Документ взят из кэша поисковой машины. Адрес оригинального документа : http://xmm.vilspa.esa.es/calibration/documentation/epic_cal_meetings/201203/nevalianen-cal-2012.pdf Дата изменения: Tue May 13 17:39:00 2014 Дата индексирования: Wed Apr 13 13:38:40 2016 Кодировка:

Clusters and effective area
J. Nevalainen, M. Guainazzi, L. David, K. Kettula et al.
BOC, March 7, 2012, Leicester


Outline


Review of IACHEC clusters Working Group results (mainly from Nevalainen et al., 2010 A&A 523, 22: "Cross-calibrating X-ray detectors with clusters of galaxies: an IACHEC study" MOS fudge as seen with cluster data




IACHEC


International Astronomical Consortium for High Energy Calibration IACHEC ( http://web.mit.edu/iachec/ ) Aims at cross-calibration between different high energy missions. Currently active: Chandra, XMM-Newton, Suzaku, Swift, RXTE, INTEGRAL Define high energy standard candles (SNR, clusters of galaxies, white dwarfs, neutron stars, blazars) Analysis of data of standard candles obtained with different instruments









calibration


Cluster sample


Hard X-ray spectra (kT ~ 2-10 keV) Nearby (z < 0.08), bright (10-12-10
-11



erg s-1 cm-2) good statistics



Clusters are stable no simultaneity requirement sample systematic effects 11 clusters: A1795, A2029, A2052, A2199, A262, A3112, A3571, A85, Coma, HydraA, MKW3S Observed with ACIS/Chandra, EPIC/XMM-Newton crosscalibration Most have a cool core, no major merger signatures Spectra extracted within r = 6 arcmin, excluding the cool core Physics well understood (bremsstrahlung continuum + collisionally excited line emission












Method


Spectral fits with 1-T MEKAL model Data for different instruments extracted from the same annular sky region for a given cluster Compare T for a given cluster obtained with different instruments cross-calibration of the shape of the effective area ( telescope effective area в filter transmission в quantum efficiency) In more detail: take pn best-fit model, fold through MOS response and divide with MOS data. If pn response is correct, the ratio tells how much MOS effarea is incorrect Compare fluxes cross-calibration of the normalisation of the effective area Fe XXV/XXVI line ratio T measurement for the hottest clusters as an additional tool












Controlling the systematics


In the bright central cluster regions the background is at a few % level. 10% variation in the bkg makes less than 1% effect to the bkg-subtracted spectrum Extraction regions are large enough so that PSF scatter from the cool core is less than 1% of the intrinsic cluster flux from the extraction region Extraction regions large enough so that PSF scattered flux to and from the extraction region is less than 1%. No effect to the flux comparison.






pn hard band spectral fits


Hard band (2-7 keV) temperatures measured by XMM-Newton/pn,XMMNewton/MOS, Chandra/ACIS, BeppoSAX/MECS, Suzaku/XIS


pn / combined MOS1+MOS2


MOS1+2 yields temperatures ~2% lower than pn on average

J


MOS1 / MOS2


MOS1 yields temperatures ~4% higher than MOS2 (~3) on average MOS1 effarea is a bit too soft if MOS2 response is correct

K


MOS1 / pn


MOS1 yields temperatures ~1% higher than pn in average MOS1 effarea is a bit too soft if pn response is correct

J




MOS2 / pn


MOS2 yields temperatures ~5% lower than pn (~3) on average MOS2 effarea is a bit too hard if pn response is correct

K


EPIC/BeppoSAX


J

pn yields temperatures consistent with BeppoSAX MECS (De Grandi & Molendi, 2002, ApJ, 567, 163) The average difference is ~1% (~0.9)




EPIC/ACIS


pn yields temperatures consistent with ACIS The average difference is ~1% (~ 0.6 )

J


EPIC/Suzaku


pn yields temperatures consistent (within 5%) with SUZAKU (K. Kettula et al., in prep.)

K


Bremsstrahlung temperature (2-6 keV band fit) ionisation temperature (67 keV band fit)
and




Fe XXV/XXVI line ratio increases with temperature Fe XXV/XXVI based T is not sensitive to the details of effarea calibration due to narrow band pn and MOS Fe XXV/XXVI based temperatures are consistent with 2-6 keV continuum fit effarea absolutely correct?





J



no significant deviations from ionisation equilibrium state and Maxwellian electron velocity distribution in the sample Fe XXV/XXVI useful for calibration


SOFT band (0.5-2.0 keV) temperatures


pn, MOS1, MOS2


pn, MOS1 and MOS2 temperatures agree within 5%

K


L


ACIS v.s. pn

ACIS yields ~20% (9) higher soft band temperatures than pn Most of the photons are in the soft band full band temperatures biased by 10% How does this look with blazars?


ACIS data / pn model


L

ACIS data / pn model exhibit a linear trend with energy In pn effarea is correct, ACIS effarea too high by ~10% at 0.5 keV




Hard band flux




MOS1 fluxes ~5% higher than pn MOS2 fluxes ~7% higher than pn





Similar to XCAL blazars

MOS/pn flux


Summary of public calibration
1) Hard band (2-7 keV) temperatures measured by XMM-Newton/pn, XMMNewton/MOS, Chandra/ACIS, BeppoSAX/MECS, Suzaku/XIS agree within a few % Hard band effective area shape cross-calibration uncertainties only at a few % level (DON'T FUDGE TOO MUCH!) 2) Bremsstrahlung and Fe XXV/XXVI temperatures of EPIC and ACIS consistent Hard band effective area shapes of EPIC, ACIS and MECS absolutely calibrated? (DON'T FUDGE TOO MUCH!) 3) MOS hard band flux 5-10% higher than pn, similar as with XCAL blazars Cluster data better explained if MOS effective area were scaled higher by a constant 5-10% (DO AHEAD AND FUDGE! WON'T HELP pn/ACIS) 4) ACIS soft band yields ~20% higher temperatures than EPIC In pn effarea is correct, ACIS effarea too high by ~10% at 0.5 keV

HOW COULD WE GET ACTION HERE?


MOS fudge


MOS fudge




Re-analysis of Nevalainen et al. (2010) cluster sample using


SAS 11.0.0 pn CCF:s on March 31 2011 MOS CCF:s on Dec 17, 2012 Fudged MOS effective area:








XRT1_XAREAEF_0009.CCF XRT2_XAREAEF_0010.CCF





Fit 2.0-6.0 keV band (i.e. exclude the Fe XXV/XXVI band) so that temperatures purely driven by bremsstahlung shape and thus better indicative of the effective area changes. (Metal abundance fixed to 0.3 Solar)


MOS1 2-6 keV band temperatures


The fudge temperatures are systematically lower than the ones obtained with public CCF:s 5% at kT = 2 keV 20% at kT = 10 keV







The effect is relatively bigger with higher temperature: the spectrum becomes harder since the effarea change has more effect at the higher energies


MOS2 2-6 keV band temperatures


The fudge temperatures are systematically lower than the ones obtained with public CCF:s 5% at kT = 2 keV 20% at kT = 10 keV







The effect is relatively bigger with higher temperature: the spectrum becomes harder and the effarea change has more effect at the higher energies


L
As a consequence, the pn/MOS hard band temperature agreement in Nevalainen et al. (2010) is lost




MOS1 fudged ("test") 2-6 keV band temperatures are ~10% smaller than pn values obtained with public calibration Relative difference rather independent of temperature






MOS2 fudged ("test") 2-6 keV band temperatures are ~10% smaller than pn values obtained with public calibration Relative difference rather independent of temperature






We also loose the MOS hard band agreement with Chandra, BeppoSAX and Suzaku Something is wrong with the fudge...

L




Fe XXV/XXVI line ratio




Best-fit MOS1+2 6-7 keV band values are almost identical when using the fudged or public effarea: calibration independent


Fe XXV / XXVI v.s. bremsstrahlung
Public


Fudged

Bremsstrahlung temperatures Bremsstrahlung temperatures higher than Fe XXV / XXVI lower than Fe XXV / XXVI values by ~7% values by ~7% Inconclusive


Temperatures are rough indicators of effarea calibration. Need to look at the residuals to get the last 10%


Fudge effect to MOS1/pn residuals


If pn effarea is correct, MOS1 fudge effarea is biased ~10% too low at 2 keV The bias decreases with energy, roughly vanishing at 7 keV Dip at 2.0-2.5...






Fudge effect to MOS1/pn residuals


If pn effarea is correct, MOS1 fudge effective area is biased ~10% too low at 2 keV The bias decreases with energy, roughly vanishing at 7 keV Dip at 2.0-2.5...






MOS1/pn public

J


MOS1/pn fudge

L


Fudge effect to MOS2/pn residuals


If pn effarea correct, MOS2 fudge effarea is biased ~10% too low at 2 keV The bias decreases with energy, roughly vanishing at 6 keV Dip at 2.0-2.5 keV and 6-7 keV...






Fudge effect to MOS2/pn residuals


If pn effarea correct, MOS2 fudge effective area is biased ~10% too low at 2 keV The bias decreases with energy, roughly vanishing at 6 keV Dip at 2.0-2.5 keV and 6-7 keV...






MOS2/pn public

J


MOS2/pn fudge

L


Fudge effect to M1/M2 residuals


If MOS2 effarea correct, MOS1 fudge effarea correct at 2.5 keV The MOS1 bias increases with energy, yielding ~10% too high effarea at 7 keV Dip at 2.0-2.5 keV...






Why do clusters and blazars tell a different story?


CLUSTERS + BLAZARS

CLUSTERS



Clusters do not show the steep rise of residuals at 3.0-4.0 keV as blazars do. PSF?




The pn/MOS1 cluster temperatures are consistent within 1% with the public effarea: no need for fudging the effarea shape



The pn/MOS1 XCAL blazar spectral shapes are inconsistent using the public effarea: need for fudging the MOS1 effarea shape harder



MOS1 temperature of A2199 changes from 4.7 keV to 4.3 keV with the fudge, corresponding to change from 2.08 to 2.15 , i.e d = +0.07 corresponds to 10% drop in temperature at 5 keV which is not allowed. WHAT IS WRONG HERE?




MOS2/pn cluster temperatures are consistent within 5% with the public effarea: no need for fudging the effarea shape much



MOS2/pn XCAL blazar spectral shapes are consistent using the public effarea: no need for fudging the MOS2 effarea shape



Yet the fudge renders MOS2 temperatures much below the public ones, while blazar spectral shapes do no change. WHAT IS WRONG HERE?


Conclusions


Public effarea yields consistent hard band temperatures for XMM, Chandra, Suzaku, BeppoSAX Public effarea yields too high fluxes for MOS Fudge makes MOS cluster temperatures 5-25% smaller Cluster temperature agreements are lost with fudge Fudge makes MOS flux into better agreement with pn Clusters and blazars consistent about fudge flux issue Clusters and blazars inconsistent about fudge spectral shape