Many local starburst galaxies show blue-shifted Na D absorption line profiles (Phillips 1993; Heckman et al. 2000). The sodium D (NaI 5890, 5896) lines probe warm neutral gas, at a temperature of a few thousand degrees. In Heckman et al's sample of IR luminous starburst galaxies 19 out of a sample of 33 showed blue shifted absorption, typically extending out to terminal velocities between vterm = - 200 to -700 km/s (Fig. 2) in galaxies with rotation velocities between 140 and 330 km/s. Absorption is seen over a wide range in velocity, from the systemic velocity of the galaxy out to vterm, suggesting gas with multiple velocity components in the outflow. Those galaxies that show blue-shifted absorption tend to be more face-on than those that do not show absorption.
Figure 2. A representative sample of Na D absorption line profiles for four starburst galaxies, adapted from the larger sample shown in Heckman et al. (2000). The dashed vertical lines show the expected centroids of the Na D doublet at the systemic velocity of the galaxy. Horizontal bars represent a blue shift of 500 km/s. Note the strongly blue-shifted Na D absorption in NGC 3526, NGC 1808 and Mrk 273 due to the superwinds in these galaxies.
This is consistent with a model where the blue-shifted absorption features arise in cool ambient gas entrained into a weakly collimated outflow along the minor axis of the galaxy. The gas initially has low velocity, giving rise to absorption near the systemic velocity of the galaxy, but is accelerated to higher velocity by the ram pressure of the SN-ejecta wind.
This cool gas dominates the total mass in the outflow. For a typical starburst in this sample (LFIR ~ 2 × 1011 L) the mass of warm neutral gas is MNaD ~ 5 × 108Msun. The mass flow rate in this component significantly exceeds the mass injection rate due to SNe and stellar winds and is comparable to the gas consumption rate due to star formation (NaD ~ 3 - 10 × SN ~ SF). Although not highly metal-enriched, this component is significant in terms of total mass of metals transported out of the disk of the galaxy.
Does this gas escape the galaxy and pollute the IGM? The observed terminal velocities are typically several times the rotational velocity of the galaxy, comparable to or greater than the galactic escape velocity assuming vesc ~ 3 × vrot. It should be stressed that even if v < vesc, this does not automatically imply that the gas is retained by the galaxy. The motion of this cool gas is not simply ballistic, as the clouds are carried along by the wind. The long term fate of this gas is unknown, and depends more on on hydrodynamic forces (wind ram pressure, retardation by halo gas) than the gravitational potential of the galaxy.