Next Contents Previous

B.10.3. Lensed Quasar Host Galaxies

One advantage of studying lensed quasars is that the lens magnification enormously enhances the visibility of the quasar host. A typical HST PSF makes the image of a point source have a mean surface brightness that declines as R-3 with distance R from the quasar. Compared to an unlensed quasar, the host galaxy of a lensed quasar is stretched along the Einstein ring leading to an improvement in the contrast between the host in the quasar of µ2 for an image magnified by µ - you gain µ3 by stretching the host away from the quasar and lose µ because the quasar is magnified. Perpendicular to the Einstein ring, the contrast becomes a factor of µ worse than for an unlensed quasar. Since the alignment of the magnification tensor relative to the host changes with each image, the segment of the host where contrast is lost will correspond to a segment where it is gained for another image leading to a net gain for almost all parts of the source when you consider all the images. The distortions produced by lensing also mean the host structure is more easily distinguished from the PSF. In a few cases, like SDSS0924+0219 in Fig. B.58, microlensing or substructure may provide a natural coronograph that supresses the flux from the quasar but not that from the host. Despite naive expectations (and TAC comments), the distortions have little consequence for understanding the structure of the host even though a lens model is required to produce a photometric model of the host.

The only extensive survey of lensed quasar hosts is that of Peng ([2004]). Fig. B.73 shows the example of PG1115+080, a zs = 1.72 radio-quiet (RQQ) quasar. The Einstein ring image is easily visible even in a short, one-orbit exposure. For comparison, we also took the final model for the quasar and the its host and produced the image that would be obtained in the same time if we observed the quasar in the absence of lensing. It is quite difficult to see the host, and this problem will carry through in any numerical analysis.

Figure 73

Figure B.73. The host galaxy in PG1115+080. The top left panel shows the 1-orbit NICMOS image from Impey et al. ([1998]). The top right panel shows the lensed host galaxy after subtracting the quasar images and the lens galaxy, The lower left panel shows the residuals after subtracting the host as well. For comparison, the lower right panel shows what an image of an unlensed PG1115+080 quasar and host would look like in the same integration time and on the same scales. The host galaxy is an H = 20.8 mag late-type galaxy (Sersic index n = 1.4) with a scale length of Re = 1.5h-1 kpc. The demagnified magnitude of the quasar is H= 19.0 mag. The axis ratio of the source, qs = 0.65 ± 0.04 is consistent with the estimate of qs = 0.58 ± 0.02 from the simpler ring curve analysis (Section B.10.1, Fig. B.71, Kochanek Keeton & McLeod [2001]).

At low redshifts (z < 1), quasar host galaxies tend to be massive early-type galaxies (e.g. McLure et al. [1999], Dunlop et al. [2003]). Over 80% of quasars brighter than MV < - 23.5 mag are in early-type galaxies with L gtapprox 2L* and effective radii of Re ~ 10 kpc for z ltapprox 0.5. Radio quiet quasars (RQQ) tend to be in slightly lower luminosity hosts than radio loud quasars (RLQ) but only by factors of ~ 2 at redshift unity. Far fewer unlensed host galaxies have been detected above redshift unity (e.g. Kukula et al. [2001], Ridgway et al. [2001]) with the surprising result that the host galaxies are 2-3 mag brighter than the typical host galaxy at low redshift and corresponded to ~ 4L* galaxies. Given that the low redshift hosts were already very massive galaxies, it was expected that higher redshift hosts would have lower masses because they were still in the process of being assembled and forming stars (e.g. Kauffmann & Haehnelt [2000]). One simple explanation was that by selecting from bright radio sources, these samples picked quasars with more massive black holes as the redshift increased, creating a bias in favor of more massive hosts. The key to checking for such a bias is to be able to detect far less luminous hosts, and the improved surface brightness contrast provided by lensing the host galaxies provides the means.

Fig. B.74 shows the observed H-band magnitudes of the lensed hosts as compared to low redshift host galaxies and other studies of high redshift host galaxies. Although 30% of the lensed quasars are radio-loud, they have luminosities similar to the lensed (or unlensed) radio-quiet hosts. There are no hosts as bright as the Kukula et al. ([2001]) radio-loud quasar hosts. Once the luminosities of the quasar and the host galaxy are measured we can compare them to the theoretical expectations (Fig. B.75). While the models agree with the data at low redshift, they are nearly disjoint by z ~ 3 in the sense that the observed quasars and hosts are significantly more luminous than predicted. The same holds for the Kukula et al. ([2001]) and Ridgway et al. ([2001]) samples, suggesting that black holes masses grow more rapidly than predicted by the theoretical models or that accretion efficiencies were higher in the past. Vestergaard ([2004]) makes a similar argument based on estimates of black hole masses from emission line widths.

Figure 74

Figure B.74. Observed H-band magnitudes of quasar host galaxies. The solid (open) circles are secure (more questionable) hosts detected in the CASTLES survey of lensed hosts. The low redshift points are from McLeod & McLeod ([2001]). All the Ridgway et al. ([2001]) systems are radio quiet. For comparison, we superpose the evolutionary tracks for a non-evolving E/S0 galaxy (solid curve), an evolving E/S0 galaxy which stars forming stars at zf = 5 with a 1 Gyr exponentially decaying star formation rate (long dashed line) and a star forming Sb/c model (short dashed line). The evolution models are matched to the luminosity of an L* early-type galaxy at redshift zero. The CASTLES observations can reliably detect hosts about 4 magnitudes fainter than the quasar. From Peng ([2004]).

Figure 75

Figure B.75. A comparison of the estimated rest frame absolute magnitudes of the quasars and hosts as compared to the theoretical models for the evolution of galaxies and the growth of black holes as a function of redshift by Kauffmann & Haehnelt ([2000]). The low redshift quasars from McLeod & McLeod ([2001]) occupy the triangle in the upper left panel. At intermediate redshift the lensed host galaxies occupy a region similar to the models, but the two distributions are nearly disjoint by z appeq 3. Both the hosts and the quasars are significantly more luminous than predicted. The horizontal line marks the luminosity of an L* galaxy at z = 0. From Peng ([2004]).

Next Contents Previous