We present a detailed study of the hydrodynamics of the matter reinserted by massive stars via stellar winds and supernovae explosions in young assembling galaxies. We show that the interplay between the thermalization of the kinetic energy provided
by massive stars, radiative cooling of the thermalized plasma and the gravitational pull of the host galaxy, lead to three different hydrodynamic regimes. These are: a) The quasi-adiabatic supergalactic winds. b) The bimodal flows, with mass accumulation in the central zones and gas expulsion from the outer zones of the assembling galaxy. c) The gravitationally bound regime, for which all of the gas returned by massive stars remains bound to the host galaxy and is likely to be reprocessed into futher generations of stars. Which of the three possible solutions takes place, depends on the mass of the star forming region its mechanical luminosity (or star formation rate) and its size. The model predicts that massive assembling galaxies with large star formation rates similar to those detected in SCUBA sources ($sim 1000$ M$_odot$ yr$^{-1}$) are likely to evolve in a positive star-formation feedback condition, either in the bimodal, or in the gravitationally bound regime. This implies that star formation in these sources may have little impact on the intergalactic medium and result instead into a fast interstellar matter enrichment, as observed in high redshift quasars.
The origin of supersonic infrared and radio recombination nebular lines often detected in young and massive superstar clusters are discussed. We suggest that these arise from a collection of repressurizing shocks (RSs), acting effectively to re-estab
lish pressure balance within the cluster volume and from the cluster wind which leads to an even broader although much weaker component. The supersonic lines are here shown to occur in clusters that undergo a bimodal hydrodynamic solution (Tenorio-Tagle et al. 2007), that is within clusters that are above the threshold line in the mechanical luminosity or cluster mass vs the size of the cluster (Silich et al. 2004). The plethora of repressurizing shocks is due to frequent and recurrent thermal instabilities that take place within the matter reinserted by stellar winds and supernovae. We show that the maximum speed of the RSs and of the cluster wind, are both functions of the temperature reached at the stagnation radius. This temperature depends only on the cluster heating efficiency ($eta$). Based on our two dimensional simulations (Wunsch et al. 2008) we calculate the line profiles that result from several models and confirm our analytical predictions. From a comparison between the predicted and observed values of the half-width zero intensity of the two line components we conclude that the thermalization efficiency in SSCs above the threshold line must be lower than 20%.
Using the hydrodynamic code ZEUS, we perform 2D simulations to determine the fate of the gas ejected by massive stars within super star clusters. It turns out that the outcome depends mainly on the mass and radius of the cluster. In the case of less
massive clusters, a hot high velocity ($sim 1000$ km s$^{-1}$) stationary wind develops and the metals injected by supernovae are dispersed to large distances from the cluster. On the other hand, the density of the thermalized ejecta within massive and compact clusters is sufficiently large as to immediately provoke the onset of thermal instabilities. These deplete, particularly in the central densest regions, the pressure and the pressure gradient required to establish a stationary wind, and instead the thermally unstable parcels of gas are rapidly compressed, by a plethora of re-pressurizing shocks, into compact high density condensations. Most of these are unable to leave the cluster volume and thus accumulate to eventually feed further generations of star formation. The simulations cover an important fraction of the parameter-space, which allows us to estimate the fraction of the reinserted gas which accumulates within the cluster and the fraction that leaves the cluster as a function of the cluster mechanical luminosity, the cluster size and heating efficiency.
