A MEGACAM SURVEY OF OUTER HALO SATELLITES. IV. TWO FOREGROUND POPULATIONS POSSIBLY ASSOCIATED WITH THE MONOCEROS SUBSTRUCTURE IN THE DIRECTION OF NGC 2419 AND KOPOSOV 2^*

Around half of the original targets of our primary survey were GCs. Interestingly, NGC 2419, Kop 2, NGC 7006, and Eridanus lie in the predicted direction of the halo substructure known as the Monoceros Ring according to the Peñarrubia et al. (2005) model (Figure 1) and their presence should be revealed by a foreground main sequence (MS) in the respective color–magnitude diagrams (CMDs). Figure 2 shows the CMDs containing both stars within a 5' radius area and those that are further away from the center of those clusters. Clear MSs are revealed in the CMDs for the surrounding areas of NGC 2419 and Kop 2 in the ranges of $19\lt g\lt 24$ and $0.3\lt g-r\lt 1.3$ (right panels). As is clear from the CMDs, the morphology of these MSs differ from the observed features associated with the GC populations (left panels). These diagrams confirm that NGC 2419 and Kop 2 lie in the same line of sight (LOS) as the vast stellar halo substructure that we identify as the Monoceros Ring. In the case of Eridanus, it is not possible to identify any features that might be associated with a population that are different from those associated with the Milky Way components. Eridanus might still be surrounded by the low surface-brightness region of Monoceros, but the complete absence of such a substructure is also a possibility. As for NGC 7006, the CMD suggests the presence of an overdensity in its $g\gt 21$ section. That subjacent MS might be associated, as proposed by Carballo-Bello et al. (2014), to the presence of stars belonging to the Hercules-Aquila cloud (Belokurov et al. 2007; Simion et al. 2014) along the LOS to the cluster. Alternatively, a higher density of halo stars in that direction of the sky, as suggested by the results of Deason et al. (2014), might produce a broader halo MS as the one observed in our data. In this work, we will focus on the well-defined foreground populations around NGC 2419 and Kop 2 to explore Monoceros.

Zoom In Zoom Out Reset image size

Since this article does not focus on the clusters themselves, we minimize the number of GC stars in the resulting diagrams, and therefore, while making the CMDs to detect and study Monoceros, we have included only those stars that are away from the cluster's center. The King tidal radii of NGC 2419 and Kop 2 are ${{r}_{{\rm t}}}=7\buildrel{\,\prime}\over{.} 5$ and $0\buildrel{\,\prime}\over{.} 8$, respectively (R. R. Muñoz et al. 2015, in preparation), though this parameter does not necessarily indicate the region beyond which cluster stars are no longer present (see discussion in Carballo-Bello et al. 2012). Thus, to work with the cleanest possible CMDs, we select stars beyond $r=10^{\prime}$ and 3' from NGC 2419 and Kop 2's centers, respectively.

The (a) and (b) panels of Figure 3 show the resulting CMDs for the area surrounding NGC 2419 and Kop 2, respectively. The observed MS widths indicate the detection of a stellar structure in a narrow distance range along the LOS. We note that the foreground sequences seem to extend blueward of the disk star's turnoff (TO) at $0.20\lt g-r\lt 0.35$ and $18\lt g\lt 19$, with a much larger separation, with respect to the tentative Monoceros TO, than the photometric error at this level ($|\delta g|\sim 0.02$). This feature is observed in both fields, though it is slightly clearer in the NGC 2419 data. To investigate whether this blue population (herein BP) corresponds to the main Monoceros population, we have visually fitted a theoretical isochrone (Dotter et al. 2008) corresponding to the nominal Monoceros age, determined to be ∼9 Gyr by Sollima et al. (2011) and with a metallicity of [Fe/H] $=-1$ (Ivezić et al. 2008; Conn et al. 2012; Meisner et al. 2012). The adopted E(B−V) values in the direction of NGC 2419 and Kop 2 are 0.035 and 0.037 mag, respectively (Schlafly & Finkbeiner 2011). As shown in the (c) panel of Figure 3, this isochrone reproduces the morphology of the foreground MS reasonably well but fails to cover them all the way up to the blue end. By using the region of the isochrone that matches the observed MSs we obtained a radial distance of ${{d}_{\odot }}{\mkern 1mu} =10.2\pm 1.6$ and 10.0 ± 1.5 kpc for the underlying system in the surroundings of NGC 2419 and Kop 2, respectively. These estimates are consistent with the predictions made by the Peñarrubia et al. (2005) model for the Monoceros Ring in that LOS and with the heliocentric distance derived by Li et al. (2012) for nearby fields on that structure. From the projected position of the clusters in Figure 14 of Li et al. (2012), it is possible to rule out the so-called anticenter stream discovered by Grillmair (2006) as the subjacent population in the fore/background of these clusters.

Zoom In Zoom Out Reset image size

To assess the possibility that these blue extensions of the Monoceros MSs do not represent an actual Monoceros population, but instead correspond to statistical fluctuations of Milky Way stars in this part of the CMD, we have explored an expanded area around the clusters using SDSS data. The right panel in Figure 3 shows all of the objects within a radius of 60' in five fields in the coordinate range $(\ell ,b)={{(182,25.35)}^{{}^\circ }}$–${{(190,25.35)}^{{}^\circ }}$, which are located between NGC 2419 and Kop 2. Though SDSS photometry is shallower than our CFHT data, the Monoceros MS is clearly visible. The blue extension observed in the CFHT data is also clear in this CMD.

To establish whether the origin of the BP is Monoceros or Milky Way stars, we study the variation of the BP star counts with Galactic longitude and compare it to those of a bona fide Milky Way and Monoceros population. To this end, we have defined three regions in the CMD that should predominantly  include disk, old Monoceros TO, and BP stars, respectively (see upper panel in Figure 4). We then counted the number of stars in those three regions for 11 SDSS circular fields of 60' of radius between the coordinates $(l,b)={{(180,25)}^{{}^\circ }}$ and $(\ell ,b)={{(230,25)}^{{}^\circ }}$, equidistantly spaced every $5{}^\circ$ in ℓ. In the bottom panel of Figure 4, we show the gradients in the star counts. To facilitate the comparison between the different boxes, the sequences have been normalized by their value at $\ell =180{}^\circ$. The number of disk stars remains nearly constant for the range of ℓ considered while TO and BP counts show a similar behavior. The observed increase in the counts is consistent with the Peñarrubia et al. (2005) model. This result supports the interpretation that the area in the CMD denoted as BP is not populated by stars associated with the Milky Way disk but by a population possibly associated with Monoceros.

Zoom In Zoom Out Reset image size

In order to show the morphology of these populations in the CMD with higher contrast, we have compared the SDSS CMDs for those positions that, according to the bottom panel of Figure 4, present a larger difference in density of stars belonging to the ring. In the left panel of Figure 5, we show the Hess diagram corresponding to the CMD, obtained at $(\ell ,b)={{(210,25)}^{{}^\circ }}$, when the Milky Way components are removed by subtraction of the CMD corresponding to $(\ell ,b)={{(180,25)}^{{}^\circ }}$. The contour of minimal density (lighter gray) indicates the bins with 4% of the highest density observed. The stellar overdensity above the traditional Monoceros TO has a lower significance with respect to the main ring population, but it clearly confirms the existence of an extra component brighter than the Monoceros TO. The right panel of Figure 5 shows the cumulative luminosity function constructed from the Hess diagram for the color range $0.3\lt g-r\lt 0.5$. The change of slope in the luminosity function at $g\sim 19.5$ is a possible hint that we are detecting the TO of the older (∼9 Gyr) population, supporting the interpretation that there are multiple populations present in these fields, though we regard this result with caution.

Zoom In Zoom Out Reset image size

Since it is expected that the regions of the CMDs where Monoceros lies are also populated by Milky Way stars, we estimate the contribution of stars belonging to Galactic components by comparing the observed diagrams with synthetic CMDs generated with the Milky Way photometric models TRILEGAL and Besançon (Robin et al. 2003; Girardi et al. 2005; Vanhollebeke et al. 2009) for the same LOS to each cluster and for a similar solid angle. We have used the set of optimized parameters provided by Gao et al. (2013) as inputs for TRILEGAL, while default values were used for Besançon.

We defined a narrow color range of $0.30\lt g-r\lt 0.45$ in the CMD to be analyzed and compared with the TRILEGAL and Besançon counts (Figure 6). We counted the number of stars in each of the 15 bins in which the $16.5\lt g\lt 23$ range was divided, and we compared this number with that obtained for the same color–magnitude bin in the synthetic CMDs.⁸ In the middle panel of Figure 6, we show the fraction ${{N}_{{\rm obs}}}/{{N}_{{\rm model}}}$ as a function of g magnitude, where ${{N}_{{\rm obs}}}$ and ${{N}_{{\rm model}}}$ are the counts obtained for the observed and synthetic CMDs, respectively. With the exception of the range of $18\lt g\lt 21$, where the presence of Monoceros MS stars is visually dominant, the models considered here adequately reproduce the observed distribution for both the disk and halo stars. The area in the CMD where the contribution of Monoceros stars stands out from the expected stellar counts includes the area defined in this work as BP; however, the significance is lower in comparison with Besançon. Similar results are obtained when we use the surroundings of Kop 2 to carry out these tests.

Zoom In Zoom Out Reset image size

We have determined that the BP corresponds to an actual overdensity and a differentiated population with respect to the Milky Way components. In order to characterize the BP, we used the same family of theoretical isochrones by Dotter et al. (2008), assuming different combinations of age and [Fe/H]. We considered a grid of possible isochrones with ages between 5–10 Gyr and $-1.6\;\lt$ [Fe/H] $\lt -0.8$ to find the best matches to Monoceros (including the BP region). Reasonable matches are obtained within a range of ages and metallicities (see Figure 7), resulting in distances in the range of $9\lt {{d}_{\odot }}\lt 12$ kpc. However, the combination that yields a heliocentric distance similar to the one previously derived for Monoceros is an isochrone corresponding to $t\sim 5$ Gyr and [Fe/H] $\sim -0.95$ (Figure 8). This shows that Monoceros might be composed of, at least, two stellar populations: a dominant component of ∼9 Gyr and a second contribution of ∼4 Gyr younger. Note that our photometry, where the faintest part of the MS is clearly defined, allows us to discard other possible solutions that would only match the upper MS, a limitation when using shallower data as those from SDSS (see Figure 3).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The presence of multiple stellar populations in streams is a natural consequence of the complex properties of their progenitor galaxies. In the case of the Sgr dwarf spheroidal, the presence of old-, intermediate-, and young-aged star formation epochs has been clearly established (e.g., Fahlman et al. 1996; Marconi et al. 1998; Bonifacio et al. 2004; Bellazzini et al. 2006; Siegel et al 2007). In the case of Monoceros, the presence of M giant stars in the structure around the anticenter (Rocha-Pinto et al. 2003) and the metallicity values reported along different LOSs suggested that possibility. Our detection of a possible younger population in that same region of Monoceros might help establish whether the Monoceros Ring is a complex halo substructure generated by the accretion of a minor satellite or by the distortion of the Galactic disk.

From Figure 7, we conclude that the BP might also be reproduced assuming the isochrone corresponding to a population with a similar age as that of the main Monoceros population but with a lower metallicity of [Fe/H] $\sim -1.5$ (see Figure 8). This result agrees with the mean metallicity value obtained by Yanny et al. (2003) and the derived distance (${{d}_{\odot }}\sim 9$ kpc) is consistent within errors with the heliocentric distance for the main Monoceros MS population in that LOS. Therefore, the presence of the BP component in the obtained CMDs might be produced by the manifestation of a metallicity spread suggested by the wide range of values reported for Monoceros in the literature. Alternatively, it is also possible that only a 5 Gyr stellar population is present along this LOS. However, the multiple detections of Monoceros in other areas of the sky have provided an age of $t\sim 9$ Gyr; therefore, this scenario might be difficult to reconcile with previous work focused on the ring.

It would be interesting to confirm the presence in other sections of the Monoceros BP using deep wide-field photometry. Even the possible non-detection of hypothetical younger stars in other sections of Monoceros might help us better constrain the location of the progenitor system of this halo substructure, taking advantage of the fact that the different stellar populations are expected to leave the satellite main body at different times (Peñarrubia et al. 2008; Walker 2011).

Since the BP is located blueward and above the TO position of Monoceros, it populates the area of the CMD where blue straggler stars are expected. Recently, Santana et al. (2013) have shown that blue stragglers are ubiquitous among Galactic GCs, classical dwarf spheroidal, and ultra-faint dwarf galaxies. Their analysis shows that in order to reproduce the observed number of blue stragglers in dwarf galaxies, this population should be composed of stars with ages of  $t\sim 2.5$ Gyr and account for up to the unlikely fine-tuned fraction of 7% of the total number of stars in the satellite. Despite the fact that the distance uncertainties and depth along the LOS of Monoceros do not allow us to estimate a precise age for the BP, we do not see a significant population of stars extending as far as the MSTO of a 2.5 Gyr old population given by the Dotter et al. (2008) theoretical isochrones. That level is estimated assuming that the brightest members of a blue straggler population should reach a magnitude of ${{g}_{{\rm TO}}}\sim 16.3$, corresponding to a stellar mass that is twice the mass of a Monoceros MSTO star, at least ∼2 magnitudes brighter than the BP MSTO.

In addition, from our data, we can estimate the specific fraction of blue straggler stars if we assume that the BP is populated entirely by them. Following the procedure described in previous work (e.g., Sollima et al. 2008; Santana et al. 2013), we selected all of the stars in the BP and along the Monoceros MS in the range of $20.5\lt g\lt 21.5$. After the decontamination of the observed stellar counts, using the synthetic CMD generated with the Besançon model for Section 3.3 as a reference, we conclude that a BP composed of blue stragglers would imply a specific fraction of $F=20\%$, an order of magnitude larger than the observed fraction of blue stragglers in GCs and dwarf galaxies. Therefore, we consider the possibility of the BP being made up entirely of blue straggler stars as unlikely.

Thus far, we have shown that the BP is likely not associated with a Milky Way disk population and that it is consistent with being part of Monoceros. However, it is still possible that the BP corresponds to a population different from both the Milky Way and Monoceros or to a different wrap of that halo substructure. Given the appearance of the BP in the CMDs, a possibility is that it corresponds to a feature along the LOS between us and Monoceros. If this was the case, we should be able to detect its presence at fainter magnitudes and redder colors, slightly above the well-defined Monoceros MS. To check this scenario, the same comparison with the Milky Way models as in Section 3.3 is performed, but this time in the color range of  $0.6\lt g-r\lt 0.75$. The right panel in Figure 6 shows the results.

In this case, both models reproduce the observed counts with the exception of the $g\gt 21$ range, where we find about four times more stars that are presumably associated with Monoceros than in the synthetic CMD generated with Besançon. The presence of more than one subjacent MS in the redder region of the CMD is not obvious, thus there is not significant evidence in our data of a foreground stream possibly associated with the BP feature.

According to the Peñarrubia et al. (2005) model (Figure 1), two different wraps of Monoceros might be present at different distances along the direction to NGC 2419 and Kop 2 and with indiscernible velocities (see the discussion below). Therefore, our results are compatible with the detection of a different wrap of the ring in that area of the sky; though, the metallicity of such a component would differ from that of the main population of Monoceros observed in these fields (Figure 7). This might support the scenario in which the generation of the BP is related to the hypothetical Monoceros metallicity spread.

To verify our interpretation that the BP represents a younger sub-population at the same distance as the Monoceros structure, we turn to spectroscopy from SDSS DR10 (Ahn et al. 2014). We select 160 stars around the position of NGC 2419 with colors of $0.2\lt g-r\lt 0.4$ in the magnitude ranges of $18.0\lt g\lt 19.0$ for the BP population and 131 stars with $0.25\lt g-r\lt 0.45$ and $19.25\lt g\lt 20.0$ for the TO population (see Figure 9). For comparison, we also generate Besançon models in the same field of view and select stars with identical color–magnitude criteria. Figure 10 shows the resulting velocities (left) and metallicities (right) for the BP (upper panels) and TO (lower) in this region. Model predictions are given as blue histograms. A clear excess population is seen in both the BP and TO populations at ${{V}_{{\rm gsr}}}\sim 20$ km s⁻¹. The velocity dispersion of this structure is clearly much less than the thick disk (which constitutes most of the Besancon stars in the BP selection box), confirming the presence of a kinematically cold structure in both the BP and TO samples, with a velocity compatible with that of the Monoceros Ring in that LOS, as predicted by the Peñarrubia et al. (2005) model (see lower panel in Figure 1). These data are selected in the same sky area as fields denoted "M3" and "M4" by Li et al. (2012), who found a narrow excess at ${{V}_{{\rm gsr}}}\sim 30$ km s⁻¹ in field M3, which is likely the same population that we see in Figure 10.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the right panels of Figure 10, we show the metallicities of stars from each population with $0\lt {{V}_{{\rm gsr}}}\lt 50$ km s⁻¹ to emphasize the velocity substructure. The mean metallicities of the BP and TO populations in this velocity range are similar, confirming our finding based on isochrone fitting in Section 3.4.