arX

iv:0

910.

5481

v2 [

astr

o-ph

.GA

] 3

0 O

ct 2

009

Mon. Not. R. Astron. Soc. 000, 000–000 (0000) Printed 30 October 2009 (MN LATEX style file v2.2)

Heated Disk Stars in the Stellar Halo

Chris W. Purcell,1⋆ James S. Bullock,1 and Stelios Kazantzidis2

1Center for Cosmology, Department of Physics and Astronomy, The University of California, Irvine, CA 92697 USA2Center for Cosmology and Astro-Particle Physics; and Department of Physics; and Department of Astronomy,

The Ohio State University, Columbus, OH 43210 USA

30 October 2009

ABSTRACT

Minor accretion events with mass ratio Msat:Mhost ≃ 1:10 are common in thecontext of ΛCDM cosmology. We use high-resolution simulations of Galaxy-analoguesystems to show that these mergers can dynamically eject disk stars into a diffuse lightcomponent that resembles a stellar halo both spatially and kinematically. For a varietyof initial orbital configurations, we find that ∼ 5 × 108

M⊙ of primary stellar diskmaterial is ejected to a distance more than ∼ 5 scale heights above the galactic plane.This ejected contribution is similar to the mass contributed by the tidal disruption ofthe satellite galaxy itself, though it is less extended. If we restrict our analysis to theapproximate solar neighborhood in the disk plane, we find that ∼ 1% of the initialdisk stars in that region would be classified kinematically as halo stars. Our resultssuggest that the inner parts of galactic stellar halos contain ancient disk stars and thatthese stars may have been liberated in the very same events that delivered materialto the outer stellar halo.

Key words: Cosmology: theory — galaxies: formation — galaxies: evolution

1 INTRODUCTION

The canonical paradigm describing the emergence of outerstellar halos around galaxies involves the accretion andtidal disruption of satellite systems (Searle & Zinn 1978;Johnston et al. 1996; Cote et al. 2000; Bullock et al.2001; Bullock & Johnston 2005; Abadi et al. 2006;De Lucia & Helmi 2008; Johnston et al. 2008; Cooper et al.2009). Empirical evidence suggests that these pre-dicted mergers have indeed populated the outskirts ofthe Milky Way and M31 halos, with diffuse, stream-like signatures apparent at low surface brightness (re-cently, McConnachie et al. 2009; Gilbert et al. 2009a,b;Starkenburg et al. 2009; Bell et al. 2008, and referencestherein).

Perhaps unsurprisingly, significant diffuse stellar halocomponents also appear to be pervasive in the population ofnearby disk galaxies outside the Local Group (Sackett et al.1994; Morrison et al. 1997; Lequeux et al. 1998; Abe et al.1999; Zibetti & Ferguson 2004). The GHOSTS survey de-scribed by de Jong et al. (2007a) has established the pres-ence of relatively luminous and extended halos surroundingthe massive disk galaxies NGC253, NGC891, and M94; theV -band surface brightness µV of this material is typically∼ 28 − 29 mag arcsec−2 at large heights (∼ 20 − 30 kpc)

⋆ E-mail: cpurcell@uci.edu

above the disk plane. This quantity of diffuse light cannoteasily be reconciled with a model where most of the in-ner stellar halo is built by the destruction of dwarf satel-lite galaxies accreted at early times, since these old andfaint subhalos can add at most only a few percent to thetotal luminosity of the host galaxy (Purcell et al. 2007).This accreted light is typically deposited over such a largevolume that the resulting ancient halo’s surface brightnessis far below that observed in the less-extended spheroidsenveloping many massive disk galaxies. Faint stellar halomaterial around M33 (McConnachie et al. 2006, see alsoChapman et al. 2006; Kalirai et al. 2006; Helmi 2008) anddiffuse extended light around dwarf galaxies like the LMC(Minniti et al. 2003) are even more difficult to explain inthe accretion scenario (Purcell et al. 2007). These generalarguments, along with direct evidence that the inner haloof M31 (Guhathakurta et al. 2005) and the Milky Way(Carollo et al. 2007) are populated by at least two distinctcomponents, make it clear that a multivalent picture of innerstellar halo formation must emerge.

There are many sources for hot stellar halo mate-rial besides direct accretion. Gas-rich mergers in the earlyepochs of galaxy assembly are likely important for the cre-ation of inner halo material (Brook et al. 2004a,b). An-other related possibility is that in-situ stars which form athigh redshift (z >

∼ 3) become ejected to large galactocen-tric radii by subsequent major mergers between 2 <

∼ z <∼ 3

2 C. W. Purcell et al.

Figure 1. Surface brightness maps of our primary system, viewed edge-on: the initial model is visualized in the left panel, while thecenter panel shows only stars belonging to the primary galaxy following the prograde infall with orbital inclination of 60◦, and the right

panel subsamples the accreted stars only. In all cases, surface densities have been converted into surface brightnesses using a stellar-mass-to-light ratio M⋆/L = 3; note that the brightness limits have been chosen in order to accentuate stellar halo substructure at theexpense of saturating the galactic disk region.

(Zolotov et al. 2009). In small galaxies, supernovae can af-fect the galactic structure in such a way that primarygalaxy stars become liberated into an extended, stellar halodistribution (Stinson et al. 2009). Here, we focus on mi-nor mergers of the type that should be common in thelast ∼ 10 Gyr of galactic assembly (Stewart et al. 2008).These accretion events involve satellites massive enough toseverely heat the host’s stellar distribution, but not massiveenough to completely eliminate the presence of the primarydisk component (Kazantzidis et al. 2008; Read et al. 2008;Villalobos & Helmi 2008; Purcell et al. 2009a; Moster et al.2009; Kazantzidis et al. 2009).

Using a suite of high-resolution collisionless simulationsas well as resimulated test cases involving a full treatmentof hydrodynamical physics, we investigate the process of dy-namical stellar ejection induced by the accretion of satellitesonto a thin galactic disk, and demonstrate that disk starscan become members of the stellar halo, as it is definedboth spatially and kinematically, via local and global heat-ing modes. In order to determine the dynamical responseof a disk to a cosmologically-motivated accretion event, wesimulate the interaction between a Galaxy-analogue pri-mary system with virial mass Mhost ≃ 1012M⊙ and an in-falling satellite subhalo much more massive than the Galac-tic disk itself, Msat ≃ 1011M⊙ ∼ 3Mdisk. This mass ratioMsat:Mhost ≃ 1:10 has been shown, via analytic formula-tions (Purcell et al. 2007, see also Zentner 2007) as well ascosmological simulations (Stewart et al. 2008), to representthe dominant mode of mass delivery into CDM halos, andis therefore of immediate concern for the question of disksurvivability (as addressed by Purcell et al. 2009a).

The majority of the simulations we present herehave been discussed elsewhere (Purcell et al. 2009a,b;Kazantzidis et al. 2009). In Purcell et al. (2009a) we ar-gued that the resultant disks are likely not good analoguesto the Milky Way. Nevertheless, it very well may be thatthe Milky Way is unusually thin, cold, and old. The re-sultant disks are not necessarily inconsistent with the ma-jority of external disk galaxies (Moster et al. 2009), whichdo seem to have thicker scale heights than the Milky Way(Yoachim & Dalcanton 2006), though observational uncer-

tainties associated with dust-lane obscuration and other ef-fects are important caveats for these considerations. Ourgoal here is to examine a plausible scenario for the emer-gence of an inner stellar halo component in most galaxies.They should not be compared directly to the Milky Waywithout some caution (see §3.1). In §2, we outline our nu-merical experiments, following with a presentation of ourresults in §3, reserving §4 for conclusions and discussion.

2 METHODS

We utilize high-resolution, multi-million-particle simulationsin order to analyze the response of a thin stellar disk to theinfall and accretion of a satellite galaxy one-tenth as massiveas the host halo itself. Using fully self-consistent numericalmodels for both the primary and satellite galaxies, as wellas initial conditions for accreting subhalos drawn from dis-tributions defined by large-scale cosmological simulations,we investigate a range of orbital infall inclinations. Alterna-tive descriptions of our simulations are provided in Purcellet al. (2009a,b); we now review the essential features forcompleteness.

All of our numerical investigations use the multi-stepping, parallel, tree N-body code PKDGRAV (Stadel2001), in which we set the gravitational softening length toǫ = 100 pc and 50 pc for dark matter and stellar particles,respectively. Each simulation is collisionless, with the ex-ception of two experiments in which we ascertain the effectof a full treatment of hydrodynamical physics on the ejec-tion of both old and newly-formed disk stars from a primarygalaxy with a modest initial gas fraction fg = 0.1. The lattersimulation was performed with the TreeSPH N-body codeGASOLINE (Wadsley et al. 2004) and is part of a campaigndesigned to investigate the effect of a dissipative componenton the dynamical response of thin galactic disks subjectto bombardment by halo substructure (Kazantzidis et al.2009b, in preparation). In these cases, the gravitational soft-ening length for the gas particles was set to ǫ = 50 pc.

Our primary galaxy components, drawn from thefully self-consistent distribution functions advanced by

Heated Disk Stars in the Stellar Halo 3

Figure 2. Central surface brightness profiles as measured alongthe minor axis for each simulation endstate, in both the colli-sionless (thick lines) and hydrodynamical (thin lines) regimes, asa function of orbital inclination angle. Here, as in Figure 1, weassign a surface brightness µ based on the stellar-mass-to-lightconversion factor M⋆/L = 3. For reference, at heights z > 5 kpc,we plot a profile with arbitrary normalization and power-law in-dex n = −2.5.

Widrow et al. (2008), represent equilibrium solutions to thecoupled collisionless Boltzmann and Poisson equations, andare thus ideally suited for simulating the complex dynam-ics involved in the accretion of a massive satellite galaxy.The host system consists of a massive disk as well as acentral bulge following a Sersic profile of effective radiusRe = 0.58 kpc and index n = 1.118 (with mass Mbulge =9.5×109M⊙ distributed among 5×105 particles); this is themodel labeled G1 in Purcell et al. (2009a). The stellar diskis initialized with an exponential scale length Rd = 2.84 kpcand a vertical distribution described by a sech2 functionwith scale height zd = 0.43 kpc, and contains 106 particlescomprising a total disk mass Mdisk = 3.6 × 1010M⊙.

In the hydrodynamical runs, a fraction fg = 0.15 ofthe stellar disk mass is converted into gas, and this gaseouscomponent is constructed with the same initial density dis-tribution as the stellar disk; the adopted methodology willbe described in detail in Kazantzidis et al. (2009b, in prepa-ration). We include atomic cooling for a primordial mixtureof hydrogen and helium, and the star formation algorithmis based on that of Katz (1992), in which gas particles incold and dense Jeans-unstable regions as well as convergentflows spawn star particles at a rate proportional to the localdynamical time. In our particular application, gas particlesare eligible to form stars if their density exceeds 0.1 cm−3

and their temperature drops below 1.5 × 104 K. Feedbackfrom supernovae is treated using the blast-wave model de-scribed in Stinson et al. (2006); in this model, the energydeposited by a Type-II supernova into the surrounding gasis 4 × 1050 erg. This choice of parameters and numericaltechniques produces realistic galaxies in cosmological simu-lations (Governato et al. 2007). Each primary galaxy modelis embedded in a dark host halo composed of 4 × 106 par-

ticles and arranged in a structure following the canonicalNFW density profile of Navarro et al. (1996), with scale ra-dius rs = 14.4 kpc and virial mass Mhost ≃ 1012M⊙. Wenote that our primary galaxy model is a Galactic-analoguesystem chosen from the Widrow et al. (2008) set of modelswith characteristics corresponding closely to criteria deter-mined observationally for the Milky Way, and that our pre-cise choice of the initial parameter set minimizes the effectsof secular evolution such as the formation of a strong centralbar, as well as artificial heating induced by the interactionof disk particles with more massive halo particles.

The satellite galaxy simulated in each isolated accre-tion event is initialized using the self-consistent formalismof Kuijken & Dubinski (1995), with which we embed a to-tal stellar mass M⋆ = 2.2 × 109M⊙ (adopting the value forM⋆/Msat at z ∼ 0.5 derived in the number-density matchingexercise of Conroy & Wechsler 2009), comprised of 105 par-ticles populating a spheroidal distribution with Sersic indexn ∼ 0.5 (according to the findings of van Zee et al. 2004, fordSph shape parameters as a function of magnitude in Virgocluster member systems), into a dark subhalo composed of106 particles with a virial mass 1011M⊙ and having a den-sity structure well-fit by an NFW profile with concentrationcvir ≃ 14 at z ∼ 0.5.

To investigate the dependence of heating efficiency dur-ing a satellite-disk interaction on the orbital parameters ofthe accreting subhalo, we simulate a range of inclination an-gles for each primary galaxy model: θ = 0◦, 30◦, 60◦, and 90◦

(where θ is defined as the angle between the angular momen-tum axes of the galactic disk and the infall orbit). In eachexperiment, the satellite galaxy is initially placed at a largedistance r ≃ 120 kpc from the center of the host halo, tominimize the shock imparted to the primary galaxy parti-cles by the sudden appearance of this new potential well.The initial velocity vector for the subhalo is motivated bycosmological analysis of substructure accretion, in which thedistributions of radial and tangential velocity componentsfor satellite systems (vr and vt) peak respectively at 90%and 60% of the host halo’s virial velocity (Benson 2005;Khochfar & Burkert 2006); therefore, our subhalo vectorshave vr = 116 km/s and vt = 77 km/s.

All fiducial simulations involve orbits that are progradewith respect to the primary galaxy’s rotation; we also per-form a retrograde simulation for the inclination angle θ =60◦ in order to assess the degree to which this case decreasesthe efficiency of disk heating while increasing the amount bywhich the primary galaxy is tilted away from its initial angleof repose, as expected according to previous experiments in-volving satellite-disk interactions (Velazquez & White 1999;Kazantzidis et al. 2009). Each prograde simulation is al-lowed to evolve for a total of 5 Gyr, by which time thesubhalo has fully coalesced with the host halo and the heat-ing process is virtually complete, leaving the galaxy in aquasi-steady state; the elapsed timescale for the retrogrademodel is slightly longer at 7 Gyr, since dynamical frictionis less efficient in this case. Lastly, the hydrodynamical sim-ulations employ prograde orbits with inclination angles forthe infalling satellite of θ = 0◦ and 60◦.

4 C. W. Purcell et al.

Figure 3. Minor-axis surface brightness µ (upper panels) and cumulative stellar mass M(> z) (lower panels) contributed by heated diskstars (solid lines) and accreted satellite stars (dashed lines) at height z from the galactic plane.

3 RESULTS

Regardless of orbital inclination, each fiducial accretionevent ejects a significant portion of stellar mass from thedisk, as shown qualitatively in the visualizations of Figure 1.This material reaches heights above the galactic plane con-sistent with the extended faint spheroids observed in manynearby massive disk galaxies, most notably in the diffuseinner halo of M31 (Guhathakurta et al. 2005; Koch et al.2008). The disk stars still located at the mid-plane of oursimulated disk remnants are much dynamically hotter thanthe initially cold and thin stellar system, yielding a distri-bution of velocities consistent with a two-component disk aswell as a non-trivial fraction of stars that have kinematicssimilar to those found in the stellar halo of the Milky Way.

Figure 2 presents the minor-axis surface-brightness pro-files of the initial disk galaxy (solid, black) and the resultantdisks after each encounter, as measured along the minoraxis extending from the galactic center1. We see that thepost-merger disk surface-brightness profile generally beginsto show a distinct transition to a diffuse power-law compo-nent at µ ∼ 28 mag/arcsec2 (for a stellar-mass-to-light ratioM⋆/L = 3). Shown for comparison is a power-law slope ofµ ∝ z−2.5, which is indicative of that expected for a stellarhalo. As measured and reported by Purcell et al. (2009a)

1 We choose the central region since it is the brightest and thusmost well-constrained area in observation of external galaxies.The minor-axis profile does not change in slope significantly asgalactocentric radius increases, although the disk outskirts atR >∼

10 kpc show pronounced flaring at faint surface brightnesses.

using a two-component sech2 fitted profile, the scale heightof the remaining thin-disk component varies along the rangezd ∼ 1−2 kpc according to the satellite’s orbital inclination.We note that the thicker component of the fit invariably hasa very large scale height (zdiffuse ∼ 4− 7 kpc), reflecting thelarge amount of energy imparted through global and directheating modes during the accretion event.

Generally, our experiments result in a significant por-tion of the initial disk’s stellar mass being ejected by thesatellite-disk interaction into a diffuse spheroidal componentat heights greater than z >

∼ 5zd ∼ 7 kpc. Figure 3 shows thesurface brightness profiles (as in Figure 2) as well as thecumulative stellar mass found above a given height z fromthe disk plane, with both quantities having been split intothe components contributed by the heated disk stars andthe accreted satellite stars. We see that low-latitude accre-tion events (θ = 0◦, 30◦) heat the disks so efficiently thata stellar mass equivalent to roughly 1 − 2% of the initialdisk mass, Mhalo

⋆ ∼ 3 − 7 × 108M⊙, is deposited at heightslarger than 7 kpc, distributed in approximately equal pro-portions between disk stars and satellite stars. For high-latitude accretion events (θ = 60◦, 90◦), the total stellarmass above 7 kpc is similar, but tidally-stripped stars fromthe accreted satellite dominate the population. Closer to thegalactic mid-plane, the very thick stellar disk becomes theprevalent factor; approximately 109M⊙ in initial-disk starsalone are ejected to a height larger than z ∼ 4 kpc, acrossthe full radial extent of the disk.

Despite some thin-disk regrowth via star formation inthe hydrodynamical simulations, the vertical disk thicken-ing is even more pronounced in those cases than in the

Heated Disk Stars in the Stellar Halo 5

Figure 4. The solar-neighborhood Toomre energy diagram for the initial model (left panel) and the endstate system including heateddisk stars (center panel) and accreted stars (right panel) following the satellite galaxy’s prograde infall at an orbital inclination of 0◦

onto a primary galaxy with an initial gas fraction fg = 0.15; the quantity T is the quadratic sum of the radial and vertical space velocitycomponents U = VR and W = Vz , and V is the rotational velocity Vφ. Disk stars are subsampled from the galactic disks in a very thinannulus (100 pc in width and 300 pc in height) centered vertically on the plane of the disk and sweeping through the circumferencedefined by the solar radius R⊙ = 8 kpc; the annulus width and height are each tripled for sampling of the accreted stars, to minimizenumerical noise. The origin of the x-axis is set in each case by the local standard of rest, defined as the mean rotational velocity V ofthe stellar subsample. We define stars inside the enclosing dashed line with a radius of 70 km/s to be kinematic members of the thindisk, while stars outside this region and enclosed by the dotted line with a radius of 180 km/s are members of the thick disk (as inNissen & Schuster 2009); subsampled stars outside this region are kinematic members of the stellar halo despite their spatial locationnear the surface brightness peak at the disk’s mid-plane. In this hydrodynamical test case, red points mark original disk stars and purple

points denote stars formed during the simulation’s evolution.

collisionless experiments, possibly because the initial stellardisk is less massive and thus more susceptible to the strongglobal resonance modes induced by the high-latitude accre-tion event. We note that in the hydrodynamical regime, theincrease in friction accelerates the subhalo infall and thusresults in less material stripped from the satellite galaxy atlarge heights above the primary galaxy plane; these accre-tion events are typically complete a few hundred Myr beforetheir collisionless counterparts.

3.1 Applicability to the Milky Way

Recent surveys indicate that the Galactic stellar halo maysomewhat possess a dual nature similar to that observedin M31, being comprised of an outer spheroid of metal-poor stars having a slightly retrograde rotation with re-spect to the Milky Way disk, and an inner component witha flatter axial ratio and a mild prograde rotation, as wellas stars more metal-rich than the classic extended haloby a factor of three (Chiba & Beers 2000; Carollo et al.2007; Kinman et al. 2007). Although most of the Galactichalo is composed of an ancient stellar population, youngerand richer stars have been observed there for decades(Greenstein & Sargent 1974; Siegel et al. 2009); a small frac-tion of these objects have peculiar velocities such that theircalculated flight times from the disk are longer than theirapparent lifetimes, and thus some debate exists over whetherthese stars formed in situ in the low-density gas of the stel-lar halo (Keenan 1992), or are possibly undergoing binaryrejuvenation (Perets 2009), or were ejected from the diskvia dynamical interactions either internal to a multiple starsystem (Blaauw 1961; Gualandris et al. 2004) or involvingthe massive black hole at the center of the Milky Way (Hills1988; Yu & Tremaine 2003). On a larger interaction scale,the simulations of Abadi et al. (2009) have shown that dis-

rupting dwarf galaxies may leave behind stars with velocitiessimilar to or exceeding the escape speed of the primary sys-tem, which may cause some degree of angular collimationand travel-time commonality in a hypervelocity population,as observed in the fast-moving Galactic halo stars clusteredin the Leo constellation.

The majority of runaway disk stars, however, are notthese hypervelocity objects but instead have moderatelyhot kinematics similar to those of the Milky Way’s classi-cal halo, although they tend to retain significant rotationalvelocity as well (Martin 2006), and can even be found farabove the Galactic plane (Ramspeck et al. 2001, see alsoAllen & Kinman 2004 and references therein). In a spectro-scopic survey of stars belonging to both the thick disk andthe stellar halo, Nissen & Schuster (2009) report a num-ber of stars with halo-like speeds, and unexpectedly highα-enrichment, surmising that this could imply either a dissi-pational collapse mechanism in the halo (as in Gratton et al.2003) or the signature of an accretion event involving a largesatellite galaxy. In this subsection we consider the impact onstellar kinematics of an accretion event involving a massivesatellite galaxy, and speculate on the extent to which thisphenomenon has operated in the context of the Milky Way,which has likely had an unusually quiescent accretion his-tory lacking in the 1 : 10 mass-ratio events simulated in ourexperiments. In order to distinguish dynamically betweenstars in the thin/thick disk structure and stars that havebeen heated into orbits similar to those found in the stel-lar halo, we employ the analysis of Toomre energy space, inwhich the radial (U) and vertical (W ) space velocities aresummed in quadrature (T = [U2+W 2]1/2) and diagrammedagainst the rotational velocity V (Sandage & Fouts 1987, seeCarollo et al. 2007 for a discussion of the solar neighborhoodas viewed in this parameter space).

In Figure 4, we extract disk stars from a thin annu-

6 C. W. Purcell et al.

lus sweeping through the solar locality and centered on themid-plane, with a radial width of 100 pc and thickness of300 pc, and plot the kinematic properties of this population.Despite being located spatially near the mid-plane of thepost-accretion thin-thick disk structure, the number of starswith thick-disk kinematics has increased significantly whencompared to the dynamical composition of the initial galaxymodels, regardless of the satellite galaxy’s orbital inclinationduring infall. Moreover, non-trivial numbers of disk starsnear the plane at the solar radius have been heated so dra-matically by the accretion event that their three-dimensionalvelocities are as high as those observed in the Galactic stellarhalo. By comparison, the kinematics of the accreted satel-lite stars are consistent with a mixed population of thick-disk and stellar-halo stars, as determined in large part bythe rotational speed of the accreted material, which lagsthe primary disk by an amount that correlates with increas-ing infall inclination angle (see Purcell et al. 2009b, for anextended discussion of this behavior as it pertains to thedynamical relationship between accreted stars and accreteddark matter).

It appears highly unlikely that the Milky Way has re-cently suffered such a dramatic accretion event as those wefiducially investigate here; not only is the Galactic disk fartoo cold and thin to have undergone much global heatingin the last several Gyr (Purcell et al. 2009a), it is also quitedeficient in stellar mass and angular momentum when com-pared to a sample of local spiral galaxies (Hammer et al.2007). These constraints indicate that the accretion historyof the Milky Way has been substantially more quiescentthan would be expected for a typical halo of virial massMhost ∼ 1012M⊙ in the ΛCDM cosmology. On the otherhand, it is plausible to suggest that minor events such asthe ongoing accretion and disruption of the Sagittarius dwarfgalaxy may also be capable of heating some small portionof disk stars to ejection, especially in the outer disk wherethe disk’s self-gravity is less powerful; we may safely assumethat the Galaxy has undergone some number of accretionsin this regime, these events being truly ubiquitous in a cos-mological context.

Regarding the Milky Way, a more salient considerationmay be the dynamical heating induced by an event occurringlong ago, such as the merger(s) that created the present-day thick disk sometime before z >

∼ 1 from a combinationof heated proto-Galactic disk stars and accreted stars. Al-though it remains to be shown self-consistently that a coldand thin Galaxy analogue can arise and persist to z = 0 afterthe thick disk’s formation at early times, we may still makea simple argument in favor of disk heating being largely re-sponsible for the complex kinematic distributions in today’ssolar neighborhood. As alluded to above, the total mass ofthe disk under assault is an integral factor in any accretionevent; according to first-order expectations, the strength ofthe stellar self-gravity will largely determine the extent towhich any deposited orbital energy (or excited global res-onance) heats the system. As an approximation, we wouldtherefore expect more severe dynamical and morphologicaldamage to be inflicted by an accretion event involving alighter disk, such as the Galactic progenitor at intermedi-ate redshift, leaving behind some fast-moving relics of thisera in nearby stellar populations at the present day. A moredetailed exploration regarding the role of disk mass in the

accretion-heating process can be found in Kazantzidis et al.(2009b, in preparation).

4 DISCUSSION

Observations of diffuse light in the inner halos of externalgalaxies have tended to inspire efforts to reconcile the stel-lar mass of the spheroid with the properties of an inferredprogenitor satellite galaxy. However, our experiments clearlydemonstrate that an inner halo component may be primarilycomposed of stars ejected to large heights above the galaxyplane during the disk heating process. In our post-accretionremnant systems, we obtain surface brightness profiles typi-fied by a power-law index n = −2.5 along the minor axis asshown in Figure 2, only slightly flatter between 5 and 10 kpcthan the projected profile found to be a general feature byZibetti et al. (2004) in their image-stacking exercise involv-ing over a thousand SDSS galaxies. In each of our simulationendstates, ejected material originally belonging to the pri-mary stellar disk accounts for ∼ 40− 80% of the luminosityfound above 5 kpc from the plane, with the accreted sub-halo’s stars being distributed over a much larger volume andthus contributing minimally to the surface brightness of theinner halo.

This morphological contamination of the inner halo oc-curs simultaneously with the energizing process that also sig-nificantly contributes stellar material to the kinematically-defined thick disk and stellar halo. Just as in models of stel-lar halo formation, the scenario of thick disk assembly mostlikely involves multiple mechanisms, although there is activedebate regarding the order of importance in these phenom-ena. The wide range of kinematic behavior displayed by low-mass thick disk galaxies may indicate, especially for cases inwhich signatures of counter-rotation appear, that these sys-tems form primarily via the accretion of stripped stars dur-ing a minor merger, or from star formation associated withsuch a merger (Yoachim & Dalcanton 2008). On the otherhand, there is solid photometric evidence that more mas-sive disk galaxies also contain thick components that aresimply too faint to discern kinematically, especially in theabsence of significant rotational lag in any particular stel-lar population. This co-rotation is a particular hallmark oflow-latitude subhalo accretion, since in this case the satel-lite’s stars will kinematically blend more efficiently into thehost’s rotation, and such uniform dynamics are also a gen-eral feature of thick disks formed largely by merger-relatedheating processes, which can achieve line-of-sight velocitydispersions similar to those found in surveys of these sys-tems. In the case of the Milky Way, the thick component’sfaintness relative to the thin disk, as well as its measurablylagging rotation, likely indicate that it is not a product ofheating that has occurred in the last several Gyr; it remainsto be seen whether the kinematic structure of the thin/thickGalactic disk is typical among systems with similar morphol-ogy and stellar content.

At least one member of the Local Group may representmore nearly the cosmological norm; the inner spheroid ofM31 has been chemically identified as having a stellar pop-ulation that is largely of intermediate age and as metal-richas that galaxy’s thick disk (Durrell et al. 2004). It may wellbe that M31, having had a more active recent accretion his-

Heated Disk Stars in the Stellar Halo 7

tory than the Milky Way (as surmised by Ferguson et al.2002; Gilbert et al. 2009b, based on density and metallicityvariations in halo substructure), has a stellar halo that hasbeen significantly contaminated by stars originally belong-ing to the primary disk. The Giant Southern Stream andassociated stellar features are clearly the tidal remnants ofan accretion event involving a massive satellite galaxy; es-timates of the progenitor’s properties suggest that it was alarge system traveling on a highly-eccentric orbit of nearlyco-planar orientation with respect to the M31 disk, and thesurface brightness of the tidal stream strongly implies thatthe event occurred within the past few Gyr (Ibata et al.2004; Fardal et al. 2006; Font et al. 2006). Our result forthe simulated infall with orbital inclination θ = 0◦ predictsa substantial amount of disk-star ejection associated withsuch an accretion event, an expectation that is consistentwith the relatively bright inner halo of M31; we can also in-fer that there may be a significant number of stars travelingat high speeds near the disk’s mid-plane.

It appears that the formation of inner galactic halosmay be heavily enhanced by the destructive heating under-gone by a stellar disk during the infall of a large satellitegalaxy. As we improve our understanding of stellar halocomposition in nearby disk galaxies, we may begin to as-sess the degree to which these diffuse spheroids are al-loyed by intermediate-age stars that are unlikely to havebeen part of the primeval halo constructed long ago fromthe debris of faint and metal-poor dwarf satellites. FutureGalactic surveys, such as SEGUE-2 and APOGEE of thethird iteration in the Sloane Digital Sky Survey (SDSS-III;Weinberg et al. 2007), will map the outer disk and stellarhalo of the Milky Way to an unprecedented degree of accu-racy, and M31 observations will continue to refine our un-derstanding of that galaxy’s complex inner structure. Theseefforts will exquisitely define the kinematics, density struc-tures, and chemical compositions of the various populationsin each component, and should easily be able to distinguishthe relative properties of ancient, metal-poor halo membersas compared to younger, more metal-rich stars that couldhave been heated to ejection from the disk during a massiveaccretion event.

We would like to thank Larry Widrow and John Dubin-ski for kindly making available the software used to set upthe initial galaxy models, and David Weinberg for useful dis-cussions. CWP and JSB are supported by National ScienceFoundation (NSF) grants AST-0607377 and AST-0507816,and the Center for Cosmology at UC Irvine. SK is supportedby the Center for Cosmology and Astro-Particle Physics atThe Ohio State University. The numerical simulations wereperformed on the IA-64 cluster at the San Diego Supercom-puting Center, with ancillary experiments performed on theGreenPlanet cluster at UC Irvine. This work was also sup-ported in part by an allocation of computing time from theOhio Supercomputer Center (http://www.osc.edu/).

REFERENCES

Abadi, M. G., Navarro, J. F., & Steinmetz, M. 2006, MN-RAS, 365, 747

Abadi, M. G., Navarro, J. F., & Steinmetz, M. 2009, ApJL,691, L63

Abe, F., et al. 1999, AJ, 118, 261Allen, C., & Kinman, T. 2004, Revista Mexicana de As-tronomia y Astrofisica Conference Series, 21, 121

Bell, E. F., et al. 2008, ApJ, 680, 295Belokurov, V., et al. 2006, ApJ, 642, L137Benson, A. J. 2005, MNRAS, 358, 551Blaauw, A. 1961, Bull. Astron. Inst. Netherlands, 15, 265Bullock, J. S., Kravtsov, A. V., & Weinberg, D. H. 2001,ApJ, 548, 33

Bullock, J. S., & Johnston, K. V. 2005, ApJ, 635, 931Brook, C. B., Kawata, D., Gibson, B. K., & Flynn, C. 2004,MNRAS, 349, 52

Brook, C. B., Kawata, D., Gibson, B. K., & Freeman, K. C.2004, ApJ, 612, 894

Carollo, D., et al. 2007, Nature, 450, 1020Chapman, S. C., Ibata, R., Lewis, G. F., Ferguson,A. M. N., Irwin, M., McConnachie, A., & Tanvir, N. 2006,ApJ, 653, 255

Chiba, M., & Beers, T. C. 2000, AJ, 119, 2843Clarkson, W., et al. 2008, ApJ, 684, 1110Conroy, C., & Wechsler, R. H. 2009, ApJ, 696, 620Cooper, A. P., et al. 2009, arXiv:0910.3211Cote, P., Marzke, R. O., West, M. J., & Minniti, D. 2000,ApJ, 533, 869

de Jong, R. S., et al. 2007, IAU Symposium, 241, 503de Jong, R. S., et al. 2007, ApJL, 667, L49De Lucia, G., & Helmi, A. 2008, MNRAS, 391, 14Durrell, P. R., Harris, W. E., & Pritchet, C. J. 2004, AJ,128, 260

Fardal, M. A., Babul, A., Geehan, J. J., & Guhathakurta,P. 2006, MNRAS, 366, 1012

Ferguson, A. M. N., Irwin, M. J., Ibata, R. A., Lewis, G. F.,& Tanvir, N. R. 2002, AJ, 124, 1452

Font, A. S., Johnston, K. V., Guhathakurta, P., Majewski,S. R., & Rich, R. M. 2006, AJ, 131, 1436

Governato, F., Willman, B., Mayer, L., Brooks, A., Stin-son, G., Valenzuela, O., Wadsley, J., & Quinn, T. 2007,MNRAS, 374, 1479

Gratton, R. G., Carretta, E., Desidera, S., Lucatello, S.,Mazzei, P., & Barbieri, M. 2003, A&A, 406, 131

Greenstein, J. L., & Sargent, A. I. 1974, ApJs, 28, 157Gualandris, A., Portegies Zwart, S., & Eggleton, P. P. 2004,MNRAS, 350, 615

Gilbert, K. M., Font, A. S., Johnston, K. V., &Guhathakurta, P. 2009a, ApJ, 701, 776

Gilbert, K. M., et al. 2009b, ApJ, 705, 1275Guhathakurta, P., Ostheimer, J. C., Gilbert, K. M., Rich,R. M., Majewski, S. R., Kalirai, J. S., Reitzel, D. B., &Patterson, R. J. 2005, arXiv:astro-ph/0502366

Hammer, F., Puech, M., Chemin, L., Flores, H., & Lehnert,M. D. 2007, ApJ, 662, 322

Helmi, A. 2008, A&A Rev., 15, 145Hills, J. G. 1988, Nature, 331, 687Ibata, R., Chapman, S., Ferguson, A. M. N., Irwin, M.,Lewis, G., & McConnachie, A. 2004, MNRAS, 351, 117

Johnston, K. V., Hernquist, L., & Bolte, M. 1996, ApJ,465, 278

Johnston, K. V., Bullock, J. S., Sharma, S., Font, A.,Robertson, B. E., & Leitner, S. N. 2008, ApJ, 689, 936

Kalirai, J. S., et al. 2006, ApJ, 648, 389

8 C. W. Purcell et al.

Katz, N. 1992, ApJ, 391, 502Kazantzidis, S., Bullock, J. S., Zentner, A. R., Kravtsov,A. V., & Moustakas, L. A. 2008, ApJ, 688, 254

Kazantzidis, S., Zentner, A. R., Kravtsov, A. V., Bullock,J. S., & Debattista, V. P. 2009, ApJ, 700, 1896

Kazantzidis, S., Purcell, C. W., Bullock, J. S., Mayer, L.,& Dotti, M. 2009, ApJ, in preparation

Keenan, F. P. 1992, QJRAS, 33, 325Khochfar, S. & Burkert, A. 2006, A&A, 445, 403Kinman, T. D., Cacciari, C., Bragaglia, A., Buzzoni, A., &Spagna, A. 2007, MNRAS, 375, 1381

Koch, A., et al. 2008, ApJ, 689, 958Kregel, M., van der Kruit, P. C., & de Grijs, R. 2002, MN-RAS, 334, 646

Kuijken, K., & Dubinski, J. 1995, MNRAS, 277, 1341Launhardt, R., Zylka, R., & Mezger, P. G. 2002, A&A, 384,112

Lequeux, J., Combes, F., Dantel-Fort, M., Cuillandre, J.-C., Fort, B., & Mellier, Y. 1998, A&A, 334, L9

Martin, J. C. 2006, AJ, 131, 3047McConnachie, A. W., Chapman, S. C., Ibata, R. A., Fer-guson, A. M. N., Irwin, M. J., Lewis, G. F., Tanvir, N. R.,& Martin, N. 2006, ApJl, 647, L25

McConnachie, A. W., et al. 2009, Nature, 461, 66Minniti, D., Borissova, J., Rejkuba, M., Alves, D. R., Cook,K. H., & Freeman, K. C. 2003, Science, 301, 1508

Mo, H. J., Mao, S., & White, S. D. M. 1998, MNRAS, 295,319

Morrison, H. L., Miller, E. D., Harding, P., Stinebring,D. R., & Boroson, T. A. 1997, AJ, 113, 2061

Moster, B. P., Maccio’, A. V., Somerville, R. S., Johansson,P. H., & Naab, T. 2009, arXiv:0906.0764

Navarro, J. F., Frenk, C. S., & White, S. D. M. 1996, ApJ,462, 563

Nissen, P. E., & Schuster, W. J. 2009, IAU Symposium,254, 103

Perets, H. B. 2009, ApJ, 698, 1330Purcell, C. W., Bullock, J. S., & Zentner, A. R. 2007, ApJ,666, 20

Purcell, C. W., Kazantzidis, S., & Bullock, J. S. 2009a,ApJL, 694, L98

Purcell, C. W., Bullock, J. S., & Kaplinghat, M. 2009b,ApJ, 703, 2275

Ramspeck, M., Heber, U., & Moehler, S. 2001, A&A, 378,907

Read, J. I., Lake, G., Agertz, O., & Debattista, V. P. 2008,MNRAS, 389, 1041

Rich, R. M., Reitzel, D. B., Howard, C. D., & Zhao, H.2007, ApJL, 658, L29

Sackett, P. D., Morrison, H. L., Harding, P., & Boroson,T. A. 1994, Nature, 370, 441

Sandage, A., & Fouts, G. 1987, AJ, 93, 74Searle, L., & Zinn, R. 1978, ApJ, 225, 357Siegel, M. H., Karatas, Y., & Reid, I. N. 2009, MNRAS,395, 1569

Stadel, J. G. 2001, PhD thesis, University of WashingtonStarkenburg, E., et al. 2009, ApJ, 698, 567Stewart, K. R., Bullock, J. S., Wechsler, R. H., Maller,A. H., & Zentner, A. R. 2008, ApJ, 683, 597

Stinson, G., Seth, A., Katz, N., Wadsley, J., Governato, F.,& Quinn, T. 2006, MNRAS, 373, 1074

Stinson, G. S., Dalcanton, J. J., Quinn, T., Gogarten, S. M.,

Kaufmann, T., & Wadsley, J. 2009, MNRAS, 395, 1455van Zee, L., Barton, E. J., & Skillman, E. D. 2004, AJ, 128,2797

Velazquez, H. & White, S. D. M. 1999, MNRAS, 304, 254Villalobos, A., & Helmi, A. 2008, MNRAS, 391, 1806Wadsley, J. W., Stadel, J., & Quinn, T. 2004, New Astron-omy, 9, 137

Wechsler, R. H., Bullock, J. S., Primack, J. R., Kravtsov,A. V., & Dekel, A. 2002, ApJ, 568, 52

Weinberg, D. H., et al. 2007, Bulletin of the American As-tronomical Society, 38, 963

Widrow, L. M., Pym, B., & Dubinski, J. 2008, ApJ, 679,1239

Yoachim, P., & Dalcanton, J. J. 2006, AJ, 131, 226Yoachim, P., & Dalcanton, J. J. 2008, ApJ, 682, 1004Yu, Q., & Tremaine, S. 2003, ApJ, 599, 1129Zentner, A. R. 2007, International Journal of ModernPhysics D, 16, 763

Zibetti, S., & Ferguson, A. M. N. 2004, MNRAS, 352, L6Zibetti, S., White, S. D. M., & Brinkmann, J. 2004, MN-RAS, 347, 556

Zolotov, A., Willman, B., Brooks, A. M., Governato, F.,Brook, C. B., Hogg, D. W., Quinn, T., & Stinson, G.2009, ApJ, 702, 1058