3. Results and discussion
The 850 µm image shows a bright nucleus and several features that clearly trace the spiral arms (in Fig. 1 the sub-mm contours are overlayed to a U-band image of the galaxy (Trewhella 1998)) As already seen in optical images (Tacconi & Young 1990), the spiral arms originating in the northeast quadrant are more pronounced than the others, where only regions with bright HII regions have detectable emission in the sub-mm. The 850 µm image presents a striking similarity to the 12CO(2-1) emission map in Sauty, Gerin & Casoli (1998), observed with the IRAM 30m radiotelescope with a comparable resolution (13"). The image of molecular line emission is also shown in Fig. 1, with the 850 µm contour overlayed. The similarity with the sub-mm image is hardly surprising, since the molecular gas is the dominant component of the ISM over the optical disk of NGC 6946 (Tacconi & Young 1986). The nucleus is elongated in the direction north-south, as observed for the central bar of molecular gas (Ishizuki et al. 1990; Regan & Vogel 1995).
Emission associated with a more diffuse atomic gas component cannot be detected, for several reasons. First of all the face-on inclination of the galaxy: since dust is optically thin to its own emission a faint component can be observed only if the dust column density is large. This is the case for the high inclination galaxies NGC 7331 (Bianchi et al. 1998) and NGC 891 (Alton et al. 1998b), where higher signal to noise were obtained coadding a smaller number of observations. The large face-on galaxy M51 has been observed using the scan-map mode and confirms the necessity of long integrations (Tilanus, private communication). Furthermore, chopping inside the source field removes not only the emission from the sky but also from possible components with a shallow gradient: this may be the case for dust associated with the flat HI distribution in NGC 6946 (Tacconi & Young 1986). Finally, a faint diffuse emission could have been masked by the mentioned artifacts and subtracted together with the curved background.
The 450 µm image is much noisier than the 850 µm one, because of the larger sky emission at this wavelength. Only a central region of x can be clearly detected, although most of the features at a 3- level correspond to regions emitting in the long wavelength image.
The temperature from the two sub-mm fluxes can be measured only for the central region with significant 450 µm flux. Sub-mm fluxes are 1.2 Jy at 850 µm and 9.3 Jy at 450 µm. We checked for the contribution of the strong 12CO(3-2) line emission at 346 GHz to the 850 µm flux using the observation of NGC 6946 centre in this line reported by Mauersberger et al. (1999) for a beam of 21". Converting from the original units to Jansky (Braine et al. 1995) and averaging over the 30 GHz bandwidth of 850 µm filter (Matthews 1999), a flux density of 80 mJy/beam is derived. As pointed out by the referee, the pointing of the Mauersberger et al. observations was offset from the strongly concentrated central emission by nearly one beamwidth in the SW direction. Using the 12CO(2-1) image as a template of the higher state emission, we corrected for the offset and derived the flux for the central region, larger than the beam. A total contribution of 0.6 Jy is derived for the 346 GHz line (50% of the 850 µm flux). However, this large contribution is due to the high density and gas temperature of the central region. In fact, for 850 µm fluxes on larger apertures, the contamination is much smaller: Israel et al. (1999) derive a contribution of only 4% to the total sub-mm flux of NGC 891. Therefore, the derivation of cold dust temperature at large galactic radii (Alton et al. 1998b) are not severely biased. We did not correct the 450 µm flux for the contribution of the 12CO(6-5) line, that lies at the edge of the filter (Israel et al. 1999).
After the correction, the dust temperature of the central region is T=346 K, where the large quoted error comes from the calibration uncertainties. Here and in the following, dust temperature and masses are computed using the emissivity law derived by Bianchi, Davies & Alton (1999a) from observation of diffuse FIR emission and estimates of optical extinction in the Galaxy. For a wavelength dependence of the emissivity , changing smoothly from to at 200 µm (Reach et al. 1995) they obtain , where is the extinction efficiency in the V-band (1.5; Casey 1991). Lacking information outside of the centre, a mean temperature for a larger aperture can be derived from the lower resolution IRAS and ISO images at 100 µm and 200 µm (Alton et al. 1998a). The total flux inside the B-band half light aperture (5´ in diameter) is 24040 Jy at 100 µm and 28040 Jy at 200 µm (Bianchi, Davies & Alton 1999b). The temperature from IRAS and ISO fluxes is T=242 K.
We derived a point-to-point correlation between the 850 µm flux and the 12CO(2-1) line, resampling the sub-mm image to the same pixel size as the line emission map (10", roughly equivalent to both beam sizes) and using all positions with signals larger than 3- in both observations. A linear correlation is found (Fig. 2). Assuming a mean dust grain radius 0.1µm and mass density =3 g cm-3 (Hildebrand 1983), the emissivity of Bianchi et al. (1999a) 1 and T=24K, the dust column density and hence the mass along the line of sight can be easily computed. The molecular gas column density has been derived from the 12CO(2-1) emission using a conversion factor appropriate for the 12CO(1-0) emission in the general ISM in the Galaxy (X=1.8 1020 cm-2 K-1 km-1 s; Maloney 1990) and a line ratio I(2-1)/I(1-0)=0.4 (Casoli et al. 1990). The slope of the linear correlation can then be converted into a gas-to-dust mass ratio of 17020, a value very close to the local Galactic one (160; Sodroski et al. 1994). This confirms the association of dust with the local dominant phase of the galactic ISM.
The dust content of NGC 6946 has been studied carrying out an energy balance between the stellar emission in the optical and the FIR dust emission, through the help of radiative transfer models. If an exponential disk is used to model the dust distribution, a central face-on optical depth is needed to explain the FIR emission (Evans 1992; Trewhella 1998; Bianchi et al. 1999b). The 850 µm image clearly show that the dust distribution is more complex, but still the column densities derived from the sub-mm flux support the idea of an optically thick dust distribution. Under the same assumption of the previous paragraph, the diffuse component of the north-east spiral arms at a 3- level corresponds to a V-band optical depth . The quite high optical depth corresponding to the sky noise () shows how difficult is to obtain sub-mm images of dust emission in the outskirts of face-on galaxies, even for a high sensitivity instrument like SCUBA. Thus, possible extended dust distributions (Alton et al. 1998a) are better revealed through deep sub-mm imaging of edge-on galaxies, where the dust column density is maximized. However, the high inclination makes the interpretation of the dust emission along the line of sight more complex.
© European Southern Observatory (ESO) 2000
Online publication: December 17, 1999