4.5. Nebulae around evolved massive stars
Evolved massive stars are associated with nebulae which result from the interaction of stellar winds and stellar ejecta with the ambient interstellar medium. By studying the chemical composition of these nebulae, together with their morphology, kinematics and total gas content, one can get insight into the previous evolutionary stages of the stars and unveil some of the nucleosynthesis and mixing processes occuring in their interiors.
Schematically, during main sequence evolution, the fast wind creates a cavity in the interstellar medium and sweeps out a shell of compressed gas. After departure from the main sequence, the nature of the mass loss changes and the star loses chemically enriched material. When the star reaches the Wolf-Rayet phase, its outer layers are almost hydrogen free. This material is lost at high velocity and catches up with material lost in previous stages (see Chu 1991 or Marston 1999 for a review).
Imaging surveys of the environments of WR stars have found that in 50% of cases a ring like nebula is seen (Marson 1999). Ring nebulae have been classified by Chu (1981) into R type - radiatively excited H II regions and subsonic expansion velocities, E type - nebulae formed out of stellar ejecta (chaotic internal motion, large velocities) and W type - wind-blown bubbles showing thin sheets or filaments. Atlases are published by Chu, Treffers & Kwitter (1983), Miller & Chu (1993) and Marston (1997). Known examples of R types are RCW 78 (amorphous, containing a WN 8 star) and RCW 118 (shell, surrounding a WN 6 star). Known cases of nebulae containing ejecta are M 1-67 (WN 8 star), RCW 58 (WN 8 star). Known W types are NGC 6888 (WN 6 star), S 308 (WN5 star), RCW 104 (WN4 star), although Esteban et al. (1992) consider NGC 6888 as an Bubble/Ejecta type in their classification.
Luminous Blue Variable stars are regarded as precursors of WR stars with the most massive progenitors. They are usually found to be associated with small ejecta type nebulae like Car, AG Car (Nota et al. 1995).
The first spatially resolved and comprehensive study of abundances in Wolf-Rayet ring nebulae is that of Esteban and coworkers (Esteban et al. 1990, 1991, 1992, 1993, Esteban & Vílchez 1992), in which 11 objects have been analyzed with similar procedures. In a plot relating the N/O and O/H differential abundances (i.e. abundances with respect to interstellar medium ones) Esteban et al. (1992) find that most objects lie close to the (O/H + N/H ) = (O/H + N/H)Orion line, indicating that oxygen has been converted into nitrogen. This is indeed what is predicted by the Maeder (1990) stellar evolution models at the beginning of the WN phase. Dividing their objects into 3 categories from their chemical composition (H II for objects with abundances close to those of the environing ISM, DN for diluted nebulae in which stellar ejecta are mixed with ambient gas and SE for pure stellar ejecta), Esteban et al. (1992) show that there is a rather good correspondence between the chemical classes and the morpho-kinematical classes. They also note that the masses of SE nebulae are small and compatible with the hypothesis of pure stellar ejecta, while the H II nebulae have larger dynamic ages, consistent with the idea of being composed of large quantities of swept up gas. Esteban et al. (1992) find that the SE nebulae surrounding WR stars are associated with stars showing variability and thus probably having unstable atmospheres. This is also true for the nebulae associated with LBVs. In plots relating the N/O mass fraction to the He mass fraction, Esteban et al. (1992) find that SE nebulae lie close to the stellar evolution tracks of Maeder (1990) for initial masses 25 - 40 M, which become WN stars after a red supergiant (RSG) phase. This is consistent with the initial masses estimated from the star luminosities (Esteban et al. 1993).
Since this pioneering study, detailed computations have been performed to simulate the evolution of the circumstellar gas around massive stars (García-Segura et al. 1996a and b), coupling hydrodynamics with stellar evolution. The fate of the circumstellar gas results from interactions between the fast wind from the star while on the main sequence, the slow wind from the red supergiant or luminous blue variable stage and the fast wind from the WR stage. The resulting masses, morphologies and chemical composition of the circumstellar envelopes strongly depend on the initial stellar masses, both because of different nucleosynthesis and different time dependence of the winds. Stars with initial masses around 35 M are predicted to go through a RSG stage, and produce massive nebular envelopes (~ 10 M) with composition only slightly enriched in He and CNO processed material. Stars with initial masses around 60 M are predicted to go through a LBV stage, and produce less massive nebular envelopes (~ 4 M) with helium representing about 70% of the total mass fraction, and CNO equilibrium abundances (C depleted by a factor 23, N enriched by a factor 13, and O depleted by a factor 18). The composition and morphology of NGC 6888 and Sh 308 well agree with the theoretical prediction of a RSG progenitor. On the other hand, Smith (1996) notes that recent abundance determinations in nebulae associated with LBV stars do not agree with the composition predicted by the García-Segura et al. (1996) model of evolution of a 60 M star through the LBV stage. The abundance paterns of these nebulae are rather similar to those of SE nebulae surrounding RSG stars, with mild enrichments in He and N and mild depletion in O, suggesting that the star went through a RSG phase. It must be noted that abundance determinations in such objects are often difficult, because few diagnostic lines are available, so that ratios like N/H or O/H may be rather uncertain, but N/O is more reliable. In a rediscussion of nebulae around LBV stars, Lamers et al. (2001) conclude that the stars have not gone through a RSG phase. The chemical enhancements are due to rotation-induced mixing, and the ejection is possibly triggered by near-critical rotation.