XMM-Newton Science Analysis System: Users' Guide

4.10.2 Creating response matrices

Analysis of EPIC data products is generally performed by specialised software packages including Xspec [17] , Ximage [19] or Xronos [18] (http://heasarc.gsfc.nasa.gov/docs/xanadu/xanadu.html). In addition to the calibrated products, some of these packages require the generation of specific files. In particular, the spectral fitting technique used by Xspec requires a characterization of the EPIC detector response to simulate an output spectrum observed by EPIC. The response function gives the probability that an incoming photon of energy E will be detected in a channel I. This discrete function can be calculated as a product of a Redistribution Matrix File (RMF) by an Auxiliary Response File (ARF). These response files shield the user from the complexity of the EPIC instrument response which varies across the field of view.

There are currently two approaches to obtain RMF redistribution matrix files:

1. The user can make use of ready made (canned) response matrices made available by the EPIC team and accessible through the EPIC Response Files page at
   http://xmm.esac.esa.int/external/xmm_sw_cal/calib/epic_files.shtml.
   They are virtually identical to the files produced by the SAS task rmfgen.

   Special care must be taken in choosing the appropriate canned RMFs as they depend on the readout mode, the pattern selection, the observation date and the distance from the readout node where the spectrum was extracted.

2. The user can also create the RMF using the rmfgen task (even though rmfgen might take some time to complete, depending on the hardware). The input spectrum file contains the necessary ancillary information to allow the correct response to be made. The rmfgen task can then reformat the detector response and energy bounds according to the information provided by the calibration access layer. It corrects for instrumental effects specific to the spectrum and writes the result to a specified dataset. It groups response data above a threshold value.

The RMFs are spatially dependent for both MOS and pn. If source extraction regions are large, e.g. for extended sources, timing mode data or complex regions containing many excluded sources, it is important to specify an appropriate number of detector map bins to allow the SAS to calculate an average response matrix (also see § 4.10.4). It is recommended to use 160 bins in each dimension by:

rmfgen spectrumset=<spectrum_file> rmfset=<rmf_file> detxbins=160 detybins=160

The ARF response file of the EPIC camera shall then be generated by the task arfgen. This task calculates an effective area curve as a function of energy, to be used in conjunction with the RMF file generated before. For each row of the RMF there is a corresponding element in the 1-D ARF. This is normally adjusted by specifying the previously generated response matrix as an input file to the arfgen task.

The arfgen task generates an ARF file taking into account the following effects:

1. Telescope effective area including vignetting by the RGA structure for the MOS cameras,
2. EPIC filter transmission,
3. EPIC CCD quantum efficiency,
4. Complex region and pattern selections,
5. Fraction of the PSF in the accumulation region (including chip gap, bad pixel and out of observing window effects),
6. Out-of-time events smearing (pn).

The above effects generally depend on the source position in the EPIC field of view. Spatial response variation over an extended source is also taken into account.