Qubit syndrome measurements with a high fidelity Rb-Cs Rydberg gate

1. Problem Statement

Neutral atom quantum computers have scaled to large qubit counts and demonstrated high-fidelity single-species entangling gates. However, implementing Quantum Error Correction (QEC) requires frequent, mid-circuit measurements of "ancilla" qubits to detect errors on "data" qubits without disturbing the data.

• Current Limitations: In single-species arrays, measuring an ancilla qubit typically disturbs neighboring data qubits due to spectral overlap (crosstalk). To avoid this, current architectures require complex, slow operations such as physically moving atoms to a dedicated measurement zone, or "shelving/hiding" data qubits in non-interacting states. These extra steps increase the error budget and reduce the logical cycle rate.
• The Goal: Develop a method for in-place, non-demolition (QND) syndrome measurements that eliminates the need for atom transport or shelving, thereby accelerating the QEC cycle and reducing overhead.

2. Methodology

The authors propose and demonstrate a solution using dual-species neutral atom arrays consisting of Rubidium (Rb) and Cesium (Cs).

• Dual-Species Architecture: By using two different atomic elements, the system exploits their distinct optical transitions. This allows for independent addressing and readout of Rb and Cs qubits using different laser wavelengths, effectively decoupling the data and ancilla registers.
• Experimental Setup:
  • Platform: A 7x7 optical tweezer array (1064 nm) generated by a Spatial Light Modulator (SLM).
  • Cooling: Atoms are pre-cooled, loaded into a 3D MOT, and then subjected to adiabatic cooling by ramping down trap depths. This reduces temperatures to ~3.2 µK (Rb) and ~4.3 µK (Cs), essential for high-fidelity Rydberg gates.
  • Qubit Encoding: Hyperfine clock states ($|f, m_f\rangle$) for $^{87}$Rb and $^{133}$Cs.
  • Gate Mechanism: A Rydberg blockade gate. Neighboring Rb and Cs atoms are excited to Rydberg states ($|63s_{1/2}\rangle$ for Rb and $|65s_{1/2}\rangle$ for Cs) via two-photon excitation using tightly focused lasers (421/1005 nm for Rb; 459/1040 nm for Cs).
• Gate Protocol:
  • The gate utilizes a parameterized, time-optimal Rydberg protocol.
  • Global microwave rotations (6.8 GHz for Rb, 9.2 GHz for Cs) and local phase shifts (via differential light shifts from Rydberg lasers) are used to implement single-qubit gates and correct phases.
  • Randomized Benchmarking (RB): Used to characterize the fidelity of the CZ gate and single-qubit operations.
• Syndrome Measurement:
  • The CZ gate is used to entangle data qubits with an ancilla qubit.
  • The ancilla is then measured independently using species-specific fluorescence imaging (780 nm for Rb, 852 nm for Cs) without affecting the data qubit.

3. Key Contributions

1. High-Fidelity Interspecies Gate: Demonstration of a Rb-Cs controlled-Z (CZ) gate with a fidelity of $F = 0.975 \pm 0.002$. This represents an order-of-magnitude improvement over previous Rb-Cs results and exceeds the threshold required for surface code error correction.
2. In-Place QND Syndrome Extraction: Successful demonstration of Quantum Non-Demolition (QND) measurements on multi-atom plaquettes (2-qubit and 3-qubit) without moving atoms or using shelving techniques.
3. Error Budget Analysis: A comprehensive Monte Carlo simulation of the error budget, identifying specific error sources (e.g., atom localization, laser noise, blockade fluctuations) and projecting a path to fidelities $>0.997$.
4. Decoupled Registers: Experimental verification that measuring one species (e.g., Cs) does not induce statistically significant decoherence in the other (Rb), validating the feasibility of crosstalk-free mid-circuit measurements.

4. Results

• Gate Fidelity:
  • CZ Gate: $F = 0.975(2)$, significantly higher than the previous Rb-Cs record of 0.69 and the heteronuclear record of 0.73.
  • Single-Qubit Gates: Randomized benchmarking yielded fidelities of $0.99963(5)$ for Rb and $0.99962(5)$ for Cs.
  • Coherence: $T_1$ lifetimes of 9.6 s (Rb) and 33.5 s (Cs); $T_2^*$ coherence times of 44.9 ms (Rb) and 27.2 ms (Cs). No significant difference was found between isolated and simultaneous (crosstalk) measurements.
• Syndrome Measurement Fidelity ($F_{QND}$):
  • 2-Qubit Plaquette: $F_{QND} = 0.933(12)$ (Rb target) and $0.865(17)$ (Cs target).
  • 3-Qubit Plaquette: $F_{QND} = 0.865(17)$. This corresponds to a weight-2 parity check (ZZ stabilizer), a building block for surface codes.
• Simulation vs. Experiment:
  • The experimental fidelity of 0.975 aligns closely with Monte Carlo simulations predicting $1-F = 0.0222$.
  • Future Projections: Simulations suggest that with realistic hardware improvements (smaller trap waists, lower temperatures, higher blockade strength, and reduced noise), the gate fidelity could reach $F > 0.997$, well within the surface code threshold.

5. Significance

• Enabling Scalable QEC: This work provides a critical pathway to measurement-based error correction without the latency and error overhead associated with atom transport or complex shelving protocols. By keeping data and ancilla qubits in place, the logical cycle rate can be significantly increased.
• Dual-Species Advantage: It validates the use of dual-species arrays as a superior architecture for fault-tolerant quantum computing, leveraging the natural spectral isolation of different elements to solve the crosstalk problem inherent in single-species systems.
• Performance Benchmark: Achieving a 0.975 fidelity for a heteronuclear gate demonstrates that interspecies interactions can be as robust as intraspecies ones, opening the door to more complex logical qubit architectures and new variants of the surface code.
• Roadmap to Fault Tolerance: The detailed error modeling and projection to >0.997 fidelity provide a clear engineering roadmap for the next generation of neutral atom quantum processors to achieve fault tolerance.