Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 327, 72-80 (1997)

Previous Section Next Section Title Page Table of Contents

3. Discussion

Independently of the details on how the photoionizing radiation is produced, the intensity of the emission lines and their response to variations in the continuum strongly argues in favour of the hypothesis that there is photoionized gas in the nuclei of active galaxies. In the standard model of AGN, such a gas is distributed in a large number of relatively small clouds surrounding an accretion disk around a supermassive black hole, whose emitted radiation heats and ionizes the gas clouds. The intensities and profiles of the emission lines are driven by the photoionization continuum and the physical conditions, distribution and kinematics of the gas. As mentioned in Sect. 1, the lack of broad wings in the optical permitted lines in NLS1 has been attributed to either the absence of high velocity gas clouds or an orientation effect that makes that these clouds have a negligible velocity component along the line of sight. However, the detection of broad wings in the permitted UV lines proves the presence of high velocity gas clouds, with a large velocity component along the line of sight. Therefore, it is necessary to conceive a scenario where a central continuum similar to that of the normal Seyfert 1 ionizes high velocity gas which emits broad UV lines that can be seen from the Earth, whereas the broad optical emission lines can not be detected.

A possibility is that both UV and broad optical lines are in fact emitted in the BLR, but hidden by the thick molecular dusty torus proposed in the unification models for AGN. The broad UV lines then will show up after reflection in a scatterer material above the torus. In this context, a natural way to show up the UV but not the optical lines is scattering by dust, which is much more efficient in the UV than in the optical (as opposed to scattering by electrons, which is essentially wavelength independent). If this picture is true and the BLR is hidden from direct view, we would expect the continuum to be hidden as well, unless the system has a very unlikely geometry. However, the SED of NLS1 discussed in previous section does not support the picture of hidden AGN. In fact, the NLS1 SED is closer to that of normal Seyfert 1 galaxies than to that of Seyfert 2 (up to now the best candidates to hidden AGN). In particular, the Seyfert 2 SED is much more strongly peaked in the FIR range than that found in NLS1. Moreover, another observational result in conflict with the hypothesis of a hidden AGN is the absence of significant cold absorption of the X-rays (Boller et al. 1996).

There is another possibility that does not consider a hidden nucleus, but a scenario in which the nuclear gas simply does not emit optical lines. The strongest permitted lines in the optical range are H [FORMULA], H [FORMULA] and the rest of the lines from the Balmer series of hydrogen. These lines are predominantly formed in regions where the hydrogen is only partially ionized (HPI), while the strongest UV lines (C [FORMULA] 1550 and He [FORMULA] 1640) form where the hydrogen is fully ionized (HFI) (H-Ly [FORMULA] may form in both fully or partially ionized regions). Therefore, the absence of broad H-Balmer lines in the optical region would be naturally explained if there is no HPI zone in the high velocity gas in the nuclei of NLS1. The presence within the BLR of hydrogen fully ionized clouds (and, hence, optically thin to Lyman continuum photons), has already been suggested by Shields et al. (1995) (see also references therein). As they show, thin clouds can contribute significantly to the emission of the broad high ionization lines in AGN without producing significant low ionization species. Shields et al. (1995) also discuss on the possibility that these thin clouds are responsible of the observed UV and soft X-ray absorption features in the AGN spectra (when crossing our line of sight to the central source).

In order to explore the physical conditions of a gas to emit the observed NLS1 line spectrum, we have made use of the photoionization code CLOUDY (Ferland 1991). We have assumed a solar abundance gas which is illuminated by an ionizing continuum similar to that derived by Mathews & Ferland (1987). The free parameters (the gas physical conditions) are selected automatically by the code to find an optimal solution fitting the observed average line spectrum: Ly [FORMULA] /CIV [FORMULA] 2, Ly [FORMULA] /H [FORMULA] [FORMULA] 100, He [FORMULA] 1640/CIV [FORMULA] 0.10. The code derives an optimal model for a region where the total hydrogen column density is [FORMULA], the hydrogen number density is [FORMULA] and the ionization parameter [FORMULA]. Under these conditions, the emitting gas is fully ionized in hydrogen, and, thus, optically thin to the Lyman continuum. We note that the detailed physical conditions would be different if, for instance, the ionizing continuum is harder than that of Mathews & Ferland (1987). However, we want to stress that, independtly of the actual numbers obtained from CLOUDY, a high ionization parameter and a relatively low column density are required to roughly reproduce the observed line ratios. A detailed model of the BLR in NLS1 is out of the scope of this paper, since higher quality data over a wider spectral range would be needed for every single object, but the CLOUDY computations give sufficiently general results for this dicussion.

The equivalent width of the Ly [FORMULA] line emitted in optically thin clouds is much smaller ([FORMULA] 100 times) than in the optically thick case, but the observed equivalent width of the broad Ly [FORMULA] in NLS1 (as derived from Tables 2 and 4) is only [FORMULA] 3 times smaller than in normal Seyfert 1 galaxies ([FORMULA] Å). If the BLR in NLS1 emits indeed in an optically thin regime, the covering factor ([FORMULA] fraction of the sky covered by clouds, as seen by the central continuum source) has to be much larger ([FORMULA]) than in normal Seyfert 1 ([FORMULA]) to keep the difference in equivalent widths not too big. As a side effect, a large value of [FORMULA] implies that the probability of finding a cloud right in the observer's line of sight is very high. However, the effect of an optically thin intervening cloud in the observed spectrum will not be very strong, except in the soft X-ray domain, which is very sensitive to relatively small amounts of neutral hydrogen and where absorption edges can be found due to some highly ionized atoms (e.g., OVI-VIII). As we noted above, this fact was already pointed by Shields et al. (1995) who discuss under which conditions the thin BLR clouds can also produce soft X-ray absorption edges. In this respect, we note that the crossing time of one of these optically thin clouds would be of the order of hours, similar to the variability time scales found in the soft X rays of most NLS1. If the variations detected in the soft X-rays are due to the passage of individual clouds across the observer's line of sight, no change in the emission lines would be expected, since their flux comes from the integration over a very large number of clouds, whose ionizing radiation has not changed.

Another implication of optically thin clouds is that the response of the emission lines to continuum variations is much weaker than in the case of thick clouds. A line fluctuation would only be detected if the change in the continuum is sufficiently large to produce a significant change in the ionization structure of the clouds. This could explain why the wings of the UV lines in the spectra of Mrk 1044 and IRAS13224-3809 remained constant while the continuum and the line cores varied (Sect. 2.2).

In their study of optically thin broad-line clouds in AGN, Shields et al. (1995) suggest that the importance of thin clouds relative to the thick clouds should be larger in low luminosity objects than in high luminosity ones. To account for this effect, they claim for outflows of the thin clouds that would proceed more efficiently in intrinsically brighter sources. This could explain the observed anti-correlation between UV emission-line equivalent width and continuum luminosity (the Baldwin effect). In this context, the tendency for the NLS1 galaxies to have lower UV luminosities than "normal" Seyfert 1 also supports the presence of optically thin broad-line clouds in NLS1.

We have finally compared the "typical" parameters that characterize normal BLR ([FORMULA]), with those found to represent the BLR in NLS1 (([FORMULA]). From the definition of U,


where [FORMULA] corresponds to 13.6 keV, and taking into account the average differences in [FORMULA] and [FORMULA] given in Table 3, the distance r of the BLR to the ionizing source should be very similar in NLS1 (within a factor of 2) and normal Seyfert 1 galaxies.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1997

Online publication: April 8, 1998