Figure

Description

The neutral atom qubit encodes quantum information in internal states (typically hyperfine ground states) of individual neutral atoms trapped in optical tweezers or optical lattices. Entangling gates exploit strong dipole–dipole interactions via transient excitation to Rydberg states.

Neutral atom platforms trap individual alkali atoms (⁸⁷Rb, ¹³³Cs) or alkaline-earth atoms (⁸⁸Sr, ¹⁷¹Yb) in tightly focused optical tweezer arrays. Qubit states are encoded in two hyperfine ground states |0⟩ and |1⟩, which have long coherence times due to weak coupling to the environment. Single-qubit gates are driven by microwave or two-photon Raman transitions. Two-qubit entangling gates use the Rydberg blockade mechanism: when one atom is excited to a high-lying Rydberg state (n ~ 50–100), the strong van der Waals interaction shifts the doubly-excited state out of resonance, creating a conditional phase.

Key advantages include identical qubits (no fabrication disorder), reconfigurable geometry via tweezer rearrangement, and mid-circuit atom reloading. Arrays of 1000+ atoms have been demonstrated, with programmable connectivity set by Rydberg interaction range (~5–10 μm), requiring atom shuttling for longer-range gates.

Hamiltonian

Single-qubit Hamiltonian in the rotating frame:

where is the Rabi frequency and the detuning.

Rydberg blockade Hamiltonian for two atoms separated by distance :

The blockade condition prevents simultaneous Rydberg excitation, enabling a controlled-Z gate.

Motivation

  • Identical qubits: Every atom of the same species is physically identical — no fabrication disorder, no frequency collisions.
  • Massive parallelism: Global Rydberg pulses entangle all non-interacting pairs simultaneously.
  • Reconfigurable connectivity: Atom shuttling via tweezer rearrangement enables arbitrary graph connectivity within a single zone.
  • Scalability: 1000+ atom arrays demonstrated; scaling beyond 10,000 atoms is plausible but requires solving control, cooling, and readout challenges at scale.
  • Mid-circuit operations: Erasure detection and atom reloading demonstrated, enabling erasure-error conversion and real-time error mitigation.

Experimental Status

First Rydberg blockade gate — Isenhower et al. (2010):

  • Demonstrated controlled-NOT gate between two individually trapped ⁸⁷Rb atoms via Rydberg blockade
  • Gate fidelity ~73% (limited by atomic motion and laser phase noise)

Large-scale programmable arrays — Ebadi et al. (2021):

  • 256-atom programmable quantum simulator using ⁸⁷Rb tweezer arrays
  • Demonstrated quantum phases of matter and optimization algorithms

High-fidelity entanglement — Bluvstein et al. (2024):

  • Two-qubit CZ gate fidelity of 99.5% in ¹³³Cs tweezer arrays
  • Single-qubit gate fidelity of 99.97%
  • Logical qubit operations with error correction demonstrated

1000+ atom arrays — Pause et al. (2024):

  • 1225-atom ⁸⁷Rb tweezer array with >97% loading efficiency

Key Metrics

MetricValueNotesFidelity reference
1Q gate fidelity99.97%¹³³Cs tweezersBluvstein et al. 2024
2Q gate fidelity (CZ)99.5%¹³³Cs tweezersBluvstein et al. 2024
T₂ (Ramsey)~1 s⁸⁷Rb hyperfineKleine Büning et al. 2011
T₂ (spin echo)~10 s¹⁷¹Yb clock statesMa et al. 2022
Readout fidelity99.8%¹³³Cs fluorescenceBluvstein et al. 2024
Array size1225 atomsRb tweezer arrayPause et al. 2024
Atom loss per circuit~0.5% per layer¹³³CsBluvstein et al. 2024

Scaling Considerations

  • Connectivity: Reconfigurable via atom shuttling; effective all-to-all within Rydberg range (~5–10 μm). Longer-range gates require physical atom transport with associated decoherence.
  • Parallelism: Global Rydberg pulses enable parallel entangling gates on non-interacting pairs.
  • Error budget: Dominated by Rydberg state decay, atomic motion (Doppler shifts), and atom loss (~0.5% per circuit layer, compounding in deep circuits).
  • Mid-circuit operations: Erasure detection via shelving to auxiliary states demonstrated; enables erasure-error conversion.

Limitations & Challenges

  • Gate speed: Two-qubit gate times (~μs) are ~1000× slower than superconducting qubits (~ns), limiting circuit depth per unit time.
  • Atom loss: Probabilistic atom loss during Rydberg excitation and imaging compounds at ~0.5% per circuit layer, requiring active reloading strategies.
  • Probabilistic loading: Initial tweezer loading is stochastic (~50% per site); defect-free assembly via rearrangement is time-costly and scales with array size.
  • Vacuum requirements: Ultra-high vacuum (~10⁻¹¹ Torr) systems are large, expensive, and have long pump-down times.
  • Laser noise & stability: Rydberg excitation requires high-power, frequency-stabilized lasers; laser phase noise is a dominant gate error source.
  • Crosstalk: At high site densities, residual Rydberg interactions and optical crosstalk between closely spaced tweezers can introduce correlated errors.

References

Original proposal

Experimental demonstrations

Linked Papers