2.1.2. The emission line spectrum of Seyfert 1 galaxies and QSOs

In addition to a narrow-line emission spectrum, Seyfert 1s have broad permitted lines (H I, He I 5876,6678,7065, He II 4686 and Fe II in the visible domain).

The narrow-line spectra of Seyfert 1s are very similar to those of Seyfert 2s; there are however significant differences, most notably much stronger high- ionization lines ([Fe VII]6087, [Fe X]6375, [Fe XI]7892) in some Seyfert 1s ([85]; [110]; [375]). [120] showed that these ``coronal lines'' form just outside the BLR, in clouds with electron temperatures T ~ 12 000-150 000 K and densities of 10²-10^8.5 cm^-3.

There seems to be an anticorrelation between R₅₀₀₇, the ratio of the intensities of the broad H component to the [O III]5007 line and the broad H component luminosity or the continuum luminosity, i.e. bright QSOs have relatively weak narrow lines (since in the NLR R₅₀₀₇ ~ 0.1, the measurement of R₅₀₀₇ is a measurement of the ratio of the broad to the narrow H component fluxes). There are however probably no ``pure'' Seyfert 1s, i.e. bare BLRs without surrounding NLRs; the objects which come closest are the Seyfert Mark231, in which the only identified narrow line is [O II]3727 and the QSO 3C273 ([85]).

The decrease in the relative amount of narrow-line emission with luminosity suggests either that the ionizing flux to the narrow-line region is proportionately smaller in objects with a high-luminosity broad-line component, or else that there is proportionately less low-density gas in the higher luminosity Seyferts ([85]).

The broad emission lines observed in AGNs have a FWHM which is typically in the range 5000-10000 km s^-1 and show different kinds of profiles that usually are not Gaussian or even symmetrical ([402]; [95]). The half-width at zero intensity of H can be extremely large, reaching 35 000 km s^-1 in the case of PG 0052+251 ([55]). Crenshaw (1986) showed that the H / H and He I 5876 / H ratios increase from the core to the wings of the lines, indicating that the broad-line region is not a thin spherical shell.

Woltjer (1959) was the first to suggest that the width of the lines could be due to fast motions in the gravitational field of a massive nucleus. Assuming the line-emitting matter is gravitationally bound, and hence has a near-Keplerian velocity dispersion (indicated by the line width), it is possible to estimate the central mass. The main problem in estimating the mass from the emission line data is to obtain a reliable estimate of the size of the BLR; this size can be measured by reverberation mapping. Both the continuum and the broad emission line fluxes are variable in Seyfert 1s and QSOs; the time lags between the emission line and the continuum light curves can be interpreted in terms of the delayed response of the spatially extended BLR to the ionizing continuum source. These observations have established that Seyfert 1 BLRs have sizes of the order of a few light-days to light-months ( ~ 100 light-days). The BLR size of these objects is consistent with the hypothesis that the BLR size grows as L^0.5 as expected if the shape of the ionizing continuum in AGNs is independent of the luminosity L, and that all AGNs are characterized by the same ionization parameter and BLR gas density ([216]; [459]). The H FWHM correlates with the H luminosity, probably reflecting a scaling of the central mass with the luminosity ([293]). Whether or not the broad emission line widths actually reflect virial motions is still somewhat problematic; if however this is the case, the BLR provides a definitive demonstration of the existence of supermassive BHs. An estimate has been made by the reverberation mapping method of the BH mass in 19 Seyfert 1s; these masses range from 0.4 to 40 10⁷ M ([461]). The average black-hole-to-bulge mass ratio is 0.0003 for AGNs ([460]) while it is 20 times larger for bright QSOs ([244]).

Despite intensive studies, the BLR is still poorly understood.

It is now widely believed that accretion of gas into a central supermassive BH lies at the heart of the phenomenon; the accretion flow takes the form of a geometrically thin disk which is the source of the X-ray, UV and optical continuum emission which ionizes circumnuclear gas in both the broad-line and narrow-line regions; the BLR is made of an assembly of small clouds, photoionized by the continuum emission of the disk; this results from the fact that the ``effective'' volume giving rise to the emission is much smaller than L³, where L is the typical size of the emitting region deduced from the variability time-scale and the ionization parameter. However, these requirements can also be fulfilled if the lines are emitted by a continuous medium whose thickness is much smaller than its lateral dimension, for instance a thin disk or the atmosphere of a disk. Assuming that the velocity is Keplerian, FWHM of ~ 5 000-10 000 km s^-1 implies that this material is located at distances of about 10³-10⁴ R_S from the center (where R_S is the Schwarzschild radius 2GM/c²); if the disk is heated by the down scattered part of the non-thermal continuum observed in AGNs, the physical parameters of the optically thin region satisfy the requirements of photoionization models for the line emission ([88]). In these conditions, the low ionization lines (Balmer lines, Fe II lines, Mg II) can be emitted mostly by the accretion disk, but not the high-ionization lines (Ly, C IV, C III]) which are likely emitted by a dilute outflowing medium; this model yields BH masses in the range 10⁷-10⁹ M for a sample of six Seyfert 1s ([358]). In NGC5548, the emission lines are variable, but are best explained by the superposition of an emission line cloud with variable lines and another which shows no variability; the non variable cloud may not be radiatively heated; it is in collisional equilibrium with a temperature ~ 10⁴ K; its emission spectrum is dominated by Balmer lines and Fe II emission (Dumont et al. 1998).

The traditional QSO BLR consists of two components: one with ~ 2000 km s^-1 FWHM, the intermediate line region (ILR), and another very broad component of width > 7000 km s^-1 FWHM, the very broad line region (VBLR); the spectra of the ILR and VBLR are very different; the high-ionization lines are relatively stronger in the VBLR ([66]). The ILR and VBLR are probably identical with the low- and high-ionization line regions identified by Rokaki et al. (1992).

A highly significant correlation has been found for radio loud QSOs between the line widths (FWHM) of broad H lines and the relative strength of the compact radio nucleus; this is expected in beaming models if the predominant motion in the line emitting gas is confined to a disk lying perpendicular to the radio axis which is the case if the ILR is in the thin disk ([476]; [55]).

There is an anticorrelation of the EW of broad lines (C IV 1550, C III]1909 and Mg II 2798) with the continuum luminosity; these correlations are substantially stronger for the radio selected sample than for the radio quiet sample ([399]; [343]). This is the so-called ``Baldwin effect''. There is no evidence for a Baldwin effect in the broad H or H component ([293]).

A small number of AGNs, mainly but not exclusively radio loud, have been found to have double-peaked line profiles which could originate in Keplerian thin disks; they have a mean FWHM of ~ 12 500 km s^-1; their spectra are characterized by large [O I]/[O III] ratios, i.e. they are Liner-like ([109]; [179]; [405]; [355]). In one of these objects, Arp 102B, UV spectra show broad Mg II present with nearly the same profile as the Balmer lines, but there is little, if any, C III], C IV or Ly emission corresponding to the displaced Balmer-line peaks, demonstrating the need to invoke different locations and different physical conditions for double-peaked and single-peaked line components in the same object. The double-peaked component could be the low temperature, collisionally excited region postulated by [107].

H₂O megamasers have been looked for in a number of nearby galaxies; they have been detected in 11 Seyfert 2s, never in Seyfert 1s; this lack of detection in Seyfert 1s indicates either that they do not have molecular gas in their nuclei with physical conditions appropriate to produce 1.3 cm H₂O masers, or that masers are beamed away from Earth, in the plane of the obscuring molecular torus; the first possibility would violate the unified scheme ([58]). Kartje et al. (1999) suggested that the maser emission regions are clumpsy; when two maser clouds having the same velocities are overlapping along the line of sight, their brightness temperature is greatly enhanced through ``self-amplification'' ([101]); this could account for the fact that only Seyfert 2s seen nearly edge-on are detected as then the probability of seeing two aligned clouds is maximized ([215]).

H₂O masers emitted from a rotating molecular disk with a radius of the order of one parsec have been observed in two Seyfert 2s: NGC1068 ([136]) and NGC4258 ([168]; [295]); both galaxies contain a massive (2-4 10⁷ M) nuclear BH. The discovery of these molecular disks provides one of the most compelling evidence today for the existence of a massive BH in the nucleus of an AGN.