gray domains, these domain walls have a longer oscillation period as indicated by the blue and red difference contrasts.
The simulation also reveals the reaction of the vortices and antivortices found in the magnetic microstructure. In the figure below, simulations for one vortex and one antivortex are shown. The red and blue color represents the z component of the magnetization.
The selected time points show the behavior of a down vortex and a down antivortex during the first half precession period of the gray domains discussed above. The z component of an antivortex is stabilized through the horizontal domain walls: Their precession frequency is lower than the one of the gray domains. On the other hand the vortices are mainly surrounded by gray domains, which all precess with the same frequency, but opposite phase with respect to the oscillation of the z component (b). This leads to a dipolar structure consisting of two vortices and an antivortex (c). Under the creation of circular spin waves the vortex–antivortex pair annihilates (d,e), and the same situation as in (b) results, but with opposite mz component in the gray domains and in the vortex.
This scenario repeats each half precession period of the gray domains, but not every time the vortex switches: Either the original or the new vortex persists. After the third vortex–antivortex annihilation the precession of the gray domains is so far damped that the amplitude of the created vortex–antivortex pair is too low to switch the vortex. The switching behavior of a vortex is the high field alternative to the vortex gyration, where no dipolar structure is created.
Publications about this research:
J. Vogel, W. Kuch, R. Hertel, J. Camarero, K. Fukumoto, F. Romanens, S. Pizzini, M. Bonfim, F. Petroff, A. Fontaine, and J. Kirschner
Influence of domain wall interactions on nanosecond switching in magnetic tunnel junctions
Phys. Rev. B 72, 220402(R) (2005).
K. Fukumoto, W. Kuch, J. Vogel, F. Romanens, S. Pizzini, J. Camarero, M. Bonfim, and J. Kirschner
Dynamics of Magnetic Domain Wall Motion after Nucleation: Dependence on the Wall Energy
Phys. Rev. Lett. 96, 097204 (2006).
J. Miguel, M. Bernien, D. Bayer, J. Sánchez-Barriga,
F. Kronast, M. Aeschlimann, H. A. Dürr, and W. Kuch
A new sample holder for laser-excited pump-probe magnetic measurements
on a Focus photoelectron emission microscope
Rev. Sci. Instrum. 79, 033702 (2008).
J. Miguel, J. Sánchez-Barriga, D. Bayer, J. Kurde, B. Heitkamp, M. Piantek, F. Kronast, M. Aeschlimann, H. A. Dürr, and W. Kuch
Time-resolved magnetization dynamics of cross-tie domain walls in Permally microstructures,
J. Phys.: Condens. Matter 21, 496001 (2009).