Birth and death go hand in hand in young SSCs, since O stars barely stop accreting before they die (Zinnecker and Yorke 2007). During their short lifetimes, O stars find many different ways to lose mass. Windy and explosive by nature, O stars emit copious numbers of destructive ultraviolet photons and are responsible for much of the mass loss and mechanical feedback within a star cluster. Outflows, wind bubbles, LBV mass ejections, and SNR from a single O star can influence a region the size of a young SSC; imagine what an SSC consisting of thousands of O stars can do to a parsec-scale volume! Figure 5 illustrates a few of the many ways that O stars can destructively interact with their environments: CO outflows in adolescence (1 Myr); LBV outflows similar to that responsible for the Homunculus Nebula in early adulthood (2 - 3 Myr); Wolf-Rayet wind bubbles at retirement (3 - 4 Myr); death by supernova (5 - 10 Myr). If the young cluster manages to survive through the paroxysms of its riotous O star siblings, then there is most likely a molecular cloud nearby to unbind it.
Figure 5. The many ways that O stars can be destructive. Circles represent a region 1 pc across, the size of the core of an SSC. (top left) Owens Valley Millimeter Array image of the CO outflow source around the massive protostar G192.16-3.82, Shepherd and Kurtz (1999). (top right) The Homunculus Nebula in Eta Carina imaged by HST, Morse et al. (1998). (lower left) Wolf-Rayet bubble RCW58 in H, Gruendl et al. (2000). (lower right) Chandra image of the 1000-yr-old SNR Cas A, Hughes et al. (2000).
There is strong evidence that most SSCs in the local universe will not survive to old age. The odds of survival even in the relatively benign environments of the Galactic disk and halo are slim. Only an estimated ~ 7% of embedded young clusters in the solar neighborhood will live to the age of the Pleiades (Lada and Lada 2003). Dynamical models of the Galactic globular cluster system suggest that as much as 75% of the Galactic stellar halo may have originally been in the form of globular clusters (Gnedin and Ostriker 1997, Shin et al. 2008), which now account for only 1% of visible halo stars (Harris 1998).
The first hurdle that a young SSC must overcome is star formation efficiency, defined here as = Mstars / (Mstars + Mgas). The gas mass contribution includes contributions from ionized gas and atomic and molecular gas from the natal clouds. Nominally > 50% is required to leave a bound cluster (Hills 1980, Mathieu 1983, Lada et al. 1984). Lower values of ~ 30% can be accommodated if the cluster loses stars, but retains a smaller bound core, leading to smaller clusters. The cluster may also survive if gas is lost quasistatically, so that the cluster adjusts to the new equilibrium (Kroupa and Boily 2002); it may also survive, although with a significantly reduced stellar mass, if it suffers rapid mass loss ("infant weight-loss") early on (Bastian and Goodwin 2006). Models suggest that clusters with masses less than 105 M lose their residual gas quickly, and that 95% of all clusters are so destroyed within a few tens of Myr, and that rapid gas expulsion may give a natural explanation for the lognormal PDMF for globular clusters (Baumgardt et al. 2008).
Based on the SFE / as observed in the Galaxy, the picture looks bleak for young SSCs. On the sizescales of giant molecular clouds, is at most a few percent (Lada and Lada 2003), a number that appears to be determined by the turbulent dynamics of clouds (Padoan and Nordlund 2002, Krumholz et al. 2006, Padoan et al. 2007). On smaller scales in the Galaxy, ~ 10% - 30% (NGC 3603; Nürnberger et al. 2002), which is still rather small to preserve a significant bound cluster on globular cluster scales.
There is both fossil and structural observational evidence that clusters dissolve. The vast majority of bright SSCs with masses over 105 M in known SSC systems are less than 10 Myr in age (Mengel et al. 2005, Bastian et al. 2005). From STIS spectroscopy, Tremonti et al. (2001) and Chandar et al. (2005) find that the field stars in the nearby starburst galaxy NGC 5253 can be modeled by dispersed cluster stars, with cluster dissolution timescales of 7 - 10 Myr. In the Antennae system, Fall et al. (2005) find that the number of clusters falls sharply with age, with ~ 50% of the stars in clusters having dispersed after 10 Myrs. They estimate that at least 20% and perhaps all, of the disk stars in the Antennae have formed within clusters. On the other hand, there is fossil evidence, in the form of globular clusters, that large clusters can and do survive for many Gyr.
Star formation efficiency and cluster disruption may imprint upon cluster mass functions. Parmentier et al. (2008) suggest that at ~ 20%, a power-law core mass function turns into a bell-shaped cluster mass function, while at higher efficiencies the power law is preserved. Gieles & Bastian (2008) suggest that the maximum cluster mass and age is a diagnostic of cluster disruption, and they see evidence in cluster mass function, that formation/disruption does vary among galaxies.
The question remains of how globular clusters have managed to live to such a ripe old age. Can we identify SSCs forming today that might indeed live to become 10 billion years old? What initial conditions favor SSC longevity? Going to deeper limits in the cluster luminosity function could illuminate the connection between today's SSCs and the older globular cluster population (Chandar et al. 2004b). This is an extremely active area of research, but there are currently few examples of high resolution studies of the efficiency of star formation on GMC sizescales in starbursts. ALMA will have the sensitivity and resolution to enable these studies in many nearby galaxies.