Up: Analysis of correlations between stars
Subsections
For statistical purposes we have compared the location of young stars with
the position of the stars which are definitely not young but also show some
polarization of radiation.
It was previously noted by Yudin ([1988]) that the classical Be stars
(40 objects) did not follow the relation derived for young stars but were
distributed in a separate group. To make statistically significant
conclusions we collect the data of polarimetry and V, L photometry for a sample
of 228 classical Be stars (for detail discussion see Yudin [2000]).
With few exceptions most of
the selected stars are dwarfs that allow us to compare their behaviour with
that of HAEBE stars. The relation between
and E(V-L) for
classical Be and HAEBE stars is shown in Fig. 17. It is safe to say that most
classical Be stars are largely concentrated in the region of small IR excesses
while their polarization varies from very low
values up to
%. The correlation between
and E(V-L) for Be
stars is weak or even absent and most of them are concentrated well away from
the above derived dependence for young HAEBE stars. It is well known that the
polarization of Be stars originates from scattering of free electrons in
a CS gaseous shell, as well as the excess in near IR, and even at 12
m is
mainly caused by free-free emission (see Waters & Marlborough [1992])
and is not due to the thermal emission from CS dust. Therefore, the nature of
IR excesses and polarization in classical Be stars and HAEBE stars is quite
different.
We are currently collecting data on optical polarization, near-IR
excesses and projected rotational velocities for a sample of
650
classical Be stars
and we are planning to publish a detailed study in the nearest future
(Yudin [2000]).
Another group of stars with IR excesses and polarization is red giants and
supergiants (hereafter RSG). The importance of the comparison of their
polarimetric behaviour with that observed for young TT stars is that the
objects in both classes are surrounded by dust shells, have approximately
the same spectral classes but are on a different stage of evolution.
RSG are post-MS objects and they are located on the
diagram near
with a considerable scatter in the
polarization values (up to 4%, see Fig. 3). It is well known that for most
RSG the dust envelopes are optically thin
(Dyck et al.
[1971]). There is a consensus now that the intrinsic polarization of RSG
originates due to scattering by nonradially oriented dust grains in the
stellar envelope. In this
case linear polarization appears even if the stellar envelope is spherical
and even for small optical depth (resulting in small near IR excesses). This
suggestion is strongly supported by the detection of high value circular
polarization in some RSG (up to
%). Comparison of the
location of TT stars and RSG on the diagram leads to some interesting
conclusions:
a) Most TT stars are located outside the box occupied by RSG which indicates
different physical conditions in their CS shells (for instance, significantly
larger optical depth in CS shells of TT stars);
b) A small fraction of TT stars (
20%) are located close to the
position derived for RSG and for some of them we may suppose the important
contribution of the scattering by nonspherical dust grains to their intrinsic
polarization (see also Lucas & Roche [1998]). This fact has been
corroborated by the detection of circular polarization in some TT stars
(Bastien et al. [1989]). Note however that the level of circular
polarization in TT stars is generally lower (
%) than that for RSG.
At present
circular polarization has been detected for only a few HAEBE stars and on the
level lower than that for TT stars. Thus a slight difference in position of
HAEBE and TT stars and respectively larger scattering of the data points for
TT stars on the diagram may also occur due to the presence of nonspherical
oriented dust grains in CS shells of some objects.
The next group is MS stars in solar neighborhood and 68 stars from Leroy's
([1993]) catalogue within 50 pc from the Sun with available L-band
photometry and
were selected. It is believed that for most of
these stars the interstellar component of polarization is negligible.
As follows from Fig. 3, all selected stars are largely concentrated in a small
region of the diagram around
and in the range of
polarization
,
i.e. in the same region as young solar-type
stars and some Vega-type stars. This is a side benefit on the small
polarization of MS stars.
Finally, we consider the Serkowski et al. ([1975]) catalogue. About 120
stars for which the data on L-band photometry were found in CIO5
(Gezari et al. [1999]) have been selected. Most of them have
MK-classification and are supergiants of early spectral type which allows
us to exclude the suggestion of their possible youth. As follows from
Fig. 2 most of these normal stars are clustered close to the value of
in the range of polarization 0.3%
8%.
This fact suggests that these stars have no near IR excesses (and no hot dust
in their CS environment) and all observed polarization has an interstellar
origin. If the component of interstellar polarization is removed from the
observed values, these MS stars will be located in the lower left
side of the diagram discussed here. In any case their position on the diagram
is very different from the dependence obtained for young stars which gives
additional support to the use of this diagram as a selection criteria for
young stars.
The conclusion of this section is that the regions of the diagram
occupied by stars which are not young are distinguishable from those occupied
by young stars of different spectral types (see Fig. 4). All the above might
be considered as a visual demonstration that the diagram can be used for
the selection of stars with hot CS dust (and most of them might be young
objects).
Up: Analysis of correlations between stars
Copyright The European Southern Observatory (ESO)