The Signal
Rhenium Transmon Reaches 407 Microseconds, but Surface Loss Persists
A rhenium transmon reached a time-averaged relaxation time of 407 ± 34 microseconds, while measured surface losses were comparable to tantalum’s. The researchers report no observable native oxide layer on the rhenium at the nanometer scale.
Quantum.world Team · Written with AI · fact-checked ·
A rhenium transmon reached a 407-microsecond relaxation time, but the absence of an observable native oxide layer did not deliver a surface-loss advantage over tantalum. The result, in a paper accepted by PRX Quantum on October 29, comes from five devices built on sapphire and places rhenium among competitive superconducting-qubit materials without establishing a performance lead.
One device reached 407 microseconds
The best-performing device averaged a relaxation time, T1, of 407 ± 34 microseconds at 5.06 GHz over measurements spanning roughly 24 hours. T1 measures how long an excited qubit takes to lose energy; it is not a measure of every form of coherence. The other four devices recorded averages of 297, 277, 201 and 205 microseconds.
Across all five transmons, the mean internal quality factor was 9.1 million, corresponding to an average T1 of 289 ± 80 microseconds at 5 GHz. The devices used rhenium capacitor pads on sapphire and aluminum Josephson junctions. The researchers measured them at 20 millikelvin, with each device’s relaxation time sampled over about a day to capture fluctuations.
No native oxide, but surface losses persist
Rhenium drew interest because it suppresses formation of a native oxide under ambient conditions. The researchers used transmission electron microscopy and energy-dispersive X-ray spectroscopy to examine the metal-air interface; they report no native oxide layer at the nanometer scale. They tested whether this material property could reduce dielectric loss at the interface, a suspected source of energy loss in superconducting circuits. Their measurements, however, found rhenium and tantalum surface losses to be comparable.
“Our loss characterization study reveals rhenium and tantalum surface losses to be comparable, demonstrating that limited native oxide formation does not guarantee reduced surface loss.”
Yanhao Wang, Suhas Ganjam, Ishan Narra, Luigi Frunzio and Robert J. Schoelkopf, the paper’s authors, write.
The loss budget also points to surfaces as the main area for improvement. In the authors’ analysis, the aluminum region near the junction has a disproportionately large role relative to its small footprint, while the rhenium-aluminum interface contributes little. The measured device-to-device and time-dependent variation further means the 407-microsecond result should be read as one device’s time-averaged performance, not as the typical result across the sample.
Rhenium competes with tantalum, not beyond it
The authors say their rhenium transmons are competitive with, but do not exceed, tantalum-based transmons. That comparison tempers the headline number: the study demonstrates that rhenium can support long-lived transmons, but does not show that suppressing native oxide alone improves qubit relaxation. The authors suggest surface contamination and material preparation may help explain why the expected oxide-related benefit did not appear.
They also report that the rhenium films cannot withstand the piranha cleaning treatment used for tantalum, and say their alternative sulfuric-acid treatment may leave organic contamination. Testing other cleaning methods is one route they identify for reducing surface loss. For now, the paper’s measured benchmark is a 407 ± 34-microsecond T1 on the best rhenium device, alongside a five-device mean of 289 ± 80 microseconds.