4.5. Galactic-Scale Outflows
The near-IR spectroscopic survey by Pettini et al. (2001) confirmed a trend which had already been suspected on the basis of the optical (rest-frame UV) data alone. When the redshifts of the interstellar absorption lines, of the nebular emission lines, and of the resonantly scattered Ly emission line are compared within the same galaxy, a systematic pattern of velocity differences emerges in all LBGs observed up to now (see Figure 30). We interpret this effect as indicative of galaxy-wide outflows, presumably driven by the supernova activity associated with the star-formation episodes. Such `superwinds' appear to be a common characteristic of galaxies with large rates of star formation per unit area at high, as well as low, redshifts (e.g. Heckman 2001). They involve comparable amounts of matter to that being turned into stars (the mass outflow rate is of the same order as the star formation rate) and about 10% of the total kinetic energy delivered by the starburst (Pettini et al. 2000b). These outflows have a number of important astrophysical consequences.
First, they provide self-regulation to the star formation process - this is the `feedback' required by theorists (e.g. Efstathiou 2000; Binney, Gerhard, & Silk 2001) for realistic galaxy formation models. Galactic winds may well be the key factor at the root of the `evolutionary sequence' for LBGs just discussed (Section 4.4).
Second, they can distribute the products of stellar nucleosynthesis over large volumes of the intergalactic medium since the outflow speeds are likely to exceed the escape velocities in many cases. As we have seen, many LBGs are already metal-enriched at z = 3 and have by then been forming stars for much of the Hubble time. There is therefore at least the potential for widespread pollution of the IGM with metals, thereby explaining at least in part the results on the metallicity of the Ly forest described in Section 3.1 and 3.2).
Third, the outflowing hot gas is likely to `punch' through the neutral interstellar medium of the galaxies and provide a route through which Lyman continuum photons can leak out of the galaxies, easing the problem of how the universe came to be reionised (Steidel, Pettini, & Adelberger 2001). Indeed it now appears (Adelberger et al. 2002, in preparation) that LBGs have a substantial impact on the surrounding IGM, and that shock-ionisation by their winds leads to a pronounced `proximity effect' - the Ly forest is essentially cleared out by these outflows over radii of ~ 100h-1 kpc.