| dc.description.abstract | Complementary Metal-Oxide-Semiconductor (CMOS) detectors and Charge-Coupled Devices (CCDs) are the two primary imaging technologies used in optical and X-ray detection. Both rely on pixel arrays that convert incoming photons into electrical charge but differ in readout architecture: CCDs shift charge across the array to a common output node, while CMOS devices incorporate amplifiers and readout circuitry at each pixel. CCDs have long been favored in astronomy for their high sensitivity, low noise, and deep depletion regions that enhance detection of higher-energy X-rays. However, they suffer from slow readout, high power demands, and susceptibility to radiation-induced charge transfer losses. CMOS detectors, in contrast, offer fast readout, low power consumption, and increased resilience in radiation environments, while enabling on-chip processing and high time resolution. These advantages make CMOS increasingly attractive for astrophysical applications, particularly in capturing faint, transient, or rapidly varying X-ray phenomena. This work evaluates the potential of two modified commercial CMOS detectors from Sony’s uEye SE series, the IMX226 and IMX662, for low- to intermediate-energy X-ray astrophysics. To enhance sensitivity, the optical windows were removed and, for the IMX226, the microlens array was eliminated to reduce absorption at low energies. The detectors were characterized at the MIT Kavli Institute X-ray Detector Lab, with performance evaluated in terms of X-ray response, readout noise, pixel-to-pixel gain variation, linearity, dark current, and contributions to overall energy resolution. Detector testing used X-ray emission lines from Polonium-250 and Iron-55 at 277 eV (C), 677 eV (F), 5.9 keV (MnKa), and 6.4 keV (MnKb). Measurements were performed in a vacuum chamber to minimize absorption, with optical linearity tested separately on an optical assembly setup using an integrating sphere. Both detectors showed strong potential as low-cost X-ray sensors, with energy resolutions approaching theoretical limits across key emission lines. Readout noise was low (2.28 e⁻ for IMX226, 3.54 e⁻ for IMX662), gain variation was minimal when measured (≤0.32%), and linearity remained stable with errors below 0.6% across high- and low-energy regimes. Dark current was negligible for the IMX662 and modest for the IMX226 (0.57 e⁻/pixel/sec). While readout noise and gain variation explain much of the measured energy resolution, additional unaccounted noise was observed, indicating that further optimization is required. | |
