Emmy Noether Group: The formation and evolution of starburst clusters

I. Phyikalisches Institut, University of Cologne
PI: Dr. Andrea Stolte

The group acknowledges funding by the German Science Foundation's Emmy Noether programme.

Topic

A comprehensive high-resolution study of the formation and evolution of resolved starburst clusters,
and the formation environment of the highest mass stars.

Summary

One of the most heavily debated, and extensively used, universal quantities in astrophysics is the initial stellar mass function (IMF). The IMF describes the relative numbers of high- vs. low-mass stars in any given star-forming region. Any stellar population observed today is a direct result of the IMF during its formation and the subsequent evolutionary history of the stellar population. In the solar neighbourhood, the IMF is surprisingly universal. Such universality is unexpected given that gaseous and stellar densities, pre-stellar core temperatures, and magnetic fields in molecular clouds vary significantly. One reason for this apparent similarity in the local stellar mass spectrum may originate from the similar conditions of star-forming clouds in the solar environment. However, this environment is not characteristic for the extreme starburst regions in external galaxies, in galactic nuclei, and in dense stellar clusters. The closest environments where stars form under enhanced temperatures, gas densities, and possibly magnetic pressures are provided in Milky Way starburst clusters. At the present epoch, these young, massive star clusters harbour 10,000 solar masses in stars or more, and are formed in rare numbers in the Milky Way's spiral arms and near the center of the Galaxy. The Emmy Noether group at the University of Cologne studies resolved starburst clusters in the Milky Way to answer the universal question as to whether their initial stellar mass functions are similar to or deviate from the IMF in the solar neighbourhood. The expected results provide the basis to understand the multitude of massive, unresolved extragalactic starburst clusters, where the massive stars contribute the majority of the light, while the total mass may be dominated by the low-mass stellar content in each cluster. Massive star clusters comprise one major building block of a galaxy's stellar population. In the Milky Way, the few known young, massive starburst clusters constitute the densest star-forming environments outside the Galactic nucleus, and provide unique testbeds for star formation in massive extragalactic clusters. The scientific aim of the project is to advance the understanding of resolved starburst clusters, their formation, survival timescales and dissolution, and the build-up of a galaxy's field population. Emphasis is placed on the differences in the formation and evolution of clusters in spiral arms and near the Galactic center. Proper motion membership, radial velocities and spectral classification from high-angular resolution observations yield the unbiased stellar mass distribution. The internal dynamical structure and global motion is derived, constraining cluster orbits and the dispersal of cluster members in the Galactic potential, and ultimately the existence of intermediate-mass black holes in the cluster cores. Orbital simulations of clusters evolving in the Galactic center tidal field constrain formation scenarios and tidal mass loss, and provide the basis for the comparison with extragalactic nuclei. The small number of only five resolved starburst clusters in the Milky Way makes it particularly exciting to use new-generation telescopes such as the LBT to extend the sample to extragalactic starburst clusters in nearby galaxies such as M31/M33.

The most puzzling questions in star cluster formation

Starburst clusters contain between tenthousand and one million stars in a unique single-age, single-metallicity population. The most striking characteristic of starburst clusters in the Milky Way and in nearby galaxies is a very compact core containing the highest stellar mass density measured outside the galactic nucleus. These starburst cores are the formation locus of the most massive stars known in the present-day universe. At the same time, the compactness of these cores implies short dynamical timescales, causing the compact clusters to dissolve rapidly. Only very few starburst clusters in our immediate vicinity, the Milky Way and the Magellanic Clouds, are observed today. The substantial extinction along the line of sight towards Milky Way starburst clusters and the high stellar densities in their cores restricted the study of their resolved stellar population until recently, when space- and ground-based near-infrared, high-angular resolution techniques became available. While those clusters in our environment are rare, they provide the tools to understand the substantial extragalactic population of young, compact star clusters as observed in the nuclei and HII regions of starforming galaxies, and the tidal interaction zones of mergers, such as the Antennae.

The Emmy Noether resarch project addresses the most puzzling questions arising from starburst clusters today: i) With the numerous giant molecular clouds observed in the Milky way, why are starburst clusters so rare in our Galaxy? What triggers their formation? ii) Why are two of the five known starbursts seen in the central molecular zone, an environment that is believed hostile to star formation? How many clusters are expected to reside near the Galactic center at the present cluster formation rate? iii) How do the total initial cluster mass and the internal dynamics of starburst clusters affect their evolution and stability? iv) Are these clusters born with normal initial stellar mass functions, or are their cores biased to high-mass stars, and are their compact cores the birth places of intermediate-mass black holes?

Starburst cluster research today

Starburst clusters, i.e. young clusters with a stellar mass exceeding 10,000 Msun formed in a single burst of star formation, represent the most efficient mode of star formation in the Galaxy. They uniquely produce the highest mass O2/O3 stars known in the present-day universe (see Zinnecker & Yorke 2007, ARAA, 45, in press, for a review). Three of the five Galactic starburst clusters, NGC 3603, Westerlund 1 and Westerlund 2, are located in spiral arms, where they are found associated with their native clouds. The two remaining clusters are special in that they formed in the central molecular zone, a ring of molecular clouds located in the inner 200 pc from the Galactic center (GC, Morris & Serabyn 1996, ARAA, 34, 645). The formation of these clusters is puzzling because high ambient and cloud temperatures, intense magnetic fields, as well as the tidal destruction of molecular clouds, render the center of the Galaxy a hostile environment to star formation (Morris 1993, ApJ, 408, 496). In combination with the increasing evidence for in-situ star formation close to the supermassive black hole (Levin & Beloborodov 2003, ApJL, 590, 33, Paumard et al. 2006, ApJ, 643, 1011, Maness et al. 2007, ApJ, accepted), these two starbursts disclose the center of the Galaxy, and the nuclei of (spiral) galaxies in general, as one of the most productive places for massive star and cluster formation. The proposed research programme focusses on the common and distinct characteristics of starburst clusters forming in intense tidal fields such as the GC vs. clusters emerging from the moderate star-forming environment of giant molecular clouds in spiral arms. Spatially resolved starburst clusters are the unique loci where cluster formation and evolution can be studied in the context of cluster environment and long-term stability or dispersal. The small population of starbursts in the Milky Way provides the templates for unresolved extragalactic systems, where conclusions on the dynamical stability and cluster survival, the stellar mass function, and especially the unobserved intermediate- and low-mass content, depend heavily on the observed modes of star formation in the Galaxy.

The initial stellar mass distribution produced inside a starburst cluster is the strongest characteristic of the cluster's fate. The initial mass function (IMF) carries the imprint of the star formation process, and the amount of high-mass cluster members and the presence of mass segregation determine its dynamical evolution and long-term survival. The favoured formation scenarios for the highest-mass stars predict a concentration of high-mass stars in the cluster cores (primordial mass segregation). At the same time, interactions between the high- and low-mass members in the dense cluster cores initiate rapid two-body relaxation with the high-mass stars sinking to the core while the low-mass stars are ejected into the cluster halo (dynamical segregation). Mass segregation was first observed in the Orion nebula cluster by Hillenbrand & Hartmann (1998, ApJ, 492, 540), and is now established in a variety of galactic and Magellanic young, open star clusters (e.g., de Grijs et al. 2002, MNRAS, 337, 597, Sirianni et al. 2002, ApJ, 579, 275, Gouliermis et al. 2004, A&A, 416, 137, Chen et al. 2007, AJ, 134, 1368). Triggered by these findings, attempts are made to quantify the changes in the mass function (MF) with time and the effects of mass segregation in the youngest clusters from simulations (Bonnell & Davies 1998, MNRAS, 295, 691, McMillan et al. 2007, ApJL, 6555, 45), which indicate that dynamical segregation in dense clusters is a rapid process taking place on timescales of a few Myr. The mixing of primordial segregation with dynamical evolution requires that the dynamical history of starburst clusters be understood before firm conclusions on a deviating initial stellar mass distribution can be made.

The formation and evolution of starburst clusters in the inner 200 pc of the Galaxy is most puzzling. In the presence of intense UV radiation causing enhanced cloud temperatures, and strong magnetic fields, the Jeans mass is expected to be elevated, causing a characteristic stellar mass of up to a factor of ten higher than in the observed initial stellar mass function (IMF) in moderate star-forming environments (Larson 2005, MNRAS, 359, 211). Scenarios for star formation in the GC predict a strong bias to high-mass stars (Nayakshin et al.~2007, MNRAS, 379, 21), and a possible low-mass cutoff in the IMF (Klessen et al. 2007, MNRASL, 374, 29), which can be invoked to explain the irregular shape of the MF observed in the Arches cluster core (Stolte et al. 2005, ApJL, 628, 113). From X-ray observations and the density profile in the nuclear stellar cluster occupying the central few parsecs around the supermassive black hole, there is increasing evidence for a top-heavy MF in the GC (Nayakshin & Sunyaev 2005, MNRASL, 364, 23, Maness et al. 2007). However, the nuclear cluster is composed of a mixed population of stars partially migrated inwards from larger radii and partially formed locally, rendering the derivation of a prestine IMF in the GC practically impossible. At a projected distance of only 30 pc, the Arches and Quintuplet clusters contain the closest single-age population near the galactic nucleus, while avoiding the crowding and mixing effects around the supermassive black hole. They provide unique samples of the stellar mass function in the environments of galactic nuclei. Their location in the strong gravitational potential in the inner Galaxy, however, exposes them to intense tidal shear. The changes in the MF imposed by the combination of tidal stripping and mass segregation need to be quantified to obtain an unbiased IMF near the GC. Deviations from the presumably uniform IMF observed in moderate star-forming regions throughout the Milky Way (Kroupa 2001, MNRAS, 322, 231) would question the use of a universal IMF to derive the dynamical properties and integrated masses of star clusters both in Galactic and extragalactic systems. From the measured present-day MFs, however, the initial mass function cannot easily be reconstructed without a detailed knowledge of the dynamical processes influencing the spatial distribution of stars in starburst clusters.

Observational constraints on the dynamical evolution of starburst clusters are scarce. Estimates of the relaxation times under the assumption of virial equilibrium contain large uncertainties of up to one order of magnitude due to the unknown velocity dispersion in the cluster core. At the present time, the dynamical evolution of starburst clusters is only constrained by N-body simulations, but predictions of the tidal losses and cluster evaporation could not be observationally addressed. Simulations of Galactic center clusters suggest cluster dispersal on timescales as short as a few 10 Myr (Portegies Zwart et al. 2002, ApJ, 565, 265, Kim et al. 1999, ApJ, 525, 228, 2000, ApJ, 545, 301), implying that only the most recently formed clusters are observed today. Even for massive clusters forming in less extreme environments, Baumgardt & Kroupa (2007, MNRAS, accepted) find that most clusters expand by factors of more than 3 during the gas expulsion phase and disperse rapidly thereafter, in agreement with the evaporation timescale of 300 Myr as derived from the mass function in NGC 3603 (Stolte et al. 2006, AJ, 132, 253). Baumgardt & Kroupa conclude that the observable velocity dispersion profile constrains the formation history of dense clusters.

The intermediate- to high-order adaptive optics systems commissioned during the past few years at 8 to 10 m class telescopes deliver high-angular resolution observations with sub-milliarcsecond astrometric accuracy. Proper motion studies of starburst clusters now contain the amazing potential to observe the velocity dispersion of massive, young clusters directly. With the derivation of the internal velocity dispersion throughout the cluster and in the core, the dynamical stability can be probed. Proper motion memberships enable the discrimination of the dispersing population around these clusters from the dense, ambient stellar field. Dynamical evolution of starburst clusters has thus become an observable quantity challenging numerial predictions. The aim of the proposed research project is to quantify the dynamical evolution of Galactic starburst clusters observationally for the first time.

One exciting prospect in measuring the internal velocity dispersion in starburst cluster cores is to probe the predicted existence of intermediate-mass black holes (Portegies Zwart & McMillan 2002, ApJ, 576, 899, Miller & Hamilton 2002, MNRAS, 330, 232). Compact cluster cores are the formation sights of the most massive stars known in the present universe. The process that leads to the formation of these stars is not yet understood. Competetive accretion (Bonnell & Bate 2006, MNRAS, 370, 488) and stellar mergers (Bonnell & Bate 2005, MNRAS, 362, 915) are the most intensely debated scenarios (e.g., Bally & Zinnecker 2005, AJ, 129, 2281). Both scenarios foster the formation of high-mass stars in the center of the gravitational well, consistent with observations of these stars in the cluster cores. However, the merger scenario offers the intriguing possibility that objects with masses beyond the stellar stability limit could form and collapse into black holes. The search for these objects in globular clusters such as M15 has now achieved upper mass limits of 2000 Msun for a potential IMBH in the cluster center (Gerssen et al. 2003, AJ, 125, 376). With the measurement of the internal velocity dispersion, the quest for IMBHs can be extended to the dense cores of young starburst clusters.

Preliminary work

The compactness of starburst clusters prohibited resolving anything but the most massive stars (Baier et al. 1985, A&A, 151, 61) in the dense cores even in the nearest Milky Way clusters until very recently. In the past ten years, substantial progress has been made as adaptive optics imaging became a generally accessible tool to the scientific community. With high-resolution, near-infrared imaging techniques, the cores of these clusters were resolved into their high-mass stellar content, mass segregation from the core to the cluster outskirts was studied, and the present-day stellar mass functions were derived.

The main focus of the previous research was on deriving photometric, present-day MFs in the Arches (Stolte et al. 2002, A&A, 394, 459, 2004, AJ, 128, 765), NGC 3603 (Stolte et al. 2006, Andersen et al.~ 005, IAU Symp.~227, 285), and Westerlund 1 cluster centers (Brandner et al. 2007, A&A, 478, 137). For the spiral arm clusters Westerlund 1 and Westerlund 2 (Ascenso et al. 2007, A&A, 466, 137), the stellar mass function down to 0.4 Msun is well fit by a standard Salpeter power law outside the core radius, as observed in less extreme star-forming regions throughout the Milky Way. In all the cluster cores, primoridal and/or dynamical mass segregation is evidenced in a significantly flattened present-day MF displaying a strong bias to high-mass stars.

For the GC clusters, the present-day MF is a matter of intense debate. In the center of the Arches cluster, the MF is measured after substantial efforts down to 2 Msun (Figer et al. 1999, ApJ, 525, 750, Stolte et al. 2002, 2004, Kim et al. 2006, ApJL, 653, 113). However, four different purely photometric approaches arrive at different conclusions (Figer et al. 1999, Stolte et al. 2005, Kim et al. 2006, Espinoza et al.~2009, A&A, submitted, astro-ph/0903.2222). The two earlier results feature a very flat MF in the cluster core suggesting an overdensity of high-mass stars with respect to a Salpeter IMF, followed by a steepening in an annulus around the core. In contrast to the flattened MF, Kim et al.~2006 obtain a normal MF in the same annulus, and the results by Espinoza et al. 2007 obtain a normal core MF which steepens far beyond the Salpeter slope at larger radii. Both in the core and in the subsequent annulus substantial irregularities are observed at intermediate masses (M ~6 Msun) (Stolte et al. 2005), indicative of either unusual dynamical evolution (Portegies Zwart et al. 2007, MNRASL, 378, 29) or a deviation in the intial MF (Klessen et al. 2007). The MF in the GC starburst clusters is a crucial probe to deviations in the initial MF in dense environments. The inconsistencies in the present-day MF derivations show the necessity for membership determination and spectral classification to obtain a reliable estimate of the cluster MF, which can form the basis for understanding its dynamical evolution, and ultimately the initial stellar mass distribution.

The formation of the GC clusters becomes ever more puzzling with increasing knowledge of their properties. The combination of Keck and VLT high-resolution imaging with resulting astrometric accuracies of 2 milliarcseconds allowed us to measure the proper motion of the Arches cluster with respect to the ambient stellar field, and derive a 3D velocity of 230 +/- 30 km/s (Stolte et al. 2007, ApJ, 675, 1278). At the young age of only 2.5 Myr, this velocity should be inherited from the parental cloud. However, the rapid orbital motion is in stark contrast to the observed radial velocities of dense molecular clouds in the central molecular zone, which are limited to less than 120 km/s (e.g., Dame et al. 2001, ApJ, 547, 792). Cloud collision scenarios are investigated as potential triggers for the Arches starburst event, but thus far, no scenario can explain all the observed cluster properties at the same time.

The structural comparison between Galactic center clusters and spiral arm clusters revealed surprising similarities. The cores of the compact clusters NGC 3603 and Arches share comparable densities, core radii, and dynamical timescales (Stolte et al. 2006). The extended population, however, appears to suffer from the influence of the environment. While the shape of the Arches cluster is influenced by tidal forces, NGC 3603 displays a seemingly unperturbed extended halo of low-mass stars (Nürnberger & Petr-Gotzens 2002, A&A, 382, 537). During the rapid dynamical evolution of GC clusters in the Galactic tidal field, such a halo of low-mass stars could not stay bound beyond the tidal radius of about 1 pc. From these previous results, the large-scale distribution of cluster members is suspected to differ significantly between GC and spiral arm clusters. More importantly, the dynamical effects of the GC on star clusters indicate that the present-day stellar MF is not representative of the initial MF in the GC environment.

The MFs derived from high-angular resolution imaging, including the study of NGC 3603 with HST/WFPC2 by Sung & Bessel (2004, AJ, 127, 1014), are the most accurate measurements of the present-day MFs in these starburst clusters to date. Despite all these efforts, the measured present-day MFs of starburst clusters suffer from large uncertainties. Substantial uncertainties in the cluster ages and distances influence the mass-luminosity conversion. Ages and distances derived from spectral classification of the high-mass stellar content in NGC 3603 (Drissen et al.~1995, AJ, 110, 2235), the Arches (Figer et al. 2002, ApJ, 581, 258, Najarro et al. 2004, ApJL, 611, 105), and Westerlund 1 (Crowther et al. 2006, MNRAS, 372, 1407) are inconsistent with the results from isochrone fitting to the pre-main sequence/main sequence transition (Stolte et al. 2004, Ascenso et al. 2007, Brandner et al. 2007). The derivation of the cluster age from evolved high-mass stars inherits the uncertainty of stellar evolution models and the understanding of the Wolf-Rayet phase, while the determination of distances from isochrone fitting to the pre-main sequence/main sequence transition depends heavily on the choice of pre-main sequence evolution models. In NGC 3603, the pre-main sequence/main sequence transition, which allowed us to tightly constrain the previously overestimated age of the cluster, nevertheless displays a large scatter in the colour distribution of stars. This scatter encompasses binarity, the survival of disks around Herbig Ae/Be objects, and possibly stellar rotation (Stolte et al. 2004), which prohibits to apply a purely photospheric mass-luminosity relation to derive the mass function in the pre-main sequence transition phase. Spectroscopic observations were obtained in Arches and NGC 3603 to minimise these uncertainties and derive an unbiased MF in the cluster cores.

The largest bias in the present starburst MFs originates from the contamination with field stars in the spiral arms (Westerlund 1 and 2, NGC 3603) and the Galactic center (Arches, Quintuplet). This problem is generally addressed with statistical field subtraction, which suffers from density and extinction variations along the line of sight, and from changes in the characteristics of the stellar population. The increasing bias by field stars with distance from the cluster center limited all existing studies to two core radii or 0.4 pc in the Arches and five core radii or 1 pc in NGC 3603. These limits are in stark constrast to the claimed cluster extent of 4.5 pc for NGC 3603 from the K-band luminosity function (Nürnberger & Petr-Gotzens 2002). The additional problem of tidal stripping in the Galactic center potential renders the cluster extent of the Arches population even more uncertain.

The problem arising from all existing studies is the reconstruction of the initial mass function from the observed present-day MF. This step requires a quantitative understanding of the dynamical evolution of starburst clusters with and without the presence of intense tidal fields. The aim of the proposed research group is to employ and extent the previously developed tools for cluster member selection, spectral classification, and the orbital evolution of GC clusters to understand the dynamical dispersal of starburst clusters into the field, and reconstruct their initial properties as prestinely as possible.

© Andrea Stolte 2009