PerspectiveApplied Physics

Addressing an antiferromagnetic memory

See allHide authors and affiliations

Science  05 Feb 2016:
Vol. 351, Issue 6273, pp. 558-559
DOI: 10.1126/science.aad8211
Stable switching.

The flow of an electrical current reorients the alternating antiferromagnetic (AF) moments to lie across the direction of electron flow. Two different states, at right angles to each other, can be used to represent the “0” or “1” of a bit of digital data.


Spintronics (1) is one of the most commercially successful nanotechnologies. The invention of the giant-magnetoresistance spin valve (2) revolutionized the magnetic recording industry, enabling the immensely cheap, high-density disk drive storage on which the data centers that support our insatiable demand for cloud computing, social networking, and video sharing technologies rely. The combination of magnetic tunneling junctions (3) and spin-transfer torque (STT) (4) has brought about the prospect of magnetic random-access memory (MRAM), written using STT, as a way to provide nonvolatile storage that can be written and operated using extremely low power. Such technologies have just entered the marketplace. Even more energy-efficient writing is possible using the recently discovered spin-orbit torque (SOT) (5). All of these technologies use ferromagnets—the type of magnetic materials that “stick to the fridge” because they possess a magnetic moment—to store the data; the direction of that magnetic moment (“north” or “south”) represents the 0 or 1 of a digital bit. In both STT and SOT, the flow of a spin-polarized electrical current exerts torques on the magnetic moments to affect the change in magnetization direction that writes this bit. On page 587 of this issue, Wadley et al. (6) show that a hitherto exotic class of magnetic materials, antiferromagnets, are candidates for an entirely new type of spintronic memory.

The antiferromagnets were described by Louis Néel, in his 1970 Nobel lecture, as being theoretically interesting but technologically useless. This is because the magnetic moments on each atomic site, rather than all pointing in the same direction as in a ferromagnet, alternate in direction. This means that antiferromagnetic order is difficult to detect externally, as there is no net magnetization to measure, and even more difficult to manipulate, as there is almost no response to the application of a magnetic field. Unless one is a connoisseur, with sophisticated techniques such as neutron diffraction at hand, these materials appear to be nonmagnetic. To date, the sole technological application of antiferromagnets has been the stabilization of the magnetization in the reference layer of spin valves and magnetic tunnel junctions, finding a commercial market almost 30 years after Néel's lecture. Although there is only one way for all the magnetic moments to point in the same direction, there is a rich variety of interesting ways in which spins can alternate in direction on a three-dimensional crystal lattice.

Wadley et al. have built on the previous theoretical insight that some of these ways of alternating have a very special property: that the flow of an electrical current can exert alternating torques on the alternating moments, rigidly rotating the whole spin structure to alternate along a different direction (7) (see the figure). Having already demonstrated the growth of an appropriate material (8), CuMnAs, in a thin film suitable for fabricating microelectronic devices by the usual planar processing methods, and knowing that the different directions of the alternating moments give rise to slightly different electrical resistances (9), the stage is set for the breakthrough reported here: an antiferromagnetic device into which a digital bit can be written and read electrically and stored without power, in a way that is insensitive to magnetic fields.

This is what Wadley et al. have now demonstrated. By patterning an eight-armed “Union Jack” device from their CuMnAs film, they have all the electrical contacts needed to rotate the alternating spin structure in the device back and forth by 90° into a pair of stable states, and then subsequently detect these rotations electrically. Furthermore, they have used synchrotron-based x-ray microscopy to directly visualize the rotations in the magnetic order, confirming the nature of the electrical signals that are read out in their devices. As expected, there is almost no sensitivity to applied magnetic fields. All this was performed at room temperature.

This breakthrough reveals fertile ground for new spintronics device concepts based on antiferromagnets. The lack of magnetic moment and concomitant insensitivity to fields means that data stored in such a memory will be stable against any attempt (planned or accidental) to externally wipe it with a magnetic field. The devices can also be packed arbitrarily densely on a chip without fear of their disturbing one another's state. Because antiferromagnets display resonances in the terahertz frequency range, versus gigahertz for ferromagnets, they offer the prospect of extremely fast operation.

However, major obstacles to a competitive memory technology remain. The device presented by Wadley et al. produces a very small readout signal, but this problem could be ameliorated by using a tunnel-junction readout technique, where it is already known that the readout signals can be large (10). Such a vertical device geometry could also help to increase memory cell density in an array architecture; at the moment, the “Union Jack” lateral geometry will occupy a prohibitively large footprint. Solving these problems will entail some interesting device engineering projects. Of more fundamental interest is the slow relaxation of the antiferromagnetic order into its rotated state, which sometimes requires many pulses, each of tens of milliseconds, to induce the full resistance change. Determining whether current-driven dynamics can truly happen at terahertz frequencies remains an outstanding challenge. Nonetheless, identifying this promising class of antiferromagnets that can display this remarkable behavior is a breakthrough that will be seized upon by many other laboratories, not least because these effects are predicted in other materials systems (11). Given the vast number of antiferromagnets, it is unlikely that the optimal material has been discovered at this early stage. The search is now on to find it.


View Abstract

Stay Connected to Science

Navigate This Article