Because the HETG illuminates regions of the ACIS-S FI CCDs with essentially monochromatic, known energy X-rays, HETG data can be used to measure parallel CTI values. Three methods/examples are illustrated here.
Some "four-function-calculator" results for Day 258 conditions (Obsid 62538 "part1") are summarized here and can be improved on with more detailed analysis (e.g., using the centroids fit by Norbert and Sara-anne.)
- S0,S2,S4,S5 CTI is energy dependent with values of order:
- 0.8 to 1.0 kev ~ 5.7 x 10-4 +/- 20%
- 1.3 to 1.5 kev ~ 3.6 x 10-4 +/- 30%
- E/dE due to CTI is approximated by:
E / dE ~=~ 1.7 / (CTI x CHIPY) +/-20%(Based on 0.8-1.0 keV range for FIs on ACIS-S)
- S1 CTI at 0.8 to 1.0 keV is ~ 1.5 x 10-4 +/-30%
The HETG spreads the incident spectrum into four spectral "parts" - the four arms of the "X" pattern. Because the "X" varies in CHIPY it is sensitive to CTI effects. Energies near 0.8 to 1.0 keV are spread out to the "tips" of the HEG arms with a delta-CHIPY value of 500-400 pixels. These same energies are spread to half these distances on the MEG arms. Similarly for the higher energy range of 1.3 to 1.5 keV - although the delta-CHIPY is smaller.
By making the assumption that all ACIS-S FIs have similar parallel CTI values, the centroids (of ENERGY, PHA, or PHAS(4) ) for these regions (or specific lines in the regions) can be measured and plotted vs CHIPY and analyzed for CTI value in the standard way:
slope-of-line
CTI ~=~ --------------------------- (fraction/pixel)
line-intercept-at-CHIPY=0
Slightly dissimilar energies (e.g., 0.826 and 1.022 keV)
can be included on the same CTI
curve by performing the analysis using the centroid normalized
by the known HETG energy, that is using PHA/E_HETG in place of PHA.
This technique produced the FI CTI values for 0.8-1.0 keV and 1.3-1.5
keV quoted above.
If the events from specific dispersed lines are extracted, the PHA distribution due to monchromatic incident energy is obtained. Then the value PHA_FWHM/PHA_Centroid can be plotted as a function of CHIPY. From preliminary data it appears that this "normalized FWHM" is proportional to CHIPY and, if the CTI value is known, the coefficient relating "dE/E" to CTIxCHIPY can be determined. Again using the HR1099 obsid 62538 part1 results, the E/dE equation presented above was determined.
This is a little risky but for the BIs on ACIS-S it is the only way short of moving the "X" pattern around to illuminate different CHIPY locations.
Basically, the assumption is made that the ENERGY in the FITS level 1 events file is correctly calibrated (or to 3%!) for NO CTI present. Taking a line or region on the diffracted X the ACIS energy can be measured and the average real (HETG) energy is known. Then, assuming the only reason for centroid change is parallel CTI the CTI can be determined:
[ 1 - (E_ACIS / E_HETG) ]
CTI ~=~ ---------------------------
CHIPY
This technique can be used on the S1 data from HR1099 (62538)
which shows E_ACIS/E_HETG = 0.78/0.826; 0.913/1.022
for CHIPY = ~400, ~700. This gives S1 CTI 0.8-1.0 keV
of order 1.5E-4.
Please send any comments to
Dan Dewey at dd@space.mit.edu.