SO AND SO2 IN MASS-LOSS ENVELOPES OF RED GIANTS - PROBES OF NONEQUILIBRIUM CIRCUMSTELLAR CHEMISTRY AND MASS-LOSS RATES

We have made a search for SO in the mass-loss envelopes of 24 mostly oxygen-rich red giants, using the 8(8)-7(7)(344.3 GHz), 6(5)-5(4) (219.9 GHz), 4(3)-3(2) (138.2 GHz), and 3(2)-2(1) (99.3 GHz) transitions. One or more of these transitions were detected toward 14 envelopes, including those of the supergiants VY CMa and IRC + 10420. The SO envelope of IK Tau was measurably extended, and that of IRC + 10420 marginally so, allowing us to determine the envelope sizes directly. The SiO (v = 0) J = 8-7, J = 5-4, and J = 3-2 lines at 347.3, 217.1, and 130.3 GHz were detected simultaneously with the 344.3, 219.9, and 138.2 GHz SO lines, respectively. From our SO data, as well as from SO2 data published elsewhere (for a few sources), we derive the circumstellar SO and SO2 abundances and compare them with theoretical predictions. We find that the photospheric abundances of SO and SO2 are insignificant, and that SO and SO2 must be produced in the circumstellar envelopes. However, the (average) SO abundances are larger than even the peak values predicted by nonequilibrium circumstellar chemistry models. The large SO abundances imply that S is not heavily depleted into grains in these envelopes, and further suggest that (i) the stellar mass-loss rates in these objects are substantially larger than accepted values, (ii) the circumstellar H/H-2 abundance ratio is quite small, and (iii) the reaction of O and SH dominates over the reaction of S and OH in the production of SO. In the peculiar bipolar nebula OH 231.8 + 4.2, the combined SO and SO2 abundance exceeds the cosmic S abundance, if the mass-loss rate in this object is smaller than or equal to 10(-4) M . yr-1. In IK Tau the data can only be fitted with models in which the SO is distributed in a hollow shell, as predicted by the nonequilibrium chemistry models. In another three, which have no measurable OH, the discovery of SO is difficult to explain by such models.