2.1. Optical/UV
The optical to UV emission of quasars is characterized by the "big blue bump" (Shields 1978; Malkan & Sargent 1982; Elvis 1985), where the peak of quasar emission is usually found. The peak energy is around the Lyman edge ( = 1216 Å), and the spectrum can be well approximated with a power law both at lower and higher frequencies. Recent observations performed with HST of over 200 quasars (compiled by Telfer et al. 2002) provide a good quality mean spectrum from ~ 300 Å to ~ 3000 Å. Composite spectra (see Fig. 2) extending from ~ 1200 Å to ~ 9000 Å have been obtained using data from ground-based optical surveys like the UK Schmidt Telescope's Large Bright QSO Survey (Francis et al. 1991), the 2dF (Croom et al. 2002), the optical follow-up of the FIRST radio survey (Brotherton et al. 2001), and the SDSS (Vanden Berk et al. 2001). The SDSS includes more than 2200 spectra at redshifts between 0 and ~ 5, providing the most accurate average optical spectrum of quasars so far, with a spectral resolution of a few Å. In the overlapping band (~ 1200 - 3800 Å), the HST and SDSS results match within the errors, providing a complete quasar spectrum in the 300 - 9000 Å band. The main results of these studies are the following:
Figure 2. Composite optical/UV spectra of quasars. (Upper panel) Results from the SDSS (Vanden Berk et al. 2001; their Fig. 3). (Lower panel) Results from HST (Telfer et al. 2002; their Fig. 4). Dotted and dashed lines are power law fits to the continuum. Horizontal thick lines in the lower panel show the spectral regions used to estimate the continuum level. |
The 300 - 5000 Å continuum can be modeled with two power laws with slopes (2) 1 = -1.76 between 300 Å and ~ 1200 Å and 2 = -0.44 between 1200 Å and 5000 Å. At longer wavelengths, the spectrum appears to flatten significantly, but this effect is probably due to the contribution of galactic emission in the quasar spectra (see Vanden Berk et al. 2001 for more details).
No correlation between optical continuum properties and redshift or luminosity has been found. An anticorrelation between the equivalent width (EW) of the main emission lines and the luminosity (the "Baldwin effect", Baldwin 1977) has been found.
Hundreds of emission lines are present in quasar spectra (a compilation of the brightest ones is given in Table 1). In addition to these lines, another major feature in the optical/UV spectra of quasars is the "small blue bump" (Wills, Netzer, & Wills 1985; Elvis 1985) between ~ 2200 Å and ~ 4000 Å. This is not a true continuum feature but is due to a forest of emission lines from the ion FeII and the Balmer recombination continuum. Permitted emission lines are "broad" (corresponding to velocities of the emitting gas of 2000 - 15000 km s-1), while forbidden lines are narrow (a few hundred km s-1).
Line | (Å) | <EW(Å)> | Ref. | Line | (Å) | <EW(Å)> | Ref. |
CIII+NIII | 980 | 9.7±0.2 | (1) | MgII | 2799 | 32.3±0.1 | (2) |
OVI + Ly | 1030 | 15.6±0.3 | (1) | H | 4103 | 5.05±0.06 | (2) |
Ly | 1216 | 91.8±0.7 | (1) | H | 4341 | 12.6±0.1 | (2) |
N V | 1240 | 18.5±0.5 | (1) | H | 4863 | 46.2±0.2 | (2) |
SiIV+OIV] | 1397 | 8.13±0.09 | (2) | [OIII] | 5007 | 13.2±0.2 | (2) |
CIV | 1549 | 23.8±0.1 | (2)a | H | 6565 | 194.5±0. | (2) |
CIII] | 1909 | 21.2±0.1 | (2) | ||||
References - (1)
Telfer et al. 2002
(for lines at
< 1300 Å);
(2) Vanden Berk et
al. 2001
(for > 1300 Å).
|
A minority (~ 10 - 20%) of quasars show broad absorption lines, often saturated, with widths and blueshifts of several thousand km s-1 and peaks of 20,000 - 30,000 km s-1.
~ 50% of bright Seyfert 1 galaxies observed in the UV with HST or FUSE show evidence of Narrow Absorption Lines (NAL), with widths of ~ 1000 km s-1, in the profiles of high ionization emission lines (OVI, CIV, Ly). The presence of such features is strongly correlated with the presence of warm absorbers in the soft X-rays (see Section 2.4).
2 Somewhat different conventions are used for the power law index: photon index in the X-rays (counts/sec/keV E-), in the radio (f ), and - or in the optical. In this review, we follow the radio convention, as it is mathematically correct in the log f versus space. Back.