# AI agents find two room-temperature magnetic semiconductor candidates

> AI agents and a researcher found two materials predicted to work for room-temperature spintronic memory, including a new compound with a 2.35 eV band gap. The second candidate, a Prussian blue analog from 1999, showed magnetism up to 376 K.

- Type: AI news
- Page: https://completeaitraining.com/news/ai-agents-find-two-room-temperature-magnetic-semiconductor/
- Published: 2026-10-06
- Updated: 2026-10-06
- Author: Jeroen Erne
- Jobs: Education, Science and research
- Reading time: 2 min

A team of AI agents working alongside a researcher has identified two candidate materials for room-temperature spintronic memory. The work, which combined automated AI search with density functional theory simulations, produced one newly designed compound and rediscovered a second from 1999. Both materials are predicted to exhibit near-zero net magnetism and spin-sorted electron windows large enough to overcome thermal noise at room temperature-properties that could enable more efficient data storage and low-power switching.

#### How the AI-led search worked

The AI agents and researcher focused on finding **Luttinger-compensated magnets**\-a class of materials where internal magnetic moments cancel out, resulting in almost no stray magnetic field. This cancellation is critical for dense spintronic devices, where cross-talk between neighboring bits must be minimized. The team ran simulations using density functional theory with PBE+U and HSE06 approximations to screen candidates for semiconductor behavior and spin-sorted electron transport.

The search surfaced two materials that met the criteria. Neither has had its spin-sorting properties confirmed experimentally, but the computational predictions point to clear electronic signatures worth testing.

#### YBaMnFeO₅: a new five-element design

The first candidate, YBaMnFeO₅, is a newly designed compound containing yttrium, barium, manganese, iron, and oxygen. Simulations predict a 2.35 eV band gap and spin windows of 1.0 eV for holes and 1.4 eV for electrons. Those windows sit well above the ~25 meV thermal noise threshold at room temperature, meaning spin-polarized currents could theoretically flow without being scrambled by heat.

A practical concern remains. The material's atomic ordering-essential for maintaining spin sorting-may scramble at typical synthesis temperatures between 900 and 1300 °C. Whether the structure can survive fabrication without losing its predicted electronic properties is an open question.

#### KV\[Cr(CN)₆\]: a compound from the archives

The second candidate, KV\[Cr(CN)₆\], belongs to the Prussian blue family of compounds and was originally reported in 1999. The team's simulations give it a 2.1 eV band gap, with spin windows of 2.6 eV for holes and 1.6 eV for electrons. Historical samples retained magnetism up to 376 K (103 °C), which is promising for room-temperature operation.

One complication is the compound's porous structure. Water molecules can occupy those pores and potentially weaken the magnetic and electronic effects. Like YBaMnFeO₅, the spin-sorting behavior has not been directly measured, so the 1999 data only goes part of the way toward validating the computational predictions.

#### Why this matters for science and research professionals

The study demonstrates a workflow where AI agents actively participate in materials discovery rather than simply accelerating existing human-led searches. For researchers in condensed matter physics and materials science, the two candidates offer concrete targets for experimental validation. The gap between simulation and synthesis remains the bottleneck-both materials face fabrication hurdles that will determine whether the predictions hold up in the lab. Courses that cover AI-driven simulation techniques, such as those found in [AI for Scientists Courses](https://completeaitraining.com/tag/science-and-research/), are becoming directly relevant to this kind of computational materials screening.

---

This is the Markdown form of https://completeaitraining.com/news/ai-agents-find-two-room-temperature-magnetic-semiconductor/. The whole site is mapped at https://completeaitraining.com/llms.txt.
