The outflow rates in superwinds should not be taken directly as the rates at which mass, metals, and energy escape from galaxies and are transported into the intergalactic medium. After all, the observable manifestations of the outflow are produced by material still relatively deep within the gravitational potential of the galaxy's dark matter halo. We know very little about the gaseous halos of galaxies, and it is possible that this halo gas could confine a wind that has blown-out of a galactic disk (Silich & Tenorio-Tagle 2001).
A necessary condition for wind escape is that radiative losses are not severe enough to drain energy from the wind, causing it to stall (e.g. Wang 1995). The X-ray luminosity of the wind is typically on of-order 1% of the rate at which the starburst supplies kinetic energy. Thus, radiative losses from hot ( T 106 K) gas will not be dynamically significant. The radiative cooling curve peaks in the so-called "coronal" regime ( T 105 to 106 K). The FUSE mission has now provided the first probe of coronal-phase gas in starbursts and their winds via the OVI1032,1038 doublet (Figure 2). Our analysis of these data imply that in no case is radiative cooling by the coronal gas sufficient to quench the outflow (Heckman et al. 2001a; Martin et al. 2001).
In the absence of severe radiative cooling, one instructive way of assessing the likely fate of the superwind material is to compare the observed or estimated outflow velocity to the estimated escape velocity from the galaxy. For an isothermal gravitational potential that extends to a maximum radius rmax, and has a circular rotation velocity vrot, the escape velocity at a radius r is given by:
In the case of the interstellar absorption-lines, Heckman et al. (2000) argued that the observed profiles were produced by material ablated off ambient clouds and accelerated by the wind up to a terminal velocity represented by the most-blueshifted part of the profile. In the case of the X-ray data, we do not measure a Doppler shift directly, but we can define a characteristic outflow speed vX corresponding to the observed temperature TX, assuming an adiabatic wind with a mean mass per particle (Chevalier & Clegg 1985):
This is a conservative assumption as it ignores the kinetic energy the X-ray-emitting gas already has (probably a factor typically 2 to 3 times its thermal energy - Strickland & Stevens 2000). Based on this approach, Heckman et al. (2000) and Martin (1999) found that the observed outflow speeds are independent of the galaxy rotation speed and have typical values of 400 to 800 km s-1. This suggests that the outflows can readily escape from dwarf galaxies, but possibly not from the more massive systems.
In all these discussions it is important to keep in mind the multiphase nature of galactic winds. It is possible (even likely) that the question of "escape" will have a phase-dependent answer. The relatively dense ambient interstellar material seen in absorption-lines, in optical line emission, and perhaps soft X-rays may be propelled only as far as the halo and then return to the disk. In contrast, the primary energy-carrying wind fluid (which could be flowing out at velocities of up to 3000 km s-1) could escape even the deepest galactic potentials and carry away much of the kinetic energy and metals supplied by the starburst. Moreover, for a realistic geometry, it is clearly much easier for a wind to blow-out of a galaxy's interstellar medium than than to blow it away (e.g. De Young & Heckman 1994; MacLow & Ferrara 1999).
How far out from the starburst can the effects of superwinds be observed? In general, such tenuous material will be better traced via absorption-lines against background QSOs than by its emission (since the emission-measure will drop much more rapidly with radius than will the column density). To date, the only such experiment that has been conducted is by Norman et al. (1996) who examined two sight-lines through the halo of the merger/starburst system NGC 520 using HST to observe the MgII2800 doublet. Absorption was definitely detected towards a QSO with an impact parameter of 35 h70-1 kpc and possibly towards a second QSO with an impact parameter of 75 h70-1 kpc. Since NGC 520 is immersed in tidal debris (as mapped in the HI 21cm line), it is unclear whether the MgII absorption is due to tidally-liberated or wind-ejected gas. We can expect the situation to improve in the next few years, as the Galex mission and the Sloan Digital Sky Survey provide us with 105 new QSOs and starburst galaxies, and the Cosmic Origins Spectrograph significantly improves the UV spectroscopic capabilities of HST.
While a wind's X-ray surface brightness drops rapidly with radius due to expansion and adiabatic cooling, its presence at large radii can be inferred if it collides with an obstacle. In the case of M 82, Lehnert, Heckman, & Weaver (1999) show that a ridge of diffuse X-ray and H emission at a projected distance of 12 kpc from the starburst is most likely due to a wind/cloud collision in the galaxy halo. An even more spectacular example (Irwin et al. 1987) is the peculiar tail of HI associated with the galaxy NGC 3073 which points directly away from the nucleus of its companion: the superwind galaxy NGC 3079 (50 h70-1 kpc away from NGC 3073 in projection). Irwin et al. (1987) proposed that the HI tail is swept out of NGC 3073 by the ram pressure of NGC 3079's superwind.