Pileup trends with source flux and grade selection

Table 6.22 lists the results of our pileup analysis of Phase H data. Column 2 reports the fraction of total events detected for grade selection All-255, defined as the ratio of the total counts detected in the observed spectrum to the expected incident counts from Table 6.21. The following three columns list the counts detected in the primary peak (col. 3), the sum of all the counts detected in the piled peaks (col. 4), and the pileup fraction (col. 5), according to eq. 6.14. In the following, we will refer to the pileup fraction estimated from the XRCF data as the ``measured'' pileup fraction. The same information is contained in the following columns of Table 6.22 for grade selections G02346, G023, and G0.
 

Table 6.22: Pileup Analysis for Phase H Data 

Test Name	Detected Events for All - 255					Detected Events for G02346					Detected Events for G023					Detected Events for G0
	Frac. of Events	Peak1	Piled Peaks	Pileup Frac.		Frac. of Events	Peak1	Piled Peaks	Pileup Frac.		Frac. of Events	Peak1	Piled Peaks	Pileup Frac.		Frac. of Events	Peak1	Piled Peaks	Pileup Frac.
a) Front-illuminated CCD
H-IAI-CR-1.001	0.93	18446	1054	0.054		0.90	18004	774	0.041		0.95	17581	575	0.033		0.95	14242	140	0.010
H-IAI-CR-1.003	0.77	17160	3639	0.175		0.68	15826	2500	0.136		0.64	14932	1809	0.108		0.66	12499	434	0.034
H-IAI-CR-1.005	0.51	8218	5563	0.403		0.37	6958	2972	0.300		0.35	6514	1763	0.213		0.30	5404	336	0.059
H-IAI-CR-1.007	0.35	6459	6844	0.514		0.21	5328	2546	0.32		0.18	5021	1235	0.197		0.16	4183	168	0.038
H-IAI-CR-1.009	0.77	21792	1520	0.065		0.74	21061	1321	0.059		0.78	20860	1079	0.052		0.87	19630	295	0.015
H-IAI-CR-1.011	0.66	15191	3079	0.169		0.59	13842	2438	0.150		0.61	13707	1841	0.118		0.60	12835	531	0.041
b) Back-illuminated CCD
H-IAS-CR-1.002	0.89	17461	1161	0.062		0.86	16911	530	0.030		0.85	10936	158	0.014		0.61	4508	11	0.002
H-IAS-CR-1.004	0.76	14355	3553	0.198		0.65	13226	1573	0.106		0.62	8651	473	0.05		0.41	3446	42	0.012
H-IAS-CR-1.006	0.47	7924	6554	0.453		0.28	6554	1935	0.228		0.24	4007	414	0.094		0.14	1484	29	0.020
H-IAS-CR-1.008	0.32	4779	6048	0.559		0.15	3755	1009	0.212		0.11	2201	135	0.058		0.06	776	9	0.012
H-IAS-CR-1.012	0.77	15573	920	0.056		0.75	15147	512	0.033		0.85	12397	264	0.021		0.85	6552	31	0.005

From Table 6.22 a number of features are apparent. For increasing source flux, the pileup fraction increases, although in a non-linear fashion, as expected due to the Poisson nature of the pileup phenomenon. This is illustrated in Figures 6.35 and 6.36, where the pileup is plotted versus the incident counts/frame as a function of grade selections for FI and BI chips, respectively.
 
 

Figure 6.35:  Plot of the measured pileup fraction
from phase H data for chip I3 of ACIS-I, versus source flux incident
on the HRMA entrance (Table 2a). Different grade selections are
shown. The dotted curves simply connect the datapoints and are plotted
to guide the eye. For increasing source flux the pileup fraction
increases, with a non-linear trend. For a fixed source flux, the
pileup decreases when events related to the higher grades are
discarded. This also corresponds to throwing away an increasingly
larger fraction of total detected events (Table 6.20a).

 
 

Figure 6.36:  Plot of the measured pileup fraction from phase H data for chip S3 of ACIS-S, versus source flux incident on the HRMA entrance (Table 2b), as a function of grade selection.  The dotted curves simply connect the datapoints and are plotted to guide the eye.  As for FI chip (see Figure 6.23), there is a non-linear trend of increasing pileup for increasing source flux. Discarding the events corresponding to the higher grades lead to a significant reduction of the pileup at the expense of a drastic loss of total events (cfr. Table 2b).

 

For a fixed source flux, the pileup fraction decreases when the higher grades are discarded, as illustrated in Figures 6.35 and 6.36. The decrease in pileup is more dramatic for BI chips than for FI at comparable source fluxes, as expected because of the way the BI chips work. Note also in Figure 6.36 the ``cutoff'' of the pileup at higher fluxes for the BI chip, which is particularly evident for grades G023 and G0. This is possibly related to grade migration at higher fluxes.

The decrease in pileup is accompanied by a loss of total detected events, i.e., going from G02346 to G0 more and more photons are thrown away. For example, at the Al-K$\alpha$ energy and for a source incident flux of $\sim$ 0.05 counts s^-1 cm^-2 (test H-IAI-CR-1.005), from Tables 6.21 and 6.22a we can see that for the FI chip the pileup is reduced from $\sim$ 40% to 6% going from no grade selection to G0, with a loss of 58% of detected photons. At a slightly lower source flux for the BI chip of $\sim$ 0.04 counts s^-1 cm^-2 (test H-IAS-CR-1.006), the pileup is reduced from$\sim$ 45% to 2%, losing a larger fraction of events, 89%. We thus conclude that, at least for the FI chip, grade selection can be used for bright sources to reduce efficiently the pileup and still retain enough photons to do some simple spectral analysis.