MNRAS 000,1–7 (2021) Preprint 18 June 2021 Compiled using MNRAS LATEX style file v3.0 No globular cluster progenitors in Milky Way satellite galaxies Pierre Boldrini1¢ and Jo Bovy2 1Sorbonne Université, CNRS, UMR 7095, Institut d’Astrophysique de Paris, 98 bis bd Arago, 75014 Paris, France 2Department of Astronomy and Astrophysics, University of Toronto, 50 St George Street, Toronto, ON M5S 3H4, Canada 18 June 2021 ABSTRACT In order to find the possible progenitors of Milky Way globular clusters, we perform orbit integrations to track the orbits of 151 Galactic globular clusters and the eleven classical Milky Way satellite galaxies backward in time for 11 Gyr in a Milky- Way-plus-satellites potential including the effect of dynamical friction on the satellites. To evaluate possible past associations, we devise a globular-cluster–satellite binding criterion based on the satellite’s tidal radius and escape velocity and we test it on globular clusters and satellites associated with the Sagittarius dwarf and with the Large Magellanic Cloud. For these, we successfully recover the dynamical associations highlighted by previous studies and we derive their time of accretion by the Galaxy. Assuming that Milky Way globular clusters are and have been free of dark matter and thus consist of stars alone, we demonstrate that none of the globular clusters show any clear association with the eleven classical Milky Way satellites even though a large fraction of them are believed to be accreted. This means that accreted globular clusters either came in as part of now-disrupted satellite galaxies or that globular clusters may have had dark matter halos in the past – as suggested by the similar metallicity between globular clusters and dwarf galaxies. Key words: galaxy dynamics - methods: orbital integrations - globular clusters - Milky Way - dwarf galaxies 1 INTRODUCTION possible to assess whether GCs formed in the MW or in a satellite galaxy that was later accreted. The origin of compact stellar clusters, also known as globular clusters (GCs), is one of the open questions in modern astrophysics. Many It was found that 62 of the present-day MW GCs likely formed years now since their first discovery, it is still unclear whether they in the MW, ∼55–65 have an extragalactic origin and the rest likely formed within dark matter minihalos or as gravitationally bound has a more heterogeneous origin (Kruijssen, et al. 2019; Massari et clouds in the early universe (Peebles & Dicke 1968; Kravtsov & al. 2019). It was established that the main contribution to the MW Gnedin 2005; Kruijssen 2015; Peebles 1984; Peñarrubia et al. 2017). came from three major accretion events: Sagittarius (Ibata, Gilmore Despite ongoing debates about their origin (Kruijssen 2015; Forbes & Irwin 1994; Law & Majewski 2010), Gaia-Enceladus (Belokurov, et al. 2018), these gravitationally-bound groupings of mainly old et al. 2018; Helmi, et al. 2018), and Sequoia (Myeong, et al. 2019). stars also have a similarly long history of being used as probes of In fact, numerous pieces of evidence indicate that 35% of the clusters the galaxy formation and assembly process, particularly in the Milky are possibly associated with these merger events and then are formed Way (MW) (Searle & Zinn 1978; Dinescu, Girard & van Altena in accreted dwarf galaxies (Leaman, VandenBerg & Mendel 2013; 1999; Brodie et al. 2006). Massari et al. 2019). It has long been recognized that the accretion of satellite galaxies A particularly intriguing set of GCs is the population of eleven has contributed to the growth of the MW (Searle & Zinn 1978)and the loosely bound GCs or at high energies in the MW as defined by Galaxy has accreted around one hundred satellite galaxies (Bland- arXiv:2106.09419v1 [astro-ph.GA] 17 Jun 2021 Massari et al.(2019). These GCs (AM1, Eridanus, Pyxis, Palomar Hawthorn et al. 2016). As a natural result of such events, GCs may 3, Palomar 4, Crater, NGC 6426, NGC 5694, NGC 6584, NGC 6934 have been accreted by the MW (Peñarrubia, Walker & Gilmore 2009). and Palomar 14) do not seem to have formed in situ in the Galaxy Indeed, several authors have used metallicities and horizontal-branch (Massari et al. 2019). In the E-MOSAICS simulations, high-energy morphologies to distinguish GC origin; those formed in satellite GCs are generally old and metal-poor and then it may not be pos- galaxies and those formed in situ within the Galaxy (Zinn 1999; van sible to unambiguously determine their origin (Pfeffer, et al. 2020). den Bergh 1993; Mackey & Gilmore 2004). Some of the 11 GCs have previously been tentatively associated with Recently, full six-dimensional phase-space information for almost MW satellite galaxies such as Fornax, Sculptor, or the Large Magel- all of the Galactic GCs from the second data release of the Gaia lanic Cloud (Irwin, Demers & Kunkel 1995; Koposov, et al. 2014). mission has offered unique insights into their dynamics (Gaia Col- Intriguingly, Fornax is the only one of the classical dwarfs to have laboration, et al. 2018; Vasiliev 2019). Thanks to these data, it is now six GCs in orbit (Wang, et al. 2019), thus demonstrating that most classical dwarfs can have hosted GCs in the past that now have been stripped. Indeed, as the 11 loosely-bound GCs reach far distances from the MW centre over their history, they could have interacted ¢ Contact e-mail: [email protected] with or belonged to MW satellites. © 2021 The Authors 2 P. Boldrini and J. Bovy In this paper, we examine if the eleven satellite galaxies of the dSph Mmin Mmax tinfall 9 9 MW could be progenitors of some of the 151 GCs. We follow the dy- [10 M ][10 M ] [Gyr] 0 namical history backward in time of these GCs in the MW+satellite LMC 100 198.8 4.0 0 environment by using orbital integration methods. The paper is or- SMC 26 73.9 4.0 Sgr 14 62 7.481 ganized as follows. Section 2 provides a description of our orbital- UMi 0.79 2.8 10.72 integration method. In Section 3, we discuss how we decide whether Draco 3.16 3.5 10.42 or not a GC was associated with a satellite galaxy in the past and Sculptor 1.99 5.7 9.92 present our results on associations between GCs and the Sagittarius Sextans 0.316 2.0 8.42 dwarf, satellites of the Large Magellanic Cloud and the Cloud itself, Leo I 1.99 5.6 2.32 and finally all Milky-Way GCs and the classical satellites. Section 4 Leo II 0.316 1.6 7.82 presents our conclusions. Carina 0.398 0.8 9.92 Fornax 0.79 21.9 10.72 Table 1. MW satellite properties: From left to right, the columns give for 2 METHOD each classical satellites: the current mass "200 (Errani, Peñarrubia & Walker 2018) and the pre-infall mass "pi (Read & Erkal 2019); the infall time from: In order to find the possible progenitors of the 151 Galactic GCs, (a) Rocha, Peter, & Bullock(2012), (b) Dierickx & Loeb(2017) and (c) we take their present-day phase-space position derived from Gaia (Fillingham, et al. 2019). DR2 data (Vasiliev 2019) and integrate them backward in time for 11 Gyr in a model for the Milky Way’s gravitational potential. It was recently demonstrated that interactions with MW satellites can orbits backwards in time for 11 Gyr in MWPotential2014 by ap- cause the orbital properties of Galactic GCs to evolve significantly plying dynamical friction to them with a constant mass of Mmin or over time (Garrow, Webb, & Bovy 2020). Therefore, we include the Mmax, given their current positions and proper motions from Fritz, gravitational potential of the eleven classical satellites in the potential et al.(2018). Dynamical friction is implemented by following the model (see Table1). semi-analytic model of Petts, Gualandris, & Read(2015). The GCs We assume that the main component of the Galactic potential is are then integrated in the combined MWPotential2014 plus the well represented by the MWPotential2014 mass model from Bovy time-dependent potential of each of the 11 classical satellites. (2015). Much of the stellar halo was likely accreted between 9 and 11 A mass of 6 × 106 M corresponds to the upper limit of the Gyr ago (Di Matteo, et al. 2019; Mackereth & Bovy 2020). Moreover, stellar mass of Galactic GCs 12 Gyr ago (Baumgardt, et al. 2019). cosmological simulations of the formation of Milky-Way-like galax- The time to change the GC apocentre substantially due to dynamical ies show that the main halo mass, which grows with time, reaches its friction is ≈ "enclosed/"GC ×Cdyn where "enclosed is the host galaxy asymptotic value 9-10 Gyr ago (Diemand, Kuhlen & Madau 2007; mass enclosed within the orbit and Cdyn is the orbital time (Binney Lux, Read & Lake 2010). As a consequence, we establish that our & Tremaine 2008). As the mass ratio between MW enclosed mass model describes the Galaxy well until 9-10 Gyr ago. and GC masses is large, dynamical friction is inefficient over our For the satellites, we assume a Hernquist(1990) density profile timescale (11 Gyr). We thus neglect the mass loss to the clusters for the DM component. Given the halo mass and redshift, both halo assuming that the GC mass remains constant during integrations. concentrations 2200 can be estimated from cosmological #-body Orbit integrations are performed with a time step of 10 Myr us- simulations (Dutton & Macciò 2014). We neglect the stellar contri- ing the publicly available code galpy1 (Bovy 2015) by applying bution of satellites as their dynamics is largely governed by the dark dynamical friction to the clusters and dwarfs.