2012A&A...538A..25N

Stellar parameters are not easily derived from M dwarf spectra, which are dominated by complex bands of diatomic and triatomic molecules and do not agree well with the individual line predictions of atmospheric models. M dwarf metallicities are therefore most commonly derived through less direct techniques. Several recent publications propose calibrations that provide the metallicity of an M dwarf from its K_s band absolute magnitude and its V-K_s color, but disagree at the ±0.1dex level. We compared these calibrations using a sample of 23 M dwarfs, which we selected as wide (>5arcsec) companions of F-, G-, or K-dwarfs with metallicities measured on a homogeneous scale and which we require to have V band photometry measured to better than ∼0.03mag. We find that the Schlaufman & Laughlin (2010A&A...519A.105S) calibration has the lowest offsets and residuals against our sample, and used our improved statistics to marginally refine that calibration. With more strictly selected photometry than in previous studies, the dispersion around the calibration is well in excess of the [Fe/H] and photometric uncertainties. This suggests that the origin of the remaining dispersion is astrophysical rather than observational.