MOS Pn cross MOS Pn cross calibration with a calibration with a sample of Galaxy sample of Galaxy
ClustersClusters
Alberto Leccardi &Alberto Leccardi &
Silvano Molendi (IASF-MI)Silvano Molendi (IASF-MI)
The ExerciseThe Exercise• Sample of 21 , high gal latitude, hot, intermediate Sample of 21 , high gal latitude, hot, intermediate
redshift Galaxy Clusters observed with XMM-redshift Galaxy Clusters observed with XMM-Newton.Newton.
• About 80 high quality spectra (bkg is not an issue)About 80 high quality spectra (bkg is not an issue)• All reduced with SAS 6.0 (rerun with 6.1 to begin All reduced with SAS 6.0 (rerun with 6.1 to begin
soon) soon) • Emchain & Epchain correction for OOT appliedEmchain & Epchain correction for OOT applied• Effective areas generated using arfgen with Effective areas generated using arfgen with
extended source optionsextended source options• Rmfs generated using rmfgenRmfs generated using rmfgen• MOS1, MOS2 (pattern 0-12) and pn (0-4) MOS1, MOS2 (pattern 0-12) and pn (0-4) • flag==0 for MOS and pnflag==0 for MOS and pn
Spectral accumulated in concentric annuli (from 5 to 8 depending on source SB). Important for testing full FOV
MOS1 image MOS1 image of A1689of A1689
The ExerciseThe Exercise
Which band?Which band?• Testing spectral calibration with thermal
spectra is somewhat different than with power-laws.
• For high T (kT > 4keV) the scientific valuable information is all in the hard band, however for EPIC and many other X-ray experiments most of the statistics is in the softer band.
• Performing spectral fits using a cumulative statistics such as χ2 on a broad band will put lots of weight on the insensitive soft part of the spectrum and relatively less on the hard scientifically important part.
Which band?Which band?Two possible points of viewObserver: forget the soft band, it
only cmplicates things!Calibrator: if I can get the right
temperature fitting the broad band than my calibration must really be good!
Spectral fits in
0.5-10 keV 1.5-10 keV 2-10 keV
Which band?Which band?
Spectral fits in
0.5-10 1.5-10 2-10 for MOS1 and MOS2
0.5-7.3 1.5-7.3 2-7.3 for pn
1. For each band comparison btwn detectors.
2. For each detector comparison btwn T measures in different bands
MOS1 vs MOS2 MOS1 vs MOS2
mean = -0.064
MOS1 vs MOS2 MOS1 vs MOS2
mean = -0.054
MOS1 vs MOS2 MOS1 vs MOS2
mean = -0.075
MOS1 vs pn MOS1 vs pn
mean = 0.063
MOS1 vs pn MOS1 vs pn
mean = -0.007
MOS1 vs pn MOS1 vs pn
mean = -0.18
Summary Summary MOS vs pn MOS vs pn
1. For hard bands reasonable agreement btwn all instruments.
2. Broad band MOS and PN are clearly discrepant
1. For each band comparison btwn detectors.
2. For each detector comparison btwn T measures in different bands
2-10 keV vs 1.5-10 2-10 keV vs 1.5-10 keV keV
(T2-10-T1.5-10)/T1.5-10
N o
f ob
j
0.5-10 keV vs 1.5-10 0.5-10 keV vs 1.5-10 keV keV
(T0.5-10-T1.5-10)/T1.5-10
N o
f ob
j
Summary Summary
1. T drops as you include softer parts of the spectrum (at least part of this is not due to calibration)
2. the Pn temperatures vary considerably more than the MOS T
3. Bigest problem is 20% pn variation when going from 1.5-10 keV to 0.5-10 keV
SAS 6.0 -> 6.1 SAS 6.0 -> 6.1 For one object A2199 we compared old and new For one object A2199 we compared old and new
SASSAS
0.5-10 keV (filled circles) vs 1.5-10 keV empty 0.5-10 keV (filled circles) vs 1.5-10 keV empty
circlescircles
SAS 6.0 SAS 6.1SAS 6.0 SAS 6.1
SAS 6.0 -> 6.1 SAS 6.0 -> 6.1
Summary Summary 1. In the hard bands we have
reasonable agreement btwn instruments, better than 10%
2. In the soft band there is a large discrepancy btwn. MOS and pn, the fact that pn in 1.5-10 keV and pn in 0.5-10 keV show a similar discrepancy indicates that the problem is a pn rather than a MOS problem.
3. Pn changes in SAS 6.1 solve only part of the problem