By InsightArea Editorial Team
Making computer components smaller can remove the very properties that make an experimental material useful. Brown University researchers report a way to keep key electrical and magnetic properties in extremely thin films of bismuth ferrite, a finding that could inform future low-power electronics.
The university described the work on October 28, 2026, saying the research was published in Nature Materials. It is a materials experiment, not an announcement that a commercial AI chip is ready or that a measured reduction in an AI system’s electricity bill has been achieved.
Why thinness creates a problem
Chip components are built from layers whose dimensions must fit into densely packed devices. A material that works well in bulk may behave differently when reduced to an extremely thin film. In multiferroics, this can create a “dead layer” where the electrical and magnetic behavior researchers want no longer survives.
Multiferroic materials are interesting because electrical and magnetic properties are linked. In principle, an electric pulse could change a state that stores information, and a magnetic signal could help read it. That combination is attractive to researchers working on memory and logic, but an attractive physical principle is not yet a finished product.
What the Brown team changed
According to Brown, the researchers grew bismuth ferrite on a dysprosium scandate substrate. The difference between the materials’ crystal structures compresses the film and helps maintain its lattice. A metallic layer also helps address electric fields that could otherwise disrupt polarization.
Brown says the resulting films retained their functional properties at very small thicknesses and showed enhanced magnetic behavior. The researchers also describe an emergent magnetic state called altermagnetism. The result offers a way to investigate material behavior that had been difficult to preserve at this scale.
From a film to a working chip
A future device would need more than a promising material. Engineers would have to demonstrate reliable writing and reading, integrate the film with other components, assess manufacturing compatibility and test performance repeatedly. Energy savings for a complete computing system would depend on the eventual design and workload.
AI’s electricity use involves chips, memory, networking and cooling. Even a useful material for one component would address only part of that system. Comparing a laboratory film with a data center’s total consumption would require measurements that this university announcement does not provide.
The useful question for readers
The finding makes a previously difficult design space more accessible: can a material remain functional when made thin enough for future devices? The next question is whether that laboratory capability can be translated into practical, scalable components. As with other materials research covered by InsightArea, the distance between a demonstration and routine use is part of the story.
For readers interested in working through challenges, explore How to Do Hard Things.
Comments are closed.