A global fitting algorithm for notch type resonator arrays in MKIDs

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Microwave kinetic inductance detectors (MKIDs) employ arrays of notch-type resonators coupled to a common feedline, enabling frequency-division multiplexing in which the transmission spectrum ( S 21 ) encodes the parameters of all resonators in a single measurement. Accurately extracting individual resonator characteristics—namely, the resonant frequency ( f r ), intrinsic quality factor ( Q i ), coupling quality factor ( Q c ), and asymmetry angle ( ϕ )—remains a major challenge. Traditional single-resonator fitting methods often yield inaccurate results due to severe localized baseline distortions and parameter degeneracies within narrow frequency windows. In this work, we introduce a novel, global fitting algorithm for the scalable and fully automated characterization of large-format resonator arrays. Compared to conventional isolated fitting approaches, our method offers two crucial advantages: (1) it definitively breaks the ambiguity between electrical delay ( τ ) and localized phase asymmetry by simultaneously optimizing over the full wideband spectrum, accurately resolving the delay and dynamic nonlinear baselines; and (2) it introduces a physically-motivated soft shared regularization for Q i . This soft constraint stabilizes the fit by pulling Q i toward a global consensus based on the uniform substrate loss, while still gracefully accommodating natural, fabrication-induced variations. Importantly, this shared regularization inherently forces noise-induced spurious peaks to adopt unphysical Q c values, providing a powerful self-correcting diagnostic that autonomously filters out false candidates. Experimental validation on a 157-pixel MKID array demonstrates highly reliable parameter extraction, yielding a converged Q i ≈ 118   648 ± 40   323 and a log-normal distribution of Q c with median values of 50 720 that matches lithographic design targets. This physically rigorous, detector-agnostic framework overcomes severe baseline ripples and manual threshold tuning, establishing a robust foundation for the rapid, high-throughput characterization of next-generation superconducting arrays.
Funding Information
  • Research Grants Council of Hong Kong
  • General Research Fund (14306323)
  • Collaborative Research Fund (C4012-20E)

This publication has 25 references indexed in Scilit: