Jülich team runs 6.8-nanosecond transmon gates above 99.98% fidelity

Researchers from Jülich, Aachen and Cologne report ultrafast single-qubit gates in Nature Communications, using a pulse method that suppresses multiphoton leakage.

Physicists at Forschungszentrum Jülich, working with RWTH Aachen University and the University of Cologne, have performed single-qubit operations on a superconducting qubit in 6.8 nanoseconds while keeping accuracy above 99.98%. The result, published in Nature Communications as Ultrafast single-qubit gates through multi-photon transition removal, relies on a new way of shaping microwave control pulses. According to Jülich's release, such operations usually take several tens of nanoseconds in superconducting machines.

What was announced

Two basic gates were run on a single transmon qubit fabricated at Jülich's Helmholtz Nano Facility, each driven by a 6.8-nanosecond microwave pulse. The measured fidelities were 99.981% and 99.989%, and the chance of the qubit leaking into a higher energy level stayed below 0.002% per operation, the institute says. Quantum Computing Report identifies the gates as X and X/2 and reports that fidelities were measured by interleaved randomized benchmarking. The work was carried out within QSOLID, the German national consortium coordinated by Jülich.

The technical details

Speed matters because qubits hold their states only briefly: faster gates fit more operations, including error-correction cycles, into that window. But shorter pulses must be stronger, and transmons have energy levels beyond the two used for computing. The Jülich team found that strong, fast pulses can push a qubit not just one level too high but two or three levels up through multiphoton transitions, an effect the release says had previously been overlooked and that can persist across computational steps and disrupt error correction.

The new method, R2D, builds on DRAG, the standard pulse-correction technique used for more than a decade. It shapes pulses so that frequency components driving the unwanted transitions are filtered out while the intended operation still happens. Compared with the standard method, R2D cuts the probability of multiphoton transitions by more than a factor of 20, according to the researchers. QCR reports that the remaining gate error is limited by decoherence rather than by control errors.

The experimental work was led by Yuan Gao and Rami Barends at the Peter Grünberg Institute's PGI-13, with theory from José da Costa Jesus, Francisco Cárdenas-López and Felix Motzoi at PGI-8.

Why it matters

Faster gates with no loss of accuracy give error-correction schemes more room before qubits decohere, and suppressing leakage is particularly valuable because leaked states are hard for standard codes to handle.

What to watch

The researchers stress that the technique has so far been tested only on a single isolated qubit. Further experiments must show whether it holds up in larger processors, where crosstalk and crowded frequencies become issues, and during two-qubit operations.