Astron. Astrophys. 349, 553-572 (1999)

1. Introduction

In order to understand the chemical evolution of our own and external galaxies, we must have a clear understanding of the evolution of their massive stars. Whilst massive stars constitute only a small fraction of the stellar population within a galaxy, they play an important and fundamental role in its chemical evolution. They are efficient processors of the chemical elements, returning nucleosynthetically enriched material to the interstellar medium (ISM) through stellar winds and supernovae (see, e.g., the review by Maeder & Conti 1994).

Our understanding of the post-main sequence evolution of massive () stars is hindered by a number of observational and theoretical uncertainties. Stellar structure calculations (for example, those of Maeder & Meynet 1989and Schaller et al. 1992) are extremely sensitive to the treatment of convection processes, mass loss and stellar rotation - phenomena which can affect both the evolutionary tracks and, more tangibly, the surface chemistry of such objects. Blue supergiants (BSG) in particular, represent a key challenge to stellar evolutionary theory, as a number of their observed properties remain at odds with current predictions; most notable is their number distribution in the Hertzsprung-Russell diagram - the ratio of blue to red supergiants (RSG) is not reproduced by evolutionary calculations (see Langer & Maeder 1995and references therein) while the predicted post-main sequence gap at high luminosity is not observed (Tuchman & Wheeler 1990).

An important indicator of evolutionary status is provided by the stellar surface chemistry and considerable qualitative evidence for photospheric chemical peculiarities in luminous stars is available. Jaschek & Jaschek (1967) and later Walborn (1972, 1976) presented evidence for anomalies in the carbon and nitrogen linestrengths in OB stars, leading to the spectral classifications OBN/OBC - analogous to the classifications of Wolf-Rayet stars which were already in use. The OBN/OBC stars have respectively strong/weak nitrogen lines for their spectral type with, in general, the nitrogen variations being negatively correlated with those of carbon; additionally, many stars with only moderate CN anomalies have been identified - thus the morphology may reflect a continuous range of surface chemistries, rather than a strict dichotomy. CNO linestrength anomalies in B-type supergiants are also found in the lower metallicity environments of the Large and Small Magellanic Clouds (Fitzpatrick & Bohannan 1993 and Lennon 1997, respectively).

Such phenomena are broadly supported by current stellar evolutionary theories, which predict that some massive stars should show variations in their surface abundances indicative of CNO core-processed material being present in the photosphere. However, the mechanism by which such matter is mixed to the stellar surface is still subject to some debate. For example, models by Schaller et al. (1992) imply that some massive stars may evolve through so-called `blue loops' (main sequence - BSG - RSG - BSG). During their time as RSGs, massive stars develop appreciably enriched photospheres through the convective mixing of core-processed material. After the commencement of core helium burning, the star evolves towards higher effective temperatures once more. Thus, two distinct populations of BSG are postulated - the pre-RSG objects which would be chemically normal and the post-RSG objects which would show evidence of nucleosynthetically processed material having been mixed to their surfaces. Alternatively, Denissenkov (1994) - see also Talon et al. (1997) - has considered the main sequence evolution of a 10 star, including the effect of rotationally induced turbulent diffusive mixing. He has shown that this important mechanism can provide efficient mixing during the main sequence lifetime , thus removing the need for blue loops to explain the OBN/OBC phenomenon.

There is some observational evidence which would favour the latter evolutionary scenario. Schönberner et al. (1988) examined the helium and CNO spectra of 4 late O-type, near main sequence objects, finding substantial changes in their CN abundances, with no evidence for a change in their oxygen abundances. As these effects were accompanied by moderate helium enhancements, they attribute them to partial CN-process mixing. This suggests that mixing of core-processed material has occurred on the main sequence. Venn (1995b) considered the CNO abundances of 22 Galactic A-type supergiants, and found N/C ratios which were also consistent with partial mixing having occurred on or near the main sequence.

As well as Schönberner et al., there are a number of authors who have estimated enhanced helium fractions for luminous, early-type stars. Voels et al. (1989), Lennon et al. (1991b), Herrero et al. (1992) and Smith & Howarth (1994) have all found considerable helium enhancements for luminous OB-stars, and in some cases a correlation with CN anomalies. This phenomenon has been referred to as the `helium discrepancy', as the magnitude of the observed helium enhancements exceeds that predicted by theory. However whilst variations in He I linestrengths are undoubtedly real, they need not necessarily reflect large variations in abundances. McErlean et al. (1998- hereafter referred to as MLD) have considered the effect of microturbulent line broadening, which was not fully included in the above papers, and for the two representative early B-type supergiants, & Ori, have found lower (and near-normal) helium fractions. If these effects are present in other luminous OB-stars, then published helium enhancements must be considered uncertain. Indeed, Smith & Howarth (1998) have independently shown that the use of microturbulence leads to a near-solar helium fraction in the O9.7 Iab supergiant HD 152003, which lends support to this idea.

The primary motivation of the work presented here was to analyse a statistically significant sample of B-type supergiant spectra in a consistent, quantitative manner and hence to constrain their possible evolutionary histories. In two previous papers (Lennon et al. 1992 , 1993- hereafter referred to as Papers I and II, respectively) a spectral atlas for B-type supergiants was presented together with a tabulation of equivalent width estimates. By examining these data as a function of spectral type, qualitative evidence for variations in their carbon and nitrogen abundances was presented. In this paper, we shall use non-LTE model atmospheres and line formation computations to investigate the physical properties of this sample in a more quantitative manner. Linestrengths will be analysed to yield estimates of absolute, non-LTE abundances and in so doing, we will be able to offer a critique of the uses and limitations of such methods. We will also attempt to quantify variations in the CNO element abundances.

To investigate the possible effect of metallicity, the sample of Galactic supergiants considered here has been supplemented by spectroscopic observations of a further 63 B-type supergiants in the SMC (Lennon 1997); a differential analysis of the two datasets will be presented in a companion paper.

© European Southern Observatory (ESO) 1999

Online publication: September 2, 1999
helpdesk.link@springer.de