The launch of NASA's Far Ultraviolet Spectral Explorer ( FUSE) mission in 1999 has finally allowed absorption from the OVI 1032, 1038 doublet to be detected in local starburst galaxies. These lines probe collisionally ionized gas at T ~ 3 × 105 K. These observations provide us with kinematic information on gas ~ 30 times hotter than probed by optical emission lines.
A theoretical prediction of both analytical and numerical models of superwinds is that hotter gas has higher outflow velocities than the cool gas. The maximum velocities are achieved by the very energetic SN-ejecta, while cooler denser ambient ISM is swept up and accelerated to a terminal velocity dependent on the column density of the clump or cloud (Chevalier & Clegg 1985). If true, hot gas might escape these galaxies even if outflow velocities in the cool phases are well below escape velocity.
If superwinds are to be stopped before reaching the IGM, it is necessary to radiate away the majority of their energy. Observations place strong limits on the radiative power loss in the X-ray band at ~ 1% or so of the SN energy injection rate SN, while optical emission lines account for a few to ~ 10% of SN. The only waveband where appreciable energy could be emerging that has not previously been explored is the far UV.
Blue-shifted absorption from OVI is seen in a variety of of starbursts of different mass and star-formation intensity, from starbursting dwarf galaxies like NGC 4214 (Martin et al, in preparation) through to typical starbursts like NGC 3310. Heckman et al. (2001) present a detailed case study of one the archetypal starbursting dwarf galaxies, NGC 1705, which shows a complicated optical morphology suggesting that hot gas is in the process of "blowing-out" of a ~ 2 kpc diameter H bubble. The FUSE observations reveal that the hot gas responsible for the OVI absorption has a higher outflow velocity than the warm ionized medium, which in turn has a higher outflow velocity than warm neutral gas (vOVI = - 77 ± 10 km/s, vWIM = - 53 ± 10 km/s, vWNM = - 32 ± 11 km/s). These observations are inconsistent with the standard superbubble model (Castor, Weaver, & McCray 1975; Mac Low & McCray 1988), but agree with the predictions gas entrainment and acceleration as hot gas flows out through holes in a fragmented superbubble shell to form a superwind. FUV radiative losses in NGC 1705 appear minimal, only ~ 5% of SN, so superwinds appear to be inefficient radiators at any wavelength.