3.3. Luminous and Ultraluminous Infrared Galaxies.

Given that the far-infrared energy output of a (dusty) galaxy is a direct measure of its star formation rate, it is not surprising a posteriori to find evidence for large-scale galactic winds in several luminous and ultraluminous infrared galaxies (LIRGs and ULIRGs; e.g., Heckman et al. 1990; Veilleux et al. 1995). Systematic searches for winds have been carried out in recent years in these objects to look for the unambiguous wind signature of blueshifted absorbing material in front of the continuum source (Heckman et al. 2000; Rupke et al. 2002). The feature of choice to search for outflowing neutral material in galaxies of moderate redshifts (z 0.6) is the Na ID interstellar absorption doublet at 5890, 5896 Å. The wind detection frequency derived from a set of 44 starburst-dominated LIRGs and ULIRGs is high, of order ~ 70 - 80% (Rupke et al. 2002, 2003 in prep.; also see Rupke's and Martin's contributions at this conference). The outflow velocities reach values in excess of 1700 km s^-1 (even more extreme velocities are found in some AGN-dominated ULIRGs).

A simple model of a mass-conserving free wind (details of the model are given in Rupke et al. 2002) is used to infer mass outflow rates in the range _tot(H) = few - 120 for galaxies hosting a wind. These values of _tot, normalized to the corresponding global star formation rates inferred from infrared luminosities, are in the range _tot / SFR = 0.01 - 1. The parameter , often called the "mass entrainment efficiency" or "reheating efficiency" shows no dependence on the mass of the host (parameterized by host galaxy kinematics and absolute R- and K'-band magnitudes), but there is a possible tendency for to decrease with increasing infrared luminosities (i.e. star formation rates). The large molecular gas content in ULIRGs may impede the formation of large-scale winds and reduce in these objects. A lower thermalization efficiency (i.e. higher radiative efficiency) in these dense gas-rich systems may also help explain the lower (Rupke et al. 2003, in prep.; see Rupke's contribution at this conference).