CTI corrector

The Penn State ACIS reduction differs in a fundamental fashion from the standard CXC pipeline reduction in that, very early in the processing, each photon is corrected (as much as possible) for the effects of charge transfer inefficiency (CTI). CTI was considerably degraded in the ACIS-I CCD chips at the beginning of the Chandra mission due to particle bombardment in the Earth's radiation belts. Magnetospheric protons cause charge traps in the buried channels of the CCD that increase CTI (Prigozhin et al., 2000). The result is row-dependent gain, event grade (spatial distribution of charge), and energy resolution.

For CTI correction, we have modeled the CTI as a function of input photon energy including the effects of de-trapping (charge ``trailing''), shielding within an event (``leading'' pixels in the 3x3 event island protect the rest of the island by filling traps), and non-uniform spatial distribution of CTI. Unlike most other CTI correctors that treat only the spatially-dependent energy degredation of the chips, the PSU CTI corrector also treats "grade migration" and thus changes the events used for scientific analysis. Together, the CTI corrector provides up to 20% increase in energy resolution and 20% increase in quantum efficiency in the ACIS-I array compared to standard Level 2 event lists, though the improvements are smaller in many cases. While the degraded energy resolution near the top of the ACIS-I chips cannot be fully recovered (requiring several rmf files in spectral analysis), a single rmf file suffices for the ACIS-S3 chip.

The corrector is described in Townsley et al. (2000). It is applied to the Level 1 event list, and is driven by a simple script, correctit as follows:

The script correctit is now available for all four ACIS-I chips and the ACIS-S chip at both -110 and -120 C. It is not available at the present time for other S-array chips.
Note 1: The fidelity of the corrector solution has been established only for a handful of observations of the on-board External Calibration Source datasets. Users are strongly encouraged to run the corrector on a calibration observation obtained during the same orbit as their science observation. A list of calibration observations (some routine external cal source, some special tests) is graciously maintained by MIT. Limited experience suggests that the corrector may alleviate most of the energy degradation, a small portion may remain (e.g. 2% from an uncorrected 20% drop in gain). Since the corrector solution is based only on the ?three? calibrator lines with energies in the range 1.5-6.4 keV, its performance outside these energy may not be excellent and has not yet been assessed.
Note 2: The IDL part of the corrector takes lots of memory for large event lists; at Penn State, one should run it on gonzo2 or pollux. On the other hand, running the correctit script also contains several ftools calls which manipulate files in your current working directory. These will run faster on the machine that's connected to that disk.
Note 3: At Penn State, the CTI corrector requires you have /bulk/pkg/xray/bin/ in your unix path and, if you use the IDL configuration environment variable IDL_PACKAGES, you must include the package ACIS_PSU.