Silicon Qubits Hit 99.99% Accuracy While Three Run at Once, RIKEN Team Reports in Nature Physics

Silicon quantum bits have a well-known weakness: they get sloppy when several are controlled at the same time. A team in Japan has now shown how to keep them accurate, according to a study published in Nature Physics.

Researchers at the RIKEN Center for Emergent Matter Science and collaborators controlled a five-qubit silicon array through a single shared microwave line. They kept single-qubit gate fidelity above 99.99% while up to three qubits ran simultaneously. Fidelity measures how accurately a quantum operation is carried out.

Silicon spin qubits are attractive because they are compatible with the semiconductor manufacturing used to make today's computer chips. In these devices, information is stored in the spin of an electron trapped in a silicon substrate, and microwave pulses flip that spin. Useful quantum computers will need to operate many qubits at once without losing accuracy, and that is where silicon devices have struggled.

The RIKEN team first pinned down the culprit. The main source of degradation, they found, is the AC Stark shift, an effect induced by the microwave control signals themselves. The shift nudges a qubit's effective energy levels, which changes the phase of its quantum state and makes its operation less precise. When several qubits are driven together, those shifts interfere with the intended control of each one.

To counter it, the team developed a mitigation protocol based on phase calibrations. "Rather than requiring an entirely separate calibration procedure for every possible combination of qubits, our method uses calibrations between pairs of qubits to compensate for the unwanted phase shifts," said Yi-Hsien Wu, the first author, who carried out the experiments as part of his Ph.D. research. That is what makes the approach scalable.

Using an optimized device design and carefully calibrated microwave pulses, the team measured single-qubit fidelities well above 99.99% on each of the five qubits when operated individually, with the best approaching 99.999%. With the protocol in place, fidelities stayed above 99.99% when up to three qubits operated at once.

When all five ran in parallel, fidelity settled at about 99.9%, a level the researchers describe as close to the practical requirements of fault-tolerant quantum computing. The remaining loss was attributed to drive-induced decoherence, a loss of quantum coherence caused by the higher microwave power needed to drive several qubits at once. Unlike the phase shifts, the new calibration does not fix that effect.

"This is a limitation, but further improvements to device design could help mitigate this source of error," said Leon Camenzind, a research scientist at RIKEN and co-first author who supervised the experiments. He called the result "an important step toward scaling up silicon-based quantum processors."

The work addresses a narrow but central problem. It does not yet show a full quantum processor, and the five-qubit array is small. But it identifies a specific, fixable error source and shows that high accuracy does not have to be sacrificed when more qubits run together.