Formation of globular clusters in a cosmological context
Item statusRestricted Access
Embargo end date12/01/2024
Globular Clusters (GCs) are among the oldest gravitationally bound stellar systems, as they are characterised by stellar populations with ages of 11.5 to 12.5 Gyr. They probably formed during, or just after, the epoch of reionisation. Therefore, the study of these systems at high redshift can not only give insight into the processes and environments that governed star formation at that time, but they can also represent valuable probes of the reionisation epoch itself. However, the exact phases of GC formation are still unknown. Due to their compact sizes, the ability to directly observe the ancestors of present-day GCs at high redshift is still difficult; thus, we must turn to cosmological simulations. In this doctoral thesis, I utilised a suite of numerical simulations of a cosmological volume from the First Billion Years (FiBY) Project and I also ran a set of simulations of isolated galaxy environments down to z ∼ 2. As a first step, I performed an “agnostic exploration” of the demographics of low-mass stellar systems within the FiBY simulations. Specifically, I identified a population of dense objects with a stellar component of up to a hundred thousand solar masses and such that their fraction of baryons (over the total mass of baryons and dark matter) is close to unity; these objects are possible “infant GC candidates”. Compared to their present-day counterparts in the local Universe, these infant stellar clusters have a higher gas fraction and appear to be more extended, consistent with the properties of the few examples of proto-globular clusters currently observed at high redshift. They are compact systems, with densities higher than the average population of FiBY systems at the same stellar mass. The infant GC candidates appear to be more massive and more abundant in massive host galaxies, indicating that the assembly of galaxies via mergers may play an important role in building several GC-host scaling relations. I assessed the present-day GC mass (GC number) – halo mass relation and found it offers a satisfactory description of the behaviour of the infant GC candidates at high redshift, suggesting that such a relation may be set at formation. I also reported a new relation between the specific star formation rate of a host galaxy and the mass of the most massive GC that holds across a large range of redshifts, thus providing a way to probe both the evolution of GCs as well as their host galaxy across time. As a second step, I examined the origin of these infant GC candidates through the use of energy-angular momentum phase-space characterisation to identify bound structures within the simulations. By analysing the evolution in phase-space, I determined the redshift at which the infant GCs first became gravitationally bound. By studying the collapse of both the gaseous and stellar components of the objects, I found evidence that they undergo radial infall and that the collapse itself is driven by the internal self-gravity of the infant GC candidates. The phase-space characterisation also revealed that some infant GC candidates shared certain features in the energy-angular momentum plane, allowing them to be grouped by their behaviour in phase-space. This classification is reflected in the formation environment: infant GC candidates of the same phase-space class are found in hosts of similar morphology, the majority of which are located in clumpy, irregular proto-galaxies. Within the set of infant GC candidates studied, there are two which consist only of stars by z = 6. By analysing these particular objects, I found that supernova feedback is the main physical mechanism behind expelling the gas and halting star formation. These candidates best represent the local Universe GCs and provide a possible future for the gas-rich infant GC candidates beyond z = 6. Finally, I explored the subsequent evolution of these infant GC candidates within their host galaxies by performing a new set of isolated, cutout numerical simulations evolved to redshift z ∼ 2. These simulations were designed to focus on the three main groupings of infantGCcandidates and to investigate whether theywould survive within their hosts, without the influence of large-scale external factors. Three host galaxies were selected, containing a total of 11 infant GC candidates; most of these candidates are gas-rich, with a single object being composed only of stars by z = 6 (the time at which the previous FiBY simulations ended). I found that all gas-rich candidates are destroyed: their star formation activity peaks at z ∼ 6 and, once the supernovae release their feedback and expel the gas, the remaining stars in the candidates can not adjust to the new potential and the objects dissolve. By comparing the destruction of the gas-rich candidates with the history of the star-only candidate (which had previously lost its gas due to supernovae as well, but survived), I found that a key factor in whether a candidate would survive the feedback phase is the ratio of gas mass to stellar mass. If this factor is too large, then the survival of the candidate is unlikely. This observation is further reinforced when comparing this ratio with the timescale for dissolution: on average, objects with a lower gas to stellar mass fraction live longer. To complete this analysis, I examined the evolution of the star-only candidate and its trajectory around its host galaxy’s barycentre. I found that the object is on an infalling orbit towards the centre of the host and, eventually, it begins to merge with the central part of the galaxy, in agreement with the GC infall formation scenario for nuclear star clusters.