XMM-Newton Users Handbook

3.3.6 EPIC filters and effective area

The next factor influencing the EPIC effective area, specifically in the low energy part of the passband, is the choice of the optical blocking filter. These filters are used, because the EPIC CCDs are not only sensitive to X-ray photons, but also to IR, visible and UV light. Therefore, if an astronomical target has a high optical flux, there is a possibility that the X-ray signal becomes contaminated by those photons. The resulting analysis of data would be impeded in four ways:

1. Shot noise on the optically-generated photo electrons will increase the overall system noise, and hence lower the energy resolution. Spectral fitting will be inaccurate, because the calibration files will assume narrower spectral lines than observed.

2. The energy scale will be incorrectly registered, because a nominally zero signal pixel will have a finite offset. For each optically generated photo electron which is registered, the energy scale shifts by about 3.6 eV. This is comparable with the accuracy with which brightest emission line features can be centroided. Consequently, contamination by visible light plays a crucial role in defining the proper energy scale. For the MOS this effect can in principle be reduced by ground processing taking into account the information on the zero level around the event which is also sent to the ground. For the pn, an offset map is computed before the beginning of each observation. This map contains also the shifts in the energy scale caused by optical photons. During the observation, the energy of each event is reduced by the value of the corresponding pixel in the offset map, before transmitted to the ground. By this technique the original energy scale is restored. Note, however, that very bright X-ray sources may contaminate the pn offset map; cf. § 3.3.2. The inclusion of such X-ray events in the offset map calculation, the so-called `X-ray loading', is discussed in the SOC document XMM-SOC-CAL-TN-0050 (available from http://xmm2.esac.esa.int/docs/documents/).

3. Excess signal and noise fluctuations can affect the detection efficiency as well, by disguising single pixel X-ray events as events split between pixels.

4. Optically-generated photo electrons can lead to a saturation of electron traps, changing (i.e. reducing) the charge transfer losses. Note that unlike items 1 to 3, this effect may be caused also by unrelated optically bright sources, which happen to irradiate the region on the detector between the X-ray source of interest and the readout node.

To prevent this, the EPIC cameras include aluminised optical blocking filters, and also an internal ``offset table'' to subtract the constant level of (optical) light or other systematic shifts of the zero level of charge measurements. For MOS the offset table values are fixed and the SAS task emchain/emproc are used to calculate the local changes in offset.

If these measures work perfectly, the above problems are minimised. The use of a thick blocking filter capable of minimising the optical light contamination for all scenarios will necessarily limit the softest X-ray energy response. Each EPIC camera is therefore equipped with a set of three separate filters, named thick, medium and thin. It is necessary for the observer to select the filter which maximises the scientific return, by choosing the optimum optical blocking required for the target of interest. At the GO's discretion a thinner filter could be used. In theory, due to the peaked optical response, a similar PSF core excising as used in pile-up cases (see § 3.3.9) might be applied to recover the desired spectra. Note, however, that the optical PSF is smoother than the X-ray PSF due to diffraction of optical light at the gaps between the XMM-Newton mirror shells. This analysis method has not yet been tested and in any case will only work for pn observations where related offset tables are available as well.

It should be noted that also an off-axis bright optical object will leak through the filters generating false X-ray events, which could contribute to degrading the effective telemetry bandwidth (see § 4.3.1).

The following guidelines apply to optical point sources (extended optical objects are not expected to be a significant problem). The optical loading is only important where a bright source is within $< 2$ arcmin of the target or along the EPIC CCD read-out direction.

The calculations below have been performed for a worst case, i.e., for the brightest pixel within the core of the PSF. Therefore, averaging the brightness of an extended object over a scale of one PSF (say, $20''$) should provide a corresponding estimate with a significant margin of safety. Note that these data apply to full frame modes only, and that a change to a partial window mode with an order of magnitude faster readout rate can allow suppression of optical contamination at 2-3 visible magnitudes brighter for ALL filters. The GO can make an estimate on optical contamination improvement based on the mode time resolution compared with full window mode (Table 3).

• Thick filter

  This filter should be used if the expected visible brightness of the target would degrade the energy scale and resolution of EPIC. It should be able to suppress efficiently the optical contamination for all point source targets up to $m_{V}$ of 1-4 (MOS) or $m_{V}$ of -2-1 (pn). The range depends on the spectral type, with only extremely red (M stars for example) or blue colours causing the change to 3 magnitudes fainter level.

• Medium filter

  The optical blocking is expected to be about 10$^{3}$ less efficient than the thick filter, so it is expected that this filter will be useful for preventing optical contamination from point sources as bright as $m_{V}$ = 6-9.

• Thin filter

  The optical blocking is expected to be about 10$^{5}$ less efficient than the thick filter, so the use of this filter will be limited to point sources with optical magnitudes about 12 magnitudes fainter than the corresponding thick filter limitations.

Figs. 29 and 30 display the impact of the different filters on the soft X-ray response of both types of EPIC camera, whereas Fig. 31 displays the combined effective area of all XMM-Newton X-ray telescopes.

Figure 29: The EPIC MOS effective area for each of the optical blocking filters.

Figure 30: The EPIC pn effective area for each of the optical blocking filters.

Figure 31: Combined effective area of all telescopes assuming that the EPIC cameras operate with the same filters, either thin, medium or thick.

The user might also want to take note of the SOC document XMM-SOC-CAL-TN-0001 (available from http://xmm2.esac.esa.int/docs/documents/) which explains in detail the PHS (Proposal Handling Subsystem) software tools, used during the technical evaluation and enhancement of a proposal to determine if the correct optical blocking filter has been chosen for a particular exposure. Technical notes on the assessment of optical loading for EPIC pn and MOS are available from the same URL as XMM-SOC-CAL-TN-0051 and XMM-SOC-CAL-TN-0043, respectively.