A space-based quantum gas laboratory at picokelvin energy scales

The study aims to demonstrate precise quantum-state engineering of ultracold atomic gases in microgravity to enable next-generation space-borne quantum technologies. While ultracold gases routinely reach sub-nK temperatures on Earth, their ultimate sensing potential requires long free-evolution times with well-controlled initial conditions. Space platforms provide extended free fall and symmetric, low-distortion trapping, allowing novel manipulations (e.g., shell topologies and uniform Bose gases) and long interrogation times. However, high-impact applications in quantum sensing, relativistic geodesy, Earth observation, and tests of fundamental physics demand exquisite control of three key parameters: the atomic ensemble’s position, center-of-mass release velocity, and expansion energy. This work uses NASA’s Cold Atom Lab (CAL) on the ISS to realize fast, accurate shuttling of Bose-Einstein condensates (BECs), tunable release velocities, and matter-wave lensing to achieve picokelvin-scale expansion energies, thereby meeting stringent requirements for space-based precision measurements.

Prior breakthroughs include creation of BECs in microgravity and in orbit, enabling long-duration free fall. Symmetric, undistorted traps in space facilitate exotic states such as shells and uniform gases. In atom interferometry, uncertainties in initial position and velocity couple to gravity gradients and Coriolis effects, limiting precision in equivalence-principle tests. Earlier approaches to sub-nK energies relied on magnetic levitation or adiabatic decompression. Delta-kick collimation (DKC) was proposed and experimentally validated as an effective atomic lensing method to reduce expansion energy, with ground-based matter-wave lensing reaching picokelvin scales. Shortcut-to-adiabaticity (STA) techniques have enabled rapid state manipulations of trapped gases and ions, though gravity complicates multi-parameter control on Earth. These advances motivate implementing ab initio STA shuttling and DKC in space to surpass Earth-bound performance.

Platform and trapping: Experiments were conducted with NASA’s Cold Atom Lab (CAL) aboard the ISS. 87Rb BECs prepared in a facility trap were transferred into a simplified two-current atom-chip configuration: current I_chip through a Z-shaped chip wire (x–y plane) and current I_coil through a Helmholtz bias-coil pair (y-axis). This produces a magnetic quadrupole field closed by the chip-wire bends, yielding an approximately harmonic trap. The initial BEC (a few thousand atoms) was trapped 267 µm from the chip with angular frequencies (ωx, ωy, ωz) = 2π·(29.3, 922, 926) Hz and exhibited z-oscillation amplitude 0.22 ± 0.05 µm (max in-trap velocity 1.3 ± 0.3 mm/s).

Shuttling via STA: Fast transport exploited shortcut-to-adiabaticity protocols derived ab initio by reverse-engineering classical trajectories with boundary conditions ensuring the cloud starts and ends at rest at the trap minimum. Microgravity simplified control to a single parameter (I_coil) determining the trap minimum position. Two final traps were implemented:

• Trap A: transport 0.42 mm in 100 ms by decreasing bias-coil current by factor 3; final (ωx, ωy, ωz) = 2π·(25.2, 109, 110) Hz.
• Trap B: transport 0.93 mm in 150 ms by decreasing the current by factor 6; final (ωx, ωy, ωz) = 2π·(14.4, 35.1, 26.9) Hz. Ramps were discretized into 80 current steps due to hardware limits. Transport quality was assessed by (i) measuring residual in-trap sloshing after varying initial hold time t_hold,i and (ii) scanning time of flight to extract release velocity.

Release-velocity control: The release velocity as a function of t_hold,i was measured after switching off the chip trap. Systematic momentum kicks were characterized, attributed to differential switch-off dynamics (coil inductance vs chip).

Delta-kick collimation (DKC): After optimizing kinematics, DKC was applied using trap A to lens the released BEC. Following free expansion, the trap was re-enabled with frequencies rescaled by 1/4 for a short lens time to reduce expansion rates. Owing to trap symmetry, y and z dimensions were collimated simultaneously; x remained weakly confined but benefited from collective-mode engineering. The lens was typically applied 20 ms after release with a 1.8 ms duration in representative data.

Calibration and modeling: A 3D Biot–Savart model of chip wires and coils computed fields, trap positions, and frequencies versus I_chip and I_coil; parameters were calibrated by measuring sloshing frequencies/amplitudes at various positions. Transport and release dynamics were modeled via Newton’s equation with time-dependent ωz(t) and Z_trap(t) from the calibrated chip model. Mean-field BEC dynamics (Castin–Dum/Kagan–Surkov–Shlyapnikov scaling approaches) described expansion and DKC. Robustness studies extended STA theory to include cubic anharmonicities and ramp discretization.

Detection and fitting: Absorption imaging along y provided 2D x–z density. Thomas–Fermi fits (for TOF >80 ms) or bimodal fits (Gaussian thermal + Thomas–Fermi BEC) extracted radii and center-of-mass motion. Science runs used 2000–4000 condensed atoms (BEC fraction 10–25%); calibration runs had up to ~12,000 condensed atoms and <10% BEC fraction.

Operational constraints: Microgravity operations were remote with limited sequence budget; atom numbers degraded over time (≤4000 BEC atoms), residual magnetic-field gradients caused accelerations in mF=2, and free-expansion times were limited (~350–400 ms).

• Fast, precise shuttling: Transported a BEC over 0.42 mm (100 ms, trap A) and 0.93 mm (150 ms, trap B). Achieved residual in-trap oscillation amplitudes of 0.068 ± 0.072 µm (A) and 0.40 ± 0.15 µm (B). Oscillation energy reduced by factors ~738 (A) and ~359 (B) relative to the initial state. Ramp B spanned only ~4 motional cycles of the final trap.
• Accurate positioning: Transport moved the ensemble roughly 1500 times its size with residual oscillations ≈5% of its spatial extent, enabling sub-micron positioning suitable for precision sensing and quantum information applications.
• Tunable release velocity with metrological-level uncertainty: By scanning t_hold,i, center-of-mass release velocity was tuned in agreement with theory. Systematic kick at release was ~1 mm/s (trap A) and tens of µm/s (trap B) due to inductance-limited coil switch-off. Minimal measured release velocity was 35 ± 117 µm/s (trap B at t_hold ≈ 2.8 ms). Overall velocity uncertainty for ramp B was 233 µm/s (165 µm/s with theory-weighted analysis), comparable to state-of-the-art atom interferometers.
• Picokelvin expansion energies via DKC: With a single 1.8 ms lens applied 20 ms after release (trap A rescaled by 1/4), the z-direction expansion energy was reduced from 3.6 ± 0.2 nK to 52 ± 10 pK (~70× reduction). Due to trap symmetry, similar performance is inferred for y; along x, collective-mode engineering yielded ~200 ± 27 pK. The total 3D kinetic energy is ~100 pK.
• Stability and reproducibility: Optimal timing (final hold ~24 ms) minimized release velocity; the BEC displacement after 20 ms free fall was 0.2 ± 5.9 µm. Results matched ab initio models across months of operation despite changing ISS conditions.
• Space-enabled simplifications: Microgravity allowed single-parameter control of transport (I_coil), enabling ab initio STA ramps without complex multi-dimensional controls required on Earth.

The demonstrated control of position, release velocity, and expansion rate addresses key systematic limitations in precision atom interferometry and other quantum-sensing modalities. Sub-micron positioning mitigates gravity-gradient and Coriolis-induced biases; picokelvin expansion energies support seconds-long free drift with high atom-light interaction efficiency and minimized wavefront-related systematics. The measured release-velocity uncertainties are commensurate with leading metrology experiments, showing that compact, multi-user platforms such as CAL can provide mission-ready sources. Collectively, these achievements satisfy stringent initial-state requirements for ambitious space tests, including dual-species universality-of-free-fall experiments and other fundamental-physics missions. The work establishes a robust, tunable, and stable space-based quantum gas source, advancing the readiness level of quantum sensing, communication, and information-processing in orbit.

This work realizes state-of-the-art quantum-state engineering of BECs in orbit using NASA’s CAL. The team achieved rapid, accurate shuttling with sub-micrometer residual motion, tunable and well-characterized release velocities with ~10^2 µm/s uncertainties, and delta-kick collimation to ≈50 pK per axis (~100 pK 3D). These capabilities meet or exceed the most demanding initial-state specifications for space-based precision measurements and open the door to long-baseline, long-duration quantum sensors in orbit. Future directions include next-generation CAL/BECCAL payloads with higher atom numbers (~10^7), more flexible current controllers enabling smoother ramps (hundreds of steps), and operation in magnetically insensitive Zeeman states to suppress residual forces. Such upgrades could push positioning accuracy to the nanometer level and support multi-second free evolution, advancing space-based quantum technologies for fundamental physics, geodesy, gravitational-wave detection, and secure quantum communication.

• Atom number limitations: Degradation of the alkali dispenser reduced BEC sizes to ≤4000 atoms, constraining signal-to-noise and limiting free expansion times to ~350–400 ms.
• Residual magnetic-field gradients: Even with coils and chip off, gradients accelerated atoms in mF=2, moving clouds out of the imaging field; magnetic curvatures were negligible but gradients affected drift and observation time.
• Spin-state transfer: Attempts to prepare magnetically insensitive mF=0 states were impeded by low atom numbers and limited CAL operation time.
• Ramp discretization and anharmonicities: Hardware allowed only 80 current steps for I_coil, causing mismatches to ideal STA ramps; the trap explored anharmonic regions during diabatic transport. These factors constrained minimum residual oscillations and required ramp durations of 100–150 ms.
• Operational constraints: Remote, multi-user operations with limited sequence counts; no hardware modifications possible during campaigns; longest free-evolution times limited by apparatus and density; data acquisition spread over months with varying ISS orientation/altitude.