2005ApJ...626..465R

We present up-to-date metallicity-dependent temperature versus color calibrations for main-sequence and giant stars based on temperatures derived with the infrared flux method (IRFM). Seventeen colors in the photometric systems UBV, uvby, Vilnius, Geneva, RI(Cousins), DDO, Hipparcos-Tycho, and Two Micron All Sky Survey (2MASS) have been calibrated. The spectral types covered by the calibrations range from F0 to K5 (7000K≳T_eff≳4000K) with some relations extending below 4000 K or up to 8000 K. Most of the calibrations are valid in the metallicity range -3.5≳[Fe/H]≳0.4, although some of them extend to as low as [Fe/H]~-4.0. All fits to the data have been performed with more than 100 stars; standard deviations range from 30 to 120 K. Fits were carefully performed and corrected to eliminate the small systematic errors introduced by the calibration formulae. Tables of colors as a function of T_effand [Fe/H] are provided. This work is largely based on the study by A. Alonso and collaborators; thus, our relations do not significantly differ from theirs except for the very metal-poor hot stars. From the calibrations, the temperatures of 44 dwarf and giant stars with direct temperatures available are obtained. The comparison with direct temperatures confirms our finding in Paper I that the zero point of the IRFM temperature scale is in agreement, to the 10 K level, with the absolute temperature scale (that based on stellar angular diameters) within the ranges of atmospheric parameters covered by those 44 stars. The colors of the Sun are derived from the present IRFM T_effscale and they compare well with those of five solar analogs. It is shown that if the IRFM T_effscale accurately reproduces the temperatures of very metal-poor stars, systematic errors of the order of 200 K, introduced by the assumption of (V-K) being completely metallicity independent when studying very metal-poor dwarf stars, are no longer acceptable. Comparisons with other T_effscales, both empirical and theoretical, are also shown to be in reasonable agreement with our results, although it seems that both Kurucz and MARCS synthetic colors fail to predict the detailed metallicity dependence, given that for [Fe/H]=-2.0, differences as high as approximately ±200 K are found.