Nanoscale multiferroic nanodots show route to electric-field magnetic memory

Japanese researchers have demonstrated control of magnetic behaviour at the nanoscale in structures made from the multiferroic oxide BiFe0.9Co0.1O3 (BFCO), offering a potential route to magnetic memory that can be written with electric fields rather than dissipative electrical currents.

Why this matters

As cloud computing, artificial intelligence and large data centres expand, global energy use for information technology continues to climb. Conventional magnetic memory is non‑volatile but typically written by magnetic fields generated from electric currents, which causes energy loss as Joule heating. Multiferroic materials that combine electric polarisation and magnetic order promise a way to flip magnetic states using an applied electric field, cutting the energy needed to write data.

What the team did

A group led by Assistant Professor Kei Shigematsu, with JSPS Postdoctoral Fellow Koomok Lee and Professor Masaki Azuma, fabricated arrays of BFCO nanodots and examined their electric and magnetic behaviour. The work was carried out at the Institute of Science Tokyo (Science Tokyo) in collaboration with the Sumitomo Chemical Next‑Generation Eco‑Friendly Devices Collaborative Research Cluster, the Institute of Integrated Research at Science Tokyo, and the Kanagawa Institute of Industrial Science and Technology (KISTEC). Their results were published in Science Advances (DOI: 10.1126/sciadv.aec2861).

At nanometre dimensions, multiferroic materials can develop complex polarisation patterns known as topological domain structures. The researchers investigated how those domain arrangements relate to magnetic behaviour when the structures are subjected to electric fields — a critical step towards practical device use.

Findings and implications

The published study reports controlled coupling between electric and magnetic order in the BFCO nanodots. That coupling means an applied electric field can alter the magnetic state at scales compatible with dense memory elements. If such control can be integrated into device architectures, it could enable magnetic memories written by voltage, avoiding the energy losses inherent in current‑driven switching.

For the technology sector and large‑scale computing operators, the attraction is clear: lower write energy for non‑volatile memory reduces both operational power draw and cooling requirements. That has knock‑on benefits for cost and carbon footprint as data centres scale to meet rising demand from AI and cloud services.

Limitations and next steps

• The study demonstrates coupling and controllability at the nanodot level, not a finished memory device; engineering challenges remain to incorporate such materials into standard memory stacks.
• Topological domain structures at the nanoscale complicate predictability and reproducibility; further work is needed to ensure consistent switching behaviour across large arrays and under real‑world operating conditions.
• Integration with existing semiconductor manufacturing and endurance testing under repeated switching cycles will determine whether the approach can be commercialised.

Nonetheless, the work marks a concrete experimental advance by showing electric‑field control of magnetisation in a cobalt‑doped bismuth ferrite at dimensions relevant to dense memory. It shifts multiferroics from a materials curiosity towards a candidate technology for low‑energy non‑volatile memories, subject to the engineering hurdles above.

For national technology strategy, the implications are twofold: first, materials‑led reductions in memory write energy can meaningfully cut the power demands of large computing installations; second, sustained investment in device integration and manufacturing research will determine whether laboratory demonstrations translate into deployed hardware. The BFCO nanodot results provide a promising direction, but much engineering work lies ahead before electric‑field‑written magnetic memory becomes a commercial reality.