Advances in instrumentation have shaped and refined our current view of "extreme star formation" (ESF). Many of the features of ESF in the local universe were known half a century ago: giant H II regions (Burbidge and Burbidge 1962), galactic winds (Lynds and Sandage 1963, Burbidge et al. 1964), young "populous clusters" (Gascoigne and Kron 1952, Hodge 1961), O-star dominated compact dwarf galaxies (Sargent and Searle 1970), and bright extragalactic radio sources (Weedman et al. 1981). The recognition that there were individual star formation events that could energetically dominate the evolution of a galaxy came with the development of infrared and high resolution radio capabilities (Rieke and Low 1975, Rieke and Lebofsky 1978, 1979, Gehrz et al. 1983). However, it was the IRAS all-sky survey in the mid and far-infrared that established the universality and energetic importance of the "starburst" to galaxy evolution. IRAS demonstrated that the luminous output of galaxies can be dominated by infrared emission and recent star formation (de Jong et al. 1984, Soifer et al. 1984, 1986, 1987a, 1987b) and that extreme star formation may even be linked to the development of nuclear activity in galaxies (Sanders et al. 1988). Many of the early IRAS results have been followed up with the subsequent ISO (Genzel and Cesarsky 2000) and Spitzer (Lonsdale et al. 2003) infrared space observatories.
The definition of starburst has evolved since the time of IRAS. In early incarnations it described systems that would deplete their gas in substantially less than a Hubble time. However, this definition can exclude galaxies with substantial reservoirs of gas that are forming stars at prodigious rates. Infrared luminosity can be used to classify these extreme star-forming systems: "Ultraluminous infrared galaxies" (ULIRGs) have luminosities of LIR > 1012 L (Soifer et al. 1987b) and "luminous" infrared galaxies have LIR 1011 L. These luminous galaxies owe most of their energetic output to star formation (Sanders et al. 1986, Genzel et al. 1998). Another definition captures the localized intensity of starbursts: Kennicutt (1998) defines starburst in terms of a star-forming surface density, 100 M pc-2 Gyr-1, or in terms of luminosity, 1038.4 - 1039.4 erg s-1 kpc-2.
High spatial resolution has also modified our view of starbursts, revealing that they often, and perhaps nearly always, consist of the formation of large numbers of extremely large clusters, "super star clusters." For the purposes of this review, we will use the term super star cluster (SSC) to denote massive young clusters of less than ~ 100 Myr in age, and globular clusters to be those systems more than 7 - 10 Gyr in age. Both starbursts and SSCs comprise "extreme star formation."
The Hubble Space Telescope (HST) is largely responsible for the recognition of the ubiquity of SSCs and their importance in starbursts. The idea that massive clusters similar to globular clusters are actually forming in large numbers at the present time was slow to germinate, probably due to lack of spatial resolution and our inability to resolve them, although the possibility was recognized early on in the large clusters of the Magellanic clouds (Hodge 1961). The cluster R136, the star cluster responsible for the lovely 30 Doradus Nebula in the Large Magellanic Clouds, was believed by many to be a single supermassive star before it was resolved with speckle observations from the ground (Weigelt and Baier 1985). Other mysterious objects included two bright sources in the nearby dwarf galaxy NGC 1569, which were difficult to classify due to the uncertainty in distance to this nearby galaxy. Regarding the two "super star cluster" candidates, Arp & Sandage (1985) stated:
A definite resolution of the present problem in NGC 1569, and for the same problem with the bright object in NGC 1705, lies in the spatial resolution into stars of these three high-luminosity blue objects using the imaging instrument of the wide-field camera of Space Telescope.
HST did indeed resolve the objects in NGC 1569, revealing that they were large and luminous star clusters. In Figure 1 is shown the HST image of NGC 1569, with objects A and B referred to by Arp and Sandage. Object A consists of two superimposed clusters; crowded conditions and confusion complicate the study of SSCs, even in the closest galaxies and with the angular resolution of HST! R136 was resolved into a compact and rich cluster of stars by HST (de Marchi et al. 1993, Hunter et al. 1996). HST imaging also revealed super star clusters in NGC 1275, M82, NGC 1705, the Antennae, and numerous other starburst galaxies (Holtzman et al. 1992, Whitmore et al. 1993, O'Connell et al. 1994, Meurer et al. 1995, Maoz et al. 1996, Whitmore and Schweizer 1995) The discovery of young, blue star clusters with luminosities consistent with those expected for young globular clusters in local starburst galaxies meant not only that conditions favorable to the formation of protoglobular clusters exist in the present universe, but also that this extreme form of star formation is close enough for the star formation process itself to be studied.
Figure 1. HST revealed that the bright sources NGC 1569-A and NGC 1569-B were large clusters of stars. Credit: ESA, NASA, and P. Anders.
Super star clusters appear to be sufficiently massive and rich to be globular clusters, differing from them only in age. Table 1 lists the general characteristics of different classes of Galactic star clusters and SSCs. The brightest young SSCs have MV ~ -14. They are brighter than globular clusters because of their youth. It is convenient to take the lower bound for SSCs to be MV ~ -11 (Billett et al. 2002), approximately the magnitude of R136 in the LMC, which is also comparable in size to the most massive young Galactic clusters. However, R136 is considered by some to be on the small side for a globular cluster. The older LMC cluster NGC 1866, at L ~ 106 L and an intermediate age of 100 Myr, is closer to a genuine globular cluster (Meylan 1993). The upper limit to the ages of SSCs is also somewhat arbitrary; while there is evidence that typical cluster dissolution timescales are about 10 Myr, there are also intermediate age clusters with ages of ~ 1 Gyr even within the Local Group.
|globular cluster||> 105||103.5 - 106||0.3 - 4||10-1 - 104.5||-3 to -10||> 1010|
|open cluster||20 - 2000||350 - 7000||2.5 - 4.5||1 - 100||-4.5 to -10||106 - 109.8|
|embedded cluster||35 - 2000||350 - 1100||0.3 - 1||1 - 5||...||106 - 107|
|SSC||> 105||105 - 106||3 - 5||...||-11 to -14||106 - 107|
|a Half mass mean density. Number of members, N*, is not as well-defined for the larger clusters as it is for the local open and embedded clusters. References. Battinelli & Capuzzo-Dolcetta 1991. Billett et al. 2002. Harris 1996. Harris & Harris 2000. Lada & Lada 2003. Mackey & Gilmore 2003. McLaughlin & Fall 2008. McLaughlin & van der Marel. 2005. Noyola & Gebhardt 2007.|