XMM-Newton Users Handbook

3.4.4.7 RGS Sensitivity Limits

The main point of the RGS sensitivity is the line detection threshold. To get an estimate of the exposure time necessary to detect a line above a certain underlying continuum (at a given significance level) detailed simulations should be performed. Then the line flux integrated over the HEW of the line profile should be evaluated. For such an estimate, four contributing components must be considered:

• the flux of the line,

• the flux of the underlying continuum of the source,

• the X-ray background, and

• the particle background (see § 3.4.4.6).

At the discussed energies the extragalactic X-ray background can be well reproduced by a power-law spectrum. This allows us to consider the X-ray background simply by adding an additional component to the continuum flux of the source.

Various components may feature in the data from RGS observations, such as:

• Internal X-ray calibration sources. Four calibration sources permanently illuminate the CCDs of the RFC. The emission is primarily of F-K and Al-K. They are evident as four horizontal distributions in the bottom panel of Fig. 77. The source intensity is 0.1 counts cm$^{-2}$ s$^{-1}$. Using both spatial and energy information, their contribution to the celestial spectra can be estimated.

• Optical load on the CCDs. This is caused by optical straylight. Detected optical photons will modify the gain calibration of the CCDs through an introduction of an additional energy offset. Using diagnostic mode data, these offsets can be measured.

• Background. This component is explained in detail in § 3.4.4.6. The diffuse cosmic X-ray background can be modelled assuming an isotropic spatial distribution. The soft-protons induced background is more difficult to remove. For point sources, it can be estimated by selecting a region on the CCDs in the cross dispersion direction and using the same windows in the CCD pulse height as for the source. The background estimation for spatially extended sources or low surface brightness sources (like galaxy clusters) is more problematic as it is not so simple to select an acceptable ``empty'' region in the cross dispersion direction. The average background described in § 3.4.4.6 could be used for quiet periods, however, as some of the components (or model parameters) are clearly variable, this is not a trivial task. The uncertainty of this method is estimated comparing the different ``blank-sky'' exposures used to obtain the average spectrum and it is about 30%. In some cases, EPIC data could be used to directly measure the background. As a different alternative, for extended, irregular, sources the user may consider the need for a particular position angle that could guarantee an ``empty'' background region in the cross dispersion direction.

• Apparent cross-talk between orders. The CCD response includes a low energy tail due to a finite number of photons producing anomalously low energy signatures. In the case of bright emission feature being measured in the second order spectrum, the intensity of this tail may become significant, as it may produce counts in the co-located first order spectrum. These effects can be easily identified in PHA or PI energy versus dispersion plane. (cf. Fig. 77).

• Effects of scattering by the gratings. Due to X-ray scattering off the gratings (about 20% at mid band), there is an additional tail of the LSF. While true source continuum emission follows the dispersion equation (the inverse relation between photon energy and wavelength), scattered light appears as an horizontal distribution in the pulse height versus dispersion plot (cf. Fig. 77). This effect can be modelled and is included in the response generator of the SAS.