Hiroyuki Tako ISHIKAWA

M dwarf is the smallest and coolest type of the main sequence star. Thanks to their small mass and size, and low luminosity, they have the advantage to detect planets using RV and transit methods, as well as to characterize small planets in the habitable zone. Recently, M dwarfs have become prime targets of many exoplanet search projects. The knowledge of their chemical composition is crucial to understand the formation process and internal structure of the planets that come to be discovered around M dwarfs. However, chemical analysis of M dwarfs has been problematic due to their faintness and complex molecular absorptions in their spectra. Most previous studies about the chemistry of M dwarfs only dealt with the overall metallicity using the empirical relations or template spectral fitting, and the attempts to determine the abundances of individual elements have been very limited. The reliability of those metallicity determinations needs to be examined.
The near-infrared spectra are relatively free from the complicated and crowded molecular bands compared to the visible spectra and thereby more suitable for the line-by-line analysis of atomic lines and the several high-resolution near-infrared spectrometers have become available in the last decade. We need to establish the procedure to determine those abundances and to verify the reliability of it.
We demonstrate the detailed chemical analysis of five M dwarfs (effective temperature T_eff ∼ 3200–3800 K) in binary systems with G/K-type stars, using the high-resolution (R ∼ 80,000) near-infrared (960–1710 nm) spectra obtained with CARMENES.
At first, we measured the equivalent widths (EW; the area between the line profile and the continuum level in the normalized spectra) of the selected (isolated & sensitive to abundance) atomic lines by fitting the Gaussian or Voigt profile. We compared the EW with the theoretical one derived from the synthetic spectra calculated assuming stellar parameters, which include the elemental abundances, the atmospheric structure such as the temperature profile (MARCS), and the data of absorption lines (VALD3). Elemental abundance is searched iteratively with changing the parameters so that the theoretical EW matches the observed EW.
We determined the chemical abundances of eight elements (Na, Mg, K, Ca, Ti, Cr, Mn, and Fe). They are evaluated by comparing with the chemical abundances reported for the G/K-type primary stars, which are reliably derived by Montes et al. (2018) from the high-resolution visible spectra. Stars in a binary system form together from the same molecular cloud, thus their chemical abundances can be regarded to be almost identical (~0.02 dex). The first figure on page 4 presents our results for the eight elements of three M dwarfs along with those abundances of G/K primary stars with the error bars of measurement uncertainty (typically ~0.2 dex). They show the agreement within the error margin.
Through the analysis process, we investigate the unique behavior of atomic lines in a cool atmosphere. Most atomic lines are sensitive to the abundance changes not only of the corresponding elements but also of other elements, especially the dominant electron donors such as Na and Ca. The Ti I lines show a negative correlation with the overall metallicity in T_eff < 3400 K due to the formation of TiO molecules (This behavior of the lines are seen in the second figure on page 4). Those findings indicate that to correctly estimate the overall metallicity or the abundance of any element, we need to determine the abundances of other individual elements consistently. The future application of our chemical analysis to the target samples of the planet search projects around M dwarfs will provide important information on the planetary systems around M dwarfs.