XMM-Newton Serendipitous Source Catalogue: 3XMM-DR4

The XMM-Newton Survey Science Centre is pleased to announce the third generation XMM-Newton Serendipitous Source Catalogue, 3XMM-DR4.

The 3XMM-DR4 catalogue is the sixth publicly released XMM-Newton X-ray source catalogue produced by the XMM Survey Science Centre (SSC) consortium on behalf of ESA. It follows the 1XMM (released in April 2003), 2XMMp (July 2006), 2XMM (August 2007), 2XMMi (August 2008) and 2XMMi-DR3 (April 2010) catalogues: 2XMMp was a preliminary version of 2XMM. 2XMMi and 2XMMi-DR3 were incremental versions of the 2XMM catalogue.

The motivations for creating 3XMM-DR4 are threefold:

1 - A substantial (~50%) increase in the number of public observations that can be included
2 - Significant improvements to the XMM-Newton Science Analysis Software and developments with the calibration
3 - The opportunity to advance the scientific quality of the catalogue through enhancements to the processing algorithms. The key science-driven gains include:
i) Improved source characterisation and reduced spurious source detections
ii) Improved astrometric precision of sources
iii) Greater net sensitivity for source detection
iv) Extraction of spectra and timeseries for fainter sources, with improved signal-to-noise

Taken together, these elements yield a catalogue that is about 50% larger than its predecessor, 2XMMi-DR3, and is, overall, of better quality and accuracy.

By number of sources, 3XMM-DR4 is the largest X-ray source catalogue ever produced. 3XMM-DR4 complements deeper Chandra and XMM-Newton small area surveys, probing a much larger sky area.

The catalogue provides an effective dataset for generating large, well-defined samples of various types of astrophysical object, notably active galaxies (AGN), clusters of galaxies, interacting compact binaries and active stellar coronae, using the power of X-ray selection. The large sky area covered by the serendipitous survey also means that 3XMM-DR4 is a rich resource for exploring the variety of the X-ray source populations and identifying rare source types.

The catalogue contains source detections drawn from 7427 XMM-Newton EPIC observations made between 2000 February 3 and 2012 December 8; all datasets were publicly available by 2012 December 31 but not all public observations are included in this catalogue. The catalogue contains 531261 X-ray detections which relate to 372728 unique X-ray sources (covering an energy interval from 0.2 keV to 12 keV). About a third of the observations have features that may cause spurious detections (mainly the wings of bright sources and large extended emission), and it is strongly recommended to use a filter (either per source, based on the summary flag column, or per observation, based on the observation class column). The following table gives an overview of the statistics of the catalogue in comparison with the 2XMMi-DR3 catalogue.

  3XMM-DR4 2XMMi-DR3 Increment
Number of observations 7427 4953 2474
Number of 'clean' observations (i.e., observation class < 3) 4553 2789 1764
Observing interval 03-Feb-00 -- 08-Dec-12 03-Feb-00 -- 08-Oct-09 3.2 years
Sky coverage, taking overlaps into account ( ≥  1ksec exposure) 794 sq.deg 504 sq.deg 290 sq.deg
Number of detections 531261 353191 178070
Number of 'clean' detections (i.e., summary flag < 3) 432321 284270 148051
Number of unique sources 372728 262902 109826
Number of 'cleanest' (summary flag = 0, not in high-background fields) extended detections 7698 4775 2923
Number of detections with spectra 123867 56017 67850
Number of detections with timeseries 123860 56017 67843
Number of detections where probability of timeseries being constant is  <  1  × 10-5 4612 3177 1435

The median flux in the total photon energy band (0.2 - 12 keV) of the catalogue detections is ~ 2.4 × 10-14 erg cm-2 s-1; in the soft energy band (0.2 - 2 keV) the median flux is ~ 5.7 × 10-15, and in the hard band (2 - 12 keV) it is ~ 1.3 × 10-14. About 20% of the sources have total fluxes below 1 × 10-14 erg cm-2 s-1.


The 3XMM-DR4 catalogue, released on 2013 July 23rd, is available in several formats:

Please read the Watchouts section before using the catalogues.

3XMM_DR4cat_v1.0.fits.gz FITS binary 434 MB MD5:3af0835623bb8d13e53f01de363a0c91 full binary catalogue; recommended for more in-depth studies (includes information of all detections)
3XMM_DR4cat_v1.0.csv.gz CSV 477 MB MD5:1e357dcbaf18e68ee0bea07a667b736c full ascii catalogue
3XMM_DR4cat_slim_v1.0.fits.gz FITS binary 51 MB MD5:5342b75bd9485d5f82f3bc955ab1f406 reduced binary catalogue, suited for studies of unique sources (individual detection information is not included)
3XMM_DR4cat_slim_v1.0.csv.gz CSV 56 MB MD5:b8588e23c29113edcbe40f5399f57e5e reduced ascii catalogue

For convenience, there are suitable SQL CREATE statements available for use with a relational DBMS in order to load the data in CSV format:

The CSV file has nulls represented by two successive commas, compatible with Postgres and (probably) other DBMS.

The slimline version of the catalogue contains one row per unique source (while the the main catalogue has one row per detection) and thus has 372728 rows. There are 44 columns, essentially those containing information about the unique sources. The catalogue also contains a column with links to the LEDAS summary pages. In the case of sources with multiple detections the summary page of the best detection is selected (i.e., the detection with the largest exposure time, summed over all cameras), and the summary page gives cross-links to the other detections.

A separate file is also provided (table below) which contains key details about the observations used in the construction of the 3XMM-DR4 catalogue. See the list of observations for a description of the columns in the file.

3XMM-DR4_summary.fits FITS binary 0.8 MB MD5: 29200a29ff8add822c7108590c84f7f1 Summary table of observations used in the catalogue

The following Web-based user interfaces will allow filtering and searching of the catalogue (and give access to all or selected associated data products):

The SSC Catalogue Home Page (this page), XSA, and HEASARC allow download of the catalogue file in (binary) FITS.


The User Guide for the 3XMM-DR4 catalogue contains details of the catalogue production process and content. Here are quick links to


1) Shortly after release, an error was noted with the vignetting values in the 3XMM-DR4 catalogue (and the OMSRLI and OBSMLI pipeline product source lists), arising from a bug in the emldetect SAS task. For a given detection (row) in the catalogue, all values of the vignetting for a given instrument (i.e. columns <instrument>_<band>_VIG where <instrument> is PN, M1 or M2 and <band> is the energy band (1 to 5)) are identical and equal to the vignetting value at the source position, but for an energy of 0 keV rather than the intended mid-band energy. It should be emphasised that this error in the vignetting values does not affect the count-rates (or fluxes) - these are computed using the exposure maps in which the vignetting corrections, based on the correct band energies, are embedded. The error only affects the quoted vignetting columns in the catalogue and the corresponding entries in the OMSRLI and OBSMLI pipeline product files

A version of the catalogue with correct vignetting columns will be made available shortly. Corrected versions of the OMSRLI and OBSMLI source list files will also be generated and re-ingested in to the XSA in due course.

2) It has been established that the ODFs of a subset of the mosaic-mode observations have been incorrectly split in to sub-pointings. Eight parent observations (involving 63 sub-pointings and 3943 detections in the 3XMM-DR4 catalogue) are known to show clear temporal overlaps of data between sub-pointings, where there should be no such overlap. In these cases all sub-pointings and instruments are believed to be affected. A further ten parent cases (involving 26 sub-pointings, of which 19 (involving 975 detections) are in the 3XMM-DR4 catalogue) are known in which some of the sub-pointings are affected and in these, only MOS1 and/or MOS2 data are impacted. The cause appears to be related to incorrect time-tagging of events due to telemetry gaps, which the splitting process handled incorrectly.

These 18 mosaic-mode parent problem-case observations (encompassing 89 sub-pointings) are listed here. Two comma-separated-variable format files, one for the cases where all data are affected and one for cases where some data are affected, are provided which contain the DETID, OBS_ID, RA, DEC and INSTRUMENT(s) affected for detections from the relevant problematic sub-pointings.

In the cases where all data are affected, this problem has most obviously manifested itself in (visibly) erroneous timeseries for detections from the affected sub-pointings. However, it is important to emphasize that in these cases all the data/products are unreliable. For those cases where only some of the data are affected, the SOC believe that only timeseries data/products are affected and only for the MOS1 and/or MOS2 instruments.

When a full analysis of the problem is completed, the SOC and SSC will consider what corrective action may be applied in terms of the processed data, the catalogue and the associated data products. In the meantime, users are advised to disregard all data/products of detections from the observations where all data are affected (and any unique source data that involve them). Users should also disregard timeseries from those cases where some of the data are affected, though it is advisable to treat other products/data from these cases with caution too.

3) Whilst the 3XMM-DR4 catalogue values have been shown to be robust, due to an uninitialised variable in a subroutine of the source detection chain, a subset of the error values in the current version of the 3XMM-DR4 catalogue have been over-estimated, and a very small portion of these errors have been under-estimated. Only when a quantity is based on an error-weighted combination of values, such as the EPIC (EP) columns, are the non-error values affected. The quality of the catalogue is therefore assured, but the overall accuracy is in general better than that which is given in the current version.

The error values affected are for the counts, the rate, the flux, the hardness ratio and the position. The worst affected column is the error on the EPIC band 9 flux (EP_9_FLUX_ERR) where 10% of the errors are erroneous, some (~0.1%) by more than a factor 1000. Other columns are in general much less affected, with less than 2% of the errors showing erroneous values in many cases. The RA/dec errors (RADEC_ERR) are erroneous for about 1.4% of detections, with 80% of those values deviating by < ± 0.1 arcsecs. Work is being carried out to rectify the problem. However, until a new version can be produced, the columns affected by this bug are:

(key: ? implies bands 1, 2, 3, 4, 5, 8 and 9; # implies hardness ratios 1, 2, 3 and 4 and * means all columns starting with SC_ and finishing _ERR)

Users working on individual sources can re-run the source detection on their fields of interest with SAS 14 tasks which will be released in November 2014.

Nomenclature and Citation

With the new, 3XMM-DR4 catalogue, IAU name nomenclature includes the designator, 3XMM. The IAU name is provided in the catalogue. The correct, full nomenclature for a specific detection from the catalogues is the IAU name, followed by a colon and the detection identification number, that is:"3XMM Jhhmmss.sSddmmss:detid".


The production of the 3XMM-DR4 catalogue is a collaborative project involving the whole XMM-Newton SSC Consortium:

The SSC acknowledges the use of the TOPCAT and STILTS software packages (written by Mark Taylor, University of Bristol) in the production and testing of the 3XMM-DR4 catalogue.