The X-ray Atmospheres of Isolated Galaxies: A View from eROSITA and Implications for Galaxy Feedback Models

This video presents an analysis of eROSITA data by Chadayammuri et al. (2022) [1]. eROSITA is making the first all-sky X-ray survey since ROSAT thirty years ago. It has twice the spatial resolution and 30-50 times the sensitivity, allowing us to see the diffuse X-ray atmospheres of galaxies like never before. All but the brightest and nearest galaxies remain below the detection limit, but stacking a large number of them reveals the average properties of each sample. We stacked >2000 galaxies at 10.2 < $\log(M_*/M_\odot)$ < 11.2, selected from SDSS to overlap with eFEDS, the 140 square degree eROSITA test field that achieves the same exposure as the equatorial patch of the full-sky survey after four years. We mask out all point sources, groups and clusters, and correct for contamination from X-ray binaries, scattered out by the Point Spread Function. We find that star-forming galaxies are brighter than quiescent ones, and that luminosity increases with stellar mass. The luminosity increases with stellar mass for quiescent galaxies, but appears to remain flat for star-forming galaxies, whereas in the EAGLE and IllustrisTNG-100 (TNG) simulations, this is a steep function. The low-mass, quiescent galaxies are 3-5x brighter than in simulations, whereas at higher stellar mass they agree with EAGLE and brighter than TNG. The low-mass star-forming galaxies are 5-20x brighter than both simulations at 40-100 kpc; inside, they agree with EAGLE but are fainter than TNG. The high-mass, star-forming stack agrees best with both simulations, although we remind readers that this agreement depends on the choice of mass bin, given how much faster the luminosity-mass relation evolves in simulations than observations. The discrepancies can be alleviated if AGN feedback in the simulations is gentler, and supernova feedback more efficient. Alternatively, or perhaps in addition, increasing the rate chemical enrichment of the CGM in simulations could help.