Astron. Astrophys. 351, 619-626 (1999)

1. Introduction

Fundamental determination of stellar masses from binary orbits is a most classical astrophysical discipline, last comprehensively reviewed by Andersen (1991; 1998). Besides defining a mean mass-luminosity relation, used at many places in astronomy to approximately convert the observable stellar light to the underlying mass (for instance to derive an initial mass function), accurate stellar masses in multiple systems provide what is perhaps the most demanding and fundamental test of stellar evolution theory (e.g. Andersen 1991). Mass, the basic input of evolutionary models, is directly measured and the models must reproduce the effective temperatures and radii (or luminosities) of both components, for a single age and a single chemical composition. Given the strong dependence of all stellar parameters on mass, this discriminating diagnostic however only shows its power for relative mass errors 1-2%. In practice this has up to now restricted its use to double-lined detached eclipsing binaries. These systems are however relatively rare: only 44 have yielded masses accurate enough to be included in Andersen's (1991) critical compilation, and few have appeared in the literature since then. Tidally induced rotational mixing may in addition affect the evolution of the short period eclipsing systems, perhaps sufficiently that they are not completely representative of isolated stars. More seriously however, detached eclipsing systems poorly fill some interesting areas in the HR diagram.

The lower main sequence is one major region with very few known eclipsing systems, as a result of the strong observational and intrinsic biases against observing eclipses in faint and physically small stars. Only three well detached eclipsing binaries are known with significantly subsolar component masses: YY Gem (M0Ve, 0.6+0.6 ; Bopp 1974; Leung & Schneider 1978), CM Dra (M4Ve, 0.2+0.2 ; Lacy 1977, Metcalfe et al. 1996), and the recently identified GJ 2069A (M3.5Ve, 0.4+0.4 ; Delfosse et al. 1999a). Detailed observational checks of evolutionary models (e.g. Paczynski & Sienkiewicz 1984; Chabrier & Baraffe 1995) have therefore heavily relied on the first two of these systems, even though they are in some respects non-ideal for this purpose: both binaries have two nearly equal mass components, so that the strength of the differential comparison of two stars with different masses but otherwise equal parameters is largely lost. Also, all three are chromospherically very active, due to tidal synchronisation of their rotation with the short orbital period. As a consequence, they may have untypical colours for their bolometric luminosity.

Angularly resolved spectroscopic binaries provide stellar masses in parts of the HR diagram where eclipsing systems are missing, though to date these systems have not matched the 1% accuracy which can be obtained in detached eclipsing systems. For M dwarfs in particular, the best representation to date of the empirical M-L relation (Henry & Mc Carthy 1993, hereafter HMcC; Henry et al. 1999) still has to rely in part on some fairly noisy mass determinations. For several years (Perrier et al. 1992) we have therefore been following up with high angular resolution some low mass spectroscopic systems found with the CORAVEL or ELODIE radial velocity spectrographs. This follow-up initially used one-dimensional (1D) IR speckle, then two-dimensional (2D) IR speckle, and now uses adaptive optics imaging. As a progress report on this program and as an illustration of our methods, we present here much improved parameters for the double-lined interferometric binary Gl 570BC. The 1% accuracy for the 2 masses improves by an order of magnitude on our earlier measurements (Mariotti et al. 1990) of this system, and is getting close to what is obtainable for eclipsing systems.

The Gl 570 system comprises the V =5.7 K4V star Gl 570A (also HR 5568, HD 131977, HIP 73184, FK5 1391), and at a projected distance of 25" the close Gl 570BC pair (also HD 131976, HIP 73182) which is the subject of the present paper. As discussed below, the orbital period of the close pair is fairly short, only 10 months. Thanks to its small distance of only 6 pc it is nonetheless usually well resolved by the diffraction limit of 4m-class telescopes. The separation within Gl 570BC on the other hand always remains less than 0.2", so that all seeing-limited measurements refer to integrated properties of the close pair. Its integrated magnitude is V =8.09, and its joint spectral type is M1V (Henry et al. 1994; Reid et al. 1995). The three components have common parallax, proper motion, and radial velocity. They are therefore gravitationnally bound, with the projected separation of the wide pair (about 120 AU at the distance of the system) corresponding to P  500 years. Formation of the system could either result from the fragmentation of a single gas clump or have involved some capture(s). The latter process however only remains efficient at the high densities which characterize very rich star forming clusters, whose lifetimes are very much shorter than the hydrogen burning timescale in a K4 dwarf. For all practical purposes, the three stars can thus in both cases be considered as coeval and as formed from the same gas.

© European Southern Observatory (ESO) 1999

Online publication: November 3, 1999
helpdesk.link@springer.de