Astron. Astrophys. 324, 51-64 (1997)

5. Discussion

5.1. Type of molecular clouds

The column density of in Cen A ranges between to (Table 6). These column densities are all higher than the onset of CO self-shielding, which occurs at column densities (Lucas & Liszt 1996). Hence the absorption in Cen A arises in gas which has a relatively low ratio. If the gas in Cen A follows the same tight correlation between and as Galactic clouds (Lucas & Liszt 1996), our measurement imply OH column densities in the range . These values are consistent with those inferred by van Langevelde et al. (1995) in Cen A.

In Fig. 9 we plot the column density of versus the column density of HCN, HNC and CS. We also include absorption line data from Galactic diffuse clouds obtained by Lucas & Liszt (1993, 1994, 1996) and somewhat denser clouds seen towards Sgr B2 (Greaves & Nyman 1996). The latter clouds are in some cases blended with gas close to the Galactic center. We have multiplied the column densities of Lucas & Liszt with a factor 1.4 in order to correct for the use of different values of the electric dipole moment. The dotted line in the figure is not a fit to the data but represents a one-to-one correspondance between the column densities. The gas in Cen A follows the same correlation as that of Galactic molecular gas. We have used an excitation temperature of 5 K when deriving column densities, while Lucas & Liszt and Greaves & Nyman have used 2.76 K. Since the molecules have almost exactly the same dependence on , this has no influence on the result.

Fig. 9. column density vs. HCN, HNC and CS column density. Filled circles represent our data on Cen A, open circles data from Galactic diffuse clouds (Lucas & Liszt 1993, 1994, private communication) and triangles data towards SgrB2 from Greaves & Nyman (1996). The column densities of Lucas & Liszt have been multiplied by a factor 1.4 due to different use of the electric dipole moment. The dotted line represents a one-to-one correspondance between the column densities.

Although the excitation temperature is likely to be low in the molecular gas seen in absorption, the HCN/HNC ratios implies that the kinetic temperature is rather high. The formation of HCN and HNC depends on the gas temperature (cf. Irvine et al. 1987), with HCN preferentially being formed in warm gas on behalf of its isotopomer HNC. In the LV complex we find HCN/HNC ratios ranging between 3-7, while the HV complex has a ratio of 2.2. This gives a kinetic temperature of the LV complex of 20-30 K, while the HV gas is characterized by a kinetic temperature of 10 K.

A more detailed comparison of the 5 main absorption components as defined in Table 2 and 6, shows that component 1 and 3 (in the LV complex) shows similar behaviour in their abundance ratios (see Table 6). The differences between the components are, however, relatively small and do not show up as significant deviations in Fig. 9. The ratios of the hyperfine components of HCN (Table 4) indicate that component 1 and 3 are close to the LTE values; R , R , while component 2 and 4 deviates from LTE. Here we have to keep in mind that in the decomposition of the hyperfine lines some components suffer from near overlap and that the F 0-1 line of component 4 is an upper limit. Nevertheless, it is tantalizing that the two components with close to LTE ratios also show similar abundance ratios, while the other two components (plus the HV complex) differs by factors of 2-3 from each other. Anomalous excitation of the HCN hyperfine lines is often seen in Galactic molecular clouds (cf. Guilloteau & Baudry 1981, Walmsley et al. 1982), with the R₁₂ ratio lower than LTE values in warm clouds and the R₀₁ ratio lower than LTE values in cold clouds. In dense regions, where the HCN(1-0) line thermalizes, both ratios tend to unity. This latter case appears to be the case for component no. 2 in Cen A, at least for the R₁₂ ratio.

In Fig. 10 we compare the (1-0) spectrum with the HI absorption obtained with the VLA (van der Hulst et al. 1983). The spectrum has been binned to the same velocity resolution as the HI data, 6.2 km s^-1. A 3-component gaussian fit to the spectrum and the residual is also shown. The appearance of the spectra agree quite well, even though the redshifted is more spread out in velocity than the HI. Column densities and N(HI)/N() ratios are given in Table 8. The component at 580 km s^-1 appears to have a higher molecular gas fraction than the other two by a factor of 3.

Fig. 10. The (1-0) spectrum binned to a velocity resolution of 6.2 km s-1, which is the same as the 21cm HI spectrum obtained by van der Hulst et al. (1983). Also shown is a 3-component gaussian fit and the residual spectrum.

Table 8. Column densities for HCO and HI .

In summary, all of the molecular gas components seen in absorption towards the nucleus of Cen A have a chemistry similar to that of Galactic diffuse molecular gas. Only the column densities are higher in Cen A than in typical Galactic diffuse clouds, which could be due to unresolved line components (i.e. more clouds) in the line of sight through the edge-on disk. The narrow lines seen in the HV complex reveal a relatively diffuse gas with lower abundances and a low kinetic temperature ( 10K).

5.2. Location of the absorbing molecular gas

The presence of two absorption complexes in Cen A, one at the systemic velocity and one redshifted relative to the systemic velocity, has led to speculations that the redshifted component arises in gas falling into the nucleus of Cen A, possibly feeding a supermassive black hole (cf. van der Hulst et al. 1983, van Gorkom et al. 1989). One argument is that the redshifted HI component is only seen towards the radio core and not against the inner jet (van der Hulst et al. 1983). The detection of 2cm H₂ CO in absorption against the inner jet (Seaquist & Bell 1990) is not necessarily a proof against the infall hypothesis, if the inner circumnuclear molecular disk is extended on scales of 300 pc. On the other hand, the lack of redshifted HI absorption against the inner jet could also mean that while the size of the HI component at the systemic velocity is 300 pc (in order to cover both the jet and the core), the extent of the redshifted gas is smaller.

HI absorption has been detected in 9 elliptical galaxies (van Gorkom et al. 1989, Mirabel 1990 and references therein). In all cases the absorption is redshifted with respect to the systemic velocity. It is not clear whether this preponderance of redshifted absorption reflects a true infall of gas or if it results from a systematic offset in the systemic velocities towards the blue. Such systematic errors are known to exist due to outflow of emission line gas, which is often used to derive the systemic velocity. In spiral galaxies the situation is different. Here HI absorption is often seen as a single broad line, extending into both the blue- and redshifted sides around the systemic velocity (cf. Dickey 1986). With higher spatial resolution, the broad HI component is decomposed into a narrow one, shifting across the finite extent of the background radio source, over all velocities defined by the rotation of the disk (Koribalski et al. 1993). These absorptions originate in fast rotating circumnuclear disks or rings, of size 200 pc, and with velocities 200 km s^-1 (e.g. NGC 253, 660, 1808, 3079, 4945, Milky Way).

The same phenomenon could be occuring in the center of Cen A, where the presence of a nuclear disk or ring of 100 pc radius and rotating with a velocity of 220 km s^-1 has been established (Rydbeck et al. 1993). High rotational velocities at small galactocentric distances are naturally occuring in elliptical galaxies due to the high mass concentration towards the center (cf. Hernquist 1990) and non-circular orbits for the gas can be generated through non-axisymmetric gravitational instabilities (e.g. bar) or in the form of tri-axiality of the elliptical galaxy itself. Since the angular extent of the continuum source in Cen A is 2 mas (6000 AU), either blue or redshifted gas, depending on the orientation of the orbits, can be seen in absorption.

This situation is reminiscent of what happens in the center of the Milky Way. Here the velocities are more strongly non-circular, maybe because the central bar is oriented at 20- from the Sun line of sight (e.g. Blitz & Spergel 1991, Weinberg 1992). absorption in front of the central continuum source SgrA reveals a broad component between -210 to -110 kms, interpreted as coming from the 200 pc nuclear disk, and four narrower features (at -51, -30, -2 and 32 km s^-1) corresponding to known spiral arms in the galactic disk (Linke et al. 1981). The broad component at negative velocities is also seen in absorption in front of SgrB2, in as well as and HCN (Linke et al. 1981, Greaves & Nyman 1996). The size of the millimeter continuum source SgrA has recently been determined through VLBI at 3 and 7 mm (Krichbaum et al. 1994), and is 0.33 and 0.75 mas respectively. At those frequencies, the interstellar scattering becomes negligible, so these figures are believed to be the actual source sizes (Krichbaum et al. 1994). Since these source sizes correspond to 5 AU at the Galactic Center distance, the absorption features prove that the apparent line of sight velocity dispersion can be quite high in a typical edge-on nuclear disk, due to the accumulation on the line of sight of differential non-circular motions.

The emitting molecular gas in Cen A is confined to the center region, consisting of the nuclear disk or ring at a radius of 100 pc and an outer ring (or spiral arm) at 750 pc. The absorption complexes are likely to be associated with these features, but since absorption is sensitive to diffuse and low excitation molecular gas which is generally not seen in emission (e.g. Lucas & Liszt 1996), it is possible that some intervening molecular gas detected in absorption is associated with the larger HI disk extending to 7 kpc (Schiminovich et al 1994). As we have seen in the previous analysis, the molecular gas in both LV and HV components is diffuse, with abundance ratios similar to Galactic values. The main difference is that gas in the LV components has a higher kinetic temperature than the HV components. We cannot differentiate between inner and outer molecular gas from this alone. There are, however, two facts which suggest that the HV components are associated with the nuclear gas and the LV components with the outer disk: (1) the LV components are close to the systemic velocity and, (2) whereas the LV components extend between 540-556 km s^-1, the HV components are spread out between 576-640 km s^-1, 4 times larger. Although the rotational velocities are similar for the inner and outer molecular gas (Rydbeck et al. 1993), the inner region has a larger velocity gradient and this gas seen in absorption should be spread over a larger velocity interval.

The blue-shifted 'wing', from 500 to 540 km s^-1, has not been considered above. This feature is seen at a very low level, and depends on the subtracted emission profile shape, and should be viewed with caution. Could this gas comes from inside the cavity delineated by the circumnuclear ring? There is a constraint on the radius where molecules can subsist, around a luminous ionizing X-ray source. Maloney et al. (1994) derived an effective ionization parameter :

where = erg s^-1, n(H₂)/10⁹ cm^-3 and is the distance from the X-ray source in parsecs. The gas will not become substantially molecular unless the effective ionization parameter is smaller than 10^-3. From ROSAT HRI measurements, Döbereiner et al (1996) have estimated an X-ray luminosity (0.1-2.4 kev) for the Cen A nucleus of erg s^-1 in 1994, after correcting for absorption. For diffuse gas with n(H₂) , and a column density N(H₂) 10²² cm^-2, the minimum distance from the center is 100 pc. For molecular gas to exist at smaller galactocentric distances in Cen A, both the volume and column densities must be considerably higher. It is therefore likely that the molecular gas corresponding to the blue-shifted wing is at least at the distance of the circumnuclear ring. It is possible that this wing corresponds to the 750 pc component, which should also possess non-circular motions. The orientation of orbits in a tumbling non-axisymmetric component change by 90 at each resonance (e.g. Contopoulos & Grosbol 1989). Along the nucleus line of sight, it is therefore possible that this blue-shift component corresponds to elliptical streamlines perpendicular to that in the circumnuclear ring.

© European Southern Observatory (ESO) 1997

Online publication: May 26, 1998

helpdesk.link@springer.de