The dynamics and high-energy emission of conductive gas clouds in supernova-driven galactic superwinds

Superwinds from starburst galaxies are multiphase outflows that sweep up and incorporate ambient galactic disc and halo gas. The interaction of this denser material with the more diffuse hot wind gas is thought to give rise to the O vi emission and absorption in the far ultraviolet (FUV) and the soft thermal X-ray emission observed in superwinds. In this paper, we present high-resolution hydrodynamical models of warm ionized clouds embedded in a superwind, and compare the O vi and soft X-ray properties to the existing observational data. These models include thermal conduction, which we show plays an important role in shaping both the dynamics and radiative properties of the resulting wind/cloud interaction. Heat conduction stabilizes the cloud by inhibiting the growth of Kelvin–Helmholtz and Rayleigh–Taylor instabilities, and also generates a shock wave at the cloud's surface that compresses the cloud. This dynamical behaviour influences the observable properties. We find that while O vi emission and absorption always arises in cloud material at the periphery of the cloud, most of the soft X-ray arises in the region between the wind bow shock and the cloud surface, and probes either wind or cloud material depending on the strength of conduction and the relative abundances of the wind with respect to the cloud. In general, only a small fraction (≲1 per cent) of the wind mechanical energy intersecting a cloud is radiated away at ultraviolet (UV) and X-ray wavelengths, with more wind energy going into accelerating the cloud. Clouds in relatively slow cool winds radiate a larger fraction of their energy, which are inconsistent with observational constraints. Models with heat conduction at Spitzer-levels are found to produce observational properties closer to those observed in superwinds than models with no thermal conduction, in particular, in terms of the O vi to X-ray luminosity ratio, but cloud life times are uncomfortably short (≲1 Myr) compared to the dynamical ages of real winds. We experimented with reducing the thermal conductivity for one set of model parameters, and found that even when we reduced conduction by a factor of 25 that the simulations retained the beneficial hydrodynamical stability and low O vi to X-ray luminosity ratio found in the Spitzer-level conductive models, while also having reduced evaporation rates. Although more work is required to simulate clouds for longer times and to investigate cloud acceleration and thermal conduction at sub-Spitzer levels in a wider range of models, we conclude that thermal conduction can no longer be ignored in superwinds.