2010MNRAS.406.1190D

We present the first published unbiased, automated approach to fitting infrared observations of interstellar dust clouds for the purposes of constraining their underlying physical properties. The dust radiative transfer problem is solved self-consistently for a given set of physical inputs. Spherical geometry is chosen to minimize the run time, given the large number (∼ millions) of models necessary. The model output is convolved with appropriate beams to produce simulated observations, which are then compared with observational data. A best fit is achieved using the Levenberg-Marquardt DNLS1 routine from the SLATEC library, to minimize the chi-squared deviation. The speed of the model allows a large number of initial starting conditions to be considered. While any number of parameters can be fitted, we concentrate on the dust density and grain property distribution. We apply the model to the well-studied source B335. While much of the envelope is well modelled by a power-law exponent ∼1.5-1.9, we find potential evidence for a different inner structure for r < ∼10arcsec, as well as an outer tenuous envelope for r > 120arcsec. We also find evidence of a variation in grain properties with radial position. The interior and exterior have larger concentrations of both thick ice mantles and bare grains. The properties in the interior would be consistent with a non-spherical structure seen by Harvey et al. (2003b) and Stutz et al. (2008), and the exterior would be consistent with incomplete refreezing of ice on to mantles after an evaporative event, such as a shock, in the last ∼10⁵ yr.