We have obtained moderate resolution (R \ few thousand) spectra of the Na I jj5890, 5896 (Na D) absorption line in a sample of 32 far-IRÈbright starburst galaxies. In 18 cases, the Na D line in the nucleus is produced primarily by interstellar gas, while cool stars contribute signiÐcantly in the others. In 12 of the 18 "" interstellar-dominated ÏÏ cases the Na D line is blueshifted by over 100 km s~1 relative to the galaxy systemic velocity (the "" outÑow sources ÏÏ), while no case shows a net redshift of more than 100 km s~1. The absorption-line proÐles in these outÑow sources span the range from near the galaxy systemic velocity to a maximum blueshift of D400È600 km s~1. The outÑow sources are galaxies systematically viewed more nearly face-on than the others. We therefore argue that the absorbing material consists of ambient interstellar material that has been entrained and accelerated along the minor axis of the galaxy by a hot starburst-driven superwind. The Na D lines are optically thick, but indirect arguments imply total hydrogen column densities of cm~2. This implies that the superwind N H D few ] 1021 is expelling matter at a rate comparable to the star formation rate. This outÑowing material is evidently very dusty : we Ðnd a strong correlation between the depth of the Na D proÐle and the line-of-sight reddening. Typical implied values are E(B[V ) \ 0.3È1 over regions several-to-10 kpc in size. We brieÑy consider some of the potential implications of these observations. The estimated terminal velocities of superwinds inferred from the present data and extant X-ray data are typically 400È800 km~1, are independent of the galaxy rotation speed, and are comparable to (substantially exceed) the escape velocities for (dwarf) galaxies. The resulting selective loss of metals from shallower potential wells can establish L * the mass-metallicity relation in spheroids, produce the observed metallicity in the intracluster medium, and enrich a general IGM to of order 10~1 solar metallicity. If the outÑowing dust grains can survive their journey into the IGM, their e †ect on observations of cosmologically distant objects would be signiÐcant.

Starburst‐driven galactic winds are responsible for the transport of mass, in particular metal‐enriched gas, and energy out of galaxies and into the intergalactic medium. These outflows directly affect the chemical evolution of galaxies, and heat and enrich the intergalactic and intercluster medium. Currently, several basic problems preclude quantitative measurements of the impact of galactic winds: the unknown filling factors of, in particular, the soft X‐ray‐emitting gas prevent accurate measurements of densities, masses and energy content; multiphase temperature distributions of unknown complexity bias X‐ray‐determined abundances; unknown amounts of energy and mass may reside in hard to observe T∼105 K and T∼107.5 K phases; and the relative balance of thermal versus kinetic energy in galactic winds is not known. In an effort to address these problems, we perform an extensive hydrodynamical parameter study of starburst‐driven galactic winds, motivated by the latest observation data on the best‐studied starburst galaxy M82. We study how the wind dynamics, morphology and X‐ray emission depend on the ISM distribution of the host galaxy, the starburst star formation history and strength, and the presence and distribution of mass‐loading by dense clouds. We also investigate and discuss the influence of finite numerical resolution on the results of these simulations. We find that the soft X‐ray emission from galactic winds comes from low filling factor (η≲2 per cent) gas, which contains only a small fraction (≲10 per cent) of the mass and energy of the wind, irrespective of whether the wind models are strongly mass‐loaded or not. X‐ray observations of galactic winds do not directly probe the gas that contains the majority of the energy, mass or metal‐enriched gas in the outflow. X‐ray emission comes from a complex phase‐continuum of gas, covering a wide range of different temperatures and densities. No distinct phases, as are commonly assumed when fitting X‐ray spectra, are seen in our models. Estimates of the properties of the hot gas in starburst galaxies based on fitting simple spectral models to existing X‐ray spectra should be treated with extreme suspicion. The majority of the thermal and kinetic energy of these winds is in a volume‐filling hot, T∼107 K, component which is extremely difficult to probe observationally because of its low density and hence low emissivity. Most of the total energy is in the kinetic energy of this hot gas, a factor that must be taken into account when attempting to constrain wind energetics observationally. We also find that galactic winds are efficient at transporting large amounts of energy out of the host galaxy, in contrast to their inefficiency at transporting mass out of star‐forming galaxies.

We measure the net energy efficiency of supernova and stellar wind feedback in the starburst galaxy M82 and the degree of mass-loading of the hot gas piston driving its superwind by comparing a large suite of 1 and 2dimensional hydrodynamical models to a set of observational constraints derived from hard X-ray observations of the starburst region (the fluxes of the Heα and Lyα-like lines of S, Ar, Ca and Fe, along with the total diffuse E = 2 -8 keV X-ray luminosity). These are the first direct measurements of the feedback efficiency and mass-loading of supernova heated and enriched plasma in a starburst galaxy. We consider a broad range of plausible parameters for the M82 starburst, varying the age and mode of star formation, the starburst region size and geometry, and supernova metal yields. Over all these varied input parameters all the models that satisfy the existing observational constraints have medium to high thermalization efficiencies (30% ≤ ǫ ≤ 100%) and the volume-filling wind fluid that flows out of the starburst region is only mildly centrally mass loaded (1.0 ≤ β ≤ 2.8). These results imply a temperature of the plasma within the starburst region in the range 30 -80 million Kelvin, a mass flow rate of the wind fluid out of the starburst region ofṀ tot ∼ 1.4 -3.6 M ⊙ yr −1 and a terminal velocity of the wind in the range v ∞ = 1410 -2240 km s −1 . This velocity is considerably larger than the escape velocity from M82 (v esc 460 km s −1 ) and the velocity of the Hα emitting clumps and filaments within M82's wind (v Hα ∼ 600 km s −1 ). Drawing on these results we provide a prescription for implementing starburst-driven superwinds in cosmological models of galaxy formation and evolution that more accurately represents the energetics of the hot metal-enriched phases than the existing recipes do.

We present a detailed case study of the diffuse X-ray and Hα emission in the halo of NGC 253, a nearby edge-on starburst galaxy driving a galactic superwind. The arcsecond spatial resolution of the ACIS imaging spectroscope on the Chandra X-ray Observatory allows us to study the spatial and spectral properties of the diffuse X-ray emitting plasma, at a height of between 3 and 9 kpc above the disk in the northern halo of NGC 253, with greatly superior spatial and spectral resolution compared to previous X-ray instruments. We find statistically significant structure within the diffuse emission on angular scales down to ∼ 10 ′′ (∼ 130 pc), and place limits on the luminosity of any X-ray-emitting "clouds" on smaller scales. There is no statistically significant evidence for any spatial variation in the spectral properties of the diffuse emission over scales from several ∼ 400 pc to ∼ 3 kpc. The spectrum of the diffuse X-ray emission is clearly thermal, although with the higher spectral resolution and sensitivity of Chandra it is clear that current simple spectral models do not provide a physically meaningful description of the spectrum. In particular, the fitted metal abundances are unphysically low. There is no convincing evidence for diffuse X-ray emission at energies above 2 keV in the halo.We show that the X-shaped soft X-ray morphology of the superwind previously revealed by ROSAT is matched by very similar X-shaped Hα emission, extending at least 8 kpc above the plane of the galaxy. In the northern halo the X-ray emission appears to lie slightly interior to the boundary marked by the Hα emission. The total 0.3 -2.0 keV energy band X-ray luminosity of the northern halo L X ∼ 5 × 10 38 erg s −1 , is very similar to the halo Hα luminosity of L Hα ∼ 4 × 10 38 erg s −1 , both of which are a small fraction of the estimated wind energy injection rate of ∼ 10 42 erg s −1 from supernovae in the starburst. We show that there are a variety of models that can simultaneously explain spatially-correlated X-ray and Hα emission in the halos of starburst galaxies, although the physical origin of the various emission components can be very different in different models. These findings indicate that the physical origin of the X-ray-emitting milliondegree plasma in superwinds is closely linked to the presence of much cooler and denser T ∼ 10 4 gas, not only within the central kpc regions of starbursts, but also on ∼ 10 kpc-scales within the halos of these galaxies.

We have analyzed Chandra ACIS observations of 32 nearby spiral and elliptical galaxies and present the results of 1441 X-ray point sources that were detected in these galaxies. The total point-source X-ray (0.3−8.0 keV) luminosity L XP is well correlated with the B-band, K-band, and FIR+UV luminosities of spiral host galaxies, and is well correlated with the B-band and K-band luminosities for elliptical galaxies. This suggests an intimate connection between L XP and both the old and young stellar populations, for which K and FIR+UV luminosities are reasonable proxies for the galaxy mass M and star-formation rate SF R. We derive proportionality constants α = 1.3 × 10 29 erg s −1 M −1