Within the enormous digital revolution, solid-state devices based on spin degree of freedom promise energy-efficient, fast and non-volatile features to contemporary complementary metal-oxide-semiconductor (CMOS) devices. Spin-based devices such as the magnetoresistive random access memory (MRAM) have been recently commercialized, whereas a complete overhaul of the charge-based counterpart with the spin devices yet requires constant discoveries and innovations.

Electrical switching of magnetic elements lies at the heart of MRAM devices. The relative spin orientation of vertically stacked two nano-magnets exhibits two distinct resistive states and thereby determines the data between ‘0’ and ‘1’. The reading current, on the other hand, inevitably undergoes the current shunting through the write channel, which disturbs the data and induces energy dissipation via thermal loss (i.e. Joule heat).

Magnetic insulators are electrically insulating but good for spin angular momentum transfer. Magnons, quanta of spin waves, carry the angular momentum without moving charges and thus enable the Joule-heat-free spin transport. Our team led by Prof. Hyunsoo Yang at National University of Singapore demonstrated, for the first time, the magnon-mediated switching based on a topological-insulator-Bi₂Se₃/antiferromagnetic-insulator-NiO/ferromagnet trilayer [Wang et al., Science 366, 1125-1128 (2019)]. The top ferromagnetic NiFe or CoFeB layer is patterned with discontinuity and electrically isolated from the bottom spin source layer by inserting the insulating NiO layer, therefore the writing current shutting effect to the ferromagnetic layer was eliminated. The writing current through the Bi₂Se₃ layer is converted to the magnon current in the NiO layer, which is followed by the magnon-torque-induced magnetization switching. The magnon-assisted electrical switching, wherein the current shutting induced thermal loss can be entirely minimized, is of importance for the low-dissipative control of spintronic devices.

For energy-efficient devices, however, not only low dissipation but also high speed is required because the power consumption is proportional to the device's operation time. Over the last few decades, it is believed that the spin transport dynamics is governed by the spin-wave vector-energy relation, i.e., magnon dispersion, by which one can estimate the magnon velocity. So far, the magnon dispersion has been estimated indirectly with inelastic neutron scattering using mm-size samples. In our group, on the other hand, we throw a fundamental question if the dynamics of magnons at the nanometer scale is still governed by the magnon dispersion that is estimated based on a micro (or even larger)-scale. For instance, it is expected that the antiferromagnetic magnons propagate at a speed of ~10 km/s and thereby it takes ~10 ps at 100 nm distance. However, a direct estimation of the propagation time remains challenging due to the lack of sufficiently fast probes.

Optical-driven terahertz (THz) emission is a well-established technique to estimate the spin-to-charge conversion and has been proposed as a new type of efficient THz emitters, wherein a ferromagnet/heavy metal (FM/HM) bilayer is typically used. Our idea is to insert an antiferromagnetic insulator between FM and HM layers. Indeed, a time-resolved nature with a temporal resolution of sub-picosecond in the THz emission system enables to measure the magnon propagation time (< 1 ps) through the antiferromagnetic insulator at nanometer distances (≤ 50 nm). The obtained magnon velocity up to 650 km/s is approximately one order magnitude larger than the group velocity indirectly obtained from the bulk dispersion.

It has been known that the dispersion becomes anomalous for dissipative (or absorbing) media, resulting in the group velocity larger than the fundamental limiting velocity, called the faster-than-light or superluminal propagation. Our collaborator, Prof. Kyung-Jin Lee in Korea Adv. Inst. Sci. Technol. (KAIST) suggests a theoretical model that a finite-damping at nanoscale induces an anomalous magnon dispersion and subsequently yields a superluminal magnon velocity for small magnon wavenumbers. This is of considerable interest from the application point of view because it allows one to operate nanoscale magnonic devices at a far higher speed than what has been thought to be the fundamental limit of magnonic devices. It is also of importance from the viewpoint of fundamental science as it has been believed that the damping always results in a detrimental effect on various spin-related phenomena.

For more information, please check out our recent publication in Nature Nanotechnology:

Kyusup Lee, Dong-Kyu Lee, Dongsheng Yang, Rahul Mishra, Dong-Jun Kim, Sheng Liu, Qihua Xiong, Se Kwon Kim, Kyung-Jin Lee & Hyunsoo Yang, “Superluminal-like magnon propagation in antiferromagnetic NiO at nanoscale distances”. Nature Nanotechnology (2021).

https://doi.org/10.1038/s41565-021-00983-4