XMM-Newton Users Handbook

4.3.2 Other considerations

Some more parameters that might influence certain XMM-Newton observations, and should therefore be taken into account, are:

• Seam losses between abutted CCDs

  Although small, there are gaps between the chips of the different X-ray detectors on board XMM-Newton. The two EPIC MOS cameras are mounted orthogonal with respect to each other, so that in final images, after adding up the data from two X-ray telescopes, the gaps should not be visible after correction for exposure. There will only be a reduced total integration time in areas imaged at the location of chip boundaries. The pn camera has a different chip pattern, leading to minimal losses in other areas of the field of view. It is also offset with respect to the X-ray telescope's optical axis so that the central chip boundary does not coincide with the on-axis position. The inter-CCD gaps of the EPIC MOS chip array are 400 $\mu$m (11”) wide. Those between neighbouring CCDs within one quadrant of the pn chip array are 40 $\mu$m (1”.1) wide, the gaps between quadrants about 150 $\mu$m (4”.1).

  The nine CCDs in each RGS also have gaps of about 0.5 mm in between them. Table 9 lists the energies affected by the gaps in the RGSs. The two RGS units have an offset with respect to each other along the dispersion direction to ensure uninterrupted energy coverage over the passband. Due to operational problems with two CCDs there are two additional gaps, one on each RGS unit. The wavelength ranges affected are from 10.6 to 13.8 Å and from 20.0 to 24.1 Å in RGS-1 and RGS-2, respectively.

• “Out of time” events

  The XMM-Newton X-ray detectors do not have shutters and are therefore exposed to incoming radiation from the sky at all times . In order to prevent photon pile-up (see § 3.3.9), the CCDs are read out frequently. During readout, photons can still be received. However, they hit pixels while their charges are being transferred to the readout nodes, i.e., when they are not imaging the location on the sky they would normally observe during the exposure. Thus, events hitting them during readout are “out of time” (and also “out of place”; see § 3.3.10). One cannot correct for this effect in individual cases, but only account for it statistically.

  The MOS CCDs have frame store areas, which help suppress the effect of out-of-time events. The frame shift times of a few ms are much shorter than the maximum frame integration time of 2.8 s. Therefore, the surface brightness background of smeared photons is only a fraction of a percent divided by the ratio of the PSF size to the CCD column height.

  For pn the percentage of “out of time” events is 6.3% for the full frame mode, 2.3% for the extended full frame mode, 0.16% for the large window mode and 1.1% for the small window mode. The large window mode has a smaller fraction of “out of time” events because half the image height is used as a storage area, but the reduced smear is penalised by a loss in live time.

• EPIC pn caveat for bright low energy sources

  The lower event threshold of the EPIC pn camera constrains its spectral capabilities for sources which show a significant flux below the threshold. These are mostly nearby white dwarfs which have absorption column densities below $10^{20}$ cm$^{-2}$ and temperatures below $\sim 40$ eV.

  This effect is seen in observations of the white dwarf GD 153, that has a temperature of 25 eV, i.e. the bulk of the photons do not directly produce events above the threshold. Spectra of this star taken with various EPIC pn readout modes and filters yield large inconsistencies in the spectra below 0.5 keV. There is a strong correlation between count rates and read-out mode and filters. Slower read-out results in higher count rate and harder spectrum. The medium filter reduces the count rate more than expected from the thin/medium ratio.

  The most likely explanation for this effect is pile-up. Three kinds of pile-up at low energies are possible: two source X-ray photons, a source photon with electronic noise and a source X-ray photon with optical photons from the source. Pile-up can bring the energy of sub-threshold events above threshold. Source photons with energies below the threshold (which would nominally not be detected) have a high probability to "gain energy" by fortuitously adding to noise. For weak sources (and/or fast readout) this is most likely the dominant pile-up effect. No calibration is available to correct for this soft pile-up effect.

• EPIC pn frame rejection due to MIPs (Minimum Ionizing Particles)

  A certain fraction of EPIC-pn frames are rejected during the standard data processing due to Minimum Ionizing Particles (MIPs, Freyberg et al., 2006, XMM-SOC-CAL-TN-0067). The exact fraction depends on the level of background measured during an observation. It has been observed to be comprised between 3% and 8%. Users are recommended to take this effect into account in their estimation of the requested time.

• EPIC pn timing mode feature

  The EPIC pn camera operated in its timing mode shows at RAWY=19 a bright line in RAWX direction. The feature only shows up for bright point sources. Its origin is related to an on-board clock sequence feature whose effect was only noticed after launch.

  There is no effect on the scientific quality of the data as long as the integration time for spectra and light curves is longer than 5.9 ms. Care should be taken for pulse phase spectroscopy with bin sizes below 5.9 ms, but only if the pulse period itself is a multiple of the frame time (5.9 ms).

• EPIC MOS1 micro meteorite column

  The post-impact offset value in MOS1 timing mode is far too large for a meaningful correction to be possible. Users are therefore advised to discard the affected column and the adjacent ones from the accumulation of any scientific products. Please, refer to the SAS Watchout for a specific recipe on how to calculate effective area files (via arfgen) in this case.

• EPIC-MOS low energy noise in two peripheral CCDs

  Two peripheral CCDs of the MOS cameras (CCD4 on MOS1, and CCD5 on MOS2) are frequently affected by a low-energy (E$<$1 keV) noise plateau. The cause of this phenomenon is unknown, and currently under investigation. There is a general trend of increasing occurrence rate of noisy CCDs (see Figure 124). There is currently no way to selectively clean the noisy events. As of SASv9.0 a task (emtaglenoise) is available which allows users to tag all the events detected in an affected CCD. This issue does not affect the quality of the calibration for the nominal pointed target, and should be taken into account by proposers only with respect to serendipitous point-like sources or for the fraction or large extended diffuse emission falling on the affected CCDs. For more information on "noisy" CCDs users are referred to Kuntz and Snowden, 2008, A&A 478, 575.

  Figure 124: Percentage of observations when the labelled MOS1 (top) and MOS2 (bottom) CCD count rate within the energy range 200-1000 eV is larger by a factor of 4 than the mean of the count rate in the other CCDs (noisy CCD excluded) as a function of the revolution number. In each plot the curves of a "noisy" and of a "clean" CCD are shown.

  

• RGA rib scattering

  Light scattered off the stiffening ribs of the grating plates of the RGAs produces diffuse ghost images in the EPIC MOS FOV in the $\pm$Y direction (i.e., the RGS cross-dispersion direction). The intensity of these images is of the order of 10$^{-4}$ relative to the intensity of the focused image. For off-axis sources at azimuth angles corresponding to the $\pm$Y direction, the intensity of the ghost images increases to a few times 10$^{-4}$.

• OM exposures

  When using FAST mode, the target coordinates must be accurate at better than about 2 arcseconds to fit and track the target in the approximately 10”$\times$10” wide windows.

  FAST mode OM exposures require a successful Field Acquisition (FAQ). FAQ could fail in crowded fields, where many extended objects are present. This limits the possibility of performing - e.g. - high resolution OM timing of active galactic nuclei in clusters.

  Ring-like loops due to scattering of out-of-field bright stars (see Fig. 109) can heavily affect the detection of faint or extended sources at the boresight. This effect is mainly due to bright stars that happen to fall in a narrow annulus 12'.1 to 13' off-axis.

  OM Grisms produce spectra of all objects in the field of view, therefore the spectrum of the target of interest can be contaminated by the zero and first orders from other objects in the field. This problem can usually be avoided by selecting an adequate position angle for the observation.

  It has been noted that for bright sources observed in the OM fast mode window, when performed in conjunction with the pre-defined EPIC/RGS Imaging mode configuration (see Section 3.5.9.2), there can be discontinuities in the mean count rate levels of the fast mode time series between each exposure. This stems from the different levels of coincidence loss that occur in each exposure because their different window configurations give rise to differing frametimes. In fact, this effect is also witnessed in the photometry of bright objects observed in the central high-resolution window that is repeteadly (five times) exposed when using the pre-defined EPIC/RGS Imaging mode configuration, for the same reason, regardless of whether fast mode is operating.