High-mass stars in clusters and associations
SStellar Clusters & Associations: A RIA Workshop on GaiaGranada, 23 rd - 27 th May 2011
High-mass stars in clusters and associations
Ignacio Negueruela , Departamento de F´ısica, Ingenier´ıa de Sistemas y Teor´ıa de la Se˜nal, Universidad deAlicante, Apdo. 99, E03080 Alicante, Spain
Abstract
High-mass stars are major players in the chemical and dynamical evolution of galaxies.Open clusters and associations represent the natural laboratories to study their evolution.In this review, I will present a personal selection of current research topics that highlight theuse of open clusters to constrain different properties of high-mass stars, such as the possibleexistence of an upper limit for the mass of a star, the evolutionary stage of blue supergiantsor the characterisation of supernova progenitors.
The study of high-mass stars (also, though not quite correctly, known as massive stars) isintimately linked to, and difficult to disentangle from, the study of young open clusters. Thisis not only because most high-mass stars are found within young open clusters and associ-ations, but also because clusters are the natural laboratories for investigating the evolutionof high-mass stars. At the most basic level, determining the fundamental parameters of amassive star requires knowledge of its distance (e.g., Herrero et al., 1992). Since very fewhigh-mass stars have accurate parallaxes (Ma´ız Apell´aniz et al., 2008), membership in a clus-ter or association is required for calibration. In view of such a close connection, this shortreview can only cover a very small fraction of the many topics subject to active study thatI could have considered for inclusion. I will concentrate on a few issues of current researchthat highlight the importance of open clusters as laboratories for understanding massive starevolution. The selection of these topics is undoubtedly biased by my personal preferences,but is representative of the areas currently generating the highest interest among researchersin the field. a r X i v : . [ a s t r o - ph . GA ] S e p Massive stars in clusters
There are several possible definitions for high-mass stars, all of them indirect. We can definehigh-mass stars as those initiating C burning non-explosively in their cores. Modern theoreti-cal models including overshooting indicate that this will happen for M ∗ > ∼ M (cid:12) (e.g., Eldridge& Tout, 2004). We can also define high-mass stars as those ending up their lives in supernovaexplosions. These two definitions are almost identical, though the latter implies a slightlyhigher mass (see, e.g., Poelarends et al., 2008, for a description of the physics involved). Thelimit for a star to produce a supernova explosion has recently been set from observations ofsupernova progenitors at > ∼ . +1 − . M (cid:12) (Smartt et al., 2009, and see also Sect. 5). A thirdpossible definition of high-mass stars is those stars with self-initiating radiation-driven winds(Kudritzki & Puls, 2000). Radiative winds become detectable for main-sequence stars withspectral type earlier than B2 V, which again roughly corresponds to M ∗ > ∼ M (cid:12) .Observationally, high-mass stars comprise the OB stars (approximately, O2–B2 V, O2-B5 III and O2-B9 I; Reed 2009; Walborn & Fitzpatrick 1990) and some later type supergiants(the most luminous supergiants of spectral types A, F, G, and K, and the M-type supergiants).For an observational review, see Massey (2003). The formation of high-mass stars is a major research topic in modern astrophysics, and willnot be discussed here. See Zinnecker & Yorke (2007) for a recent review. High-mass starsmay have an effect on the formation of low-mass stars, and thus on the Initial Mass Function(IMF). In addition, high-mass stars may play a role in triggering further star formation. Thisissue has been controversial since the emergence of the classical theory (Elmegreen & Lada,1977), because a causal relationship is difficult to assess, even if indications of sequentialformation are widespread and strong indirect evidence for triggering has been found (e.g.,Walborn, 2002; Zavagno et al., 2005). I will not discuss star formation here, but simply notethat sequential star formation within a cluster or association may give rise to populationswith different ages, which might be difficult to disentangle, meaning that accurate ageing isnot always possible for young open clusters.
The possible existence of an upper limit to the IMF has been a hotly debated issue in recenttimes. Several stars believed to be extremely luminous, and thus extremely massive, havebeen resolved into two or more components (e.g., Ma´ız Apell´aniz et al., 2007). Based on ananalysis of the stellar population in the compact young Arches open cluster, Figer (2005)concluded that there was clear evidence for an upper limit to stellar masses around 150 M (cid:12) .This result is dependent on several factors, such as the extinction law to the Galactic Centre,but has also been reproduced for the Large Magellanic Cloud (Koen, 2006).Recent modelling advances have revealed that hydrogen-rich Wolf-Rayet (WR) stars ofthe nitrogen sequence (WN) are really main-sequence objects with heavy mass loss. Theseare candidates to be the most massive stars in their hydrogen core burning phase (Langer etal., 1994; de Koter et al., 1997; Crowther & Dessart, 1998). The fact that they are commonly . Negueruela ∼ M cl = 10 M (cid:12) (Ascenso et al., 2007),a result supported by the recent discovery of two very massive stars likely ejected from thecluster core (Roman-Lopes et al., 2011). One of the most luminous members of Westerlund 2,the WN6h star WR20a, was identified as a massive binary by Rauw et al. (2004) and as aneclipsing binary by Bonanos et al. (2004). Combination of the radial velocity curve and thelightcurve provides an accurate solution with P orb = 3 . ± .
01 d. The components havemasses M = 83 ± M (cid:12) and M = 82 ± M (cid:12) . They have the same spectral type and thesame mass, which is the highest stellar mass measured with high accuracy.Even more massive than Westerlund 2, the cluster at the core of the giant H ii regionNGC 3603, also located in the Sagittarius arm (at ∼ ≈
35 early O or WR stars (Moffat et al., 1994) and likely has a mass M cl =1 . × M (cid:12) (Rochau et al., 2010, and references therein). Its central concentration contains3 WN6ha stars, the brightest of which is an eclipsing binary with P orb = 3 . K -band spectroscopy (Schnurr et al., 2008). The radial velocitycurve of the secondary has large uncertainties and therefore the masses are not very tightlyconstrained. The solution implies M = 116 ± M (cid:12) and M = 89 ± M (cid:12) . Component 1is thus the most massive star with a dynamically determined mass (Schnurr et al., 2008).Outside the Milky Way, the 30 Dor complex in the Large Magellanic Cloud is likelythe most massive starburst in the Local Group. Its nuclear cluster, Radcliffe 136, is 2.7 pcacross and contains hundreds of OB stars. Its mass has been estimated to be, at least, M cl = 1 × M (cid:12) (Andersen et al., 2009). Using updated stellar models, Crowther etal. (2010) found that the dynamical masses of the WN6ha stars in NGC 3603 could be wellreproduced by evolutionary tracks. If these models are then extrapolated to the even brighterWN5h stars at the core of 30 Dor, they indicate enormous masses. Three of the WN5h starsare fitted with models implying current masses ≥ M (cid:12) , with Star a1 having 265 +80 − M (cid:12) (Crowther et al., 2010). This would be the most massive star known, well above the proposedupper mass limit. Unfortunately, none of the stars at the core of 30 Dor appears to be aspectroscopic binary to confirm this determination with a dynamical measurement. High-mass stars play a decisive role in driving the chemical evolution of galaxies. Heavy massloss through all their life stages and in the final supernova explosion provides an importantfraction of the heavy elements in the interstellar medium. In order to understand the chemicalenrichment of the medium, we need to constrain how and when mass is lost by high-massstars, and this means being able to map observed phases on to theoretical evolutionary tracks.Open clusters and associations allow us to explore the evolutionary context of massivestars in different evolutionary stages (e.g., Walborn, 2010). This process starts with the
Massive stars in clusters youngest stellar systems. For example, the Carina nebula provides many of the MK standardsfor the earliest O subtypes (Walborn et al., 2002). The fact that O3–4 stars are still on themain sequence indicates that the complex is very young, while this very youth constrains itsonly evolved star, η Car, to be extremely massive.At a slightly older age, Cyg OB2 represents an important laboratory for the study ofO-type stars. Considered a very promising target since its discovery (Johnson & Morgan,1954), it could only be studied in detail after the advent of CCD detectors because of thehigh obscuration, in spite of its relative low distance (see Fig. 1). Massey & Thompson (1994)identified ∼
60 stars more massive than 15 M (cid:12) , finding that Cyg OB2 is very compact for anOB association and seems to occupy a more or less unique position somewhat intermediatebetween an open cluster and a normal OB association (cf. Kn¨odlseder, 2000). The centralregion contains two cluster-like stellar concentrations (Bica et al., 2003). They form anelongated figure ∼ (cid:48) × (cid:48) ( ∼ × >
50 (and likely ∼ ∼ M (cid:12) in the core region (Currie et al., 2010). Most recentworks seem to converge on an age ∼
14 Myr, in good agreement with an apparent main-sequence turn-off at B1 V. The luminous supergiants in the cluster, however, have spectraltypes too early for this age (Marco & Bernabeu, 2001), at B2 Ia (HD 14143) and B3 Ia(HD 14134), and evolutionary masses approaching 40 M (cid:12) (McErlean et al., 1999; Crowtheret al., 2006a). A similar situation is found in many other clusters (Marco et al., 2007). Inmost cases, there are no observational reasons to suggest that these supergiants are membersof a younger (second-generation) population. This clearly shows that we are still very farfrom understanding exactly which evolutionary phase blue supergiants represent.Because of the huge size of its stellar population, the young cluster Westerlund 1 (Wd 1)is currently our best laboratory for studying the evolution of high-mass stars. Located at ∼ × M (cid:12) (Gennaro et al., 2011) and may approach 10 M (cid:12) .With more than twenty WR stars (Crowther et al., 2006b) and several dozen OB supergiants(Negueruela et al., 2010), Wd 1 provides stringent tests on current theoretical models. Thepopulation observed is consistent with a single burst of star formation. There may be some . Negueruela V − K ) term was added twice, because of a mistake with thespreadsheet. When this is taken into account, the distance modulus DM = 10 . DM = 11 .
3, as displayed. Continuous lines are non-rotating isochrones for log t = 6 .
2, 6 . . t = 6 . t = 6 . Massive stars in clusters mild blue stragglers, but the blue supergiants in the cluster (including some very luminouslate-B hypergiants with M V ∼ −
9) fit rather well the theoretical isochrones for ages between5 and 6 Myr (Negueruela et al., 2010). Moreover, the dynamical determination of the massfor the two components of the eclipsing binary Wd1-W13 shows good agreement with theexpected theoretical masses, with the B1 I component having a mass M ∗ = 35 ± M (cid:12) (Ritchieet al., 2010). This good agreement is comforting in view of the difficulties in understandingblue supergiants in other clusters and provides strong support for the basic assumptionsunderlying current evolutionary tracks (e.g., Meynet & Maeder, 2000). The large number of high-mass stars present in Wd 1 offers an ideal opportunity to studytheir binary fraction. The population of WR stars offers strong indirect evidence for a veryhigh binary fraction (Crowther et al., 2006b). Targeted observations of the blue supergiantpopulation revealed that at least 40% of the stars observed are binaries (Ritchie et al., 2009).Observations of a larger sample, including stars closer to the main sequence, are in progress.Searches for binaries in Cyg OB2 are, to date, more complete. A radial velocity surveycarried out over several years (Kiminki et al., 2007; Kobulnicky & Kiminki, 2011) shouldbe able to reveal, except for very unfavourable inclinations, all massive companions and asubstantial fraction of low-mass companions. The survey has so far detected 20 spectroscopicbinaries, while 20 other objects show radial velocity variations, and might be binaries. Com-parison to theoretical expectations suggests a very high binary fraction, perhaps approachingunity. The masses of the companions seem to follow a flat distribution (Kiminki et al., 2009;Kobulnicky & Kiminki, 2011).The binary fraction and companion mass distribution do not seem to be universal,though. Variability in both observables is high among the few clusters with dedicated studies.In NGC 6231, the complete sample of O-type stars (16 objects) presents f bin > .
63, withcompanions having consistently M ∗ > M (cid:12) (Sana et al., 2008). In contrast, the smallercluster NGC 2244, with 6 O-type stars, has a much smaller binary fraction, f bin > . f bin > .
20, with the possibility of massive companionsalmost ruled out (Mahy et al., 2009; de Becker et al., 2006).Meanwhile, spectroscopic monitoring of a large sample of O and WN stars has revealed avery high fraction of spectroscopic binaries (Barb´a et al., 2010). Interferometric observations,which sample a different range of orbital sizes, have also demonstrated a very high degreeof multiplicity amongst high-mass stars. O-type stars in clusters display a much higherbinary fraction than field O stars (Mason et al., 2009). This can be naturally explained ifmany (most) field stars have been ejected from open clusters (in accordance with theoreticalpredictions presented in several contributions to these proceedings).
After the blue supergiant phase, the evolutionary tracks of massive stars become more un-certain. An important fraction of high-mass stars (perhaps all with M ∗ > ∼ M (cid:12) ) becomeWR stars. Before reaching this hydrogen-depleted phase, the stars must lose their envelopes, . Negueruela ∼ M (cid:12) are expected to end their lives as red supergiants(RSGs). Modelling of RSGs is very complex due to a number of factors, among which wecan cite their huge size, poorly determined molecular opacities and the expected heavy massloss. Because of these difficulties, their fundamental properties are not very well known,though important advances have been obtained with new model fits to a sample of RSGs inopen clusters and associations (Levesque et al., 2005). Spectroscopic monitoring of RSGs inclusters has allowed the detection of irregular radial velocity variations, which may be thesignature of pulsation, the most likely source of mass loss (Mermilliod et al., 2008).In recent years, RSGs have shown their usefulness as signposts of massive clusters. Asthey are very bright in the infrared, they can be seen through heavy absorption. Severalclusters rich in RSGs have been discovered towards the inner Galaxy (e.g., Figer et al., 2006;Clark et al., 2009a). Determination of their chemical composition has provided extremelyvaluable clues to the Galactic chemical distribution (Davies et al., 2009). These studies areallowing the discovery of an increasing number of massive young open clusters in the Galaxy(see Portegies Zwart et al., 2010, for a review). Unfortunately, the actual masses of theseobscured clusters are not known, as we can only see the RSGs and not the associated mainsequence (see Fig. 2). The most extreme case is the open cluster Stephenson 2, which containsat least 26 RSGs (Davies et al., 2007), and might have a mass approaching 10 M (cid:12) , if currentpredictions for the duration of the RSG phase are correct. Unfortunately, this is not certain,as the case of the open cluster NGC 7419 (Marco & Negueruela, 2011, and Fig. 2) highlights.One important test for stellar evolution models is the ratio of blue to red supergiantsin a population. This ratio is very sensitive to input physics, such as treatment of massloss, convection and mixing processes. Unfortunately, in most Galactic clusters, the numberof supergiants is so small that the ratio is not statistically significant. Studies so far haveconsidered average values over a number of open clusters, finding an apparent dependenceon Galactocentric radius. When observations of the Magellanic Clouds are also included,there is strong evidence for an increase in the blue to red supergiant ratio with increasingmetallicity (Eggenberger et al., 2002). This is exactly the opposite behaviour to what currentevolutionary tracks would predict (Meynet et al., 2011).However, simply counting all red and blue supergiants may provide a distorted picture,as not all “supergiants” sample the same population. As an example, the open clusterNGC 6649, located at ∼ ∼
50 Myr (Turner, 1981)and contains at least three low-luminosity RSGs (with spectral types around K1 Ib) andone supergiant Cepheid variable. According to modern evolutionary tracks (e.g., Marigo etal., 2008), the progenitors of these supergiants had initial masses ∼ M (cid:12) . Stars of suchlow masses should never appear as blue supergiants (according to the same tracks), and soincluding them in counts to determine the red to blue ratio may be misleading.Nevertheless, these low-mass RSGs are very interesting objects in themselves, as theymust be the progenitors of most Type II supernovae. This is a simple consequence of the shape Massive stars in clusters
Figure 2: Left: 2MASS image of the open cluster NGC 7419. This obscured cluster inthe Perseus Arm contains 5 red supergiants for no blue supergiants and presents the highestfraction of Be stars among Galactic clusters (Marco & Negueruela, 2011). Current theoreticalmodels predict that this cluster should have a mass approaching 10 M (cid:12) to contain 5 RSGs(Clark et al., 2009b), but its mass is unlikely to exceed 3 × M (cid:12) . Right: 2MASS imageof RSGC1, a cluster of red supergiants at d ∼ ∼
10 Myr and may have ≥ × M (cid:12) (Davies et al., 2008).Comparison of this image to the one in the left panel highlights the brightness of RSGs inthe near infrared and their role as signposts of recent star formation. The population of bluestars easily seen in NGC 7419 fades out in RSGC1 due to the higher distance, much higherextinction, and confusion. . Negueruela M ∗ ∼ M (cid:12) than with M ∗ ∼ M (cid:12) . AsType II supernovae are very important contributors to the dynamical and chemical evolutionof galaxies, determining the lower mass for a supernova explosion becomes a fundamentalissue to understand the history of the Universe. In recent years, an important observationaleffort has been dedicated to identifying supernova progenitors and deriving their properties(see Smartt, 2009). In some cases, it has been possible to identify the actual star in pre-explosion images. Sometimes, the supernova seems to have taken place within an opencluster or association and properties can be inferred from the parent population.Based on these observations, the minimum mass for an exploding star has been esti-mated at ≈ . +1 − . M (cid:12) (Smartt et al., 2009). Efforts are also being dedicated to determiningwhether there is a direct relation between the initial mass of a star and the class of super-nova explosion it will undergo. Of course, the final fate of a massive star depends on severalfactors, such as mass loss rates during the different evolutionary phases, initial rotationalvelocity and its evolution, and, perhaps most decisively, binary interaction (e.g., Meynet etal., 2011). In spite of this, attempts have been made to the develop “typical” scenarios (e.g.,Smith et al., 2011).Another interesting fact is the apparent correlation between the fraction of Be starsamong B-type stars and the number of RSGs. Both seem to increase in a similar way withdecreasing metallicity (Meynet et al., 2007), suggesting a possible evolutionary connectionwith fast rotation. In this sense, the open cluster NGC 7419 (Fig. 2), which has the highestfraction of Be stars amongst Milky Way clusters, presents a blue to red supergiant ratio of 0/5(Marco & Negueruela, 2011), atypical for Milky Way clusters. In contrast, NGC 663, whichalso has a very high Be-star fraction and is located at approximately the same Galactocentricdistance, presents a ratio 5/0 for the core region (6/2 when the halo is also considered),showing that correlations are not always direct or simply that statistical fluctuations maydominate the observables even in moderately massive clusters. In the near future,
Gaia is bound to play a decisive role in furthering our knowledge of high-mass stars. Accurate parallaxes will result in much better distances to open clusters, implyingimproved luminosity determinations. In addition, confirmation of membership for individualpeculiar objects will give them an evolutionary context, providing strong constraints forevolutionary tracks.
Gaia may also contribute strongly to the study of binarity amongsthigh-mass stars, as it will obtain accurate lightcurves for huge samples and may detect orbitalmotions in wide binary systems.Of course,
Gaia cannot solve every problem. We will need other data and also newtheoretical developments. High-resolution spectroscopy coupled with accurate stellar atmo-sphere models will be needed to determine stellar parameters beyond luminosity (see, e.g.,Sim´on-D´ıaz et al., in these proceedings). Improved evolutionary tracks will result from deeperinteraction between theory and observations. Radial velocity curve solutions for spectroscopicbinaries will sample the intermediate range of binary separations.But, above all, we have to be conscious of one main limitation:
Gaia will not be ableto sample the inner Galaxy, where high extinction and crowding will impede the acquisi-0
Massive stars in clusters tion of high-quality data. Almost all the massive young clusters known will be beyond itsreach. Fortunately, a full complement of new instruments will help us to study star formationin the inner Galaxy. Among them, the first generation of infrared spectrographs with highmultiplexing capabilities, such as KMOS at the VLT or MIRADAS at GTC, will play a funda-mental role in studying obscured clusters and massive star-forming regions. Meanwhile, newspace telescopes will provide the tools to resolve clusters in the galaxies of the Local Group.JWST may be an ideal tool for detecting RSGs in distant galaxies (and hence identifyingsupernova progenitors), while WSO/UV will be used to study low-extinction populations ofmassive hot stars. In all, with this formidable set of new missions and instruments, the nextfew years are likely to see important developments in this field.
Acknowledgments
This research is partially supported by the Spanish Ministerio de Ciencia e Innovaci´on (MICINN)under grants AYA2010-21697-C05-05 and CSD2006-70. I thank Jes´us Ma´ız Apell´aniz and SergioSim´on D´ıaz for comments on the manuscript.The Two Micron All Sky Survey (2MASS) is a joint project of the University of Massachusettsand the Infrared Processing and Analysis Center/California Institute of Technology, funded by theNational Aeronautics and Space Administration and the National Science Foundation.
References
Andersen, M., Zinnecker, H., Moneti, A., et al. 2009, ApJ, 707, 1347Ascenso, J., Alves, J., Beletsky, Y., & Lago, M.T.V.T. 2007, A&A, 466, 137Barb´a, R.H., Gamen, R., Arias, J.I., et al. 2010, RMxAC, 38, 30de Becker, M., Rauw, G., Manfroid, J., & Eenens, P. 2006, A&A, 456, 1121Bica, E., Bonatto, Ch., & Dutra, C. M. 2003, A&A, 405, 991Bonanos, A.Z., Stanek, K.Z., Udalski, A., et al. 2004, ApJ, 611, L33Clark, J.S., Negueruela, I., Davies, B., et al. 2009a, A&A, 498, 109Clark, J. S., Davies, B., Najarro, F., et al. 2009b, A&A, 504, 429Clark, J.S., Ritchie, B.W., & Negueruela, I. 2010, A&A, 514, A87Comer´on, F., Pasquali, A., Rodighiero, G., et al. 2002, A&A, 389, 874Crowther, P.A., & Dessart, L. 1998, MNRAS, 296, 662Crowther, P.A., Lennon, D.J., & Walborn, N.R. 2006a, A&A, 446, 279Crowther, P.A., Hadfield, L.J., Clark, J.S., et al. 2006b, MNRAS, 372, 1407Crowther, P.A., Schnurr, O., Hirschi, R., et al. 2010, MNRAS, 408, 731Currie, T., Hernandez, J., Irwin, J., et al. 2010, ApJS, 186, 191Davies, B., Figer, D.F., Kudritzki, R.-P., et al. 2007, ApJ, 671, 781 . Negueruela Davies, B., Figer, D.F., Law, C.J., et al. 2008, ApJ, 676, 1016Davies, B., Origlia, L., Kudritzki, R.-P., et al. 2009, ApJ, 696, 2014Eggenberger, P., Meynet, G., & Maeder, A. 2002, A&A, 386, 576Eldridge, J.J., & Tout, C.A. 2004, MNRAS, 353, 87Elmegreen, B.G., & Lada, C.J. 1977, ApJ, 214, 725Figer, D.F. 2005, Nature, 434, 192Figer, D.F., MacKenty, J.W., Robberto, M., et al. 2006, ApJ, 643, 1166Gennaro, M., Brandner, W., Stolte, A., & Henning, Th. 2011, MNRAS, 412, 2469Hanson, M.M. 2003, ApJ 597, 957Herrero, A., Kudritzki, R.-P., Vilchez, J.M., et al. 1992, A&A, 261, 209Johnson, H.L., & Morgan, W. W. 1954, ApJ, 119, 344Kiminki, D.C., Kobulnicky, H. A., Kinemuchi, K., et al. 2007, ApJ, 664, 1102Kiminki, D.C., Kobulnicky, H.A., Gilbert, I., et al. 2009, AJ, 137, 4608Kn¨odlseder, J. 2000, A&A, 360, 539Kobulnicky, H.A., & Kiminki, D.C. 2011, BSRSL, 80, 616Koen, C. 2006, MNRAS, 365, 489de Koter, A., Heap, S.R., & Hubeny, I. 1997, ApJ, 477, 792Kudritzki, R.-P., & Puls, J. 2000, ARA&A, 38, 613Langer, N., Hammann, W.-R., Lennon, M., et al. 1994, A&A, 290, 819Levesque, E.M., Massey, P., Olsen, K.A.G., et al. 2005, ApJ, 628, 973Mahy, L., Naz´e, Y., Rauw, G., et al. 2009, A&A, 502, 937Ma´ız Apell´aniz, J., Walborn, N.R., Morrell, N.I., et al. 2007, ApJ, 660, 1480Ma´ız Apell´aniz, J., Alfaro, E., & Sota, A. 2008, arXiv:0804.2553