Citation: Maniv, Eran, et al. "Antiferromagnetic switching driven by the collective dynamics of a coexisting spin glass." Science advances 7.2 (2021): eabd8452.
Web: https://arxiv.org/abs/2008.02795
Tags: Physical, Materials-applications, Phase-transition
I learned about this talk after James Analytis gave a talk at the IQIM seminar. I find the big idea fascinating. The idea is to use critical phenomena for ultra low-power computing. Near a critical point, a small change in parameters can drive a phase transition. That is, a small input can induce a huge response. This is exactly what one wants when making an ultra low-power component! The idea is to store your information in the phase of the sample - 0 if it is in one phase, 1 if it is in the other. A small change in parameters can drive a change in phase, and thus can be used to write-in information. The different phases will have different macroscopic properties, which can allow you to read-out the information.
This paper gives a great example of a specific material where you can apply this principle - Fe_(1/3)NbS_2 (one-third iron niobium disulfide), an iron-niobium-sulfide compound. At low temperatures (less than 42K), Fe_(1/3)NbS_2 is an antiferromagnet. This means that its ground state has an interesting crystalline spin texture. It has a stable critical point. Around this critical point there are two antiferromagnetic phases, distinguished by their spin textures. The idea is to store the information in the specific antiferromagnetic phase of the sample. One thing that is very nice about this material is that the phase transition can be induced by an electric pulse, and using a different electric pulse the phase transition can be driven the other way. An order paramter for this phase transition is the Neel vector, which defines an orientation of the domain. Current can flow in the direction of the Neel vector (low resistance), but it cannot flow perpendicular to the Neel vector (high resistance). Driving the phase transition corresponds to rotating the Neel vector 90 degrees. Probing this resistance allows for a totally electronic read-out of the stored information.
Remarkably, even though the phase transition can be driven entirely electronically the authors find that the phase is quite resistant to external magnetic fields (which is a problem for other proposals), as emphasized in this paper:
> Nair, Nityan L., et al. "Electrical switching in a magnetically intercalated transition metal dichalcogenide." Nature materials 19.2 (2020): 153-157.
One of the main problems for this particular approach is that Fe_(1/3)NbS_2 is only antiferromagnetic at low temperatures (less than 40K). Anything that needs to be hyper cooled is not going to fly for making large-scale classical computers. However, the principle is still really cool and there are other candidates in the same family as Fe_(1/3)NbS_2 which are antiferromagnetic at room temperatures.