Which interface dominates dielectric loss in millisecond transmons?
In plain words
Energy leaks from qubits into defects at several places: metal-air, metal-substrate and substrate-air surfaces, the substrate interior and the junction itself. Which of these limits the best current qubits is not settled.
Precise statement
For Ta- or Nb-based transmons with $T_{1}$ from $\sim 0.3\,\mathrm{ms}$ to above $1\,\mathrm{ms}$ at $\sim 4-5\,\mathrm{GHz}$ on sapphire or high-resistivity Si, determine the loss tangents $\tan \delta_{i}$ and participation ratios $p_{i}$ of the metal-air, metal-substrate, substrate-air, substrate bulk and junction regions, and identify which term dominates $1/Q = \sum_{i} p_{i}\tan \delta_{i}$ plus non-dielectric channels.
What would settle it
Participation-resolved loss measurements on resonators and qubits of varied geometry from one fabrication process, with surface chemistry characterized by x-ray photoelectron spectroscopy and bulk substrate loss measured separately.