Source-linked AI summary
High-fidelity parallel entangling gates on a neutral atom quantum computer
Simon J. Evered, Dolev Bluvstein, Marcin Kalinowski, Sepehr Ebadi, Tom Manovitz, Hengyun Zhou, Sophie H. Li, Alexandra A. Geim, Tout T. Wang, Nishad Maskara, Harry Levine, Giulia Semeghini, Markus Greiner, Vladan Vuletic, Mikhail D. Lukin
TL;DR
Neutral-atom entangling gates need lower errors while retaining parallelism and scalable connectivity. The paper combines optimal-control pulses, dark-state scattering suppression, and hardware improvements, achieving approximately 99.5% two-qubit CZ fidelity at up to 60 qubits in parallel and demonstrating a three-qubit CCZ gate.
Problem
Entangling gates in neutral-atom systems needed substantially improved fidelity while preserving parallel operation and scalability.
Method
The paper uses optimal-control single-pulse gates, dark-state engineering, and improved Rydberg excitation and atom cooling, with repeated-gate benchmarking.
Results
The time-optimal CZ reaches 99.48(2)% fidelity on 60 qubits in parallel, while a CCZ gate reaches a fidelity consistent with 97.9(2)% across 21 qubits.
Takeaways & Limitations
The demonstrated gates support high-fidelity parallel entangling operations and extend the approach to native three-qubit gates.
Takeaways & Limitations
Further fidelity scaling requires suppressing adjacent-Rydberg-state coupling, optimizing pulse rise times, and managing higher laser intensity.
Abstract
from arXiv · showhide
The ability to perform entangling quantum operations with low error rates in a scalable fashion is a central element of useful quantum information processing. Neutral atom arrays have recently emerged as a promising quantum computing platform, featuring coherent control over hundreds of qubits and any-to-any gate connectivity in a flexible, dynamically reconfigurable architecture. The major outstanding challenge has been to reduce errors in entangling operations mediated through Rydberg interactions. Here we report the realization of two-qubit entangling gates with 99.5% fidelity on up to 60 atoms in parallel, surpassing the surface code threshold for error correction. Our method employs fast single-pulse gates based on optimal control, atomic dark states to reduce scattering, and improvements to Rydberg excitation and atom cooling. We benchmark fidelity using several methods based on repeated gate applications, characterize the physical error sources, and outline future improvements. Finally, we generalize our method to design entangling gates involving a higher number of qubits, which we demonstrate by realizing low-error three-qubit gates. By enabling high-fidelity operation in a scalable, highly connected system, these advances lay the groundwork for large-scale implementation of quantum algorithms, error-corrected circuits, and digital simulations.
Neutral atom entangling gates
The paper combines optimal-control single-pulse Rydberg gates with dark-state engineering and experimental improvements to reduce entangling-gate errors in parallel neutral-atom arrays.
- Gate operation: Rydberg-mediated gates place atoms at designated sites and selectively excite |1⟩ atoms through an intermediate state while leaving |0⟩ atoms unexcited.Two- and three-qubit gates are implemented by modulating Rabi frequency and phase profiles.
- Error sources: The main physical error sources are Rydberg decay, intermediate-state scattering, atomic temperature, dephasing, miscalibration, laser noise, and inhomogeneity.These mechanisms motivate both pulse optimization and apparatus improvements.
- Gate design: Single-pulse gates use numerically optimized continuous phase profiles and globally tunable parameters for robust calibration.The gate family includes a parameterized time-optimal gate and a smooth-amplitude gate.
- Error suppression: Dark-state optimization suppresses intermediate-state scattering by minimizing bright-state population through detuning-sign choices and pulse shaping.The dark state excludes the short-lived intermediate state |e⟩.
- Experimental improvements: Higher intermediate-state detuning, a 4.6 MHz two-photon Rabi frequency, and colder atoms with radial phonon occupation approximately 1–2 reduce scattering and motional decoherence.The upgrades use a lower-lying n = 53 Rydberg state, higher laser power, gray-molasses cooling, and improved optical pumping.
Entangling gate characterization
Multiple repeated-gate characterization methods measure CZ fidelities near 99.5%, with consistent results across Bell-state and randomized-benchmarking protocols.
- Bell-state characterization: 99.52(4)% CZ fidelity is extracted from exponential Bell-state-fidelity decay after repeated CZ applications.The raw single-gate Bell-state fidelity is 98.0(2)%, while the SPAM-corrected estimate is 99.4(4)%.
- Randomized benchmarking: 99.54(2)% CZ fidelity is obtained using global randomized benchmarking with random single-qubit rotations between gates.The protocol keeps the number of single-qubit rotations fixed across data points.
- Phase calibration: 99.48(2)% fidelity is measured with randomized benchmarking without interleaved X gates, enabling calibration of the single-particle phase.The method additionally benchmarks the gate while retaining sensitivity to that phase.
- Gate variants: 99.55(3)% fidelity is achieved by the smooth-amplitude CZ gate, comparable to the parameterized time-optimal gate.The smooth-amplitude profile can strongly suppress scattering under closer-detuned excitation.
Scaling up
The time-optimal CZ gate maintains homogeneous fidelity across parallel gate sites and reaches 60-qubit operation without site-specific calibration, while further scaling is constrained mainly by laser resources.
- Scaling up: 99.48(2)% CZ fidelity is achieved on 60 qubits in parallel with good homogeneity across the array.The larger system uses larger Rydberg beams while maintaining the same intensity.
- Scaling up: Gate fidelity remains constant across 10 individual gate sites within statistical error despite globally shared calibration and control.More gate sites therefore do not increase calibration overhead in this experiment.
- Error analysis: Modeling attributes residual CZ infidelity to Rydberg decay, coupling to another Rydberg level, intermediate-state scattering, and ground–Rydberg dephasing.The measured ground–Rydberg T*2 is 3 µs and is dominated by laser light-shift fluctuations and finite atomic temperature.
- Correlated errors: High-weight correlated errors are largely absent, although small covariance growth appears between neighboring gate sites.The possible sources are correlated detuning fluctuations or long-range interactions following Rydberg decay.
- Scaling constraints: The dominant challenge for larger parallel arrays is scaling laser power while maintaining beam homogeneity.Other analyzed decoherence mechanisms appear independent of system size.
Fast multi-qubit gates
Optimal control generalizes the gate scheme to native multi-qubit controlled-Z operations, including a demonstrated three-qubit CCZ gate with low error across parallel sites.
- Fast multi-qubit gates: 97.9(2)% fidelity is consistent with a three-qubit CCZ gate demonstrated across 21 qubits in parallel.The fidelity is inferred from repeated entangling and disentangling of a three-qubit GHZ state, not rigorous randomized benchmarking.
- Fast multi-qubit gates: The time-optimal CCZ gate takes only 44% longer than the time-optimal CZ and is faster than other known CCZ profiles.The gate is realized by arranging atom triplets into triangular sites and applying a global pulse.
Discussion and outlook
The results support high-fidelity digital circuits and future exploration of large-scale error correction with neutral atoms. Further progress depends on improving fidelity and integrating additional readout and control capabilities.
- Future improvements: 99.9% gate fidelity is theoretically projected at 3x higher Rabi frequency and 2x further detuning.Achieving this projection requires suppressing adjacent-Rydberg-state coupling, optimizing pulse rise times, and managing high laser intensity.
- Future improvements: Microscopic error-source analysis can guide decomposition into Pauli channels, atom loss, and leakage.
- Quantum error correction: Efficient parallel control of logical qubits opens opportunities to explore large-scale quantum error correction.Proposed supporting capabilities include moving atoms to readout zones, using ancillary species, shelving data qubits, and nondestructive readout.
- Multi-qubit gates: The CCZ experiment uses triangular three-qubit gate sites within a 21-qubit entangling zone.The triangular configuration provides strong, symmetric interactions between three qubits.
- Multi-qubit gates: Repeated CCZ applications produce GHZ-state fidelity decay fitted by 0.979(2)^N_CCZ.The figure also compares theoretical gate duration scaling with qubit number for native Rydberg-blockade multi-qubit gates.
- Future directions: Alkaline-earth atoms offer additional opportunities including single-photon Rydberg excitation, nuclear-spin control, and erasure conversion.
METHODS
The methods combine neutral-atom preparation, optimized Rydberg gate control, microscopic error modeling, and parallel-gate benchmarking. The study also examines scalability, correlated errors, and three-qubit CCZ operation.
- Experimental system: Two-qubit gates use optically prepared and Raman-controlled 87Rb clock-state qubits in rearranged, cooled optical-tweezer arrays.Atoms are prepared in hyperfine clock states and entangled after state-selective two-photon excitation to Rydberg states.
- Gate physics: Opposite-sign detunings route Rydberg excitation predominantly through a dark state, suppressing intermediate-state population and scattering.The dark-state condition is δ∆ < 0, while the bright-state contribution produces more intermediate-state scattering.
- Gate physics: Smooth-amplitude pulses reduce scattering by maintaining dark-state occupation, with suppression factors of 1.2 at equal duration and roughly 2.5 when twice as long.The slower gate is useful when scattering dominates, whereas faster operation can be preferable when dephasing dominates.
- Error analysis: Microscopic modeling projects 99.9% fidelity through combinations including 2x larger detuning, a 3x longer Rydberg lifetime, and 2x longer T∗_2.An alternative projection uses 3x higher Rabi frequency with suppressed coupling to the other mJ state and 2x larger detuning.
- Error analysis: The modeled error mixture is dominated by Z-type, atom-loss, and leakage errors, while global randomized benchmarking is expected to faithfully capture fidelity despite symmetry.The benchmarking procedure is less sensitive to some error types because it uses global rotations.
- Parallel operation: Across 20- and 60-atom datasets, error counts are approximately Poissonian and large-scale correlated errors are uncommon, although small nearby-site covariance appears.After one gate, these correlations are smaller than the main error sources; atom transport can further suppress effects from leftover Rydberg decay.