Stuttgart Physicists Measure an 11.5-Millisecond Lifetime for Circular Rydberg Atoms at Room Temperature. The Group’s Best Circular-Qubit Coherence Time Is Still 278 Microseconds.

September 18, 2026 – Physicists at the University of Stuttgart kept individual strontium atoms in circular Rydberg states for more than 10 milliseconds at room temperature, according to a paper published in Nature Communications on 15 September 2026. The longest lifetime they measured, 11.5 milliseconds for the state with principal quantum number n = 101, is about 21 times what the same state would last in free space at room temperature.

The team, led by group leader Florian Meinert at the university’s 5th Institute of Physics, held each atom in an optical tweezer between two glass plates coated with indium tin oxide and spaced 10.5 millimeters apart. The coating reflects microwaves but lets laser light through. The plates suppress the thermal microwave radiation that otherwise knocks a circular Rydberg atom out of its state within a fraction of a millisecond. The atoms themselves were laser-cooled, and the room-temperature condition refers to the apparatus around them and the thermal radiation its surfaces emit.

The authors reported that the suppression produced lifetimes that, in free space, would require surroundings cooled to 14 kelvin. They raised the principal quantum number of the circular states from 79 to as high as 103 using up to 12 microwave pulses, giving an electron orbit about 1.1 micrometers across, and measured a trapping lifetime of 133 milliseconds for the Rydberg atoms in the tweezers.

In a press release issued on 16 September, Meinert said the group had set three records at once: the longest lifetime measured for individual Rydberg atoms, the largest controlled circular Rydberg atoms and the longest storage time for such atoms in optical tweezers. Tilman Pfau, head of the institute, said the team had made the atoms about 20 times more stable. Meinert added that the results were obtained without the liquid helium cooling that earlier experiments of this kind had required.

The authors wrote that the lifetimes, combined with control of interactions between circular Rydberg atoms that a Paris group demonstrated in 2025, could support quantum simulators running for orders of magnitude longer than current Rydberg-atom simulators. They also wrote that the coherence time of qubits encoded in circular states, which technical noise currently limits to the 100-microsecond scale in their setup, must first be raised into the millisecond range.

The paper’s authors are Einius Pultinevicius, Aaron Götzelmann, Fabian Thielemann, Christian Hölzl and Meinert. A preprint appeared on arXiv on 31 October 2025, and the ion-arrival data are published on Zenodo. Germany’s Federal Ministry of Research, Technology and Space funded the work through the CiRQus and QRydDemo projects, alongside the EU’s Horizon Europe project EuRyQa.

My Analysis

An 11.5 ms circular Rydberg state is a real atomic-physics record whose most direct application is quantum simulation. It changes nothing yet for the neutral-atom machines on the market, which work with ordinary laser-excited Rydberg states: gate-based designs excite them only during entangling gates, and analog simulators such as QuEra’s Aquila encode each qubit in the ground state and one of those Rydberg states. Meinert’s group documents the record with care. The team even tested its fitting method against a full simulation, in which the method returned slightly low lifetimes.

The university’s press office goes further than the paper. Its release says the stability of the atoms makes it possible to “preserve quantum information for longer,” and says researchers expect this to let quantum computers calculate for longer with fewer errors. Both claims depend on coherence, which the paper doesn’t measure. The best coherence Stuttgart has published for a circular-state qubit is 278 µs, measured in 2024 on the n = 77 and 79 levels, about 40 times shorter than the new lifetime record.

Circular Rydberg Atoms and Room-Temperature Blackbody Radiation

A Rydberg atom has one electron lifted into an orbit with a very high principal quantum number n, far from the nucleus. Neutral-atom machines use such states for their two-qubit gates. Two Rydberg atoms a few micrometers apart interact so strongly that exciting one blocks the excitation of the other, the blockade effect I described in my overview of the neutral-atom modality.

A circular Rydberg atom is the extreme case. Its electron has the largest angular momentum its energy level allows, which confines it to a thin ring-shaped cloud around the nucleus, the quantum counterpart of a circular orbit. States of the same level with less angular momentum correspond to elliptical orbits. Physicists call them elliptical states. Randall Hulet and Daniel Kleppner made the first circular states at MIT in 1983. At zero temperature, an electron in that state can decay in only one way, by emitting a single microwave photon that drops it to the next circular level down. With only that one weak channel open, a circular state decays slowly. The Laboratoire Kastler Brossel group in Paris, which has worked with circular atoms since Serge Haroche’s Nobel-winning cavity experiments, puts the zero-temperature lifetime of the n = 60 circular state at about 71 ms, against about 500 µs for the ordinary 60P state.

Room-temperature radiation removes most of that advantage. Every surface at 300 K fills the space around the atom with thermal microwave photons, which drive the electron up and down the ladder of circular levels. By my calculation, a single mode at 6.4 GHz, near the transitions out of n = 101, contains on average $$\bar{n} = 1/(e^{hf/k_BT}-1) \approx 976$$ photons at 300 K. The Paris group calculates that room-temperature radiation cuts the n = 60 circular lifetime to about 170 µs, barely better than the 140 µs of a 60P state. For Stuttgart’s n = 101 state, the free-space figure at 300 K is 545 µs. Most circular-atom experiments have therefore run in cryostats.

How Two Coated Glass Plates Suppress Blackbody Decay

Kleppner proposed the remedy in a 1981 paper, and in a companion experiment that year his group, with A. G. Vaidyanathan and W. P. Spencer, observed plates suppressing the absorption of blackbody radiation. Two ideal parallel conducting plates cannot support microwaves polarized parallel to them when the wavelength exceeds twice their spacing, so an atom between them can neither emit nor absorb photons of that kind. In 1985 Hulet, E. S. Hilfer and Kleppner showed the same effect for spontaneous emission, using cesium atoms in circular states at n = 22 flying between two plates, and measured the natural lifetime rising by a factor of at least 20.

Meinert proposed Stuttgart’s version in 2020. It replaces metal with glass plates coated in indium tin oxide, a transparent conductor that reflects about 96% of the microwave power while letting the trapping and imaging lasers reach the atom. The plates are 10.5 mm apart, which sets the cut-off near 14 GHz. Above about n = 78, the transitions from a circular state to its neighbors fall below that frequency, and at n = 101 they are near 6.4 GHz, a wavelength of about 4.7 cm. A circular atom’s decay photons are circularly polarized in the plane of its orbit, and the group uses electric and magnetic fields to keep that plane parallel to the plates. In that orientation, the plates block the channel that dominates at room temperature.

The lifetime curve across n is not smooth. The plates do not suppress photons polarized perpendicular to them, and the four ring electrodes around the atom turn the gap into a cylindrical cavity, 12 mm in radius, whose lowest resonances strengthen exactly those transitions. The lifetime dips near n = 90, where the group calculates that about 80% of decays from n = 89 end in non-circular, elliptical states. Above about n = 95, the last weak channels, to elliptical states two levels up, also fall below the cut-off, and the lifetime climbs past 10 ms. A finite-element model of the full electrode stack reproduces the dip, which an idealized infinite-plate model misses.

Against free space at 300 K, the plates extend the n = 101 lifetime by a factor of about 21. Hulet, Hilfer and Kleppner reported a factor of at least 20 in 1985, for spontaneous emission from a beam of atoms at n = 22. Stuttgart’s factor applies to the thermal transfers that dominate at room temperature, measured on one atom in a laser trap at n = 101. Stuttgart’s press office describes the method as a concept from the 1980s without naming its author. Kleppner, who never received a Nobel Prize but with Dave Pritchard mentored four physicists who did, died in June 2025 at 92.

Checking the Three Records Against the Paper

An 11.5 ms Lifetime at n = 101

I find the lifetime record sound. The group extracts each lifetime by fitting a rate model to the way population spreads from the starting level into its neighbors. Because the plates divert much of the decay into elliptical states, the team checked that model against a full simulation that includes 229 extra elliptical states. In that simulation, the fitted lifetimes came out within 16% of the true values and consistently low.

The authors describe these as the longest-lasting Rydberg states produced in a laboratory, three times longer than the best cryogenic result. That benchmark is the Paris group’s 3.7 ms for rubidium at n = 52, measured in a cryostat in 2020. Because circular-state lifetimes in free space grow with n at any temperature, the n = 101 state in a cryostat would last far longer than 3.7 ms. Cohen and Thompson’s 2021 computing proposal, discussed below, assumes lifetimes above a second in cryogenic microwave cavities. Stuttgart reached its record by pairing the plates with a very high n at room temperature.

In the release, Meinert also says the results needed none of the costly liquid helium cooling earlier work required. The room-temperature method predates this paper. The Paris group reported more than 1 ms at n = 60 in 2023, using a capacitor with one fully transparent electrode, and Stuttgart’s 2024 paper reached 2.55 ms at n = 79 with the same kind of plates. In that paper the group predicted lifetimes above 10 ms once n passed 96. Meinert’s team has now confirmed that prediction.

Circular Orbits up to n = 103

The group starts each atom at n = 79 and climbs two levels at a time with two-photon microwave transitions, sweeping an electric field through each resonance so the population transfers adiabatically. At n = 103 the orbit is about 1.1 µm across. The authors write that circular states much above n = 50 had remained elusive before their own n = 79 work in 2024. A University College London group had already prepared helium circular states at n = 70 in 2018, in an atomic beam rather than a trap. They also verified, with a separate spectroscopic test demonstrated at n = 89, that the sweeps end in the circular state and not in an elliptical neighbor. Rydberg atoms in general have been made far larger. F. B. Dunning’s group at Rice University worked with potassium Rydberg atoms at n ≈ 305, about 10 µm across, in 2012. Stuttgart’s record is for circular states prepared through successive microwave transfers on a trapped atom.

The last step, from n = 101 to 103, was only probed spectroscopically and not optimized, and the lifetime series ends at n = 101. The other transfers average 94% per step. By my arithmetic, treating the eleven steps from n = 79 as independent and equally efficient, $$0.94^{11} \approx 0.51$$, so roughly half the population reaches |101C⟩. The group reported in 2024 that about 70% of excited atoms reach |79C⟩ in the first place, which leaves, on this rough estimate, about 35% in the target state. The climb also takes about 400 µs, which the authors note is already longer than the free-space lifetime of the lowest circular levels in the ladder. The authors expect better voltage ramps and more microwave power to raise the transfer efficiency.

A 133 ms Trap Lifetime

Laser light usually pushes a Rydberg atom out of an ordinary tweezer, because the loosely bound electron experiences the beam as a repulsive potential. Rubidium and the other alkali atoms therefore usually need hollow beams. The Paris group confined circular rubidium atoms in a hollow laser beam for up to 10 ms in 2020. Strontium’s second valence electron stays in the Sr⁺ core, and the light pulls on that core with enough force to trap the whole Rydberg atom in a standard Gaussian tweezer. Starting from |97C⟩, Stuttgart measured a 1/e trapping lifetime of 133 ms. With the tweezer switched off, the signal disappeared within a millisecond.

The 133 ms figure counts the atom in any Rydberg state the detector can still see, circular or not. The authors call it a lower bound, because levels above n = 104 and states of low angular momentum fall outside the detection window. The 133 ms is how long a Rydberg atom stays trapped, about ten times the lifetime of the |97C⟩ state it started in.

Lifetime Versus Qubit Coherence

A qubit needs more than a long lifetime. The lifetime, T1 in the usual notation, is how long the atom stays in its state. A qubit also needs a long T2, the time over which a superposition of two states keeps a stable phase, and any noise that shifts the energy difference between the two states erodes that phase long before the atom decays. I covered the mechanisms in The Many Faces of Decoherence.

Stuttgart measured T2 for a circular-state qubit in 2024, using the levels |77C⟩ and |79C⟩. A Ramsey measurement gave 43 µs. A spin-echo sequence, which cancels slow drifts, gave 278 µs, defined as the time for the echo contrast to fall to half. Both figures come from atoms released from the tweezer. Trapped atoms dephased slightly faster. The group traced the limit to electric-field gradients across its tweezer array and to magnetic-field noise. Circular states are very sensitive to electric fields: at the 2 V/cm field used in that work, the qubit frequency shifted by 8.8 kHz for every millivolt per centimeter of field change.

The 2026 paper contains no coherence measurement at any n. The authors write that technical noise still limits T2 to the 100 µs scale in their setup, and they list the fixes they have in mind: active field feedback, cancellation of gradients, laser cooling on the ion core and dynamical decoupling. They expect coherence close to the circular-state lifetime to be within reach. They haven’t yet published a measurement showing it.

Stuttgart’s press office gets its numbers right, from the 1.1 µm orbit to the 133 ms trap lifetime, and then says the atoms’ stability could keep quantum information intact for longer, a claim that depends on T2. If T2 at n = 101 matched the 2024 figure, it would be about 40 times shorter than the lifetime there.

The published paper is more careful than its own preprint. Both arXiv versions presented the lifetimes as the basis for quantum simulation essentially free of dissipation, on timescales two orders of magnitude longer than today’s. The Nature Communications version makes that claim conditional on fixing the dephasing. It also adds the paragraph on T2 and the 94% transfer figure. The journal publishes the peer review file with the paper, so readers can check whether the referees asked for these changes.

Circular States and Today’s Neutral-Atom Computers

Neutral-atom machines use Rydberg states in two ways. Gate-based designs, Infleqtion’s among them, store qubits in ground or metastable levels with long lifetimes and excite atoms to a Rydberg state only during entangling gates. Infleqtion’s prototype reported a best-estimate coherence time of 2.8 seconds for its ground-state qubits under dynamical decoupling, with a lower bound of 128 ms at 97% confidence. An atom that decays out of the Rydberg state mid-gate produces an error. Research systems have reported two-qubit fidelities between 99.5% and 99.85%, as I noted in my coverage of the July strategic plan for neutral-atom computing, and Junlan Jin and colleagues at Princeton argue that gate error budgets are increasingly dominated by Rydberg-state decay.

Analog simulators such as QuEra’s Aquila work differently. Each qubit is the ground state paired with an ordinary Rydberg state, whose lifetime limits how long a simulation can run. The Stuttgart authors aim their result at this kind of simulation.

Those gates use S, P or D Rydberg states, which also decay by emitting optical photons toward low-lying levels. A plate capacitor suppresses only modes with wavelengths longer than twice its gap, so it can’t touch that optical decay. Pasqal’s researchers put the gain from removing blackbody radiation entirely at a factor of two to three for such states. For the n = 60 circular state, the Paris figures above imply a factor of about 400.

The route aimed at today’s gates is cryogenic. Jin and colleagues, working in Jeff Thompson’s and Waseem Bakr’s groups, enclosed a cesium tweezer array in a 4 K radiation shield and measured 406 µs for the 55P3/2 state, 3.3 times its room-temperature lifetime. Their paper has since been published in PRX Quantum. Pasqal has built a 4 K platform, since published in PRX Quantum, that assembles defect-free arrays of up to 1,024 atoms in just over 10% of attempts. Trap lifetimes for its ground-state atoms reach about 80 minutes, and the team plans Rydberg experiments on it next. Its two-stage pulse-tube cooler reaches 4 K, not the millikelvin temperatures of the dilution refrigerators that superconducting processors need. It’s still cryogenics, and neutral-atom vendors have marketed room-temperature operation as one of the modality’s advantages.

Using circular states for computing would take a different encoding, preparation and control scheme. Sam Cohen and Jeff Thompson at Princeton proposed one in 2021, with qubits encoded across several circular levels and two-qubit gate errors around 10⁻⁵ projected for arrays of hundreds of atoms. Their projection assumes lifetimes above a second in cryogenic microwave cavities, nearly a hundred times Stuttgart’s room-temperature record. Room-temperature operation is useful to simulation experiments today. Cohen and Thompson planned their computing architecture around a cryostat from the start.

What Circular-State Qubits Still Need

To turn this lifetime result into a many-atom platform near n = 100, the Stuttgart group still needs four results:

• Millisecond coherence at high n. The group names this as its next step. The 278 µs from 2024 is the number to beat, and the 2026 paper reports no coherence measurement near n = 100.
• Interactions between two atoms at high n. The Paris group measured the dipolar interaction between two trapped circular atoms at n = 51 and 52 in 2025. Stuttgart calculates that at its n the van der Waals blockade is about a thousand times stronger than at n ≈ 50 and the dipole-exchange coupling more than ten times larger. Stuttgart has measured neither between two atoms.
• Fast, near-deterministic preparation. The current climb takes about 400 µs and, by my rough estimate above, brings about 35% of excited atoms to |101C⟩. For the first step, circularization, the group pointed in 2024 to optimal-control methods demonstrated in Paris at lower n as a route to about 100 ns at 99% fidelity.
• Arrays at high n. The Paris group trapped arrays of individual circular rubidium atoms in optical tweezers in 2023. Stuttgart’s 2026 lifetime measurements prepare one atom at a time.

In a preprint posted on 7 August 2026 and not yet peer reviewed, the Stuttgart team worked on a different gap: control and readout inside one atom. Strontium’s second electron stays in the Sr⁺ core while the first circles far outside, and the team encoded a second qubit on the core’s narrow 674 nm transition, which clocks and trapped-ion qubits based on Sr⁺ also use. With the tweezers switched off, that optical qubit kept its phase for 400 µs under spin echo and 550 µs under an XY8 decoupling sequence. The two electrons couple through an electrostatic quadrupole interaction that the team could tune to zero at the magic angle of about 54.7°, and the team ran a Mølmer–Sørensen-like sequence that, by the authors’ account, takes the pair transiently through an entangled Bell state.

The team encoded the circular qubit in that work in n = 79 and 81, and the n = 79 level has a lifetime of about 2 ms. Both electrons belong to the same atom, and neither Stuttgart paper entangles two separate atoms. Circular atoms have been entangled with each other before: in 1997 Serge Haroche’s group paired two of them through a microwave cavity.

Meinert proposed the transparent plates in 2020. The group reached n = 79, with 2.55 ms lifetimes and a 278 µs qubit, in 2024. It posted the n = 101 result in October 2025 and put a second qubit inside the atom in August 2026. The next measurements I will look for are a spin-echo T2 near n = 100 and a controlled interaction between two atoms at that n. A T2 near a millisecond would make room-temperature circular-state simulators a serious proposition. A T2 that stays near 300 µs would keep coherent simulations to a few hundred microseconds, whatever the circular-state lifetime.