ar X iv : a st ro - p h / 05 11 17 4 v 1 7 N ov 2 00 5 The metallicity dependence of WR wind models

With the advance of stellar atmosphere modelling during the last few years, large progress in the understanding of Wolf-Rayet (WR) mass loss has been achieved. In the present paper we review the most recent developments, including our own results from hydrodynamic non-LTE model atmospheres. In particular, we address the important question of the Z-dependence of WR mass loss. We demonstrate that models for radiatively driven winds imply a rather strong dependence on Z. Moreover, we point out the key role of the L/M-ratio for WR-type mass loss. 1. WR-type stellar winds WR stars show exceptionally strong winds with mass loss rates of the order of the single-scattering limit or above. The observed wind efficiency numbers η = ˙ M v ∞ /(L ⋆ /c), which denote the ratio between the wind momentum and the momentum of the radiation field, are typically in the range of 1-5. Therefore, if WR-type winds are driven by radiation, photons must be used more than once, i.e., after absorption photons must either be redistributed into different wavelength regimes or they must be scattered more than once by overlapping lines in the extended wind. Because of their high wind densities, WR stars develop extended atmospheres where the hydrostatic layers are completely hidden to the observer. In combination with their strong radiation fields, they develop extreme non-LTE conditions that require sophisticated modeling techniques. The key question concerning WR mass loss is why a star becomes a Wolf-Rayet star. The observed parameters of early-type O stars, namely their lumi-nosities and effective temperatures, are in principle comparable with late-type WN stars. O supergiants even tend to show a similar He-and N-enriched surface composition. Because of their higher mass loss rates, however, WR stars show dramatically different spectra with strong and broad wind emissions where O stars show tiny photospheric absorption lines. If the WR winds are radiation-driven like OB star winds, one would also expect a similar Z-dependence of their mass loss. In this case, however, the question must be addressed why WNL stars are so different from O stars. 2. Monte-Carlo models The first wind models which were able to reproduce the high efficiency factors of WR winds were obtained by Monte-Carlo techniques (Lucy & Abbott 1993; Springmann 1994). In this approach the path of single photon energy packets,