2012ApJ...761..161S

We present a medium-resolution spectroscopic survey of late-type giant stars at mid-Galactic latitudes of (30° <|b| < 60°), designed to probe the properties of this population to distances of ∼9 kpc. Because M giants are generally metal-rich and we have limited contamination from thin disk stars by the latitude selection, most of the stars in the survey are expected to be members of the thick disk ( < [Fe/H] > ∼ -0.6) with some contribution from the metal-rich component of the nearby halo. Here we report first results for 1799 stars. The distribution of radial velocity (RV) as a function of l for these stars shows (1) the expected thick disk population and (2) local metal-rich halo stars moving at high speeds relative to the disk, which in some cases form distinct sequences in RV-l space. High-resolution echelle spectra taken for 34 of these "RV outliers" reveal the following patterns across the [Ti/Fe]-[Fe/H] plane: 17 of the stars have abundances reminiscent of the populations present in dwarf satellites of the Milky Way, 8 have abundances coincident with those of the Galactic disk and a more metal-rich halo, and 9 of the stars fall on the locus defined by the majority of stars in the halo. The chemical abundance trends of the RV outliers suggest that this sample consists predominantly of stars accreted from infalling dwarf galaxies. A smaller fraction of stars in the RV outlier sample may have been formed in the inner Galaxy and subsequently kicked to higher eccentricity orbits, but the sample is not large enough to distinguish conclusively between this interpretation and the alternative that these stars represent the tail of the velocity distribution of the thick disk. Our data do not rule out the possibility that a minority of the sample could have formed from gas in situ on their current orbits. These results are consistent with scenarios where the stellar halo, at least as probed by M giants, arises from multiple formation mechanisms; however, when taken at face value, our results for metal-rich halo giants suggest a much higher proportion to be accreted than found by Carollo et al. and more like the fraction suggested in the analysis by Nissen & Schuster and Schuster et al. We conclude that M giants with large RVs can provide particularly fruitful samples to mine for accreted structures and that some of the velocity sequences may indeed correspond to real physical associations resulting from recent accretion events.