Fundamental stellar parameters such as masses and radii of well detached double-lined spectroscopic eclipsing binaries can be determined very accurately (Andersen 1991). Therefore, from accurate (1-2%) stellar mass and radius determinations of such objects, one can compute surface gravities to very high and otherwise inaccessible confidence levels. Indeed, Henry & Mc Carthy (1993) have discussed the data available for visual binaries of solar mass and below. Only Cen B (G2V,0.90 ) has a mass known with an accuracy comparable to that for favorable eclipsing binaries. This point shows the importance of choosing such double-lined eclipsing binaries in order to obtain surface gravities with the highest possible accuracy. Moreover, these binaries are of great interest to perform accurate tests of stellar evolutionary models (see e.g. Lastennet et al. 1996, Pols et al. 1997, Lastennet & Valls-Gabaud 1998) used to derive cluster ages. The knowledge of all possible stellar parameters for such single stars is the basis of the modelling of the global physical properties and evolution of star clusters or galaxies. Nevertheless, while masses and radii are accurately known, the effective temperatures - and consequently, the luminosities of these stars - strongly depend upon the calibration used to relate photometric indices with . As a matter of fact, for such binaries the temperatures given in the literature come from various calibration procedures and are indeed highly inhomogeneous. Furthermore, due to the lack of empirical calibrations at different metallicities, solar metallicity is often assumed for photometric estimations of . In this regard, synthetic photometry performed from large grids of stellar atmosphere models, calibrated in , , and , provides a powerful tool of investigation. In this paper, we explore simultaneous solutions of and , and we address the question of the reliability of metallicity-independent effective temperature determinations.
We have selected 20 binary systems (40 stars) for which we have uvby Strömgren photometry with estimated errors (see Table 1). For this sample, previous estimates of effective temperature are not homogeneous, originating from various calibrations established for different spectral-type domains: Morton & Adams (1968), Relyea & Kurucz (1978), Osmer & Peterson (1974), Grosbol (1978), Davis & Shobbrook (1977), Popper (1980), Moon & Dworetsky (1985), Saxner & Hammarbäck (1985), Jakobsen (1986), Magain (1987), Napiwotzki et al. (1993), and Edvardsson et al. (1993). Moreover, all these studies are of course historically not fully independent. As an example, the of Moon & Dworetsky (1985) is estimated using the , (B-V)0 calibration of Hayes (1978) and the , c0 calibration of Davis & Shobbrook (1977). However, this does not mean that these calibrations allow to derive very similar temperatures. As highlighted by Andersen & Clausen (1989) concerning the O-type components of EM Carinae, the temperature calibration of Davis & Shobbrook (1977), Jakobsen (1986), and Popper (1980) do not agree particularly well. A similar comparison of these three calibrations made by Clausen & Giménez (1991) with the massive B-type components of CW Cephei leads to 5500K ! Thus, a new and homogeneous determination of effective temperature is of primordial interest for such well-known objects. In order to re-derive in a homogeneous way the of these stars, we have used the Ba sel S tellar L ibrary (hereafter "BaSeL" ) which provides empirically calibrated model spectra over a large range of stellar parameters (Lejeune et al. 1997, 1998a).
Table 1. Strömgren photometry for the sample (after Table 5 of Jordi et al. 1997). Some useful notes about reddening are given in the last column.
In Sect. 2, we will describe the models used to perform our calculation of from uvby Strömgren photometry. Sect. 3 will be devoted to the description of the method and the presentation of the results.
© European Southern Observatory (ESO) 1999
Online publication: December 16, 1998