2.6. Luminosity and Redshift Effects

The SED described above is representative of local, optically selected quasars. Here we summarize the evidence for luminosity or redshift dependence in the emission of quasars. The optical and UV spectra of quasars observed with HST and the SDSS show no evidence of a dependence on redshift or luminosity. On the other hand, the optical-to-X-ray ratio _OX shows clear evidence of a luminosity or redshift dependence in optically selected samples (Zamorani et al. 1981; Avni & Tananbaum 1982, 1986; Wilkes et al. 1994; Yuan et al. 1998; Bechtold et al. 2003; Vignali et al. 2003). Yuan et al. (1998) discussed the reality of this effect in the ROSAT sample and concluded that the luminosity and/or redshift dependence could be due to selection effects, provided that the intrinsic dispersion in the X-ray emission of quasars is greater than in the optical.

The latest results from the SDSS (Vignali et al. 2003), however, strengthen the observational evidence for a dependence. The statistical analyses performed on the SDSS quasars (Vignali et al. 2003) and on the sample of optically selected quasars observed with ROSAT (Yuan et al. 1998) suggest that the dependence is only on luminosity and not on redshift. However, it remains difficult to disentangle the dependence on redshift and luminosity, which are strongly correlated in flux-limited samples. In Figure 7, we show the _OX-luminosity correlation for a sample of SDSS quasars observed (mostly serendipitously) with ROSAT and Chandra (Vignali et al. 2003). The best-fit linear correlation is _OX = -0.11 × log L(2500 Å) + 1.85.

Figure 7. Dependence of OX on optical luminosity in SDSS quasars observed with ROSAT. Triangles are averages using the number of quasars indicated next to each point (first number - total number of quasars in the luminosity interval; second number - number of X-ray upper limits in the luminosity interval). Circle is an average using all quasars at redshifts z > 3. Figure from Vignali et al. (2003; their Fig. 7b).

The dependence of the IR emission of quasars on luminosity is much harder to estimate, mainly because of the possible contribution from star formation. In Figure 6b, we plot the distribution of the IR-to-bolometric ratio for the same sample as Figure 6a, but for two luminosity ranges: L_IR < 3 × 10¹² L (shaded histogram) and L_IR > 3 × 10¹² L (open histogram). Apparently, higher luminosity sources have, on average, a smaller fraction of their emission in the IR. Haas et al. (2003) concluded that most of the observed emission is due to the AGN. The same conclusion was reached by Kuraszkiewicz et al. (2003) for ISO SEDs of X-ray selected AGN. However, we cannot exclude the possibility that the effect in Figure 6b is due to a higher contamination by nuclear star formation in lower luminosity sources.