All-optical switching of magnetic domains moves one step closer to application

The ability to switch magnetic domains with femtosecond laser pulses (i.e., all-optical switching, AOS) demonstrates the potential for optical control over magnetism. First observed in a rare-earth-transition metal alloy (gadolinium iron cobalt, GdFeCo),¹ similar effects have recently been discovered in other materials, including rare-earth-free magnetic multilayers. The exploitation of this behavior could enable the sustainable development of faster magnetic recording technologies.^{2, 3}

AOS was originally achieved using circular-optical polarization. Linear polarization has also been found to display AOS above a threshold fluence (i.e., radiant energy received by the surface).⁴ For this to be technologically meaningful, however, AOS must be able to compete with the bit densities of conventional storage devices. The technology must therefore be capable of restricting optically switched magnetic areas to sizes well below the diffraction limit. In the first demonstration of magnetic switching using all-optical methods,¹ switched areas with a diameter of 10μm (defined by the size of the laser spot used) were achieved. Using sample patterning in GdFeCo, AOS was subsequently localized to 200nm.⁵ Domain sizes of 300nm have been achieved in a magnetically harder material (terbium iron cobalt, TbFeCo) by focusing the light with a microscope objective and exploiting the AOS-threshold character (i.e., the sharp switching-probability increase, from zero to one, that occurs at a particular laser intensity).⁶

We have employed a two-wire plasmonic gold nanoantenna to obtain localized enhancement of the optical field, achieving unprecedented nanoscale control of AOS.⁷

We embedded the nanoantennas within a 10nm silicon-nitride capping layer on top of the active magnetic layer (TbFeCo), thereby enabling the near-field intensity enhancement to be exploited. The magnetically switched area is confined to dimensions that are defined by the antenna geometry. The two-wire antennas resonate with light that is linearly polarized parallel to the antenna axis.

To resolve the optically switched magnetic domain structure, we used an x-ray holographic imaging technique with a spatial resolution of 16nm. We obtained the magnetic contrast via x-ray magnetic circular dichroism⁷ at the SLAC National Accelerator Laboratory, Stanford. The reversible magnetic information was written using plasmonic nanoantennas via AOS with a laser fluence of 3.7mJ/cm². This is a lower energy than is required to induce AOS without the antennas.

Using a single-femtosecond optical laser pulse, we were able to write a reversible magnetic domain in a uniformly magnetized region below one end of the 230nm antenna arms: see Figure 1. The diameter of the switched area (53nm) is almost six times smaller than has been previously achieved. This domain size compares favorably to the track width of 55nm obtained in a recent demonstration of 1+Tb/in² heat-assisted magnetic recording.⁸

Figure 1. Reproducible and reversible antenna-mediated switching in terbium iron cobalt. (a) The initial magnetic contrast around an antenna after it is magnetically saturated in a field of 1.6 Tesla. The gold antennas consist of two aligned wires of 100nm each, with a narrow gap between them. (b) Magnetic contrast after the first laser pulse of 3.7mJ/cm². The polarization of a small domain (with a full width half-maximum of 53nm) is switched from negative to positive. (c) The magnetic domain is switched back in the original direction by the next laser pulse. M+: Positive magnetic polarization. M-: Negative magnetic polarization.

We have realized optically induced magnetic switching on the nanoscale using plasmonic nanoantennas. Our system has an energy requirement of less than 80fJ per bit. However, switching a single 53nm domain is not the same as writing information at Tb/in² densities. To achieve our aim of implementing AOS in real-world hard-disk drives, the antenna structure must be incorporated in a writing head similar to that developed recently for next-generation heat-assisted magnetic recording.⁸ Currently, sample inhomogeneities strongly affect switching probabilities, thereby reducing the reliability of AOS. It may be possible to overcome this problem using multilayer samples.

Optically driven random-access memory represents another potential application that we plan to explore. In this novel photonic-spintronic device, a photonic switching unit would be integrated with magnetic tunnel-junction stacks, the free layer of which would be optically switchable. We hope that our future work will enable the development of fast and energy-efficient memory as well as optically driven radio-frequency oscillators or sensors.

This work was supported by the US Department of Energy, the Office of Basic Energy Sciences, the Stichting voor Fundamenteel Onderzoek der Materie, the Netherlands Organisation for Scientific Research (NWO), the EU European Research Area project FENOMENA, and European Research Council Grant agreements 257280 (Femtomagnetism), 339813 (EXCHANGE), and EC FP7 281043 (FEMTOSPIN).