Understanding and minimizing resonance frequency deviations on a 4-inch kilo-pixel kinetic inductance detector array

TL;DR

This study investigates resonance frequency deviations in a large LEKID array, demonstrating that capacitor trimming significantly reduces deviations and improves array yield, which benefits future large-scale detector arrays.

Contribution

The paper introduces a capacitor trimming technique that reduces resonance frequency deviations and enhances yield in large LEKID arrays, supported by optical mapping analysis.

Findings

Fractional deviation explained within ±25×10^{-3}

Capacitor trimming reduces deviation by a factor of 14

Mapping yield improved from 69% to 76%, further to 81% after system updates

Abstract

One of the advantages of kinetic inductance detectors is their intrinsic frequency domain multiplexing capability. However, fabrication imperfections usually give rise to resonance frequency deviations, which create frequency collision and limit the array yield. Here we study the resonance frequency deviation of a 4-inch kilo-pixel lumped-element kinetic inductance detector (LEKID) array using optical mapping. Using the measured resonator dimensions and film thickness, the fractional deviation can be explained within $\pm 25\times 10^{-3}$, whereas the residual deviation is due to variation of electric film properties. Using the capacitor trimming technique, the fractional deviation is decreased by a factor of 14. The yield of the trimming process is found to be 97%. The mapping yield, measured under a 110~K background, is improved from 69% to 76%, which can be further improved to 81% after updating our readout system. With the improvement in yield, the capacitor trimming technique may benefit future large-format LEKID arrays.