Proton-induced traps in electron multiplying charge-coupled devices

Abstract

Charge-coupled device (CCD)-based technologies exposed to high-energy radiation are susceptible to the formation of stable defects within the charge transfer channel that defer signal to subsequent pixels and limit the lifetime of the detector. Performance degradation due to these defects depends upon the interplay between the clock timings used to operate the device and the properties of defects introduced by irradiation. Characterization of both the type and number of post-irradiation defects makes it possible to minimize charge loss though the appropriate selection of clock timings for a given operating temperature. This technique has the potential to increase nominal mission lifetimes by several years for CCD-based instruments and is of particular significance to electron multiplying charge-coupled devices (EMCCDs) for photon counting applications where the effect of charge traps on low signal levels is expected to be most severe. We present a study of charge traps within CCDs, specifically within EMCCDs irradiated at room temperature to proton fluences up to and including 1.45 × 10 10 p þ ∕cm 2 (74 MeV). Defects are characterized through the "single-trap pumping" technique, with clocking schemes specifically designed for the 2-phase pixel architecture of the EMCCD. Five dominant trap species are thought to be introduced by the irradiation, the Si-E center, Si-A center, double and single acceptor charge states of the silicon divacancy (VV −− , VV − ), and an as yet unidentified defect referred to here as the Si-U center (the "unknown" trap). Energy-level and crosssection values are presented that allow inference of the defect landscape for a range of proton fluences and operating temperatures. While the study focuses specifically on EMCCDs, in more general terms, the results for trap properties are interpreted as being applicable to all CCD types following irradiation and can serve as a foundation for future charge loss correction and optimization techniques.

Developing the active trap model for charge-coupled device charge transfer optimization in large-scale space missions

“…Figure 17(b) shows the distribution of emission time constants for both temperaturecorrected landscapes, with the peak of the divacancy and unknown defect aligning with the known literature values 12 (energy level and cross-section). At 193 K, the theoretical emission time constant of the divacancy is ∼2 × 10 −4 s (using E ¼ 0.225 eV and a cross-section of 5 × 10 −16 cm 2 ), which is within error (compared with the data in Fig.…”

“…The defect is well studied, however, with a known cross-section and energy level. 12 As a result, the defect can be added artificially to the landscape via the use of SRH to calculate the emission time constants of the Si-E, with the shape of the peak being estimated by a Gaussian function. Although the position and shape of the Si-E peak can be added accurately, the defect density is less known.…”

“…18 CCD280 trap landscape with a Gaussian shaped Si-E peak added based upon known literature energy level and cross-section values. 12 Gaussians with differing peak heights are shown here; these will subsequently be used as input to the ATM.…”

“…Previous work utilized this optimization method, known as the active trap model (ATM), as part of the Nancy Grace Roman Space Telescope to optimize charge transfer performance of an electron-multiplying CCD201. 12 More specifically, a model was developed to optimize a single phase of readout for one operating mode (full-frame readout) at a range of temperatures. However, CCD-based space missions are wide-ranging, operating using different CCDs from different manufacturers (and hence different trap landscapes), at different clocking speeds, and at different temperatures.…”

Nancy Grace Roman Space Telescope coronagraph instrument observation calibration plan

“…The bases of estimate are a published trap pumping analysis of radiation damaged EMCCDs, similar to the Coronagraph flight EMCCD (Bush et al 2021), and simulations of image corruption with the CTI simulation/correction software ArCTICPY, which yield a CBE <1.6% RSME in Fp/Fs (>41% margin) . About 98% of the CTI trailing in Hubble images can be corrected in post-processing; the resulting fractional residual flux errors are near 0.3% (Massey et al 2014)).…”