Extended Rydberg Lifetimes in a Cryogenic Atom Array

We report on the realization of a $^{133}$Cs optical tweezer array in a cryogenic blackbody radiation (BBR) environment. By enclosing the array within a 4K radiation shield, we measure long Rydberg lifetimes, up to $406 (36),μ$s for the $55 P_{3/2}$ Rydberg state, a factor of 3.3(3) longer than the room-temperature value. We employ single-photon coupling for coherent manipulation of the ground-Rydberg qubit. We measure a small differential dynamic polarizability of the transition, beneficial for reducing dephasing due to light intensity fluctuations. Our results pave the path for advancing neutral-atom two-qubit gate fidelities as their error budgets become increasingly dominated by $T_1$ relaxation of the ground-Rydberg qubit.

💡 Research Summary

In this work the authors demonstrate a substantial improvement in Rydberg state lifetimes for a 133 Cs neutral‑atom quantum processor by operating the optical tweezer array inside a cryogenic environment. The experimental platform consists of a room‑temperature atom‑source chamber feeding a ultra‑high‑vacuum science chamber that is surrounded by 35 K and 4 K radiation shields. The inner 4 K shield is fitted with indium‑tin‑oxide (ITO) coated windows that transmit the required laser beams while strongly attenuating microwave radiation in the 10–300 GHz band, which is responsible for black‑body‑radiation (BBR)‑induced transitions between Rydberg levels. By this design the effective BBR temperature at the atom location is reduced to below 25 K, with an inferred value of 10 K (±10 K) for the measurements presented.

Atoms are loaded from a 2‑D MOT into a 3‑D MOT and subsequently transferred into a 7 × 7 tweezer array with 12 µm spacing. After polarization‑gradient cooling and optical pumping into the stretched ground state |g⟩ = |6 S₁/₂, F = 4, m_F = 4⟩, the qubit is formed with a single‑photon transition to an nP₃/₂ Rydberg state. The authors employ a 319 nm UV laser generated by sum‑frequency generation and second‑harmonic conversion, delivering up to 300 mW. This single‑photon scheme eliminates the intermediate‑state scattering that limits two‑photon excitation, thereby preserving the intrinsic T₁ of the Rydberg level.

Coherent control is demonstrated with Rabi oscillations at Ω = 2π × 1.35 MHz and a π‑pulse fidelity of 98–99 %. Ramsey interferometry yields a dephasing time T₂* = 6.2 µs, limited by residual Doppler broadening corresponding to an atomic temperature of ≈2 µK. The authors also characterize the differential light shift of the Rydberg transition as a function of Ω², finding a small coefficient κ = 29 kHz/MHz², indicating that even at higher Rabi frequencies the light shift remains well below the drive strength.

The central result is the measurement of the 55 P₃/₂ lifetime. Using a sequence where the tweezer light is turned off, a π‑pulse excites the atom to the Rydberg state, a variable delay is introduced, and a second π‑pulse returns population to the ground state, the authors obtain an exponential decay with a 1/e lifetime of 406 µs (±36 µs). This is a factor of 3.3(3) longer than the room‑temperature value of 122 µs and approaches the calculated spontaneous‑emission limit of 429 µs. Measurements for n = 46, 50, 55 all show a similar three‑fold increase, consistent with an effective BBR temperature ≤ 25 K. The authors note that further cooling would yield diminishing returns because the lifetime becomes dominated by spontaneous decay rather than BBR‑induced transitions.

From a quantum‑gate perspective, the error contribution from T₁ scales as 1/(Ω T₁). With the present parameters (Ω ≈ 2π × 1.35 MHz, T₁ ≈ 400 µs) the T₁‑limited gate error is estimated at 6.4 × 10⁻⁴ for a time‑optimal gate, already below the typical error floor of room‑temperature experiments (1–3 × 10⁻³). If the beam waist is reduced to increase Ω to ≈2π × 5 MHz, the error could be suppressed further to ≈1.6 × 10⁻⁴. In surface‑code error‑correction simulations, such a reduction in the physical two‑qubit error rate for a code distance d = 7 translates into roughly two orders of magnitude lower logical error rates, dramatically easing the overhead for fault‑tolerant computation.

Overall, the paper establishes that cryogenic suppression of BBR, combined with single‑photon Rydberg excitation and careful control of light shifts, can push the Rydberg‑ground‑state T₁ to its fundamental limit. This advancement directly benefits neutral‑atom quantum processors by enabling higher‑fidelity two‑qubit gates and reducing the physical error budget required for scalable quantum error correction.