University of Heidelberg

My Research:
1. Star Formation and Young Star Clusters in the Milky Way and Magellanic Clouds


Preamble

The globular clusters Palomar 5 and M3 on the same scale (SDSS) Star clusters cover a wide range of gravitationally bound stellar agglomerates ranging from massive, compact, long-lived globular clusters to low-mass, diffuse, short-lived associations, from metal-poor objects to clusters with super-solar metallicity, and from very old populations to embedded young clusters. It has been suggested that all stars originally formed in clusters or asociations. The study of the many types of clusters is worthwhile in its own right and reveals important information on star formation processes and the impact of galactic environment. A crucial advantage of star clusters relative to field star populations is that they represent, to first order, coeval populations and can be easily age-dated when they are resolved into individual stars. Star clusters have been identified in all galaxy types except for the least massive ones. As ensembles, star clusters can serve as a very useful tracer of galaxian star formation history.

The work presented in this section was done with my past and present graduate students Andrea Dieball, Katharina Glatt, Andrea Kayser, Andreas Koch, Xiaoying Pang, Andrea Stolte, and Peter Zeidler.

Subtopics

Back to research overview


1.1 The Stellar Initial Mass Function (IMF) and Mass Segregation

The young massive star cluster Westerlund 2 (HST) Both the slope(s) of IMF as a function of stellar mass and the question of the universality of the IMF are topics of major importance for many areas of astronomy, ranging from the details of star formation to chemical enrichment and to the interpretation of integrated stellar populations in distant galaxies. While there is clear evidence for stochastic fluctuations at the high-mass end of the IMF it is not clear whether there are also systematic variations as a function of, e.g., environment.

One way of addressing these questions is by directly determining the IMF in very young clusters; clusters young enough to not yet have been significantly affected by dynamical or stellar evolution. Our studies of young massive clusters with ground-based adaptive optics or with HST revealed such empirical measurements of the IMF to be more difficult than expected. My graduate students Andrea Stolte, Xiaoying Pang, and Peter Zeidler and I investigated several young massive clusters in detail (Arches (age 2 - 4 Myr), NGC 3603 (~ 1 Myr), and Westerlund 2 (~ 1 Myr)). They all turned out to show mass segregation despite their young age. We found radially varying mass functions and a concentration of high-mass stars in the cluster centers (e.g., Stolte et al. 2002, 2005; Pang et al. 2013). Interestingly, our results favor dynamical over primordial mass segregation, which may have implications for formation scenarios of very massive stars (Pang et al. 2013). In the Hubble Tarantula Treasury Project (HTTP; Sabbi et al. 2013), we are now extending these studies to the young clusters in 30 Doradus in the Large Magellanic Cloud (LMC).

Radially varying mass function and mass segregation in NGC 3603 (Pang et al. 2013).


The young massive star cluster NGC 3603 (HST) 1.2 Effects of Massive Star Formation on Low-Mass and Sequential Star Formation

Massive star formation in very dense environments may affect low-mass star formation in its vicinity and/or trigger subsequent star formation activity. In NGC 3603 we found the disk fraction for pre-main-sequence stars to increase from the center to the outskirts, indicating that photoevaporation caused by massive stars shortens the mass accretion time scale (e.g., Stolte et al. 2004). In Westerlund 2 we found the mass accretion rate of the pre-main-sequence stars to be 25% lower in close spatial proximity to main-sequence OB stars (Zeidler et al., in prep.). A young cluster's ionizing radiation also has a substantial impact on the surrounding nebula, clearing a cavity around the cluster and eroding the remaining pillars with higher gas and dust density as well as the envelopes of proplyds (e.g., Brandner et al. 2000; Pang, Pasquali, & Grebel 2011).

In very massive star-forming regions like 30 Doradus star clusters with a range of ages can be found (e.g., Grebel & Chu 2000). Here, we discovered the first pre-main-sequence stars detected in a galaxy other than the Milky Way and found evidence of sequential star formation presumably triggered by the central massive R136 cluster (Walborn et al. 1999; Brandner et al. 2001). The infrared HST data even revealed a very young, embedded, compact cluster in the vicinity of R136. We are now resuming these studies in the framework of the HTTP imaging survey with HST (e.g., Cignoni et al. 2015; Sabbi et al. 2015).

Embedded triggered star formation in 30 Doradus



Photometric identification of H alpha-emitting Be stars 1.3 Emission-line Stars in Young Clusters

Young star clusters may contain various types of emission-line stars, for instance He I and He II emission-line stars such as O and B supergiants or He II-emitting WN stars, C III-emitting WC stars, or Balmer emission-line stars such as Be stars or pre-main-sequence stars. We developed a simple photometric method for the detection of such stars, which uses a narrow-band filter centered on the desired emission line in combination with broad-band filters to measure the continuum and colors of the stars (Grebel, Richtler, & de Boer 1992).

Applying this method to search for main-sequence Be stars in young clusters, we found that the Be star fraction among B main-sequence stars is highest among early-type B stars (Grebel et al. 1992, 1993; Grebel 1997; Grebel & Chu 2000; Keller et al. 2001). Furthermore, comparing clusters the Milky Way, the Large Magellanic Cloud, and the Small Magellanic Cloud, we found the Be star fractions to increase with lower metallicities (Grebel 1997; Maeder, Grebel, & Mermilliod 1999). We suggested that that the average rotation is faster at low metallicities, which in turn implies more rotational mixing and could explain the higher relative N enrichment found in massive metal-poor stars (Maeder et al. 1999).

Pre-main-sequence stars also show varying amounts of Hα emission depending on their mass accretion rate. In the young Galactic cluster Westerlund 2, we used our photometric technique to identify pre-main-sequence stars, to quantify their mass accretion rate, and to show that circumstellar disks around these low-mass stars get rapidly destroyed by the UV radiation emitted by massive OB stars (Zeidler et al., in prep.)


1.4 Age Spreads

Proper-motion-selected color-magnitude diagram of NGC 3603 How coeval are the stars in a given star clusters? How long does star formation proceed in a forming star cluster? We know that extended stellar associations may show subgroups of different ages, age gradients, and/or sequential star formation (e.g., Blaauw 1991). In compact, young, massive star clusters the ages of the main-sequence and pre-main-sequence stars hold important clues. However, it is difficult to accurately age-date the upper main sequence photometrically and spectroscopically due to uncertainties in the life times of massive OB stars and additional effects introduced by rapid rotation and by the common binarity of these stars (e.g., Grebel, Roberts, & Brandner 1996). The locus of pre-main-sequence stars, on the other hand, is affected by the continuing and variable accretion of circumstellar material and by the individual inclination angles of the protostellar disks, which widen the observed pre-main-sequence locus. Moreover, as already mentioned in Section 1.2, once massive stars reach the main sequence their radiation starts to evaporate the circumstellar disks of lower-mass pre-main-sequence stars in their vicinity.

Nonetheless, the currently available data suggest that in young compact massive clusters age spreads, if any, seem to be small. For instance, a study of a careful proper-motion-based sample of member stars in the massive Galactic star cluster NGC 3603 by my graduate student Xiaoying Pang indicates an approximate mean age of 1 Myr for this object. The broad main-sequence turn-on region of the cluster supports an age spread of possibly up to two Myr, and the few older main-sequence member stars with lower luminosities than the bulk of the main sequence also suggest an earlier onset of star formation (Pang et al. 2013). However, it is also evident that star formation already progressed in that region prior to the formation of the massive cluster as evidenced by, e.g., the evolved supergiants found in its vicinity (e.g., Brandner et al. 1997a, 1997b). That young star clusters are embedded in environments with earlier as well as continuing star formation in their surroundings is also observed in many other extended massive star-forming regions, for instance in 30 Doradus (e.g., Grebel & Chu).

Duration of star formation in clumps of different density (Parmentier, Pflanzer, & Grebel 2014) There appears to be a trend that massive, high-density star clusters have more narrow age distributions, which may be caused by star formation progressing more rapidly in gas clumps with a higher density Parmentier, Pfalzner, & Grebel 2014). This yields automatically smaller age spreads in the resulting higher-density clusters. The time scale for star formation seems to be of the order of one to four free-fall times (Parmentier et al. 2014).

The discovery of multiple stellar populations in globular clusters (see Section 2.1) may indicate the presence of two or more stellar generations of different ages. This has reinvigorated the interest in possible age spreads in star clusters, particularly with respect to populations separated in age. If there are indeed multiple stellar generations, when and how did they form? Our Milky Way does not contain massive intermediate-age clusters in which one could catch such events as they are happening, but the Magellanic Clouds are rich in such clusters.

In a study to age-date intermediate-age clusters in the Small Magellanic Cloud (SMC) led my my graduate student Katharina Glatt, we detected the first cluster with possible multiple main-sequence turn-offs ever to be found in the SMC: NGC 419 (Glatt et al. 2008). Since then a growing number of such Magellanic clusters with ages typically between one to two Myr have been identified, but whether the extended turn-off regions seen in their color-magnitude diagrams are indeed caused by age spreads is a matter of debate. Other explanations, such as stellar rotation, have been proposed as well, which obviate the need for an age spread (e.g., Niederhofer et al. 2015 and references therin). Indeed, somewhat older massive intermediate-age clusters with ages of several Gyr appear to be entirely consistent with being single stellar populations (Glatt et al. 2008).

Possible multiple main-sequence turn-offs in the intermediate-age massive star cluster NGC 419 in the Small Magellanic Cloud (Glatt et al. 2008).


1.5 Binary and Multiple Star Clusters

Binary cluster candidate SL 538 / NGC 2006 in the LMC (Dieball) Individual young clusters may show substructure, for instance in the form of two distinct density peaks (e.g., Zeidler et al. 2015), but one may also find other clusters in their immediate neighborhood. This apparent proximity may occur due to chance superpositions along our line of sight without implying any physical connection, but the reverse may also be true (for example, in the famous case of the double cluster h and χ Persei in the Milky Way). But generally such constellations are rare in the Milky Way. In the Magellanic Clouds, however, an unexpectedly large number of apparent binary and multiple clusters was found. Using the criteria laid out by Bhatia and collaborators, LMC clusters with a separation less than 18 pc may be gravitationally bound to each other (e.g., Bhatia & Hatzidimitriou 1988).

A map of the distribution of binary cluster candidates in the LMC (Dieball et al. 2002) As part of her PhD thesis work, Andrea Dieball identified possible binary clusters in the LMC based on this proximity criterion and determined the ages of the candidates via color-magnitude diagrams. A bit more than 4000 clusters fulfilled Bhatia et al.'s proximity criterion. We found that multiple cluster candidates are predominantly younger than 300 Myr. Many, though not all, are coeval within the age-dating accuracy, which would support a physical connection and possibly formation from the same giant molecular cloud. An alternative, though probably less common possibilty for forming a cluster pair is via tidal capture. Depending on the cluster density, between 56% (high-density bar region) and 12% (outer LMC) of the detected pairs would be expected statistically as chance superpositions, which leaves a large number of possible physical pairs (Dieball & Grebel 1998, 2000; Dieball, Grebel, & Theis 2000; Dieball, Müller, & Grebel 2002). The question of the long-term evolution of physical pairs remains open - in principle, mergers are possible (which might have interesting implications for multiple stellar populations in older clusters), or both components may gradually dissolve.


References



Back to research overview

to top of page