The photospheres of main-sequence B-type stars are often assumed to be uncontaminated by products of internal nuclear processes and to directly represent the chemical composition of their progenitor interstellar medium. This is because stellar evolutionary calculations (see, for example, Schaller et al. 1992) predict that these young objects are in a stable core hydrogen-burning phase.
However recent results have suggested that the atmospheres of some B-type stars could be contaminated by CNO-processed core material mixing to the surface. Gies & Lambert (1992) have studied 39 solar neighbourhood B-type stars with a range of surface gravities, , and effective temperatures, . They concluded that the nitrogen rich stars in their sample had abundances which were possibly consistent with surface enrichment by CNO-cycled material, although they could not rule out the possiblility that the nitrogen anomalies originated in the progenitor gas. Hence they were unable to confirm the strong claim by Lyubimkov (1991; and references therein) that CN-cycled material is clearly detectable in the atmospheres of many B-type core hydrogen burning stars. Some further preliminary evidence for mixing on the main sequence has come from the analysis of boron abundances in early-type objects (Venn et al. 1996, Fliegner et al. 1996).
Quantitative predictions can be made about the redistribution of photospheric abundances from evolutionary models including mixing mechanisms. For example, Maeder (1987) predicted that contamination of a stellar atmosphere by CNO-cycled material can occur as a star is in the process of evolving off the main sequence. In such a model, a nitrogen enhancement of 0.6 dex should be accompanied by a 0.2 dex increase in helium and a 0.2 dex decrease in carbon (with oxygen only slightly depleted). Dennisenkov (1994) has produced a model in which mixing from the core regions to the radiative envelope of a B-type star is achieved by rotationally induced turbulent diffusion. In either case for both models an observed anti-correlation between carbon and nitrogen abundances visible in the stellar photosphere is predicted as a tracer for contamination by core CNO-cycled material.
In two previous papers (Smartt et al. 1996a, 1996b), chemical compositions for a sample of nine main sequence stars with low projected rotational velocities and similar atmospheric parameters has been presented; using differential methods, it was possible to derive relative abundances to an accuracy of better than 0.2 dex. These results were used, for example, to correlate metal abundances with spiral structure and to search for different patterns between individual elements.
In general, a good correlation was found between the differential abundances of silicon and magnesium and those of oxygen, whereas nitrogen and oxygen did not appear to be well correlated. For example, three of the stars studied (S283-2, S289-4 and RLWT-41) had atmospheres significantly enhanced in nitrogen compared with the other -processed elements (i.e. oxygen, magnesium, silicon), beyond that which could be explained by observational uncertainty. Additionally two stars in the cluster S289 (-4 and -2) appeared to have significantly different nitrogen abundances, whereas the abundances of oxygen, silicon and magnesium were found to be consistent. Hence this stellar sample suggested either a significantly different spatial variation in the progenitor interstellar nitrogen abundance (compared with the other metals), or the products of CNO-cycle nuclear processes are being mixed to some of the stellar surfaces.
As a method of distinquishing between these possibilities we have gathered further high quality spectroscopic data to determine accurate carbon abundances in this homogeneous sample of B-type stars. We shall present carbon abundances from the C II lines at 6578.1 and 6582.9 for comparison with the previously published nitrogen and oxygen abundances (Smartt et al. 1996a). These lines are among the strongest carbon absorption features in the optical spectra of early B-type stars, are known to be relatively unaffected by non-LTE effects (Eber & Butler 1988) and to give reliable differential abundances (Barnett & McKeith 1988).
© European Southern Observatory (ESO) 1998
Online publication: March 23, 1998