There are a number of important theoretical uncertainties, for example the treatment of convection processes, mass loss and rotation, concerning the evolution of massive stars. Such processes generally lead to varying degrees of surface contamination by nucleosynthetically produced material from the stellar interior (see Maeder & Meynet 1989 for example). Therefore the surface chemical composition of massive early-type stars is a vital diagnostic tool for understanding their evolution. Blue supergiants in particular may be considered as key tests of stellar evolution since neither their very distribution in the Hertzsprung-Russell diagram, nor their number ratio relative to red supergiants, both as a function of metallicity, are accounted for by theory (for recent discussions of these problems see Maeder 1994 and Langer & Maeder 1995). It is therefore vital that reliable surface abundances be obtained for such stars, in particular for the elements helium, carbon, nitrogen and oxygen (see also Fliegner et al. 1996 for a discussion of the importance of boron).
For O- and early B-type stars, one is in the fortunate position of having access to profiles of both neutral and ionised helium lines, which, together with the hydrogen Balmer lines, may be used to estimate the helium abundance simultaneously with the surface gravity and effective temperature. There have been a number of previous analyses using plane-parallel non-LTE models. For example, Voels et al. (1989) used wind-blanketed models to analyse four stars of spectral type O9.5 that form a sequence in luminosity class (Ia, Ib, II, V). They found an enhanced helium fraction, y, of 0.18 0.03 by number, only for their most luminous Ia supergiant, Cam. Lennon et al. (1991) also found enhanced helium abundances for the B0.5 Ia supergiants Ori, Sk-68 41 and Sk 159 (y = 0.20, 0.23 & 0.35 respectively). More recently, Herrero et al. (1992) analysed a sample of 25 Galactic luminous O-type stars, finding enhancements for approximately half of these objects. Finally, Smith & Howarth (1994), in their analysis of three O9.5 Iab supergiants, gave evidence for more marginal helium enhancements and a correlation with the stars' CN abundances.
A common problem encountered in this previous work was that a consistent fit to all the available He I lines could not be obtained for a given helium abundance. Voels et al. (1989) attributed this to the extended spherical nature of the supergiant photosphere leading to a strengthening of triplet absorption lines relative to singlet lines - the so-called `generalized dilution effect'. They therefore gave greatest weight to the weakest lines, which they argued formed deeper in the atmosphere where extension (and wind) effects are less important. Smith & Howarth (1994) used the same line of argument in their analysis of O9.5 supergiants and it is noticeable that both these papers derive only moderate helium enrichments compared to the results of Lennon et al. (1991), where more substantial overabundances are found. These latter authors however adopted a mean abundance from all lines considered, but commented in detail on the difficulties involved in obtaining a consistent fit for all the He I lines, a problem which was further elaborated upon by Lennon (1994). All these analyses however have neglected the effect of microturbulence on the line formation process, despite the fact that microturbulent velocities comparable to the sound speed are consistently deduced from analyses of metal lines in B-type supergiants (see, for example, Lennon et al. 1991).
© European Southern Observatory (ESO) 1998
Online publication: December 8, 1997