A team of researchers from Claude Opus 5.5 has identified two candidate magnets for next-generation computer memory. These materials are predicted to exhibit zero net magnetism while still being able to sort electrons by spin. One candidate is a newly designed compound, while the other is a material first synthesized in 1999. The full calculations, code, and known caveats have been shared publicly.
The two types of magnets commonly recognized are ferromagnetic and antiferromagnetic. Ferromagnetic materials, such as fridge magnets, have atomic magnets that align in the same direction, resulting in a net magnetic effect. In contrast, antiferromagnetic materials have neighboring atomic magnets that point in opposite directions, canceling each other out.
Research in computer memory has focused on developing materials that fall between these two categories. The researchers provided a primer on magnetism, explaining that each electron possesses a quantum mechanical property called 'spin,' which contributes to its magnetic moment. This property is utilized in storage technologies, where information is stored based on the spin orientation of electrons.
Ferromagnetic materials naturally sort electrons by their spin orientation, while ordinary antiferromagnets do not. The new category identified, Luttinger compensated (LC) materials, features antiferromagnets where the spin-up and spin-down atoms possess equal magnetism, resulting in a net spin moment of zero. However, the atoms are situated in different environments, allowing for the separation of spins by energy levels.
The research team utilized quantum-mechanical simulations, specifically density functional theory, to analyze the crystal structures of potential candidates. The first candidate designed by the AI agents requires a perfect arrangement of manganese and iron atoms, which may be difficult to achieve in practice due to temperature-induced disorder.
The second candidate, KV[Cr(CN)₆], was previously reported in 1999 and has properties that align with the researchers' criteria. It has a zero net magnetism design and has been noted for its spin sorting capabilities. However, the only available sample showed a small residual magnetic moment due to water in its pores.
Both materials have been identified as promising candidates for spintronic devices, combining characteristics of ferromagnetic and antiferromagnetic materials. They are semiconductors with zero net spin moment and larger spin sorting windows than the thermal energy at room temperature, suggesting minimal interference with neighboring materials. The next steps include synthesizing KV[Cr(CN)₆] again and directly measuring its spin sorting properties. The research findings and computational methods have been made available on GitHub for public access.