Neutral Atom Quantum Computers
Neutral-atom quantum computers trap thousands of individual atoms using focused laser beams (optical tweezers) and perform gates by exciting atoms to high-energy Rydberg states. When two atoms both enter the Rydberg state, the strong dipole-dipole interaction creates a controlled phase gate — the "Rydberg blockade." The tweezer array is software-reconfigurable, allowing arbitrary connectivity patterns. QuEra's Aquila (256 atoms) and PASQAL's Fresnel (100 atoms) are currently accessible via cloud.
GATE SPEED
Single-qubit gates: ~100 ns–1 µs
CONNECTIVITY
Reconfigurable — atoms can be physically…
ENVIRONMENT
Vacuum chamber
How Neutral Atom Qubits Work
Neutral-atom quantum computers use individual electrically neutral atoms — typically Rubidium-87 or Caesium-133 — trapped by tightly focused laser beams called optical tweezers. Unlike trapped ions, neutral atoms do not interact with each other via long-range Coulomb forces. Instead, entanglement is achieved by temporarily exciting atoms to extremely high-energy electronic states called Rydberg states.
A Rydberg atom has an electron in a shell far from the nucleus, giving it an enormous electric dipole moment. When two nearby atoms are both excited toward the Rydberg state, their dipole-dipole interaction is so strong that the excitation of one atom prevents the other from being excited — the Rydberg blockade. This blockade mechanism is the engine of two-qubit entangling gates in neutral-atom systems. By carefully timing the Rydberg excitation laser pulses, a controlled-Z or controlled-X gate is implemented between pairs of atoms within the blockade radius (typically 5–15 micrometers).
The key architectural advantage is reconfigurability: optical tweezers are controlled by acousto-optic deflectors or spatial light modulators, so the atom array can be rearranged between gate operations. Atoms can be moved to create arbitrary connectivity patterns, or to sort a randomly loaded array into a defect-free configuration. This enables both digital gate-based computation and analog quantum simulation of spin models, Hubbard models, and other many-body Hamiltonians.
Physical Implementation
Connectivity and Gate Operations
Connectivity
Reconfigurable connectivity — atoms can be physically moved between gate operations to create arbitrary qubit coupling geometries. Within a single gate zone, all atoms within the Rydberg blockade radius (~10 µm) can be entangled. Global Rydberg pulses can address the full array simultaneously. This enables both local nearest-neighbor interactions and longer-range gates via atomic transport.
Native Gate Set
Single-qubit: arbitrary rotations via global or locally addressed microwave and optical pulses. Two-qubit: CZ gate (Rydberg blockade), CCZ gate (multi-control). Analog mode: programmable Ising Hamiltonian evolution. Mid-circuit measurement and erasure detection demonstrated; qubit reinsertion from reservoir demonstrated by Atom Computing.
Gate Speed
Single-qubit gates: ~100 ns–1 µs; two-qubit (Rydberg) gates: ~1–10 µs
Potential Advantages
- ✓Reconfigurable qubit array — atoms can be repositioned to create any desired connectivity graph without SWAP overhead
- ✓Demonstrated arrays of 1000+ atoms (Atom Computing, 2023); scaling by adding more tweezers is straightforward in principle
- ✓No dilution refrigerator required — significantly lower infrastructure cost and operational complexity than superconducting
- ✓Natural analog quantum simulation for spin models and lattice gauge theories in 2D and 3D geometries
- ✓Mid-circuit measurements and erasure error detection have been demonstrated, improving error correction efficiency
- ✓Stochastic atom loss is detectable as erasure errors (much easier to correct than depolarizing errors)
Limitations & Challenges
- —Atom loading is stochastic — each experimental cycle has random atom loss requiring active reloading from a reservoir
- —Rydberg state lifetimes limit gate fidelity; two-qubit gate fidelities (~99–99.5%) currently lag trapped-ion by ~0.5–1%
- —Rydberg blockade radius of ~10 µm limits which atoms can directly interact; long-range gates require atom transport
- —Atom transport adds latency; moving atoms adds heat and risks atom loss
- —Gate clock rates are currently slower than superconducting (µs scale gates vs. ns scale)
- —Large-scale integration of independent-site laser addressing is technically challenging
Scaling Considerations
Neutral-atom arrays scale naturally by expanding the tweezer array — demonstrated from small (10-atom) systems to 1,180-atom arrays (Atom Computing, 2023). The challenge is maintaining high-fidelity, site-resolved addressing as arrays grow. Proposed three-dimensional tweezer arrays would dramatically increase qubit count. For error correction, the built-in erasure detection from atom loss substantially reduces the qubit overhead compared to depolarizing noise models — a compelling advantage for reaching fault tolerance. QuEra and Harvard demonstrated 48-logical-qubit error-corrected circuits using neutral atoms in 2023, a significant milestone for error-corrected computation.
Coherence: Coherence times in hyperfine states: seconds. Rydberg state lifetimes limit gate fidelity; T2* for qubit register: typically milliseconds.
Available QPUs
View all →Leading Companies
Workload Suitability
Full guide →Simulating molecular electronic structure to compute ground-state energies, reac…
Simulating solid-state physics phenomena including high-temperature superconduct…
Solving NP-hard combinatorial problems including vehicle routing, supply chain s…
Monte Carlo simulation acceleration via quantum amplitude estimation, derivative…
Training and inference on quantum neural networks (QNNs), kernel methods using q…
Shor's algorithm can factor large integers and solve discrete logarithms exponen…
WELL SUITED FOR
- ›Analog quantum simulation of condensed matter and AMO physics problems
- ›Combinatorial optimization with Rydberg blockade encoding (MaxCut, graph problems)
- ›Quantum chemistry simulation with reconfigurable connectivity
- ›Machine learning circuits with variable connectivity per layer
- ›Error correction research leveraging erasure detection advantage
LESS SUITED FOR
- —High-throughput applications requiring microsecond cycle times
- —Algorithms requiring extremely deep gate sequences without mid-circuit reloading
- —Problems with strict requirements on deterministic, loss-free qubit behavior
Frequently Asked Questions
Key References
- [1]Logical quantum processor based on reconfigurable atom arrays — Nature / Harvard & QuEra, 2023
- [2]High-fidelity parallel entangling gates on a neutral-atom quantum computer — Nature / Lukin Group, 2022
- [3]Observation of thermalization and information scrambling in a superconducting quantum processor — QuEra Computing, 2023