Typical starburst galaxies (LFIR ~ L*) show high velocity (200-700 km/s) multi-phase outflows. The observed mass flow rates in the wind are comparable to the gas consumption rate due to star formation (wind *), and are dominated by relatively cool ambient gas (T ~ few thousand K) that has been swept-up and accelerated by the ram pressure of the hotter wind of SN-ejecta.
Outflow velocities are typically comparable or greater than estimates of the galactic escape velocity, but caution should be exercised in making claims about mass loss rates. The gas motions are not ballistic, making it impossible to give quantitative observational mass loss rates. Observations of gas at much larger galactocentric radii (100 kpc) are needed to directly observe mass loss. Nevertheless, the observed mass flow rates are considerable, and does seem likely that some significant fraction of even the coolest phases may well escape even moderately massive starburst galaxies. Speaking as a practitioner of hydrodynamical simulations of superwinds, I am not convinced that we know mass-loss rates theoretically. Existing models have yet to be meaningfully tested against observations. The lack of sub-parsec numerical resolution in current simulations prejudices the ability of these models to treat mass transport in winds.
FUSE observations of OVI absorption provide vital information of the kinematics and radiative losses of coronal gas at T ~ 3 × 105 K. In the dwarf starburst NGC 1705 the FUSE observations support the theoretical prediction of superwind models that the hotter phases in superwinds have higher outflow velocities than the cooler phases. This suggests that the even hotter material holding the metals is more likely to escape than the warm neutral and ionized gas. Radiative energy losses within the wind appear minimal compared to the energy injection rate from SNe, even within the FUV and X-ray wave-bands.
With the sub-arcsecond spatial resolution provided by Chandra it is now clear that the soft thermal X-ray emission seen in superwinds is due to some form of interaction between the (still invisible) high velocity hot wind and cooler denser ambient gas swept-up or overrun by the flow. This is unfortunate in the sense that X-ray observations do not provide a direct probe of the energetic metal-enriched gas driving these winds. Nevertheless the data from Chandra allow us to obtain more accurate estimates of the physical properties of the X-ray emitting gas than ever before, and provide deeper insight into the conditions within these winds.
Starburst-driven winds are difficult objects to study, due to the range of different gas phases involved and the faintness of the emission. Nevertheless, very significant progress is being made, most notably due to the new observational capabilities provided by FUSE & Chandra. The wealth of multi-wavelength data will place extremely strong constraints upon numerical models of superwinds. This is an exciting time to study superwinds, and there is no prospect of an end to new discoveries about these fascinating and important objects.
Acknowledgments. It is a pleasure to thank Tim Heckman for countless enlightening discussions over the years, Crystal Martin and Gerhardt Meurer for providing a variety of spectra and images, and Francesca Matteucci for organizing a stimulating workshop. DKS gratefully acknowledges the support from Chandra Postdoctoral Fellowship Award Number PF0-10012, issued by the Chandra X-ray Observatory Center, operated by the SAO on behalf of NASA.