Following the initial discovery of tidal features in the M31 halo, concerted follow-up observations and detailed modelling have been carried out in order to develop a complete understanding of this material. In this section, we review and summarise some highlights from this work.
Over the last decade, the GSS has been the subject of intense study. On the observational side, efforts have concentrated on deriving quantities (e.g. distance, velocity) that can be used to model the orbit of the progenitor and on stellar populations constraints that can be used to establish its nature. On the theoretical side, work has focused on reconstructing the orbital history of the progenitor and using this knowledge to measure the halo mass of M31.
Line-of-sight distances to the stream are one of the key inputs for GSS orbit models. McConnachie et al. (2003) used measurements of the TRGB in a series of CFHT12K pointings to show that the GSS lies ≳ 100 kpc behind M31 at a projected radial distance of 60 kpc and moves progressively closer to galaxy at smaller radii, with their distances being indistinguishable (within the uncertainties) at ≤ 10 kpc. These authors were also able to detect the stream in two fields on the northern side of M31, where it lies ∼ 40 kpc in front of the galaxy, but no further, suggesting that the stream wraps fairly tightly around the M31 center towards the north-east.
Another key observable is the radial velocity distribution of stream stars. Individual RGB stars at the distance of M31 are sufficiently faint that an 8-m class telescope is required in order to measure their line-of-sight velocities; the DEIMOS multi-object spectrograph on the Keck II telescope has been the source of almost all the GSS radial velocity measurements to date (Ibata et al. 2004, Guhathakurta et al. 2006, Kalirai et al. 2006, Gilbert et al. 2009). These studies have shown that the radial velocity of the GSS becomes increasingly positive with increasing distance from the center of M31, ranging from vhelio = −320 km s−1 at a projected distance of 60 kpc to vhelio = −524 km s−1 at a projected distance of 17 kpc. In all fields studied so far, the velocity dispersion is quite narrow – in the range of 10−30 km s−1. Intriguingly, a second cold kinematic component has been detected at several locations along the GSS (Kalirai et al. 2006, Gilbert et al. 2009). It has the same velocity gradient (and dispersion) as the primary GSS over the range in which the two have been mapped (∼ 7 kpc) but it has a radial velocity that is offset by ∼ +100 km s−1. It is presently unclear whether this component is due to M31's disturbed disk or a forward wrap or bifurcation of the main stream.
The combination of line-of-sight distance and radial velocity measurements indicates that the GSS progenitor fell almost straight into M31 from behind. These data have motivated a variety of efforts to model the accretion of a dwarf satellite galaxy on a highly radial orbit (Ibata et al. 2004, Font et al. 2006, Fardal et al. 2006, 2008, 2013, Mori & Rich 2008, Sadoun et al. (2014). While these models differ in various aspects, they generally agree on the fact that the progenitor's initial stellar mass was in the range 1−5 × 109 M⊙ and that its first pericentric passage came within a few kpc of the M31 center less than 1-2 Gyr ago. Some properties of the observed GSS, in particular the asymmetric distribution of stars along the stream cross section and the internal population gradient, are better reproduced in N-body models in which the progenitor possessed a rotating disk (e.g. Fardal et al. 2008, Sadoun et al. 2014).
In this near head-on collision, the progenitor experiences significant destruction at the first pericentric passage. Much of the satellite's mass is stripped off to form leading and trailing tidal streams, and a generic prediction is that much of M31's inner halo should be littered with this debris. Fig. 5 shows the sky distribution of debris in a set of recent N-body models of the GSS's orbit within the M31 potential (Fardal et al. 2013). Each panel differs in the adopted values of M200, the M31 mass inside a sphere containing an average density 200 times the closure density of the universe, and Fp, the orbital phase of the progenitor at the present day. Although the exact pattern of debris depends on these (and a few other) parameters, all panels exhibit a similar morphology and show remarkable consistency with some of the features seen in Figs. 2 and 3. The GSS, the NE Shelf and the W Shelf are all naturally reproduced in this scenario, with the GSS representing the trailing stream of material torn off in the progenitor's first pericentric passage while the NE and W shelf regions contain material torn off in the second and third passages, respectively. These predictions are in excellent agreement with observations of the stellar populations of these substructures – which show striking similarities in their star formation histories (SFHs) and metallicity distributions to those of the GSS (e.g. Ferguson et al. 2005, Richardson et al. 2008) – as well as with the observed kinematics of the shelves (e.g. Gilbert et al. 2007, Fardal et al. 2012). The extensive pollution of M31's inner halo by material stripped from the GSS progenitor provides at least a partial explanation for why this region is dominated by more metal-rich stellar populations compared to the Milky Way (e.g. Mould & Kristian 1986, Ibata et al. 2014).
 |
Figure 5. Spatial distribution of stellar debris from the tidal disruption of the GSS progenitor, as predicted by a set of 9 N-body simulations. The dashed lines show the observed boundaries of the NE and W shelves. In each case, different combinations of M200, the virial mass of M31, and Fp, the orbital phase of the progenitor at the present epoch, are adopted. While all panels share similar morphologies and resemble the main features seen in Figs. 2 and 3, some systematic differences can be seen as a function of the parameters. The present-day state of the progenitor varies from being tightly bound to highly dispersed; in all models shown here, it lies in the region of the NE shelf. Reproduced from Fardal et al. (2013). |
Based on the projected positional alignment with the GSS, M31's luminous dwarf elliptical satellites M32 and NGC 205 were initially considered as prime candidates for the progenitor (Ibata et al. 2001, Ferguson et al. 2002). Indeed, both these systems possess unusual properties and, as will be discussed in Section 7, are tidally distorted in their outer regions (Choi et al. 2002, Ferguson et al. 2002). However, even the earliest attempts at orbit models ruled out a straightforward connection between the GSS progenitor and either of these two satellites (Ibata et al. 2004) 2. Current models agree that any existing remnant should lie in the region of the NE Shelf, although thus far no candidate has been identified.
Although the location of the GSS progenitor remains a mystery, analysis of the stellar populations in the stream has provided important insight into its nature. Fig. 6 shows deep HST CMDs of a variety of tidal debris fields in the inner halo of M31, including those associated with the GSS. One can clearly see two predominant CMD morphologies – those which contain a narrow tilted red clump and a prominent horizontal branch extending quite far to the blue (labelled ‵SL' in Fig. 6), and those which contain a rounder red clump, a well-populated blue plume and exhibit no horizontal branch (labelled ‵DL' in Fig. 6). Those pointings which directly sample material associated with the GSS progenitor uniformly exhibit the morphology of the former type (Ferguson et al. 2005, Richardson et al. 2008). A third category, labelled ‵C', appears to be a composite of the previous morphologies.
 |
Figure 6. HST/ACS Hess diagrams of 14 fields in the inner halo of M31, many of which lie on substructures seen in Fig. 2. The ridge line of 47 Tuc ([Fe/H] = -0.7 and age 12.5 Gyr) has been shifted to the distance of M31 and overlaid in each case. Based on their CMD morphologies, the fields can be segregated into two classes; stream-like (SL) fields with tilted red clumps and extended horizontal branches and disk-like (DL) fields with round red clumps and a population of blue plume stars. Composite (C) fields have features in common with both and likely represent cases where different material is projected along the line-of-sight. Reproduced from Richardson et al. (2008). |
Quantitative measures of the SFH and age-metallicity relation (AMR) of the GSS have been derived from deep CMDs using synthetic modelling techniques (Brown et al. 2006, Bernard et al. 2015a). The bottom panels of Fig. 7 show the most extensive analysis carried out to date, based on the 5 fields lying on GSS debris located throughout the inner halo. It appears that star formation in the progenitor got underway early on and at a fairly vigorous pace, peaking 8–9 Gyr ago. Star formation remained active until about 6 Gyr ago when there was a very rapid decline; this ‵quenching' may indicate the time when the progenitor first entered the halo of M31. Roughly 50% of the stellar mass in the GSS fields was in place ∼ 9 Gyr ago, and the metallicity had reached the solar value by 5 Gyr ago, consistent with direct spectroscopy constraints from RGB stars (e.g. Guhathakurta et al. 2006, Kalirai et al. 2006, Gilbert et al. 2009). In addition, all the GSS fields probed thus far reveal a large spread in metallicity (≥ 1.5 dex). Taken together, these properties suggest an early-type progenitor, such as a dwarf elliptical galaxy or spiral bulge. This further supports inferences from N-body modelling which suggest that the GSS progenitor was a fairly massive object with some degree of rotational support. Indeed, Fardal et al. (2013) argue it was likely the fourth or fifth most massive Local Group galaxy as recently as 1 Gyr ago.
 |
Figure 7. (Left) The star formation histories (SFHs) and (Right) age-metallicity relations (AMRs) of M31 inner halo substructure fields, derived from quantitative fitting of the CMDs in Fig. 6. The SFHs for the individual fields are shown in light grey and are normalised to the total mass of stars formed in each field. Overlaid in bold are the average behaviours of the normalised SFHs. The filled circles in the AMRs show the median metallicity in each age bin, with grey circles indicating those bins which contain ≤ 1% of the total stellar mass and hence carry significant uncertainties. There is more star formation at early (late) times in the stream (disk) fields and the stream fields also exhibit a more rapid early chemical evolution. |
5.2. Other Inner Halo Substructure
While tidal debris from the GSS progenitor can explain the origin of some of the M31 inner halo substructure, other features require a different origin. In particular, the substructures lying near the major axis – such as the NE and G1 Clumps and the Claw – do not arise naturally in models of the dissolution of the stream progenitor.
A first step to understanding this low latitude substructure was taken by Ibata et al. (2005) who analysed the kinematics of stars in numerous fields around M31, including the G1 and NE Clumps. They noted a strong signature of rotation in almost all fields out to a galactocentric radius of ∼ 40 kpc, with some further detections out to 70 kpc. Stars are observed to move with velocities close to those expected for circular orbits in the plane of the M31 disk and with a typical velocity dispersion of 30 km s−1. Based on stacked spectra, Ibata et al. (2005) estimated the mean metallicity of these rotating outer populations to be [Fe/H] ∼ −0.9.
The irregular morphology yet coherent rotation observed in the outer disk regions led Ibata et al. (2005) to speculate that a vast disk-like structure was being assembled as a result of multiple accretion events. However, this interpretation faced a number of challenges, such as the homogeneity of the constituent stellar populations and the need for the accreted satellites to be sufficiently massive so that dynamical friction could circularise their orbits before disruption.
An alternative explanation is that much of the inner halo substructure is material that has either been torn off of the M31 outer disk or dynamically heated from the disk into the halo. It has long been known that the accretion of a low mass companion can have rather a disruptive effect on a stellar disk (e.g. Quinn et al. 1993, Walker et al. 1996) and more recent work has quantified the way in which disk stars can get ejected into the halo through such events (e.g. Zolotov et al. 2009, Purcell et al. 2010). Kazantzidis et al. (2008) demonstrated how the accretion of a population of satellites with properties drawn from a cosmological simulation can produce distinctive morphological features in the host galaxy's disk, similar to the inner halo substructures seen in M31. Additionally, they confirmed earlier work that showed the final distribution of disk stars exhibits a complex vertical structure that can be decomposed into a thin and thick disk (see also Villalobos & Helmi 2008).
Examination of the constituent stellar populations in the non-GSS debris fields supports the idea that this material has originated in the disk (e.g. Ferguson et al. 2005, Brown et al. 2006, Richardson et al. 2008, Bernard et al. 2015a). The ‵DL' fields in Fig. 6 are rather homogeneous in appearance, all displaying a round red clump with significant luminosity width, a well-populated blue plume and no apparent horizontal branch – features that indicate continuous star formation and a moderately young mean age. The quantitative SFHs and AMRs of the ‵disk-like' debris fields further strengthen this assertion; roughly ∼ 65% of the stars formed in the last 8 Gyr and chemical evolution proceeded at a modest pace, starting from a pre-enriched level (see top panels of Fig. 7). Most importantly, these trends are strikingly similar to those that have been measured for populations in the M31 outer disk (Bernard et al. 2012, 2015b).
It is also notable that both the stream-like and disk-like fields in Fig. 7 show evidence for an enhancement in the rate of star formation roughly 2 Gyr ago. This is surprising given that the constituent stellar populations in these fields have very different origins, and that many of them are substantially displaced from the main body (and the gas disk) of M31. There is now strong evidence that the M31 outer disk underwent a burst of star formation around this epoch, likely triggered by the relatively close passage of M33 (Bernard et al. 2012, 2015b). The existence of trace populations from this episode scattered throughout the inner halo, including up to 20 kpc along the minor axis, argues for a redistribution of disk material in the intervening time. It would thus appear that, in addition to being heavily polluted by GSS debris, the M31 inner halo also contains a widespread component of heated disk stars (Bernard et al. 2015a). This idea was independently raised by Dorman et al. (2013) who find that a non-negligible fraction of the inner halo stars identified kinematically in M31 show a luminosity function consistent with an origin in the disk.
Both the highly structured nature of the outer disk and the presence of displaced disk stars in the halo could be explained by one or more violent accretion events. Given the likely transitory nature of the outer disk substructures (Ibata et al. 2005) and the fact that stars as young as 2 Gyr have been displaced into the halo, the event responsible for disrupting the disk must have been rather a recent one and it is tempting to speculate that it has been the head-on impact of the GSS progenitor roughly 1 Gyr ago (e.g. Mori & Rich 2008, Sadoun et al. 2014). If this scenario is correct, it implies that, in spite of its extremely messy appearance, all of M31's inner halo substructure can be traced to the direct and indirect effects of a single event.
The outer halo debris features in M31 are much more poorly understood than those of the inner halo. While the inner halo is populated by tidal debris with a variety of morphologies and moderate metallicities, the outer halo is dominated by fairly narrow stellar streams and arcs which are, with the exception of the metal-rich component of Stream C, only apparent in maps constructed from metal-poor stars (see Fig. 3). The outer halo streams are of such low surface brightness that detailed characterisation of their stellar populations and kinematics has thus far been difficult. As will be discussed in the following section, GCs offer a very exciting way to probe the tidal features in these parts.
There are the four main structures visible in the far outer halo – Stream A, the E Cloud, the SW Cloud and the NW Stream. All of these features lie at radii ≳ 100 kpc from the center of M31 and subtend at least a few tens of kpc in length. Their CMDs indicate a similar metallicity of [Fe/H] ∼ −1.3 (Ibata et al. 2007, Ibata et al. 2014, Carlberg et al. 2011, Bate et al. 2014). The most luminous of the outer halo structures is the SW Cloud which Bate et al. (2014) estimate contains ∼ 5.6 × 106 L⊙ or equivalently MV ≈ −12.1. This luminosity is approximately 75% of that expected for the feature on the basis of its measured metallicity and the Kirby et al. (2011) luminosity-metallicity relation. While this might indicate that a sizeable fraction of the luminosity of the parent object has been detected, it remains unclear at present whether these most distant halo features originate from distinct accretion events or material torn off from a single progenitor. Indeed, an interesting question is whether any of the outer halo debris can be traced to the accretion of the GSS progenitor. Although the metallicity of these features is considerably lower than that of the core of the GSS, it is a good match to that of the stream envelope (Ibata et al. 2007, Gilbert et al. 2009).
It is curious to note that the dwarf spheroidal satellite And XXVII appears projected on the upper segment of the NW Stream. Discovered by Richardson et al. (2011), this faint (MV ≈ −7.9) system is highly morphologically disturbed and it is tempting to speculate that it may be the source of the NW Stream debris. However, the metallicity of the stream stars appears somewhat higher than that of the dwarf galaxy which complicates the interpretation (Carlberg et al. 2011, Collins et al. 2013). Furthermore, Collins et al. (2013) note additional kinematic substructure in the vicinity of And XXVII which is not yet understood.
Ibata et al. (2014) have recently conducted a global analysis of the large-scale structure of the M31 halo using data from the PAndAS survey. Despite the presence of copious substructures throughout, they find that the stellar halo populations closely follow power-law profiles that become steeper with increasing metallicity. The smooth metal-poor halo component (defined as the population with [Fe/H] < −1.7 that cannot be resolved into spatially distinct substructures with PAndAS), has a global (3D) power-law slope of γ = −3.08 ± 0.07 and an almost spherical shape, but accounts for a mere ∼ 5% of the overall halo luminosity. By far, most of the luminosity of the halo out to the edge of the PAndAS survey resides in moderate-metallicity substructure. Ibata et al. (2014) estimate that the total stellar mass of the M31 halo at distances beyond 2∘ is ∼ 1.1 × 1010 M⊙ and that the mean metallicity decreases from [Fe/H] = −0.7 at R = 30 kpc to [Fe/H] = −1.5 at R = 150 kpc for the full sample. An alternative approach to studying the outer halo in M31 has been taken by the SPLASH team who have obtained spectroscopy of RGB stars in many pencil-beam fields extending out to 175 kpc (Guhathakurta et al. 2006, Kalirai et al. 2006, Gilbert et al. 2006, 2012, 2014). In contrast to photometric studies, their approach allows them to identify and remove kinematic substructure in their fields, at least out to 90 kpc. Although it is not possible to directly compare the results of the PAndAS and SPLASH surveys, which are based on very different sample selections, their inferences on the global halo properties of M31 appear to be largely consistent (Ibata et al. 2014, Gilbert et al. 2014).
2 Meanwhile, Block et al. (2006) argue that a head-on collision between M31 and M32 about 200 million years ago could be responsible for the formation of two off-center rings of ongoing star formation seen in the M31 disk. Back.