XMM-Newton slew survey Source Catalogue, version 3.0

The median flux in the total photon energy band (0.2-12keV) of the catalogue detections is ~2.7x10^-12^erg/cm^2^/s; in the soft energy band (0.2-2keV) the median flux is ~1.1x10^-12^erg/cm^2^/s, and in the hard band (2-12keV) it is ~9.0x10^-12^erg/cm^2^/s. It is possible that there are more counts in the soft or the hard band than in the full band. This is because detections are carried out in each band and then combined using a match radius of 30 arcseconds to identify sources detected in more than one band.

The match radius is approximately 4 sigma and is reasonable for point sources because of the scarcity of sources (1 detection per square degree on average) at the flux levels probed by the slew survey. Therefore the extraction region maybe placed at slightly different positions, resulting in this incoherance. This also explains why band 8 (full band) may have many more counts than the sum of the soft and the hard band.

The catalogue is manually screened to identify possibly spurious detections. We recommend to filter the catalogue on the VAL_FLAG column, keeping only those detections with the CLEAN_SAMPLE qualification to obtain a clean catalogue. The XMMSL3 catalogue has a new major version, as there are twice as many sources as in XMMSL2 and the recommended method for producing a cleaned catalogue has been modified with respect to the previous version, XMMSL2, notably through filtering the catalogue on the VAL_FLAG column, keeping only those detections with the CLEAN_SAMPLE qualification.

However, in other respects, it is an incremental version, retaining all detections in XMMSL2 and adding a further 8.5 years of data. Since XMMSL2, there have been a number of revisions made to the source detection tasks and notably to the SAS task emldetect. All detections are tested for their potential spatial extent during the fitting process.

The instrumental point-spread function (PSF) is convolved with a {beta} extent model, fitted to the detection, and the extent likelihood is calculated as described by Section 4.4.4 of Watson et al. 2009A&A...493..339W). A detection is classified as extended if its core radius (of the {beta}-model of the PSF), rc>6arcsec and if the extended model improved the likelihood with respect to the point source fit such that it exceeded a threshold of L_ext,min_=4.

In the XMMSL2 version, when a detection was identified as pointlike, the extent maximum likelihood (EXT_ML_B6, EXT_ML_B7 or EXT_ML_B8) was set to -999. Since version 7.5.2 of emldetect, a negative maximum likelihood is attributed to point-like detections. L_ext,min_>-=4 indicates that a source is probably extended, whilst negative values indicate a clear preference of the point-like over the extended fit.

A minimum likelihood difference of four has been chosen to mark a detection as extended. This threshold makes sure that the improvement of the extended over the point-like fit is not only due to statistical fluctuations but from a more precise description of the source profile. Where no detection in a particular band has been made, there is no extension maximum likelihood provided for that band.

Therefore, the extension maximum likelihood column is composed of positive and negative values, as well as the value -999 and nulls.