MAGIS (Measuring Abundances of red super Giants with Infrared Spectroscopy) project: I. Establishment of an abundance analysis procedure for red supergiants and its evaluation with nearby stars

Context. Given their high luminosities (L ≳ 10⁴ L_⊙), red supergiants (RSGs) are good tracers of the chemical abundances of the young stellar population in the Milky Way and nearby galaxies. However, previous abundance analyses tailored to RSGs suffer some systematic uncertainties originating in, most notably, the synthesized molecular spectral lines for RSGs. Aims. We establish a new abundance analysis procedure for RSGs that circumvents difficulties faced in previous works, and test the procedure with ten nearby RSGs observed with the near-infrared high-resolution spectrograph WINERED (0.97‑1.32 µm, R = 28 000). The wavelength range covered here is advantageous in that the molecular lines contaminating atomic lines of interest are mostly weak. Methods. We first determined the effective temperatures (T_eff) of the targets with the line-depth ratio (LDR) method, and calculated the surface gravities (log 𝑔) according to the Stefan-Boltzmann law. We then determined the microturbulent velocities (v_micro) and metallicities ([Fe/H]) simultaneously through the fitting of individual Fe I lines. Finally, we also determined the abundance ratios ([X/Fe] for element X) through the fitting of individual lines. Results. We determined the [X/Fe] of ten elements (Na I, Mg I, Al I, Si I, K I, Ca I, Ti I, Cr I, Ni I, and Y II). We estimated the relative precision in the derived abundances to be 0.04‑0.12 dex for elements with more than two lines analyzed (e.g., Fe I and Mg I) and up to 0.18dex for the other elements (e.g., Y II). We compared the resultant abundances of RSGs with the well-established abundances of another type of young star, namely the Cepheids, in order to evaluate the potential systematic bias in our abundance measurements, assuming that the young stars (i.e., both RSGs and Cepheids) in the solar neighborhood have common chemical abundances. We find that the determined RSG abundances are highly consistent with those of Cepheids within <0.1 dex for some elements (notably [Fe/H] and [Mg/Fe]), which means the bias in the abundance determination for these elements is likely to be small. In contrast, the consistency is worse for some other elements (e.g., [Si/Fe] and [Y/Fe]). Nevertheless, the dispersion of the chemical abundances among our target RSGs is comparable with the individual statistical errors on the abundances. Hence, the procedure is likely to be useful to evaluate the relative difference in chemical abundances among RSGs.