Silicon Spin Qubits Hit Fault-Tolerance Threshold While Running in Parallel

Quantum computers promise to transform everything from cryptography to drug discovery, but only if their qubits can be controlled with extraordinary precision — and, crucially, all at the same time. A team of researchers working on silicon spin qubits has now taken a major step toward that goal, demonstrating simultaneous high-fidelity single-qubit gates across a five-qubit array, with fidelities that hold above the widely targeted fault-tolerance threshold of 99.9 percent even when all five qubits are driven in parallel.

The work, published in Nature Physics, addresses one of the most stubborn bottlenecks in scaling semiconductor quantum processors. Silicon spin qubits are attractive because they are tiny — each one is essentially a single electron trapped in a quantum dot — and because they can, in principle, be manufactured using the same industrial processes that produce ordinary computer chips. But as more qubits are packed onto a chip and driven simultaneously, their performance typically degrades, a problem known as crosstalk.

The researchers, led by Yi-Hsien Wu and Leon C. Camenzind at RIKEN’s Center for Emergent Matter Science, together with collaborators in Taiwan, Switzerland, and the Netherlands, identified the dominant culprit behind this degradation: microwave-drive-induced a.c.-Stark shifts. When a microwave pulse is applied to one qubit through a shared control line, neighbouring qubits are exposed to off-resonant fields that subtly shift their resonance frequencies, introducing phase errors that accumulate and corrupt gate operations.

What makes this discovery particularly significant is that the team did not simply identify the problem — they solved it with a remarkably efficient calibration protocol. Instead of characterising every possible combination of simultaneous drives, which grows exponentially with qubit number, the researchers showed that the crosstalk phase shifts can be measured pairwise and then summed linearly. For a system of N qubits, this reduces the calibration overhead from an intractable 2^N minus N minus 1 combinations to just N times N minus 1 pairwise measurements.

The device itself is a five-qubit array fabricated in isotopically purified silicon-28 enriched to a residual silicon-29 concentration of just 800 parts per million. Three overlapping aluminium gate layers define the quantum dots, while a cobalt micromagnet deposited above the device creates a magnetic field gradient that separates the qubit resonance frequencies by roughly 150 megahertz between neighbours, allowing each qubit to be individually addressed despite sharing a single microwave control line.

Operating each qubit individually with carefully shaped Kaiser-window microwave pulses, the team measured primitive X and Y gate fidelities well above 99.99 percent for all five qubits, with some approaching 99.999 percent. These numbers were obtained through randomized benchmarking, a rigorous statistical protocol that involves applying thousands of randomly selected gate sequences and measuring how quickly the qubit state departs from the ideal outcome.

The Kaiser-window pulse shape is itself a key innovation. Unlike a simple square pulse, which leaks microwave energy into neighbouring qubit frequencies and causes resonant crosstalk, the Kaiser window has an engineered excitation spectrum that suppresses this spectral leakage. The team also had to operate within a careful speed regime: too slow, and charge-noise-induced dephasing dominates; too fast, and microwave-induced decoherence takes over. They settled on a minimum gate time of 83 nanoseconds, where the Rabi frequency still scales linearly with drive amplitude.

When the researchers moved to simultaneous operation, the a.c.-Stark shifts became the dominant error source. For a two-qubit demonstration on qubits Q3 and Q4, the uncorrected crosstalk phase on Q3 was large enough to cap its fidelity at 99.9 percent, even though Q4 remained above 99.99 percent. After applying the pairwise phase and amplitude calibrations, both qubits exceeded 99.99 percent under simultaneous operation. The team then extended the approach to three qubits, again achieving fidelities above 99.99 percent for all three.

Pushing to all five qubits simultaneously, the fidelities settled at or above 99.9 percent — right at the practical fault-tolerance threshold used in surface-code error correction schemes. The researchers note that at a gate time of 500 nanoseconds, the total microwave power matches that of the five-qubit operation at 250 nanoseconds, suggesting that the same approach could extend to roughly ten qubits within the current device architecture.

The implications for scalable quantum computing are substantial. Driving multiple qubits from a single impedance-matched control line dramatically reduces wiring overhead, a critical concern as processors grow toward the millions of qubits envisioned for fault-tolerant machines. The pairwise calibration strategy avoids the exponential explosion of full crosstalk characterisation, and the researchers point out that for sufficiently detuned qubit pairs, the corrections become negligible altogether, further reducing the burden in larger arrays.

There are still challenges ahead. Two-qubit gates, which are essential for universal quantum computation, introduce additional complications when microwave driving is combined with activated exchange coupling, and the team notes that resonantly driven two-qubit gates would require further calibration of conditional phases. Exchange-controlled CZ gates implemented with baseband pulses sidestep this issue. The researchers also suggest that improvements to the micromagnet design — potentially doubling the driving-field gradient — could enable faster gates at lower power, pushing fidelities above 99.999 percent at modest gate speeds and supporting parallel control of even larger qubit numbers.

For now, the demonstration stands as a clear benchmark: high-fidelity parallel control across multiple silicon spin qubits is achievable, and the path to scaling it further is mapped out. As the race to build practical quantum computers intensifies, results like this one show that the humble silicon chip — the workhorse of the digital age — may yet become the foundation of the quantum one.

Subject of Research: High-fidelity parallel control of silicon spin qubits in a five-qubit array

Article Title: Simultaneous high-fidelity single-qubit gates in a spin qubit array

Article References: Wu, Y.-H., Camenzind, L. C., Bütler, P., Jin, I. K., Noiri, A., Takeda, K., Nakajima, T., Kobayashi, T., Scappucci, G., Goan, H.-S., & Tarucha, S. (2026). Simultaneous high-fidelity single-qubit gates in a spin qubit array. Nature Physics. https://doi.org/10.1038/s41567-026-03466-2

Image Credits: AI Generated

DOI: 10.1038/s41567-026-03466-2

Keywords: silicon spin qubits, quantum computing, a.c.-Stark shift, crosstalk, randomized benchmarking, fault tolerance, quantum dots, shared control line, Kaiser-window pulses, quantum error correction, micromagnet, Nature Physics