Dec 01, 2009

Fig. 1 : Magnetic nanostructure patterned at the PTA. Magnetic Force Microscopy images with (in dark) the up domains, and (in bright) the down domains. The propagation dynamics of the wall (red arrow) occurs on the central wire. The width of the wire is around 200 nm, the up domain (dark) appears in the upper left reservoir and expands into the nanostructure.

In collaboration with the Institut d’Electronique Fondamentale (Orsay) and Spintec, we have studied in detail the effect of an electric current within a magnetic nanostructure, in order to better understand the spin transfer phenomenon. This complex effect, involving interactions between electron spins and the magnetization of a magnetic material, is at the root of several promises of spintronics. Our experiments allowed us to clearly reveal this effect and to make progress in the theory.


Ferromagnetic materials are made of uniform magnetic domains separated by walls. (see inset). The displacement of these walls plays a major role in the magnetization reversal of thin magnetic films. A wall can be pinned, i.e. its movement can be temporarily stopped by crystal defects. It thus propagates discontinuously, jumping from defect to defect.


Spin transfer


In the last few years, both theoretical and experimental studies have shown the relevance of electronic transport for manipulating and detecting walls in nano-patterned samples. At the origin of this displacement is the spin transfer, a phenomenon which is still poorly understood but which allows us to foresee many applications in spintronics, ranging from logical devices to memories. Most studies, so far, have focused on samples with in-plane magnetization. In that case, the walls are thick (100 nm), the spin transfer has little efficiency, and the currents needed to displace the walls are too large for practical applications. Moreover, heating effects can come on top of spin transfer mechanisms.


For this reason, we work on systems magnetized perpendicularly to the thin film layers, in which domain walls are very thin (~1 nm). Most theories of spin transfer predict, in this case, an increase in its efficiency. We have deposited thin layers of FePt and CoNi that were then patterned, as seen in figure 1. In such a nanostructure, the resistance of the central wire indicates the position of the wall along the wire.


Fig. 2 : Position of the domain wall along the central wire, in the zone between the two crosses, in arbitrary units and as a function of time. Two successive measurements are displayed on this graph: in the first measurement (red), the wall pins to a defect for a duration t1. In the second measurement (blue), done in the same experimental conditions, the pinning duration t2 is very different, and we could show that it is actually random.

An effect that depends on the current direction


Let us now apply a magnetic field: the wall propagates within the nanostructure so as to favor the growth of the domain whose magnetization is in the same direction as the applied field. Then it bumps into a defect, onto which it gets pinned. After some times, it will then become depinned, due to the current influence. We have shown that this phenomenon is stochastic, i.e., that the pinning time is random from one measurement to another (Fig. 2). This stochasticity has two causes. On one hand, the wall can be pinned in various ways to a given defect, the various configurations corresponding to a more or less strong pinning. On the other hand, the depinning is a hermetically activated process. As a consequence, the distribution of depinning time probability distribution is exponential, with characteristic parameters linked to the pinning
strength. We observed in these experiments a clear spin transfer effect: for a positive current, i.e. for which the electrons push the wall out of the defect, the mean pinning time decreases. On the contrary, it increases for negative currents. This method allows us to measure the effect of spin transfer for small currents, without being bothered by Joule heating. Beyond the fact that we have shown that this stoschasticity has to be taken into account for the development of new applications, these experiments have allowed us to measure an important parameter of the spin transfer theory.



Spin transfer and domain wall


In a magnetic domain, the magnetization has a fixed direction. If the system is split into domains, the domain wall is the zone where the magnetization changes from one constant direction to another. Depending on the material properties, these domain walls can be more or less thick. In a ferromagnetic material, the electrical current is spin-polarized, meaning that there are more electrons with a spin in the direction of the magnetic orientation of the domain than in the opposite direction. An electron that flows from one domain to another crosses a domain wall. Its spin thus wants to align along the domain magnetization. Within the domain wall, the magnetization varies too rapidly and the spin can not follow the magnetization reversal. There is thus a torque between the electron spin and the local magnetization within the domain wall. This torque rotates the magnetization of the domain wall. As a consequence, the wall moves!


Further reading: C. Burrowes, et al., Nature Physics 6 (2009) 17



Last update : 02/20 2014 (970)


Retour en haut