Characterization of Si hybrid CMOS detectors for use in the soft X-ray band

Abstract

We report on the characterization of four Teledyne Imaging Systems HAWAII Hybrid Si CMOS detectors designed for X-ray detection. Three H1RG detectors were studied along with a specially configured H2RG. Read noise measurements were performed, with the lowest result being 7.1 e -RMS. Interpixel capacitive crosstalk (IPC) was measured for the three H1RGs and for the H2RG. The H1RGs had IPC upper limits of 4.0 -5.5 % (up & down pixels) and 8.7 -9.7 % (left & right pixels), indicating a clear asymmetry. Energy resolution is reported for two X-ray lines, 1.5 & 5.9 keV, at multiple temperatures between 150 -210 K. The best resolution measured at 5.9 keV was 250 eV (4.2 %) at 150 K, with IPC contributing significantly to this measured energy distribution. The H2RG, with a unique configuration designed to decrease the capacitive coupling between ROIC pixels, had an IPC of 1.8 ± 1.0 % indicating a dramatic improvement in IPC with no measurable asymmetry. We also measured dark current as a function of temperature for each detector. For the detector with the lowest dark current, at 150 K, we measured a dark current of 0.020 ± 0.001 (e -sec -1 pix -1 ). There is also a consistent break in the fit to the dark current data for each detector. Above 180 K, all the data can be fit by the product of a power law in temperature and an exponential. Below 180 K the dark current decreases more slowly; a shallow power law or constant must be added to each fit, indicating a different form of dark current is dominant in this temperature regime. Dark current figures of merit at 293 K are estimated from the fit for each detector.

Soft x-ray quantum efficiency of silicon hybrid CMOS detectors

“…This spectrum has had the scaled-background subtracted and shows a fit to the large peak (Green line) made up of three Gaussian components fitting the line at 1.49 keV. The three Gaussians are due to the inter-pixel capacitance (IPC) effects in H1RG detectors, described in [4], [5]. IPC causes the signal from an event to spread into neighboring pixels.…”

“…Knowledge of the window transmission, gas properties (temperature, pressure, composition), active detector depth, along with appropriate X-ray cross-sections are all that are needed. The PC QE can then be used to calculate the HCD QE for a given energy when data is collected simultaneously in each detector, assuming that the geometry of the detectors with respect to each other is well understood; described in [5] and further in Bongiorno 2013 [7]. The QE HCD is then:…”

“…H1RG data analysis has been described in previous work [5], that was focused on making event cuts to improve energy resolution. For QE calculations, every photon event needs to be counted, making the spectrum more straight forward to measure.…”

“…For QE calculations, every photon event needs to be counted, making the spectrum more straight forward to measure. In this case, once each HCD raw image has been processed using techniques described in [2], [5], events are counted using a primary pixel threshold (100 dN, where dN is the channel number) which must also be a local maxima. The charge for each event is then totaled by adding up all charge above a secondary threshold (set at the median of the noise floor, approximately 30 dN) in the 3x3 pixel 2 event stamp centered on the primary pixel.…”

“…This capability makes these devices fast with lower high speed noise than CCDs [1] (speeds of 10 Mpixel/sec are possible), inherently radiation hard, and lower power. Penn State University (PSU) and Teledyne Imaging Sensors (TIS) have been collaborating on the design of such HCDs [2], [3] and have recently reported the characterization of X-ray energy resolution, dark current, read noise, and inter-pixel capacitance (IPC) for some of these devices [4], [5].…”