Was the Andromeda Stream Produced by a Disk Galaxy?
Mark A. Fardal, Arif Babul, Puragra Guhathakurta, Karoline M. Gilbert, Cara Dodge
aa r X i v : . [ a s t r o - ph ] M a r Submitted to Astrophysical Journal Letters
Preprint typeset using L A TEX style emulateapj v. 08/22/09
WAS THE ANDROMEDA STREAM PRODUCED BY A DISK GALAXY?
Mark A. Fardal , Arif Babul , Puragra Guhathakurta , Karoline M. Gilbert , Cara Dodge Submitted to Astrophysical Journal Letters
ABSTRACTThe halo region of M31 exhibits a startling level of stellar inhomogeneities, the most prominent ofwhich is the “giant southern stream”. Our previous analysis indicates that this stream, as well asseveral other observed features, are products of the tidal disruption of a single satellite galaxy withstellar mass ∼ M ⊙ less than 1 Gyr ago. Here we show that the specific observed morphology of thestream and halo debris favors a cold, rotating, disk-like progenitor over a dynamically hot, non-rotatingone. These observed characteristics include the asymmetric distribution of stars along the stream cross-section and its metal-rich core/metal-poor sheath structure. We find that a disk-like progenitor canalso give rise to arc-like features on the minor axis at certain orbital phases that resemble the recentlydiscovered minor-axis “streams”, even reproducing the lower observed metallicity of these streams.Though interpreted by the discoverers as new, independent tidal streams, our analysis suggests thatthese minor-axis streams may alternatively arise from the progenitor of the giant southern stream.Overall, our study points the way to a more complete reconstruction of the stream progenitor and itsmerger with M31, based on the emerging picture that most of the major inhomogeneities observed inthe M31 halo share a common origin with the giant stream. Subject headings: galaxies: M31 – galaxies: interactions – galaxies: kinematics and dynamics INTRODUCTIONThe relative proximity of the Andromeda galaxy (M31)and the global perspective from our external vantagepoint make M31 an excellent laboratory for studyingthe stellar halos of large galaxies. Resolved stellarmaps of M31’s halo, assembled over the past decade,have revealed highly complex inhomogeneities, the moststriking of which is the Giant Southern Stream (GSS),extending ∼
150 kpc away from M31’s center in thesoutheast direction (Ibata et al. 2001; Ferguson et al.2002; McConnachie et al. 2003; Ibata et al. 2007, here-after I07) and falling towards M31’s center with rela-tive radial velocities as high as ∼
250 km s − (Ibata et al.2004; Guhathakurta et al. 2006; Kalirai et al. 2006).Other significant morphological and kinematic featuresin the M31 halo include stellar shelves (Ferguson et al.2002; Fardal et al. 2007; Gilbert et al. 2007, hereafterF07/G07) as well as the recently discovered minor axis“streams” (I07). The GSS is especially notable because itoffers an opportunity to precisely measure M31’s poten-tial (Ibata et al. 2004; Fardal et al. 2006, hereafter F06)and provides a view into the most significant Local Groupgalaxy disruption in the last Gyr.Models detailing the formation of the GSS agree re-markably well with most aspects of the observations,and suggest the progenitor had a stellar mass of ∼ × M ⊙ (Font et al. 2006, F06; F07). Our kinematicanalysis in F07 finds that seemingly unrelated featureslike the “Northeast Shelf” and less prominent “West- Electronic address: [email protected] Dept. of Astronomy, University of Massachusetts, Amherst,MA 01003, USA UCO/Lick Observatory, Dept. of Astronomy & Astrophysics,Univ. of California, 1156 High St., Santa Cruz, CA 95064, USA Dept. of Physics & Astronomy, University of Victoria, ElliottBuilding, 3800 Finnerty Rd., Victoria, BC, V8P 1A1, Canada Astronomy Department, Smith College, Clark Science Center,Northampton, MA 01060, USA ern Shelf” are also the result of the same disruptionprocess (F07), a conclusion supported by independentstudies of their stellar populations (Ferguson et al. 2005;Richardson et al. 2008). The observed GSS’s most strik-ing point of contrast with the models is its asymmetryin the transverse direction. As shown both with photo-metric samples (McConnachie et al. 2003) and spectro-scopic surveys (G07), its stellar distribution is sharplytruncated on the NE side and falls off much more slowlyon the SW side. In addition, the current models do notaddress the observed stellar population gradients withinthe GSS (Ferguson et al. 2002; McConnachie 2006, I07).In this letter, we show that this structure in the GSScan be accounted for if the progenitor hosted a cold, ro-tating stellar disk, unlike the simple spherical progeni-tors used in previous simulations. Surprisingly, we findthat the disruption of a disk galaxy can also give rise tofeatures similar to the recently discovered arc-like minor-axis “streams”, leading to the tantalizing possibility thatmost of the major inhomogeneities observed in the M31halo are tidal debris from the same galaxy that causedthe GSS. In Section 2, we briefly describe our model forthe progenitor and our N -body study of its tidal dis-ruption. In Section 3, we show results from these sim-ulations, focusing on the transverse density profile ofthe GSS, the metallicity gradient, and arc-like structuresthat overlap the minor axis. Section 4 summarizes ourconclusions. SIMULATION METHODOur simulations are based on the methods worked outin our earlier papers: Geehan et al. (2006), F06, andF07. We use the orbit and potential from Table 1 ofF07 and their spherical Plummer model to represent anon-rotating progenitor. For runs with a disk progen-itor, we use the same initial position and velocity, butsubstitute a different initial structure of the satellite.Briefly, our disk models assume the satellite is com- Fardal et al. (a)
Ibata et al 2007 (b)Disk A (c)Disk B(d)Plummer (e)Disk A (f)Disk B
Fig. 1.— (a): Stellar surface density/metallicity map of M31 from I07. The shallower INT/WFC survey is used inside the large ellipse,and the deeper CFHT/MegaCam survey outside it. The minor-axis “streams” C and D are visible at the lower left, projecting from thelarger plume of the GSS. These streams (green) and the GSS cocoon (red) are observed to be more metal poor than the GSS core (yellow).(b): Mass surface density map from model Disk A, 840 Myr into the run. The map is 160 kpc on a side, and a dotted contour indicatesM31’s disk orientation. The square indicates the region shown in Figure 2. (c): Same for model Disk B, at 680 Myr. (d): Same for thePlummer model, at 840 Myr. (e): Map of the metallicity as a function of position in Disk A (at 840 Myr), with red denoting the highestmetallicity, dark blue intermediate, and light blue the lowest (see Figure 3a for a quantitative scale). Boxes indicate the regions used forthe metallicity histograms in Figure 3. (f): Same for Disk B (at 680 Myr). posed of a bulge and rotating disk of stars. For sim-plicity we assume that the dark matter associated withthe galaxy has been tidally stripped before the encountermodeled here. We use a hot exponential sech disk withmass 1 . × M ⊙ , radial scale length 0 . . × M ⊙ and scale length 0 . RESULTS3.1.
Stream morphology
Figures 1b,d show surface density maps based on theDisk A and Plummer models, respectively. Both modelsreproduce the main feature of a stream extending to theSE. They also reproduce the observed line-of-sight dis-tances and velocities along the GSS. However, the trans-verse distribution of GSS stars is strikingly different be-tween the two models—Disk A displays a much sharperNE edge. The observed star-count maps (Ferguson et al.2002, I07) are not directly comparable since they con-tain both non-GSS-related M31 components and non-M31 contaminants and are not explicitly calibrated tostellar surface density, but the morphology of the GSS inthese maps appears much closer to our disk model.Figure 1d shows that the Plummer model results ina large amount of stars spilling over as far as the SEminor axis, located to the NE of the GSS. When G07as the Andromeda Stream Produced by a Disk Galaxy? 3 ξ M31 (degrees)-2.0-1.5-1.0-0.50.0 η M ( d e g ree s ) Inner minor axis a0, outer minor axis f135, near stream
Fig. 2.—
Comparison of the minor-axis contamination to theobservations of G07. The G07 DEIMOS masks (rectangles) aregrouped into inner minor-axis masks, outer minor-axis masks, anda single mask (f135) offset from the minor axis. M31’s center isat (0, 0). The inset plots for each group show the ratio R m of thestrength of the GSS component to the peak of the GSS at the same R proj . R m is measured as discussed in the text for Disks A and Band the Plummer model 840 Myr into the runs. The observationalestimates and ± σ error bars from Gilbert et al. (in preparation)are plotted as horizontal solid and dotted lines. The Plummermodel clearly contributes too much debris on the minor axis. compared their Keck/DEIMOS spectroscopic data nearM31’s SE minor axis to this model, they noted muchless spillover from the GSS than predicted by the model.Gilbert et al. (in preparation) has quantified this by di-viding the number of stars moving with GSS-like veloc-ities on the minor axis to those in the GSS core at thesame projected radius R proj . For the nine innermostDEIMOS masks on the minor axis combined, this ra-tio R m = 0 . ± .
02; for the three outermost maskson the minor axis, R m = 0 . ± .
02; and for the maskf135 located somewhat nearer the GSS, they find a likelydetection of GSS material with R m = 0 . ± . N -body models to these results. We have selected“GSS” particles by defining the trend of radial veloc-ity v with R proj and then taking stars that fall within ±
80 km s − of this velocity in the given field. We also re-strict the particles to those actually in the GSS’s “shell”.We then repeat the procedure for a control field locatedat the peak of the GSS at the same R proj , using a smallerinterval ±
40 km s − as the GSS core has a sharper veloc-ity distribution. Clearly the two disk models are in betteraccord with the observations than the Plummer model.The sharper NE edge and smaller minor-axis contami-nation of the disk models thus imply that the progenitorwas rapidly rotating. We will explore this argument inmore detail in Fardal et al. (2008, in preparation).3.2. Metallicity pattern
The mean color of GSS RGB stars is observed tovary in the transverse direction: the GSS is significantlybroader in blue than in red stars (Ferguson et al. 2002;McConnachie 2006). This is probably due to a metallic- ity gradient. I07 quantified the metallicity distributionin two GSS-dominated regions, one in the center of theGSS and one in a less dense “cocoon” region to the SW,and showed that the latter has a lower mean metallicity.Disk galaxies, of course, tend to have metallicity gra-dients. Therefore it is interesting to see how a plausiblegradient in our disk progenitor translates to the metal-licity pattern on the sky.We use a simple parametric model to produce a plau-sible metallicity gradient in our initial disk model. Wefirst find the specific orbital energy E i of each parti-cle. We then assign it a metallicity using [Fe / H] = A Z + B Z log (cid:2) − E i / (50 km s − ) (cid:3) , setting A Z and B Z to agree with the results of I07 as explained below. Thisproduces the metallicity gradient seen in Figure 3a. Ob-servational results for the stars in the small disk galaxyM33 are also plotted, with the radius for both galaxiesnormalized by the disk scale length; the metallicity pat-tern of our disk model agrees quite well. The GSS pro-genitor should perhaps be lower in metallicity than M33by a few tenths of dex due to its lower inferred mass,but the photometric metallicity measurements probablyhave systematic uncertainties at this level in any case.Figure 1e shows the sky view of the resulting modelmetallicity pattern. The gradient along the stream isvery weak, but the mean metallicity along the densercentral part is clearly higher than in the broad wingto the SW, similar to the pattern seen in M31’s GSS.Using I07’s Figure 27, we estimate the “core” and “co-coon” regions (at R proj ∼
60 kpc) have mean metallic-ities of [Fe / H] = − .
54 and − .
71, respectively (mean[Fe / H] = − .
51 was obtained at the GSS’ sharp NEedge by Guhathakurta et al. 2006). Figure 1e shows“broad wing” and “central GSS” boxes chosen at a sim-ilar radius, but better matching the slightly differentmodel stream position. Once we set A Z = − .
70 and B Z = 1 .
06, the metallicities in these boxes are also − . − .
71. The bare fact we can match two metallici-ties with two parameters is not in itself meaningful, butit is significant that the magnitude and sign of our ini-tial metal gradient are very reasonable (Fig. 3a). Fig-ure 3b shows that within each box there is a wide rangeof metallicities; the distributions in I07 appear somewhatbroader, but given measurement errors and the contribu-tions from other halo components this is not surprising.3.3.
Minor-axis arcs
Using their MegaCam photometric survey of M31’shalo, I07 found multiple surface density ridges along theminor axis which they called “streams”. Streams C andD (the two closest to M31) form a pair of curving ridgesat slightly different orientations, which appear to mergeas they approach the survey boundary (see their Fig. 22).Stream C appears to be slightly broader than stream D,and slightly more metal-rich, though not as metal-rich asthe GSS core/cocoon. From I07’s Figure 33 we estimatethe mean metallicity of streams C and D to be − . − .
91 respectively. Mori & Rich (2008) suggestedthese “streams” might be shell features from a satellitedisruption, similar to the event that created the GSS butfrom a different progenitor.While studying our overall sample of runs based on12 disk orientations, we noticed one (Disk B) contain-ing two curious “arcs” crossing the minor axis. These Fardal et al.
Fig. 3.— (a): Metallicity values in the original disk prior to disruption are shown by black dots, where the radius is plotted in units of diskscale length. For comparison, observed results for M33’s disk stars are plotted as symbols and lines: Stephens & Frogel (2002) (square);linear approximation to points of Kim et al. (2002) (straight line); Galleti et al. (2004) (diamonds); McConnachie et al. (2006) (cross);Barker et al. (2007) (triangles). The colorbar translates [Fe / H] to the color scale of Figures 1ef. (b): Histogram of particle metallicityvalues in model Disk A within the “core” and “cocoon” regions marked in Figure 1e. (c): Histogram of particle metallicity values in modelDisk B within the “core”, “cocoon”, “N arc”, and “S arc” regions marked in Figure 1f. arcs are clearest at the step 680 Myr into the run shownin Figure 1c. Morphologically, the two arcs somewhatresemble streams C and D, with a fatter southern arcnearly merging into a sharper northern arc. Like the ob-served “streams”, neither arc crosses the GSS to the SW.Compared to the observed arcs, the simulated arcs aresignificantly further from M31’s center.As Figure 1f shows, the simulated arcs are significantlyless rich in metals than the GSS. Using the same metal-licity model as for Disk A and the regions defined byboxes in this figure, the mean [Fe / H] is − .
78 for thesouthern arc and − .
90 for the northern arc. Thus thereis considerable if inconclusive evidence that these arcsare close analogues of the “streams” in I07.In our model, these two arcs originate from the outerregions of the disk, and are sharp mainly because of therelatively cold velocity field of the disk. Both arcs consistof material that takes a path around M31’s center nearlyopposite to the bulk of the progenitor, explaining whythey lie so far from the GSS. The large size of our diskis thus crucial; a compact progenitor resembling M32,for example, would be unable to produce similar arcs.The southern arc consists of a group of particles sharingnearly the same energy, and come from fairly far outin the progenitor’s disk. The northern arc consists ofparticles that lie even further out (explaining its lowermetallicity on average), which form a tidal tail duringthe interaction with M31.We cannot yet explore the full parameter space of thecollision for the presence and properties of these arc-likefeatures. However, we did conduct a few additional runswith changes to the disk mass, radius, and orientationof Disk B, finding the arcs were sensitive to the exactinput parameters. Thus we will require more theoreticalinvestigation as well as more observational constraintsto determine whether the arcs explain some of the I07minor-axis streams, or are merely a fortuitous similar-ity. If the arcs are shown to be related to the GSS, theywill be a very solid argument for the disk nature of theprogenitor, and will place strong constraints on the pa-rameters of the collision. CONCLUSIONSIn summary, several strands of observational evidencesuggest that the GSS originated from a progenitor witha strong sense of rotation, such as a disk galaxy. Thetransverse density profile of the GSS is more easily pro-duced by a rotating satellite. The observed decline inmean metallicity from the central core of the GSS to its“cocoon” to the SW suggests that the progenitor hada strong radial metallicity gradient, of the sort foundmainly in disk galaxies. Furthermore, several observedarcs lying across the minor axis in M31 have very sugges-tive analogues in one of our runs. If shown to be relatedto the GSS in the manner suggested by our model, thesefeatures would be definite confirmation of a disk-like pro-genitor.The notion of a disk galaxy progenitor is somewhat atodds with age measurements of the GSS, which suggestslittle star formation during the last 4 Gyr (Brown et al.2006a,b). However, the fields used to infer this wereplaced in the central, metal-rich part of the GSS; it ispossible that the progenitor had an age gradient as wellas a metallicity gradient, with the older stars on the in-side. Age measurements in the GSS cocoon would there-fore be interesting. It is also possible that the GSS pro-genitor was more similar to an S0 galaxy than a spiral,perhaps due to stripping of its gas in an earlier phase ofits encounter with M31.Many papers have used metallicity to assess the rela-tionship among various M31 disk and halo features. Oursuggestion that the GSS progenitor had a strong metal-licity gradient means that metallicity can no longer beused as a reliable fingerprint of origin. This obviouslycomplicates the forensic reconstruction of M31’s mergerhistory. Despite this, the rapidly growing databaseon M31 halo structure is a fascinating puzzle, offeringunique insight into the life of a typical disk galaxy andthe death of its unfortunate former companions.We thank Tom Quinn and Joachim Stadel for the useof PKDGRAV, Josh Barnes for the use of ZENO, andAlan McConnachie and Roger Davies for helpful conver-as the Andromeda Stream Produced by a Disk Galaxy? 5sations.
REFERENCESBarker, M. K., Sarajedini, A., Geisler, D., Harding, P., &Schommer, R. 2007, AJ, 133, 1125Brown, T. M., Smith, E., Guhathakurta, P., Rich, R. M.,Ferguson, H. C., Renzini, A., Sweigart, A. V., & Kimble, R. A.2006a, ApJ, 636, L89Brown, T. M., Smith, E., Ferguson, H. C., Rich, R. M.,Guhathakurta, P., Renzini, A., Sweigart, A. V., & Kimble,R. A. 2006b, ApJ, 652, 323Fardal, M. A., Babul, A., Geehan, J. J., & Guhathakurta, P.2006, MNRAS, 366, 1012 (F06)Fardal, M. A., Guhathakurta, P., Babul, A., & McConnachie,A. W. 2007, MNRAS, 380, 15 (F07)Ferguson, A. M. N., Irwin, M. J., Ibata, R. A., Lewis, G. F., &Tanvir, N. R. 2002, AJ, 124, 1452Ferguson, A. M. N., Johnson, R. A., Faria, D. C., Irwin, M. J.,Ibata, R. A., Johnston, K. V., Lewis, G. F., & Tanvir, N. R.2005, ApJ, 622, L109Font, A., Johnston, K. V., Guhathakurta, P., Majewski, S. R., &Rich, R. M. 2006, AJ, 131, 1436Galleti, S., Bellazzini, M., & Ferraro, F. R. 2004, A&A, 423, 925Geehan, J. J., Fardal, M. A., Babul, A., Guhathakurta, P. 2006,MNRAS, 366, 996Gilbert, K. M., et al. 2007, ApJ, 668, 245 (G07)Guhathakurta, P., et al. 2005, arXiv preprint (astro-ph/0502366)Guhathakurta, P., et al. 2006, AJ, 131, 2497Ibata, R., Irwin, M. J., Ferguson, A. M. N., Lewis, G., & Tanvir,N. 2001, Nature, 412, 49 Ibata, R., Chapman, S., Ferguson, A. M. N., Irwin, M., Lewis, G.,& McConnachie, A. 2004, MNRAS, 351, 117Ibata, R., Martin, N. F., Irwin, M., Chapman, S., Ferguson,A. M. N., Lewis, G. F., & McConnachie, A. W. 2007, ApJ, 671,1591 (I07)Kalirai, J. S., Guhathakurta, P., Gilbert, K. M., Reitzel, D. B.,Majewski, S. R., Rich, R. M., & Cooper, M. C. 2006, ApJ, 641,268Kim, M., Kim, E., Lee, M. G., Sarajedini, A., & Geisler, D. 2002,AJ, 123, 244McConnachie, A. W., Irwin, M. J., Ibata, R. A., Ferguson,A. M. N., Lewis, G. F., & Tanvir, N. 2003, MNRAS, 343, 1335McConnachie, A. 2006, The Observatory, 126, 144, available at