Quantifying surface losses in superconducting aluminum microwave resonators

Abstract

The recent realization of millisecond-scale coherence with tantalum-on-silicon transmon qubits showed that depositing the Al/AlOx/Al Josephson junction in a high purity, ultrahigh vacuum environment was critical for achieving lifetime-limited coherence, motivating careful examination of the aluminum surface two-level system (TLS) bath. Here, we measure the microwave absorption arising from surface TLSs in superconducting aluminum resonators, following methodology developed for tantalum resonators. We vary film and surface properties and correlate microwave measurements with materials characterization. We find that the lifetimes of superconducting aluminum resonators are primarily limited by surface losses associated with TLSs in the 2.7 nm-thick native AlOx. Treatment with 49% HF removes surface AlOx completely; however, rapid oxide regrowth limits improvements in surface loss and long term device stability. Using these measurements we estimate that TLSs in aluminum interfaces contribute around 27% of the relaxation rate of state-of-the-art tantalum-on-silicon qubits that incorporate aluminum-based Josephson junctions.

Figures (4)

• Figure 1: Diagnosing sources of loss in superconducting resonators. a) Top: Optical image of chip containing eight $\lambda$/4 CPW resonators ranging from 2 to 16 $\mu$m in pitch capacitively coupled to a microstrip feedline. Scale bar is 500 $\mu$m. Bottom: Scanning electron microscope image showing sidewall profile of 200 nm Al deposited on c-plane sapphire deposited at 25$\degree$C and fabricated using wet etch. The sidewall is vertical and the film has a lumpy texture. Scale bar is 400 nm. b) Pipeline for measuring materials-related losses using iterative resonator measurement and materials characterization feedback. Microwave measurements provide loss tangents associated with different loss mechanisms, while structural and chemical composition characterization provides candidates for sources of loss. Combined, these measurements allow for estimation of the quantitative contribution of each surface and contaminant to the loss. The cartoon cross section of the superconducting resonator (top) consists of substrate (blue), metal (green), a thin native oxide (yellow), photoresist contamination residue after fabrication (orange), and adventitious hydrocarbons from the atmosphere (red). The metal-substrate (MS) interface and substrate-air (SA) interfaces are represented by a continuous black line. These colors extend to those in the bar graph in the loss budgeting panel.
• Figure 2: Aluminum film characterization. a) XRD patterns of three aluminum thin films on c-Al$_2$O$_3$ deposited at -72$\degree$C (brown), 25$\degree$C (green) and 200$\degree$C (blue). Dotted lines show Al $<111>$ (2$\theta$ = 38.5$\degree$) and c-Al$_2$O$_3$$<0006>$ (2$\theta$ = 41.7$\degree$) crystal directions. b) AFM of 200 nm Al film on c-Al$_2$O$_3$ deposited at -72$\degree$C with S$_q$ = 5.11 nm and average grain size of 15.63 nm. c) AFM of 200 nm Al film on c-Al$_2$O$_3$ deposited at 200$\degree$C with S$_q$ = 1.19 nm and average grain size of 87.34 nm. d) AFM of 200 nm Al film on $<100>$ Si deposited at 25$\degree$C with S$_q$ = 0.5 nm and average grain size of 12.58 nm. e) TEM of the interface of Al 200 nm film deposited at -72$\degree$C with c-Al$_2$O$_3$. Scale bar is 5 nm. f) TEM of the interface of Al 200 nm film deposited at 200$\degree$C with c-Al$_2$O$_3$. Scale bar is 5 nm. g) TEM of the interface of Al 200 nm film deposited at room temperature with $<100>$ Si. Scale bar is 5 nm. h) Resistivity as a function of temperature for films deposited at -72$\degree$C (brown), 25$\degree$C (green) and 200$\degree$C (blue).
• Figure 3: Aluminum resonator microwave characterization. a) Power and temperature dependence of 10 $\mu$m pitch CPW resonator. High power (-10 dB, red) decreases by 10 dB until the lowest power (-90 dB, purple). Three regimes are seen: 1/T dependence until 25 mK, TLS saturation and increase until 125 mK, and quasiparticle rolloff at 150 mK. Scale bar is 500 $\mu$m. b) Q$_\mathrm{{TLS,0}}$ as a function of surface participation ratio. Surface loss tangents are fitted according to surface treatment (tantalum surface loss tangents have been added from Ref. crowley2023disentangling); shaded region indicates one standard deviation on fit parameters. From the top, fitted loss tangents are : Ta treated with buffered oxide etch (gray), Ta treated with piranha solution (black), HF-treated Al on sapphire or silicon (green), H$_{2}$O$_{2}$-treated Al on sapphire (purple), solvent-treated Al on sapphire (blue), and solvent-treated dry etch Al on sapphire, chlorine chemistry (brown). Red points represent HF-treated resonators remeasured after 3 months of oxide and hydrocarbon saturation.
• Figure 4: Aluminum surface characterization using X-ray Photoelectron Spectroscopy. a) Example fit of Al2p peak for a film with 2.69 $\pm$ 0.07nm of native oxide. Each oxidation state (0, 3+, interface) has a 1/2 (dotted) and 3/2 (solid) spin orbital contribution. Using the intensity of oxide peaks to the metal peak, we calculate the oxide thickness using the Strohmeier equation (Online Resource S6). Plot is normalized to intensity. b) Oxide thicknesses calculated for a native oxide on aluminum film (green) as a function of time. The fitted line (blue) shows a linear regime with rapid oxide growth kinetics for roughly the first 12 hours, then an intermediate regime until 24 hours after exposure to air with slower, but still relatively fast growth, followed by a logarithmic dependence which slowly saturates the oxide. Points in red show oxide thicknesses measured from devices. Inset: The same data plotted on semilog axes.c) Al2p XPS intensity as a function of binding energy. H$_2$O$_2$ and solvent-cleaned devices show 2.66 $\pm$ 0.07 and 2.69 $\pm$ 0.07 nm of oxide growth, respectively, while the HF-cleaned device shows 1.87 $\pm$ 0.05 nm of growth. Plots are normalized to intensity, and then scaled so the Al2p peak height is set to 1 to more easily see differences in oxide thickness. All spectra are shifted to the Al$^0_{3/2}$ = 72.6 eV. d) C1s XPS intensity as a function of binding energy normalized to the corresponding intensity of the Al2p peak. The solvent-cleaned device shows a greater intensity than H$_2$O$_2$, which shows a greater intensity than the HF-cleaned device. e) Numerical comparisons of the integral intensities of the C1s peak across devices and films at each of the growth regimes. f) O1s XPS intensity as a function of binding energy normalized to the corresponding intensity of the Al2p peak. The HF-treatment produces a double peaked structure with a small increase in binding energy, signifying a slight decrease in interfacial oxide between the metallic and 3+ aluminum states. We normalize the peak positions for both Al2p and O1s to the C1s peak at 284.8 eV and peak intensities for both C1s (Fig. \ref{['fig:fig3']}d) and O1s (Fig. \ref{['fig:fig3']}f) to the corresponding Al2p intensity to accurately compare carbon contamination between different post-fabrication cleaning treatments.