Understanding quasar host galaxies

Low redshift quasars in the SDSS Stripe 82. The host galaxies

Faloma et al. (2014), arXiv: http://arxiv.org/abs/1402.4300


The phenomenon of Active Galactic Nuclei (AGN) are understood to be the result of matter accreting onto a central super massive black hole (SMBH) in the nucleus of a host galaxy. Quasars are a very energetic and distant form of AGN. Quasars are impressive in their luminosity, as they can outshine the entire galaxy in which they reside, and yet this luminosity originates from a comparatively small place. Typically, a quasar accretion disk is about 1 light-week in diameter. It may well be that galaxies have phases of extreme activity in their nuclei, indicating AGN are an evolutionary phase of galaxy growth. This inexorably links activity in the nucleus to the overall growth and evolution of the galaxy itself. A number of questions are raised based on this logical connection surrounding AGN: where does the fuel come from? how does an ‘active phase’ start? what turns it off? is there feed back?
In order to answer these questions, we need to study both the nucles of activity as well as the surrounding host galaxy in which it resides. Unfortunately, in the case of quasars, the galaxy is outshone by the nucleus by as much as 10x or 100x. This problem is even more difficult with high luminosity, high redshift quasars.
The Sloan Digital Sky Survey (SDSS) offers an opportunity to study a large sample of well understood quasars. SDSS contains 105 783 quasars in DR7. However, most of these are too faint to disentangle the host galaxy from the bright AGN. Luckily, Stripe 82 is available, which was imaged >60 times, greatly improving the depth of the imaging.

stripe map

12 434 quasars in Stripe 82 DR7


After coadding all available images from Stripe 82 for each object, the different in depth is notable.

The Sample

The author’s used the low redshift quasars in SDSS Stripe 82 (S82) found in the fifth release of the SDSS Quasar Catalog (Schneider et al. 2010). Two cuts made. First, avoid objects that are closer than 2 degrees on the sky. Second, chose upper limit of z=0.5 in order to be able to resolve the host. After all these cuts, 416 quasars. Dominated by radio quiet (24 are radio loud).


The distribution of quasars in the redshift-absolute magnitude plane. The mean redshift is 0.39. Red are radio loud, blue are radio quiet.

Image Analysis

Assume the object is a superposition of two components: nucleus (AGN) and host galaxy. The nucleus would be described by a Point Spread Function. The host galaxy they assume a galaxy model described by a Sersic law convolved with the proper PSF. The analysis was then done using something called the Astronomical Image Decomposition Analysis (AIDA).
Need to determine a suitable PSF. Luckily there are a bunch of stars in the fields of the quasars in question. The author’s selected between 5 and 15 stars nearby to the quasar+host in question. There’s a lot of effort into doing this, but in the end they simultaneously fit the stars using a 2D model (gaussian + exponential).


Modelling the PSF. a) selection of stars, b) example of definition of fit area and background, c) masked areas to avoid contamination, d) example of the model fit to radial brightness profile of one selected star.

Once the average PSF of the image is found (using 5 to 15 stars) the authors fit the quasars using both a scaled version of this PSF and a 2 component model (point source plus galaxy).


Top left: fitting the quasar with the PSF determined earlier. Top right: fitting the quasar with a PSF+Sersic model of an underlying galaxy.

In the paper, the authors refer to ‘resolved’ nucleus/host as those where the PSF can be disentangled easily from the Sersic model. Further, ‘unresolved’ indicates the PSF and the Sersic model are indistinguishable. Note that 46/60 quasars that are ‘unresolved’ are those that are above redshift of 0.4, indicating an angular separation/luminosity issue. Cuts of ‘resolved’ vs. ‘unresolved’ vs. ‘marginally resolved’ were done by looking at the $\chi^2& of each fit.

Host Galaxy Properties

Colour and K-corrections were applied assuming a composite quasar spectrum (Francis et al. 1991) for the nucleus and an elliptical-type SED for the host. Doing this you can measure the absolute magnitude of the host galaxy, the following figure is made:


Absolute magnitude vs. redshift. Green triangles come from a separate study of 100 quasars by HST for comparison.

Controversial topic: relationship between the nucleus and host galaxy luminosity
Assuming quasars emit in a relatively narrow range of Eddington ratio, and that the BH mass is correlated with the mass of the galaxy one would expect to find a correlation between nucleus and host galaxy luminosity. however they do not. they find no significant correlation between the two quantities.


Comparison of the host luminosity vs. nucleus luminosity. The blue line refers to the loci of fixed ratio of 1, 2.5, and 6.25 (i.e., the same, 1 mag, 2 mag).

Do quasars inhabit both disc and bulge dominated galaxies? When HST came online, it was able to show that quasars are in both ellipticals and spirals. Stating the morphology of a host galaxy is very difficult, even with the above analysis ‘resolving’ the nucleus from the host. In order to classify the galaxies, the author’s used the value of the Sersic model from the best fits done above. They also redid all the fits as if the galaxy was an elliptical, and all as if the galaxy was a disk. This, however, can only do a preliminary indication, since BOTH components can be present. So they tempered this with visual inspection of all possible available data: images, contour plots, fit of brightness profile, ellipticity following Nair & Abraham (2010) to classify the morphological type. Results:
113 dominated by bulge component
129 dominated by disk structure
64 exhibited mixed bulge+disk equally
100 exhibited complex features
More detailed analysis of the quasars host galaxy morphology is in forthcoming paper.

Black Hole Mass vs. Host Galaxy Relationship

Black hole masses correlate with: stellar velocity dispersion, the luminosity of host galaxy, mass of spheroidal component. The author’s investigated these relationships for low redshift quasars. The author’s investigate these claims at low redshift. The BH mass is taken from Shen et al. (2011, see section 3.7), who measured the virial mass from H\beta /luminosity relationships.


Black hole mass vs. absolute magnitude of host galaxy

See no correlation really, argue this is because the ‘abs mag of host galaxy’ includes both spheroidals, disks, and disk+spheroidals. Check any redshift dependence?

sees to be no evolution from redshift 0.2 to 0.5


BH mass vs. absolute magnitude of the bulge component of the host galaxy only. This is also supported by the 25 lowest redshift quasars in the sample who’s BH mass is measured via reverberation mapping


The authors investigated 416 galaxies that host quasars in Stripe 82. Using a 2D image analysis they were able to well resolve the quasar from the host fro approximately 75% of the sample; selected findings:
1. Morphology of the host galaxies turned out to be rather complex with both bulge and disk dominated galaxies, about one third of the objects in sample show both bulge/disk components.
2. Irrespective of host morphology the size of galaxy ranges from compact to extended (3-15 kpc). no trend of galaxy size with redshift is mentioned.
3. Nuclear and host galaxy luminosities do not correlate
4. BH mass (from H\beta considerations) poorly correlates with total luminosity/mass of WHOLE host galaxy, but correlates with the bulge luminosity, though not particularly strongly.

Posted in Research and tagged as

Comments are closed.