Electron umbrellas and photon rain: an investigation into several noise and charge effects in CMOS detectors

Abstract

CMOS detectors are becoming widely used in astronomy because of their low noise and high-speed readout. Projects like the upcoming Argus Array, an optical survey instrument consisting of 1,200 28-cm telescopes which will image 8,000 square degrees every minute, depend on the low-noise characteristics of these sensors to reach their desired depth goals. The ability to combine many frames is necessary to achieve these depths; therefore, the behavior of stacked data products from these detectors needs to be verified against the system science requirements. With an average dark current of order 0.01 e−/pixel/second and read noise of 1.1-1.2 e−, the Sony IMX455 and IMX461 sensors meet the performance requirements for Argus science. However, during the characterization of these sensors, after collecting thousands of dark frames in a controlled environment, we identified and analyzed an infrequent but curious phenomenon in the form of errant clusters of pixels. These clusters of pixels, labeled “charge umbrellas” due to their shape, appeared in all cameras that we tested from three different manufacturers. They consist of a primary hot pixel, whose value varies directly with sensor temperature and exposure time, and a surrounding grouping of warm pixels. The severity of the surrounding shape depends on the central pixel, but in the worst cases, a saturated central pixel has 10-15 pixels immediately adjacent that are 10s to 100s of counts above the local bias level. Analysis of dark frames showed these umbrellas around O(10s) to O(100s) of hot pixels near saturation, a very small proportion of a 61-MPix or 102-MPix chip. The locations of these artifacts on each individual sensor are effectively random, but are consistent in severity and position in successive images. We will discuss these sites, show how they impact our data, explore how the existing literature references potentially similar phenomena, and hypothesize mechanisms causing these artifacts.