Diffuse soft X-ray emission in superwinds is considerably easier to observe than diffuse hard X-ray emission, by virtue of being orders of magnitude more luminous and spatially extended. Nevertheless Chandra has also revolutionized our understanding of the origin and properties of soft diffuse X-ray emission in starbursts. The majority of starburst activity in the local universe occurs in galaxies with log LBOL(L) ~ 10.5, classic starbursts M82, NGC 253, or NGC 3628. Thus edge-on starburst galaxies such as these objects, at distances of 20 Mpc, provide the best observational data on superwinds.
Models such as Starburst99 (Leitherer et al., 1999) predict that the ratio of mechanical energy released by SNe and stellar winds to the bolometric radiated power from a starburst is LW / LBOL 0.01 to 0.015, where LBOL is proportional to the star formation rate. The fraction of the superwind mechanical power expected to emerge as soft X-ray emission is rather more uncertain. There are a variety of different theoretical models for the origin of the soft X-ray emitting gas. The merged SN ejecta driving the wind is too tenuous to provide significant X-ray emission, unless its density has been significantly increased by mass-loading (Suchkov et al., 1996). In our multi-dimensional hydrodynamical simulations of superwinds (Strickland & Stevens, 2000), where the emission is primarily due to shock heating of ambient disk and halo gas, we found LX / LW ~ 0.03, (with higher values associated with mass-loaded models). Similar values can be found if the Weaver et al. (1977) wind-blown bubble model is applied to superwinds (Strickland et al., 2004b), where the soft X-ray emission is primarily from conductively-heated ambient gas at the wall of the bubble. Combining these models, and including the scatter, theoretical models of starburst-driven winds predict log LX / LBOL -3.5 ± 0.6.
Chandra observations of many classic starbursts have now demonstrated that the soft thermal X-ray emission from the superwinds is highly-structured and spatially correlated with H emission, as expected if the X-ray emission is generated by the interaction of the superwind with cool ambient gas. In many cases both the X-ray and H emission is strongly limb-brightened, indicating that the emission comes predominantly from the walls of the cavity carved by the superwind into the disk and halo medium of the host galaxy (Strickland et al., 2002; Strickland et al., 2004a; Strickland et al., 2000). The fraction of the wind volume occupied by soft X-ray emitting gas is relatively low, with upper limits of < 20% of the soft X-ray emission coming from any volume-filling mass-loaded wind (Strickland et al., 2003; Strickland et al., 2000). Both the total diffuse soft X-ray luminosity LX, TOT and the superwind emission at heights z > 2 kpc from the disk LX, HALO, scale in direct proportion to the the host galaxies bolometric luminosity, log LX, TOT / LBOL -3.6 ± 0.2 and log LX, HALO / LBOL -4.4 ± 0.2 (Strickland et al., 2004b), remarkably tight correlations.
Chandra and XMM-Newton observations of many dwarf starbursts (LBOL 9) and ULIRGs (log LBOL 12) are now available (Ptak et al., 2003; Summers et al., 2003; Summers et al., 2004; Hartwell et al., 2004; Martin et al., 2002; Ott et al., 2003). The characteristics of the diffuse X-ray emission in these objects are essentially the same as those of the classic starbursts, with similar spectral properties (relative -to-Fe abundances) and LX / LBOL ratios (Grimes et al., in preparation).