Документ взят из кэша поисковой машины. Адрес оригинального документа : http://xmm.vilspa.esa.es/docs/documents/CAL-SRN-0176-1-1.ps.gz Дата изменения: Wed Jul 28 18:35:00 2004 Дата индексирования: Mon Oct 1 22:29:43 2012 Кодировка: Поисковые слова: m 87 jet

XMM-Newton CCF Release Note
XMM-CCF-REL-176
RGS QUANTUMEF Shortest-Wavelength Correction Factors
A.M.T. Pollock
July 27, 2004
1 CCF components
Name of CCF VALDATE List of Blocks changed XSCS ag
RGS1 QUANTUMEF 0013 1998-01-01T00:00:00 RGA EFFAREACORR NO
SI1
SI2
SI3
SI4
SI7
RGS2 QUANTUMEF 0014 1998-01-01T00:00:00 RGA EFFAREACORR NO
SI1
SI2
SI3
SI4
SI5
SI7
2 Changes
Near the short-wavelength end of the RGS bandpass, the instrument model breaks down and the
detected count rate falls o more quickly with decreasing wavelength than expected from the scalar
diraction theory that forms the basis of the calibration. This is illustrated in Fig.1 which shows
a variety of uxed spectra from Mkn421. The ux decrease at the shortest wavelengths is clear.
What is less clear is where the trouble starts. Fig.2 shows details of the empirical approach adopted
to model the decit. By inspection, the spectrum falls into three parts: between 7  (  A) 10
the spectrum is OK and serves as the basis for extrapolation to shorter wavelengths; down to
  5:7  A the spectrum falls quite steeply; while at even shorter wavelengths, where no claims are
made for the reliability of the instrument model, the observed ux rst attens before becoming
1

XMM-Newton CCF Release XMM-CCF-REL-176 Page: 2
Figure 1: Mkn421 has been observed many times for calibration purposes with XMM. It is variable
both within and between observations. Shown here are averaged uxed spectra below 10  A calculated
by the SAS task rgsfluxer that combine the rst and second orders of RGS1 and RGS2. The
brightest, faintest and two intermediate observations are shown along with a total spectrum resulting
from 88 spectra in 22 observations.
entirely dominated by noise. The data below about 6  A are shown for completeness and are not
recommended for use.
Assuming no dierences between RGS1 and RGS2 and orders 1 and 2, the reliable spectrum be-
tween 7  (  A)  10 was modelled using the IDL svdfit routine with a combination of Legendre
polynomials that was extrapolated to shorter wavelengths to provide the reference ux for the decit
determination. The extrapolation follows the data a little further down to 6.7  A, close to the Silicon
edge, where the decit seems to start. The observed decit was smoothed in turn with two further
sets of Legendre polynomials as shown by the green and blue stretches of data in Fig. 2.

XMM-Newton CCF Release XMM-CCF-REL-176 Page: 3
Figure 2: The empirical ux reduction factor in red calculated from the mean uxed spectrum of
Mkn421. The green and blue stretches of data delineate wavelength ranges in which the decit
apparently behaves dierently as modelled by separate sets of polynomial coeфcients.

XMM-Newton CCF Release XMM-CCF-REL-176 Page: 4
The changes described above concern the rst use of the RGA EFFAREACORR extensions for
supplying empirical short-wavelength correction factors to the eective area model. At the same
time the opportunity has been taken to bring up-to-date data concerning the thicknesses of some
CCDs held for completeness in the MTHICK header keywords of the extensions [SI1,...,SI9]. These
data are not used currently in the calibration model. The CCD PI redistribution model relies instead
on a method parameterized according to independent data held in the RGS% REDIST CCFs.
3 Scientic Impact of this Update
The wavelength range between 6 and 7  A contains important lines of SiXIII and SiXIV which,
especially in the case of the He-like triplet at 6.648,6.688,6.740, are often strongly detected in
RGS spectra.
4 Estimated Scientic Quality
This CCF update should allow users to assign SiXIII or SiXIV line uxes that are reliable to 10 or
20% or to model spectra down to 6  A with a similar level of reliability.
5 Test procedures & results
Although the CCF update is expected to be useful for the analysis of He-like and H-like lines of
Silicon, the quality of the correction is best assessed through analysis of smooth continuum blazar
spectra which are likely, over the restricted wavelength range of interest here, to be well described
by power-laws. Fig. 3 shows the XSPEC power-law models derived using response matrices based
on the new RGS QUANTUMEF CCFs for one observation of PKS2155-304 and three of Mkn421.
With the new corrections, the power-law models work quite well even below 6  A.
6 Expected Updates
None.

XMM-Newton CCF Release XMM-CCF-REL-176 Page: 5
Figure 3: Some example XSPEC analysis of blazar spectra using response matrices calculated with
the new short-wavelength correction CCFs. Short wavelength parts of RGS1 and RGS2 1st (black
and red) and 2nd order (green and blue) spectra are compared with power-law models of one
observation of PKS2155-304 and three of Mkn421, which attens as it brightens. The three Mkn421
spectra are the brightest, the faintest and an intermediate case. The best-t power-law indices, ,
are shown. The absorbing column densities were xed at their galactic values as indicated, although
these are of little relevance at such high energies.