Characterizing the Dark Count Rate of a Large-Format MKID Array

We present an empirical measurement of the dark count rate seen in a large-format MKID array identical to those currently in use at observatories such as Subaru on Maunakea. This work provides compelling evidence for their utility in future experiments that require low-count rate, quiet environments such as dark matter direct detection. Across the bandpass from 0.946-1.534 eV (1310-808 nm) an average count rate of $(1.847\pm0.003)\times10^{-3}$ photons/pixel/s is measured. Breaking this bandpass into 5 equal-energy bins based on the resolving power of the detectors we find the average dark count rate seen in an MKID is $(6.26\pm0.04)\times10^{-4}$ photons/pixel/s from 0.946-1.063 eV and $(2.73\pm0.02)\times10^{-4}$ photons/pixel/s at 1.416-1.534eV. Using lower-noise readout electronics to read out a single MKID pixel we demonstrate that the events measured while the detector is not illuminated largely appear to be a combination of real photons, possible fluorescence caused by cosmic rays, and phonon events in the array substrate. We also find that using lower-noise readout electronics on a single MKID pixel we measure a dark count rate of $(9.3\pm0.9)\times10^{-4}$ photons/pixel/s over the same bandpass (0.946-1.534 eV) With the single-pixel readout we also characterize the events when the detectors are not illuminated and show that these responses in the MKID are distinct from photons from known light sources such as a laser, likely coming from cosmic ray excitations.