Modelling the dust emission from dense interstellar clouds: disentangling the effects of radiative transfer and dust properties

Context. Dust emission is increasingly used as a tracer of the mass in the interstellar medium. With the combination of Planck and Herschel observatories, we now have both the spectral coverage and the angular resolution required to observe dense and cold molecular clouds. However, as these clouds are optically thick at short wavelengths but optically thin at long wavelengths, it is tricky to conclude anything about dust properties without a proper treatment of the radiative transfer. Aims. Our aim is to disentangle the effects of radiative transfer and dust properties on the variations in the dust emission at long wavelengths. This enables us to provide observers with tools to analyse the dust emission arising from dense clouds. Methods. We model cylindrical clouds with visual extinctions between 1 and 20 mag, illuminated by the standard interstellar radiation field, and carry out full radiative transfer calculations using a Monte Carlo code. Dust temperatures are solved using the DustEM code for amorphous carbons and silicates representative of dust at high Galactic latitude (DHGL), carbon and silicate grains coated with carbon mantles, and mixed aggregates of carbon and silicate. We also allow for variations in the optical properties of the grains with wavelength and temperature. We determine observed colour temperatures, T-colour, and emissivity spectral indices, beta(colour), by fitting the dust emission with modified blackbodies using a standard chi(2) fitting method, in order to compare our models with observational results. Results. Radiative transfer effects can explain neither the low T-colour, the increased submillimetre emissivity measured at the centre of dense clouds, nor the observed beta(colour) - T-colour anti-correlation for the models considered. Adding realistic noise to the modelled data, we show that it is unlikely to be the only explanation of the beta(colour) - T-colour anti-correlation observed in starless clouds, which may instead be explained by intrinsic variations in the grain optical properties with temperature. Similarly the higher submillimetre emissivity and the low T-colour have to originate in variations in the grain optical properties, probably caused by their growth to form porous aggregates. We find it difficult to infer the nature of the grains from the spectral variations in their emission, owing to radiative transfer effects for lambda less than or similar to 300 mu, and to the mixture of different grain populations for longer wavelengths. Finally, the column density is underestimated when determined with modified blackbody fitting because of the discrepancy between T-colour and the true dust temperature in the innermost layers of the clouds.