1994A&A...288..231S

We present the first line blanketed hydrodynamic models of spherically expanding atmospheres of hot stars. The models are characterised by a simultaneous solution of the equation of motion, the non-LTE populations of hydrogen and helium, and radiation transfer in a line blanketed atmosphere. The entire domain from the optically thick photosphere out to the terminal velocity of the wind is treated. The radiative forces are evaluated consistently with the depth-dependent radiation field, taking into account multiple scattering by metal lines and line overlap. This allows us to determine mass loss rates and the velocity field resp. density structure, as well as to predict the line blanketed energy distribution, the photospheric absorption lines, and the emission features emerging from the wind. The major improvements over unified non-LTE model atmospheres advocated by the Munich group (Gabler et al. 1989) are twofold: 1) The effects of line blanketing for the radiation transfer and statistical equilibrium of hydrogen and helium are included in the atmosphere calculations. 2) The radiative force (resp. line force parameters k, α) is evaluated using the depth-dependent radiation field of the model atmosphere We present a detailed discussion of the influence of the photosphere-wind transition zone on line profiles and the effects of line blanketing on a hydrodynamic non-LTE model atmosphere. Two important results are obtained from our study: (1) We quantify the influence of line blanketing on the atmospheric structure and on the predicted spectrum. In particular, we qualitatively confirm the results obtained with core-halo models and find that the corrections of Abbott & Hummer (1985) and Bohannan et al. (1986, 1990) are also quantitatively correct. (2) We show that even "purely" photospheric lines, on which spectroscopic determinations of basic stellar parameters rely, are strongly affected by the velocity field in the transition zone between the photosphere and the wind, and not only by the mass loss rate. Thus, for the more luminous OB stars spectroscopic analyses not only depend on three parameters (g, T_eff, H/He abundance), but also on the atmospheric structure of the wind (i.e. {dot}(M), v(r)). Therefore, we add new evidence to the previously stated finding that for precise determinations of stellar parameters and abundances of hot luminous stars, the use of plane parallel models may lead to systematic errors. This implies that the recent finding of discrepancies of spectroscopic masses and helium abundances compared to predictions of standard evolutionary models could be due to the inappropriateness of photospheric models for the analysis of luminous stars. The stellar parameters of our models are those thought to be representative for the O4 I(n)f star ζ Puppis. A comparison of the synthetic spectra with the observations shows that our model fits are not satisfactory. We find good agreement only for the key lines of a spectroscopic analysis, i.e. Hγ, HeI λ4471, and HeII λ 4542. However, for all lines that show wind features our predictions are clearly not correct. Since a spectroscopic analysis is a multi-dimensional problem it is impossible to single out one stellar parameter that is responsible for the failure of the model. We tentatively interpret our result as an indication that the calculated wind structure is not correct. The reason is not obvious, but it could be simply that the commonly adopted distance to the star is wrong. In any case, the spectrum of ζ Puppis should be carefully reanalysed with hydrodynamic model atmospheres.