1998A&A...329.1059O

We have observed a sample of 61 AGB-stars (39 M-stars and 22 C-stars) in circumstellar CO, CS, HCN, SiO, SiS, and SO radio line emission. The main results presented are based on the use of line intensity ratios, a well defined observational quantity that can be used to infer important conclusions as well as to provide constraints on models. Taken together the data are fully consistent with the facts that for this sample the circumstellar envelopes have the same basic chemistry (i.e., C/O<1 or >1) as the central stars, and that the mass loss rates have not changed drastically over periods between 10²-10³ years. The HCN(J=1->0)/SiO(J=2->1) intensity ratio discriminates unambiguously between ``normal'' circumstellar envelopes with C/O<1 (O-CSEs) and >1 (C-CSEs), while the CS(J=2->1), HCN(J=1->0), SiO(J=2->1), and SiS(J=5->4) intensity ratios with respect to CO(J= 10-) are not perfect for this purpose, and neither is the SiS(J=5->4)/SiO(J=2->1) intensity ratio. The data further shows that SO and the C-bearing molecule HCN are ubiquitously present in O-CSEs, and that their line intensities in O-CSEs are qualitatively consistent with the fact that the molecules are formed in a photo-induced circumstellar chemistry in a quantity that depends on the mass loss rate. Hence, both species can in principle be used to estimate the mass loss rate, and the tight relation between the SO(J_K=3₂->2₁) and CO(J=1->0) intensities in O-CSEs shows that SO line emission may even be a good mass loss rate estimator. On the contrary, the SiO(J=2->1) luminosity appears to be essentially independent of the mass loss rate in O-CSEs, possibly due to a larger influence from molecular adhesion onto grains. These results explain why the HCN(J=1->0)/SiO(J=2->1) intensity ratio increases with the mass loss rate in O-CSEs, and there is no need to invoke e.g. a spread in C/O-ratios for the M-stars to explain the large range of this ratio. Maser emission is very likely present in the HCN(J=1->0) line in C-CSEs, and it seems to be sensitively dependent on the mass loss rate, i.e., it appears only for {dot}(M)≲5x10^–7M_☉/yr. Based on time monitoring of this emission towards the C-stars W Ori and X TrA, we suggest that the strongest maser features are due to radial amplification in the F=2->1 transition. The predominance of redshifted maser emission could be caused by an additional amplification in the F=1->1 transition. We find no evidence for a similar maser in O-CSEs.