Narrow-line Seyfert 1 (NLS1) galaxies have low-mass black holes and mass accretion rates close to (or exceeding) Eddington, so a standard blackbody accretion disc should peak in the extreme ultraviolet. However, the lack of true absorption opacity in the disc means that the emission is better approximated by a colour temperature corrected blackbody, and this colour temperature correction is large enough (∼2.4) that the bare disc emission from a zero spin black hole can extend into the soft X-ray bandpass. Part of the soft X-ray excess seen in these objects must be intrinsic emission from the disc unless the vertical structure is very different to that predicted.
None the less, this is not the whole story even for the extreme NLS1 as the shape of the soft excess is much broader than predicted by a bare disc spectrum, indicating some Compton upscattering by warm, optically thick material. We associate this with the disc itself, so it must ultimately be powered by mass accretion. We build an energetically self-consistent model assuming that the emission thermalizes to a (colour temperature corrected) blackbody only at large radii. At smaller radii the gravitational energy is split between powering optically thick Comptonized disc emission (forming the soft X-ray excess) and an optically thin corona above the disc (forming the tail to higher energies). We show examples of this model fit to the extreme NLS1 RE J1034+396, and to the much lower Eddington fraction broad-line Seyfert 1 PG 1048+231. We use these to guide our fits and interpretations of three template spectra made from co-adding multiple sources to track out a sequence of active galactic nucleus (AGN) spectra as a function of L/LEdd.
Both the individual objects and template spectra show the surprising result that the Compton upscattered soft X-ray excess decreases in importance with increasing L/LEdd. The strongest soft excesses are associated with low mass accretion rate AGN rather than being tied to some change in disc structure around Eddington. We argue that this suggests a true break in accretion flow properties between stellar and supermassive black holes.
The new model is publicly available within the xspec spectral fitting package.