Long-Term Radiation Damage Effects on the Spectral Energy Resolution of the CCD Solid-state Imaging Spectrometer (SIS) on ASCA--by A. Rasmussen, G. Crew, G. Ricker & the SIS Team (MIT)
IntroductionThe CCD-based, Solid-state Imaging Spectrometer (SIS) on ASCA has continued to provide good energy-resolved spectroscopy for a wide variety of cosmic X-ray sources observed by ASCA. At the present time, both SIS cameras are operating reliably. With the development and maturation of ever more powerful image and spectral processing software for the SIS as the ASCA mission has progressed, the full capabilities of the SIS are now being generally realized.
Since the previous ASCANews report, there have been continuing increases in the number of "flickering pixels", as well as a broadening of the effective energy resolution of the SIS CCDs. Both of these effects are thought to be due to the accumulated energetic particle radiation dose from the 18 months in-orbit exposure. For the flickering pixels, effective identification and rejection tools exist for correcting ASCA data during on-ground processing. For the broader energy-resolution, a set of restoration algorithms are being developed, and are currently undergoing testing by the SIS Team. In this report, we briefly review the status of the SIS energy resolution, and outline the tools in preparation for resolution restoration.
Effects of Radiation Damage on the SIS Energy ResolutionThere appear to be at least two ways in which the energy resolution of the SIS CCDs on ASCA is affected by energetic particle-induced radiation damage:
- Increases in dark current, and evolution to a non-Gaussian distribution (referred to as "residual dark-current distribution," or RDD)
- Decline in X-ray signal charge transfer efficiency (referred to as "charge transfer inefficiency," or CTI)
The RDD appears to be intimately related to the flickering pixel numbers, and effectively increases the pixel noise. From what is presently known, it appears that the additional broadening introduced by the RDD scales with exposure time (4 times worse for 4-CCD mode than for 1-CCD mode), and also scales proportionally with the 0.4 power (approximately) of the elapsed time since launch.
The CTI primarily affects the gain solution in a position-dependent fashion. Currently, the best (position-dependent) CTI measurements result from 1-CCD mode observations of Puppis-A on March 30, 1993, and May 30, 1994. Although the CTI effects are in principle partially correctable via pulse-independent (PI) binning, we are aware of other CTI dependencies (e.g., local count-rate) whose corrections are much less straightforward. In the following analyses, we have underestimated some of the CTI effects and overestimated others. Underestimates result from assuming a simple, linear dependence to the photon arrival position, by considering single pixel events only, and by neglecting local count-rate effects. Overestimates should result from assuming a linear relation with photon energy, a linear growth of the effect with time since launch and no attempt to correct the simple, position dependence. The overall resolution estimates are therefore conservative, although they are tabulated here only for 1-pixel calculation events (grade 0, split threshold = 40 ADU).
The following table compares SIS CCD sensor S0C1 and S1C3 spectral resolution (FWHM) at 1 and 6 keV, for 1 and 4 CCD modes, single pixel events only, at intervals of 0.5 years following launch, and extrapolated out to 3 years post-launch (i.e., to February, 1996). The 1-CCD mode energy resolution (FWHM) is altered at low energies primarily by the evolving RDD, while the CTI strongly affects the high energy resolution. In contrast, the 4-CCD mode resolution is dominated by the RDD effects over all energies.
Table1. Modelled Energy Resolution Changes in the ASCA SIS CCD sensors as a Function of Elapsed Time since Launch for X-ray Single Pixel Detections.
From the table, it is apparent that for observations requiring the best energy resolution, 4-CCD mode has already been significantly affected: In addition to suffering larger "flickering pixel" numbers, the spectral resolution is also inferior compared to that of 1-CCD mode. The spectral resolution available to 2-CCD mode lies somewhere between the values for 1 and 4-CCD modes. The above table makes it clear that 1-CCD mode is strongly recommended for ASCA observations for which optimum spectral resolution is most important, and a smaller SIS field-of-view can be tolerated.
Checking and refinements by the SIS Team of our semi-empirical models for energy resolution are continuing, with further insights expected from detailed analyses of just-completed calibration observations of the supernova remnant Cas A (July 1994), and of deep "blank pixel" comparisons using North Ecliptic Pole observations carried out in March 1993 and March 1994.
New SIS ToolsFor implementing the corrections which are derivable from the above analyses of the effects of RDD and CTI, a suite of tools are being developed by the SIS Team and GOF personnel. In the upcoming FTOOLS 3.0 release, two such tools are:
SISPI (written by Koji Mukai) which populates a pulse-invariant (PI) column in Bright or Bright2 mode science files. The pulse-height (PH) to PI transformation is based on preliminary analyses of CTI calibration observations of the supernova remnant Puppis A in June 1994.
SISRMG (written by MIT) has been enhanced in several ways for the FTOOLS 3.0 release. It includes a preliminary CTI gain calibration which allows the creation of response matrices appropriate to a given observational epoch. The detection efficiencies of the individual CCDs have been adjusted according to the most recent analyses at ISAS. Finally, the response model now includes a high-PH tail model to correct for partial charge collection events occurring deep in the CCD depletion layer.
In the coming months, we anticipate further refinements of these two tools, as well as development of even more efficient spectral resolution restoration algorithms, by the SIS Team. These tools will be made available to the ASCA user community by the ASCA GOF, as AO-3 progresses.
Proceed to the next article Return to the previous article