Документ взят из кэша поисковой машины. Адрес оригинального документа : http://xmm.vilspa.esa.es/docs/documents/CAL-SRN-0304-1-1.pdf Дата изменения: Mon Oct 21 17:10:57 2013 Дата индексирования: Mon Mar 3 13:09:44 2014 Кодировка: Поисковые слова: http astrokuban.info astrokuban

XMM-Newton CCF Release

Page:

1

XMM-Newton CCF Release Note XMM-CAL-SRN-0304 Post-XRL Rate-Dep endent CTI correction for EPIC-pn Timing M o de
Matteo Guainazzi 29 July 2013

1

CCF comp onents
Name of CCF EPN CTI 0029.CCF VALDATE 2000-01-01 EVALDATE Blocks changed RATE DEPENDENT CTI XSCS flag NO

This CCF constituent includes an update of the Rate-Dependent CTI (RDCTI) correction for EPIC-pn Timing Mo de. This change is required by the implementation of the "X-Ray Loading (XRL)" correction, described in an accompanying CCF Release Note (Guainazzi & Smith 2013). It has been recently discovered that almost all the offset maps taken in scientific filters prior to exposures in EPIC-pn Fast Mo des (Burst and Timing) are contaminated by the pointed source (see Guainazzi et al., 2012 for a discussion of this effect). XRL distorts the energy scale in PHA space. This means that the calibration of the RDCTI publicly available so far was based on data, whose energy scale was corrupted in a rate-dependent way. This CCF fixes this issue. The RDCTI has been now calibrated using XRL-corrected spectra. As of the 23r d of May 2012, offset maps prior to EPIC-pn exposures in Fast Mo des (Burst and Timing) are taken in CLOSED filters to avoid XRL. The recalibration of the RDCTI is adequate for these exposures as well.

2

Changes

The principle of the RDCTI calibration are described in the original CCF RN (Guainazzi et al. 2009). Readers are referred to this do cument for a description of the data reduction and analysis, as well as of the algorithm. The RDCTI calibration embedded in this CCF constituents was based on an updated sample of 49 sources with featureless spectra in the 1.5­3 keV energy band, covering the


XMM-Newton CCF Release

Page:

2

Rate dependent CTI - pn timing mode
1.10

1.05 Gain

1.00

0.95

0.90 1

fit: y=a0(xa1)+a2 a0= 0.0010(0.0002) a1= 0.6404(0.0303) a2= 0.9940(0.0013) 10 100 Shifted electrons/pixel/sec

Figure 1: RDCTI calibration curve in EPN CTI 0029.CCF. The solid line represent the best-fit function (inset label): the dashed lines represent the envelope of the same function corresponding to the 1 error on the best-fit parameters (in brackets in the inset). a0 0.0010 ± 0.0002 a1 0.640 ± 0.030 a2 0.9940 ± 0.0013

Table 1: Parameters of the RDCTI calibration function (and 1 statistical uncertainties) in EPN CTI 0029.CCF

whole mission up to April 2012. The sample do es not include either sources with a count rate exceeding the pile-up limit (800 counts per second on the 9-pixel RAWY extraction region), nor observations whose ODF do es not contain an offset map, because the XRL correction is not reliable in these cases (Guainazzi & Smith 2013). The RDCTI calibration curve is shown in Fig. 1. The data points were fit with the function: G = a0 (Ne )a1 + a2 , where G is the gain factor (calculated around the Si and Au instrumental edges, i.e. around 2 keV), Ne is the total number of shifted electrons (a proxy for for the source count rate), and ai are fit parameters. The best-fit parameters are listed in Tab. 2 (the CCF contains only the best-fit values, not the asso ciated statistical uncertainties).


XMM-Newton CCF Release

Page:

3

3

Scientific Impact of this Up date

Together with the asso ciated calibration of the XRL correction (Guainazzi & Smith 2013), the update of the RDCTI calibration is intended to finalise a full recalibration of the energy scale in EPIC-pn Timing Mo de.

4

Estimated Scientific Quality

Together with the asso ciated calibration of the XRL correction (Guainazzi & Smith 2013), the update of the RDCTI calibration is expected to yield a reconstructed energy scale in EPIC-pn Timing Mo de better then 25 eV on most of the calibrated energy bandpass (0.7­10 keV) and for any level of count rate below the pile-up limit (800 total counts per second). Doubts have been expressed on the accuracy of the energy-dependence of the RDCTI (Walton et al. 2012). As this correction is calibrated at 2 keV, a wrong energydependence could yield significant errors at the iron line energies. Admittedly, the true origin of the RDCTI is unknown. It is therefore possible that the assumptions underlying its algorithm could yield mistakes in the energy reconstruction at energies different from those where the RDCTI had been calibrated. We address this point from an experimental perspective in §5­6. If the rate-dependent correction to the energy scale is due to a CTI effect, and taking into account the known dependence of EPIC-pn CTI with energy as measured through the calibration source (K.Dennerl, in preparation), we estimate that the systematic error due to the energy dependency of the RDCTI algorithm is lower than 10 eV. This is within the residual systematic uncertainties of the energy scale in EPIC-pn Timing Mo de. On the other hand, not applying the RDCTI (i.e., not correcting the EPIC-pn Timing Mo de event list with epfast) may yield errors on the energy scale as large as a few hundreds eV. Extreme care should be exercised in deriving scientific conclusions on RDCTI-uncorrected spectra, especially in case of complex continua, because the spectral deconvolution could be completely falsified.

5

Test pro cedures and results

The performances of the new RDCTI have been tested using two metrics: - the deviation between the energy scale in the data and in the calibration at the Au edge, applying the same algorithm to calibrate the RDCTI on spectra extracted from epfast-corrected event list. The results are summarised in the top panel of Fig. 2. The energy scale is accurate within ±10 eV up to count rates below the pile-up threshold, with a scatter of ±20 eV - the measured energies of atomics transitions over the whole energy range between 1


XMM-Newton CCF Release

Page:

4

Figure 2: Top panel: accuracy of the e panel: percentage difference between transitions observed in the spectrum The red crosses indicate the observed

nergy scale at the Au edge (2.3 keV) as a function of Ne . Bottom the centroid and the laboratory energy for a number of atomic of the outbursting nova RSOph (Obs.#0410180101; blue points). spectrum (scale on the left y-axis)


XMM-Newton CCF Release

Page:

5

and 8 keV in the spectrum of the outbursting nova RSOph observed by XMM-Newton in February 2006 (Obs.#0410180101; cf. the bottom panel in Fig. 2). The difference between the energies measured by the EPIC-pn Timing Mo de spectrum and the laboratory energies of the closest transition is within 10 eV from Si to Ni. A larger difference (25 eV) is measured for MgXi and Mgxii. It is possible that uncertainties in the spectral deconvolution asso ciated to the degraded resolution play a role. At lower energies Ne ix and Nex could constrain the energy scale. It is, however, difficult to constrain these features given the level of photo electric absorption in this source (NH 8в1021 cm-2 ).

6

Exp ected up dates

Guainazzi (2013) describes another algorithm to correct for the intrinsic rate-dependence of the energy scale: the Rate-Dependent PHA correction (RDPHA). It is intended to eventually replace the RDCTI, because it acts on the spectra in PHA, and it is independent on any astrophysical assumptions. Its calibration is, however, not mature enough yet. In the meantime, the RDCTI correction have been therefore updated, not to delay unnecessarily the public delivery of an energy scale correction consistent with the XRL correction. Studies are ongoing to make the RDPHA publicly available in a SAS version later than 13.0.2. A study is ongoing to directly measure the energy-dependence of the RDCTI using a sample of X-ray binaries exhibiting absorption lines in the Fe band. This study may lead to a further enhancement of the RDCTI (or of the RDPHA) in a future SAS version.

7

References

Guainazzi M., 2013, XMM-SOC-CAL-SRN-0299 (available at: http://xmm2.esac.esa.int/docs/documents/CAL-SRN-0299-1-1.ps.gz) Guainazzi M. & Smith M., 2013, XMM-SOC-CAL-SRN-0302 (available at: http://xmm2.esac.esa.int/docs/documents/CAL-SRN-0302-1-5.pdf) Guainazzi M., et al., 2012, XMM-SOC-CAL-TN-0083 (available at: http://xmm2.esac.esa.int/docs/documents/CAL-TN-0083.pdf) Guainazzi M. et al., 2009, XMM-CCF-REL-256 (available at: http://xmm2.esac.esa.int/docs/documents/CAL-SRN-0256-1-0.ps.gz) Walton D., et al., 2012, MNRAS, 422, 2510