There are a number of interesting correlations between the photometric indices of Be stars observed here which may shed some light on the nature of the Be star phenomenon. Starting with Figs. 1 to 6, we see that within the clusters, as well as in the field, Be stars exist over a wide range of luminosities. Since the Be stars in the clusters are all of essentially the same age (field contamination of the cluster Be star populations should be <10% based on the small cluster area), the wide range in Be star luminosities means that the Be phenomenon can occur at any time throughout the main-sequence lifetime, and need not be confined to particular evolutionary phases such as the core contraction phase at the end of the main-sequence.
Another interesting correlation is seen in the (, V-I) plots (Figs. 1c to 6c). Here we see a significant trend for the strongest emitters, as measured by the colour, to be those with the reddest V-I. This effect was also noted by Grebel (1997). The fact that Be stars are redder in V-I than normal B stars is also obvious from Figs. 1a,b to 6a,b where, at a given V, the Be stars are clearly redder than the normal main-sequence stars. An increase in redness of Be stars with increased strength of H emission has also been noted in the infrared (e.g. Dachs et al. 1988).
It is widely accepted that the Be phenomenon is associated with rapid rotation of the stellar photosphere and the presence of a circumstellar disk of comparatively cool material which gives rise to the observed Hemission. The observed increase in V-I of Be stars with increased strength of H emission is almost certainly the result of one or more of four factors: a real reduction in stellar effective temperature due perhaps to rotation, a changed spectral energy distribution due to rotational distortion of the stellar atmosphere, continuum and line emission from the disk, or absorptive reddening caused by circumstellar matter. The first two factors result from changes in the structure of the star, while the latter two involve circumstellar matter only. The viewing angle will also effect the colours in the non-spherical geometry.
In this regard our photometric colours do not constitute a conclusive test of the mechanism responsible for the apparent reddening of Be stars. Figure 7 shows plots of B-V and V-R against V-I for Be stars and normal B stars within and around the cluster. The B-V values come from Balona & Jerzykiewicz (1993). The zero point of V-R is arbitrary (since the R magnitude has not been zero-pointed) but has been adjusted to be close to the value expected for the normal hot B stars.
In the (V-R, V-I) plot, the Be stars fall on a sequence that is essentially indistinguishable from the sequence followed by non-Hemitting stars. This overlap is coincidental and is the consequence of the spectral distribution of the continuum emission from the circumstellar envelope (Kaiser 1989) within the R and I bands. The position of the Be stars in the (B-V, V-I) in Fig. 7 is suggestive of a systematic shift to the right of the sequence of normal main-sequence stars, shown by the continuous line. In this case significant emission due to Paschen continuum emission from the circumstellar envelope is located in the I band alone. Hence the Be stars appear shifted to the right relative to the non-Be stars.
Also shown in Fig. 7 are the reddening lines from Taylor (1986). On the (B-V, V-I) plot, it is clear that the Be stars do not follow the reddening line, suggesting that the red colours of the Be stars are not due to circumstellar dust absorption.
|Figure 7: The (V-R, V-I) and (B-V, V-I) diagrams for stars within and around the cluster NGC 2004, and brighter than V=16.5. Non-emission B stars are shown as dots and Be stars are shown as solid squares. The B-V colours are taken from Balona & Jerzykiewicz (1993). Due to the relatively small field studied by Balona and Jerzykiewicz, there are fewer stars in the (B-V, V-I) diagram. The solid line in the (B-V, V-I) diagram is the theoretical main-sequence ( = 4.5) from Bessell et al. (1998). The arrows are interstellar reddening vectors from Taylor (1986), corresponding to E (B-V)=0.1|
The effect on B-V colour of rotational distortion of the atmosphere of a Be star has been examined by Collins et al. (1991). However, even in the case of a star rotating near breakup, the maximum predicted change in B-V colour (from the colour of a non-rotating B star) is only 0.05 magnitudes (Zorec & Briot 1997), compared to changes of up to 0.15 magnitudes seen in Fig. 7. It therefore seems that this effect is not a major contributor to the reddening of Be stars.
The final possibility is that the red colours of the Be stars compared to the colours of non-Be stars at the same luminosity (as seen in Figs. 1a to 6a) are due to a real change in the photospheric flux distribution through the BVRI bands caused by a change in stellar effective temperature. However, since the bulk of the flux of these stars is emitted shortward of B, it remains to be seen whether the Be stars really do have lower effective temperatures than B stars of similar luminosity. This question will be examined in forthcoming papers which will give spectral types of the stars, and HST ultraviolet magnitudes from which values can be accurately compared between the Be and normal B stars.
Copyright The European Southern Observatory (ESO)