Abstract
The relativley large spread in the derived metallicities ([Fe/H]) of M dwarfs shows that various approaches have not yet converged to consistency. The presence of strong molecular features, and incomplete line lists for the corresponding molecules have made metallicity determinations of M dwarfs difficult. Furthermore, the faint M dwarfs require long exposure times for a signal-to-noise ratio sufficient for a detailed spectroscopic abundance analysis. We present a high-resolution (R~50,000) spectroscopic study of a sample of eight single M dwarfs and three wide-binary systems observed in the infrared J-band. The absence of large molecular contributions allow for a precise continuum placement. We derive metallicities based on the best fit synthetic spectra to the observed spectra. To verify the accuracy of the applied atmospheric models and test our synthetic spectrum approach, three binary systems with a K-dwarf primary and an M-dwarf companion were observed and analysed along with the single M dwarfs. We obtain a good agreement between the metallicities derived for the primaries and secondaries of our test binaries and thereby confirm the reliability of our method of analysing M dwarfs. Our metallicities agree well with certain earlier determinations, and deviate from others. We conclude that spectroscopic abundance analysis in the J band is a reliable method for establishing the metallicity scale for M dwarfs. We recommend its application to a larger sample covering lower as well as higher metallicities. Further prospects of the method include abundance determinations for individual elements.