A catalog lists 350 possible ultraluminous X-ray sources, or ULXs, in HECATE galaxies. It separates them into 90 main-sample candidates, including 53 first-time identifications, and 260 extended-sample candidates, including 245 first-time potential identifications. The total is a candidate count, not a confirmation that every entry is a genuine ULX.
The two groups reflect different levels of detection evidence. Candidates with a detection-likelihood score above 20 formed the main sample. Those scoring from 10 to 20 were placed in the extended sample, which is pending confirmation. The full list should therefore be read as two levels of candidate evidence, not a settled census.
A broad search, then a tight filter
Researchers combined the eRASS1 main catalog, the source list from eROSITA's first survey, with HECATE positions and distances. The starting data included 930,203 point sources detected between 0.2 and 2.3 keV and 204,733 galaxies. They kept sources inside HECATE's D25 ellipses, the galaxy regions used for matching, and excluded any source within 12 arcseconds of a galaxy's D25 core. That left 10,474 sources in 866 galaxies for the next stage.
For each remaining source, the team modeled its flux with an absorbed power-law spectrum, using a fixed slope of 1.7 and an absorption column of 10^21 cm^-2. It then kept sources whose modeled luminosity in the 0.2 to 5.0 keV band exceeded 10^39 erg per second. This initial cut produced 838 preliminary candidates in 784 galaxies, before contaminant removal and final screening.
At least two distinct astronomers, assigned at random, manually reviewed every candidate. The review and likelihood thresholds produced the two cleaned catalogs. Because the extended sample covers likelihoods from 10 to 20 and awaits confirmation, it remains a lower-confidence set of possibilities rather than an equally secure second batch of ULXs.
The survey could not see all candidates equally well
To estimate how often the X-ray search would recover sources, the study ran SIXTE simulations with eROSITA observing data. It placed 105 artificial sources at random positions, assigned fluxes from 10^-16 to 10^-11 erg s^-1 cm^-2 and used the same assumed spectrum, with a slope of 1.7 and absorption of 10^21 cm^-2. The eSASS detection software then tried to find them. This tested the X-ray selection, rather than the completeness of the galaxy catalog.
The simulated X-ray selection was complete within about 7 megaparsecs for ULXs and out to about 70 megaparsecs for sources labeled HLXs, short for hyperluminous X-ray sources. These distances describe the modeled reach of the X-ray search. They do not show that HECATE is complete across its full galaxy catalog.
Even within the galaxies covered by the survey, the usable area varied. The effective search area was zero for about 36% of observed galaxies, while at least half the area was covered for about 22%. The measure depends on the D25 geometry and the 12-arcsecond nuclear exclusion, so candidate counts need to be read alongside how much of each galaxy was actually searched.
Contamination and blends remain possible
One uncertainty is contamination by unidentified AGN, or active galactic nuclei, that fall in the search areas. Separate modeling predicted 26 such interlopers in the main sample and 38 in the extended sample, 64 in all. That corresponds to estimated contamination fractions of 29% for the main sample and 18% for the combined catalog. The authors describe those fractions as conservative upper limits, not direct measurements.
Another issue is source confusion, when an apparent point source may include emission that eROSITA cannot separate. The analysis placed the transition between point-source and extended-structure confusion at roughly 13 Mpc. At least about 95% of candidates exceeded the luminosity expected from their host galaxies, and about 40% exceeded it by at least 1 dex, meaning a full step on a logarithmic luminosity scale. Even so, the data could not rule out multiple luminous X-ray binaries appearing as one source.
Earlier catalogs hint at changing brightness
A comparison with earlier ULX catalogs added a possible clue about changing brightness. Thirteen of the 90 main-sample candidates had prior observations but had not previously been classified as ULXs. In the other direction, several candidates listed earlier, whose reported fluxes suggested eRASS1 should have detected them, were absent from the new survey. The authors interpret this pattern as variability over months to years.
That is an indirect cross-catalog inference, not a direct measurement of brightness changes in each source. It does not establish that every candidate is variable.
A target list, not a final inventory
The result is best read as a starting point for follow-up and population work. The main catalog offers the higher-likelihood group, while the extended catalog remains a lower-confidence list awaiting confirmation. Any population analysis will need to account for simulated X-ray completeness, uneven search area, modeled AGN upper limits and possible unresolved blends.
The main and extended catalogs, along with the identified-contaminant list, are available through the online journal, eROSITA DR1 and CDS. Follow-up remains necessary to establish which candidates are genuine ULXs and whether any apparent source contains multiple luminous X-ray binaries.
Paper data and sources
Original title: Ultraluminous X-ray sources in the first eROSITA survey I. Candidate catalogs
Authors: P. Weber, T. P. Roberts, S. Hämmerich et al.
Journal/Repository: Astronomy and Astrophysics, 712, A112 (2026)
Status: Peer-reviewed
First online: 2026-08-26
DOI: 10.1051/0004-6361/202659942
Original paper · Full text