Astron. Astrophys. 363, 1065-1080 (2000)

1. Introduction

It has long been recognized that in the coolest stellar atmospheres a significant fraction of the hydrogen forms molecular hydrogen (Tsuji 1964). The dissociation energy of the H₂ molecule is 4.48 eV and in layers cooler than 2500 K molecular hydrogen is the predominant atmospheric constituent. Nevertheless, spectral lines from molecular hydrogen are notoriously difficult to observe. As noted by Herzberg (1938), this is because homonuclear diatomic molecules such as H₂ have no permitted electric dipole spectrum and hence no permitted rotation or vibration-rotation spectrum. There is a permitted quadrupole spectrum but the lines are weak. In the case of H₂ a typical quadrupole oscillator strength is 310^-14. By comparison, a typical molecular dipole or atomic oscillator strength in the same spectral region as the H₂ quadrupole lines is 10⁶ times larger.

Detection of H₂ lines in cool giant spectra, especially in the spectra of AGB stars, is important for a number of reasons: Due to the large predicted abundance of H₂ in the stellar atmosphere detection of spectral lines allows a definitive test of stellar model atmospheres (Lambert et al. 1973). Planetary nebulae are known in which hydrogen is depleted (Kaler 1985). Since luminous AGB stars are the immediate precursors of planetary nebulae, it would be interesting to investigate the hydrogen abundance in AGB atmospheres. Finally, in the case of Mira variables, H₂ has been shown to be critical in computing the energy dissipated as the pulsation induced shock moves through the atmosphere. H₂ dissociation can absorb up to 50% of the thermal kinetic energy in a shock (Fox & Wood 1985). While H₂ will be present in a cool equilibrium atmosphere, Bowen (1988) argues that H₂ will not be present in Mira atmospheres because the recombination time (through three body collisions) is much longer than the Mira pulsation period.

The quadrupole vibration-rotation lines of H₂ lie near the flux maximum in cool stars and the continuous opacity minimum at 2 µm. The much stronger electronic transitions are situated in the 1200-1500 Å ultraviolet. Since in cool stars the flux is small in the ultraviolet, the infrared is a more practical region in which to look for this molecule. In spite of the expected weakness of the H₂ infrared quadrupole transitions, there have been a number of searches for these lines. These were based on predictions from model atmosphere calculations that H₂ would reach a large enough column density in cool giant atmospheres to be observable (Lambert et al. 1973; Goorvitch et al. 1980). Early attempts found only upper limits resulting in a gross mismatch between predicted and observed equivalent widths, with the predictions far in excess of the observations. With improved observations, H₂ lines were detected in carbon stars (Johnson et al. 1983) and improved models have resulted in a reasonable agreement between observation and prediction for the atmospheres of carbon stars (Lambert et al. 1986). However, Tsuji (1983) concluded that H₂ was not convincingly present in the oxygen rich cool giants and that this result was not in agreement with model atmospheres.

In two earlier papers we noted the presence of very strong molecular hydrogen lines in the spectra of the S-type Miras (Hall & Ridgway 1977; Hinkle et al. 1982[HHR]). These strong H₂ lines in stars with C/O 1 are generally in agreement with carbon star results showing that equivalent widths of the H₂ S(0) line increase with decreasing effective temperature and increase strongly as C/O approaches unity (Lambert et al. 1986). In this paper we explore in detail the behavior of the H₂ lines as a function of phase in Mira variables and as a function of C/O and effective temperature in cold long period variables (LPVs). Our goal is to establish when and why this useful atmospheric diagnostic can be detected in late type giants.

© European Southern Observatory (ESO) 2000

Online publication: December 5, 2000
helpdesk.link@springer.de