Next Contents Previous

6. GLOBULAR CLUSTERS AS PROBES OF TIDAL STREAMS IN M31

An intriguing question regarding the substructures seen in the halo of M31 is whether they show any degree of spatial correlation with members of the M31 GC system. This is motivated in part by the long-held suspicion that a substantial number of the GCs in the Milky Way halo did not form in situ, but rather in small satellite dwarf galaxies that subsequently fell into the Galactic potential well and disintegrated. This idea was first suggested in the seminal paper by Searle & Zinn (1978) and was spectacularly verified in the early 1990s with the discovery of the disrupting Sagittarius dwarf, which is in the process of depositing at least five GCs into the outer halo of the Milky Way (e.g. Bellazzini, Ferraro, & Ibata 2003, Law & Majewski 2010). Modern studies of the Galactic GC system have only served to add further weight to the assertion – it is now known that the abundances, velocities, ages, horizontal branch morphologies, and sizes of perhaps up to a third of Milky Way GCs are consistent with an external origin (e.g., Zinn 1993, Mackey & Gilmore 2004, Mackey & van den Bergh 2005, Marín-Franch et al. 2009, Forbes & Bridges 2010, Dotter et al. 2010, 2011).

Historically almost all work on the M31 GC system has been confined to regions comparatively close to the galactic center, typically within R ∼ 20−25 kpc; however the situation has changed thanks to the aforementioned wide-field mapping surveys of M31. In particular, the INT/WFC and PAndAS surveys, but also to some extent the SDSS, have facilitated the first detailed and uniform census of the outer halo GC system of M31 (Huxor et al. 2005, 2008, 2011, 2014, Martin et al. 2006, di Tullio Zinn & Zinn 2013, 2014). Remote clusters are abundant in M31; there are now more than 90 objects known to reside at projected radii beyond 25 kpc, 13 of which sit outside 100 kpc in projection, with the most the most distant at ∼ 140 kpc in projection and up to 200 kpc in three dimensions (Mackey et al. 2010a, 2013b). This is many more than is seen in the outskirts of the Milky Way – while the disparity in the number of GCs in the Milky Way and M31 within ∼ 25 kpc is roughly 3:1 in favour of M31, outside this radius it is more like 7:1 in favour of M31 (Huxor et al. 2014).

Figure 8 shows the positions of all known M31 GCs plotted on top of the PAndAS spatial density map of metal-poor RGB stars. In the outer parts of the halo, where large, coherent tidal debris streams are readily distinguished, there is a striking correlation between these features and the positions of a large fraction (∼ 50−80%) of the GCs. Substantial numbers of clusters are seen projected onto the NW Stream, the SW Cloud, the E Cloud, and the overlapping portion of Streams C and D. There is, in addition, a statistically significant overdensity of clusters (“Association 2”) sitting near the base of the NW Stream, that cannot (as of yet) be identified with a visible tidal stream in the field halo.

Figure 8

Figure 8. Map showing extant radial velocity measurements for M31 outer halo GCs, projected on top of the PAndAS metal-poor RB map. Most of the outer halo GCs can be seen to preferentially lie along stellar streams. Inner halo GCs from the Revised Bologna Catalogue are shown as grey points. The GCs are colour-coded by their radial velocity in the M31-centric frame (white points indicate those objects with no radial velocity measurement), and the inner and outer dashed circles correspond to radii of 30 and 100 kpc. A clear rotational signature is seen with GCs in the NE side of the galaxy receding while those in the SW quadrant approach us. Additionally, coherent velocities are seen for GCs which lie along specific debris features, strongly suggesting that the GCs have been brought into M31 along with their host galaxies.

Mackey et al. (2010b) have demonstrated that the probability of this global alignment between clusters and streams arising randomly is low – well below 1% for a GC system possessing an azimuthally uniform spatial distribution. This implies that the observed coincidence represents a genuine physical association and hence direct evidence that much of the outer M31 GC system has been assembled via accretion. Moreover, at least some of the properties of the accreted M31 GCs appear to be consistent with those exhibited by ostensibly accreted Galactic members – particular examples being those of younger ages (Mackey et al. 2013a) and extended structures (Huxor et al. 2011, Tanvir et al. 2012).

The argument made by Mackey et al. (2010b) is based entirely on statistical grounds; to determine on an object-by-object basis which GCs are associated (or not) with a given substructure requires kinematic information. The most extensive kinematic study of the M31 outer halo GC system to date is by Veljanoski et al. (2013, 2014), who acquired spectra for 71 clusters outside a projected radius of 30 kpc (representing 86% of the known population in these parts); the velocities of these objects in the M31-centric frame are color-coded in Fig. 8. It can be readily seen that GCs projected onto a given outer halo substructure tend to exhibit correlated velocities (see left panel of Fig. 9 for an example). Those objects on the NW Stream and SW Cloud reveal strong velocity gradients from one end of the substructure to the other, while clusters on the E Cloud form a close group in phase space. Members of the Stream C/D overlap area, and those in Association 2 split into additional sub-groups by velocity. A remarkable feature of many of the ensembles considered by Veljanoski et al. (2014) is the coldness of their kinematics, with all GC groupings exhibiting velocity dispersions consistent with zero given the individual measurement uncertainties. These results strongly reinforce the notion that a substantial fraction of the outer halo GC population of M31 has been accreted, and that these clusters trace the velocities of the tidal streams from their progenitor systems. Indeed, while definitive measurements have thus far only been possible for two substructures, the velocities of the GCs sitting on Stream C and on the SW Cloud have been shown to be in excellent agreement with those of the underlying stream stars (Collins et al. 2009, Mackey et al. 2014).

Fig. 8 also demonstrates the surprising result that the M31 GC system as a whole possesses bulk rotation – those GCs to the west of M31 appear to systematically possess negative velocities in the M31-centric frame, while those to the east typically have positive velocities. Veljanoski et al. (2014) compared a variety of kinematic models to the data and found a rotation amplitude of 86 ± 17 km s−1 around an axis aligned with the M31 optical minor axis provided the best match. This rotation velocity is quite substantial – for comparison, Veljanoski et al. (2014) also found evidence that the velocity dispersion in the cluster system decreases from 129−24+22 km s−1 at 30 kpc to roughly 75 km s−1 at 100 kpc. The right panel of Figure 9 further elucidates this rotation by showing the GC velocities in the M31-centric frame as a function of projected radius along the major axis. It is apparent that the rotation of the outer halo GC system is in the same sense as for the inner halo clusters (and indeed the M31 disk), albeit with smaller amplitude. Importantly, the rotation does not seem to be driven purely by clusters that sit on the particular halo substructures nor by those sitting well away from any underlying halo feature – both groups apparently share in the pattern equally.

Figure 9

Figure 9. (Left) Radial velocities for seven GCs that lie along the NW Stream. A clear signature of radial infall is evident. (Right) GC velocities, in the Galactocentric frame and corrected for the M31 systemic motion, versus distance along the M31 major axis. GCs which lie along specific debris features in the outer halo are colour-coded accordingly, while inner GC velocities taken from the literature are shown in light grey. The outer halo GCs rotate in the same sense as the inner GCs but with a somewhat smaller amplitude. The rotation is exhibited by the ensemble system of outer halo GCs and is not driven by specific stream features. Both panels are reproduced from Veljanoski et al. (2014).

Understanding the origin of the angular momentum in the outer halo GC system of M31 presents a significant challenge. One possibility is if a large fraction of the halo GC system was brought into M31 by just one relatively massive host galaxy on a low inclination orbit. However, in this scenario it is difficult to explain the observed presence of distinct dynamically cold subgroups of GCs as well as the typically narrow stellar debris streams in the halo. Another possibility is that the outer halo GC system results from the assimilation of several dwarf galaxies, but that these were accreted onto M31 from a preferred direction on the sky. This scenario might well be related to the recent discovery that many dwarf galaxies, both in the Milky Way and M31, appear to lie in thin rotating planar configurations such that their angular momenta are correlated (Ibata et al. 2013, Pawlowski, Pflamm-Altenburg, & Kroupa 2012). In this context it is relevant that almost all of the dwarf galaxies thought to be members of the planes presently observed in both M31 and the Milky Way are insufficiently massive to host GCs, and furthermore that the rotation axes of the M31 outer halo GC system and the M31 dwarf plane are misaligned with each other by ∼ 45.

Next Contents Previous