Vals AI Uses Claude Opus 5.5 Agents to Find Two Room-Temperature "Spin-Sorting" Magnetic Semiconductor Candidates: One Newly Designed, One Hiding in a 1999 Paper

In an October 4 post, AI evaluation company Vals AI said researcher Geby Jaff and a team of Claude Opus 5.5 agents found two candidate "Luttinger compensated" magnetic semiconductors that may work at room temperature: materials with zero net magnetic moment that can still sort electrons by spin, seen as a direction for next-generation spintronics. One is the newly designed YBaMnFeO₅, with a simulated band gap of 2.35 eV and a calibrated magnetic ordering temperature of about 490 K, but it has never been synthesized and its required atomic arrangement is likely to scramble during high-temperature synthesis. The other is KV[Cr(CN)₆], a Prussian blue compound the agents identified in 1999 literature, whose original sample stayed magnetically ordered up to 376 K. All conclusions rest on density functional theory (DFT) simulations, spin sorting has not been confirmed experimentally, and input files, raw outputs and analysis code are public on GitHub.

What material they were after

Spintronics aims to store and process information using an electron's spin, not just its charge. Ferromagnets can sort electrons by spin but carry a magnetic field, so devices can't be packed densely; ordinary antiferromagnets have no net field but can't separate spins. Vals AI went looking for "Luttinger compensated" materials: spin-up and spin-down atoms with equal magnetism, zero net moment, yet still able to sort electrons by spin. To be useful, such a material also has to be a semiconductor and stay magnetically ordered above room temperature.

Two candidates

In an October 4 post, Vals AI researcher Geby Jaff said Jaff and a team of Claude Opus 5.5 agents found two candidates. The first is the newly designed YBaMnFeO₅: across DFT simulations at two levels of accuracy (the faster PBE+U and the usually more accurate HSE06), it has a 2.35 eV band gap and a magnetic ordering temperature of about 420 K in the raw simulation, about 490 K after calibration. The problem is that it has never been made, and the checkerboard atomic arrangement it needs is likely to scramble at synthesis temperatures of 900–1300°C.

The second is more intriguing: the agents recognized KV[Cr(CN)₆], a Prussian blue compound, in literature from 1999. The original sample stayed magnetically ordered up to 376 K, with a band gap of about 2.1 eV. But that sample contained water and showed a residual moment of 0.125 Bohr magnetons, and the two simulation methods disagreed about water's effect.

What it means

This is not a settled case of "AI discovers a new material": both conclusions come only from simulation, spin sorting has not been confirmed experimentally, and the post doesn't say how many agents ran, for how long or at what cost, or whether outside experts confirmed the novelty. The author's stated next step is to make KV[Cr(CN)₆] again and measure it directly. What's worth noticing is where the second candidate came from: a useful material may already sit in the literature, unexamined from this angle, and combing papers and running simulations is exactly the grind agents are good at. Input files, raw outputs and analysis code are on GitHub for peers to check.

via: Vals AI post, GitHub data and code