Innsbruck physicists probed Bethe strings, bound states predicted in 1931, in a one-dimensional cesium gas. How the experiment worked, with the binding energies checked by hand.
Bethe strings are bound states that exist only in one dimension, held together not by chemical bonds but by the quantum mechanics of interacting particles confined to a line. Hans Bethe’s 1931 solution method predicted them. Physicists at the University of Innsbruck, in a team led by Hanns-Christoph Nägerl, have now probed Bethe strings of six or more particles in an ultracold gas of cesium atoms, reported in Nature Communications this year.
The result has been framed as a century-old prediction coming to life. The more precise story is that Bethe strings had been detected before in solid-state magnetic systems, and this experiment brings them into a clean, tunable atomic gas where the interaction can be set on demand. The sections below explain what a string is, how the gas was built, how the team detected strings without photographing them, and then check the preprint’s binding energies by hand. Figures come from the arXiv version of the paper (arXiv:2505.10550, v2).
What a Bethe String Is
Bethe’s 1931 technique, now called the Bethe ansatz, solves certain quantum many-body problems exactly. It works in one dimension because particles in a line cannot get around each other. When they collide, they can only trade places or bounce, and the set of momenta describing the whole system is preserved. Physicists call such a system integrable.
The model relevant here, named for Elliott Lieb and Werner Liniger, describes bosons in a line that interact only when they touch. In the Bethe-ansatz solution each particle carries a label called a rapidity, a generalized momentum. With repulsive interactions the rapidities are ordinary real numbers. Make the interaction attractive and they become complex: they bunch into clusters that share a real part, with imaginary parts spaced by the interaction strength. Each cluster is a string, and a string of n rapidities is a bound state of n particles.
The preprint gives the size and energy of an isolated string. Its spatial extent is roughly 2 divided by the product of the interaction strength and (n − 1). Its binding energy is ħ²c̃² multiplied by n(n² − 1) and divided by 24 times the atomic mass, where c̃ is the one-dimensional coupling. For N attractive particles, the true ground state is a single giant string holding all N. Smaller strings have to be created out of equilibrium, which shapes the entire experimental design.
Building a One-Dimensional Gas From Cesium
Real atoms live in three dimensions, so the experiment has to freeze two of them. The Innsbruck team starts with a Bose-Einstein condensate of cesium atoms, cooled to between 10 and 35 nanokelvin. A two-dimensional optical lattice then slices the cloud into about 4,000 independent tubes. Atoms are squeezed so tightly across each tube that they can move freely only along its length.
| Parameter | Value in the preprint |
|---|---|
| Atoms in the initial condensate | About 2 to 6 × 104 |
| Condensate temperature | 10 to 35 nK |
| Number of tubes | About 4,000 |
| Atoms per tube | About 7 to 30 |
| Lattice spacing | 532.2 nm |
| Transverse trap frequency | About 10.5 kHz (oscillator length near 83 nm) |
| Longitudinal trap frequency | About 29.3 Hz (oscillator length near 1.6 μm) |
| Interaction control | Broad Feshbach resonance, tuned with a magnetic field near 20.8 G at loading |
A Feshbach resonance is the control knob. Adjusting the magnetic field changes the three-dimensional scattering length of cesium atoms, and with it the effective one-dimensional coupling c̃. The same atoms can therefore be made to repel one another strongly, ignore one another, or attract.
Sweeping the Interaction Through Zero
The team begins on the repulsive side, with a coupling of about 3.4 per micrometre. That is moderately interacting and close to the Tonks-Girardeau regime, in which strongly repelling bosons behave like non-interacting fermions. They then ramp the field so the coupling passes through zero, at roughly −0.038 per micrometre every millisecond, and into attraction.
That sweep is the out-of-equilibrium step. The gas does not have time to condense into one giant string, so it ends up populated by strings of several sizes. Atoms in the tubes are also subject to loss through short-range molecular processes, which the idealized model leaves out. The authors say so, and it limits how far the comparison with theory can be pushed.
Detection Without a Camera: One Dimension Versus Three
The team did not image strings one by one. They inferred them from energy, using a two-step expansion. First the cloud expands along the tube for a time, with the transverse confinement still on. Then the confinement is switched off and the three-dimensional scattering length is set near zero, so the atoms fly freely in all three dimensions. From how the cloud’s width grows, the team extracts a release energy for each stage: E1D and E3D.
The logic is clean. If the particles are unbound, switching to three dimensions changes nothing along the tube, so the two energies agree. If bound strings exist, they survive the one-dimensional expansion with their binding energy locked up inside them. Once the interaction is switched off, the strings come apart and that stored energy appears as motion. In that case E3D exceeds E1D.
| Regime | E1D | E3D |
|---|---|---|
| Repulsive, coupling approaching zero | Falls from about h × 100 Hz to about h × 60 Hz | Same as E1D |
| Attractive, coupling −1.0 per μm | Plateaus near h × 35 Hz | Rises to about h × 110 Hz |
The repulsive data show no gap, as expected. The attractive data show a large one. The momentum distribution tells the same story: its full width at half maximum starts near 6.0 ħ per micrometre and narrows to about 3.0 as the coupling approaches the non-interacting point. Both sets of measurements agree well with simulations based on generalized hydrodynamics, a theory built for integrable systems.
Checking the Binding Energies by Hand
The preprint’s formula is easy to test. Take the cesium mass of about 132.9 atomic mass units and a coupling of −1 per micrometre. Binding energy then grows roughly with the cube of the string size. The values below are my own arithmetic from the formula; the paper quotes the n = 4 row.
| String size n | Bond length (μm) | Binding energy (h × Hz) | Binding energy (nK) |
|---|---|---|---|
| 2 | 2.0 | 19 | 0.9 |
| 3 | 1.0 | 76 | 3.7 |
| 4 | 0.67 | 190 | 9.1 |
| 5 | 0.5 | 380 | 18 |
| 6 | 0.4 | 665 | 32 |
The n = 4 row reproduces the preprint’s quoted figures: a bond length of about 0.67 micrometres and a binding energy near h × 190 Hz, or about 9 nanokelvin in temperature units. That agreement confirms the formula and the units.
Two observations follow. First, the larger strings have binding energies comparable to the gas temperature, which the paper puts between about 3 and 35 nanokelvin in one dimension. Strings survive anyway because, in an ideal integrable system, ordinary collisions cannot break them apart. The real gas is only approximately integrable, which is another reason to read agreement with theory with some care. Second, a rough check on the energy gap: a release-energy difference of about 75 Hz per particle matches the per-particle binding of an n = 5 string, which is 76 Hz. The paper’s own analysis uses the average binding energy and puts the largest string size at about six. The two estimates are in the same neighbourhood, which is all a back-of-envelope comparison can claim.
What This Does and Does Not Show
- It is not the first sighting. The press summary itself notes that Bethe strings had been detected before in solid-state magnetic systems. What is new is the platform: a clean atomic gas with an adjustable interaction.
- The detection is indirect. The evidence is an energy gap and a momentum width, backed by theory, not a picture of individual strings.
- Strings need room. They are only well defined when the bond length is much larger than the transverse confinement scale of about 83 nanometres, which the bond lengths above satisfy.
- The model is an approximation. Short-range molecular loss and the weak departure from perfect one-dimensionality are not in the Lieb-Liniger description.
Why Exactly Solvable Models Matter
Exact solutions are rare in many-body physics, and rarer still in systems anyone can build. A solvable model realized in a laboratory lets theorists test their approximations against a case where the right answer is known. Generalized hydrodynamics, which agrees with the Innsbruck data, was developed precisely to describe how integrable systems evolve, and this kind of experiment is its proving ground.
It also fits a wider pattern in quantum physics, where exotic excitations are being engineered in controlled platforms instead of waiting to be found in nature. Our claim check on Quantinuum’s anyon experiment looks at another example.
Sources and Verification Notes
- arXiv:2505.10550 v2: Probing Bethe strings in an attractive one-dimensional Bose gas (preprint)
- ScienceDaily: A quantum prediction from 1931 just came to life (October 9, 2026; University of Innsbruck)
Could not verify: the published Nature Communications paper (volume 17, 2026, DOI 10.1038/s41467-026-76018-0 as listed in the press summary), so numbers from the preprint may differ in the final version. The binding-energy table is my own calculation from the preprint’s formula, and the n = 5 comparison is a rough check of mine, not an analysis from the paper.
