In order to study the first overtone rotation-vibration bandheads of SiO we observed the spectral range from 3.94 to m for a sample of 23 oxygen-rich late-type giants, most of which are Mira or Semiregular (SR) variables. Also two Irregular variables have been included. The sources have been selected due to their variability types, their bright L-magnitude (), the existence of reliable JHKLM as well as IRAS 12 and m photometry and the existence of a good IRAS-LRS spectrum. As it is shown in Table 1 the stars cover a large variety of photometric periods and infrared colors. Due to their pulsational properties at least most of them are expected to be situated on the AGB (e.g. Alvarez et al. 1997).
Table 1. -values, pulsational and photometric properties of the observed stars. and are the temperatures from a fit of two blackbodies, while corresponds to the effective temperature derived from the (J-K) color (see text).
The spectra have been observed in June 1993 with the cooled grating spectrograph IRSPEC at the ESO NTT and reduced with MIDAS. They are composed of 5 overlapping parts corresponding to different grating positions. The final resolution is approximately 4000. To eliminate atmospheric features and effects of detector response, spectra of nearby O and B stars were taken at the same time and included into the reduction. The hot objects were mainly used for the determination of the stellar continua, since, apart from Brackett-, they do not show any distinct features in the studied wavelength range. However, concerning the overall slopes of the measured energy distributions we did not always get consistent results. Thus, we decided to use normalized spectra divided by a polynomial fit to the wavelength ranges m, m m, m m, m m, and m, where there are no strong SiO or OH features. In this context it should be mentioned that in stars with strong SiO bands the absorption will extend into all the defined continuum regions at m, since there is no frequency not affected by SiO at our resolution. Nevertheless, this effect can easily be corrected, if one compares the data to results from theoretical calculations, as it is discussed in Sect. 4.1. Another problem may be caused by the presence of a large number of stellar water lines in the whole spectral range between 3.94 and m (Aringer et al. 1997b). But water absorption will only appear in the coolest objects and never become very strong. In addition, we do not expect it to change the overall slope of the continuum.
Since the stars in our sample are extremely bright in the infrared, the signal/noise ratio is always very high. In addition, almost all of the objects have been observed more than once (usually 3 to 4 times). The different spectra, which have been averaged in order to get the final data, are generally identical concerning their features. But - even after a careful reduction - there were some variable spikes produced by the detector. Because of their fixed positions the stronger ones could be easily identified and removed using the O and B stars. Nevertheless, the weak ones and those, which did not appear in the spectra of the hot objects, remain in the final data. A good example is the relatively strong spike at m that can be seen in some of our stars (see Fig. 1).
The normalized spectra of all objects are shown in Fig. 1. In most of them one can see the first three bandheads of the main isotope 28Si16O situated at m (V = 2 0), m (V = 3 1) and m (V = 4 2) as the dominant feature. But there are also stars like X Hya, R Oph, RR Sgr and S Vir, all of which are Mira variables, having only a very weak or no SiO absorption. Some of them show Brackett- emission at m, which is a sign for the appearance of atmospheric shock waves. The distinct depression between 3.96 and m is caused by OH (e.g. Decin et al. 1997). It can only be seen in objects with intense SiO bands. Finally, as it has already been mentioned, there are a lot of H2O lines in the whole observed spectral range, which become important for the coolest stars with . However, they will not give rise to any remarkable intense features (Aringer et al. 1997b). Instead, they are expected to reduce the continuum flux level and introduce many small absorption dips into the spectrum.
Isotopic bandheads of SiO are situated at m (V = 2 0, 29SiO), m (V = 2 0, 30SiO), m (V = 3 1, 29SiO), m (V = 3 1, 30SiO) and m (V = 4 2, 29SiO). According to our synthetic spectra published in Paper I, the most distinct of them are expected at 4.0687 and m. Especially at the first of these positions, there is a quite strong feature that can be seen in all sources with large SiO absorption. But also the second band, which appears to be weaker, seems to be present at least in some of the stars. However, it should be noted that around 4.07 as well as around m an accumulation of some intense OH lines can be found (e.g. Decin et al. 1997). Thus, we assume these features to be formed by 29SiO and OH together. The remaining isotopic bands are only very weak or not visible. This is especially true for 30SiO, because its bandheads fall into an area with intense absorption from the main isotope. Maybe they produce small peaks seen in the spectra of some objects like SU Sgr. But that could also be due to another species like water in the earth's or stellar atmosphere. A possible exception concerning the appearance of isotopic bands is the strange spectrum of RR Aql showing intense features at all of their positions.
In order to evaluate the intensity of the SiO absorption we have measured , which is the sum of the equivalent widths corresponding to the V = 2 0, V = 3 1 and V = 4 2 bandheads of the main isotope. This has been done exactly in the same way as in Paper I for the synthetic spectra. The results are listed in Table 1. The typical uncertainty of these values, which is mainly due to the definition of the local continuum around the bandheads, amounts to Å.
© European Southern Observatory (ESO) 1999
Online publication: February 23, 1999