The Signal
Researchers report bosonic qubit coherence beyond break-even
A Nature Physics paper reports that an autonomous error-correction protocol extended a bosonic logical qubit’s coherence time to 1.04 times that of the system’s best physical qubit. The experiment used engineered drives and dissipation to transfer photon-loss information to ancilla qubits and reset them without repeated measurements or classical feedback.

A logical qubit passes its physical benchmark
A bosonic logical qubit retained coherence for 1.04 times as long as the best physical qubit in the same system, according to a Nature Physics paper published on 16 November. The experiment used an autonomous quantum error-correction protocol built from engineered drives and dissipation, rather than repeated measurement and classical feedback. The result is a beyond-break-even demonstration: the encoded logical qubit lasted longer than any single physical qubit in the test system.
The authors implemented the protocol with a binomial code stored in a bosonic mode. Their paper says the approach is derived from quantum-channel simulation and is, in principle, applicable to arbitrary recovery maps; the reported experiment, however, used the binomial code. Lida Sun, Yifang Xu and Yilong Zhou are listed as equal contributors.
Correction through ancillas
In the protocol, photon loss moves the encoded state into an error subspace. A recovery operation transfers information about that error to an ancilla qubit, while a second ancilla receives the resulting excitation and is reset. This sequence removes error information without measuring the encoded logical state, according to the paper.
The experimental system used three three-dimensional microwave cavities and four transmon qubits. One cavity stored the logical qubit; ancillas handled the transfer and reset steps. The paper describes this separated reset process as a way to avoid unwanted back-action on the stored state.
Measured performance
The researchers let the logical state evolve freely for 220 microseconds during each correction cycle, a period chosen for their experimental system. After one complete cycle, the state fidelity was 87.6%; in a separate process-fidelity measurement, the corrected logical qubit reached 89.1% after the same interval.
Across 20 correction rounds, a fit to process-fidelity decay gave the logical qubit a coherence time of 1,627 microseconds, compared with 629 microseconds for the unprotected logical qubit. The best physical-qubit reference was 1,565 microseconds, leaving the protected logical qubit at 1.04 times that benchmark.
Where errors remain
The paper’s error analysis identifies finite transmon coherence and residual multi-photon loss as the main performance limits. It also breaks down contributions from recovery operations, excitation transfer, and ancilla thermal and reset errors. The authors say the protocol can be extended to additional recovery operations and higher-dimensional bosonic codes, while the demonstrated result remains specific to their tested system and encoding.
Sources
Related stories
- Nature Physics paper reports 18% longer autonomous qubit lifetime
- Shaghaghi Sets Out Six-Step Z-Rotation Protocol for Bosonic Qubits
- Researchers synthesize four-circuit fault-tolerant ancilla preparation
- ETH Zurich team teleports logical states on a 41-qubit circuit
- Researchers propose fault-tolerant atom arrays with global control