Short answer: Niobium superconducting circuits lose microwave energy in the native oxide that grows on their surface. A 2021 Physical Review Applied paper reported that etching the oxide away raised internal quality factors from 1 million to 7 million, a 2024 paper traced loss to oxygen vacancies in niobium pentoxide, and a recent Applied Physics Letters paper reports that a vapor-phase HMDS treatment improved resonator quality and kept it for 60 days. These are resonator and cavity studies. Qubit coherence results are not in these abstracts.
The problem
The 2021 abstract says the coherence of state-of-the-art superconducting qubit devices is predominantly limited by two-level-system defects, found primarily in amorphous interface layers. Reducing microwave loss from these interfaces by surface treatment is key to improving devices.
2021: etching and regrowth
The authors study niobium resonators after removing native oxides with a hydrofluoric acid etch, and investigate how losses return as oxides regrow in air. They find microwave loss reduced by about an order of magnitude: internal quality factors rise from 1 million to 7 million in the single-photon regime when oxides are removed and air exposure lasts no more than 16 minutes. They find that niobium pentoxide is the only surface oxide that grows significantly in the first 200 hours, following the extended Cabrera-Mott growth model. In that time, loss scales linearly with its thickness, with an extracted loss tangent of about 0.01. They say the findings apply to qubits, quantum-limited amplifiers, kinetic inductance detectors and single-photon detectors.
2024: oxygen vacancies
The second Physical Review Applied abstract identifies what the authors call a major source of decoherence in three-dimensional superconducting radio-frequency resonators and two-dimensional transmons made of oxidized niobium: oxygen vacancies in the niobium pentoxide, which drive two-level-system loss. They use sequential in situ vacuum baking on bulk niobium resonators and probe the oxide structure with time-of-flight secondary ion mass spectrometry. They find a nonmonotonic evolution of cavity quality factor that correlates with the interplay of vacancy generation and oxide thinning. They localize the effect to the oxide itself, show a small role for interstitial oxygen diffusing from the niobium below by regrowing the oxide through wet oxidation, and hypothesize that vacancies act as magnetic impurities causing loss. This last point is a hypothesis.
HMDS passivation
The Applied Physics Letters accepted manuscript describes vapor-phase hexamethyldisilazane (HMDS) as a dry surface treatment compatible with standard nanofabrication. In five independent fabrication runs, treated resonators consistently show improved internal quality factor and a narrower spread than paired untreated controls, across the full microwave power range. Electron energy loss spectroscopy shows smaller chemical shifts and shallower oxidation depth in treated samples. After 60 days of ambient storage, treated resonators keep their quality factor. The authors say this gives strong potential for superconducting quantum devices. The text read was the manuscript abstract, so the final published wording may differ.
Limits
The three studies use different devices, from planar resonators to bulk cavities, and treatments, so their numbers are not interchangeable. None reports a qubit lifetime from a treated device.
FAQ
What is the oxide problem?
The native oxide hosts two-level defects, which absorb microwave energy.
How much did etching help?
Quality factors rose from 1 million to 7 million in 2021.
Which oxide matters?
Niobium pentoxide, per the 2021 and 2024 abstracts.
Can chemistry stabilize it?
The HMDS manuscript reports stability over 60 days of ambient storage.
Is this proven in qubits?
Not in these abstracts.
All field notes