2011MNRAS.411.2596S

We estimate the distribution of ejection velocities for the known population of high Galactic latitude runaway stars. The initial sample is a collection of 174 early-type stars selected from the literature. The stars are first classified according to their evolutionary status in order to obtain a homogeneous sample of 96 genuine main-sequence stars. Their present velocities and flight times are then estimated using proper motion data from various astrometric catalogues (including Tycho-2, UCAC2 and USNO-B) and the ejection velocities are computed by tracing their orbits back in time, based on a Galactic potential. The potential used is constructed from a mass density model chosen to fit the most recent observational constraints.

We find evidence for two different populations of runaway stars: a `high' velocity population, with a maximum ejection velocity of about 400-500 km/s, and a `low' velocity population, with a maximum ejection velocity of about 300 km/s. We argue that the observed limit of 500 km/s and the bimodality of the observed ejection velocity distribution are natural consequences of the so-called Binary Ejection Mechanism. We discuss the connection between the `high' velocity population and the so-called hypervelocity stars, showing how previously studied hypervelocity stars are consistent with the results obtained.

We also find that some stars that were once thought to be best explained as being formed in the halo are compatible with a runaway hypothesis once proper motions are included in the analysis. However, three stars in the selected sample appear to be inconsistent with ejection from the Galactic disc. Possible scenarios are discussed, including a possible formation in the Galactic halo.