Illustrative image: a superconducting transmon chip patterned at the Minnesota Nano Center — the same qubit hardware family Jülich's team used, not Jülich's own device. Image: Wikimedia Commons / OJB Quantum (made by Onri Jay Benally), CC BY 4.0.
TL;DR: Jülich 6.8 nanosecond quantum gate explained in one line: researchers in Germany have run a single-qubit operation in just 6.8 nanoseconds at over 99.98% accuracy, using a new R2D pulse-shaping method that suppresses the unwanted multi-photon jumps that normally appear when gates are driven this fast.
📌 KEY HIGHLIGHTS
- Record speed, accuracy kept: X and X/2 gates in 6.8 ns each, with fidelities of 99.981% and 99.989% — while typical superconducting gates take several tens of nanoseconds.
- The trick is R2D: "Double Recursive DRAG" pulse shaping cuts unwanted multi-photon transitions by more than 20× compared with the standard DRAG correction used for over a decade.
- Leakage nearly eliminated: the chance of the qubit jumping to a higher, non-computational energy state stayed below 0.002% per gate.
- Peer-reviewed, not a preprint: the work is published in Nature Communications (Y. Gao et al., 2026), from Forschungszentrum Jülich with RWTH Aachen University and the University of Cologne, under Germany's QSOLID project.
- One-qubit caveat: the result is on a single isolated transmon qubit — multi-qubit processors and two-qubit gates are still to be tested.
What happened?
6.8 nanoseconds is how long light takes to travel about two metres in a vacuum. In that sliver of time, researchers at Forschungszentrum Jülich's Peter Grünberg Institute for Functional Quantum Systems (PGI-13) and Quantum Control (PGI-8) have now performed an elementary computing operation on a qubit — with more than 99.98% accuracy. The paper, Ultrafast single-qubit gates through multi-photon transition removal, is published in Nature Communications, and the result was reported widely on 29–30 October 2026.
The team is led on the experimental side by Yuan Gao and Prof. Rami Barends (PGI-13) and on the theory side by José da Costa Jesus, Dr. Francisco Cárdenas-López and Prof. Felix Motzoi (PGI-8), with RWTH Aachen University and the University of Cologne. The qubit was a single superconducting transmon fabricated at the Helmholtz Nano Facility at Jülich, and the work was done as part of QSOLID, Germany's large national quantum-computing project.
Speed matters in quantum computing because qubits forget. A superconducting qubit holds its state for only tens of microseconds before decoherence scrambles it, so every gate must finish as fast as possible to leave room for more operations — and for error correction — before the information decays. The catch, until now, was that pushing gates faster made them sloppier. This result breaks that trade-off, at least for one qubit.
Jülich 6.8 nanosecond quantum gate explained: how R2D works
A transmon qubit is a superconducting circuit whose lowest two energy levels, |0⟩ and |1⟩, store the bit of quantum information. But the circuit has higher levels too — |2⟩, |3⟩ and beyond — which are not part of the computation. A gate is performed by a microwave pulse that rotates the qubit's state, and to go faster you must drive harder. In the strongly driven regime the drive field shifts the qubit's own energy levels, and that opens a forbidden-looking shortcut: a single pulse can now push the qubit two or three "floors" up at once, through direct two-photon transitions (|0⟩ → |2⟩ and |1⟩ → |3⟩) that bypass the usual selection rules.
That leakage is particularly treacherous, the team notes, because it can persist across several computational steps and corrupt error correction — a leaked qubit is not just a wrong answer, it is an answer outside the model the error-correction code assumes. The established fix, DRAG (Derivative Removal by Adiabatic Gate), has been standard practice for more than a decade, but it was designed for gentler driving and misses these multi-photon jumps — an effect the Jülich team says had previously been overlooked.
R2D — Double Recursive DRAG — extends DRAG with a double recursive derivation that notches out the spectral components of the pulse at exactly the multi-photon transition frequencies, like a frequency filter tuned to silence the unwanted jumps. It uses three parameters optimised in place on the device (α12, α02, α13). Paired with it is a new measurement trick, Amplified Leakage Error (ALE), which coherently accumulates tiny leakage signals so the team could measure leakage down to one part in a million and separate real coherent leakage from ordinary thermal excitation.
| Benchmark | Jülich result | Why it matters |
|---|---|---|
| Gate duration | 6.8 ns (X and X/2 gates) | Several times faster than the typical tens of nanoseconds for superconducting gates |
| Average gate fidelity | 99.981% (X) / 99.989% (X/2) | Measured by interleaved randomised benchmarking; errors are now decoherence-limited, not control-limited |
| Leakage per gate | Below 2.0 × 10−5 (under 0.002%) | Leaked states no longer threaten error correction at this scale |
| Improvement over DRAG | Multi-photon transitions cut by more than 20× | The first method to handle the strongly driven regime (Ω/|Δ| = 0.33) |
| Qubit coherence (device) | T1 = 40 µs, T2Echo = 56 µs | Remaining errors come from the environment, not the pulse shaping — the method itself is no longer the bottleneck |
Illustrative image: the Bloch sphere, the standard geometric picture of a qubit's state — a gate is a rotation of this sphere, and leakage is the state escaping it to higher levels. Image: Wikimedia Commons / Smite-Meister, CC BY-SA 3.0.
Why it matters for Europe and India readers
No India-specific result comes with this paper, and this post claims none. The relevance is structural, and it runs in two directions. For Europe, this is a national-programme result: QSOLID is Germany's bid to build a sovereign superconducting quantum stack, and a gate-speed record with a clear, publishable method is exactly the kind of capability such programmes are judged on. It also sharpens the competition among superconducting approaches — IBM, Google and European labs all use transmons, so a better pulse-shaping method is portable knowledge, not a single lab's secret.
For Indian readers, the lesson is about where quantum advantage is actually won. Headlines usually count qubits; this result is about the quality of each operation. India is building quantum capacity under its National Quantum Mission, and the same trade-off — faster gates versus leakage and error correction — will decide whether Indian hardware efforts can run long, reliable computations. The honest framing also matters: this is a single-qubit demonstration, and the team's own caveat is that scaling to multi-qubit chips brings spectator-qubit dephasing, microwave crosstalk and frequency crowding. Readers should treat 6.8 ns as a speed limit proven in isolation, not a processor you can buy. For context on the road to reliable machines, see our coverage of Sandia and Quantinuum's first full map of logical qubit noise, Quantinuum's Helios error-correction record and IBM's quantum L-coupler advance toward fault-tolerant computing.
Should you care?
✅ Verdict: Yes, if you follow quantum computing's road to useful machines — this removes a real obstacle to faster superconducting processors, it is peer-reviewed, and the remaining errors are now set by physics, not by control. ⏳ But keep it in proportion: one isolated qubit is not a processor, and the multi-qubit test — where crosstalk and crowding bite — has not happened yet.
FAQs
What is a transmon qubit? A transmon is a superconducting circuit — a Josephson junction shunted by a large capacitor — whose two lowest energy levels act as the 0 and 1 of a qubit. Introduced in 2007, it is the dominant qubit design at IBM, Google and other superconducting labs because the large capacitor makes it insensitive to charge noise.
What are multi-photon transitions, and why do they hurt quantum gates? When a gate is driven very hard, the microwave pulse can jump the qubit straight to higher energy levels (|2⟩, |3⟩) instead of rotating it within |0⟩ and |1⟩. That "leakage" escapes the computational space, persists across later steps, and breaks the assumptions error-correction codes rely on.
What is the R2D method? R2D (Double Recursive DRAG) shapes the control pulse so that its frequency components at the multi-photon transition frequencies are cancelled out, like a notch filter. It builds on DRAG, the standard correction technique for over a decade, and in Jülich's tests cut unwanted multi-photon transitions by more than 20 times.
What does 99.98% gate fidelity mean? Fidelity measures how close the gate's actual operation is to the ideal one; 99.981% and 99.989% for the X and X/2 gates mean fewer than 2 errors per 10,000 operations from the gate itself. Crucially, Jülich's remaining errors are decoherence-limited — they come from the qubit's environment, not from imperfect pulses.
When will ultrafast gates reach real quantum computers? Not yet known. The method has only been demonstrated on a single isolated qubit, and the team says adapting R2D to multi-qubit chips will need multi-frequency spectral notch filters in automated calibration, plus tests on two-qubit entangling gates. Watch for a multi-qubit demonstration as the next milestone.
Takeaway: Faster quantum gates no longer have to mean sloppier ones — Jülich's R2D method proves 6.8-nanosecond operations can clear 99.98% fidelity, but the multi-qubit test is still to come.
Follow UK & Europe Tech on TechByCountry for tomorrow's update.
📸 Get tech in the last 24 hours on Instagram — follow @_techinlast24hrs for daily posts (9 AM–9 PM) and reels at 12:30 PM & 8:30 PM.
Sources
- Y. Gao et al., "Ultrafast single-qubit gates through multi-photon transition removal", Nature Communications (2026) — peer-reviewed primary source
- Forschungszentrum Jülich press release, "Quantum gates set speed record at 6.8 nanoseconds with over 99.98% accuracy" (Phys.org republication, October 2026) — institutional primary source
- Quantum Computing Report, "Forschungszentrum Jülich Demonstrates 6.8-Nanosecond Single-Qubit Gates" (30 October 2026) — independent trade coverage
Related: princeton cryogenic atom array quantum gates explained — Princeton's cryogenic caesium array runs a full gate set at 99.76% two-qubit fidelity.
Comments
Post a Comment