Magnetic domains get a DUI for driving under the influence of DMI

Magnetic domains get a DUI for driving under the influence of DMI

This Week’s Pick of the Past article took a retrospective look at the excitement that magnetic bubble memory caused in the 1970s.  The basic principle of this bubble memory was the creation and expansion of cylindrical magnetic domains with magnetization orientations perpendicular to the plane of the magnetic material surface.  Come 2015, and some of the principles of magnetic bubble memory are still being investigated for potential use in modern day magnetic memory logics.  Certainly, as IBM were once excited about using domains for magnetic bubble memory, so too are they now excited about the use of magnetic domains in racetrack memory.  This week’s Pick of the Week, “Two-dimensional control of field-driven magnetic bubble movement using Dzyaloshinskii–Moriya interactions” in APL by Petit et al., looks at how the devilishly hard to pronounce Dzyaloshinskii–Moriya interaction (DMI) can be capitalised upon to move magnetic bubble domains in any direction in a 2D plane.

Accessing the magnetic state

Current approaches to magnetic memory in real world devices rely on the polarisation of a ferromagnetic layer.  The writing of the layer (i.e. setting of its magnetization orientation) is usually performed by applying a stray field, and the reading (i.e. detecting its magnetization orientation) is usually performed by a TMR type measurement.  For hard disk drives, the read and write operations require mechanical movement of a read head to access the relevant physical location on the spinning ferromagnetic disk for the data in question.  For MRAM, the read and write operations require the read and write lines to a specific cell to be activated.  In both cases, changing the magnetization orientation of a specific (part of a) ferromagnetic layer requires some mechanism to physically access that specific location.  Racetrack memory, which has previously been written about in Pick of the Week, is a completely different concept whereby magnetic domains move about within a ferromagnetic layer, usually driven by a spin-polarised current.  In this case, while the ferromagnetic layer itself remains static, the magnetic polarisation (domain) itself moves towards the polariser/detector as opposed to the polarised detector physically moving towards it.  For the concepts of racetrack memory currently under development, the thin width of the ferromagnetic wire which hosts the domains restrict their movement to one dimension (it is worth pointing out that this does not hinder, but in fact helps the intended operation).  In this week’s Pick of the Week paper, Petit and colleagues demonstrate the controlled movement of magnetic domains along two dimensions in the plane of the ferromagnetic layer.  This could have implications on the design of potential novel future data storage architectures.

DMI the key

The DMI has been know about for more than 50 years, but has only recently received a sizeable amount of attention from the physics community.  The DMI occurs at the interface between certain magnetic materials and non-magnetic metals, and is at heart caused by the combination of the spin-orbit interaction in the non-magnetic metal and the broken inversion symmetry of the interface.  The DMI manifests itself as a preferred relative alignment of spins in the magnetic material.  For example, it is responsible for the stabilization of the skyrmion spin texture.  The DMI has also recently been shown to be the reason for the reported high efficiency domain wall motion of Néel domain walls.

It has previously been shown that by applying  both an in and out of plane magnetic field, the unique symmetry of the DMI allows asymmetric expansion in the surface plane of magnetic “bubble” domains of parallel and anti-parallel to the plane magnetization orientations that are separated by a Néel wall.  Due symmetry considerations, reversing the directions of both the in and out of plane fields shrinks the domain in an asymmetrically opposite manner to that of the case of its expansion.  Petit uses this feature to design a clever process of magnetic field pulse steps that asymmetrically expand and shrink a magnetic domain such that at the end of the sequence, the size and shape of the domain is the same as it was at the beginning, but its position is not.  The pulse sequence is demonstrated on a Pt(10nm)/Co(0.8nm)/Pt(3nm) sample, where the Co layer has perpendicular magnetic anisotropy, and MOKE is used to image the magnetic domains.  A bubble domain is nucleated on a naturally occurring defect in the sample, expanded to a size of around 100μm in diameter, and then by applying the pulse sequence is moved by a distance of around 120μm at a speed of about 40 μm/s.  The speed that the bubble moves is proportional to its domain wall velocity, and therefore can be increased with larger field pulses.

New approach to magnetic memory?

For current memory architectures, Petit’s experiment is not particularly relevant.  However, when developing innovative new approaches to magnetic memory design it may one day bear fruit.  Racetrack memory, with its 1D domain wall motion, has shown that the movement of domains is a viable approach to storage.  Moving domains in 2D may offer an additional degree of freedom (and therefore considerably more memory) if a complementary architecture for reading can be conceived.  From a device standpoint, one obvious drawback to Petit’s experiment is the need to apply fields along the x, y, and z axes – this would require a more complex arrangement of write lines for the application Oersted fields.  I am not sure whether the symmetry of the DMI would allow the same domain movement if only an in-plane spin polarised current were applied – I suspect some combination of damping and anti-damping torques would be required, and this may not be achievable.  Nevertheless, Petit’s work is an interesting demonstration of how a correct understanding of the DMI can allow for such a nice experimental result.  Moving bubble domains in any in-plane direction without the need for any type of lithographic patterning could be a result that lives long in the memory.


Image: Thomas Anderson, some rights reserved

Petit, D., Seem, P., Tillette, M., Mansell, R., & Cowburn, R. (2015). Two-dimensional control of field-driven magnetic bubble movement using Dzyaloshinskii–Moriya interactions Applied Physics Letters, 106 (2) DOI: 10.1063/1.4905600