Are dust shell models well-suited to explain interferometric data of late-type stars in the near-infrared?

Abstract

Recently available near-infrared interferometric data on late-type stars show a strong increase of diameter for asymptotic giant branch (AGB) stars between the K (2.0-2.4 m) and L (3.4-4.1 m) bands. Aiming at an explanation of these findings, we chose the objects Orionis (Betelgeuse), SW Virginis, and R Leonis, which are of different spectral types and stages of evolution, and which are surrounded by circumstellar envelopes with different optical thicknesses. For these stars, we compared observations with spherically symmetric dust shell models. Photometric and 11 m interferometric data were also taken into account to further constrain the models. -We find the following results. For all three AGB stars, the photosphere and dust shell model is consistent with the multi-wavelength photometric data. For Orionis the model dust shell has a very small optical depth (0.0065 at 11 m); the visibility data and model in K and L are essentially entirely photospheric with no significant contribution from the dust, and the visibility data at 11 m show a strong dust signature which agrees with the model. For SW Virginis the model dust shell has a small optical depth (0.045 at 11 m); in K the visibility data and model are essentially purely photospheric, in L the visibility data demand a larger object than the photosphere plus dust model allows, and at 11 m there was no data available. For R Leonis the model dust shell has a moderate optical depth (0.1 at 11 m); in K and L the visibility data and model situation is similar to that of SW Vir, and at 11 m the visibility data and model are in agreement. -We conclude that AGB models comprising a photosphere and dust shell, although consistent with SED data and also interferometric data in K and at 11 m, cannot explain the visibility data in L; an additional source of model opacity, possibly related to a gas component, is needed in L to be consistent with the visibility data.