3. SiO and stellar properties
A very important attribute of most AGB stars is their pulsation. In Fig. 2 we show as a function of the photometric periods listed in Table 1, which are taken from the GCVS (General Catalogue of Variable Stars). The variability types are separated using different symbols. As one can see, the Miras are located at periods longer than 200 days, whereas all SRb stars have shorter cycles. The SRa variables represent an intermediate class between those two groups (Kerschbaum & Hron 1992). It is obvious that objects with weak or no SiO absorption (Å) are only found among the Miras. Since there are also some of them showing quite intense bandheads (Å), we get a large scatter of the values for the longer periods. This may be due to the spectral variability discovered by Rinsland & Wing (1982), who noticed that the SiO bands become very weak or disappear in Mira stars about the time of the visual light maximum. On the other hand, all SR objects have a large SiO absorption, which also indicates that the weak bands of the Miras are caused somehow by their strong pulsations. It should be noted that the most intense bandheads (Å) appear only among the SR stars of our sample. Because of the expected variability and the small number of objects with longer periods it is not sure if this represents a general trend.
For those Miras in Table 1 where periods from the GCVS as well as information on the light maxima from the AAVSO (Mattei 1996) are available we have also listed the visual phase at the time of our observations. We did not give a value for R Cen, since according to the GCVS it has two periods. Although the sample is very small, it again confirms the scenario that the SiO bands become weak or disappear around the time of the light maximum (see also Fig. 7).
Another interesting property of AGB stars is their photometric colors in the near and mid infrared. We have taken the 12, 25 and m fluxes from the IRAS-PSC (Joint IRAS Science Working Group 1988b) and combined them with near infrared measurements from Fouqué et al. (1992), Kerschbaum & Hron (1994, for EP Vel) and Catchpole et al. (1979, for R Cen). Some of the resulting colors are listed in Table 1. We did not attempt to correct the values for the effects of interstellar or circumstellar reddening. In addition, one should always bear in mind that the SiO spectra and the photometric fluxes, which are expected to be variable, have not been taken at the same time. This will introduce some scatter in all of the diagrams.
In Fig. 3 we present as a function of the (12-25) color from the IRAS photometry, which mainly probes the properties of the circumstellar shell. Like in Fig. 2 the variability types are shown with different symbols. For the IRAS magnitudes we have used the definition given in the Explanatory Supplement (Joint IRAS Science Working Group 1988a). No color correction has been applied. As one can clearly see, generally seems to be increasing for higher values of (12-25). This is caused by the fact that the Miras in our sample, and especially those with weak or no SiO absorption, have bluer colors than the SR variables. If one neglects the objects with Å, the trend will disappear. However, it is interesting that just these Miras show such blue colors. The only source which does not follow this general behaviour is the very red Irregular variable V2234 Sgr situated at a (12-25) value of 1.20. But it also deviates significantly from the rest of the objects for all other infrared colors and is probably not an AGB star.
We have also studied the SiO band intensities as a function of the (K-12) color, the latter depending very much on the thickness of the circumstellar envelope. Although the Miras are statistically redder than the SRb variables, we did not find any trend of the values in the corresponding diagram. This agrees with the fact that there is also no correlation with the IRAS LRS class probing the optical depth of the dust shell. Finally, we have studied the SiO absorption as a function of the (J-K) color, which depends mainly on the stellar temperature, and we got the same behaviour as for (K-12): the Miras are statistically redder, but there is no correlation with . It should be mentioned that, like in (12-25), V2234 Sgr is also extremely red in (K-12) and (J-K).
The photometric data can also be used to determine temperatures for the stars. One of the simplest methods is to fit two blackbodies to the spectral energy distribution in the near and mid infrared. They are attributed to the stellar photosphere and the circumstellar gas and dust shell, respectively (Kerschbaum & Hron 1996). Of course, neither the star nor its envelope radiate like a blackbody and this technique does not provide a direct measurement of the effective temperature. Nevertheless, the results are interesting, since they contain information on the overall spectral energy distribution. We have obtained such blackbody temperatures using the already mentioned IRAS and near infrared photometry as well as visual magnitudes published in the GCVS. But due to their large variability the latter have only been given a very low statistical weight. The temperatures for the first blackbody attributed to the stellar photosphere () are listed in Table 1. Their typical uncertainty being mainly due to stellar variability is about . A plot of the values versus is shown in Fig. 4a. There one can see that the Miras in our sample generally have lower blackbody temperatures than the SRb stars. As a consequence objects with weak or no SiO absorption appear only at and for the cooler sources there seems to be a large scatter in the diagram, whereas all hotter objects have more intense bands. It should be mentioned that in stars with the first blackbody may be strongly affected by the circumstellar shell.
In order to determine effective temperatures from photometric data we have used a semi-empirical relation by Ng et al. (in preparation), which connects them with the (J-K) color. This method is based on angular diameter measurements by Ridgway et al. (1980), Dyck et al. (1996) and van Belle et al. (1996). The results are shown in Table 1 and Fig. 4b. For some of the objects we could not determine a -value, since they are too red for the relation. It is obvious, that there is no visible trend in the diagram. In this context it should be noted that, at least for the cooler stars in our sample, we expect large uncertainties concerning the effective temperatures. First, in extremely extended objects dominated by pulsation and mass loss the definition of such a quantity becomes problematic. This affects the interpretation of any kind of observations, which requires a detailed knowledge about the stellar atmospheres being not available at the moment (see next section). A good example is the strong wavelength dependence of the measured angular diameters of cool Miras. In addition, in dynamical models the effective temperature loses its unique role for characterizing the atmospheric structure, and thus, the observed spectrum: for a given stellar mass and chemical composition, the physics of a hydrostatic environment may be described by the gravitational acceleration and the effective temperature. But this will not work for dynamical atmospheres, since their structure is dominated by pulsation and mass loss. Another problem concerning the -values listed in Table 1 is the fact that especially the cooler stars in our sample may have thick envelopes producing a considerable amount of circumstellar reddening. This could be the reason some objects are too red for the relation by Ng et al. (in preparation).
© European Southern Observatory (ESO) 1999
Online publication: February 23, 1999