Astron. Astrophys. 334, 1016-1027 (1998)

1. Introduction

Kuiper's detection of water vapour vibration-rotation absorption bands in 1963 proved the presence of H₂ O in the circumstellar regions around Mira type stars that are found nearest to the photosphere. The most intense bands were detected at 1.4, 1.9 and 2.5 µm, and showed strong intensity variations throughout the cycle of the star. Hinkle & Barnes (1979) were able to establish the column density of water vapour in R Leo, inferring H₂ O abundance similar to that of CO in those regions. This conclusion coincided with the results from the calculations for the molecular abundances in thermodynamic equilibrium made by Tsuji (1964, 1973).

By and large, theoretical models predict that, in the regions nearest to the photosphere, almost all the oxygen which has not combined with carbon to form CO contributes to the formation of water vapour. There are two processes in circumstellar envelopes (hereafter CEs) which may affect significantly the abundance of water vapour: the formation of ice mantles and -in the most external regions of the CE- the photodissociation of the molecules due to interstellar ultraviolet radiation. The latter process triggers the formation of OH, whose masers are usually observed in O-rich CEs, and it has been studied by Goldreich & Scoville (1976) and Huggins & Glassgold (1982), amongst others.

The presence of water vapour in layers of CEs located farther away from the central star than those associated to the H₂ O absorption bands was discovered, more than two decades ago, thanks to the observations of the maser line near 22 GHz. This line has been detected in more than 300 evolved stars (see Benson et al., 1990), most of which are Mira type stars. Several interferometric observations have shown that the bulk of the emission in Mira and SR stars originates from small condensations within an area whose radius is less than cm, or (e.g., Lane et al., 1987). In supergiant stars the radius of the emitting region is about 10 times greater. The spatial morphology of the emission provides an evidence of a lack of spherical symmetry in some stars (e.g., R Aql, see Lane et al., 1987). Although, in others such as W Hya, the emission appears to be much more symmetrical (Reid & Menten, 1990). The spectral emission usually covers a velocity range which is significantly more restricted than that covered by the CO rotational emission or the OH maser emission. The above observation suggests that the line forms in the inner regions where the gas has not reached the terminal velocity of the envelope.

The lack of spherical symmetry and the fact that the bulk of the emission originates from compact condensations indicate that maser propagation effects -not produced by the systematic velocity field of the envelope- can play an important role in the amplification of the emission at 22 GHz. The scarcity of information about the physical conditions and the geometrical parameters related to these condensations has been the principal obstacle for a detailed modeling of this emission; from maser observations alone, the water abundance could not be derived. Nevertheless, it is firmly established that the maser at 22 GHz is pumped by collisions between H₂ O and H₂ molecules (Cooke & Elitzur, 1985).

Apart from the lines at 22 and 183 GHz, the latter being the main subject of the present study, other H₂ O pure rotational transitions from CEs have been detected at (sub)millimeter wavelengths, using ground based telescopes: the at 321 GHz ( K; Menten & Melnick, 1991) and the and lines of the bending mode =1 in VY CMa (Menten & Melnick 1989). Other recently discovered H₂ O submillimeter lines in CEs are reported in Melnick et al. (1993) and Menten & Young (1995). Since the levels involved in these transitions have high energies, the lines are formed in regions near the photosphere of the star, in areas where probably the SiO masers are also formed. The emission of the above mentioned H₂ O lines also shows clear indicia of being maser type. The detection of these lines indicates a high water vapour abundance in regions close to the photosphere, but it does not provide any information about the water abundance in more external layers of the envelopes.

Obviously, the solution to this problem requires the observation of low-lying H₂ O transitions; however, the strong absorption of these lines by the atmospheric water vapour constitutes a serious obstacle. Nonetheless, the airborne observations are on the way to circumvent this problem in some cases. Thus, Kuiper et al. (1984) attempted to detect the p-H₂ O line at 183 GHz in stars by using the 0.91m diameter telescope aboard the Kuiper Airborne Observatory. The results proved negative due to the lack of sensitivity of the on board instrument. The situation changed radically when Cernicharo et al. (1990) detected the above mentioned line in CEs using the 30m-IRAM telescope located on Pico Veleta. Since the levels involved in this transition lie at a lower energy than the levels of any of the transitions previously cited, one might expect that the line originates in the external regions of the CEs, at least in the case of stars with high mass loss rates. The above conclusion is suggested by the line profile observed toward RX Boo (Cernicharo et al., 1990). The spectral emission at 183 GHz covers almost the full CO velocity range, and shows a broad spectral component or "emission plateau" and narrow spectral components as well. The most intense of these narrow components is blue-shifted in frequency relative to the stellar velocity, and lies at a velocity close to the terminal velocity of the envelope. These properties resemble the emission of the OH masers, and suggest that the main direction of the maser amplification at 183 GHz in RX Boo is radial.

Recently, the launch of the Infrared Space Observatory (ISO) has provided for the possibility of observing numerous rotational and ro-vibrational water vapour lines at infrared wavelengths, among them most of the low-lying rotational water transitions. Barlow et al. (1996) have shown that, in the O-rich evolved star W Hya, the spectral emission in the Long Wavelength Spectrometer (LWS) is dominated by tens of water lines. Depending on the energy of the levels involved in the transitions, the various water lines trace different regions of the envelope, thus allowing in principle to infer the water abundance profile. From LVG models, Barlow et al. (1996) were able to explain most of the water line intensities with an H₂ O abundance of in the inner parts of the envelope, and with some decrease of abundance in more external regions. However, the inference of the water abundance from observations of very optically thick lines is difficult because the line fluxes are not very sensitive to the water abundance. Despite the advantage of observing a great number of lines, this problem introduces a factor of great uncertainty in the estimation of the H₂ O abundance.

Therefore, one might hope that the emission at 183 GHz from CEs, which arises from regions of the envelope more external than those where the other water maser lines originate, may allow for a valuable estimate of the water abundance in these sources. The gas movement in the most external regions of a CE is more orderly and stationary than in the internal regions. Furthermore, the density and the velocity profiles are also better described than in the inner regions. Consequently, and despite the maser nature of the emission at 183 GHz, it is possible to estimate the abundance of water vapour in a large sample of objects. The confirmation of this postulate has been the main purpose of the present study and that of a forthcoming paper (González-Alfonso & Cernicharo, 1998).

In continuation we present observations at 183 GHz in a total of 23 CEs, and compare them with the emission of other two water lines, at 22 GHz and 325 GHz, as well as with the emission of the CO and SiO v =1 lines. Some objects could be observed at 183 GHz in different observation periods, allowing us to evaluate the time variations of this line. We find correlations between some spectral characteristics of the 183 GHz line -the line width, the velocity of the peak emission relative to the stellar velocity and the power emitted in the line- and the mass loss rate of the star. In a forthcoming paper (González-Alfonso & Cernicharo, 1998) we will model the emission at 183 GHz from CEs by means of radiative transfer models, explain the observational correlations discussed here and provide an estimate of the water abundance in CEs.

© European Southern Observatory (ESO) 1998

Online publication: June 2, 1998

helpdesk.link@springer.de