Stellar halos provide a window into the formation histories of individual galaxies. Stellar halos are built largely from the accretion and disruption, influenced by gravitational tidal forces, of smaller satellite galaxies. Debris features from these disruption events remain identifiable for billions of years, providing observable signatures of the merger histories of individual galaxies. However, star formation also plays a part in building up the inner regions of stellar halos, either via stars formed in situ in the early disk of the host galaxy, or formed from gas deposited in the host galaxy by disrupted satellites.

With current instrumentation, the only stellar halos that can be studied in great detail are those of the Milky Way and Andromeda (M31), the two large spiral galaxies of the Local Group. While M31's distance has historically precluded measurements that are common-place for the Milky Way's halo, the last decade has seen tremendous progress. This progress has largely been due to the photometric and spectroscopic observations obtained by the PAndAS (Pan-Andromeda Archaeological Survey; McConnachie et al. 2009), SPLASH (Spectroscopic and Photometric Landscape of Andromeda’s Stellar Halo; Guhathakurta et al. 2005, 2006; Gilbert et al. 2006), and PHAT (Panchromatic Hubble Andromeda Treasury; Dalcanton et al. 2012) collaborations. The combination of large-scale, contiguous photometric surveys and pointed spectroscopic surveys has been particularly powerful for discovering debris features and disentangling M31's structural components.

Observational strategy

The SPLASH collaboration has imaged more than 80 fields and obtained more than 20,000 stellar spectra in M31's disk, dwarf galaxies, and halo, in fields ranging from 2 to 230 kpc in projected distance from the galaxy's center. The photometric data are primarily taken with the Mosaic camera on the Mayall 4-m telescope on Kitt Peak, and include intermediate-width selected band imaging that allows us to select spectroscopic targets likely to be stars at the distance of M31. The SPLASH spectroscopic data are taken with the DEIMOS multi-object spectrograph on the Keck II 10-m telescope. The spectra allow us to identify stars belonging to the M31 system (Gilbert et al. 2006), analyze the velocity distributions of stars in each field, and identify debris features from disrupted satellite galaxies via their relatively small line-of-sight velocity dispersions (Guhathakurta 2006; Kalirai et al. 2006; Gilbert et al. 2007; Gilbert et al. 2009a,b).

We are using the SPLASH halo dataset (Figure 1) to learn about the global properties of M31's stellar halo, M31's surviving dwarf galaxy satellites (Majewski et al. 2007; Kalirai et al. 2009; Howley et al. 2008; Kalirai et al. 2010; Tollerud et al. 2012; Ho et al. 2012; Howley et al. 2013), and the ensemble of disrupted dwarf galaxies that built the halo.

Figure 1: The SPLASH Keck/DEIMOS spectroscopic fields in M31's stellar halo, overlaid on the PAndAS star count map (McConnachie et al. 2009). The spectroscopy targets fields on and off halo substructure, and cover a large range in radius (from Gilbert et al. 2014).

Global properties of M31's halo

We have shown that M31's stellar halo extends to at least 180 kpc in projection from the galaxy's center with no indication of halo truncation, even though we are now tracing it to two-thirds of the estimated virial radius (Gilbert et al. 2012). The data also show a clear gradient in metallicity that extends to 100 kpc (Gilbert et al. 2014).

This large-scale metallicity gradient, when compared to the results of simulations of stellar halo formation, implies that the bulk of M31's stellar halo was likely built primarily from one to a few relatively massive dwarf galaxies. We also observe significant field-to-field scatter in the mean metallicities and surface brightnesses of fields at large radius. This implies that recently accreted, small dwarf galaxies have contributed substantially to the outermost regions of M31's stellar halo.

To make more concrete statements about the luminosity function and time of accretion of the satellites that formed M31's stellar halo, we will need to make more detailed measurements of the chemical abundances of halo stars, including the ratio of alpha element abundances (Ne, Mg, Si, S, Ar, Ca, and Ti) and iron abundance. Vargas et al. (2014a,b) published the first [α/Fe] measurements of M31 stars (including four halo stars and measurements in nine dwarf galaxies). The SPLASH dataset provides a rich archive for measuring [α/Fe] in many more halo fields.

Leveraging Hubble imaging

The combination of deep, resolved Hubble imaging and ground-based stellar spectroscopy provides a powerful mechanism for enhancing our understanding of M31's stellar populations. One early example is the discovery that M31's giant stellar stream and prominent shell features are tidal debris from a single merger (Brown et al. 2006; Fardal et al. 2007; Gilbert et al. 2007).

We have extended the SPLASH spectroscopic survey to cover fields throughout M31's stellar disk (Figure 2; Dorman et al. 2012, 2013, 2015; Hamren et al. 2015). The vast majority of these spectroscopic masks overlap with the coverage of the PHAT survey (Figure 3; Dalcanton et al. 2012), a Hubble Space Telescope Multicycle Treasury Program that obtained contiguous ultraviolet (UV) to infrared (IR) imaging over one-third of M31's disk.

Figure 2: Spatial coverage of the PHAT imaging (white rectangles) and SPLASH spectroscopy (magenta rectangles)in M31's disk, overlaid on a GALEX UV image (from Dorman et al. 2015).

Figure 3: The full PHAT mosaic (left) and a detail showing the resolved stellar populations in M31's disk (right). The PHAT photometry has been used for spectroscopic target selection, as well as joint spectroscopic and photometric analyses of the properties of the various stellar populations in M31's inner regions.

The addition of Hubble PHAT imaging in M31's disk has proven to be a powerful tool in both obtaining and analyzing the stellar spectra. The PHAT imaging was used in selecting targets for spectroscopy, both to identify isolated stars in crowded regions and to target rare stellar populations in the existing survey. Spectroscopic measurements from the SPLASH survey were analyzed in conjunction with the photometric measurements from the PHAT survey to derive exciting results about the nature and origin of the stellar populations in M31's disk and inner halo described below.

The luminosity function of resolved stars in M31, as observed by PHAT, was used in conjunction with the stellar velocity distribution from SPLASH spectra and integrated-light surface photometry to model the strength of M31's stellar bulge, disk, and halo as a function of radius (Dorman et al. 2013). We found that the number of stars with a disk-like luminosity function is larger than the number of stars with disk-like kinematics. This is the first direct evidence that there are stars that were born in M31's disk and subsequently dynamically heated into the halo, resulting in a population of stars in the galaxy's inner regions that have a disk-like luminosity function, but a spheroid-like velocity distribution.

We have also used the colors and magnitudes of stars, obtained from PHAT photometry, to measure the velocity dispersion of M31's disk as a function of stellar age (Dorman et al. 2015). The observed line-of-sight velocity dispersion increases with the age of the stellar population, as has been observed in the Milky Way. However, the relationship between age and velocity dispersion in M31's disk is significantly steeper than that found in our Milky Way, and the average velocity dispersion of M31's stellar disk at young ages is significantly higher as well. Both findings lend yet more observational support to the hypothesis that M31 has experienced a more violent merger history than the Milky Way.

Next steps

The combination of large-scale, contiguous imaging of resolved stellar populations and resolved stellar spectroscopy has revolutionized our view of the M31 galaxy and its stellar populations. These data are providing us with a window into the formation and evolution of M31, together indicating that the galaxy has experienced an active merger history. With next generation instrumentation, including Webb, WFIRST, and extremely large ground-based telescopes, similar studies can be performed in more distant galaxies, thereby allowing us to determine just how unique (or common) such an active merger history is.


Brown, T. M., et al. 2006, ApJ, 652, 323

Dalcanton, J. J., et al. 2012, ApJS, 200, 18

Dorman, C. E., et al. 2013, ApJ, 779, 103

Dorman, C. E., et al. 2015, ApJ, 803, 24

Gilbert, K. M., et al. 2006, et al ApJ, 652, 1188

Gilbert, K. M., et al. 2007, ApJ, 668, 245

Gilbert, K. M., et al. 2009a, ApJ, 701, 776

Gilbert, K. M., et al. 2009b, ApJ, 705, 1275

Gilbert, K. M., et al. 2012, ApJ, 760, 76

Gilbert, K. M., et al. 2014, ApJ, 796, 76

Guhathakurta, P., et al. 2005, arXiv preprint, astro-ph/0502366

Guhathakurta, P. 2006, AJ, 131, 2497

Hamren, K. M., et al. 2015, ApJ, 810, 60

Ho, N., et al. 2012, ApJ, 758, 124

Howley, K. M., et al. 2008, ApJ, 683, 722

Howley, K. M., et al. 2013, ApJ, 765, 65

Kalirai, J. S., et al. 2006, ApJ, 641, 268

Kalirai, J. S., et al. 2009, ApJ, 705, 1043

Kalirai, J. S., et al. 2010, ApJ, 711, 671

Majewski, S. R., et al. 2007, ApJL, 670, L9

Tollerud, E., et al. 2012, ApJ, 752, 45

Vargas, L. C., et al. 2014a, ApJ, 790, 73

Vargas, L. C., et al. 2014b, ApJ, 797, L2