2011A&A...532A..56G

We are studying the effects of submm observations on the total dust mass and thus dust-to-gas mass ratio measurements. We gather a wide sample of galaxies that have been observed at submillimeter (submm) wavelengths to model their spectral energy distributions using submm observations (>160µm) and then without submm observational constraints in order to quantify the error on the dust mass when submm data are not available. Our model does not make strong assumptions on the dust temperature distribution to precisely avoid submm biaises in the study. Our sample includes 52 galaxies observed at submm wavelengths. Out of these, 9 galaxies show an excess in submm which is not accounted for in our fiducial model, most of these galaxies being low-metallicity dwarfs. We chose to add an independant very cold dust component (T=10K) to account for this excess. We find that metal-rich galaxies modelled with submm data often show lower dust masses than when modelled without submm data. Indeed, these galaxies usually have dust SEDs that peaks at longer wavelengths and require constraints above 160µm to correctly position the peak and sample the submillimeter slope of their SEDs and thus correctly cover the dust temperature distribution. On the other hand, some metal-poor dwarf galaxies modelled with submm data show higher dust masses than when modelled without submm data. Using submm constraints for the dust mass estimates, we find a tightened correlation of the dust-to-gas mass ratio with the metallicity of the galaxies. We also often find that when there is a submm excess present, it occurs preferentially in low-metallicity galaxies. We analyse the conditions for the presence of this excess and find a relation between the 160/850µm ratio and the submm excess.