PAndAS IN THE MIST: THE STELLAR AND GASEOUS MASS WITHIN THE HALOS OF M31 AND M33

In ΛCDM cosmologies, galactic halos are built up over time through the accretion and cannibalization of smaller mass systems. Given the complexity of the dynamical interactions, these are studied in detail using computer simulations (e.g., Ibata & Lewis 1998; Fardal et al. 2008), although, due to the difficulties in modeling baryonic physics, these typically only consider the evolution of dark matter. However, these predict that the outer region of galactic halos should be dominated by extensive tidal streams, whereas the shorter dynamical timescales of the inner halo result in complete destruction and a smooth stellar distribution (e.g., Bullock & Johnston 2005; Cooper et al. 2010; Font et al. 2011).

The progenitors of substructure could possess a mix of stars and gas, generally more concentrated than the dark matter halos in which they reside. The precise composition of any infalling halo will depend upon its mass and evolutionary history, and we would expect dwarf galaxies to be stripped of their gas due to interactions within the Local Group, whereas larger galaxies will hold on to its gaseous component. While stars are effectively collisionless, more complex internal physics influences the evolution of gas, including internal shocking, cooling and collapse, star formation and its associated feedback, as well as ram-pressure stripping due to the presence of hot galactic halo gas (e.g., see Moore & Davis 1994; Murakami & Babul 1999; Bekki 2008; Mastropietro et al. 2009). This is quite apparent in the two major accretion events within the Milky Way halo, with the body and stream of the Sagittarius Dwarf Galaxy possessing no gas, having potentially lost it during its initial interaction with the Galaxy (Burton & Lockman 1999, but see also Putman et al. 2004), while the Magellanic Stream appears to be completely devoid of stars, consisting of stripped gas, although the question of whether this is the result of tidal or ram-pressure forces has yet to be decided (Stanimirović et al. 2008; Besla et al. 2010; Diaz & Bekki 2011). However, while we should expect different distributions for various disrupted components in M31 and M33, the complexities of ram-pressure stripping, which may require full magnetohydrodynamic approaches to simulate (Ruszkowski et al. 2012), make drawing robust conclusions difficult.

We are now in an era where large-scale surveys are providing a global picture of the formation and evolution of galaxies over cosmic time (e.g., see Abraham & van den Bergh 2001). Given their distances, unraveling the fine details of galactic evolution is below the resolution achievable with the vast majority of galaxies in the universe. This is unfortunate, as it is on small scales, with the structure within the halos of large galaxies, that have presented the major challenge to the prevailing ΛCDM paradigm (e.g., the missing satellite problem; Klypin et al. 1999; Moore et al. 1999).

Luckily, the large members of the Local Group, the Milky Way, and the Andromeda (M31) and Triangulum (M33) galaxies are close enough for such fine-scale detail to be resolvable, although their immense angular scale presents a significant challenge to the observability of their extensive halos. To this end, the last decade has seen the advent of large, detailed surveys of both the stellar and gaseous components of Local Group members, for the first time providing a panoramic view of galaxy formation in action; for the Milky Way, this includes surveys such as RAVE (Siebert et al. 2011), SEGUE (de Jong et al. 2010), GASS (McClure-Griffiths et al. 2009), and others.

In this paper, we present a detailed comparison of the stellar and gaseous matter in the halos of our nearest large companions in the Local Group, M31 and M33, using a new map of stellar structure obtained through the Pan-Andromeda Archaeological Survey (PAndAS) and the most detailed maps of H i, obtained during several telescope campaigns. In Section 2, we outline the observational campaigns that resulted in the data presented, with a summary of their key scientific results. In Section 3, we compare the spatial and kinematic distributions of gaseous and stellar substructure. We present the interpretation of these correlations in Section 4 and our conclusions in Section 5.

Some galaxies of the Local Group have been known since pre-history (e.g., see Section 3.1 of Geehan et al. 2006)¹⁵ and have played a pivotal role in our explorations of the workings of the universe. Reproducing a detailed account of this history is beyond the scope of this article, and the interested reader is directed to recent reviews (e.g., Tolstoy et al. 2009; Wyse 2010; Tolstoy 2011; McConnachie 2012). In the following we will focus upon the recent optical and radio surveys of M31 and M33, with a review of the observational programs and key scientific discoveries to date.

2.1. Stellar Observations

2.1.1. Observational Program

The PAndAS is a survey of 400 deg², covering the halo of M31 out to a distance of 150 kpc, and M33 to a corresponding distance of 50 kpc, undertaken as a large program on MegaCam,¹⁶ mounted on the 3.6 m Canada–France–Hawaii Telescope (CFHT). Integrations were sufficient to achieve photometric limits g = 25.5 and i = 24.5 at an S/N ∼ 10, reaching several magnitudes below the tip of the red giant branch (RGB) at the distance of M31/M33 (D ∼ 780–900 kpc; Conn et al. 2011).¹⁷ The photometric data taking was completed in early 2011, with the first published map, covering roughly half of the total observed area, presented in McConnachie et al. (2009), with the (almost) entire data set first presented in Richardson et al. (2011). The final map and high-level data products will be made publicly available in a forthcoming publication (McConnachie 2012; A. W. McConnachie et al. 2013, in preparation).

This study focuses upon RGB stars at the distance of M31 and M33, selected with cuts in color and magnitude (see Ibata et al. 2001, 2007; McConnachie et al. 2009). Figure 1 presents the distribution of RGB stars, overlain by a schematic map of the prominent stellar substructure; note that a nonlinear scaling has been applied to the RGB density, to bring out faint substructure, and contains a broad swath of metallicities to reveal metal-poor (− 3.0 < [Fe/H] < −1.7), intermediate (− 1.7 < [Fe/H] < −0.7), and metal-rich (− 0.7 < [Fe/H] < 0.0) substructures (cf. Ibata et al. 2007). The thick solid line is the entire PAndAS footprint, with stellar substructure as labeled thin solid lines. The dashed curve represents a significant overdensity of globular clusters identified as the NW Group by Mackey et al. (2010). The large dot-dashed line corresponds to a circle of radius of 150 kpc from the center of M31, whereas the smaller dot-dashed circle represents a distance of 50 kpc from the center of M33. It is very apparent that accompanying M31 and M33 is a wealth of substructure consisting of extensive streams and dwarf galaxies (these will be discussed in more detail in Section 2.1.2, and see Richardson et al. 2011).

Zoom In Zoom Out Reset image size

The stellar catalog derived from PAndAS is built upon earlier observations with the CFHT/MegaCam and CHFT/CFH12k (see McConnachie et al. 2003; Ibata et al. 2007). In parallel, a number of fields, targeting prominent substructure and dwarfs, were targeted with DEIMOS (Faber et al. 2003) on the 10 m Keck-II Telescope (e.g., Chapman et al. 2008; Collins et al. 2011). With a moderate resolution (R ∼ 6000), observations of 60–90 minutes around the prominent CaT absorption lines ∼8600 Å resulted in an S/N ∼ 5 for targets of i ∼ 21, with a corresponding velocity resolution of ${\rm \sim 5\hbox{--}10\,{\rm km}\,{\rm s}^{-1}}$; these will be described in more detail in Section 3.2.

2.1.2. Stellar Properties

Detailed analysis of the stellar substructure in PAndAS will be presented in R. Ibata et al. (2013, in preparation) and A. W. McConnachie et al. (2013, in preparation), so here we focus on the key features.

There is a wealth of substructure apparent in the outer disk/inner halo regions of M31, revealed in the earlier Wide-Field Camera on the 2.5 m Isaac Newton Telescope survey of the galaxy (Ibata et al. 2001; Richardson et al. 2008; see Figure 1), with the most significant structure being an extensive stellar stream, the Giant Stellar Stream (GSS), in the halo of M31. This is seen as a metal-rich substructure wrapped within a more metal-poor halo, and is apparently wrapped around the disk of M31 in the northeast (e.g., Ferguson et al. 2002; Ibata et al. 2007; Fardal et al. 2012). Accompanying the GSS is a series of streams, labeled B, C, and D, which lie perpendicular to the giant stream (Ibata et al. 2007), plus Stream A, a stellar overdensity ∼125 kpc (in projection) from M31. Other features include the South West (SW) Cloud and the North West (NW) Stream (McConnachie et al. 2009; Mackey et al. 2010). As revealed in Figure 1, and as shown by Richardson et al. (2011), this latter feature appears to loop back toward M31 as the North East (NE) Stream and intercepts the And XXVII dwarf galaxy; following Carlberg et al. (2011), we identify this entire feature as the NW Stream, and label the two components of it as NW1 and NW2. Finally, also apparent in this image is another significant overdensity of stars, named the Eastern (E) Cloud, located a distance of ∼125 kpc from M31.

As well as the extensive substructure, the PAndAS observations revealed the presence of a smooth stellar halo of M31 out to 150 kpc (Ibata et al. 2007; see also the SPLASH characterization of the smooth halo by Gilbert et al. 2012). Furthermore, the data reveal a wealth of dwarf galaxies (e.g., Martin et al. 2006, 2007; Richardson et al. 2011), globular clusters, and extended clusters (Huxor et al. 2008, 2011; Cockcroft et al. 2011; Tanvir et al. 2012); while these extended clusters (ECs) are more diffuse than normal globular clusters (Huxor et al. 2005), they appear to have very similar stellar populations (Mackey et al. 2006, 2007) and do not appear to be dominated by dark matter (Collins et al. 2009).

The structure around M33, apparent as a distorted outer disk, was originally presented by McConnachie et al. (2009, 2010). This structure is quite clearly asymmetric about M33, being significantly more prominent in the north than in the south.

The extension of the PAndAS survey toward the north of M31, in the region of the two satellite galaxies, NGC 147 and NGC 185, reveals the presence of stellar debris that appears to have been torn from NGC 147, as well as a new dwarf galaxy, Cass II (also known as And XXX), which is a potential satellite of the 147/185 subgroup, and will be discussed in more detail in a forthcoming publication (M. J. Irwin et al. 2013, in preparation).

2.2. H i Observations

2.2.1. Observational Program

The extended environment of M31 and M33 was observed by Braun & Thilker (2004) using the Westerbork Synthesis Radio Telescope (WSRT) as 14 single dish telescopes. A region of 60° × 30° in R.A.×decl. was imaged in the H i emission line at an effective resolution of 49 arcmin with an rms sensitivity corresponding to a column density of 5 × 10¹⁶ cm⁻² over ∼30 km s⁻¹. This unprecedented sensitivity permitted detection of H i in emission from column densities that have previously only been probed by Lyα absorption toward background QSOs. This low-resolution survey has been supplemented by higher resolution total power observations made with the Green Bank Telescope (GBT). A region covering 7° × 7° centered on M31 and 5° × 5° centered on M33 was observed with GBT during 2002 October with multiple perpendicularly scanned coverages yielding images with angular resolution as high as 9 arcmin. The M31 data have been presented previously by Thilker et al. (2004), while the M33 data were obtained with the same setup and have been reduced in a similar fashion. These moderate resolution data are supplemented with interferometric mosaic observations consisting of a 163 pointing WSRT coverage of M31 (Braun et al. 2009) and Very Large Array (VLA) coverage of M33 consisting of 6 pointings in the B and CS configurations and a 20 pointing mosaic in the D configuration. Some early M33 results based on the B and CS configuration data were presented in Thilker et al. (2002) while an independent reduction of all three VLA configurations has been presented by Gratier et al. (2010). The interferometric data provides resolution as high as 15 arcsec in M31 and as high as 5 arcsec in M33.

2.2.2. H i Properties

The H i distribution in the extended M31 and M33 environment is overlain as contours on the surface density of RGB stars in the PAndAS survey in Figure 2. The H i contours delineate the diffuse gaseous filament that connects M31 and M33 as well as other filamentary features extending both to the northwest of the M31 disk and to the southwest. Heliocentric radial velocities of various features are indicated with the annotations. The intrinsically diffuse nature of these features has been verified with follow-up observations using the GBT directed at the brightest positions along the M31–M33 filament (Braun & Thilker 2004). Despite employing a 25 times smaller beam area in these GBT observations, the features were detected at the same low column densities of only a few times 10¹⁷ cm⁻². This rules out the possibility that the filament is simply a collection of unresolved clumps in the discovery observations. Several discrete features are also seen in the northern portion of the field. It is noteworthy that the M31–M33 filament connects the systemic heliocentric velocities of M31 (∼ − 300 km s⁻¹) and M33 (∼ − 180 km s⁻¹).

Zoom In Zoom Out Reset image size

The more immediate H i environment of M31 is illustrated in Figure 3 where the GBT image (left panel) and WSRT mosaic (right panel) are similarly overlain on the PAndAS survey. At these higher angular resolutions, of 18 and 4 arcmin respectively, much of the diffuse gas is not detected but rather only the discrete High Velocity Cloud (HVC) features within about 100 kpc of the disk. Of note is that the radial velocity of discrete HVC features follows the basic pattern of disk rotation with the most negative velocities occurring in the southwest and most positive in the northeast.

Zoom In Zoom Out Reset image size

The immediate gaseous environment of M33 is illustrated in Figure 4 where the GBT image at 9 arcmin resolution (left panel) and the VLA mosaic with 2 arcmin resolution (right panel) are overlain on the PAndAS survey. M33 is not as rich in discrete HVC features as M31 but rather displays a very strong clumpy warp of the outer H i disk oriented toward the northwest and southeast, strongly suggestive of accretion fueling of the M33 disk by fall back from the M31–M33 filament.

Zoom In Zoom Out Reset image size

3.1. Spatial Correlations

3.2. Kinematic Correlations

What does the general lack of spatial and kinematic correlations between stellar substructure and H i tell us about the ongoing accretion in the M31–M33 system? Clearly, the axis connecting these two galaxies contains the majority of the substructure, but it appears that, in this case, the stellar and H i are the result of distinctly different accretion events; the main gaseous bridge through this region arises from the interaction between M31 and M33 alone, with any stellar material too sparsely distributed to be detected (Bekki 2008; McConnachie et al. 2009). Given M33's relatively large mass, and potentially large impact parameter, it has been able to retain the bulk of its stars and gas during the interaction with M31, although the interaction has distorted both the stellar and gaseous disk of M33.

The progenitor of the GSS, however, would have been a lower mass system that has been interacting with M31, and due to this interaction any progenitor gas was lost long ago, while the remnant stellar debris exists in the form of tidal streams and shell-like caustics. At first glance, the lack of significant gaseous structures corresponding to the GSS and other associated stellar features may seem surprising. However, as noted previously, while the gaseous and stellar components of an orbiting satellite are both subject to gravitational (tidal) forces, the gas is also subject to shocks and hydrodynamic drag. One possibility is that ram-pressure stripping led to nearly complete loss of gas from the GSS progenitor well before the formation of the observed GSS and associated structures. However, even if the ram pressure is not sufficiently efficient to completely strip the progenitor of its gas, hydrodynamic interactions can erase potential spatial and kinematic correlations between the two components. We are presently studying the gaseous and stellar streams using high-resolution numerical simulations (M. Shannon et al., in preparation) and preliminary results suggest that the two components can experience significant dislocation over the course of a single orbit, with the loss of angular momentum and energy due to hydrodynamic forces causing the gas to sink to the center of M31, unlike the much longer-lived stellar structures.

One intriguing correlation occurs where the GSS meets the disk of M31 (see Figure 3). While the gas is spatially offset by ∼15 kpc, its velocity of v_h ${\rm \sim -500\,{\rm km}\,{\rm s}^{-1}}$ is remarkably close to that of the GSS at this point. While they may be unrelated, it is worthwhile to examining if it is plausible that they are physically connected. Such a question is complicated by the fact that there appears to be very little gas associated with the GSS, consistent with a lack of recent star formation (Brown et al. 2006), but this could be solved if the enhancement of H i marks the (as yet unidentified) progenitor of the GSS, with the offset in the spatial location illustrating the action of ram-pressure stripping; such a conclusion would indicate that the accretion of the GSS is a relatively recent event and has not undergone several complete orbits, or by a progenitor massive enough to retain some of its gas through a prolonged accretion. This latter option is in agreement with the prediction of Fardal et al. (2008) who suggested that the source of the GSS was a rotating disk galaxy with a stellar mass of ∼10⁹ M_☉.

An alternate explanation also requires that the progenitor of the GSS be gas-rich, but that some gas remain orbiting with the stars. This gas would be too tenuous to be visible in the observations presented in this paper, but as the stream is funneled into the central regions, orbits converge and the local densities of both stars and gas increase. Given this density enhancement, the gas could become visible to our observations. Again, the offset in spatial location probably indicates ongoing ram pressure, although there may be a delicate balance between the amount of ongoing ram pressure and the dynamical shepherding by the associated stream of stripped dark matter, so that any gas can remain associated with the stream. Detailed simulations of the formation and evolution of the GSS, considering realistic gas physics, are required to address these questions.

Away from M31, there is a tantalizing correlation between H i and stellar material in the vicinity of the SW Cloud. There are two potential interpretations for this, one with the SW Cloud representing an outwardly moving agglomeration of shredded stars, left over from a close interaction with M31 ∼250 million years ago. This interaction would have to have been with a relatively pristine, gas-rich progenitor to leave the shredded debris with gas. On the other hand, the SW Cloud could represent the enhancement at the turning point of a stellar stream, and this stream would again have to be gas-rich to exhibit a similar enhancement in the H i. Both scenarios require a relatively recent first passage of M31, ensuring that the progenitor does not lose all its gas through ram-pressure stripping through an extended interaction. The estimated stellar and gas masses in the region of the SW Cloud are ∼10⁷ M_☉ and ∼5 × 10⁵ M_☉, respectively, suggesting that if this is representative of a recent accretion event, then the progenitor would have properties similar to the dwarf irregular IC10 (McConnachie 2012).

While we do not have kinematics for the stellar component of the SW Cloud, it does appear to be spatially correlated with three globular clusters (see Mackey et al. 2010) for which velocities have been recently determined (these will be presented in detail in forthcoming publications by Mackey et al. 2012 and J. Veljanoski et al. 2013, in preparation). If we take their velocities, of ∼ − 433 km s⁻¹ (PAndAS-7), ∼ − 411 km s⁻¹ (PAndAS-8), and ∼ − 363 km s⁻¹ (PAndAS-14), each with an error of less than ${\rm 10\,{\rm km}\,{\rm s}^{-1}}$, as being representative of that of the SW Cloud in general, we can compare this gaseous spur velocity of ${\rm \sim -470\,{\rm km}\,{\rm s}^{-1}}$, which is broadly consistent with the globular cluster velocity, especially if we expect a velocity gradient along any putative stream connecting the two components. It should be noted, however, that the width of the H i in the SW Spur is ${\rm \sim 200\,{\rm km}\,{\rm s}^{-1}}$, again indicating that we may be looking at both outbound and inbound gaseous streams, although until we have stellar kinematics for the SW Cloud and detailed dynamical modeling, the link between the SW Cloud and SW Spur remains circumstantial.

The distribution of gas within the M31–M33 system, especially the warped gas disk of M33 and the apparent gas bridge connecting M31 to M33, strongly suggests a past interaction of the two galaxies. It has been argued that the misalignment between the inner and outer gaseous disk of M33 and its interpretation as a warped disk could be the result of a recent tidal interaction (this is discussed in more detail below). The faint extension of the stellar disk with an alignment similar to the outer gas disk can be reproduced in interaction models (McConnachie et al. 2009). The gas bridge between the two galaxies has also been modeled in the context of a past interaction (e.g., Bekki 2008).

While the interaction hypothesis is a plausible explanation for the phenomenology of the gas in the M31–M33 system, it ultimately depends on knowing the orbit of M33. Recent measurements of both the proper motion of M33 (Brunthaler et al. 2005) and M31 (Sohn et al. 2012) provide strong constraints on the M33 orbit within assumed mass models of M31. We have been recently extending the dynamical models described in McConnachie et al. (2009) to describe the distorted outer stellar disk of M33 to a range of mass models for M31 and M33 using a Bayesian analysis to determine the distribution of orbits that are statistically consistent with the observed distance, radial velocities, and proper motions of the galaxies. This is a work in progress but we present some preliminary results here relevant to the observed gas distribution. In Figure 5, we present the distribution of orbits and the resulting line-of-sight velocity field consistent with the observations with the constraints that the pericenter for the M33 orbit is in the range of 30–60 kpc in a mass model of M31 with M = 2.5 × 10¹² M_☉ with r₂₀₀ = 280 kpc. The orbits are computed within the model potentials including a Chandrasekhar drag term to model dynamical friction. The drag coefficients are calibrated against live N-body simulations to ensure their applicability to the Bayesian analysis. We further constrain the distribution to lie within 20°of the plane containing the gas bridge with the implicit assumption that the bridge is created from tidally stripped gas from M33. We have also overlaid the stellar distribution from one N-body simulation including dark matter and stars without gas to orient the orbital distribution as well as illustrate the expected distortion of the outer disk of M33. We note that the complex stellar features seen around M33 in the simulation would be undetectable using current observations but we present an enhanced view to illustrate the complex dynamics. This model succeeds in two points with respect to the gas distribution. First, the model predicts that the outer disk of M33 will be warped and twisted as the result of an interaction while the inner bright disk will remain unscathed. Second, tidally stripped stars result in a tidal tail extending to the SE and a tidal bridge that falls back onto M31. While we have not modeled the gas in this simulation, we expect stripped gas to follow approximately ballistic trajectories similar to the stars. The heliocentric velocity of the distribution of orbits is consistent with the observed velocities of −240 and −280 km s⁻¹ seen in prominent gas features in the bridge. A tidal bridge of gas falling back onto M31 leading the position of M33 on its orbit after a past interaction is a plausible description of these features.

Zoom In Zoom Out Reset image size

Finally, we look more closely at the stellar and gaseous material in the outskirts of M33, with distinct features visible in both the NW and SE (see Figure 4). The position angle of the H i feature is offset from that of the disk by ∼30°–40° and has been interpreted as a warped disk in projection and modeled in terms of tilted rings by Corbelli & Schneider (1997). The stellar features lie roughly between the disk and H i position angles.

The origin of both the H i and stellar features has been attributed to tidal interactions with M31 during a close encounter (Putman et al. 2009; McConnachie et al. 2009; but see Reakes & Newton 1978 for an early discussion of this possibility), though models of a steady state precessing warp in a model with a flattened halo have also been considered (e.g., Kuijken 1991). Competing with tidal effects is the gravitational field of the M33 disk and dark halo. In addition, the H i disk experiences ram pressure due to its interaction with the gaseous halo of M31. We can distinguish between three regions: the outermost part of the disk where tidal interactions and ram pressure strip material from M33 leaving behind a stream, which roughly traces the M33 orbit; an intermediate region, where material is stripped from the M33 disk but remains bound to the galaxy; and a close-in region, where the disk is warped but material remains on roughly circular orbits.

In this simplified argument, we consider an element of the M33 gas disk at galactocentric radius R with surface area ΔΩ and surface density Σ_gas. The tidal force on this element due to M31 is

$$\begin{equation} F_{\rm tid} \simeq \Sigma _{\rm gas} \frac{V_{{\rm M}31}^2}{R_p^2}R \Delta \Omega2 , \end{equation} \tag{ 1 }$$

where R_p is the perigalactic distance between M31 and M33. In deriving this expression we have assumed that R_p ≫ R and that M31 has a flat rotation curve with a circular velocity V_M31. The ram-pressure force is

$$\begin{equation} F_{{\rm rp}} \simeq \rho _{\rm gas} V_{\rm rel}^2 \Delta \Omega , \end{equation} \tag{ 2 }$$

where ρ_gas is the density of the M31 gaseous halo. The gravitational restoring force due to the M33 disk is

$$\begin{equation} F_{d,{\rm res}} \simeq 4\pi G\Sigma _{\rm disk} \Sigma _{\rm gas}\Delta \Omega \end{equation} \tag{ 3 }$$

where Σ_disk is the total surface density of the disk. If we assume an exponential disk, then Σ_disk = (M_d/4πGR_d)exp (− R/R_d). The restoring force due to M33 as a whole is

$$\begin{equation} F_{t,{\rm res}} = \frac{V_{{\rm M}33}^2}{R}\Sigma _{\rm gas} \Delta \Omega , \end{equation} \tag{ 4 }$$

where we assume that M33 also has a flat rotation curve. Setting F_tid = F_{t, res} yields an estimate for the tidal radius: R_t ≃ R_p(V_M33/V_M31).

The different forces are shown in Figure 6, and for purely illustrative purposes, we set V_M31 = 250 km s⁻¹, V_M33 = 100 km s⁻¹, V_rel = 400 km s⁻¹, R_p = 50 kpc, M_d = 2.6 × 10⁹ M_☉, R_d = 1.1 kpc, and ρ_gas = 8 × 10⁻⁵ cm⁻³ m_H. The forces due to the M31 tidal field and due to M33 itself are equal at the tidal radius R_t ≃ 20 kpc. Ram pressure can disrupt the disk at R ≳ 7 kpc and strip material from M33 at R ≳ 14 kpc.

Zoom In Zoom Out Reset image size

While these arguments represent order-of-magnitude estimates, they appear to roughly correspond to the observed structures in M33. However, a full exploration of the M31–M33 interaction, especially with regards to the stripping of M33, using models including gas, stars, and dark matter, are required to fully understand the complex dynamics on display; this will form the basis of an upcoming paper (J. Dubinski et al. 2013, in preparation).

We have presented the spatial and kinematic correlations of the stellar and gaseous substructures within the halos of M31 and M33. While significant substructure is apparent in each component, and there is a gross alignment of significant features, mainly along the axis connecting M31 and M33, there is a distinct lack of correlation between the detailed structures.

The lack of a truly global kinematic survey of the stellar substructure within the halos of M31 and M33 limits a detailed comparison of stars and gaseous material, although the growing number of "key-hole" observations with 10 m class telescopes is addressing this. However, this situation will be resolved with the advent of wide-field spectroscopy on large telescopes, such as the proposal to build the ngCFHT,¹⁹ ushering in a new era in understanding galactic archaeology.

The resulting conclusion of this study, therefore, is that the gaseous and stellar substructures within the halos of M31 and M33 have been built through a number of distinct accretion events, but differing physical processes have driven the disruption of the stellar and gaseous components of any particular substructure, with the latter suffering the additional forces of shocking and ram-pressure stripping. Given that, through the use of high-resolution numerical simulations, our understanding of the differing processes driving the disruption of stellar and gaseous material, a comparison of their distribution through the halo should allow us to dynamically date the accretion; this will be the subject of a future contribution.

G.F.L. gratefully acknowledges financial support for his ARC Future Fellowship (FT100100268) and through the award of an ARC Discovery Project (DP110100678). Furthermore, G.F.L. thanks the Institute of Astronomy at the University of Cambridge for their support and hospitality during his sabbatical during the latter half of 2010, where a substantial portion of this work was undertaken.