CORDIS - Forschungsergebnisse der EU

The dynamical evolution of Globular Clusters: From Brown Dwarfs to Blue Stragglers

Final Report Summary - BDINGC (The dynamical evolution of Globular Clusters: From Brown Dwarfs to Blue Stragglers)

Globular clusters (GCs) are the oldest stellar systems in the Milky Way and comprise up to a few million stars at all evolutionary stages. As such, GCs are the ideal natural laboratories to study large, co-eval stellar populations at a known distance, age and chemical composition. The main objectives of this project are:

(1) to study the lowest mass main sequence (MS) stars and detect for the first time brown dwarfs (BDs) in a Globular Cluster,
(2) to study the formation and evolution of exotic stellar populations,
(3) to investigate the dynamical evolution of GCs by means of N-body simulations.


Much of our understanding of star formation and evolution has been derived from observational studies of GCs. However, we still lack an understanding of the lowest-mass stars around the Hydrogen-burning limit and of sub-stellar objects that are not massive enough to sustain Hydrogen burning, so called "Brown Dwarfs" (BDs). BDs present a link between the lowest mass stars and planets and thus are vital for our understanding of both stellar and planetary formation and evolution. BDs have complex atmospheres, like giant gas planets, but their formation might be more similar to stellar formation (e.g. Elmegreen 1999, ApJ, 522, 915). However, the formation of BDs is a matter of considerable dispute: they might form just like stars, or via dynamical processes (disk fragmentation, ejection of stellar embryos), or from photo-evaporation of protostars by nearby massive stars (e.g. Reipurth & Clarke 2001, AJ, 122, 432; Kroupa & Bouvier 2003, MNRAS, 346, 369; Thies et al. 2010, ApJ, 717, 577). These processes prevent protostellar clumps to accumulate enough mass to ignite Hydrogen, resulting in BDs. The latter scenarios might also suggest an increase of BDs in massive and/or dense clusters like GCs, because more massive clusters also have more O-type stars (massive stars with strong stellar winds) which can produce more BDs via photo-evaporation, and dynamical interactions are increased in dense clusters which might lead to more disk fragmentation and ejection of sub-stellar clumps from protostars.
Nearly all BDs detected so far are either young and metal-rich, or their age and chemical composition are unconstrained (see e.g. In fact, determining the physical properties of such cool objects is a major hurdle in BD research. On the other hand, we need metal-poor and old BDs if we are to test theoretical models on BD and star formation, atmospheres and evolution.
GCs are massive and thus might have produced BDs in large numbers. They are also the oldest and most metal-poor stellar aggregates in our Milky Way. Thus, GCs are the ideal hunting grounds for old metal-poor benchmark BDs (and lowest-mass stars).

At the start of this research project, no BDs had been detected in any GC at all. We used near-infra-red (NIR) observations of the GC M4, obtained with the Hubble Space Telescope. The best-photometry NIR colour magnitude diagram (CMD) of M4, see Fig. 1, is the deepest NIR CMD of a GC to date and clearly shows the MS extending beyond the Hydrogen-burning limit towards fainter sources. A white dwarf (WD) sequence can also be identified and crosses the extension of the MS. Archival ultra-deep optical data have been used to proper-motion clean the CMDs and to distinguish cluster members from field sources, and cluster WDs from BD candidates, see Fig. 2. Based on theoretical models (Fig. 3), we conclude that we have reached beyond the H-burning limit in our NIR CMD and are probably just above or around this limit in our optical-NIR CMDs. We found in total four sources which are all clear detections in the NIR, but no detection in the optical, see Fig. 4. We used the optical data to estimate an upper magnitude limit for these four sources, and found that all four are good BD candidates, see Fig. 5. These are the first BD candidates detected in a GC. As such, this project proved to be a pioneering and ground-breaking work, and has established European leadership in this particular research area.


Stellar densities in the cores of GCs can be extremely high, so that dynamical interactions between the cluster members are ubiquitous, boosting the formation of new and exotic stellar sources like blue stragglers (BSs), cataclysmic variables (CVs) and other exotica.
BSs appear as an extension of the MS and have been found in essentially all GCs. They occupy a region in the CMD that should be devoid of stars, according to normal stellar evolution. They are thought to be the merger product of two or more single main sequence stars, either via direct collisions, or via the evolution of primordial binaries (McCrea 1964, MNRAS, 128, 147; Hills & Day 1976, ApL 17, 87). Which formation channel dominates is still a matter of dispute.
CVs contain a WD accreting from a low-mass star. CVs can be used as tracers of the close binary (CB) population in a GC. CBs, in turn, are important for the dynamical evolution of the GC as they heat the cluster and cause cluster expansion, preventing or delaying core collapse. Theory predicts large numbers of CVs in GCs (e.g. di Stefano & Rappaport 1994, ApJ, 423, 274; Ivanova et al. 2006, MNRAS, 372, 1043), but the numbers detected observationally are still considerably smaller. Theoretically, also WDs accreting from a red giant (so called symbiotic stars), or accreting from another low-mass WD donor (AM CVn stars) can be expected, however, no such system had been discovered in a GC prior to the start of this project.
Most exotic stellar populations emit most of their light in the far- or near-ultraviolet (FUV or NUV). In addition, MS stars and red giants, which make up the bulk of the cluster population, are much fainter at such short wavelengths, and as a consequence FUV images are considerably less crowded than optical images. Interacting CBs like CVs are also strong X-ray emitters, so that the combination of both X-ray and FUV observations is an ideal approach to detect and study the interacting CBs in GCs.

During the course of this research project, the two GCs NGC 6397 and NGC 1851 have been investigated using FUV observations obtained with the Hubble Space Telescope.
In NGC 1851, we detect an exotic FUV bright variable star with an 18.05 minutes period, see Fig. 6. This source could either be an intermediate polar (a CV with a moderate magnetic field) or an AM CVn star (an interacting system consisting of two WDs). However, intermediate polars are also X-ray sources, but no X-ray emission has been detected for this system. We thus suggest that this source is an AM CVn binary, the first detected in a GCs.
We present the first FUV-NUV CMD for NGC 6397, see Fig. 7. A NUV-V CMD is presented in Fig. 8. Both CMDs were used to distinguish the various stellar populations. BSs and CVs are the most centrally concentrated populations, whereas the horizontal branch (HB) stars are the least concentrated population (Fig. 9). In fact, HB stars are absent in the innermost 5.5 arcseconds. The central concentration of BSs and CVs reflects their larger masses, compared to the smaller masses of the more widely distributed HB stars. The preferred location of the BSs and CVs towards the cluster centre might be an effect of mass segregation, i.e. more massive stars travel towards the cluster centre, and/or of their formation, as BSs and CVs likely form via dynamical processes in the dense cluster centre.

(3) N-body simulations:

The main objective of the N-Body simulations is to study the distribution of BDs and BSs as the cluster evolves. For this purpose, we used the NBODY6-GPU code (Aarseth 1999, PASP, 111, 1333; Nitadori & Aarseth 2012, MNRAS, 424, 545) and assumed an initial mass function as described in Thies & Kroupa (2007, ApJ, 671, 767), ranging from 0.01 to 5 solar masses, a Plummer model to describe the cluster profile, and a half-mass radius of 1 pc at the start of the simulation. BDs were introduced as a separate population with masses 0.01 to 0.15 solar masses (to account for BDs formed via disk fragmentation), while initial masses of stars range from 0.07 to 5 solar masses. We further assumed a number ratio of BDs to stars of 1:5 at the start of the simulations (Thies & Kroupa, 2007, ApJ, 671, 767). We varied the initial number of bodies, ranging from 50 000 to 200 000 bodies. The simulations run for a simulated time frame of 10 Gyr.

Simulations for the 50 000, 75 000 and 100 000 clusters have finished, and the distribution of BDs, stars and WDs were analysed. The analysis regarding the BS populations and their evolution still needs to be done and might require additional simulations that go beyond the funding period of this project. The 200 000 source simulation is still running and will continue to do so until 10 Gyr are reached in the simulation.
Fig. 10 shows snapshots of the spatial distribution of the 100 000 bodies cluster at different times. As can be seen, at the start of the simulations the distribution of BDs follows the distribution of stars, but towards the end of the simulations (at 10 Gyr) the number of BDs has considerably declined. Most BDs have evaporated from the cluster and only a few BDs are left. The simulation also suggests that the very core of the cluster is largely devoid of BDs, so the best place to search for BDs in GCs is not the core region, but also not the very outskirts of the cluster, as most BDs will have been ripped away by the galactic tidal field. On the other hand, the number of WDs increases with time. This is expected, because WDs are the end product of stellar evolution (for stars less massive than 8 solar masses, more massive stars will produce a neutron star or a black hole).
The radial distribution of stars, WDs and BDs of the 100 000 bodies simulation is plotted in Fig. 11, and again suggests that the cluster core becomes largely devoid of BDs as the cluster evolves.


During the funding period of this research project, BD candidates were detected in a GC for the first time. This shows that the search for BDs in GCs is feasible and can provide the missing metal-poor benchmark sources that can then be used to calibrate theoretical models. Further observations with the Hubble Space Telescope have been awarded and will be carried out in 2017. Thus, the research fellow will continue with and expand on this project and will also further establish European leadership in this particular research area.
The presented N-body simulations are the first simulations that include BDs, and provide insight into the dynamical evolution of various stellar populations from stars, WDs and - for the first time - for BDs. The numbers obtained from the series of N-body simulations need to be scaled, so that predictions about the expected number and location of BDs in GCs, which are considerably more massive, can be made. This is essential in planning and designing future observations.
FUV studies of GCs are an ideal tool to study exotic stellar populations, and each study detects new and unusual objects. We detected a likely AM CVn star in a GC for the first time. Also, our finding that HB stars are absent from the very core region of the cluster NGC 6397 has triggered a new project that will look into the dynamical evolution of HB stars in GCs, led by the research fellow.

Verwandte Dokumente