MNRAS 428, 1248–1262 (2013) doi:10.1093/mnras/sts112 Unearthing foundations of a cosmic cathedral: searching the stars for M33’s halo Robert Cockcroft,1‹ Alan W. McConnachie,2 William E. Harris,1 Rodrigo Ibata,3 Mike J. Irwin,4 Annette M. N. Ferguson,5 Mark A. Fardal,6 Arif Babul,7 Scott C. Chapman,4 Geraint F. Lewis,8 Nicolas F. Martin9 and Thomas H. Puzia10 1Department of Physics and Astronomy, McMaster University, Hamilton, Ontario L8S 4M1, Canada 2NRC Herzberg Institute of Astrophysics, 5071 West Saanich Road, Victoria, British Columbia V9E 2E7, Canada 3Observatoire Astronomique, Universite´ de Strasbourg, CNRS, 11, rue de l’Universite,´ F-67000 Strasbourg, France Downloaded from 4Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA 5Institute for Astronomy, University of Edinburgh, Blackford Hill, Edinburgh EH9 3HJ 6Department of Astronomy, University of Massachusetts, 710 North Pleasant Street, Amherst, MA, USA 7Department of Physics and Astronomy, University of Victoria, Victoria, BC, V8P 1A1, Canada 8Institute of Astronomy, School of Physics A28, University of Sydney, NSW 2006, Australia 9Max-Planck-Institut fur¨ Astronomie, Knigstuhl 17, D-69117 Heidelberg, Germany http://mnras.oxfordjournals.org/ 10Department of Astronomy and Astrophysics, Pontificia Universidad Catlica de Chile, Av. Vicuna Mackenna 4860, 7820436 Macul, Santiago, Chile Accepted 2012 September 28. Received 2012 August 28; in original form 2012 May 22 ABSTRACT We use data from the Pan-Andromeda Archaeological Survey to search for evidence of an extended halo component belonging to M33 (the Triangulum Galaxy). We identify a population at Pontificia Universidad Cat�lica de Chile on May 19, 2016 of red giant branch (RGB) stars at large radii from M33’s disc whose connection to the recently discovered extended ‘disc substructure’ is ambiguous, and which may represent a ‘bona fide’ halo component. After first correcting for contamination from the Milky Way foreground population and misidentified background galaxies, we average the radial density of RGB candidate stars over circular annuli centred on the galaxy and away from the disc substructure. We find evidence of a low-luminosity, centrally concentrated component that is everywhere −2 in our data fainter than μV ∼ 33 mag arcsec . The scalelength of this feature is not well constrained by our data, but it appears to be of the order of rexp ∼ 20 kpc; there is weak evidence to suggest that it is not azimuthally symmetric. Inspection of the overall colour–magnitude diagram for this region that specifically clips out the disc substructure reveals that this residual RGB population is consistent with an old population with a photometric metallicity of around [Fe/H] ∼−2 dex, but some residual contamination from the disc substructure appears to remain. We discuss the likelihood that our findings represent a bona fide halo in M33, rather than extended emission from the disc substructure. We interpret our findings in terms of an upper limit to M33’s halo that is a few per cent of its total luminosity, although its actual luminosity is likely much less. Key words: galaxies: evolution – galaxies: individual: M33 – galaxies: haloes – Local Group – galaxies: spiral. of that material forms the stellar halo of the larger galaxy. Stellar 1 INTRODUCTION AND BACKGROUND haloes therefore contain the remnants of past interactions between cold dark matter (CDM) cosmology predicts that larger galax- galaxies, and their properties can indicate the approximate time, ies are built through the hierarchical merging of smaller galaxies. size and frequency of past mergers (e.g. Purcell, Bullock & Zentner Infalling components are disrupted partially or entirely, and part 2007). We can directly observe only a few relatively nearby haloes in any great detail due to their faint nature, and only a small number E-mail: [email protected] C 2012 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society Searching the stars for M33’s halo 1249 of haloes are directly observed through their resolved stars outside 2 HALOS OF THE LOCAL GROUP GALAXIES of the Local Group. Due to their faintness, it is problematic to de- termine whether or not the haloes are smooth and/or symmetric. 2.1 The Milky Way Galaxy There is likely a continuum of scenarios that we observe between The Local Group provides the closest opportunity to study a stellar newly accreted objects (creating streams, shells, etc.; e.g. Mart´ınez- halo but even the MW is problematic to observe because of the Delgado et al. 2010) and smooth haloes, and our interpretation will restrictions and biases associated with viewing our Galaxy from depend on the time since the accretion and the spatial resolution and within – although it has obviously been studied in depth (e.g. see depth of the observations. The long dynamical time-scales for struc- the annual review by Helmi 2008, and references therein). Current tures outside of the disc implies that they are long lived (Johnston, seemingly contradictory evidence means it is unclear whether the Hernquist & Bolte 1996). MW stellar halo is oblate, prolate or triaxial (Newberg & Yanny Outside of the Local Group, halo detections are extremely chal- 2006; Deason, Belokurov & Evans 2011), although models of the lenging as it becomes more difficult with increasing distance to dark matter (DM) halo seem to favour triaxiality (Law, Majewski distinguish the halo from other stellar components (e.g. Dalcanton & Johnston 2009; Law & Majewski 2010). Numerous detections of Downloaded from & Bernstein 2002; de Jong, Radburn-Smith & Sick 2008) and even substructure beyond the stellar bulge and disc are another reason that more so in the absence of kinematical data (Barker et al. 2009, 2012). this ambiguity remains – substructure such as the Sagittarius dwarf Scattered light and non-stellar pollution of counts also interfere with galaxy (Ibata, Gilmore & Irwin 1994) and associated tidal streams halo detections (e.g. de Jong 2008). (The surface brightness detec- (Ibata et al. 2001b, 2002; Majewski et al. 2003), the Monoceros ring tion limits generally needed are ≥7magfainterthantheskywhere (Crane et al. 2003; Ibata et al. 2003; Yanny et al. 2003), overdensities the ‘darkest’ skies (20th percentile), at Mauna Kea, are fainter than http://mnras.oxfordjournals.org/ in Canis Major (Martin et al. 2004a,b) and Virgo (Vivas et al. 2001; μ 21.3.)1 V Newberg et al. 2002; Xu, Deng & Hu 2006; Juric´ et al. 2008), Searches for haloes around distant galaxies began with deep ob- clouds in the Triangulum–Andromeda region (Rocha-Pinto et al. servations around single galaxies using surface brightness photom- 2004; Martin, Ibata & Irwin 2007) and the Hercules–Aquila region etry and, as recent studies continue to do, focused on late-type (Belokurov et al. 2007a), and finally the Orphan (Belokurov et al. edge-on galaxies (e.g. Sackett et al. 1994; Shang et al. 1998; Zibetti 2006, 2007b; Grillmair 2006) and Cetus Polar (Newberg, Yanny & & Ferguson 2004; Tikhonov & Galazutdinova 2005; Buehler et al. Willett 2009) streams. 2007; de Jong et al. 2007; Mouhcine, Rejkuba & Ibata 2007; Seth There is growing evidence to suggest that the MW halo has a dual et al. 2007; Rejkuba, Mouhcine & Ibata 2009; Mouhcine, Ibata & halo, with the different components being a result of their different Rejkuba 2010; Radburn-Smith et al. 2011). Each stellar component formation processes (e.g. Chiba & Beers 2000; Carollo et al. 2007; at Pontificia Universidad Cat�lica de Chile on May 19, 2016 is revealed more easily in the cross-section rather than the face- Miceli et al. 2008; de Jong et al. 2010; Beers et al. 2012), such on view. An alternative technique stacks many re-scaled images of as satellite accretion and in situ formation (e.g. Bell et al. 2008; galaxies together before looking for a halo signal (Zibetti, White & Schlaufman et al. 2009, 2012; Zolotov et al. 2009; Oser et al. 2010; Brinkmann 2004; de Jong 2008; Bergvall, Zackrisson & Caldwell McCarthy et al. 2012). 2010; Zackrisson & Micheva 2011; Zackrisson, de Jong & Micheva The total (dark plus luminous) mass of the Galaxy within 300 kpc 2012), again highlighting the difficulty of detecting haloes because is estimated to be in the range 0.7 ≤ M ≤ 3.4 × 1012 M of their extreme faintness. MW (Baiesi Pillastrini 2009; Watkins, Evans & An 2010). The MW Haloes have also been observed around other types of galax- stellar halo luminosity, including all substructure, is estimated to ies, not just late-type edge-on galaxies: for example, the Virgo 9 be of the order of L , ,V ∼ 10 L (Carney, Latham & Laird Cluster’s central elliptical galaxy, M87 (Weil, Bland-Hawthorn & MW halo 1990; Bullock & Johnston 2005, and references therein), com- Malin 1997), nearby starburst galaxies (Bailin et al. 2011; Rys´ +1.0 10 pared to the MW host luminosity, L , ,V = 2.1 × 10 L et al. 2011; Rich et al. 2012), the Leo elliptical NGC 3379 (Harris MW host −0.6 (Sackett 1997). et al. 2007) and the giant elliptical NGC 5128 (Centaurus A; see Malin, Quinn & Graham 1983; Peng et al. 2002; Rejkuba et al. 2011). 2.2 The Andromeda Galaxy Section 2 provides a literature review of stellar haloes in the Local Group. Details of the Pan-Andromeda Archaeological Sur- Observations of the Andromeda and Triangulum Galaxies (M31 vey (PAndAS) observations around M33 are given in Section 3. and M33, respectively) are free from the problems inherent with We are ultimately concerned with identifying the red giant branch viewing the MW from within, but are still close enough to resolve (RGB) stars in the M33 halo (if it exists). However, we must ex- individual stars (Crotts 1986; Mould & Kristian 1986) – and many clude the regions associated with the extended optical substruc- ground-based studies are now also resolving individual stars be- ture surrounding the disc identified in McConnachie et al.