Quantum Gate Hits 6.8 Nanoseconds With Better Than 99.98% Accuracy
German researchers cut the time of a single-qubit operation to a fraction of the usual speed by suppressing an overlooked error that appears when microwave pulses get strong.
Researchers from Jülich, Aachen and Cologne have performed a quantum operation on a qubit in just 6.8 nanoseconds, with an accuracy above 99.98%. Operations like that typically take several tens of nanoseconds in superconducting quantum computers. The results were published in Nature Communications.
For scale, 6.8 nanoseconds is about how long light takes to travel two meters in a vacuum. In that span, teams at the Peter Grünberg Institute for Functional Quantum Systems (PGI-13) and for Quantum Control (PGI-8) at Forschungszentrum Jülich carried out elementary computing operations on a single qubit.
Speed matters because qubits are fragile and hold their quantum states only briefly. The faster each operation runs, the more of them fit inside that window, which is crucial for complex algorithms and for the resource-heavy process of error correction. But there is a catch. The shorter the microwave pulse used to switch a qubit, the stronger it has to be, and stronger pulses raise the risk of unwanted excitation. Superconducting qubits have more than the two energy states used for computation, and a hard push can knock them into higher levels.
The researchers use an elevator analogy. A lift meant to carry its load only between the ground floor and the first floor can overshoot if the load is raised too briskly. The team found that strong, fast pulses can push information not just one floor too far but two or three, through so-called multiphoton transitions, an effect previously overlooked.
Those transitions are especially troublesome because they can persist across many computational steps and degrade quantum error correction. Working together, the experimental group of Yuan Gao and Professor Rami Barends and the theory group of José da Costa Jesus, Francisco Cárdenas-López and Professor Felix Motzoi developed a way to suppress them by shaping the control pulses with great precision.
The method, called R2D, acts like a frequency filter, removing the frequency components that drive the unwanted transitions while leaving the desired operation intact. It builds on DRAG, the correction technique that has been standard for controlling qubits for more than a decade. Compared with that standard, R2D reduces the probability of unwanted multiphoton transitions by a factor of more than 20.
The test used a single superconducting transmon qubit fabricated at the Helmholtz Nano Facility at Jülich, as part of the large collaborative project QSOLID. Two fundamental operations, each lasting 6.8 nanoseconds, were measured at accuracies of 99.981% and 99.989%. The chance of the qubit slipping into a higher energy state stayed below 0.002% per operation.
There is a limit to the claim. So far the method has only been demonstrated on one isolated qubit, and further experiments are needed to find out whether it works as well in larger quantum processors and during operations involving two qubits. Still, the team says the work sets a new benchmark for quantum gates and shows that extremely short switching times and high precision can be achieved together.
