Astron. Astrophys. 339, 19-33 (1998)

5. Discussion and conclusions

We have shown that with simple assumptions about the birth of massive stars and their relationships with the ISM, we can reproduce qualitatively and quantitatively the characteristics of the UV, H and FIR emissions of a particular object,the Sc galaxy NGC 6946. For such a galaxy with a prominent spiral structure, having a large mass of neutral gas, and forming stars actively, the observed far infrared emission is produced both in molecular gas and in the diffuse atomic gas. More precisely, 54% of the FIR (60-200 ) emission comes from dust grains in giant molecular clouds. Dust in the diffuse neutral atomic gas contributes to about 46% of the total FIR luminosity.

We have evaluated the respective contributions of the UV radiation from massive stars and of the radiation field from the old stellar population. We find that 72% of the FIR luminosity can be attributed to UV heated dust grains, which reside mostly in molecular clouds envelopes. The remaining 28% is due to dust heated by the radiation field from the old stellar population at locations far away from OB associations.

We have calculated the emission of the model galaxy in the fine structure line of C⁺ at 158 . In the spiral arms, photon dissociation regions at the surfaces of molecular clouds are the main source of the emission. We have found a large arm-interarm contrast in this line. This effect could be tested by high angular resolution maps of galaxies. It results naturally from the combination of a lower gas density and lower radiation field in the interarm regions, because of the short mean free path for UV photons, 440 pc. As a whole, PDRs represent 76% of the emission. The contribution from the diffuse phase is found to be 24%. Our model is able to account for about 40% of the observed C⁺ emission of NGC 6946. The emission from PDRs should be viewed as a lower limit since we use the model by Le Bourlot et al. (1993) with low abundances of carbon and other elements in the gas phase: [C]/[H]= 3 10^-5. The average value is 1.3 10^-4 for Galactic diffuse clouds (Snow & Witt 1996), a factor of 4 larger than the value used in the model. Since the C⁺ 158 emission scales roughly with the column density, hence the carbon abundance, the total C⁺ luminosity from PDRs could be larger by at least a factor three than our current model prediction. This would increase the contribution from PDRs to the total C⁺ emission of the model galaxy: with this scaling factor, the predicted C⁺ luminosity of PDRs would reach . Moreover, the C⁺ emission from the diffuse gas is overestimated, because the model of Wolfire et al. (1995) assumes [C]/[H] 3 10^-4 in the gas. Thus the diffuse emission could be 2-3 times smaller than in our standard model. Adopting [C]/[H]=1.3 10^-4 in both phases would thus enhance the differences of between dense and diffuse gas.

The knowledge of the cloudy nature of the ISM, and of the global structure of the galaxy, is important to determine how far UV photons can travel away from OB associations. The filling factor and the mass/radius scaling law appear to be major parameters for the transfer of stellar radiation in the galaxy disk, because they determine at the same time the obscuration and the size of the emitting regions. Other important parameters are the number of OB associations and the sizes of HII regions, because with a large number of OB associations or with small HII regions, molecular clouds are on average closer to massive stars, and are thus more efficiently heated.

In all the models we ran, we have found that the average internal UV opacity is of the order 0.8. The discs are therefore moderately opaque in the UV, as measured by Buat & Xu (1996). This moderate opacity holds for face-on discs. Edge-on discs are quite opaque, with a small fraction of the luminosity escaping, less than 1.0% of the face-on luminosity. This fraction corresponds to an equivalent extinction of 5 magnitudes in the UV.

These results have been obtained using a crude description of the interstellar medium. The adopted spatial resolution results from a compromise between astrophysical requirements and computational needs, but is certainly very poor compared to the complexity of the interstellar medium. The good agreement of the observed and predicted large scale properties shows nevertheless that the transfer of UV radiation, and the role of the radiation for the gas and dust heating, are correctly described at the 12 pc scale. This is in agreement with previous works estimating that dust heating by UV radiation occurs principally at large distances from massive stars (Murthy et al. 1992, Leisawitz & Hauser 1988).

© European Southern Observatory (ESO) 1998

Online publication: September 30, 1998
helpdesk.link@springer.de