The physical and statistical properties of the X-ray absorber in AGNs are far from being fully understood, and several questions remain unanswered yet.
We have shown that there is clear evidence for a compact absorber, located on the scale of the BLR, in a few sources. However, it is not clear whether such a compact absorber is ubiquitous to all AGNs or not. A related issue is whether all AGNs have Compton thick gas around them (including those which are Compton thin along our line of sight) or not. For what concerns the structure of the absorber, it is not clear yet whether the temporal variations of NH are tracing a medium with two phases (cold, dense clouds inside a warm medium), or a more homogeneous, cold gas with density gradients. In terms of stability of the X-ray absorber, it is not clear how its vertical structure (required to account for the large covering factor) is supported.
Another class of issues is related to some puzzling comparisons of the X-ray absorption with other observed quantities. [Zhang et al. (2006)] found that AGNs with nuclear H2O maser disks are not preferentially Compton thick. However, the detection of maser emission requires large columns of gas (NH > 1023-24 cm-2) and very small disk inclination (< 10°), the combination of which are expected to produce Compton thick absorption, that instead is observed for only half of the AGNs in the maser sample of [Zhang et al. (2006)] (i.e. the same fraction observed in optical samples). On the large scales there is an opposite issue. [Maiolino et al. (1999)] found that the X-ray absorption is correlated with the presence of a bar in the host galaxy. In particular, Compton thick AGNs appears preferentially hosted in barred galaxies. As discussed above, Compton thick absorption probably occurs on the sub-parsec scale, therefore it is hard to understand how the physics of the sub-parsec medium can be affected by the dynamical properties of the host galaxies on the kpc scale. We have verified the [Maiolino et al. (1999)] result by using the more recent data by [Cappi et al. (2006)] and [Guainazzi et al. (2005)]; but even in these samples the correlation remains, as illustrated in Fig. 4.
Figure 4. Column density distribution of the sources in the samples of [Cappi et al. (2006)] and [Guainazzi et al. (2005)], divided as a function of the strength of the stellar bar in the host galaxy. There is a significant progression of the average NH as a function of the bar strength. In particular, most Compton thick nuclei are hosted in barred galaxies.
Acknowledgements. This work was supported by the Italian Space Agency. RM is grateful to the organizers of the conference for their kind invitation.