HOney-BeeS II. Be-X-ray binaries as testbeds for spectroscopic studies of Be stars

Abstract

The majority of massive classical Be stars are thought to be binary interactions products. Their rapid rotation and often strong, variable, and emission-line dominated spectrum, make spectroscopic analysis challenging. Hence, robust binary properties and statistical constraints are still lacking for the Be population. In this study, we use seven Be-X-ray binaries, and their orbital periods derived from the X-rays, to investigate the reliability of different spectral lines and numerical methods for the measuring of radial-velocities and orbital period determination of Be stars. We use multi-epoch high-resolution HERMES spectra and compare absorption- and emission-line radial velocities obtained with cross-correlation, line-profile fitting, and the bisector method. Line-profile variability affects the bisector method and line-profile fitting requires model templates that do not encompass the complexity of Be-star line profiles. Therefore, we recommend using cross-correlation: it is independent of models and easily compatible with line blends seen in Be-star spectra. The obtained statistical uncertainty on the radial-velocities from cross-correlation is 0.2-0.3km/s for H$\alpha$ (emission) and ~5km/s for absorption lines, excluding potential systematics. In general, whether the goal is to do binary statistics of a population or an in-depth study of a specific system, we suggest using emission lines, due to a higher precision and less scatter than absorption lines. Here, H$\beta$ is preferred over H$\alpha$ because of its lesser variability. However, large-scale variability may cause large shifts in emission-line radial velocities, resulting in spurious eccentricities. In this case, orbital solutions should ideally be compared to lower-signal absorption lines (if present). Finally, we highlight the need for understanding how companion-disc interactions alter emission-line appearance.