Two-Level System Microwave Losses in Chemically Pure Bulk Niobium Oxide Samples
Vishal Ganesan, Jiankun Zhang, Drew G. Wild, Alexey Bezryadin · 2026
Read Two-Level System Microwave Losses in Chemically Pure Bulk Niobium Oxide Samples by Vishal Ganesan, Jiankun Zhang, Drew G. Wild, Alexey Bezryadin (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-06668.txt.
Introduction
Vishal Ganesan and co-authors measure microwave losses in niobium oxide in 2026. Two-level systems in amorphous oxides limit superconducting qubits and cavities. These tiny defects absorb microwave energy and shorten coherence times.
Niobium resonators are widely used in quantum science experiments today. Niobium's natural oxide layer contains various types of oxides mixed together. Their relative contributions to loss had not been clearly distinguished. The team studies chemically pure bulk niobium oxide samples to separate them. Pure samples let one oxide type speak at a time. Quantum hardware builders need exactly this attribution of loss sources. Reducing the worst oxide could extend qubit lifetimes in future devices.
Catalog record
| CREATOR | Vishal Ganesan, Jiankun Zhang, Drew G. Wild, Alexey Bezryadin |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-06668
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Two-Level System Microwave Losses in Chemically Pure Bulk Niobium Oxide Samples Vishal Ganesan, Jiankun Zhang, Drew G. Wild, Alexey Bezryadin Abstract Losses from two-level systems (TLS) associated with amorphous oxides remain one of the primary limitations to the performance of superconducting qubits and microwave cavities. Niobium resonators are widely used in quantum science experiments, yet niobium’s natural oxide layer contains various types of oxides whose relative contributions to TLS loss have not been clearly distinguished. Here, we use a superconducting 3D microwave cavity to measure chemically pure oxides Nb 2 O 5 \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} and NbO 2 \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} . Using this approach, we directly compare the loss characteristics of Nb 2 O 5 \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} and NbO 2 \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} . Our measurements show that the Nb 2 O 5 \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} oxide exhibits losses which have the power and temperature behavior expected for TLS. Moreover, the measurements agree with existing theoretical models. Analogous measurements performed on NbO 2 \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} do not show any detectable TLS loss signatures. These results provide direct experimental evidence that Nb 2 O 5 \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} is the dominant TLS host in niobium resonators and establish a general framework for separating oxide-specific dissipation channels. I Introduction Superconducting microwave resonators are essential components in quantum information processing, particle accelerator technology, and searches for new physics via dark matter experiments. Their performance is characterized by the intrinsic quality factor ( Q_{\text{i}} ), which limits photon lifetime and coherence 20 ; 17 . At low temperatures and excitation powers, dielectric losses arise from microscopic two-level systems (TLS) in amorphous materials, such as native oxides or dielectric interfaces. These TLS limit the achievable Q_{\text{i}} and consequently the device performance 21 ; 27 . Despite numerous mitigation strategies, including substrate removal 2 , oxide passivation 6 , and surface treatments 15 , TLS losses remain a persistent issue. For niobium, a common superconducting cavity material, TLS losses are generally attributed to its native oxide layers that form spontaneously in air. The niobium oxide system is arranged in a stack: metallic Nb covered by suboxides ( NbO , \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} ) and capped by an amorphous pentoxide ( \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} ) 13 ; 22 . Although previous studies have suggested that the outer \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} layer dominates TLS losses 28 , direct experimental verification has been hindered by the difficulty of isolating individual oxide phases. To distinguish their individual contributions, we employ a 3D superconducting Nb cavity into which a bulk amount of either \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} or \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} , purchased from a commercial supplier of pure chemicals, is inserted at the electric-field antinode of the fundamental resonant mode. Thus, we can measure the quality factor of the cavity versus temperature and microwave power. The results obtained on pure \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} are in agreement with the existing TLS models. At the same time, measurements on pure \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} do not show any dependence on power. Thus, we present direct evidence that only \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} is responsible for the TLS effect, while \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} is not. II Experimental Methods Both oxide powders ( \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} and \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} ) were commercially obtained from Sigma-Aldrich with stated purity exceeding 99.9%. Approximately 50 mg of powder was weighed and mixed with one drop ( \sim 200 mg) of clear nail polish to form a uniform viscous suspension. The resulting mixture was applied to a 10 mm \times 10 mm a-plane sapphire substrate, used to support the oxide load, with a clean toothpick producing a slug-shaped deposit at the center of the sapphire substrate. Sapphire provides a low-loss crystalline dielectric background, ensuring that observed dissipation originates primarily from the oxide film. The sapphire substrates were ultrasonically cleaned in acetone and isopropanol (IPA) prior to oxide deposition. The resulting sample contained about 20 mg of oxide. The nail polish acted as a mechanical binder and inert host that preserves the oxide microstructure and prevents particulate contamination of the cavity interior. The coated substrates were left to dry under ambient laboratory conditions for 24 hours to ensure complete solvent evaporation. Once dried, the samples were mounted into a rectangular niobium cavity, with GE varnish securing the substrates within the machined groove at the electric field antinode for the fundamental resonant cavity mode, as shown in Fig. 1 . Samples are labeled by the oxide type (Nb02 or Nb25 for \text{NbO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} and \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} respectively), mixture medium type (NP for nail polish), and version number. The one exception is the first sample created with a label ”AP” standing for the cryogenic grease Apiezon (ex. Nb25_NP3 indicates \text{Nb}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}}\text{O}{\vphantom{\text{X}}}_{\smash[t]{\text{5}}} , mixed with nail polish, being the 3rd sample made). Before installation, the Nb cavity was cleaned by sonication in acetone and IPA for five minutes each, followed by a vacuum bake at 400\text{\,}\mathrm{\SIUnitSymbolCelsius} in flowing argon to remove surface oxides and adsorbates. This annealing procedure is known to improve Q_{\text{i}} and reduce TLS-related losses 25 ; 30 . The assembled cavity was anchored on the cooling stage of our He-3 cooler with a base temperature of 390 mK, and transmission ( S_{21} ) measurements were performed using a Keysight P9374A vector network analyzer (VNA). VNA microwave power was varied to achieve an average power resonantly circulating, P_{\text{circ}} (to be discussed below), in the cavity from about -75 to +35 dBm (corresponding to a power resonantly applied at the cavity SMA connector jack, P_{in} , from -110 to 0 dBm). Note that our system was equipped with attenuators anchored at different stages of the cooling system. The total attenuation of the input line was 18 dB and the total attenuation of the output line was also 18 dB. Figure 1 : Sample Nb25_NP5 on a sapphire substrate installed in the Nb cavity. The transmission S_{21} near resonance is modeled as a complex Lorenztian given by 16 S_{21}(f)=\frac{Q_{\text{L}}/|Q_{\text{e}}|e^{i\phi}}{1+2iQ_{\text{L}}\frac{f-f_{0}}{f_{0}}}, (1) where f is the frequency, f_{0} is the resonance frequency, \phi accounts for impedance mismatch, Q_{\text{L}} is the loaded quality factor, |Q_{\text{e}}| is the external coupling quality factor, and the intrinsic quality factor Q_{\text{i}} is obtained from the fit parameters Q_{\text{L}} and Q_{\text{e}} using the formula \frac{1}{Q_{\text{i}}}=\frac{1}{Q_{\text{L}}}-\frac{1}{Q_{\text{e}}} . Each resonance was calibrated by removing (mathematically) the effects of line attenuation and amplification, then fit with Eq. 1 to both the real and imaginary components of S_{21} (see Fig. 2 ) at each power applied by the VNA. For low powers, many scans are averaged to improve the signal-to-noise ratio. This circular fit allows us to obtain accurate values for the intrinsic quality factor 26 . (a) (b) Figure 2 : Representative complex Lorentzian fit to the transmission coefficient of the cavity with the sample Nb25_NP5 installed, at P_{\text{circ}}=-45 dBm. (a) This plot shows S_{21} magnitude as a function of frequency. (b) This plot shows the corresponding complex-plane circle representation of S21 in the Argand plane. The fit to Eq. 1 are shown by solid black lines. This plot is used to extract Q_{\text{i}} and Q_{\text{e}} . Standard errors for the frequency and the quality factors obtained from this circular fit are several orders of magnitude below the mean values for these quantities. III Results III.1 Control Measurements In Fig. 3 we show control measurements of Q_{\text{i}} versus VNA output power for the empty cavity and the cavity loaded with a bare sapphire substrate. This figure shows