Two-Level System Microwave Losses in Chemically Pure Bulk Niobium Oxide Samples

TL;DR

The paper identifies amorphous Nb2O5 as the dominant TLS host in niobium oxide stacks and NbO2 as essentially TLS-free, using a cavity-based approach that loads chemically pure oxide powders into a 3D Nb resonator. The authors demonstrate that Nb2O5 exhibits TLS-like power and temperature dependence, consistent with the standard TLS model and exhibiting TLS–TLS interactions, while NbO2 shows no such losses. This provides direct, oxide-specific dissipation characterization and confirms that removing or thinning the outer Nb2O5 layer can substantially improve resonator quality factors. The work introduces a framework to disentangle oxide-specific dissipation channels, with practical implications for improving superconducting quantum devices.

Abstract

Losses from two-level systems (TLS) associated with amorphous oxides remain one of the primary limitations to the performance of superconducting resonators in quantum information science and precision measurements. Niobium resonators are widely used for these purposes, yet niobium's natural oxide stack 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 \ch{Nb2O5} and \ch{NbO2}. Using this approach, we directly compare the loss characteristics of \ch{Nb2O5} and \ch{NbO2}. Our measurements show that the \ch{Nb2O5} oxide exhibits TLS-like power and temperature dependence. Analogous measurements performed on \ch{NbO2} do not show any detectable TLS loss signatures. These results provide direct experimental evidence that \ch{Nb2O5} is the dominant TLS host in niobium resonators and establish a general framework for separating oxide-specific dissipation channels

Figures (8)

• Figure 1: (a) Control measurements of $Q_\text{i}$ versus microwave power (VNA output power) of the empty Nb cavity (black) and the cavity loaded with the bare sapphire substrate (red). In these measurements there were no Nb oxide sample. The inset shows the cavity with the bare sapphire substrate. (b) Sample Nb25_NP5 on a sapphire substrate installed in the Nb cavity.
• Figure 2: Representative complex Lorentzian fit to sample Nb25_NP5 inside the cavity at circulating power of -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. \ref{['eq:lorentzian']} 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.
• Figure 3: Power dependence of the inverse of the intrinsic quality factor $Q_\text{i}$ for Nb2O5 samples as a function of circulating power (bottom axis) and photon number N (top axis). Solid black lines show the TLS saturation model fit of Eq. (\ref{['eq:TLSmodel']}). Two samples, Nb25_NP3 and Nb25_NP5, were measured twice (indicated by $1^{st}$ and $2^{nd}$) to confirm TLS losses are reproducible after removing the samples, cleaning the cavity, and re-installing the samples.
• Figure 4: Sample Nb25_NP2. (a) Loaded quality factor $Q_\text{L}$, normalized to its value at $P_\text{circ}=-10$ dBm, as a function of circulating power for various temperatures. (b) Inverse $Q_\text{i}$ versus circulating power (bottom axis) and photon number (top axis) with fits to Eq. \ref{['eq:TLSmodel']} shown as solid black curves. The inset highlights a pronounced TLS-induced reduction in $Q_\text{i}$, measured at $1.1$ K as a function of circulating power. This is in contrast to the absence of an obvious corresponding decrease in $Q_\text{L}$.
• Figure 5: Sample Nb25_NP2. Measured resonance frequency shift (red squares) at $P_\text{circ}=-45$ dBm, as a function of temperature. The solid black curve shows the fit to Eq.\ref{['DiGamma']}, extrapolated to wider range outside the measured temperature range.