Evidence for very extended gaseous layers around O-rich Mira variables and M giants

Nine bright O-rich Mira stars and five semiregular variable cool M giants have been observed with the Infrared and Optical Telescope Array (IOTA) interferometer in both K' (similar to2.15 mum) and L' (similar to3.8 mum) broadband filters, in most cases at very close variability phases. All of the sample Mira stars and four of the semiregular M giants show strong increases, from similar or equal to20% to similar or equal to100%, in measured uniform-disk (UD) diameters between the K' and L' bands. (A selection of hotter M stars does not show such a large increase.) There is no evidence that K' and L/ broadband visibility measurements should be dominated by strong molecular bands, and cool expanding dust shells already detected around some of these objects are also found to be poor candidates for producing these large apparent diameter increases. Therefore, we propose that this must be a continuum or pseudocontinuum opacity effect. Such an apparent enlargement can be reproduced using a simple two-component model consisting of a warm ( 1500 2000 K), extended (up to similar or equal to3 stellar radii), optically thin (tau similar to 0.5) layer located above the classical photosphere. The Planck weighting of the continuum emission from the two layers will suffice to make the L' UD diameter appear larger than the K' UD diameter. This two-layer scenario could also explain the observed variation of Mira UD diameters versus infrared wavelength outside of strong absorption bands as already measured inside the H, K, L, and N atmospheric windows. This interpretation is consistent with the extended molecular gas layers (H2O, CO, etc.) inferred around some of these objects from previous IOTA K'-band interferometric observations obtained with the Fiber Linked Unit for Optical Recombination (FLUOR) and from Infrared Space Observatory and high-resolution ground-based FTS infrared spectra. The two-component model has immediate implications. For example, the Mira photosphere diameters are smaller than previously recognized this certainly implies higher effective temperatures, and it may favor fundamental mode pulsation. Also, the UD model fails generally to represent the brightness distribution and has very limited applicability for Mira stars. The presence of a very extended gas layer extending up to 3 stellar radii seems now well established on a fair sample of asymptotic giant branch stars ranging from late-type giants to long-period variables, with some probable impact on stellar model atmospheres and mass-loss mechanisms.