Astron. Astrophys. 319, 607-616 (1997)

3. Results

In Table 1 we list the observed sources and the corresponding CO and SiO line parameters. In subtable a we gather the Mira and optical semiregular variables and in subtable b we present the OH-IR stars and other infrared objects. For SiO, the main beam brightness temperature and velocity correspond to the strongest feature. For CO, the main beam brightness temperature corresponds to the intensity averaged over a few channels around the line center; the corresponding velocity is the line profile centroid. For each line, the full velocity coverage above 2 has been determined and the corresponding values are given in Table 1. The SiO and CO lines have been observed with different backends at different spectral resolutions. Consequently, it was easy to check that the observed SiO line wings were real and not an artifact in one of the spectrometers. In addition, when the SiO line wings were particularly prominent the observation was repeated after the sideband noise level of the phase-lock loop-gain was checked and ajusted to the minimum aceptable value (-40 dB). Figs. 2 and 3 show the line profiles, smoothed to 0.4 km s^-1 for both molecules, in selected objects showing weak red and/or blue SiO emission. Fig. 4 shows the observed CO and SiO line profiles in the remaining objects. Fig. 5 shows a comparison of the SiO spectra for the stars observed in both periods.

Fig. 2. CO J =2 1 and SiO v =1 J =2 1 line profiles toward selected objects from Table 1 with R 0.95 (see Table 1). The intensity scale is in main beam brightness temperature and the velocity is relative to the LSR. For each object the CO and SiO full profiles are given in the top and bottom panels, respectively. The center panel shows an enlargement of the CO (thin line) and SiO (thick line) profiles. For µ Cep the CO blue profile is contaminated by galactic molecular cloud emission. The SiO profiles in the center panels are clipped when intersecting the CO ones. Zero level emission for both CO and SiO lines in the center panel are marked with horizontal dashed lines. Original spectra in this panel have been smoothed by adding two or three adjacent channels to improve the signal to noise ratio in the line wings.

Fig. 3. Same as in Fig. 2, but for stars with R 0.95 (see Table 1).

Fig. 4. CO (histogram plot) and SiO v =1 (thick continuous line) spectra for stars of Table 1 with R 0.7. The left and right Y-scales correspond to CO and SiO intensities respectively. For both lines the intensity scale is in main beam brightness temperature. Some stars are contaminated by galactic CO emission (TX Cam, VX Sgr, R Aql -mainly outside the line core-, OH104.9+2.4, and OH127.8-0.0); the corresponding channels have been omitted in the plots.

Fig. 5. Comparison of the SiO emission in 1994 (histogram plots) and 1995 (thick continuous line) for some objects observed in both periods. For each star the top panel shows the full main beam brightness temperature scale and the bottom panel corresponds to an enlargement of the same data. The bottom panel for each star also shows the CO spectrum (dashed line).

In Table 1, the SiO velocity peak agrees for most of the objects with the CO velocity centroid which is a good estimate of the systemic stellar velocity (see also Jewell et al. 1991); this suggests that the bulk of the maser amplification occurs in the immediate neighbourhood of the star, where the gas has not yet been fully accelerated, or that the velocity vectors are perpendicular to the paths of maximum amplification. In addition, all strong peaks with narrow SiO emission lie inside, and spread over, the full velocity range of the thermal CO emission. Presumably, the line wings of v =1 SiO emission are of maser nature, as no trace of high velocity gas emission has been observed in thermal v =0 SiO emission (see Fig. 4 in Cernicharo et al. 1994). Moreover, a full track in parallactic angle of R Leo with the 30 m IRAM telescope (unpublished data) has revealed that its SiO v =1 J =2 1 line wings are strongly linearly polarized.

Because most stars in our sample have high galactic latitudes, the blending of CO stellar lines with interstellar CO is not a major problem. However, in those stars with low galactic latitudes, contamination by interstellar cloud emission may be present. Such an emission is narrow compared to the broad stellar CO profile and, when present, may fall outside (e.g. S Per, see Fig. 2) or within the stellar profile (e.g. VX Sgr, TX Cam, OH104.9+2.4, OH127.8-0.0 in Fig. 4). The coincidence with the edge of the stellar CO profile perturbs the CO widths analysis in the case of µ Cep (see Figs. 1 and 2). This object shows galactic contamination in the blue part of the CO profile; from the shape of the profile we have adopted a CO velocity range between 0 and 50 km s^-1 (between 14 and 50 km s^-1 above 2 ); in any case, due to this problem this object will not be considered in the discussion.

In order to find out whether the SiO emission range is broader than that of the CO emission, we derive the SiO and CO line widths to 2 intensity and compute the ratio R = v(SiO)/ v(CO). This ratio is given in column 12 of Table 1, and varies between 0.18 for OH104.9+2.4, and for µ Cep (however, see comments above for this object). The average R value for Miras (Table 1a) is 0.7. 52% of these objects have R 0.7 and tend to show prominent red wings (see comments in Table 1 and Discussion), and 17% (four Miras) have R 1 (see Figs. 2 and 3). We will consider that an object shows high velocity wings when it displays red and/or blue wings which can occur independently of value of R. All the objects with R 1 have SiO wings exceeding the CO emission both in the blue and in the red with the exception of S CMi where only the SiO red wing surpasses the CO terminal velocity. The three supergiants in subtable a have R in the range 0.8 to 1.8 and the four semiregular variables have R between 0.6 and 1.3. The infrared objects in subtable b have an average R value of 0.29 when one excludes the irregular supergiant NML Cyg.

Stars with SiO velocities more negative than in CO are not easily interpreted due to trapping of CO blue photons in the outer layers of the circumstellar envelope (which leads to a self-absorption in the blue part of the CO profile, see e.g. Huggins & Healy 1986). Hence, the measured CO velocity extent toward the blue must be considered a lower limit. Therefore, for stars with blue SiO wings, we can only conclude that SiO emission reaches or approaches the terminal velocity of the envelope. In the red part of the CO line profile it is difficult to produce self-absorbed effects similar to those mentioned above, unless we invoke non-standard velocity fields. For R larger than 0.7 most of the stars show SiO line velocities exceeding the CO velocities in the red, and hence this SiO emission represents a gas component that is not detected in the thermal emission of CO and other molecular species.

A plot of the ratio R versus the mass loss rates deduced from CO observations (Loup et al. 1993) shows that large values of R tend to be found in stars with low mass loss rates. This is also true when one plots R versus the the IRAS 25 to 12 µm flux ratio which is also a measure of the dust mass loss rate (Loup et al. 1993; however, this trend may need to be corrected for a possible dependence of R on the optical light phase; see below). The bulk of our sample lies in regions II and IIIa of the two-colour plot by van der Veen & Habing (1988). Stars with R 1 do not show any particular behaviour in this plot and all lie near the Mira evolutionary track. It is worth noting that stars with R 0.7 are Mira or semiregular variables, whereas those with the lowest values of R are OH-IR stars. Finally, all stars in our sample showing blue and/or red wings (independently of the value of R) are characterized by moderate values for the terminal velocity of the envelope, 10-12 km s^-1, except S Per and µ Cep (both have 17.5 kms^-1 and both are known to be supergiants).

A previous comparison between SiO and CO line profiles has been done by Nyman & Olofsson (1986), but due to limited signal to noise ratio they compared the long term averaged profile of SiO with CO expansion velocities. They concluded that the SiO velocity coverage could be larger than the CO velocity extent for R Leo, U Ori and o -Cet. However, the CO expansion velocities in their Table 3 are clearly underestimated when compared with our CO data (see our Table 1). For instance, for R Leo and o -Cet they found a CO velocity coverage of 12 and 9.8 kms^-1 while in our data it is of 15.7 and 18.7 km s^-1 respectively (see Table 1). With our CO data the R values of Nyman & Olofsson for these stars are smaller than 1.

© European Southern Observatory (ESO) 1997

Online publication: July 3, 1998
helpdesk@link.springer.de