In the long search for computing that asks less of the world's energy, a film of bismuth ferrite only a few atoms thick now holds both its electrical polarization and its magnetism, a result from Brown University that seems to close a chapter in which materials simply fell silent when made small. For years, multiferroics promised memory that could be written with a trace of voltage and read by its magnetism, yet they faded at the very scales modern chips demand. The team's answer was not to force the material but to shape its surroundings, letting a carefully chosen foundation and a counteract
Brown University Team Eliminates 'Dead Layer' in Multiferroics, Opening Path to Low-Power AI Chips
The trick is learning how to engineer them into a form that's useful.
Start with the basics. Why does a material going "dead" when thin matter so much?
Because chips are built from extremely thin layers. A material that loses its electrical and magnetic behavior at that scale can't be used, no matter how good it looks in bulk.
And what did the Brown team actually change?
Two things. They grew bismuth ferrite on dysprosium scandate, whose slightly smaller crystal structure compresses the film and keeps its atomic arrangement intact. Then they added a metallic layer to counter the electric fields that kill the polarization.
Worth noting the claim here is that the dead layer is eliminated down to a few angstroms. That comes from Caretta, who led the work, and it's published in Nature Materials. I'd want to see other labs reproduce it.
Fair. What about the altermagnetism? That sounds like a bonus.
It is, and potentially a big one. Ferromagnets give strong readable signals but are slower and leak stray fields. Antiferromagnets are fast and compact but hard to read. Altermagnets are described as combining the strengths of both.
Careful, though. "Can be fast, densely integrated" is a description of potential. The reporting doesn't give speed or density figures for these films.
So what about the AI energy angle?
That's the motivation. Lower-power memory and logic could help with the electricity demands of AI hardware.
And it's a long way off. The source says the technology needs further development before commercial chips, and it offers no timeline or energy-savings number.
El Pulso
- A film of bismuth ferrite just a few angstroms thick still holds its electrical polarization and magnetism, ending the 'dead layer' that had blocked multiferroics from chips.
- Shrinking chip components had stripped these materials of their behavior, because atomic distortion and internal electric fields kept undermining polarization.
- Brown engineers, led by Lucas Caretta, pressed the film onto a dysprosium scandate base that squeezes its atomic structure, and added a metal layer to cancel disruptive fields.
- The layers also produced altermagnetism, a hybrid magnetism that could be fast, densely packed and still readable, promising denser chips.
- Commercial use remains distant, as the approach has yet to be developed into something manufacturable for everyday chips.
In the long search for computing that asks less of the world's energy, a film of bismuth ferrite only a few atoms thick now holds both its electrical polarization and its magnetism, a result from Brown University that seems to close a chapter in which materials simply fell silent when made small. For years, multiferroics promised memory that could be written with a trace of voltage and read by its magnetism, yet they faded at the very scales modern chips demand. The team's answer was not to force the material but to shape its surroundings, letting a carefully chosen foundation and a counteracting metal layer restore what thinness had taken away. As the demand for artificial intelligence swells, such quiet recoveries of lost behavior may matter as much as any new invention.
A film of bismuth ferrite barely a few atoms thick, from a team at Brown University, is still electrically polarized and still magnetic. The researchers say this ends the 'dead layer' problem, the obstacle that has kept multiferroics out of computer chips.
Multiferroics store information through polarization, a charge arrangement that a small voltage can flip. Because their magnetism is tied to that polarization, the stored bits can also be read magnetically. Together these traits could make memory and logic that use far less electricity. But modern chips need extremely thin layers, and at those sizes the atomic structure distorts and internal electric fields disrupt polarization, erasing both electrical and magnetic behavior.
Led by Lucas Caretta, the team attacked both causes. They grew bismuth ferrite on dysprosium scandate, whose slightly smaller crystal lattice compresses the film and preserves the structure it needs. A second metallic layer counters the unwanted fields. Caretta says the material can now be made "basically as thin as you want," and the work appeared in Nature Materials.
The films also displayed altermagnetism, which Caretta calls "the best of both worlds" for combining the readability of ferromagnets with the speed and density of antiferromagnets. The timing matters as AI pushes demand for lower-energy hardware. Still, the researchers stress that the approach must be developed further before it reaches commercial chips.
A film of bismuth ferrite just a few angstroms thick, roughly the width of a handful of atoms, is still electrically polarized and still magnetic. That result, from a team at Brown University, is what the researchers describe as the end of the "dead layer." It matters because that problem has kept a promising class of materials out of computer chips.
The materials are called multiferroics. Their atomic structure is built from repeating units in which positive and negative charges sit on opposite sides, which gives the material an electrical polarization. A small voltage can reverse that polarization by swapping the positions of the charges. That offers a way to store ones and zeros using very little electricity. Because the material's magnetism is tied to its polarization, the stored information can also be read back magnetically. The combination could allow memory and logic components that need far less energy than current technology.
The obstacle has been size. Modern chips need exceptionally thin layers, and multiferroics have traditionally stopped working properly at those dimensions. Scientists trace the breakdown to two effects. The atomic structure becomes distorted, which disrupts the arrangement and chemical environment the properties depend on. And internal electric fields interfere with the polarization until it can no longer be maintained. Together, they wipe out both the ferroelectric and the magnetic behavior, leaving the dead layer.
The Brown team, led by Lucas Caretta, an assistant professor of engineering, attacked both causes. They deposited bismuth ferrite, a widely studied multiferroic, onto a foundation of dysprosium scandate, a compound derived from rare earth elements. Its crystal structure is slightly smaller than that of bismuth ferrite, and the mismatch squeezes the bismuth ferrite under compression, which helps preserve the atomic structure its properties require. They also added a second, metallic layer designed to counteract the unwanted electric fields that would otherwise undermine the polarization.
The result, Caretta said, is that the dead layer is gone. "We show that you can make this material basically as thin as you want -- down to a few angstroms -- and it's still ferroelectric, and it's still magnetic," he said. The findings were published in the journal Nature Materials.
The team also found something they had not simply been aiming to preserve. The engineered layers caused the bismuth ferrite to show altermagnetism, a form of magnetism that combines useful traits of ferromagnetism and antiferromagnetism. Caretta explained the appeal. Ferromagnets produce strong signals that are easy to read, but they are relatively slow, and their stray magnetic fields make dense packing on a chip difficult. Antiferromagnets are much faster and avoid stray fields, but their signals are much harder to detect. Altermagnets, he said, are "kind of the best of both worlds": fast, densely integrated, and still able to give a strong electrical signal for readout.
The work arrives as artificial intelligence drives rising demand for computing power, which has made lower-energy hardware an increasingly important goal. Caretta believes the findings could bring multiferroics closer to practical use in low-power memory and computing devices. The technology still needs further development before it could appear in commercial chips. "We know the underlying properties are already there in the bulk materials," he said. "The trick is learning how to engineer them into a form that's useful -- and sometimes even better than what nature gives us."
The research was funded by the U.S. Army Research Office, the National Science Foundation and the U.S. Department of Energy. What to watch next is whether the approach can be developed into something manufacturable in commercial chips, which the researchers say has not yet happened.
Citas Notables
We've established that we can get rid of the dead layer.— Lucas Caretta, assistant professor of engineering, Brown University
Altermagnets are kind of the best of both worlds -- they can be fast, densely integrated and still provide a strong electrical signal for readout.— Lucas Caretta