Luminosities and mass-loss rates of carbon stars in the Magellanic Clouds

Dust radiative transfer models are presented for 60 carbon stars in the Magellanic Clouds (MCs) for which 5-35 mu m Spitzer infrared spectrograph (IRS) spectra and quasi-simultaneous ground-based JHKL photometry are available. From the modelling, the luminosity and mass-loss rate are derived (under the assumption of a fixed expansion velocity and dust-to-gas ratio), and the ratio of silicon carbide (SiC) to amorphous carbon (AMC) dust is also derived. This ratio is smaller than observed in Galactic carbon stars, as has been noted before. Light curves for 36 objects can be retrieved from the massive compact halo object (MACHO) and optical gravitational lensing experiment (OGLE) data bases, and periods can be derived for all but two of these. Including data from the literature, periods are available for 53 stars. There is significant scatter in a diagram where the mass-loss rates are plotted against luminosity, and this is partly due to the fact that the luminosities are derived from single-epoch data. The mass-loss rates for the MC objects roughly scatter around the mean relation for Galactic C-stars. The situation is better defined when the mass-loss rate is plotted against pulsation period. For a given period, most of the Large Magellanic Cloud (LMC) and Small Magellanic Cloud (SMC) stars have mass-loss rates that are in agreement with that observed in Galactic carbon stars (under the assumption that these objects have an expansion velocity and dust-to-gas ratio typical of the mean observed in Galactic carbon Miras). For some SMC sources only, the IRS spectrum at longer wavelengths falls clearly below the model flux predicted by a constant mass-loss rate. An alternative model with a substantial increase of the mass-loss rate to its present-day value over a time-scale of a few tens of years is able to explain the spectral energy distribution (SED) and IRS spectra of these sources. However, the probability to have two such cases in a sample of 60 is small, and makes this not a likely explanation (and testable by re-observing these objects near the end of the lifetime of Spitzer). Alternative explanations are (ad hoc) changes to the dust emissivity at longer wavelengths, and/or deviations from spherical symmetry.