Astron. Astrophys. 323, 442-448 (1997)

Astron. Astrophys. 323, 442-448 (1997)

4. Discussion

The presence of the strong molecular bands of SiO and H₂ O show that the object is oxygen-rich. This is consistent with the emission feature at 9µm being due to silicates. The spectral observations around 2µm show two deep CO absorption bands, at 2.925µm and 2.323µm (the ¹² CO (2,0) and (3,1) band heads). Using Kleinmann and Hall (1986), the star is classified as a late M giant or supergiant, owing to the deep CO bands. The strong water bands imply that the star is a red giant, probably a Mira variable, and the great depth of the 1.9µm band (discussed earlier) argues that the object is of a later spectral type than any previously observed Mira variable. Lewis et al. (1990) found that only the brightest stars were detected in the OH molecule, so that although the system contains water the absence of detectable OH is not surprising in view of the distance of the star (1.9kpc according to Loup et al., 1993; although this distance is uncertain because they used a standard luminosity for their sample).

The IRAS colours from the 12/25/60 µm data show the star to be a late O-rich star with a moderate dust shell. The colours ([12] - [25] = 1.30, [25] - [60] = -0.06), using Walker et al. (1989) place the system on the extreme red edge of the region of O-rich stars with silicate emission and the region of O-rich stars with silicate absorption. These stars are mostly Miras and other AGB stars. Using Van der Veen and Habing (1988), their IRAS colours ([12] - [25] = -0.26, [25] - [60] = -1.93) put the source in region IIIa (variable stars with more evolved O-rich circumstellar shells), close to the "evolutionary track". The IRAS colours are consistent with the conclusion from the spectra that the star is a late type Mira variable.

The source IRAS 18530+0817 has an IRAS likelihood of variability [FORMULA] 90%, indicating significant variation of 12 and 25µm fluxes with time. In Table 2 we see that the source remained relatively constant during a time interval of a few days (8 days) but showed a large flux decrease six months later, towards the end of the survey. Soon after there is possibly a further flux decrease, this one within a 26 hour time-interval. We observe a similar decrease in the IRAS LRS spectra shown in Fig. 3. The spectrum in Fig. 3a was taken near the beginning of the survey, the spectrum in Fig. 3b was taken about six months later and there is a large decrease in energy output as well as changes in the morphology of the spectrum. The bright emission feature at 9.5µm remains relatively unaltered whereas the emission component around 11µm has almost disappeared.

Ground-based spectra in the 10µm window are generally uncertain in the 9.4µm - 10.0µm region, due to the difficulties in correcting for the strong telluric ozone absorption band at around 9.7µm (note the larger error bars for the affected data points in Fig. 2d). Thus the unresolved absorption edge at 9.5µm, present in the CGS3 spectrum (Fig. 2b), should be questioned, especially since it was not seen subsequently by CGS3 or HIFOGS. However, the consistency of the shift in flux level longward and shortward of 9.5µm in this spectrum, which is composed of three sub-spectra shifted by 1/3 resolution element and obtained sequentially in time, is strong evidence that the feature is real. Moreover, the shape of the IRAS LRS spectrum is consistent with the CGS3 spectrum in Fig. 2b and with the smoothed HIFOGS spectrum in Fig. 4a.

It may be that IRAS 18530+0817 has a mild silicate self-absorption feature, but it does not appear very similar to other examples. The unusual shape could be caused by an unknown absorption feature. Sequences of CGS3 10µm spectra of Mira and IK Tau (a very cool Mira variable) were examined; they showed SiO absorption near 8µm and smooth silicate emission profiles, but no silicate self-absorption or sharp absorption features, such as the one at 9.5µm seen in IRAS 18530+0817. We have not been able to identify the absorber responsible for the feature. The edge could be the band head of an unidentified molecule that becomes prominent at a certain phase in the period of IRAS 18530+0817.

Another alternative is that the feature (and its variability) could be an unusual manifestation of circumstellar silicate absorption in this object. Stars, with conventional 10µm silicate features, sometimes show slight evidence of structure in the feature around 10 - 10.5 µm (e.g. Tielens, 1990; Barlow, 1993). Tielens (1990) suggested that some Miras have silicates other than the usual astronomical silicate (olivine) present. He proposed the presence of disordered silicates (also oxides) but these mainly influence the shape longward of 9.7µm. Bradley et al. (1992) have an interesting alternative, from their work on Interplanetary Dust Particles (IDPs). Some of their samples have a strong emission component around 9µm. It would appear that a mixture of olivine (normal astronomical silicate) with some pyroxene could explain the unusual shape of IRAS 18530+0817. Alternatively, the glass-rich IDPs, which produced silicate features similar to comets Halley and Bradfield, also have quite a conspicuous 9µm component to their broad silicate feature.

We conclude that although the central star of IRAS 18530+0817 may appear to be a normal very-late Mira star (with perhaps the coolest photosphere yet detected in such stars), the dust around it is somewhat unusual. The shape of the 9µm feature appears to vary, apparently cut off by an absorption edge at 9.5µm at some times, and is partly self-absorbed at other times. Features may arise from the presence of dust which is similar to interplanetary dust, in addition to the normal astronomical silicate dust shell.

Online publication: June 5, 1998

helpdesk.link@springer.de