Comparison of Nb and Ta Pentoxide Loss Tangents for Superconducting Quantum Devices

TL;DR

This work targets dielectric losses that limit superconducting qubit coherence by directly comparing Nb2O5 and Ta2O5 pentoxides deposited on CPW resonators. Using PLD, comprehensive structural and chemical analyses (XRR, XPS, XANES, STEM, EELS) confirm that the deposited oxides are chemically homogeneous pentoxides closely resembling native Nb and Ta pentoxides. Microwave measurements at ~10 mK reveal a linear, thickness-dependent TLS loss: Nb2O5 contributes $3.5\times 10^{-6}$ per nm while Ta2O5 contributes $2.3\times 10^{-6}$ per nm, with PI losses being comparatively small ($0.10\times 10^{-6}$ and $0.20\times 10^{-6}$ per nm, respectively). The findings show Nb2O5 has higher dielectric loss than Ta2O5, suggesting mechanisms beyond suboxide content (potentially hyperfine coupling) influence TLS losses and guiding oxide choice for improved qubit performance.

Abstract

Superconducting transmon qubits are commonly made with thin-film Nb wiring, but recent studies have shown increased performance with Ta wiring. In this work, we compare the resonator-induced single photon, millikelvin dielectric loss for pentoxides of Nb (Nb2O5) and Ta (Ta2O5) in order to further understand limiting losses in qubits. Nb and Ta pentoxides of three thicknesses are deposited via pulsed laser deposition onto identical coplanar waveguide resonators. The two-level system (TLS) loss in Nb2O5 is determined to be about 30% higher than that of Ta2O5. This work indicates that qubits with Nb wiring are affected by higher loss arising from the native pentoxide itself, likely in addition to the presence of suboxides, which are largely absent in Ta.

Figures (7)

• Figure 1: Pentoxide electron microscopy characterization. (a) Schematic of control and pentoxide-coated resonator samples. (b-c) Representative HAADF-STEM images of the 30 nm (b) Nb2O5-coated and (c) Ta2O5-coated Nb films on Al2O3 substrates. (d-e) EDS maps corresponding to (b) and (c), respectively, confirm the elemental composition of all layers. (f-g) High-magnification oxygen K-edge EELS spectrum images of the deposited (f) Nb2O5 and (g) Ta2O5. (h-i) ELNES of the oxygen K-edge of the deposited (h) Nb2O5 and (i) Ta2O5, where the spectrum is summed over the entire deposited pentoxide region. The t2g-eg crystal field splitting is labeled. (j-k) EELS line profiles of the oxygen K-edge across the oxide region, including the native Nb oxide and the deposited (j) Nb2O5 and (k) Ta2O5, corresponding to the color codes in (f) and (g), respectively.
• Figure 2: Pentoxide X-ray reflectivity and photoelectron characterization. (a-b) X-ray reflectivity (XRR) patterns of resonator chips with (a) deposited $\mathrm{Nb_2O_5}$ and (b) deposited $\mathrm{Ta_2O_5}$. (c-d) Extracted electron density profiles, indicating the thickness of the deposited oxide (green), the native Nb oxide (purple), and, for the 3 nm $\mathrm{Ta_2O_5}$, the combined oxide layer (maroon), as measured by XRR. (e-f) X-ray photoelectron spectroscopy (XPS) of resonator chips with (e) deposited $\mathrm{Nb_2O_5}$ and (f) deposited $\mathrm{Ta_2O_5}$, measuring the Nb3d and Ta4f regions, respectively. Only peaks corresponding to the 5+ oxidation state of Nb and Ta are observed for the respective deposited oxides. For the 3 nm $\mathrm{Ta_2O_5}$ sample, additional Nb4p peaks are observed, corresponding to the underlying Nb film of the resonator.
• Figure 3: X-ray absorption near-edge structure (XANES) spectra of the (a) Nb K edge for amorphous $\mathrm{Nb_2O_5}$ films and (b) Ta L3 edge for amorphous $\mathrm{Ta_2O_5}$ films, both deposited by pulsed laser deposition on Si(100) substrates. The position of the edge was determined to be the maximum in the first derivative of the spectrum. In both spectra, the position of the respective edge feature is constant, indicating a similar average chemical state of the Nb or Ta atoms, regardless of the thickness of the film.
• Figure 4: Microwave measurements of pentoxide-coated superconducting resonators at a temperature of $\sim$10 mK. (a) Power-dependent loss tangent. Total resonator loss tangent $\delta$ as a function of average photon number $\langle n_{ph} \rangle$ in the resonator. Points are obtained by fitting complex $S_{21}$ data with the diameter correction method, Khalil2012 with fit uncertainty shown as error bars. Lines represent fits to the resonator TLS loss tangent equation, McRae2020 and 95$\%$ confidence prediction curves are shown as shaded regions. (b) Linear fits of TLS and PI loss tangents for Nb and Ta pentoxide-coated samples. Blue circle represents the Nb on sapphire control sample, green circles represent Nb pentoxide-coated samples, and purple squares represent Ta pentoxide-coated samples. Pentoxide top surface thickness $t$ for each sample is determined by STEM and shown in the supplementary material.
• Figure 5: HAADF-STEM images of resonator sidewalls for each of the six devices with nominally 3, 15, and 30 nm deposited $\mathrm{Nb_2O_5}$ and $\mathrm{Ta_2O_5}$, as described in the main text. The thicknesses of the deposited oxides measured from these images are additional corroboration of the thickness values extracted from XRR in Fig. \ref{['fig:XRR_eDensity_XPS']}, and serve as a more accurate representation of the overall volume of the pentoxides when measuring and extracting the loss tangents in Fig. \ref{['fig:device']}. Variations in the measured pentoxide thicknesses along the MA-top and SA interfaces are due to the roughness of the Nb film, whereas variations along the sidewall are representative of gradual changes in the actual deposited thickness due to self-shadowing effects in the PLD.