On a breezy afternoon last autumn in Cambridge, Mass., in a laboratory thrumming with the huff-whish-huff sound of refrigeration pumps, Massachusetts Institute of Technology graduate student Jiaruo Li was crafting a new device for storing digital data. She was aiming to use an exotic kind of magnetism discovered in the same lab the previous year to make the device faster and more energy-efficient than any competing technology. Her goal was timely given the current AI-driven boom in data centers and the exploding demand for power it portends.
At that moment Li was focused on finding her version of a needle in a haystack: a barely visible flake of nickel bromide with just the right attributes. To get to this point, she’d grown a dime-sized crystal of the compound by baking a glass tube containing nickel bromide powder for 10 days at high temperatures in a computer-controlled oven in an M.I.T. lab. Then, seeking an atomically thin sample, she’d applied a special tape to her creation, peeled it off and transferred the flakes on the tape to a shiny silicon wafer. Now, holding the wafer up to the light, she eyed a galaxy of thousands of tiny golden crystals against a purple mirrored background. “From all these,” she said, “only one or two of them is going to be thin enough.”
Nickel bromide is a sibling compound to nickel iodide, which made news in the spring of 2025 for displaying so-called p-wave magnetism, a phenomenon that had been predicted by theorists in early 2024. P-wave magnets exhibit behaviors that traditional magnets lack, including imparting special properties to electric currents passed through them. The breakthrough was just the latest in a series of revelations over the previous few years related to the discovery of a new class of magnets called altermagnets. These materials surprised many scientists by displaying a combination of attributes that could not only revolutionize computer hardware but rewrite our understanding of the physics of magnets. Equally remarkable: the new magnets weren’t actually new at all. Many were well-known, widely studied compounds with heretofore unrealized superpowers, and their abilities can be explained by simple geometry.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
The magnets of our everyday experience, the ones with north and south poles that keep children’s artwork stuck to refrigerator doors, are called ferromagnets and have been used extensively since prehistoric times. Still, it wasn’t possible to understand them until modern quantum theory was developed in the 1920s. In fact, says University of Oxford physicist Stephen Blundell, “the birth of quantum mechanics could have come from the observation of magnetism.” Physicists Niels Bohr and Hendrika Johanna van Leeuwen proved, independently and before modern quantum theory was devised, that magnetism is incompatible with classical—aka nonquantum—physics.
Magnetism originates in the quantum-mechanical property of electrons called spin. Spin makes an electron behave like a little rotating ball of charge, which furnishes it with a magnetic field similar to that of a tiny bar magnet. (That electrons are, as far as anyone knows, infinitesimally small and therefore not balls at all underlines how spin is an essentially quantum property.) When the spins of a large number of electrons in a crystalline solid align en masse so that their many minuscule magnetic fields combine to produce macroscopic effects, voilà, a ferromagnet is born. A ferromagnet’s most salient feature is its magnetization, or macroscopic magnetic field, consisting of lines of force that can become visible in the self-arrangement of iron filings sprinkled around the magnet.
The essence of magnetism is the organization of electron spins in a material, and ferromagnetism isn’t the only possibility.
Ferromagnets are immensely important in technology; Blundell calls them “the engine of the modern world.” Power plants, for instance, whirl magnets around to convert mechanical energy into electrical energy. And although most personal computers now rely on solid-state, nonmagnetic memory, the vast majority of the information stored in the world’s data centers takes the form of bits encoded in the magnetization of tiny regions of ferromagnetic hard-disk drives. “All of your data are stored in the cloud, and the cloud is all magnets,” says Jairo Sinova, a professor of physics at Johannes Gutenberg University Mainz in Germany and a key figure in the discovery of altermagnets.
The essence of magnetism is the mass organization of electron spins in a material, and ferromagnetism isn’t the only possibility. In the 1930s French physicist Louis Néel predicted that forces called exchange interactions could drive the spins in neighboring atoms to point in opposite directions rather than aligning, resulting in an up-down-up spin pattern, in contrast to ferromagnetism’s up-up-up. This alternating pattern would cancel out the magnetization generated by the spins so that the material would have no net magnetic field. Néel was awarded the 1970 Nobel Prize in Physics after experiments confirmed his prediction. Although “extremely interesting” theoretically, Néel said in his Nobel lecture, the “antiferromagnets” he’d discovered appeared to have no applications because of their lack of magnetization.
Jiaruo Li uses tiny wires to bond a p-wave magnetic tunnel junction sample to a silicon chip.
In the late 1980s, however, a different feature of magnets was revealed to be useful: they can adjust the electrical resistance of a material. Physicists Peter Grünberg and Albert Fert independently showed that if you place two parallel layers of ferromagnets in a device and then change the orientation of their magnetic fields, you can control how much electricity the device conducts. The phenomenon results from the fact that when a current passes through a ferromagnet, the spins of the electrons inside the current tend to align with the direction of magnetization. Physicists say the current has become “spin-polarized,” and a current polarized by one magnet will have an easier time passing through another magnet with the same direction. A resistance change achieved in this way is called giant magnetoresistance, or GMR, because a small magnetic field used to rotate the magnetization of one of the two layers can cause a huge change in resistance.
Grünberg’s and Fert’s discoveries won them the 2007 physics Nobel and launched the field of spintronics, in which spin is used to convey and store information, analogous to electronics doing the same with electric charge. Spintronics made a splash in the 1990s when IBM developed GMR-based “read heads” for extracting data from hard-disk drives: their exceptional sensitivity to magnetism led to more densely packed bits and a 1,000-fold increase in disk memory capacity. The same effect used in GMR—the differential conductance of rotated ferromagnets—has been used to represent 1’s and 0’s in a memory technology called MRAM, which has found a niche in the kinds of specialized computer chips used in cars, appliances and smartwatches.
For decades spin polarization seemed to be mainly a ferromagnetic effect. But the discovery of altermagnets has changed that. In fact, it has transformed how many experts understand magnetism generally. Rafael Fernandes, a physicist at the University of Illinois Urbana-Champaign, says this is a new way of thinking. “It’s like when I was a kid; I had to get glasses when I was 13 years old, and the moment I put on the glasses, I realized how much I couldn’t see before.”
The discovery of altermagnetism began with a mystery: an antiferromagnet that seemed to display a signature property of ferromagnetism. Sinova in Mainz and his collaborators, including Tomas Jungwirth, a professor at the Institute of Physics of the Czech Academy of Sciences, the University of Nottingham in England and Tohoku University in Japan, were studying the theoretical characteristics of the antiferromagnetic material ruthenium dioxide in 2018. Their calculations predicted that the material should exhibit a property called the anomalous Hall effect, which arises when a current passes through a material and creates a voltage—an electric force—much stronger than you would expect if the voltage were caused only by the magnetic fields in the material. This effect was thought to be a signature of ferromagnetism, but ruthenium dioxide was an antiferromagnet, so where was the phenomenon coming from? The theorists and their team had calculated the effect, but they didn’t feel like they understood it. “What the hell is going on here?” Jungwirth recalls thinking. Having predicted the effect in just one material, Sinova says, “you need to start asking: Is this a new type of magnetism? And for that you need to understand the symmetry that drives it.”
Symmetry, to a physicist, is a quality that lets an object undergo certain transformations without its properties changing. A circle, for example, is highly symmetrical because it can be rotated by any angle and still look the same. A square also has symmetry but less: a rotation leaves it unaltered only if the angle is an integer multiple of 90 degrees.
Li holds the sample, which is about 10 microns wide, in the fabrication area of the laboratory.
A signal achievement of 20th-century physics was the recognition of the role symmetry plays in the laws of nature. “In the 19th century we had these grand principles, which were kind of handed down on tablets from the great physicists: conservation of energy, conservation of momentum, conservation of angular momentum,” Blundell says. “But we didn’t quite know where they came from.” That changed in 1918, when German mathematician Emmy Noether showed that these three conservation laws followed from symmetries of physical laws under time shifts, spatial translations and rotations, respectively. Around the same time, Albert Einstein’s requirement that the laws of physics work the same irrespective of an observer’s motion led him to his theories of special and general relativity. And the specific medley of elementary particles and forces that is foundational for much of physics is now understood to stem from a particular combination of symmetries.
But it’s when symmetries are broken that things really get interesting. “If the universe had the highest-possible symmetry, which is full rotational symmetry and full translational symmetry, it would be completely featureless,” says Riccardo Comin, the physicist in charge of Li’s research group at M.I.T. “There would be no life, no planets, nothing.” Although the rotational symmetry of the physics governing subatomic forces, for example, implies that there is no special direction in space, at low-enough temperatures, in a process called spontaneous symmetry breaking, those laws can produce phenomena such as magnetism, which does single out a direction. And “it’s generally the case that the less symmetrical things are, the more rich and varied they are,” Comin says.
Case in point is ferromagnetism versus antiferromagnetism: the latter’s relative dearth of interesting and useful features follows from the fact that antiferromagnets retain a symmetry that ferromagnets break. Imagine flipping all the spins in a magnet so that they point in the opposite direction. An antiferromagnet will essentially stay the same, whereas a ferromagnet’s magnetization will reverse direction, and its north and south poles will be interchanged. Physicists say the ferromagnet breaks “time-reversal symmetry” and the antiferromagnet (mostly) doesn’t, because such a spin flip is what you would see if you could somehow reverse the arrow of time and thus the direction of each electron’s “rotation” [see graphic below].
Obviously we can’t really turn back time, but ruminating on the theoretical effects is a useful thought experiment. And it’s this breaking of time-reversal symmetry that enables ferromagnets to exhibit spin-polarized currents, the anomalous Hall effect, and other properties that antiferromagnets don’t have—or at least didn’t seem to until Sinova and his collaborators came along.
Libor Šmejkal, a former student of Sinova’s and Jungwirth’s who is now a researcher at the Max Planck Institute for the Physics of Complex Systems in Germany, ultimately came to realize—after a mess of complicated calculations requiring a supercomputer—that the solution to ruthenium dioxide’s mystery was hidden in the shapes of its atoms. The compound’s ruthenium atoms carry the spins that make it magnetic, but their electron clouds are deformed from their natural spherical shape by their oxygen neighbors. And, it turns out, the atomic clouds with spins in one direction are rotated by 90 degrees with respect to those with opposite-pointing spin. The resulting pattern of spins and shapes breaks time-reversal symmetry because a spin flip no longer leaves the magnet unchanged.
Moreover, Šmejkal observed, ruthenium dioxide retains a symmetry that endows it with special powers even ferromagnets don’t have. Reversing its spins (akin to theoretically reversing time) and then rotating the magnet’s atoms by 90 degrees brings the arrangement of spins and shapes back to where it started, and that symmetry gives it the ability to produce spin-polarized currents with spins that alternate with the current’s rotation—the property after which altermagnets are named.
In short, Šmejkal showed how the origin of a magnet’s magic lies in symmetries rather than its magnetization. Like antiferromagnets, altermagnets lack magnetization and have no net magnetic field. Yet because they break time-reversal symmetry, they can create many useful magnetic effects, such as the ability to polarize spins.
A signal achievement of 20th-century physics was the recognition of the role symmetry plays in the laws of nature.
Šmejkal went on to apply group theory—an area of mathematics that describes symmetries—to develop a system for classifying magnets and pinpointing their properties. “This type of symmetry [system] turned out to be supernutritious,” Šmejkal says, “because I was able to identify all these materials very systematically.” The system revealed, for example, that the three distinct types of magnets (ferromagnets, antiferromagnets and the new altermagnets) are the only possibilities for magnets whose spins are collinear, or parallel to one another (in these three cases, spins may point up or down but never off to the side at an angle). The system also provided a way to identify new magnetic materials. Šmejkal and his collaborators found more than 200 potential altermagnets by surveying databases of known materials. Many of these, like ruthenium dioxide, were well known, and no one had suspected they had any special powers at all.
At that point much of this research was still theoretical. Things changed in 2024, when a team led by Juraj Krempasky of the Paul Scherrer Institute in Switzerland made the first conclusive confirmation of altermagnetism. The researchers shot carefully calibrated light from a synchrotron particle accelerator at a crystal of manganese telluride to measure the energy, momenta and spin of its electrons. Their results proved that these properties conformed to the predictions made by Šmejkal and other theorists. Although manganese telluride is too fragile for commercial use, the result bodes well for spintronics because there are so many other potential altermagnets, says Peter Wadley, a physicist at the University of Nottingham who participated in the experiment. These materials “unite the advantages of ferromagnets and antiferromagnets in such a beautiful way,” Wadley says. “It’s like your fantasy magnet; it’s almost too good to be true.”
Qian Song dreams of one memory technology to replace them all. Computers typically use several, including speedy but volatile RAM that requires power to function, slower solid-state drives that store data for extended periods, and the magnetic hard-disk drives that make up the majority of the cloud. As a graduate student in Comin’s research group, Song—now a postdoc at the University of California, Berkeley—started the project that Li is working on. He first demonstrated the spiral-shaped p-wave variant of altermagnetism in nickel iodide in 2025 and believes it could be the key to a one-size-fits-all solution. “Why do we need all these types of memory?” Song asks. “I want to unify all the memories and push the speeds. The question is, Is there any physical limit?”
Altermagnets could come closer to that limit than anything else by combining the desirable properties of ferromagnets and antiferromagnets. Like ferromagnets, they can generate spin-polarized currents and effects such as GMR. And like antiferromagnets, they have spins that can be rotated about 1,000 times faster than those of ferromagnets, which could mean memory devices operating at terahertz versus the current gigahertz speeds. The absence of magnetization and the lack of sensitivity to magnetic fields that altermagnets share with antiferromagnets are also advantageous because they may allow engineers to pack many more of them into a small space. And the p-wave magnetism Song found in nickel iodide could be the key to a potentially huge increase in energy efficiency.
Li transfers p-wave magnet layers to build a magnetic tunnel junction, using a microscope and micromanipulators inside an argon-filled glovebox for handling atomically thin materials.
P-wave magnets belong to the fourth and final category of Šmejkal’s symmetry system: antialtermagnets, which break symmetry for a transformation called inversion that basically entails turning an object inside out. This kind of magnet does not have collinear spins—instead of pointing in parallel directions, the spins in nickel iodide, for instance, assume a triangular shape that rotates through the crystal. The resulting pattern is a helix, like the shape of a screw or a molecule of DNA. And like that of a screw, the helix’s rotation can be either right- or left-handed, a property called chirality that breaks inversion symmetry and enables a handy feature for Song and Li’s device: the ability to efficiently switch the magnetism’s chirality by applying an electric field.
While listening to Comin give a talk about Song’s experiments with nickel iodide in early 2024, Fernandes recalled a recent paper in which Šmejkal and his collaborators predicted p-wave magnetism, and he realized he was hearing something familiar. “It smells like a p-wave,” Fernandes recalls thinking. P-wave magnetism hadn’t yet been seen experimentally, and the idea was so new that Comin hadn’t even heard of it. But after collaborating with Fernandes, he realized that the theory underlying p-wave magnetism could explain the unusual and potentially useful properties of nickel iodide that Song had found. Meanwhile Song had already set his sights on building an all-purpose memory device: a p-wave version of a magnetic tunnel junction, the component in MRAM memory that stores a single bit.
Song estimates that such a device could write data using a mere one-hundred-thousandth (or less) of the energy required by any of the existing memory technologies—mostly because of the efficient electrical switching the p-wave magnet makes possible. But for the reading of data to also be as effective as possible, the nickel bromide crystal that will comprise one layer of the device must be atomically thin. That’s the quest that had Li looking for just the right flake.
Back in the lab, her arms were now shoulder deep in a pair of black rubber gloves that reached into a giant acrylic box. In her gloved hands lay a silicon disk dusted with nickel bromide crystals. She positioned the disk under an atomic force microscope whose needle would, over the next 10 minutes, trace out a path a few nanometers above the disk. That path would enclose a 25-by-25-micron area, and Li would then map the thicknesses of the crystals inside it. A computer screen by her side tracked the needle’s progress over a landscape of yellow, blue and green geometric shapes. Once the microscope had finished, Li drew a line with her mouse across a promising crystal’s edge on the screen and read off a number: 10 nanometers. Close but not thin enough. “That’s all I wanted to know,” she said.
The altermagnet discovery highlights the fact that there isn’t just one kind of unconventional magnetism but many. For one thing, magnets can be hybrids of Šmejkal’s four basic types and can therefore commingle their characteristics, much the way labradoodles combine traits of Labrador retrievers and poodles. Also, within Šmejkal’s categories are potentially many subcategories, which some theorists are now endeavoring to detail. And in the case of noncollinear magnets—including p-waves as well as oddities such as the whirlpool-like formations known as skyrmions—classifying them precisely is a fool’s game, says Igor Mazin, a professor of physics at George Mason University, because the possibilities are endless. “With collinear magnets, you can say that it’s up and down,” Mazin says. “When you step out to noncollinear things, then you can rotate more, less, in this direction, that direction.”
The big revelation from altermagnets isn’t just that there is a new type of magnetism but that in fact there are many.
Nickel iodide is a case in point. Fernandes says detailed symmetry analysis has revealed that despite “smelling” like a p-wave magnet, “it is not a proper p-wave,” because it doesn’t satisfy all of Šmejkal’s criteria for that category. It is the magnetic equivalent of a labradoodle that happens to look and act a lot like a p-wave magnet. And although experiments have now confirmed a handful of materials as altermagnets, they have also cast doubt on the status of ruthenium dioxide, the progenitor of them all. “The material itself probably is not magnetic,” Jungwirth says. “We chose a [somewhat] unfortunate first example in our theoretical papers.”
Despite those twists and turns, the identification of altermagnets has stimulated a surge of related research and high hopes in the spintronics community. The 2022 paper by Šmejkal, Sinova and Jungwirth that first used the term “altermagnetism” has since been cited more than 1,500 times. The reality, however, is that altermagnetism was just one in a series of breakthroughs over the past decade or so that have revealed a huge frontier in magnetism research. Pedram Khalili, a professor of electrical and computer engineering at Northwestern University, traces the start of the new era back to 2016, when the University of Nottingham’s Wadley and his collaborators showed that they could switch an antiferromagnet electrically, something Khalili says was previously thought impossible. “Gradually this understanding emerged that a lot of these things that we are used to associating with ferromagnets actually do not emerge from their magnetization,” he says. “That’s just a very superficial way of understanding them.”
What impact this new understanding will have on technology is an open question. Li and Song’s device, for example, will never see the inside of a computer or data center. Nickel iodide’s p-wave magnetism can survive only at temperatures below 60 kelvins, colder than liquid nitrogen, and Li made her measurements in an argon-filled glove box because nickel bromide, like nickel iodide, is a salt that melts if exposed to the water vapor that’s inevitably in air. “It’s not really practical, but what we think is that what we learn from nickel iodide will inform the search for new materials,” Comin says. “That’s one of the directions we’re exploring.”
The path from scientific breakthrough to commercial applications more generally is strewn with obstacles, says Stuart Parkin, director of the Max Planck Institute of Microstructure Physics in Germany and the scientist who developed GMR-based read heads for IBM in the 1990s. “Often what happens is a university person will find a phenomenon and think that’s the be-all and end-all, and you could use that for something,” he says. Usually, though, “you need not just one property [but] several properties.” Those properties include durability and cost-effective manufacturability, in contrast to Li’s artisanal efforts in the lab. Then there are matters of inertia and cost. “In the end, you typically need a device that somehow is superior to other devices by one or two orders of magnitude to warrant all the investment” in making a change, Parkin says. Even then, he estimates, it can take 10 to 20 years for a discovery to have commercial impact. GMR, for example, took about a decade, and the superstrong synthetic fiber Kevlar took two.
In any case, the notion of altermagnetic memory supplanting all competing technologies is “wildly unrealistic,” says Daniel Worledge, a senior manager at IBM who leads the company’s MRAM research and development effort. “I used to hear 25 years ago that MRAM was going to be a universal memory. It was going to replace flash, DRAM and SRAM and be best of all,” he says. “It’s just not the case because each of those is incredibly specialized and really good at what it does. And MRAM is really good at what it does, and there’s no one memory that’s going to be good at everything.”
Khalili leads a team that recently fabricated a promising magnetic tunnel junction based on noncollinear antiferromagnetism. It works at room temperature and is made from application-friendly materials. But he prefers not to speculate on its implications. “It’s a completely new device with completely new physics,” Khalili says. “It’s also an opportunity to really rethink the computing architecture completely. Maybe the biggest wins will be things we don’t even anticipate now.”