Vienna's 2D Quantum Ground State: How a 150-Nanometer Silica Rotor Was Pinned to Heisenberg's Limit in Two Directions at Once

Imagine a compass needle, only now the needle is a hundred million atoms fused together, floating in a laser beam in a vacuum chamber. Tap the trap, and the tip wobbles. Keep cooling, and the wobble shrinks — first to the width of a bacterium, then to a hundredth of the diameter of a single atom. Push further, and the wobble stops getting smaller. Not because the apparatus has run out of headroom, but because quantum mechanics has. At that floor, the needle is no longer mostly still. It is as still as nature allows a compass needle to be.

That is the floor a team at the University of Vienna, TU Wien, and Ulm University says it has reached — and reached in two directions at once. According to coverage by Phys.org of the group's April 6, 2026 Nature Physics paper, the researchers cooled a levitated silica nano-dumbbell to its two-dimensional librational quantum ground state, driving the orientational uncertainty down to roughly 20 microradians — the Heisenberg-limited zero-point fluctuation for both tilt axes simultaneously.

Why the "2D" Matters More Than It Sounds

Quantum ground-state cooling of levitated nanoparticles is not new. Back in 2020, a group including Uroš Delić — then at the University of Vienna, now leading his own team at TU Wien — reported cooling a silica sphere's translational motion to the quantum ground state along a single axis. A 2023 follow-on extended that to two translational modes at once. Earlier single-axis librational cooling was demonstrated by Lukas Novotny's group at ETH Zürich, credited as the prior benchmark by Interesting Engineering.

The 2026 Vienna result is a different animal. It is not translational; it is orientational. And it is not one-dimensional; it is two-dimensional. A levitated nano-dumbbell can tilt in more than one direction — imagine the compass needle wobbling not just east-west but also up-down. Cooling one axis while ignoring the other is an incomplete description of quantum motion. Both axes need to be in their ground states simultaneously, with neither stealing thermal energy from the other through cross-coupling, for the object to genuinely occupy what physicists call a two-dimensional librational ground state. That is the milestone Phys.org describes as a first.

The distinction between rotation and libration is easy to fumble in headline prose. The rotor is not spinning freely. Caught by the optical trap, its long axis oscillates back and forth — librates — around a mean orientation. Cooling shrinks the amplitude of that oscillation. At the floor, the amplitude is set by Heisenberg's uncertainty principle, not by how cold the apparatus is. Push harder; nothing further happens. The uncertainty is a feature of the quantum description, not a residual of imperfect engineering.

What Got Frozen

The object at the center of this experiment is deliberately simple. Per the University of Vienna release carried by Mirage News, the rotor is a nano-dumbbell formed from two silica spheres, each roughly 150 nanometers in diameter, trapped in ultra-high vacuum by laser light at intensities of about 100 megawatts per square centimeter. Stephan Troyer, the paper's lead author, offered two analogies widely quoted in coverage: the tip of the rotor "moves less than one hundredth of the diameter of a single atom," and the alignment precision is "like a compass needle oriented to better than the width of a bacterium."

Those analogies, stripped of the poetry, describe an orientational uncertainty of about 20 microradians in each librational axis. That is the zero-point fluctuation of the two coupled harmonic librational modes when each is in its ground state. According to Phys.org, the effective temperature hovers at a few tens of microkelvin above absolute zero, and the total object in question comprises roughly 100 million atoms. That number matters. Most quantum experiments use single atoms, single ions, or single molecules. The Vienna rotor is massive by quantum standards — several orders of magnitude heavier than the largest molecules used in matter-wave interferometry to date — and it still behaves as a quantum object with respect to its orientation.

How Single Photons Pulled Heat Out of a Tumbling Dumbbell

The cooling mechanism is coherent scattering into an optical resonator — a technique the Vienna group has been refining for the better part of a decade. A laser beam traps the nanoparticle and, crucially, the particle scatters light coherently into a nearby high-finesse optical cavity. When a scattered photon enters the cavity, it can leave the cavity at a slightly higher frequency than it entered. The extra energy has to come from somewhere. It comes from the particle's mechanical motion — in this case, its librational oscillation.

Each such scattering event removes one phonon — one quantum of librational vibration — from the rotor. Repeat the process millions of times per second, and the rotor's librational energy drops steadily. As the press release notes, at the floor, the particle is emitting and absorbing photons symmetrically, and the librational mode can no longer give up a phonon without being in a superposition. That balance — zero-point fluctuation against spontaneous emission — is the quantum ground state.

Coherent scattering has one property that distinguishes it from feedback cooling or cold damping: it is passive with respect to measurement noise. There is no detector whose shot noise gets fed back into the particle. The information flows the other way; the cavity carries entropy away from the mechanical mode. That matters for scaling to two dimensions. Two modes cooled by two independent feedback loops can develop cross-talk and heat each other. Two modes cooled by coherent scattering into the same cavity ride the same cooling channel, in different but compatible projections. The press release quotes Troyer on this point: "The beauty of our 2D cooling method is that it works across scales."

The Prior Art Ladder — Credit Where It Belongs

Breakthrough framing has a way of flattening prior results. The 2026 paper sits on top of a decade of incremental cooling achievements, and singling out the Vienna work without placing it in that ladder misrepresents the field.

The first rung was single-mode translational ground-state cooling, achieved in levitated silica by Delić and colleagues in Science in 2020. That paper demonstrated the center-of-mass quantum ground state for a trapped nanoparticle — a milestone that converted levitated optomechanics from a promising platform into a full quantum testbed. The second rung was two-mode translational ground-state cooling in 2023, showing that the technique could be extended to more than one degree of freedom. A third rung came in 2025 with the demonstration of a high-purity two-dimensional translational levitated oscillator in Nature Communications, clarifying what "2D ground state" even means quantitatively in a trap with realistic asymmetries.

What had not been done — and what Interesting Engineering credits the Vienna team with — was extending any of this from translational to orientational degrees of freedom in more than one dimension at the quantum limit. Single-axis librational cooling existed. Two-axis librational cooling at the ground state did not, until 2026.

Reframing the result this way is not nit-picking. It is how physicists themselves read the paper. The genuinely novel claim is narrow and specific — the full two-dimensional orientational ground state — and the claim is strengthened by placing it next to what was already known rather than hiding those antecedents behind a press-release first.

Why Anyone Would Want a Quantum-Limited Compass Needle

Three applications drive the field, and each one changes shape slightly once both librational axes are in the ground state at the same time.

The first is rotational matter-wave interferometry. A particle that can be prepared in a superposition of two orientations is, in principle, detectable through interference fringes in the angular variable — an angular analog of the double-slit experiment. For massive objects, such interference is the most direct test of whether quantum mechanics holds at scales approaching the classical limit. Interferometry requires a well-defined starting point. A 2D ground state is that starting point in the orientational sector; without it, the object's initial alignment carries thermal noise that blurs any subsequent superposition.

The second is quantum torque sensing. Torque — the rotational analog of force — is what acts on a compass needle near a magnet, or on a molecular rotor near a charged surface. At Heisenberg-limited alignment precision, even very weak torques produce detectable orientational shifts. The press coverage singles out Casimir torques, surface forces, and vacuum-friction effects — phenomena whose magnitudes are small enough that today's instrumentation barely sees them. A rotor pinned to 20 microradians of alignment uncertainty is, at least in principle, a torque sensor operating at the quantum limit. Whether it can be turned into a useful instrument depends on calibration, stability, and integration into the kind of vacuum apparatus that fits outside a dilution fridge — none of which the 2026 paper resolves.

The third, more speculative application is fundamental-physics tests. Rotational degrees of freedom couple to gravity differently from translational ones; a levitated rotor in a quantum-limited orientation could, in principle, probe gravitationally induced decoherence or deviations from standard quantum behavior in the high-mass regime. These proposals predate the Vienna result by years, and the paper does not claim to test them. But the result removes one of the remaining technical obstacles that kept such proposals on paper rather than in the lab.

The Macroscopic-Quantum Question, Honestly Stated

It is tempting to describe 100 million atoms behaving as one quantum oscillator as a "Schrödinger's cat" system. The research team does not say this, and neither should an honest write-up. A ground state is an incoherent population of the lowest quantum state; it is not a superposition of two macroscopically distinct states. The Vienna rotor at rest in its 2D librational ground state is quantum-limited in its orientation, but it is not in two orientations at once. The next step — preparing the rotor in a superposition of two different mean orientations and looking for angular interference — is what would warrant the cat terminology. None of the coverage claims this has been achieved.

What the 2026 paper does is remove the first, and historically the hardest, obstacle to that next step. Superpositions prepared on top of a thermal state are scrambled by the thermal fluctuations; superpositions prepared on top of a ground state are not. Quantum ground-state cooling is the gating condition for everything that would follow.

Scalability — and Its Limits

One of Troyer's more striking claims, quoted in the press release, is that the 2D cooling method "works across scales." Interesting Engineering and Mirage News both amplify this into a projection that objects roughly a hundred times lighter than the current rotor — the mass range of certain viruses, with tobacco mosaic virus as the commonly cited example — could be the next generation of targets. That projection should be treated qualitatively, not quantitatively. It rests on the generic scaling of coherent-scattering cooling with particle volume and polarizability, and says nothing about whether nonrigid bodies, polar molecules, or biological objects would couple to an optical trap in the same benign way that a rigid silica nano-dumbbell does.

Across-scales scalability is a hypothesis the 2026 paper enables, not one it proves. The honest reading is that the experimental platform is now unambiguously good enough for larger and smaller silica rotors, with a plausible pathway to heavier or lighter variants if the engineering and the physics hold. Each of those is a separate experiment. None of them would be trivial.

What This Does Not Tell Us — Yet

1. The exact phonon occupation numbers of the two librational modes are not given in the open press coverage. The Nature Physics abstract presumably states the mean occupation for each mode, but the Nature article page was not directly retrievable during research. Treating the 2026 result as "ground state" rests on the press description; finer-grained claims — "0.3 phonons," "0.4 phonons" — are not established by the public-facing sources.

2. The librational mode frequencies and their symmetry are not disclosed in press coverage. That matters because the spectral separation of the two modes controls how cleanly they can be cooled independently. Without those numbers, a reader cannot judge how close the two modes are in frequency, nor whether the 2D ground state is a near-degenerate or well-separated configuration.

3. Coherence times in the ground state — how long the rotor remains in its quantum-limited orientation before ambient gas, laser noise, or residual cavity heating drives it back up — are not specified. That number governs whether interferometry or torque sensing is feasible in practice or only in principle.

4. Scaling to non-silica or non-rigid objects is a projection, not a demonstration. Applications to viruses, polar molecules, or biological assemblies are conceivable under the same general cooling framework, but each introduces absorption, birefringence, or internal-mode degrees of freedom that the current silica rotor avoids by construction.

5. The 2026 result does not constitute a macroscopic quantum superposition. Ground-state cooling and superposition preparation are distinct experimental tasks. The former is a prerequisite for the latter, not a substitute.

Key Takeaways

• A team at the University of Vienna, TU Wien, and Ulm University reports in Nature Physics the first simultaneous quantum ground-state cooling of two librational modes of a levitated silica nano-dumbbell — a genuine advance over single-axis librational cooling and over two-mode translational cooling.
• The nano-dumbbell comprises two 150-nanometer silica spheres totalling roughly 100 million atoms, trapped in an optical cavity at light intensity of about 100 megawatts per square centimeter, and cooled by coherent scattering to an alignment precision of approximately 20 microradians in each librational axis.
• The 20-microradian floor is set by Heisenberg's uncertainty principle, not by the apparatus. Pushing the light harder would not shrink it further; it is the zero-point fluctuation of a quantum harmonic oscillator.
• The result sits on prior rungs: 2020 single-mode translational ground state, 2023 two-mode translational ground state, ETH Zürich single-axis librational cooling, and 2025 two-dimensional translational oscillator. Full 2D librational ground state is what was missing, and what 2026 supplies.
• Applications in rotational matter-wave interferometry, quantum-limited torque sensing, and tests of gravitationally induced decoherence are enabled — not demonstrated — by this result. The difference between "enabled" and "demonstrated" is the honest boundary around what the paper actually establishes.