Comparative high-resolution spectroscopy of M dwarfs: Exploring non-LTE effects

Context. M dwarfs are key targets for high-resolution spectroscopy and model atmosphere analyses because of the high incidence of these stars in the solar neighbourhood and their importance as exoplanetary hosts. Several methodological challenges make such analyses difficult, leading to significant discrepancies in the published results.
Aims: The aim of our work is to compare M dwarf parameters derived by recent high-resolution near-infrared studies with each other and with fundamental stellar parameters. We also assess to what extent deviations from local thermodynamic equilibrium (LTE) for iron and potassium influence the outcome of these studies.
Methods: We carry out line formation calculations based on a modern model atmosphere grid appropriate for M dwarfs along with a synthetic spectrum synthesis code that treats formation of atomic and molecular lines in cool-star atmospheres including departures from LTE. We use near-infrared spectra collected with the CRIRES instrument at the ESO VLT as reference observational data.
Results: We find that the effective temperatures obtained with spectroscopic techniques in different studies mostly agree to better than 100 K and are mostly consistent with the fundamental temperatures derived from interferometric radii and bolometric fluxes. At the same time, much worse agreement in the surface gravities and metallicities is evident. Significant discrepancies in the latter parameters appear when results of the studies based on the optical and near-infrared observations are intercompared. We demonstrate that non-LTE effects are negligible for Fe I in M-dwarf atmospheres but are important for K I, which has a number of strong lines in the near-infrared spectra of these stars. These effects, leading to potassium abundance and metallicity corrections on the order of 0.2 dex, may be responsible for some of the discrepancies in the published analyses. Differences in the temperature-pressure structures of the atmospheric models may be another factor contributing to the deviations between the spectroscopic studies, in particular at low metallicities and high effective temperatures.
Conclusions: High-resolution spectroscopic studies of M dwarfs are yet to reach the level of consistency and reproducibility typical of similar investigations of FGK stars. Attention should be given to details of the line formation physics as well as input atomic and molecular data. Collecting high-quality spectra with a wide wavelength coverage of M dwarfs with known fundamental parameters is an essential step in benchmarking spectroscopic parameter determination of low-mass stars.