Mg I emission lines at 12 and 18 in K giants^*

¹ Institut für Astronomie und Astrophysik der Universität München, Scheinerstr. 1, 81679 München, Germany e-mail: jon@usm.uni-muenchen.de
² Department of Astronomy & Space Physics, Uppsala University, Box 515, 751 20 Uppsala, Sweden
³ Lund Observatory, Box 43, 221 00 Lund, Sweden
⁴ Center for Astrophysics and Space Astronomy, Astrophysics Research Lab, 593 UCB, University of Colorado, Boulder, CO 80309-0593, USA
⁵ Department of Physics, University of California at Davis, CA 95616, USA

Received: 13 March 2008
Accepted: 22 May 2008

Abstract

Context. The solar mid-infrared metallic emission lines have already been observed and analyzed well, and the formation scenario of the Mg I 12 m lines has been known for more than a decade. Detections of stellar emission at 12 m have, however, been limited to Mg I in very few objects. Previous modeling attempts have been made only for Procyon and two cool evolved stars, with unsatisfactory results for the latter. This prevents the lines' long predicted usage as probes of stellar magnetic fields.

Aims. We want to explain our observed Mg I emission lines at 12 m in the K giants Pollux, Arcturus, and Aldebaran and at 18 m in Pollux and Arcturus. We discuss our modeling of these lines and particularly how various aspects of the model atom affect the emergent line profiles.

Methods. High-resolution observational spectra were obtained using TEXES at Gemini North and the IRTF. To produce synthetic line spectra, we employed standard one-dimensional, plane-parallel, non-LTE modeling for trace elements in cool stellar atmospheres. We computed model atmospheres with the MARCS code, applied a comprehensive magnesium model atom, and used the radiative transfer code MULTI to solve for the magnesium occupation numbers in statistical equilibrium.

Results. The Mg I emission lines at 12 m in the K giants are stronger than in the dwarfs observed so far. We present the first observed stellar emission lines from Mg I at 18 m and from Al I, Si I, and presumably Ca I at 12 m. We successfully reproduce the observed Mg I emission lines simultaneously in the giants and in the Sun, but show how the computed line profiles depend critically on atomic data input and how the inclusion of energy levels with n ≥ 10 and collisions with neutral hydrogen are necessary to obtain reasonable fits.