Conclusions from XRCF pileup analysis

The primary results from the pileup analysis for a subset of Phase F and H data taken at the Al-K$\alpha$ and O-K$\alpha$ energies with ACIS-I and ACIS-S in focus, and comparison to model predictions, are as follows:

• The measured pileup fraction increases with the source intrinsic flux, as expected, with a non-linear trend.
• The pileup is a function of grade selection, decreasing significantly when only grade G0 is considered (Figures 6.35 and 6.36). For bright sources and for the FI chip, grade filtering could provide an effective method to control pileup without losing the capability of doing some (modest) spectral analysis.
• The analytical model proposed by McNamara (1997) overestimates the true pileup. This model should thus be used to derive a conservative upper limit for the pileup.
• The MARX tool pileup underestimates the true pileup by as much as 50% at lower source fluxes and for no grade selection (see above).
• Subframing (i.e., reading out a reduced number of CCD rows centered on the aimpoint) can be used to control pileup. Available calibration observations show that pileup can be reduced by 44% for grades G02346 by going from 114 to 18 rows.
• For extremely bright sources either the gratings should be used, or Continuous Clocking mode used (or both).
• In cases where spatial resolution is not critical, it is possible to avoid pileup problems by intentionally positioning the source off-axis.