Astron. Astrophys. 324, L33-L36 (1997)

4. Discussion and Conclusions

The observed distribution of stars fits well with what is expected for pre-main sequence stars: they are almost all located between (or at) the zero-age main sequence (ZAMS) and the birthline. There is a good correlation between the luminosity class (Table 1) and the position in the HRD: All giants are located to the right of the ZAMS. The clustering of stars around the evolutionary track of mass 2.5  in Fig. 1 is probably due to a selection effect: There are no Herbig Be stars in the nearest star forming regions. Since the pre-main sequence evolution of B-type stars proceeds much faster than those of their A-type counterparts this could reflect a difference in age between nearby and more distant star forming regions.

Peculiar is the position of BD+ 4124 in Fig. 1, far below the ZAMS. Such a position seems impossible for a young stellar object. BD+ 4124 illuminates a reflection nebula, in which many other young stellar objects can be found (Hillenbrand et al. 1995), so the possibility of this being an extremely rare evolved object seen projected towards a star forming region can be excluded. Therefore one or more of our assumptions used for deriving this position in the HRD must be wrong. Different authors in literature agree well on the spectral type of BD+ 4124, B2-3e. Also, a spectral type which would place it near the zero-age main sequence is not compatible with the observed spectral energy distribution. However, the distance as measured by Hipparcos (  pc) differs greatly from the literature value of the distance to the star forming region in which BD+ 4124 is located (1000 pc). Therefore we conclude that, although formally the Hipparcos data products give a 4 detection for the parallax of BD+ 4124, this must be wrong. We note that for BD+ 4124 and HD 53367 the percentage of measurements that was rejected to arrive at the solution for the trigonometric parallax given in the Hipparcos Catalogue is very high, (indicated by open circles in Fig. 1). Furthermore, BD+ 4124 was flagged as a variability induced mover in the Hipparcos Catalogue, most probably indicating problems with the astrometric solution. This case may reflect the difficulty of parallax determinations by Hipparcos in nebulous regions. For HD 53367 the discrepancy between the literature and Hipparcos distance that can be seen in Table 1 is statistically hardly significant.

A fair number of stars in our sample cluster around the 2.5  track (Fig. 1). This allows us to construct an evolutionary sequence for their circumstellar environment. We select the objects HD 104237, AB Aur and HD 100546 for which reliable ages were estimated (Table 1), and plot their spectral energy distributions (SEDs) in Fig. 2. Included in Fig. 2 is the SED of  Pic which is even older than HD 100546 (Crifo et al. 1997), but has a slightly lower mass. The younger stars in our sample have SEDs that tend to be more or less flat in the infrared, whereas for the more evolved ones the dust excess tends to have a double-peaked structure, with a broad dip in the energy distribution around 10 µm. Extrapolating such a behaviour, we would expect to end up with a higher-mass equivalent of a star like  Pic. Interestingly, Waelkens et al. (1992) showed that the SEDs of Herbig stars can be classified as "flat" or "double-peaked" in the infrared and proposed an evolutionary scenario in which a broad dip around 10 µm develops with time. The ages derived from the Hipparcos data are consistent with such a scenario.

Fig. 2. Observed (squares) and extinction-corrected (circles) SEDs for (a) HD 104237, (b) AB Aur, (c) HD 100546, (d) Pic. Also shown are Kurucz (1991) models fitted to the extinction-corrected SEDs.

Spectroscopic evidence for evolution of the circumstellar dust component comes from observations with the Infrared Space Observatory ISO (Waelkens et al. 1996): HD 100546 shows a strong crystalline silicate component, similar to those seen in the ground-based 10 µm spectra of  Pic (Knacke et al. 1993) and of comets (Hanner et al. 1994). A recent ISO spectrum of the comet Hale-Bopp also shows strong crystalline silicates (Crovisier et al. 1997). In contrast, the object HD 104237 shows little evidence for such a crystalline dust component (Malfait, private communication). The occurrence of crystalline dust is an indication for the degree of "processing" which the circumstellar material underwent after accretion from the molecular cloud, and is thus a measure of age. We find that for the small sample discussed here, the age derived from Hipparcos is consistent with the degree of processing seen in the ISO spectra.

We stress that the timescale for the evolution of the central star and the circumstellar material are probably not well coupled: studies of very young open clusters show that only a small percentage of the cluster stars, many of which are undoubtedly pre-main sequence in nature, show circumstellar gas and dust (e.g. van den Ancker et al. 1997a). Furthermore, it is known that some giants, with presumably lower stellar ages, do show a double-peaked infrared excess, making the construction of evolutionary sequences difficult. Provided that the initial configuration and spatial distribution of the circumstellar dust envelopes of the Herbig Ae/Be stars are similar, some unknown physical mechanism regulates the timescale at which the circumstellar matter is dispersed. So, while the timescales may be very different for individual stars, it is quite possible that the sequence is as proposed in Fig. 2.

It is tempting to speculate about the origins of the formation of a 10 µm dip. From the presence of 10 µm amorphous silicate emission in most Herbig Ae/Be stars, we know that the dust envelopes surrounding these stars must be optically thin, and that the presence of a broad dip around 10 µm must correspond to a physical gap in the radial distribution of dust around HAeBes. Such a gap can only be caused by perturbations by another body surrounding the central star. This could either be a low-mass star outside the dust shell, clearing the region between the inner and outer Lindblad resonances through long-range resonance effects (Lin & Papaloizou 1979), or a Jupiter-sized planet forming inside the dust envelope, clearing Roche lobe radii on either side of the planet (Artymowicz 1987). Since many of the HAeBes with faint companions detected in recent years by speckle interferometry as well as by Hipparcos (Paper II) do not show a dip in their energy distributions, we consider the latter explanation more likely.

© European Southern Observatory (ESO) 1997

Online publication: May 26, 1998

helpdesk.link@springer.de