One Large Blob and Many Streams Frosting the nearby Stellar Halo in Gaia DR2

Figure 2 shows the velocity distribution of the kinematically selected halo stars. This distribution is rather complex, particularly in the V_x − V_y projection (left panel of Figure 2). This is in part due to the sharpness of the selection criterion applied on the Toomre diagram. This figure also shows that slightly relaxing the value of V_cut would lead to the inclusion of more stars from the tail of the velocity distribution of the disk(s) (Schönrich & Binney 2009; Bonaca et al. 2017). Their imprint is a noticeable asymmetry in the V_x-distribution, which is characteristic of the perturbation induced by the Galactic bar (e.g., Antoja et al. 2015). In our sample, this asymmetry has a smaller amplitude than for the thin disk, but it is nonetheless clearly, and possibly unexpectedly, present at V_y ≳ 100 km s⁻¹, where the number of stars with V_x < 0 is 389, whereas for V_x > 0 there are 322 stars (∼3σ excess). Properly understanding the dynamical properties of this transition region of velocity space would likely require a multi-dimensional probabilistic analysis to assign stars to different physical components using also chemical and age information (Binney et al. 2014; Posti et al. 2017), which is beyond the scope of this Letter.

Zoom In Zoom Out Reset image size

The left panel of Figure 2 shows a broad overdensity of stars for positive V_x at V_y ∼ 50 km s⁻¹, just where the contribution from the low-velocity tail of the disks would be expected to die away. Although to understand its nature requires an in-depth analysis, the location of this overdensity seems unlikely to be related to the Toomre-diagram-based selection criterion.

Figure 2 shows the presence of a prominent component with a large dispersion inV_x (equivalent to V_R near the Sun) that has a slightly retrograde mean motion of a few tens of km s⁻¹. Although this could be considered as the "traditional" halo, it is slightly too retrograde and asymmetric toward more negative V_y. Figure 2 also reveals that the high-velocity tails of the distribution of the stars in our halo sample are populated by several cold clumps. The structure at V_y ∼ 150 km s⁻¹ and ${V}_{z}\sim -250$ km s⁻¹ (in green, 25 stars), can be associated with one of the streams found by Helmi et al. (1999), and also reported in Gaia Collaboration et al. (2018c) using only proper motion information (their Figure 25). The second stream found by Helmi et al. (1999) is less conspicuous with 12 stars, but present at V_y ∼ 150 km s⁻¹ and V_z ∼ 230 km s⁻¹. The asymmetry in the number of stars in each of the streams implies that the accretion event from which these streams originate must have happened 6 to 9 Gyr ago according to the models presented in Kepley et al. (2007).

There are also other clumps in Figure 2 and these are marked with different colors. Some appear to overlap with previously reported hints of substructure (e.g., Re Fiorentin et al. 2015). For example, the orange circles are probably related to the "retrograde outlier stars" of Kepley et al. (2007), and those that are blue overlap with the structure VelHel-4 from Helmi et al. (2017).

Now we explore the distribution of stars in the space of "integrals of motion" defined by the z-component of the angular momentum¹ ${L}_{z}$, the perpendicular component ${L}_{\perp }\,=\sqrt{{L}_{x}^{2}+{L}_{y}^{2}}$, and the energy E. For the stars in our sample we compute their total energy E as the sum of a kinetic and a potential term, where the amplitude of the latter is estimated using a suitable Galactic potential² (see Helmi et al. 2017 for details). Note that ${L}_{\perp }$ is not really conserved in an axisymmetric potential like that of our Galaxy, but nonetheless it has proven to be a useful proxy for a third integral, and helps to discriminate substructures with different orbital properties (Helmi et al. 1999; Helmi & de Zeeuw 2000).

Figure 3 shows the distribution of the halo stars in our sample in the E −  L_z space. Although this figure is reminiscent of that obtained using a TGAS × RAVE sample (Helmi et al. 2017) or perhaps even TGAS × SDSS (Myeong et al. 2018), it is much more spectacular.

Zoom In Zoom Out Reset image size

Figure 3 shows a clear prominent "blob" or "plume" that is slightly retrograde, the counterpart of the structure seen in velocity space (Figure 2) and also in the Toomre diagram (Figure 1). This region has been previously associated with where a possible progenitor of OmegaCen would deposit debris (Dinescu 2002; Bekki & Freeman 2003; Majewski et al. 2012; Helmi et al. 2017). Whether this is the only progenitor populating this region of phase-space remains to be seen. Because of its large extent in energy, this structure contributes stars to the outer halo, and so may well be at least partly responsible for the "retrograde" component previously reported by Carollo et al. (2007, and subsequent work). If a single event, its size in "integrals of motion" space suggests it was very significant.

Figure 4 shows a scatter plot of the distribution of kinematically selected stars in the E −  L_z (top) and L_⊥ − L_z (bottom) spaces. We have plotted here as a density map the contribution of all the stars in the Gaia DR2 6D sample located within 1 kpc from the Sun and with distance errors smaller than 20%. These figures clearly show where the disks fall, as well as the sharpness of our kinematic selection criterion in the transition region between the disks and the (traditional) halo.

Zoom In Zoom Out Reset image size

Several tight overdensities are also apparent in the panels of Figure 4. There is a close correspondence between these overdensities and those identified in velocity space. The boxes shown here mark the clumps that are easy to identify in the "integrals of motion" space, while the remaining colored clumps have been identified in velocity space. All of the structures have been highlighted using the same color coding as in Figure 2.

To establish the significance level of these structures we randomize the whole halo data set by reshuffling the velocities of the stars, while keeping their spatial distribution. We make 10.000 such random realizations and recompute for each the distribution in "integrals of motion" space. We find that in none of these realizations is a substructure apparent that has a similar extent and location as any of the overdensities identified in the various figures.

The stars that are part of the large "blob/plume" identified in Figure 3 (inside of the gray box in the top panel of Figure 4) define a blue sequence in the HR diagram of kinematically selected halo stars, as indicated by the red dots in the large panel in Figure 5. The absolute magnitude given here is calculated using the parallax from Gaia and has not been corrected for extinction. This sequence coincides with that reported by Gaia Collaboration et al. (2018a). These authors estimated an age of 13 Gyr and a metallicity [M/H] ∼ −1.3 dex for this population, obtained by comparison to the blue isochrone shown here and obtained from Marigo et al. (2017), considering an α-enhancement of 0.23 (Salaris et al. 1993). The redder stars are fitted better by an isochrone with [M/H] ∼ −0.5 dex and age of 11 Gyr (in black in Figure 5), as already shown in Gaia Collaboration et al. (2018a).

Zoom In Zoom Out Reset image size

The stars associated with the blue sequence (and the corresponding isochrone) are more metal rich than the "outer halo" component of Carollo et al. (2007), which has [Fe/H] ∼ −2.2 dex. This could imply that this region of phase-space is more complex than anticipated (see also Nissen & Schuster 2010). On the other hand, if this overdensity is due to the merger of a single large progenitor system, this could have had a metallicity gradient. This would imply that stars at larger distances could be more metal poor on average (as seen also for the streams of the Sagittarius dwarf; Bellazzini et al. 2006), thus explaining the findings of Carollo et al. (2007).

The HR diagrams for the stars belonging to the other substructures are plotted as small panels in Figure 5. We do not attempt to fit isochrones to their distribution because of the relatively low number of member stars (ranging from 12 to 37 stars). Note, however, that their HR diagrams are very coherent, and suggest low metallicities and old ages, similar to those of the large "plume" shown in the large panel of the figure. Exploration of the full Gaia DR2 should disclose additional members. These could be either more distant giants stars or fainter nearby dwarfs. These dwarf stars would not have full phase-space information in Gaia DR2 (because of the magnitude cut of the spectroscopic sample; Katz et al. 2018), but should have accurate proper motions and parallaxes, and be present in large numbers.