To investigate the atmospheric dynamics and mass loss of C-rich long-period variables we have calculated radiation-hydrodynamical models of the atmosphere and circumstellar dust shell which include a detailed description of the dust condensation. Simulating the stellar pulsation by a variable inner boundary we have varied the physical parameters of the models systematically to demonstrate their influence on the time-dependent behaviour and time-averaged mass loss characteristics of our models.
The dependence of the mass loss rate on stellar parameters predicts a strong increase of mass loss as stars evolve along the AGB. As the stellar luminosity increases, while simultaneously the mass and effective temperature decrease, the atmosphere becomes more extended which leads to more favourable conditions for dust formation and higher densities in the acceleration region of the wind. In accordance with Bowen & Willson (1991) we argue that the development of a "superwind" () is a natural consequence of the evolution of stellar parameters on the AGB. The increase of mass loss with luminosity in our models is much steeper than predicted by Reimers' law (which was found empirically for the RGB, not the AGB) and is in reasonable agreement with the description of mass loss used by Blöcker (1995) for calculating stellar evolution on the AGB.
The models presented in this paper agree nicely with mean mass loss-period relations deduced from observations (Groenewegen 1995, Whitelock 1990). The observed scatter of mass loss rates for a given period (about one order of magnitude) can be reproduced with the models, e.g. by differences in the pulsation amplitude (which affects the shock strength and, consequently, the levitation of the atmosphere) or in the abundance of condensible material (non-linearities of the wind mechanism) for otherwise identical models. The wind velocities are well correlated with , a quantity characterizing the strength of radiation pressure on dust relative to gravitation.
The shape and long-term behaviour of IR light curves are determined both by stellar pulsation and dust formation. While the first leads to bolometric variations the second causes rather a spectral redistribution of the stellar radiation. Thus, in principle, it should be possible to disentangle the effects of pulsation from the physical processes in the circumstellar envelope by simultaneously monitoring LPVs in different wavelengths, probing different regions of the atmosphere and circumstellar dust shell. Since the time-dependent behaviour of the circumstellar envelope depends strongly on various model parameters this could help to restrict physical parameters of individual stars.
© European Southern Observatory (ESO) 1997
Online publication: July 3, 1998