The far-UV spectral region includes line transitions that are sensitive to both the very hot (~ 105 K) and very cold (~ 102 K) interstellar material. Coronal gas with temperatures of several 100,000 K can be probed with the O VI line whose corresponding ionization potential is 114 eV. On the other hand, the rotational/vibrational transitions of H2 trace cool molecular gas.
The combined effects of multiple stellar winds and supernovae are capable of heating the interstellar medium (ISM) and initiating large-scale outflows. Such outflows are a significant sink for the gas reservoir. They have been known for some time from optical and X-ray imagery and have recently been analyzed with absorption-line spectroscopy (Heckman 2005). Heckman et al. (2001) obtained FUSE far-UV spectra of the nearby dwarf starburst NGC 1705, probing the coronal (105 - 106 K) and the warm (104 K) phases of the outflow. The kinematics of the warm gas are compatible with a simple model of the adiabatic expansion of a superbubble driven by the supernovae in the starburst. Radiative losses are negligible so that the outflow may remain pressurized over a characteristic flow time scale of 108 to 109 yr, as estimated from the size and velocity. The same conclusion was reached for M82 by Hoopes et al. (2003) who used FUSE to search for O VI emission in the starburst superwind of M82. No O VI emission was detected at any of the pointings. These observations limit the energy lost through radiative cooling of coronal phase gas to the same magnitude as that lost in the hot phase through X-ray emission, which has been shown to be small.
The total mass transported out of the starburst region via galactic superwinds is hard to constrain, given the uncertain ionization corrections and the strength of the observable spectral lines. Attempts were made by Johnson et al. (2000) and Pettini et al. (2000) for the nearby dwarf starburst galaxy He 2-10 and the luminous Lyman-break galaxy MS1512-cB58, respectively. In both cases the mass-loss rate of the ISM is rather similar to the star-formation rate. Taken at face value, this suggests that the available gas reservoir will not only be depleted by the star formation process but, more importantly, by removal of interstellar material. Starbursts may determine their own fate by their prodigious release of kinetic energy into the ISM.
The spectral range of FUSE includes numerous transitions of molecular hydrogen, making it possible to study H2 in diffuse interstellar environments directly through absorption measurements, rather than relying on the indirect CO technique. Hoopes et al. (2004) searched for H2 absorption in the five starburst galaxies NGC 1705, NGC 3310, NGC 4214, M83, and NGC 5253. Weak H2 absorption was detected in M83 and NGC 5253, and upper limits on the H2 column density were derived in the other three galaxies. The upper limits on the mass of molecular gas are orders of magnitude lower than the H2 mass inferred from CO emission measurements. The interpretation is that almost all of the H2 is confined to clouds with column densities in excess of 1020 cm-2 that are opaque to far-UV light and cannot be detected in the FUSE spectra. The far-UV continuum seen with FUSE originates from sightlines between the dense clouds, which have a covering factor < 1. This morphology is consistent with that of the interstellar dust, which is thought to be clumpy. The complex observational biases related to varying extinction across the extended UV emission in the FUSE apertures make it difficult to characterize the diffuse H2 in these starburst galaxies.