4. Observed and theoretical IRAS LRS classes
In Kerschbaum et al. (1996 ) the IRAS LRS classes of the carbon-rich AGB variables have also been investigated with the following result: the ratio of 4n/1n spectra which should be an indicator for the importance of mass loss is for Lbs 1.46, for SRas 1.33, for SRbs 2.44 and for Miras 3.22. Moreover, in the cases of 4n objects the mean classes are 42.7, 43.5, 43.4 and 44.4, respectively. At least for the Miras the trend seems clear that they contain a much larger fraction of higher mass loss objects than the other 3 groups of variables.
For comparison we have plotted the LRS classes of known, carbon-rich variables together with our models containing amorphous carbon particles as well as -SiC grains (in the ratio of 4:1). Fig. 6 depicts the IRAS LRS 4n class according to the definitions of the Joint IRAS Science Working Group (1988) as a function of the colour where the LRS class characterizes the strength of the SiC-feature around m. The solid line at the bottom of the plot denotes a linear fit through all models without SiC-dust. Hence, we expect all observations above this line. For clarity we plot only the time-averaged values of the dynamical models. At most we find a change of one LRS class over time. Our dynamical models form a nearly linear sequence through the bulk of the observations of carbon-rich Miras, SRa-, SRb- and Lb- variables plotted as small crosses. The agreement between observations and models would be even better if the slight colour shift noted in the two-colour diagram is taken into account. The small number of observed stars at small LRS classes and red colours is probably due to an observational bias against the detection of variability for stars with high mass loss rates (Guglielmo et al. 1993 ).
All stationary dust-driven outflows (models VS, A23, B19 and L12) are concentrated around an LRS class of 43 with a value of about 0.1. These symbols together with model R7C18 correspond nicely to the left boundary seen in the observations. Note that these bluest models exhibit also the smallest mass loss rates of our theoretical sample (cf. Table 1). The plot thus indicates that dynamical models have a higher LRS class at a given mass loss rate than the stationary ones.
If the stationary models are associated with stars with small pulsation amplitudes and/or irregular variability, the observed blue objects with weak LRS features should thus all be such stars. Indeed this is seen in the pulsation characteristics of the observed stars. However, large amplitude variables can also show blue IRAS colours and a weak SiC feature, indicating that the pulsation amplitude and regularity alone are not sufficient to distinguish between time-dependent and stationary cases.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998