The high velocity stars in the Local Stellar Halo from Gaia and LAMOST
Cuihua Du, Hefan Li, Shuai Liu, Thomas Donlon, Heidi Jo Newberg
DD raft version J uly
3, 2018
Preprint typeset using L A TEX style emulateapj v. 12 / / THE HIGH VELOCITY STARS IN THE LOCAL STELLAR HALO FROM GAIA AND LAMOST C uihua D u , H efan L i , S huai L iu , T homas D onlon , H eidi J o N ewberg College of Astronomy and Space Sciences, University of Chinese Academy of Sciences, Beijing 100049, China; [email protected] School of Physical Sciences, University of Chinese Academy of Sciences, Beijing 100049, China Department of Physics, Applied Physics and Astronomy, Rensselaer Polytechnic Institute, Troy, NY 12180, USA, [email protected] Key Laboratory of Optical Astronomy, National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China
Draft version July 3, 2018
ABSTRACTBased on the first Gaia data release and spectroscopy from the LAMOST Data Release 4, we study thekinematics and chemistry of the local halo stars. The halo stars are identified kinematically with a relativespeed of at least 220 km s − with respect to the local standard of rest. In total, 436 halo stars are identified.From this halo sample, 16 high velocity (HiVel) stars are identified. We studied the metallicity and [ α / Fe]distribution of these HiVel stars. Though most of HiVel stars are metal-poor, there are several stars that havemetallicity above − . Subject headings:
Galaxy:abundance-Galaxy:halo-Galaxy:kinematics and dynamics-Galaxy:formation INTRODUCTION
High velocity (HiVel) stars, discovered in the Galactic halo(Brown et al. 2005; Hirsch et al. 2005; Edelmann et al. 2005),are moving su ffi ciently fast so that they could escape from theGalaxy. The orbits of HiVel stars can provide useful informa-tion about the environments in which they are produced. Ingeneral, the extreme velocities of high velocity stars suggestthat they were ejected from the Galactic center (GC) by theinteractions of stars with a massive black hole (MBH Hills1988) or a hypothetical binary MBH (Yu & Tremaine 2003).For either scenario, the binary stars could be injected into thevicinity of the MBH from the young stellar disk in the GC(e.g., Lu et al. 2010; Zhang et al. 2010) or from the Galacticbulge (Perets 2009). It is also possible that high velocity starscould originate from the interaction of a BH binary with a sin-gle star (Yu & Tremaine 2003), or a star cluster (Fragione &Capuzzo-Dolcetta 2016). Other models proposed to explainthe HiVel stars that do not originate in the GC include: thesurviving companion stars of type Ia supernova explosions(Zubovas et al. 2013; Tauris 2015); the tidal debris of anaccreted and disrupted dwarf galaxy (Abadi et al. 2009) orglobular cluster; the result of the interactions between multi-ple stars (Gvaramadze et al. 2009); and runaways ejected fromthe Large Magellanic Cloud (Boubert & Evans 2016, 2018).Recent studies have used the chemical and kinematic infor-mation to determine the origin of HiVel stars (e.g., Wang et al.2013; Hawkins et al. 2015; Li et al. 2012; Geier et al. 2015).A few studies have used only the kinematics of HiVel starsto obtain an estimate of the Galactic mass and Galactic es-cape speed(e.g., Smith et al. 2007; Pi ffl et al. 2014). Since thefirst hypervelocity star was discovered by Brown et al. (2005),more than 20 hypervelocity have been found (e.g., Brown etal. 2006, 2009, 2012, 2014; Zheng et al. 2014; Geier et al.2015; Huang et al. 2017). Most of these are 2 − M (cid:12) lateB-type stars in the Galactic halo. Some studies suggest thatHiVel stars are also metal-poor (e.g., Schuster et al. 2006);Ryan et al. (2003) studied a sample of intermediate metallic-ity HiVel stars and found that most of these stars resemble the stars in the thick disk. In order to put constraints on the originof the HiVel stars, it is necessary to study the chemical distri-bution of late type HiVel stars in the local halo. These studieswill also help to better understand the structure and formationof the Galactic halo, in which many of these HiVel stars cur-rently reside. For example, if the HiVel stars are more metal-rich ([Fe / H] > − .
5) than expected for the inner halo, and the[ α / Fe] measurements are consistent with those of disk stars,it may suggest that these metal-rich HiVel stars formed in thedisk and were subsequently dynamically ejected into the halo(Bromley et al. 2009; Purcell et al. 2010).In the standard hierarchical model of galaxy formation, stel-lar halos are thought to form via the accumulation of starsfrom infalling dwarf galaxies. This merging process createsmany stellar streams in the Galactic halo (Searle & Zinn 1978;Freeman & Bland-Hawthorn 2002). However, there are manysources for stellar halo material besides direct accretion frominfalling galaxies. Some simulations suggest that a fractionof kinematically defined halo stars are in situ stars (Zolotovet al. 2009; Font et al. 2011; Brook et al. 2012; Cooper etal. 2015) that formed in the initial collapse (Samland et al.2003) of a galaxy, or ‘runaway’ stars (Boubert & Evans 2018)that formed in the disk and were subsequently kinematicallyheated (Bromley et al. 2009; Purcell et al. 2010). These ‘run-away’ disk stars are a subclass of HiVel stars that can provideimportant clues to Galactic halo formation.Although there is some evidence that both in situ and ac-creted stars are present in the Milky Way halo, the origin ofthe in situ stars is still unclear due to poorly measured propermotions and parallaxes. However, as ongoing and future sur-veys such as Gaia (Perryman et al. 2001) provide us with largenumbers of radial velocities and proper motions of Galacticstars which are much more precise than previously available,it will be possible to construct accurate three-dimensional ve-locity distributions for nearly complete samples of nearbyhalo stars. These 3D maps allow us to identify the HiVel starswith higher fidelity and subsequently explore their origins.In this study, we use Gaia proper motions combined with a r X i v : . [ a s t r o - ph . GA ] J u l Du Cuihua et al.radial velocities and metallicities derived from LAMOST stel-lar spectra (Zhao et al. 2012) to search for HiVel stars in thesolar neighborhood. Section 2 introduces the observationaldata from Gaia and LAMOST, describes the sample selection,and defines the coordinate systems in the study. In Section 3,we kinematically split the sample into disk and halo compo-nents, and extract the local halo sample stars. In Section 4,we identify these rare HiVel stars in the solar neighborhoodand explore their origin, including an analysis of the chem-ical abundances and orbital properties. The conclusions andsummary are given in Section 5. DATA
Studying the kinematics and chemistry of the stellar samplerequires 6D phase space information. The first year of Gaia(DR1) provides 5D measurements in the solar neighborhood;radial velocity measurements are not included. We comple-ment the data with radial velocity and metallicity from theLAMOST survey.
Gaia and LAMOST
Gaia is a space-based mission which is obtaining accu-rate parallaxes and proper motions for more than one billionsources brighter than G ∼ .
7. The first the Gaia data release(Gaia DR1) was released in September 2016 (Gaia Collabo-ration, et al. 2016a,b), and contains positions, parallaxes andproper motions for ∼ ∼
12 (Høg et al. 2000). Thejoint catalog is known as Tycho-Gaia Astrometric Solution(TGAS; Lindegren et al. 2016). The 5-parameter astrometricsolutions for TGAS stars were obtained by combining Gaiaobservations with the positions and their uncertainties of theTycho-2 stars (with an observation epoch of around J1991)as prior information. The resulting catalog has median par-allax uncertainties of ∼ . ∼ . / yrin proper motions). Most of these stars are within a few kpcfrom the Sun, while a few objects such as supergiants exist atdistances of ∼
50 kpc.The Large Sky Area Multi-object Fiber Spectroscopic Tele-scope (LAMOST) is a 4 meter quasi-meridian reflectiveSchmidt telescope with 4000 fibers within a field of view of5 ◦ . The LAMOST spectrograph has a resolution of R ∼ ∼ . ∼ .
33 mil-lion are AFGK-type stars with estimated stellar atmosphericparameters as well as α -element abundances and radial veloc-ities. The survey reaches a limiting magnitude of r = . r denotes magnitude in the SDSS r -band), but mosttargets are brighter than r ∼
17. The scientific motivation andsurvey target selection are described in Zhao et al. (2012),Deng et al. (2012), and Liu et al. (2014).The LAMOST Stellar Parameter Pipeline at Peking Uni-versity [LSP3] (Xiang et al. 2015, 2017) determines atmo-spheric parameters by template matching with the MILESspectral library (S´anchez-Bl´azquez et al. 2006). Comparedto the ELODIE spectra (Prugniel & Soubiran 2001) whichare secured using an echelle spectrograph with a very high spectral resolution (R ∼ ∼ . − α -element to iron abundance ratio [ α / Fe] isa good indicator of the Galactic chemical enrichment history.LSP3 also gives the overall α -element (Mg, Si, Ca and Ti) toiron abundance ratio [ α / Fe](Li et al. 2016; Xiang et al. 2017).For LSS-GAC spectra of FGK stars of SNRs per pixelhigher than 10, the current implementation of LSP3 hasachieved an accuracy of 5.0 km / s, 150 K, 0.25 dex, and 0.15dex for the radial velocity, e ff ective temperature, surface grav-ity and metallicity, respectively. To provide a realistic er-ror estimate for [ α/ Fe], the random error induced by spectralnoises is combined with the method error, which is assumed tohave a constant value of 0.09 dex, estimated by a comparisonwith high-resolution measurements. The detailed descriptionof the parameters determination can be found in Xiang et al.(2015, 2017).
Sample selection and Coordinate Systems
The data used in our work are from two catalogues; the stel-lar parameters ([Fe / H], log g , [ α / Fe] ) and the line-of-sight ve-locities are from the LSS-GAC DR4 catalog, and the propermotions and parallaxes are from TGAS catalog (Gaia Collab-oration, et al. 2016a,b). We adopt the distance estimated byAstraatmadja & Bailer-Jones (2016), who applied a Bayesianmodel to derive the distance from the parallax, taking into ac-count the Milky Way prior and systematic uncertainties in theGaia catalog.Our initial sample was obtained by cross-matching betweenthe LAMOST and TGAS catalogs based on stellar position.Stars with large observational uncertainties were excludedfrom the sample. To ensure a sizable halo sample, we choseto use generous cuts rather than stringent cuts on observa-tional uncertainties. There are in total more than 230,000 starsin common with SNR ≥
20 and radial velocity uncertaintiessmaller than 10 kms − . Although it is not a very large sam-ple, it can lend insights into the stellar kinematics in the solarneighborhood.For the following analysis, we transform the Galactic ( l , b )and distances for the stars into a Cartesian coordinate system( X , Y , Z ). We use a right-handed, Cartesian Galactocentriccoordinate system defined by the following set of coordinatetransformations: X = R (cid:12) − D cos( l ) cos( b ) Y = − D sin( l ) cos( b ) (1) Z = D sin( b ) , where R (cid:12) = . D is distancefrom the star to the Sun, and l and b are the Galactic longitudeand latitude. Note that the X axis is oriented toward l = ◦ , the Y axis is oriented toward l = ◦ (the Sun’s motion in the diskis toward l ∼ ◦ ), and the Z axis toward the north Galacticpole.The tangential velocity v , is obtained from the proper mo-tion µ and the distance D by v = . µ mas · yr − D kpc km s − . (2)The proper motions together with line of sight velocities areused to calculate the Galactic velocity components ( V X = U , V Y = V , V Z = W ) and their errors, according to the formu-lae and matrix equations presented in Johnson & Soderblom(1987). Here, we adopt a Local Standard of Rest ve-locity V LSR =
220 kms − , and the solar peculiar motion( V (cid:12) , pec X , V (cid:12) , pec Y , V (cid:12) , pec Z ) = (10 . − , . − , . − )(Tian et al. 2015; Bland-Hawthorn & Gerhard 2016) anduse these values to obtain the Galactocentric velocity com-ponents: V X = V obs X + V (cid:12) , pec X V Y = V obs Y + V (cid:12) , pec Y + V LSR (3) V Z = V obs Z + V (cid:12) , pec Z We can now use this 6D phase space information to study thekinematics of local stars in the Milky Way. THE LOCAL HALO SAMPLE STARS
The space distribution in the Toomre diagram has beenwidely used to distinguish the thin-disk, thick-disk, and halostars (e.g., Venn et al. 2004; Bonaca et al. 2017). Figure 1shows the Toomre diagram of stars in the solar neighborhoodfrom the LAMOST and TGAS catalogs, where the X axis rep-resents the Galactocentric Y velocity component, V Y , whereasthe Y axis represents the perpendicular Toomre component, (cid:113) V X + V Z . As shown in Figure 1, disk stars dominate a largeoverdensity at V Y ∼
200 kms − ; the density of disk stars de-creases smoothly in both directions from this V Y value, anddoes not populate retrograde orbits ( V Y < V Y ∼ − , as can be seen in the top por-tions of Figure 1. Following Nissen & Schuster (2010) andBonaca et al. (2017), we kinematically divide the sample starsinto disk and halo components according to this Toomre dia-gram. The halo stars are defined as having | V − V LSR | > − , where V LSR = (0 , ,
0) kms − in the GalactocentricCartesian coordinates. Here, we employ the halo definitionfollowing Bonaca et al. (2017) , which is more conservativethan similar cuts adopted by Nissen & Schuster (2010). Thevelocity cut ensures that the contamination from thick diskstars is minimized. The dividing line between the componentsis marked with a red line in Figure 1. The left panel of Figure1 also shows the distribution of sample stars with a measuredmetallicity in the Toomre diagram, with color coding corre-sponding to the average metallicity of stars. The right panelshows the relative density of stars in each portion of the dia-gram and the color coding corresponds to the number densityof stars in each pixel. In total, we identified 436 local halostars within 3 kpc of the Sun.Surprisingly, there are many stars with disk-like metallici-ties ([Fe / H] > − .
0) found in the halo region of the Toomrediagram. Some metal-rich stars are very far from the regionof the diagram populated by disk stars; some are on stronglyretrograde orbits, and some of those have large V XZ veloci-ties as well. Bonaca et al. (2017) found a similar result in their study using the Gaia data combined with RAVE andAPOGEE spectroscopic surveys. The existence of metal-richstars in kinematically-defined samples of halo stars impliesthat metallicity alone cannot be used to separate halo starsand disk stars.Since the metal-rich ([Fe / H] > − .
0) halo identified in thestudy has metallicities consistent with the thick disk, we there-fore quantify the thick disk contamination to our halo sampleunder the assumption that the Galactic space velocities ( U , V ,and W ) of the stellar populations in the thin disk, the thickdisk, and the halo have Gaussian distributions: f ( U , V , W ) = k · exp( U σ U − ( V − V asym ) σ V − W σ W ) , (4)where k = π ) / σ U σ V σ W . (5)Here, σ U , σ V , and σ W are the characteristic velocity disper-sions, and V asym is the asymmetric drift. The values of thethree populations are listed in Table 1 (Bensby et al. 2003). TABLE 1Observed fraction of stars for the populations in the solar neighborhood,characteristic velocity dispersions ( σ U , σ V , and σ W ) and the asymmetricdrift ( V asym )X σ U σ V σ W V asym [km / s]Thin disk (D) 0.94 35 20 16 -15Thick disk(TD) 0.06 67 38 35 -46Halo (H) 0.0015 160 90 90 -220 To determine the probability that a given star belongs to aspecific population, we multiply the probabilities from Eq. (4)by the observed fractions ( X ) of each population in the solarneighborhood. We then obtain the relative probabilities forthick-disk-to-halo (TD / H) as follows:TD / H = X TD · f TD X H · f H (6)According to the thick disk and halo probability distribu-tions, calculated with Eq. (4) and Eq. (5), only 35 stars withTD / H > . / H] > − . / H] > − .
0. So the thick disk still doesn’t explain all metal-rich stars identified in this sample, and particularly those withhigh velocities ( | V Y | > V XZ . This suggests that there exits a metal-rich halo compo-nent in addition to metal-poor inner and outer halo compo-nents.Figure 2 presents the metallicity distribution of local halostars; there is a wide metallicity distribution ranging [Fe / H] ∼− . / H] ∼ .
5, we fit the distribution with Gaussianmodel which peak at near [Fe / H] ∼ − . / H] ∼ − .
1, than inthe APOGEE sample, [Fe / H] ∼ − .
600 400 200 0 200 400 600 V Y (km / s) V X Z ( k m / s ) [ F e / H ]
600 400 200 0 200 400 600 V Y (km / s) V X Z ( k m / s ) D e n s i t y ( k m / s ) − F ig . 1.— Toomre diagram of stars in the solar neighborhood from the LAMOST and TGAS catalogs. The dividing line between the components is marked witha red line. The left panel shows the distribution of sample stars with a measured metallicity in the Toomre diagram. The color coding corresponds to the averagemetallicity of stars. Note that there are some halo stars that are quite metal-rich. The right panel shows the relative number of stars in each part of the diagramand the color coding corresponds to the number density of stars in each pixel. [Fe/H] N u m b e r F ig . 2.— The metallicity distribution of local halo stars is fitted by a Gaus-sian model with a peak near [Fe / H] ∼ − . with [Fe / H] > − .
0. In this study, we found that about 30 per-cent of halo sample stars are metal-rich with [Fe / H] > − . ff ect the metallic-ity distribution.We decide the optimal number of Gaussian functions usingthe Bayesian information criterion (BIC): BIC = − ln [ L ( M )] + klnN where L ( M ) represents the maximum value of the likelihoodfunction of the model, N is the number of data points, and k represents the number of free parameters. More details aboutBIC can be found in Ivezi´c et al. (2014). As shown in Figure2, we adopt one-peak Gaussian models to fit the metallicitydistribution of local halo stars as the model with the lowestBIC. HIGH VELOCITY STARS IN THE LOCAL STELLAR HALO
Selection of HiVel star candidates
Before selecting HiVel star candidates, we removed starswith a higher likelihood of erroneous parameters. First, weselected only stars with calibrated T e ff between 3500 and 8000 K and estimated log g larger than 0.5 dex. In addition, starswith extremely low metallicities ([Fe / H] < − . − in thefinal sample of halo stars. In order to derive reliable space ve-locities, we constrain the sample to stars with relative errorsin the proper motions and distance smaller than 50 percent.We subsequently derive the velocity in the Galactic rest frameV gsr . Our final selection criterion of V gsr >
300 km s − gaveus a HiVel candidate sample containing 16 stars.Atmospheric parameters and position for the HiVel starscan be found in Table 2. Table 3 presents the space positionsand velocities of the 16 HiVel stars. From the spatial distri-bution in Galactic coordinates, these HiVel stars are locatedin di ff erent Galactic directions. Therefore, it is possible thatthese HiVel stars have di ff erent origins. Figure 3 gives thespace velocity distribution of our HiVel stars, showing thatthese local stars are not clumped in velocity. Chemical abundances of HiVel stars
As discussed in detail by Gilmore & Wyse (1998), chem-ical abundances have been used to discern di ff erent compo-nents of the Galaxy. Many recent surveys have shown that thedi ff erent components of the Galaxy can be partially separatedin [ α / Fe] vs. [Fe / H] distribution (Nissen & Schuster 1997;Stephens & Boesgaard 2002; Nissen & Schuster 2010; Lee etal. 2011; Feltzing & Chiba 2013; Haywood et al. 2013). Thedistribution in [ α / Fe] space also provides information aboutthe star formation rate in the stellar population. The high[ α / Fe] found in halo and thick disk stars suggests that theyformed in regions with a high star formation rate, so that onlytype II SNe contributed to their chemical enrichment. On theother hand, low − α stars originate in regions with relativelyslow chemical evolution so that type Ia SNe have had time toform, and thus contribute iron to the interstellar medium be-fore [Fe / H] ∼ − .
5. Since there is a higher iron abundance, the[ α / Fe] is lower at these higher metallicities (Nissen & Schus-ter 2010). Therefore, the abundance space of [ α / Fe] versus[Fe / H] is particularly useful in tracing the origin of individualstars (Lee et al. 2015).Figure 4 shows the chemical abundance distribution [ α / Fe]vs. [Fe / H] for all stars in this study. The red triangles rep- F ig . 3.— Velocity distribution of local halo stars in the LAMOST and TGAS catalog. The blue dots represent the halo sample stars selected from the Toomrediagram, and the red triangles represent the HiVel stars. resent the HiVel stars. The halo stars are shown individuallyas blue points and the disk stars are shown as yellow plussigns for comparison. Notice that for the halo stars, there existhigh − α stars, with [ α / Fe] scatter from 0.2 to 0.6, and low − α stars, with [Fe / H] > − . α / Fe] as a functionof increasing metallicity. The metal-poor halo is α -enhanced,while the metal-rich halo follows the abundance pattern of thedisk. The large dispersion in the [ α / Fe] could result from theuncertainty of the individual [ α / Fe] estimates. The large un-certainty in the [ α / Fe] estimates, particularly for metal-poorstars, is a result of the relatively low resolution of LAMOSTspectra.We can see from Figure 4 that our HiVel stars are are metal-poor and α − enhanced, except for HiVel7 ([Fe / H] = − . α / Fe] = − α population might haveformed as the first stars in a dissipative collapse of a proto-Galactic gas cloud (Gilmore et al. 1989; Schuster et al. 2006). Orbits of the HiVel stars
For each of the stars in our local halo sample, we inves-tigate their orbital properties by adopting a Galaxy potentialmodel. In this study, we use a recent Galactic potential modelprovided in McMillan (2017). This new model includes com-ponents that represent the contribution of the cold gas discsnear the Galactic plane, as well as thin and thick stellar discs,a bulge component and a dark-matter halo. We estimated themaximum distance above the Galactic plane (denoted Z max )and the eccentricity, e , from the orbital integration. The ec-centricity is defined as e = ( r apo − r peri ) / ( r apo + r peri ), where r peri F ig . 4.— Chemical abundance distribution [ α / Fe] vs. [Fe / H] of halo starsin the TGAS-LAMOST sample. The red triangles represent the HiVel stars.The halo stars are shown individually as blue points and the disk stars shownas yellow plus signs for comparison. The metal-poor halo is α -enhanced,while the metal-rich halo follows the abundance pattern of the disk. denotes the closest approach of an orbit to the Galactic center(i.e., the perigalactic distance), and r apo denotes the farthestextent of an orbit from the Galactic center. Figure 5 showsthe e - Z max plane, which allows us to characterize the orbitsof our sample stars; e describes the shape of the orbit and Z max describes the amplitude of the vertical oscillations (Boeche etal. 2013). Figure 5 shows that the HiVel stars have e > . max >
10 kpc, rea ffi rming that these starsare decidedly not associated with a disk (Schuster et al. 1988;Ryan et al. 2003; Schuster et al. 2006). Their orbits wouldtake them into the outer halo. However, there are two starswith Z max < e > . TABLE 2Atmospheric parameters and positions for the 16 HiVel stars.Notation source-id l b µ α cos ( δ ) µ δ RV (cid:12) T e ff log(g) [Fe / H] [ α / Fe](deg) (deg) (mas yr − ) (mas yr − ) (km s − ) (K)HiVel1 3266449244243890176 179.06 -47.69 40.15 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
10 6473 3.96 -2.18 ± ± ± ±
10 6259 4.14 -1.9 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
10 5733 4.36 -0.35 ± gsr D (cid:12) e Z max (kpc) (kpc) (kpc) (km s − ) (km s − ) (km s − ) (km s − ) (pc) (kpc)HiVel1 9.0 -0.0 -0.9 -305 ±
18 -48 ±
113 -260 ±
14 398 1199.5 ± ± ± ± ± ± ± ± ± ±
43 157 ±
69 177 ±
15 366 1224.1 ± ± ± ±
11 78 ± ± ± ± ± ± ±
22 5 ±
32 316 328.3 ± ± ± ±
43 -158 ±
63 408 ± ± ± ± ±
14 207 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
11 -177 ± ± ± ± ± ± ± ± ± ± ± ±
12 -19 ±
23 0 ± ± ± ± ±
24 -169 ±
19 641 ± ± ± ± ± ± ±
12 -712 ±
10 -166 ± ± F ig . 5.— Eccentricity, e , as a function of the maximum height above theGalactic plane, Z max . The red triangles represent the HiVel stars. The halostars are shown individually as blue points for comparison. To understand the origin of high velocity stars, we calcu-lated the backwards orbits of individual stars to see if theyconverge somewhere, and in particular whether they originatefrom the Galactic center. We did this by integrating the orbitin our Galactic potential model, starting with the current posi-tion of each star and the negative of its current velocity. In all16 HiVel stars, 3 HiVel stars (Hivel14, HiVel15 and HiVel16)are unbound due to their very high velocity and only the or-bit of 13 HiVel stars could be determined. Figure 6 gives thederived backward orbits for 13 HiVel star, integrated back 1Gyr. The red dot represents the present position, and the blackdot represents the Galactic Center. As seen in Figure 6, a fewhigh velocity stars appear to originate from the Galactic cen-ter, but others are not consistent with a GC origin, and mustbe produced by another mechanism.According to the orbital integration of HiVel stars ( shownin Figure 6), HiVel4, HiVel7, HiVel9 and HiVel10 could notoriginate in the Galactic Center. Combining their chemicaland orbit information, we conclude that HiVel4, HiVel9 andHiVel10 could originate from the tidal debris of an accretedand disrupted dwarf galaxy (Abadi et al. 2009) or globularcluster. While for HiVel7, the disrupted dwarf galaxy or glob-ular cluster explanation are unlikely due to the chemical com-position of the stars. The star likely originates in the thickdisk where one would expect a richer metallicity and lower α -abundance. HiVel5 ,HiVel13, HiVel3 and HiVel8 possiblelyare ejected from near the Galactic Center. For the rest of theHiVel stars (HiVel1, HiVel2, HiVel11 and HiVel12), it is pos-sible that they were kicked from the Galactic disk. The mech-anism by which these stars were ejected from the disk, namelybinary supernova explosion, interaction of a dwarf galaxy or aglobular cluster with the disk, or interaction between multiplestars or other gravitational mechanisms, is unclear. CONCLUSIONS AND SUMMARY
Based on the first year of Gaia data combined with obser-vations from ground-based spectroscopic survey LAMOST DR4, we analyzed a sample of local halo stars within ∼ / H] > − .
0, which ismore metal-rich than expected in the inner halo. For the halostars, there also exist high − α stars, with [ α / Fe] scatter from0.2 to 0.6, and low − α stars, with [Fe / H] > − . α / Fe] as a function of increasing metallicity. For each ofthe stars in our local halo sample, we also adopt an Galacticpotential model to derive their orbital parameters, particularlyZ max and eccentricity, to study the kinematics.From this halo sample, 16 high velocity stars are identified.We studied the metallicity and [ α / Fe] distribution of our HiVelstars. While most of the HiVel stars are metal-poor, there areseveral stars that have metallicity above − . ACKNOWLEDGEMENTS
We thank especially the referee for insightful commentsand suggestions, which have improved the paper significantly.This work was supported by joint funding for Astronomyby the National Natural Science Foundation of China andthe Chinese Academy of Science, under Grants U1231113.This work was also by supported by the Special funds ofcooperation between the Institute and the University ofthe Chinese Academy of Sciences, and China ScholarshipCouncil (CSC). HJN acknowledges funding from NSF grantAST 16-15688. Funding for SDSS-III has been providedby the Alfred P. Sloan Foundation, the Participating In-stitutions, the National Science Foundation, and the U.S.Department of Energy O ffi ce of Science. This project wasdeveloped in part at the 2016 NYC Gaia Sprint, hosted bythe Center for Computational Astrophysics at the SimonsFoundation in New York City. The Guoshoujing Telescope(the Large Sky Area Multi-Object Fiber SpectroscopicTelescope, LAMOST) is a National Major Scientific Projectbuilt by the Chinese Academy of Sciences. Funding forthe project has been provided by the National Develop-ment and Reform Commission. LAMOST is operatedand managed by the National Astronomical Observatories,Chinese Academy of Sciences. This work has made useof data from the European Space Agency (ESA) missionGaia (http: // / gaia), processed by theGaia Data Processing and Analysis Consortium (DPAC,http: // / web / gaia / dpac / consortium).Funding for DPAC has been provided by national institu-tions, in particular the institutions participating in the GaiaMultilateral Agreement. REFERENCESAbadi, M.G., Navarro, J.F., Steinmetz, M., 2009, ApJL, 691, L63Astraatmadja, T. L., & Bailer-Jones, C. A. L. 2016, ApJ, 833, 119Bensby, T., Feltzing, S.,Lundstrm, I., 2003, A&A, 410, 527Blaauw, A. 1961, BAN, 15, 265 Bland-Hawthorn, J., Gerhard, O., 2016, ARA&A, 54, 529Boeche, C. et al., 2013, A&A, 553, A19Bonaca, A., Conroy, C., M., Wetzel, A., Hopkin, P.F., & Kereˇ s , D., 2017,ApJ, 845, 101 Du Cuihua et al. F ig . 6.— 1 Gyr backwards orbit of the individual HiVel stars in XYZ
Galactocentric Coordinates. The red dot represents the current position and the black dotrepresents the Galactic Center. The thin grey lines show 100 orbits drawn at random from the uncertainties in the positions and velocities of each HiVel star,showing the uncertainty in the orbits.Boubert, D., & Evans, N. W. 2016, ApJL, 825, L6Boubert, D., & Evans, N. W., 2018, MNRAS, 477, 5261Bromley, B. C., Kenyon, S. J., Brown, W. R., & Geller, M. J. 2009, ApJ,706, 925Brook, C. B., Stinson, G. S., Gibson, B. K., et al. 2012, 426, 690Brown, W. R., Geller, M. J., Kenyon, S. J., & Kurtz, M. J. 2005, ApJL, 622,L33Brown, W. R., Geller, M. J., Kenyon, S. J., & Kurtz, M. J. 2006, ApJ,647,303Brown, W. R., Geller, M. J., & Kenyon, S. J. 2009, ApJ, 690, 1639Brown, W. R., Geller, M. J., & Kenyon, S. J. 2012, ApJ, 751, 55Brown, W. R., Geller, M. J., & Kenyon, S. J. 2014, ApJ, 787, 89Cenarro, A. J., et al. 2007, MNRAS, 374, 664Cooper, A. P., Parry, O. H., Lowing, B., Cole, S., & Frenk, C. 2015,MNRAS, 454, 3185Cui, X.Q., Zhao, Y. H., Chu, Y. Q., et al. 2012, RAA, 12, 1197Deng, L. C., Newberg, H. J., Liu, C., et al. 2012, RAA, 12, 735Edelmann, H., Napiwotzki, R., Heber, U., Christlieb, N., & Reimers, D.,2005, ApJ, 634, L181Feltzing, S., Chiba, M., 2013, New Astron. Rev., 57, 80 Font, A. S., McCarthy, I. G., Crain, R. A., et al. 2011, MNRAS, 416, 2802Fragione, G., Capuzzo-Dolcetta, R. 2016, MNRAS, 458, 2596Freeman, K., Bland-Hawthorn, J., 2002, ARA&A, 40, 487Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2016a, A&A, 595,A2Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., et al. 2016b, A&A, 595,A1Gilmore, G., & Wyse, R. F. G., 1998, AJ, 116,748Gilmore, G., Wyse, R. F. G., & Kuijken, K. 1989, ARA&A, 27, 555Geier, S., F ´’urst, F., Ziegerer, E., et al. 2015, Science, 347, 1126Gvaramadze, V. V., Gualandris, A., & Portegies Zwart, S. 2009, MNRAS,396, 570Haywood, M., Di Matteo, P., Lehnert, M. D., Katz, D., G´o mez A., 2013,A&A, 560, A109Hirsch, H. A., Heber, U., O’Toole, S. J., & Bresolin, F. 2005, A&A, 444,L61Ivezi´c, ˇZ., Connolly, A. J., VanderPlas, J. T., Gray, A. 2014, Statistics, DataMining, and Machine Learning in Astronomy, (Princeton: PrincetonUniversity Press)
Hawkins, K., Kordopatis, G., Gilmore, G., Masseron, T., et al. 2015,MNRAS, 447, 2046Hills, J.G., 1988, Nature, 331, 687Høg, E., Fabricius, C., Makarov, V. V., et al. 2000, A&A, 355, L27Huang, Y., Liu, X.W., Zhang, H.W., et al. 2017, ApJ, 847, 9Johnson, D.R.H., & Soderblom D.R., 1987, AJ, 93, 864Lee, D.M., Johnson, K.V., Sen, B., & Jessop, W., 2015, ApJ, 802,48Lee, Y. S., Beers, T.C., An, D., et al. 2011, ApJ, 738, 187Li, Y., Luo, A., Zhao, G., et al. 2012, ApJL, 744, L24Li, J., et al. 2016, Res. Astron. Astrophys., 16, 010Lindegren, L., Lammers, U., Bastian, U., et al. 2016, A&A, 595, A4Liu, X. W., et al. 2014, in Feltzing S., Zhao G., Walton N., Whitelock P.,eds, Proc. IAU Symp. 298, Setting the scene for Gaia and LAMOST,Cambridge University Press, p. 310Lu, Y., Zhang, F., & Yu, Q. 2010, ApJ, 709, 1356McMillan P. J. 2017, MNRAS, 465, 76Moultaka, J., Ilovaisky, S. A., Prugniel, P., Soubiran, C., 2004, PASP, 116,693Nissen, P. E., & Schuster, W. J., 1997, A&A, 326, 751Nissen, P.E., & Schuster, W.J., 2010, A&A, 511, L10Perryman, M. A. C., de Boer, K. S., Gilmore, G., et al. 2001, A&A, 369, 339Perets, H. B. 2009, ApJ, 698, 1330Pi fflffl