A NEW DISTANT MILKY WAY GLOBULAR CLUSTER IN THE PAN-STARRS1 3π SURVEY

As compact stellar systems that can be discovered at large distances, globular clusters (GCs) located in the outskirts of massive galaxies are valuable tracers of the hierarchical formation of their host (Searle & Zinn 1978; Law & Majewski 2010; Mackey et al. 2010; Pota et al. 2013). In particular, the detailed study of nearby GCs has shown that the most distant Milky Way (MW) GCs preferentially belong to the class of so-called "young halo" clusters (Mackey & van den Bergh 2005). These clusters are expected to have formed in dwarf galaxies (DGs) that were later accreted onto our Galaxy and destroyed by tidal forces (Dotter et al. 2011). Young halo GCs are preferentially younger (8–12 Gyr old), more metal-rich ([Fe/H] ∼ −1.5), and more extended than other halo GCs (Dotter et al. 2010). Similar conclusions are reached for some GCs in the outskirts of M31 (Mackey et al. 2013).

Although large sky surveys like the Sloan Digital Sky Survey (SDSS) have transformed our view of the MW satellite DG system (e.g., Willman et al. 2005; Belokurov et al. 2007), only a handful of GCs were discovered within the survey (Koposov et al. 2007; Belokurov et al. 2010; Balbinot et al. 2013); all of these are extremely faint (M_V ≳ −2.0) and within the inner ∼60 kpc of the halo. Additionally two other GCs within this spatial regime were found on sky survey plates (Pyxis and Whiting 1; Irwin et al. 1995, Whiting et al. 2002), yet no new distant MW GC has been discovered since the searches that led to the discovery of the Palomar clusters (e.g., Arp & van den Bergh 1960) and AM-1, the most distant MW GC to date at a galactocentric distance of ∼125 kpc (Madore & Arp 1979; Aaronson et al. 1984; Dotter et al. 2008).

Here, we report the discovery of the most distant MW GC, PSO J174.0675-10.8774, found in the Pan-STARRS 1 (PS1) 3π photometric survey. We argue that PSO J174.0675-10.8774 shares the properties of known young halo GCs. The Letter is structured as follows: in Section 2 we briefly describe the PS1 survey and the satellite search that led to the discovery of PSO J174.0675-10.8774. Section 3 focuses on the analysis of follow-up wide-field imager data. We derive the properties of the cluster in Section 4 and discuss their implication on the nature of the cluster in Section 5.

Whenever necessary, magnitudes are dereddened using the Schlegel et al. (1998) maps, assuming the extinction coefficients of Schlafly & Finkbeiner (2011). We also assume a heliocentric distance of 8 kpc to the Galactic center.

We note that the same stellar system was discovered independently by Belokurov et al. (2014) using VST ATLAS data. In their pre-print, these authors favor a DG classification for this object.

The PS1 3π Survey (K. Chambers et al., in preparation) targets three quarters of the sky (δ > −30°) in five photometric bands, g_P1, r_P1, i_P1, z_P1, y_P1, with the 1.8 m PS1 telescope, located in Haleakala, Hawaii (Tonry et al. 2012). The sky is surveyed with a 1.4 gigapixel camera covering a 3.3 degree field of view, which, combined with short exposures four times per year per filter over the course of 3.5 yr, yields a deep, panoptic view of the MW's surroundings. Once the individual frames are downloaded from the summit, they are automatically processed with the Image Processing Pipeline (Magnier 2006, 2007; Magnier et al. 2008) to generate a photometric catalogue. The PS1 survey is currently at the stage where stacked photometry and images have become available, reaching a similar depth to the SDSS in g_P1 and reaching deeper magnitude for r_P1 (∼0.5 mag) and i_P1 (∼1 mag; Metcalfe et al. 2013). The PS1 survey represents a significant increase of the search volume for small scale stellar systems that orbit the MW.

PSO J174.0675-10.8774 was discovered in an on-going search for small-scale substructures in the PS1 data. A comprehensive paper about this search is in preparation (B. Laevens et al., in preparation), but we provide here a broad outline of the discovery method. Inspired by previous searches for stellar overdensities (Koposov et al. 2007; Walsh et al. 2009), we adopt a hybrid method of the two aforementioned papers. First, we identify stars which have colors and magnitudes compatible with old, metal-poor stars possibly pertaining to a GC or DG by constructing color–magnitude masks from typical isochrones describing such systems and accounting for the possible distance to the system. The subsample of stars for a given distance is then convolved with Gaussian kernels. A positive convolution is performed on the data with a Gaussian of 4' or 8' dispersion, tailored to the size of typical MW satellites, while a convolution with a large negative Gaussian kernel accounts for the slowly-varying contamination (Koposov et al. 2009). After this two-step process, the resultant differential density maps track localized stellar over- and underdensities over the PS1 footprint, which we then translate into detection significance by comparison with the local density values. This led to the discovery of PSO J174.0675-10.8774 as an unambiguous 10σ detection.

At this location, the PS1 stacked images reveal the presence of a compact stellar system, which is emphasized by the distribution of PS1 sources on the sky. In panel a of Figure 1, the color–magnitude diagram (CMD) of this compact overdensity confirms that these stars are blue ((g_P1 − r_P1)₀ ≲ 0.8) and more numerous than in a field region (panel b) of the same coverage. Despite the detection of the object in the PS1 data, these are too shallow to reliably determine the properties of the system, which motivated us to gather deeper photometric data.

Zoom In Zoom Out Reset image size

During the night of 2013 December 31–2014 January 1, we obtained follow-up imaging with the Wide Field Imager (WFI) on the MPG/ESO 2.2 m telescope located in La Silla, Chile. Equipped with eight CCDs the camera has a field of view of 34' × 33'. Given the small angular size of PSO J174.0675-10.8774, we focus here on the photometry from CCD 1, which contains the full extent of the new stellar system. We obtained imaging in the B (five six-minute dithered sub-exposures), V, and R bands (three seven-minute dithered sub-exposures) in good seeing conditions with a median image quality of 09. The individual sub-exposures were bias-subtracted, flat-fielded, and cleared of cosmic rays. The initial astrometric solution for sub-exposures was obtained using the Astrometry.net package (Lang et al. 2010), with the PS1 catalog acting as the astrometric reference catalog. We then used the software packages SExtractor and SCAMP (Bertin 2006) to align each individual frame to the frame with the best seeing before constructing composite images for each band by average combining them with SWARP (Bertin 2006). The resulting BVR image of PSO J174.0675-10.8774 is shown in Figure 2 and unequivocally confirms that it is a compact stellar system.

Zoom In Zoom Out Reset image size

We performed the photometry using the daophot/allstar/allframe suite of programs (Stetson 1994) as described in Martin et al. (2013; E. J. Bernard, in preparation). A point-spread-function model was built for each individual frame and the flux was extracted by applying the model to fixed positions of stars as measured from the deep BVR image. The instrumental magnitudes were then calibrated to the first sub-exposure in a band, averaged out. We use the exquisite photometry of the single epoch PS1 data (Schlafly et al. 2012) to derive our own transformations from the WFI filters to the PS1 g_P1- and r_P1-band magnitudes.

The final WFI photometry reaches ∼1.5 mag deeper than the stacked PS1 data, as shown in panel c of Figure 1. The CMD within two half-light radii of PSO J174.0675-10.8774 (see Section 4 for the determination of the structural parameters) unveils a sparse red giant branch (RGB), a red horizontal branch (HB), and the beginning of the system's main sequence turn off at the magnitude limit of the data. All of these features are absent from a field region (panel d) of similar coverage shown for comparison. The spatial distribution of stars in the WFI data is shown in panel g.

A search for possible RR Lyrae stars varying in the multi-epoch PS1 photometry revealed no good candidate. However, although the CMD of PSO J174.0675-10.8774 is only sparsely populated, the presence of a well defined (red) HB allows for an accurate determination of the heliocentric distance to the system. We therefore measure the median apparent magnitude of the HB through a Monte Carlo resampling of the magnitude of stars in this part of the CMD, yielding m_gP1 = 21.41  ±  0.06. To determine the absolute magnitude of the red HB in this band, we extract from the PS1 stacked photometry database the photometry of four GCs that share the morphological properties and CMD of PSO J174.0675-10.8774 (Pal 3, Pal 4, Pal 5, and Pal 14). We repeat the same procedure on the red HB of these systems, shifted by their distance modulus as listed in Harris (2010). The weighted average of these four HBs' absolute magnitude measurements yields M_gP1 = 0.60 ± 0.10. The distance modulus of PSO J174.0675-10.8774 is therefore (m − M)₀ = 20.81 ± 0.12, yielding both a heliocentric and a Galactocentric distance of 145 ± 17 kpc.

Having pinned down the distance to the system, we can now estimate its age and metallicity through a comparison with a set of isochrones. Since the photometric data does not reach the main sequence of the cluster, these estimates should be taken with caution, but they nevertheless give a broad understanding of the age and metallicity of the cluster. Fixing the distance modulus at (m − M)₀ = 20.81, we cycle through the Parsec isochrones for a metallicity range −2.3 < [Fe/H] < −1.5 and an age between 8 and 13 Gyr (Bressan et al. 2012). The top-right panel (e) of Figure 1 presents the best fit to the data, obtained for the isochrone with an age of 8 Gyr and [Fe/H] = −1.9. Allowing the distance to vary within the formal distance uncertainties, we obtain a metallicity and age range of 8–10 Gyr and [Fe/H] varies between −1.5 and −1.9. These are fairly typical properties of young, outer halo GCs and this impression is further bolstered by panels f and h of Figure 1, which present literature photometry (Saha et al. 2005) of Pal 14 (11.3 Gyr, [Fe/H] = −1.5; Dotter et al. 2011) and Pal 4 (10.9 Gyr, [Fe/H] = −1.3) shifted to the distance of the new GC. The CMDs of the three GCs exhibit similar features, with sparsely populated RGBs and red HBs.

To determine the structural parameters of PSO J174.0675-10.8774, we use a variant of the technique presented in Martin et al. (2008), updated to allow for a full Markov Chain Monte Carlo treatment. Briefly, the algorithm uses the location of every single star in the WFI data set to calculate the likelihood of a family of radial profiles with flattening and a constant background. The parameters are: the centroid of the system, its ellipticity⁸, the position angle of its major axis from N to E, the number of stars in the system, and one or two scale parameters. We use three different families of radial density models (exponential, Plummer, and King), for which the scale parameters are the half-light radius, the Plummer radius, and the core and King radii, respectively. The resulting structural parameters are listed in Table 1. Figure 3 also shows the probability distribution function (pdf) of the exponential model parameters, as well as the comparison between the favored radial distribution profile and the data, binned with the favored structural parameter model. They show a very good agreement, testament to the quality of the structural parameter's inference. The results for the Plummer profile and the King profile are equally good. From the determination of the structural parameters, we find that PSO J174.0675-10.8774 is a round, compact system with a half-light radius of only $0\space'\hspace{-0.5em}. 47^{\,+\,0.04}_{-0.03}$ or 20 ± 2 pc at the distance of 145 ± 17 kpc.

Zoom In Zoom Out Reset image size

Table 1. Properties of PSO J174.0675-10.8774

α (J2000)	11:36:16.2
δ (J2000)	−10:52:38.8
ℓ	2748
b	+478
Distance modulus	21.81 ± 0.12
Heliocentric distance	145 ± 17 kpc
Galactocentric distance	145 ± 17 kpc
MV	−4.3 ± 0.2
[Fe/H]	∼−1.9
Age	∼8 Gyr
E(B − V)a	0.026
Exponential profile
Ellipticity	$0.00^{+0.10}_{-0.00}$
Position angle (from N to E)	$-56^{+88}_{-68}$°
rh	$0.47^{+0.04}_{-0.03}$'
	20 ± 2 pc
Plummer profile
Ellipticity	$0.0^{+0.11}_{-0.00}$
Position angle (from N to E)	−64° ± 64°
rh	052 ± 004
	22 ± 2 pc
King profile
Ellipticity	$0.00^{+0.11}_{-0.00}$
Position angle (from N to E)	$+70^{+48}_{-36}$°
rc	061 ± 018
	$24^{+9}_{-6}{\rm \,pc}$
rt	$1.30^{+0.19}_{-0.12}$'
	$56^{+9}_{-6}{\rm \,pc}$

Note. ^aFrom Schlegel et al. (1998).

Download table as:  ASCIITypeset image

The absolute magnitude of PSO J174.0675-10.8774 is also derived using the technique presented in Martin et al. (2008), which accounts for "CMD shot-noise" that stems from the impact of the sparsely populated CMD on the derivation of its total magnitude. Using the favored Parsec isochrone presented above (8 Gyr, [Fe/H] = −1.9) and its associated luminosity function shifted to the distance of the cluster, we build a color–magnitude pdf of where PSO J174.0675-10.8774 stars should lie given the photometric uncertainties. We then populate a mock CMD from this pdf until the number of stars brighter than r_P1 = 23.5 equals the number of stars, N*, that we determine from the estimate of the structural parameters for the same region of the CMD. Summing up the flux of all the stars in the mock CMD yields the absolute magnitude of this realization. Finally, we iterate this procedure 300 times, accounting for the uncertainties on the heliocentric distance and on N*, as well as the shot-noise uncertainties that come from randomly populating the mock CMDs. This procedure yields absolute magnitudes of M_g = −4.02 ± 0.22 and M_r = −4.49 ± 0.25, which converts to M_V = −4.3 ± 0.2 when using the Tonry et al. (2012) color transformations. The total luminosity of the system is 4.5 ± 0.7 × 10³ L_☉, which, assuming a typical mass-to-light ratio in solar units of ∼1.5 for GCs of this metallicity (e.g., Strader et al. 2011), yields a total stellar mass of ∼6.8 ± 1.1 × 10³ M_☉.

We have presented a new stellar system found within the PS1 3π data in the outer halo of the MW. Located at a heliocentric distance of 145 ± 17 kpc, this system is rather faint (M_V = −4.3 ± 0.2), compact (r_h = 20 ± 2 pc), round ($\epsilon =0.0^{+0.10}_{-0.00}$), younger than the oldest GCs (∼8 Gyr), and fairly metal-poor ([Fe/H] ∼ −1.9). Figure 4 compares the properties of PSO J174.0675-10.8774 with MW GCs and DGs. Although it is slightly more distant than any known MW GC, all the other properties of this system are similar to those of outer halo GC systems (e.g., Mackey & Gilmore 2004; Mackey & van den Bergh 2005). In particular, the size and the roundness of the system differentiate it from DGs at the same magnitude, which are all larger than at least r_h = 60 pc and tend to favor elliptical radial density profiles (Martin et al. 2008; Sand et al. 2012). Furthermore, the age and metallicity we determine from the comparison with isochrones are quite typical of young, outer halo GCs, as made evident by the direct comparison of PSO J174.0675-10.8774's CMD with that of Pal 4, and Pal 14 (Figures 1(e), (f), and (h)).

Zoom In Zoom Out Reset image size

Beyond our local environment, this new stellar system is also similar to the population of extended GCs recently found in the vicinity of M31, many of which lie on faint stellar streams (Huxor et al. 2005; Mackey et al. 2010). Those systems are similarly round, rather fluffy for GCs, but compact for DGs, and can also exhibit red HBs (e.g., Mackey et al. 2006, 2013; Huxor et al. 2011). The discovery of this most distant MW GC supports the idea that very remote GCs are a common feature of large spiral galaxies as has been shown in M31 (Huxor et al. 2011), in M33 (Cockcroft et al. 2011) and M81 (Jang et al. 2012). Further investigation in the vicinity of this object would therefore be interesting in the context of stellar streams and their association with GCs.

After the work for this Letter had been completed, we learned of the independent discovery of this object by Belokurov et al. (2014) from the VST ATLAS survey and follow-up data. Their determination of the system's properties are consistent with ours, even though they derive a slightly larger distance (∼170 kpc) and size (r_h ∼ 06 = 30 pc at their distance). Their interpretation of the nature of the system nevertheless differs from ours as they favor a scenario in which the system is a DG. Their conclusion is partly driven by the larger size they measure, but also by their interpretation of a handful of blue stars being blue loop stars indicative of recent star formation. We do indeed retrieve these stars in our data set; however, we are cautious as to their interpretation. Performing our structural parameter analysis on these blue stars only yields a detection of an overdensity at the ∼2σ–3σ detection level. Furthermore, if these were truly blue loop stars, one would expect the presence of a higher number of their low-mass analogs at fainter magnitudes and bluer colors, which do not appear in the CMD. Finally, it cannot be ruled out that the two blue stars residing barely above the red HB could in fact be asymptotic giant branch stars. Taking into account the derived properties of this stellar system, we rather conclude that we are in the presence of a distant GC and not of a faint and extremely compact DG. We stress that this conclusion must be confirmed or infirmed with radial velocities through follow-up spectroscopy of these blue loop stars, thus ascertaining the definite nature of this new distant satellite.

We would like to thank the editor and referee for their celerity. We wish to thank P. Bianchini, M. A. Norris, and D. Martinez-Delgado for fruitful discussions about globular clusters and R. A. Ibata for a careful reading of the manuscript. B.P.M.L. acknowledges funding through a 2012 Strasbourg IDEX (Initiative d'Excellence) grant, awarded by the French ministry of education. N.F.M. and B.P.M.L. gratefully acknowledge the CNRS for support through PICS project PICS06183. N.F.M., H.-W.R. and E.F.S. acknowledge support by the DFG through the SFB 881 (A3). E.F.B. and C.T.S. acknowledge support from NSF grant AST 1008342.

The Pan-STARRS1 Surveys (PS1) have been made possible through contributions of the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, the Johns Hopkins University, Durham University, the University of Edinburgh, Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation under grant No. AST-1238877, the University of Maryland, and Eotvos Lorand University (ELTE).