7.1. Search for correlations between bars and AGN
There have been several observational works revealing a correlation between nuclear activity and bars (Dahari 1984, Simkin et al 1980, Moles et al 1995). But the correlation is weak and depends on the definition of the samples, their completion and other subtle effects. Near-infrared images have often revealed bars in galaxies previously classified unbarred, certainly due to gas and dust effects. However, Seyfert galaxies observed in NIR do not statistically have more bars nor more interactions than a control sample (McLeod & Rieke 1995, Mulchaey & Regan 1997).
It is however evident from observations that bars are efficient to produce radial gas flows: barred galaxies have more H2 gas concentration inside their central 500 pc than un-barred galaxies (cf Sakamoto et al 1999). Also, the radial flows level out abundance gradients in barred galaxies (Martin & Roy 1994).
Peletier et al (1999) and Knapen et al (2000) have recently re-visited the question, and took a lot of care with their active and control samples. Their Seyfert and control samples are different at 2.5 , in the sense that Seyferts are more barred. They also measure the bar strength by the observed axial ratio in the images. In Seyferts, the fraction of strong bars is lower than in the control sample (Shlosman et al. 2000). Although a surprising result a priori, this is not unexpected, if bars are believed to be destroyed by central mass concentrations (cf section 4).
Regan & Mulchaey (1999) have studied 12 Seyfert galaxies with HST-NICMOS. Out of the 12, only 3 have nuclear bars but a majority show nuclear spirals. However their criterium for nuclear bars is that there exist leading dust lanes along this nuclear bar. This is not a required characteristic, since these secondary bars in general are not expected to have ILRs themselves.
A morphological study of the 891 galaxies in the Extended 12 µm Galaxy Sample (E12GS) has confirmed that Seyfert galaxies and LINERs have the same percentage of bars as normal spirals, contrary to HII/Starburst galaxies that have more bars (Hunt & Malkan 1999). However, active galaxies show rings significantly more often than normal galaxies or starbursts. The LINERs have more inner rings (by a factor 1.5), while Seyferts have more outer rings (by a factor 3-4) than normal galaxies. This might be due to the different time-scales for bar and ring formation: bars form relatively quickly, in a few 108 yr; they can drive matter to the central regions and trigger a starburst there, in the same time-scale. Outer rings form then, also under the gravity torques of the bar, but in the dynamical times of the outer regions, i.e. a few 109 yr. Since Seyferts are correlated with them, they would be associated to delayed consequences of the starburst, or of the bar, which by this time begins to dissolve.
The percentage of AGN in the E12GS sample is 30%. As for interactions, 25% of the Seyferts are "peculiar" (disturbed), while 45% of the HII/Starbursts are. There is also a correlation between AGN and morphological types along the Hubble sequence. Seyferts tend to lie in early-types (Terlevich et al 1987, Moles et al 1995). This has to be related to the existence on inner Linblad resonances in early-types, favoring the fueling of the nucleus (e.g. section 4, Combes & Elmegreen 1993).
In summary, there might be some evidence of the role of large-scale dynamics on AGN fueling, but it is in general weak, except for the most powerful AGN. A good correlation between bars/interactions and AGN is not expected, from several arguments:
- there must be already a massive black hole in the nucleus, and this might be the case only for massive-bulge galaxies (not all barred galaxies)
- again a large central mass concentration (bulge) is necessary to produce an ILR and drive the gas inwards (early-types)
- other parameters, like geometrical parameters, control the fueling efficiency of interaction
- time-scales are not fitted: the AGN fueling is postponed after the interaction/bar episode
- there are other mechanisms to fuel AGN such that a dense nuclear cluster