PaperlayerView exact revision on arXiv

Source-linked AI summary

Quantum computing with atomic qubits and Rydberg interactions: Progress and challenges

M. Saffman

arXiv:1605.05207v2quant-phphysics.atom-ph

TL;DR

Neutral-atom quantum computing must meet stringent requirements beyond demonstrating high-fidelity gates, including scalable arrays, atom-loss management, and low-crosstalk measurement. This review examines these requirements, Rydberg-mediated gates, and simulation approaches, concluding that neutral atoms could provide thousands of qubits while facing substantial technical challenges.

  • Problem

    Useful neutral-atom quantum computing requires scalable architectures, error-corrected gates, reliable atom management, and low-crosstalk initialization and measurement.

  • Method

    The review evaluates neutral-atom architectures, qubit-array preparation, Rydberg-mediated two- and multi-qubit gates, measurement, and resources needed for scalability.

  • Results

    Neutral atoms can conceivably provide many thousands of qubits in a footprint smaller than 1 mm^2, while scalable computation requires two-qubit gate fidelity F ∼0.9999.

  • Takeaways & Limitations

    Neutral-atom platforms show progress toward larger arrays and higher-fidelity entanglement, but engineering large-scale systems will require scalable laser resources and continued advances in loading, measurement, and error correction.

  • Takeaways & Limitations

    Large-scale systems face a laser-power requirement that grows with the number of parallel site-selected gates, creating a technical and economic challenge.

Abstract

from arXiv · show

We present a review of quantum computation with neutral atom qubits. After an overview of architectural options and approaches to preparing large qubit arrays we examine Rydberg mediated gate protocols and fidelity for two- and multi-qubit interactions. Quantum simulation and Rydberg dressing are alternatives to circuit based quantum computing for exploring many body quantum dynamics. We review the properties of the dressing interaction and provide a quantitative figure of merit for the complexity of the coherent dynamics that can be accessed with dressing. We conclude with a summary of the current status and an outlook for future progress.

1. Introduction

The review evaluates neutral-atom quantum computing against the requirements for scalable quantum computation, emphasizing that high-fidelity gates are necessary but not sufficient. It surveys architectures, array scaling, gate operations, and alternative simulation approaches.

  • 1. Introduction: Neutral-atom quantum computing is assessed as a candidate platform for solving problems that may be classically intractable.The review places neutral atoms among several physical platforms developed for quantum computation and simulation.
  • 1. Introduction: A high-fidelity two-qubit entangling gate remained to be demonstrated with neutral atoms at the review’s stated point.The authors caution that gate fidelity is only one of several scalability challenges.
  • 1. Introduction: The review uses the DiVincenzo criteria to organize discussion of scalable arrays, trap lifetime, coherence, initialization, measurement, and logic gates.Later sections also cover multi-qubit operations, experimental gate challenges, quantum simulation, and Rydberg dressing.
  • 1. Introduction: Neutral-atom architectures include arrays of single-atom qubits, with Rydberg interactions providing routes to entangling and multi-atom operations.The review also considers ensemble encodings and alternatives to circuit-based computation.

2. Neutral atom architecture

Neutral-atom architectures combine optically or magnetically trapped atoms with several qubit-encoding and loading strategies. Scaling is constrained by atom preparation, trap lifetime, addressability, optical power, loss correction, and measurement crosstalk.

  • 2. Neutral atom architecture: Neutral-atom registers can use one two-level atom per qubit, ensembles of N atoms per qubit, or collective encoding in one ensemble.The collective encoding uses K qubits with atoms having K + 1 internal levels.
  • 2. Neutral atom architecture: Heavy alkali atoms such as Rb and Cs offer long-coherence ground-state qubits, GHz-scale hyperfine frequencies, and resolved excited-state structure for preparation and readout.These properties support optical pumping and resonance-fluorescence measurements.
  • 2.1. Qubit arrays: Arrays of several hundred sites have been implemented, and scaling to N ∼ 10^4 sites is considered realistic with appropriate laser development.Optical-array size is limited partly by laser power, while detuning trades trap depth against photon-scattering-induced heating and decoherence.
  • 2.1. Qubit arrays: 91% single-site and 90% four-site loading have been achieved, while movable tweezers have enabled nearly 100% filling of ∼50-atom arrays on ∼100 ms timescales.Rearrangement can convert partially filled lattices into smaller fully loaded arrays.
  • 2.1. Qubit arrays: Addressing large arrays requires optical control technologies that trade off speed, crosstalk, and the number of addressable sites.Candidate approaches include electro-optic, acousto-optic, piezo, micro-electromechanical, and spatial-light-modulator systems.
  • 2.2. Trap lifetime: Atom loss is an unavoidable part of the neutral-atom error budget, making trap lifetime, rapid diagnosis, replacement, and error correction essential.A 20-qubit code with threshold ǫ = 0.0001 requires τvac = 400 s, while demonstrated loading rates remain about 10^3 s−1 in 1D and 2D arrays.
  • 2.2. Trap lifetime: Cryogenic environments can lengthen atom and Rydberg lifetimes, but deep optical traps impose substantial scattering, laser-power, and heat-load constraints.Magnetic traps avoid some optical power and dissipation limits but require challenging field control for deep traps at micrometre scales.
  • 2.2. Trap lifetime: Ensemble encoding reduces the fidelity loss from losing one atom to O(1/N), but current ensemble qubits show worse gate fidelity and shorter coherence times than single-atom qubits.This creates a trade-off between loss tolerance and operational performance.

3. Coherence

Neutral-atom coherence is limited by relaxation, dephasing, and changes in trapping potentials during Rydberg excitation. Hyperfine qubits can nevertheless store states for many seconds, while magic-trap designs aim to reduce heating and decoherence during gates.

  • Relaxation and noise: Hyperfine qubits suppress low-frequency magnetic-field transitions because their energy separation is several GHz, unlike Zeeman qubits with MHz-scale splittings.Unshielded microwave-frequency fields can still cause transitions, making RF shielding important.
  • Coherence times: Hyperfine-qubit T2 dephasing arises from magnetic noise, optical-trap intensity noise, and motional effects.Appropriate hyperfine states and choices of optical intensity, polarization, and magnetic field enable coherence times of many seconds.
  • Rydberg trapping: Rydberg gates can heat or anti-trap atoms because Rydberg and ground states experience different trapping potentials.Turning the trap off during the short Rydberg interaction mitigates this effect, but future arrays would benefit from gates with the trap left on.
  • Rydberg magic trapping: Magic traps can match ground- and Rydberg-state potentials over a broad wavelength range by shaping the trap to the Rydberg electron wavefunction.The early proposals required relatively small detuning, producing excessive scattering and suboptimal coherence; later designs avoid this restriction.
  • Rydberg magic trapping: A negative Rydberg polarizability enables dark trapping when the ground state also has negative polarizability and |αRyd| < |αground|.These conditions are satisfied over a broad wavelength range for Rb and Cs.
  • Rydberg magic trapping: Landscape-modulated trapping with positive ground-state polarizability is limited to long wavelengths near 10 µm and high principal quantum numbers such as n ≥154 for Rb ns states.It accounts for the three-dimensional overlap of the Rydberg wavefunction with the repulsive ponderomotive potential.

4. Initialization and Measurement

Scalable neutral-atom computation requires initialization and measurement that remove entropy while preserving qubit states and avoiding crosstalk. Existing fluorescence-based approaches can be high fidelity but may lose atoms, while QND, Stark-shift, and dual-species strategies address these constraints with important remaining experimental boundaries.

  • Requirements: Scalable computation requires dissipative initialization and measurement for error correction, alongside coherent gate operations.Optical pumping initializes qubits, while resonant-fluorescence detection is a principal measurement approach.
  • Fluorescence readout: Fluorescence measurement preceded by blow-away light infers the qubit state from whether an atom remains after repumping.The method avoids relying on Raman-free scattering throughout the full measurement but removes one hyperfine state before readout.
  • QND measurement: High-fidelity fluorescence measurements lose an atom half the time on average, requiring mechanical reloading and reinitialization before computation can continue.Lossless QND measurements are possible when so few photons are scattered that Raman-transition probability is negligible.
  • QND measurement: Multiplexed QND detection of more than two atoms in an array has not yet been demonstrated, and EMCCD cameras add excess noise compared with discrete photon detectors.Recent work has produced initial demonstrations of EMCCD-based QND state measurements.
  • Crosstalk: Measurement crosstalk remains an outstanding challenge because photons scattered from a selected atom can be absorbed by proximal qubits.At lattice spacings d ∼5λ, the cited estimates give ηabs ∼0.0015 and ηdet ∼0.034, with ηabs/ηdet ∼0.04.
  • Crosstalk mitigation: Focused-beam Stark shifting can add a calculated crosstalk suppression factor of >100 relative to no Stark shifting.Other proposed routes include shelving surrounding qubits or using a two-species architecture.
  • Dual-species architecture: A proposed dual-species architecture computes with Cs qubits and measures auxiliary Rb qubits after transferring the Cs state to them.Interspecies Förster resonances provide strong Cs–Rb Rydberg coupling, while the species occupy interleaved dark and bright traps.
  • Open challenge: None of the proposed crosstalk-free initialization and measurement solutions has yet been demonstrated experimentally.Demonstrating one would be an important step toward scalable quantum computation.

5. Quantum gates

Neutral-atom quantum gates have achieved high-fidelity single-qubit control, but two-qubit entanglement remains substantially below the fidelity needed for scalable error-corrected computation. The review examines intrinsic limits, experimental errors, and protocols aimed at approaching F ∼0.9999.

  • One-qubit gates: Single-qubit Clifford fidelities of 0.992 and 0.996 were demonstrated, with crosstalk errors of 0.014 and 0.0046, respectively.Pulse sequences reduced sensitivity to fourth order in beam-pointing errors.
  • Two-qubit gates: Two-qubit neutral-atom entanglement fidelities remain far below those achieved with trapped-ion and superconducting qubits.The review identifies this fidelity gap as a central challenge for neutral-atom quantum computation.
  • Two-qubit gates: Scalable computation is evaluated against a target two-qubit gate fidelity of F ∼0.9999, reflecting the requirements of small quantum-error-correction codes.Codes with 25 qubits or fewer have thresholds near 0.001, motivating gate errors at least ten times smaller.
  • Intrinsic gate fidelity: For Cs, the estimated intrinsic error floor is Emin ≃2 × 10−5, but coherent phase variations and leakage reduce calculated process fidelities to about 0.9988–0.9989 at 300 K.The error estimate based on computational-basis inputs does not capture all errors affecting superposition inputs and entangled outputs.
  • Intrinsic gate fidelity: Intrinsic blockade-gate error scales as 1/(Bτ)2/3, but neither blockade nor interaction protocols saturates its corresponding error bound.The review therefore identifies improved protocols that saturate the bound as an open challenge.
  • Experimental issues for Rydberg gates: A ∼100 ns gate with leakage-suppressing pulses enabled a predicted Rydberg-DRAG fidelity above 0.9999 for one-photon Cs excitation at room temperature.Two-photon excitation could achieve similar fidelity if sufficient laser power permits fast, far-detuned excitation that suppresses spontaneous emission.

6. Other approaches

Beyond circuit-model gates, neutral-atom platforms support ensemble encodings, quantum simulation, and Rydberg dressing for engineering many-body interactions. Dressing produces soft-core interactions whose accessible coherent dynamics can be quantified, but collective effects and decoherence constrain scalability.

  • Alternative encodings and paradigms: Neutral-atom computation can use ensemble encodings, decoherence-free logical subspaces, topological-state preparation, one-way computation, and adiabatic computation.These approaches extend beyond single-atom circuit-model qubits and standard gate operations.
  • Quantum simulation: Rydberg interactions provide long-range, anisotropic interactions for quantum simulation and tailored many-body spin models.Choosing interacting Rydberg states controls the interaction structure, while experiments have observed long-range and spin-dependent interactions.
  • Rydberg dressing: Far-off-resonant Rydberg dressing creates a soft-core potential through blockade, with its sign determined by the detuning and Förster defect.The interaction is attractive for δ > 0, ∆ > 0 and repulsive for δ < 0, ∆ < 0; other sign combinations can produce finite-distance excitation resonances.
  • Rydberg dressing: The two-atom dressing model does not directly apply to dense gases when |Ω/∆| is not sufficiently small, because collective many-body effects emerge.The cited crossover conditions distinguish two-body dressing from collective behavior.
  • Rydberg dressing: Exact dipole-dipole and van der Waals descriptions agree in weak dressing, but differ substantially when |Ω| is comparable to |∆|.Near the origin, the full interaction varies as R^3, whereas the van der Waals approximation gives an R^6 dependence and a flatter core.
  • Coherent-dynamics resources: The dressing figure of merit scales asymptotically as n^19/3, n^20/3, and n^7 in 1D, 2D, and 3D, respectively.For |∆dr(0) − ∆dr(∞)| = 2π × 20 kHz and τdr = 16 ms, the estimate gives 320 coherent operations per atom.
  • Coherent-dynamics resources: A modified figure of merit scales as F′ ∼ n^6 independently of dimensionality, while available coherent evolution steps per atom decrease as system size increases.Using the stated parameters gives F′ = 640 and 95, 18, and 4 operations per atom in 1D, 2D, and 3D.

7. Outlook

Neutral-atom platforms offer an attractive path toward very large, compact quantum systems, but substantial experimental and engineering challenges remain. The review identifies promising gate-fidelity projections while emphasizing unresolved hardware, loading, measurement, crosstalk, and noise requirements.

  • Scaling prospects: Many thousands of neutral-atom qubits could fit within less than 1 mm^2, an attractive scaling capability that is difficult for other technologies to match.The review also notes progress in larger arrays, deterministic loading, higher-fidelity entanglement, and ensemble-qubit preparation.
  • Outlook: The review concludes that many challenges remain before neutral-atom quantum computers become reality, while finding no principle-level reason they cannot be overcome.This outlook is presented alongside ongoing experimental progress and unresolved questions about Rydberg-dressing dynamics.
  • Remaining challenges: Primary challenges include higher-fidelity entangling gates, atom loading and reloading, QND measurement, low-crosstalk operations, and electric-field-noise control.These requirements apply to arrays with qubit spacings of only a few microns.
  • Gate performance: Rydberg gate protocols could reach Bell state fidelity F = 0.9999 in real atoms at room temperature if Doppler, laser, and electric-field noise are addressed.The review anticipates still higher fidelity in a cryogenic environment with increased Rydberg lifetime.
  • Systems integration: High-fidelity control of multi-qubit arrays and error correction require closely integrated ultrahigh-vacuum, electro-optical, laser, and classical-computing hardware not currently available.The authors frame this integration challenge as substantial but comparable to the challenges facing other promising quantum-computing technologies.
Loading 1605.05207v2…