Having most of the H I column located in front of the counterjet at a projected distance of pc is consistent with a number of previous results on NGC 4261. First, at distances closer to the nucleus we do not expect H I absorption, since we know the circumnuclear material is mostly ionised. This is shown by the free-free absorption at a projected radius of 0.2 pc inferred by Jones & Wehrle (1997). Secondly, NGC 4261 harbours an X-ray source ( erg/s in the 0.2-1.9 keV range, Worrall & Birkinshaw 1994), which puts a limit on the total column density on the line of sight to the nucleus. The model presented by Worrall & Birkinshaw (1994) yields an upper limit on the total column density of . Given that the X-rays preferentially originate in the nucleus, this fits in comfortably with the constraints from our VLBI H I absorption.
The model fitting from which we derive the optical depth does not allow the positions of the components to vary (see Sect. 3), nor is there any sensitivity for H I beyond the end of the continuum counter-jet. We are therefore forced to make a simplifying assumption, namely that 18 mas is the mean radius of the H I absorbing structure. This is supported by the fact that most of the VLA absorption is recovered by the VLBI observations (Sect. 3). Given the HST dust disk inclination, this implies a distance 5.7 pc away from the nucleus. The FWHM of the line is comparable with other H I absorption observations of circumnuclear gas (e.g. in Cyg A, Conway & Blanco 1995). Therefore, in the next step we assume that the atomic gas is part of such a circumnuclear rotating structure and not due to individual clouds randomly distributed in front of the continuum source. Such a model of the H I disk is supported by the nuclear parameters derived by Ferrarese et al. (1996) from HST data on optical transitions. They found a central mass of , which implies a rotational velocity of 610 km/s at the location of the H I . Under the standard assumption that the linewidth provides an estimate of the isotropic turbulent velocity, we use the thin disk relation to estimate the disk thickness h. We estimate the velocity dispersion at radius r to be which gives pc. So the H I is likely to reside in a thin circumnuclear disk with an opening angle of . The average density can, assuming a volume filling factor f of unity, be estimated to be for a spin temperature of 100 K. It follows that a more clumpy distribution () will increase the estimated density () and decrease the estimated mass (). However, adopting for simplicity, a mass estimate of H I inside an homogeneous disk of radius 6 pc is . Such a mass would be enough to supply material to the source for years (assuming a radiative efficiency ), given that the total luminosity of the radio source is erg/s (e.g. Ferrarese et al. 1996). Using the correlation between FRI source sizes and their age (Parma et al. 1999), the size of NGC 4261 (Jaffe & McNamara 1994) implies an age years. The same correlation shows other FRIs with ages years. Hence, on this time-scale the H I mass we estimate is barely sufficient to fuel the source. It seems more plausible that there is a continuous flow of accreting material being transported from the 100 pc scale dust disk onto the central nucleus.
The circumnuclear torus- or disk-structures observed in H I are usually found on slightly larger scales (50 - 100 pc; e.g. Gallimore et al. 1999 and Conway 1999). Only in a few other cases the H I is found to lie on very small scales ( pc in NGC 4151; Mundell et al. 1996 and Gallimore et al. 1999) and it is not obvious that H I survives so close to the nucleus. For gas irradiated by X-rays an effective ionisation parameter can be defined, which governs the physical state of the gas (Maloney et al. 1996). For the gas is likely to be molecular with gas temperatures close to or below 100 K, while higher values of correspond to a hotter atomic gas phase. Following Maloney et. al (1996), we use , where is the hard ( keV) X-ray luminosity, r is the distance from the nucleus to the irradiated gas, n is the gas density and is the column density in units of cm-2. At the distance of 6 pc a gas density of cm-3 yields an (atomic) obscuring column density of . Using the hard X-ray luminosity of NGC 4261 ( erg/s, Roberts et al. 1991) this results in ; thus implying a mainly atomic gas phase where the gas temperature is likely to exceed 1000 K (Maloney et al. 1996). As a consequence the spin temperature is probably larger than 100 K, and our estimates of the H I mass and density will only be lower limits.
We conclude that within the scope of this model, it is indeed possible to have an atomic structure on the scales sampled by our VLBI observations. The inner boundary of this region is naturally set by the location of the free-free absorption, which also must be geometrically thin in order to leave the core unattenuated. On the outside, the structure changes over into a dust disk which is visible to HST from its innermost pixel, at out to 240 pc. However, since one would think that the mm radiation originates from the flat spectrum core, it is difficult to reconcile the reported CO absorption (Jaffe & McNamara 1994) with a thin molecular disk. Apart from the unknown location of the CO gas, the evidence points to the FRI radio-source in NGC 4261 being powered by gas infall through a relatively thin disk with a clear gradient of excitation conditions.
© European Southern Observatory (ESO) 2000
Online publication: January 31, 2000