Optical and Near-Infrared Radial Velocity Content of M Dwarfs: Testing Models with Barnard's Star

TL;DR

This study compares observed and modeled spectra of Barnard's star across optical and near-infrared bands, revealing significant discrepancies in predicted radial velocity content crucial for exoplanet detection around M dwarfs.

Contribution

It provides the first detailed comparison of observed and theoretical spectral line profiles of M dwarfs across multiple bands, highlighting model inaccuracies.

Findings

Models over-predict Y and J band RV content by over a factor of 2.

Models under-predict H and K band RV content by approximately 50%.

Results impact the planning of RV surveys for exoplanets around M dwarfs.

Abstract

High precision radial velocity (RV) measurements have been central in the study of exoplanets during the last two decades, from the early discovery of hot Jupiters, to the recent mass measurements of Earth-sized planets uncovered by transit surveys. While optical radial-velocity is now a mature field, there is currently a strong effort to push the technique into the near-infrared (nIR) domain (chiefly $Y$, $J$, $H$ and $K$ band passes) to probe planetary systems around late-type stars. The combined lower mass and luminosity of M dwarfs leads to an increased reflex RV signal for planets in the habitable zone compared to Sun-like stars. The estimates on the detectability of planets rely on various instrumental characteristics, but also on a prior knowledge of the stellar spectrum. While the overall properties of M dwarf spectra have been extensively tested against observations, the same is not true for their detailed line profiles, which leads to significant uncertainties when converting a given signal-to-noise ratio to a corresponding RV precision as attainable on a given spectrograph. By combining archival CRIRES and HARPS data with ESPaDOnS data of Barnard's star, we show that state-of-the-art atmosphere models over-predict the $Y$ and $J$-band RV content by more than a factor of $\sim$$2$, while under-predicting the $H$ and $K$-band content by half.