PRX Quantum (Apr 2022)

Localization and Mitigation of Loss in Niobium Superconducting Circuits

  • M. Virginia P. Altoé,
  • Archan Banerjee,
  • Cassidy Berk,
  • Ahmed Hajr,
  • Adam Schwartzberg,
  • Chengyu Song,
  • Mohammed Alghadeer,
  • Shaul Aloni,
  • Michael J. Elowson,
  • John Mark Kreikebaum,
  • Ed K. Wong,
  • Sinéad M. Griffin,
  • Saleem Rao,
  • Alexander Weber-Bargioni,
  • Andrew M. Minor,
  • David I. Santiago,
  • Stefano Cabrini,
  • Irfan Siddiqi,
  • D. Frank Ogletree

DOI
https://doi.org/10.1103/PRXQuantum.3.020312
Journal volume & issue
Vol. 3, no. 2
p. 020312

Abstract

Read online Read online

Materials imperfections in planar superconducting quantum circuits—in particular, two-level-system (TLS) defects—contribute significantly to decoherence, ultimately limiting the performance of quantum computation and sensing. The identification of specific parasitic layers and their associated loss contributions has, however, proven elusive. Using a combination of x-ray photoemission spectroscopy (XPS) and analytical scanning transmission electron microscopy (STEM), we determine the thickness, chemical composition, and location of the oxides present in niobium-on-silicon coplanar-waveguide (CPW) resonators and quantify their respective contributions to the measured single-photon quality factor (Q). Using selective chemical etching, we reduce first the substrate-air oxide then the metal-air oxide thickness, dramatically reducing both TLS (δ_{TLS}) and non-TLS (δ_{hi}) losses, resulting in a median Q value over 5×10^{6}, with individual devices approaching 6×10^{6}. We find that silicon surface oxides host 70% of TLS losses, with a δ_{TLS}:δ_{hi} loss-density ratio near 11:1. In contrast, niobium surface oxides host 77% of non-TLS losses, uniformly distributed within the oxide layer, with a δ_{TLS}:δ_{hi} loss-density ratio of 3:4 for the superconducting circuits investigated in this work. Only 7% of losses come from other sources, including the niobium-silicon interface, which is sufficiently clean in our devices to allow epitaxial Nb nucleation on Si. As we mitigate surface losses through selective modification of the interface dielectrics, we arrive in a regime where TLS losses are no longer dominant, which will allow other types of losses in superconducting circuits to be investigated in more detail, including nonequilibrium quasiparticles.