Astron. Astrophys. 323, 513-523 (1997)

4. Discussion

The main results of our investigation can be summarized as follows. We confirm the discovery by Zuckerman et al. (1995) that giants with infrared excesses exist (with an estimated frequency of 8%), but that they are less common than Vegatype stars ( 20%), a result anticipated by Jura (1990); the excesses appear at spectral type G5 and from there on involve a rather constant fraction of stars; the circumstellar material is very cold, i.e. only few 25 µm excesses are observed; except for a few cases, the excess objects appear point-like at the IRAS wavelengths.

Before discussing the possibility that the excesses have the same origin as those of the Vega-like objects, it is appropriate to investigate other hypotheses, such as the presence of circumstellar matter due to mass ejection by the giant. Mass loss is known to occur from AGB stars, and leads to important infrared excesses once luminosities of the order of (i.e. the absolute visual magnitude for Mira) are attained. For a relatively large fraction of the giants in this study, the parallax is known. In Fig. 5a we display these objects in the observational HR diagram; the data are taken from the Hipparcos Input Catalogue Version 2.0 (Turon et al. 1993); since it concerns nearby stars, the reddening can be neglected.

Fig. 5. Observational HR-diagram of a G, K and M giants with infrared excess with known parallax (left); b all G, K and M giants with known parallax (right). Legend: : G giants; : K giants; : M giants.

It can be seen on the figure that all stars with excesses for which a parallax is known are less luminous than about . In fact, it turns out that these objects cover the full domain in the HR diagram that is defined by all giants in the Hipparcos Input Catalog (see Fig. 5b). We conclude, therefore, that the giants with infrared excesses essentially all are first-ascent giants, and thus that the sample is not significantly affected by the presence of asymptotic giant branch stars.

Whether first-ascent red giants are also subject to mass loss, can be inferred from the fact that RR Lyrae stars consistently have lower masses than the highest initial mass for a star that cannot yet have left the main sequence in a Hubble time. How and when red-giant mass loss occurs is not well known. For AGB stars, the current idea is that mass loss is a discontinuous, recurring phenomenon, probably connected with the thermal pulse phase (Zijlstra et al. 1992). However, we consider it unlikely that the excesses we observe are due to mass loss. First, it is most likely that most of the mass is lost at the Helium flash; if that were the circumstellar mass we observe, one would expect that the distribution of our stars in the HR diagram is more clumpy than on Fig. 5a. Second, one of the striking characteristics of our stars, is that their IR excesses occur essentially at larger wavelengths, indicating detached shells, and so are difficult to reconcile with recent mass loss.

Since in addition there is no evidence that a larger than average binarity occurs in the giants with excesses, it then seems most natural to conclude that the excesses are of a similar nature than those observed in Vega-like stars. By all means, the fraction of excess stars among giants is distinctly smaller than among main-sequence stars. It is tempting to attribute this, as Jura (1990) did, to the evaporation of the circumstellar debris due to the higher radiation field of the giant. On the other hand, it has to be said that little is known about the evolution of the fraction of Vega-like stars on the main sequence. The smaller fraction of giants with excesses may also be partly due to the gradual destruction of the circumstellar bodies and the subsequent decrease of the likelihood of collisions.

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998

helpdesk.link@springer.de