In November 2025, at the National High Magnetic Field Laboratory in Tallahassee, Florida, a water-cooled resistive magnet hummed at 35 tesla—roughly 35 times the field inside a hospital MRI machine, and one of the strongest continuous fields producible on Earth. Inside the magnet's bore sat a crystal smaller than a centimeter square: ytterbium dodecaboride (YbB₁₂).
By the textbook, this is an insulator. It should not conduct electricity, should not conduct heat, and above all should not show quantum oscillations.
Yet University of Michigan physicist Lu Li and his team were watching clear oscillation waveforms appear on their instrument screens—and the waves were coming not from the crystal's surface, but from its interior.
"I wish I could say this has some great application," Li said, "but my work has kept pushing that dream further away. What we found, though, is genuinely bizarre and exciting."
A Metal's Spring, an Insulator's Silence
To understand why this is strange, consider what "quantum oscillations" are.
Electrons in a metal are not bound to atoms; they roam the entire lattice like wandering children. Apply a strong magnetic field and these electrons begin circular motion—fields bend charged particles. The circular motion produces oscillations whose frequency depends on field strength.
Li likes the metaphor of a spring: electrons are the spring, the magnetic field the hand plucking it. The stronger the field, the faster the spring vibrates. In experimental data this appears as periodic fluctuations in resistivity, magnetic susceptibility, or heat capacity versus field—the de Haas–van Alphen or Shubnikov–de Haas effects, classic phenomena discovered in the 1930s.
The key point: quantum oscillations require freely moving charged particles. Insulators have no free electrons—everything is locked in atomic orbitals or chemical bonds. Insulators should not show quantum oscillations, just as deserts should not have tides.
But over the past decade or so, physicists have observed oscillations in several insulators. The first assumption: it must be a surface effect. Topological insulators conduct on the surface while remaining insulating inside—if oscillations come only from that thin conductive skin, the story holds: surfaces are metallic, metals oscillate, all is well.
The question Li's team wanted to answer: surface, or bulk?
The Trial at the Strongest Magnetic Field
Answering this required extreme conditions. The 35-tesla field was not decoration—it was the instrument of judgment.
At such fields, surface and bulk effects leave different "fingerprints." Specifically, heat capacity versus magnetic field shows different oscillation frequencies and phases for surface versus bulk carriers. Li's team measured YbB₁₂'s heat capacity—a property that reflects the whole material, unlike surface-sensitive conductivity.
The result was clear: the oscillations come from the bulk, not the surface.
"We naively thought there was a good conducting layer on the surface that could be used for electronic devices—that picture is completely wrong," Li said. "The whole compound behaves like a metal, even though it's an insulator."
In other words, every inch of the crystal oscillates, every inch plays the role of a "metal"—but when you stop interrogating it with a magnetic field, it obediently goes back to being an insulator.
A New Duality
Li calls this a "new duality."
The naming is ambitious. A century ago, quantum mechanics introduced the "old duality"—wave-particle duality—which rewrote our understanding of matter. Light is both wave and particle; the electron is both particle and wave. Not "sometimes one, sometimes the other," but "both at once, depending on what question you ask."
Li suggests conductor-insulator may share this structure. Some materials, under some conditions, are both conductor and insulator—not a "division of labor" where the surface conducts and the bulk insulates, but the entire material holding both identities simultaneously.
The power of the analogy: wave-particle duality had no "applications" when discovered. It was just a headache-inducing paradox. Later came semiconductors, lasers, solar cells, electron microscopes—nearly all of 20th-century technology built on that paradigm shift.
The "new duality" also has no applications now. A 35-tesla field requires a national facility, far from everyday electronics. But, Li says: "We now have experimental evidence of this remarkable phenomenon. We've recorded it, and hopefully at some point we'll figure out how to use it."
What Exactly Is That "Thing"?
There is a deeper mystery still.
In metals, the carrier of quantum oscillations is the electron—a charged, massive spin-1/2 fermion. But YbB₁₂ is an insulator; its electrons are locked up. So what is oscillating?
The team acknowledged in the paper: they don't know.
"We still don't know what neutral particle is responsible for this observed phenomenon," said team member Yuan Zhu. "We hope this discovery will stimulate more experimental and theoretical work."
"Neutral particle"—that is the key clue. If the carrier is neutral, it is not an electron. It could be an exciton—a bound state of an electron and a hole, electrically neutral overall. Or some more exotic quasiparticle, perhaps a neutral fermion. In condensed matter physics, quasiparticle "identity" is flexible—collective excitations can behave like almost anything, as long as the quantum numbers match.
It resembles the dark matter problem in astrophysics: we see the effect but don't know the particle. In YbB₁₂, we see the oscillations but don't know the quasiparticle. We know it is electrically neutral, we know it is in the bulk, we know it only appears at 35 tesla—but its true identity remains a mystery.
Tides in the Desert
Return to the metaphor. Tides in a desert could mean: first, there's a sea under the desert (the surface explanation—now excluded); second, the desert itself becomes a sea under some conditions (the "new duality" explanation); third, something we don't yet recognize makes the sand ripple like water (the "neutral particle" explanation).
The third possibility is the most exciting, because it means our fundamental understanding of what can happen inside insulators is incomplete. The textbook says insulator electrons are locked up, hence no free carriers—but that "lockup" may only be a low-energy approximation. Under extreme conditions (35-tesla fields, near absolute zero), new collective modes may be excited—modes that never appear under ordinary conditions.
This resembles a phenomenon in AI research: emergent capabilities. Large language models show no certain abilities (multi-step reasoning, code generation) at small scale, but these appear suddenly past a threshold. Predicting large-model behavior from small-model behavior misses them entirely—they are not linear extrapolations.
YbB₁₂ is a textbook insulator at low fields, but at 35 tesla it "emerges" metal-like oscillatory behavior. Predicting high-field behavior from low-field data misses it entirely.
Extreme conditions are physics' "scale." Like parameter count in AI, timescales in evolution, energy scales in cosmology—many phenomena only appear under extreme conditions, and those are often the most profound.
A Discovery Without Applications
To close with Li's words: "I wish I could say this has some great application, but my work has kept pushing that dream further away."
It sounds self-deprecating, but it is science's most honest posture. Wave-particle duality had no applications at discovery. General relativity's only "application" was explaining Mercury's tiny orbital deviation. No one imagined CRISPR when the DNA double helix was found.
World-changing discoveries often start without applications—because they change our basic understanding of what the world is, not a concrete list of what we can do. Applications come later, bridges built by engineers and inventors in the new paradigm. But bridges need real ground, and the scientist's job is to discover that the "ground" is stranger, deeper, and less well-behaved than we assumed.
At 35 tesla, an insulator admitted it was also a metal. We don't yet know what that means. But we now know: the wall between conductor and insulator is not as thick as the textbook drew it.
Maybe the next edition will change "conductor or insulator" to "conductor and insulator"—just as today's textbooks already changed "wave or particle" to "wave and particle."
One word's difference—but the difference between entire worldviews.
---
References
- University of Michigan News: 'Really bizarre and exciting': The quantum oscillations are coming from inside (2025-11)
- Physical Review Letters: Quantum Oscillations in the Heat Capacity of Kondo Insulator YbB₁₂ (DOI: 10.1103/ms3x-pjsk)
- National High Magnetic Field Laboratory (35 Tesla magnet facility)