Figure 2

SAW TLS measurements (a) The decreased resonant frequency of different SAW modes, and each data set’s fit to Eq. 1. The modes of SAW devices that were argon ion milled with GCIB prior to cooldown had the largest redshift. The devices that had been annealed at 500 \(^{\circ }\)C for 8hr at atmospheric pressure had the next largest redshift, followed by the 5% MgO co-doped device. All other surface treatment steps yielded similar low temperature redshift. (b) Fit \(F \delta ^0_\text {TLS}\) products compared across various surface treatments. Three CLN data sets are plotted in blue. Error bars shown on the first CLN device are defined by measuring a single device multiple times per cooldown and recording changes in the redshift fit. A summary of the mean loss and error from three separately fabricated CLN devices is plotted in the black-outlined blue circle. (c) External and (d) internal quality factors are plotted as a function of mean phonon number for each of the surface processed SAW devices. A zoomed in image of the internal quality factor for the GCIB device is shown in figure (e). The internal quality factors are fit to Eq. 2 up to an average phonon number of \(2.8\times 10^7\), plotted in gray. Beyond that, the \(Q_i\) decreases once again, likely due to heating or other nonlinearities.