Next Contents Previous

4.4. HST Imaging: the Quasar Host Galaxy

Improving the astrometry over and over, helps, but does not solve the fundamental problem imposed by the mass sheet degeneracy. The number of lensed images available can not be increased, so that potential well of the lens, and hence the arrival-time surface, are probed only at a very limited number of points. This is quite not true as soon as one considers extended sources. Quasars have host galaxies. They are faint, but they have a much larger angular size than the quasar they harbourgh. The effect of lensing on their shape, distorted and stretched, is much stronger than on the point source quasar. The role of high angular resolution observations therefore extends well beyond the simple aim of measuring the image positions and fluxes. Each detail discovered in the lensed image of the host galaxy is an additional point in the mass map.

PG 1115+080 has been observed with the refurbished HST, in the infrared (Impey et al. 1998), with a resolution and depth sufficient to unveil an almost full ring joining the quasar images (Fig. 9). While the host galaxy provides more constraints for the models, it also forces to introduces more degrees of freedom. Indeed, a model has to be chosen to represent the "unlensed" host, for example an exponential disk, as done in Impey et al. (1998). Feeding the lensing model with the image positions, time-delays, flux ratios, lens ellipticity, position of the nearby group of galaxies, and the model for the host, one can reproduce the observed ring.

Figure 9

Figure 9. An HST image of PG 1115+080 is shown in the upper left panel (Impey et al. 1998), with the four quasar images and the lens G. Removal of the quasar images, as done in the upper right panel, unveils an almost full ring: the lensed host galaxy of the quasar. It is even more evident after removal of the lensing galaxy, in the lower left panel. If a model is adopted for the "unlensed" host galaxy, the shape of the lensed ring can be predicted (lower right panel), and compared with the observations.

The actual gain brought by the ring is significant but not always sufficient to discriminate between lens models. Impey et al. (1998) give two estimates for H0 for two different type of lens models. One assumes that light traces mass, i.e., that the mass-to-light ratio is constant in every point of the lens. The second, more realistic according to our current knowledge of the mass distribution in galaxies, includes a dark halo component. Using these two models, they find H0 = 66 km s-1 Mpc-1and H0 = 44 km s-1 Mpc-1, respectively.

A complete Einstein ring was found in PG 1115+080, but with no resolved details, only the surface brightness can be used to constrain the models. Because the of the high degree of symmetry of the system, and given the resolution of the observations, it is not clear whether full rings are actually useful for sorting out between different mass models, or not (see, e.g., Saha & Williams, 2001).

Many other lensed hosts galaxies have been discovered using HST images (Lehar et al. 2000), in systems that are less symmetric than PG 1115+080, and with higher degree of detail. A good example is the double Q 0957+561 in which the host galaxy reveals clumps, distorted arcs and for which a different approach of the modeling was used, not involving any prior model for the host's surface brightness (Keeton et al. 2000). Since the observations of the host of Q 0957+561 are probably the most detailed existing so far, we will briefly leave PG 1115+080 aside and use Q 0957+561 to illustrate the use of hosts in quasar lensing. The observations are shown in Fig. 10, while Fig. 11 illustrates the modeling method. Based on the image of the brighter component A of the host, mapping it back onto the source plane, the unlensed host is reconstructed. It is then re-imaged, in order to produce the fainter image B, and a chi2 is minimized, between the observed image and the prediction from the model. The method takes advantage of the high degree of details in the host, and on the assymetric configuration of the system. It does not require any strong prior knowledge on the host. The method allows to rule out several models previously published in the literature.

Figure 10a Figure 10b

Figure 10. Left: HST NICMOS observation obtained by the CASTLE group, where the two quasars and the lensing galaxy are well separated. Right: the quasars and the lens have been removed to unveil the host galaxy of Q 0957+561. On the contrary of PG 1115+080, many details are seen in each image of the host (Keeton et al. 2000).

Figure 11a Figure 11b

Figure 11. Left: the A image of the host is mapped back into the source plane to reconstruct the unlensed image of the host. Right: the source is imaged again in order to produce image B given the lens model. The image B, as predicted by the model, is the solid contour in the figure. Grey levels are the observations (Keeton et al. 2000).

Next Contents Previous