2.2.3. The X-ray properties of Seyfert 2 galaxies

The ``unified model'' for AGNs assumes Seyfert 2s and Seyfert 1s to be identical physical objects, while the orientation of the line of sight with respect to an obscuring torus accounts for all the observed differences between the two classes.

In agreement with this model, the X-ray emission of many Seyfert 2s is characterized by a power-law spectrum similar to that observed in Seyfert 1s, with a cut-off at low energies due to absorption by gas column densities between 10²² and > 10²⁴ cm^-2 ([278]; Sambruna et al. 1999). Such high absorbing column densities have been identified with the torus. In addition to this hard component, there is a soft component due to reflection from an optically thick disk or a molecular torus ([393]).

In low X-ray luminosity Seyfert 2s, in addition to the compact X-ray source associated with the nucleus, an extended X-ray component due to starburst activity can also be present ([423]; [390]).

For N_H < 10²⁴ cm^-2, X-rays above a few keV can penetrate the torus, making the nuclear source visible to the observer; for values of N_H around 10²⁴ cm^-2, only X-rays in the 10-100 keV range pass through the torus. These sources are called ``Compton thin''; their column density is measurable. For values of N<sub>H</sub> higher than 10<sup>24</sup> cm<sup>-2</sup>, the matter is optically thick to Compton scattering and the nucleus becomes practically invisible also in hard X-rays; it can be observed only in scattered light; such sources are called ``Compton thick'' ([285]; [27]).

A number of Seyfert 2s have been found with column densities in the range 1-5 10²⁴ cm^-2: NGC4945 ([203]; [105]), NGC6240 ([451]), Mark3 ([74]), while others are Compton-thick ([424]; [27]). About 3/4 of all Seyfert 2s are heavily obscured (N_H > 10²³ cm^-2) and almost half are Compton-thick ([353]).

When the column density increases to a few 10²³ cm^-2, the Fe K EW increases since it is measured against a depressed continuum; the EW can be higher than 1 keV for column densities N_H > 10²⁴ cm^-2 ([27]).

Strong observational evidences suggest that the [O III] luminosity is one of the best independent measures of the intrinsic luminosity of the nuclei of AGNs. Indeed, Seyfert 1s and 2s have the same ratio of far-infrared (FIR) to [O III] luminosities which is consistent with the hypothesis that the [O III] and FIR emissions are isotropic in both types of objects ([53]); the narrow-line luminosities of NLRGs and radio loud QSOs having the same extended radio luminosities are similar ([387]); moreover a strong correlation has been found between [O III] and HX luminosities for objects with low N_H ([429]; [483]). The 5007 emission line luminosity can therefore be used to roughly infer the intrinsic power in AGNs, even in Compton-thick Seyfert 2s. It seems however that the strength of the [O III] lines is somewhat aspect-dependent, core-dominated QSOs having stronger lines than lobe-dominated QSOs ([16]). If the observed X-ray luminosity is substantially less than that predicted by [O III], this is an indication of the presence of a hidden X-ray continuum source ([429]; [277]; [27]). As a few per cent of the soft X-rays are expected to be scattered into our line of sight, Seyfert 2s should be underluminous in the soft X-ray band by a factor of 10-100 relative to Seyfert 1s having the same [O III] luminosity ([306]; [181]).

NGC3147 and NGC7590 are bona-fide Seyfert 2s and yet have negligible X-ray absorption; the low absorption could be reconciled with their Seyfert 2 nature only if they were Compton-thick which is not supported by their large soft X-ray (0.1-2.5 keV) to 5007 flux ratio ([27]); these authors concluded that there might be a few objects in which the Seyfert 2 appearance is intrinsic and not due to obscuration; but what would then be the mechanism of ionization of the Seyfert 2 nebulosities which are generally believed to be photoionized by the hidden QSO?

We have collected all known AGNs with N_H > 7 10²¹ cm^-2, from Bassani et al. (1999), excluding NGC1365 for which the nuclear source has a low column density ([225]). From the most recent available data, four of these objects are now classified as Seyfert 1.9s (NGC526a, NGC2992, NGC7314 and F49; two more NGC5674 and ESO103-G35 could also be S1.9, but their published spectra are of poor quality; moreover in ESO103-G35 the X-ray column density is suspected to be variable; [278]), four more are S1is (NGC2110, NGC5506, ESO434-G40 and IRAS20460+1925). Thirty six objects are Seyfert 2s or S1hs (and one possible Seyfert 2); NGC7582 is discussed below. All the Seyfert 1.9s and S1is have moderate column densities (7-37) 10²¹ cm^-2, while the pure Seyfert 2s have very large column densities (> 4010²¹ cm^-2). The ranges of column densities of the S1hs and the S2s are about the same.

RX J1343.4+0001 is a type 1.9 QSO at z = 2.35; the X-ray column density is ~ 10²³ cm^-2 which is large for a type 1.9 QSO; the nature of the UV continuum in this object is also puzzling ([140]).

In table 1 we give the range of visible extinction found for the Seyfert 1.8s and 1.9s, the S1is and the pure Seyfert 2s, including the S1hs (col. 2), the column densities inferred from the X-ray observations (col. 3) and the visible extinction calculated from A_V = 0.5N_H10^-21 cm^-2 ([340]) (col. 4). There is a clear correlation between the optical and X-ray derived values of the extinction; it seems however that the extinctions deduced from the X-ray column densities are consistently larger by about a factor of two than the extinctions obtained from the optical or infrared emission lines, in agreement with Mushotzky (1982). To explain this finding, [165] suggested the existence of dust-free X-ray absorbing gas, lying inside the dust sublimation radius.

In conclusion, it seems that the hard X-ray properties of Seyfert 2s depend on a single parameter, the absorbing column density along the line of sight, in accordance with the unified model ([27]). All Seyferts, from S1.0 to S1.9 to S2, could be quantified by their X-ray column density.

A few hard X-ray sources turned out to be associated with galaxies with strong emission lines undistinguishable from those of Seyfert 2s ([465]; [376]). These objects were called Narrow emission Line X-ray Galaxies (NLXGs), or more often NELGs; they more closely ressemble Seyfert 1s than Seyfert 2s at high energies, having HX luminosities in the range 10⁴²-10⁴⁴ erg s^-1; they were thought to represent a new class of X-ray galaxies ([465]; [302]; [467]).

Most of them turned out to have relatively large X-ray absorbing column densities (of the order of a few times 10²² cm^-2), implying a visual absorption A_V ~ 10 mag. for the central engine and the broad emission line region ([468]). Several of these objects were indeed found to have a weak broad H or Pa component showing that they are Seyfert 1s with heavily reddened broad lines ([448]; [381]; [159]).

Although today the X-ray NELGs fit nicely into the Seyfert classification, being considered as either Seyfert 1.8s, 1.9s, or 2s with a relatively small column density, they are still sometimes considered as constituting a special class of objects.

The X-ray column densities observed in AGNs are usually variable, with time-scales of the order of one year or less ([278]). On the other hand, some Seyferts have spectra changing from Seyfert 1.9 to 1.0 (Mark530, Mark993, Mark1018); the changes in flux of the broad lines and continuum near H and H are consistent with changes in the extinction in each of these cases (Goodrich 1989a; 1995; [416]). Broad Balmer lines have appeared in the nucleus of the Seyfert 2 galaxy NGC7582 ([13]); the X-ray column density, although not measured at the time where the broad lines were visible, was found to vary in the range 0.9-4.810²³ cm^-2 ([468]; [484]); the apparition of the broad lines could be explained by a drop of the column density to ~ 10²² cm^-2 or by holes appearing in the obscuring screen ([426]). It seems likely that the observed changes of optical types are due to changes of the column densities. The shortest observed time scales for spectral changes imply high transverse velocities for the dusty clouds, high enough that they must be close (but outside of) the bulk of the BLR itself ([157]). The variability of both the X-ray column densities and the visible extinctions implies some dispersion in the correlation between these two quantities if they are not measured simultaneously, which is generally the case.