Testing the IDL implementation (v5.3) of the temperature-dependent CTI correction

Testing with temperature dependence set to zero

Here I compare an event list CTI corrected with acis_process_events to a version CTI corrected with the IDL code with the temperature dependence set to zero (should be identical).

Percent of Flight Grades and PHAs Correct
ObsidSourceModeFltgradePHA
60501E0102 on I3VFAINT100%100%
57036ECS 012357FAINT100%99.9982%
10098G18.95-1.1 on ACIS-IVFAINT100%99.9991%
11713Abell 426 on ACIS-I, S2-3FAINT100%99.9993%

The small percentage of PHAs that are not correct differ by only +/- 1.

Testing with temperature dependence

E0102 Obsid 6050

Focal plane temperature:

PHA for G02346 events with different CTI corrections:


G18.95-1.1 Obsid 10098

Focal plane temperature:

PHA for G02346 events with different CTI corrections:


ECS Obsid 57036

Focal plane temperature:

PHA for G02346 events with different CTI corrections:

Fits to ECS lines:
CCD / locationLine*Centroid: Energy (keV)Centroid: PHA (ADU)
a_p_etcticorrdifferencea_p_etcticorrdifference
I3, node 3, top 64 rowsAl K-alpha1.44+-0.021.52+-0.02 +0.08 (5.5%)369+-5388+-5+19 (5.1%)
Mn K-alpha5.78+-0.015.90+-0.01 +0.12 (2.0%)1478+-41512+-4+34 (2.3%)
S3, node 0, middle 64 rowsAl K-alpha1.494+-0.0061.493+-0.007-0.001 (0.07%)330+-1330+-10
Mn K-alpha5.906+-0.0065.894+-0.006-0.012 (0.20%) 1320+-11317+-1-3 (0.23%)

* The Al K-alpha line is at 1.487 keV; Mn K-alpha is at 5.898 keV.
Line centroids determined by local Gaussian fit. Uncertainties are 90% confidence interval.


Coma (ObsID 9714)
Alexey's plot
After applying the temperature-dependent CTI correction, fits to the Coma cluster show the same result as the other clusters with respect to contamination modeling. Before using the temperature-dependent CTI correction this observation was an outlier.

Perseus (ObsID 11713)
Larry's presentation
For half of this observation the ACIS focal plane temperature was within the desired range; the other half was warm. Fits to the warm and cool portions CTI-corrected with either acis_process_events or tcticorr show that the temperature-dependent CTI correction results in best-fit redshifts that are closer to the optical redshift.