5.1. Heating & Enriching the IGM
The ejection of highly metal-enriched material by superwinds offers a very natural explanation for the long-standing puzzle that the intra-cluster medium in clusters of galaxies contains at least as much mass in the form of metals as do the galaxies in the cluster (e.g., Sarazin 1986). This idea has been explored by many authors, including Larson & Dinerstein (1975), DeYoung (1978), and (more recently) White (1991) and David, Forman, & Jones (1991).
Superwinds can heat the JCM as well as enrich it. The combination of gravitational and superwind heating of the ICM could readily account for the result that the temperature of the ICM seems to exceed the 'temperature' of the galaxies as they move within the cluster potential. That is, the parameter defined as µ mH2gal / kTICM may be systematically less than unity in clusters (cf. David, Forman, & Jones 1991; White 1991). A good quantitative match of White's model to the data on superwinds and the ICM was found by HAM.
The idea that the ICM has been enriched and heated by superwinds makes several specific and testable predictions (cf. David, Forman, & Jones 1991; White 1991; and Donahue's contribution to this volume):
While direct evidence for a hot, all-pervasive intergalactic medium remains scant, there are compelling theoretical reasons to believe it exists (cf. the recent review by Shapiro 1989). In order that a cosmologically significant IGM not violate Gunn-Peterson limits, it must be ionized at quite early epochs (e.g., already by the time at which the first known quasars appear). The mechanical energy input from superwinds driven by protogalaxies may be capable of heating and collisionally ionizing the IGM in the early 'pre-quasar' era of the universe (cf. Miralda-Escude & Ostriker 1991).
5.2. Galaxy Formation & Evolution
The most remarkable and profound constraints on our ideas of how galaxies actually form are the relatively tight relationships between such disparate quantities as velocity dispersion, mass and luminosity, length-scale, surface brightness, and chemical abundances (e.g., Djorgovski & Davis 1987; Dressler et al. 1987; Franx & Illingworth 1990). Recent attempts to understand the processes that regulated the conversion of gas into stars and the subsequent chemical enrichment in proto-galaxies have led to the concept of 'self-regulated galaxy formation' or 'feedback' (e.g., Silk 1987; White & Frenk 1991): the effect of the kinetic energy supplied by massive stars and supernovae on the gas in a (proto)galaxy may lead naturally to the creation of the specific properties of galaxies we see today.
For example, Dekel & Silk (1985), Vader (1986), and Silk, Wyse, & Shields (1987) have argued that the striking differences between many of the global properties of dwarf and massive galaxies may reflect the existence of a critical escape velocity (potential well depth) below which galaxies suffer catastrophic superwind-driven loss of their ISM following the initial burst of star-formation that accompanies their formation. This process leads naturally to the low surface brightnesses and very low metal abundances of dwarf galaxies (see also DeYoung & Gallagher 1990). The catastrophic destruction of dwarfs by starburst-driven winds might be related to the apparently rapid disappearance of the 'faint, blue galaxies' (starbursting low mass systems) during the last few Gyr of the history of the Universe (cf. Broadhurst, Ellis, & Glazebrook 1992; Cowie et al. 1992).
Wyse & Silk (1985) have examined the consequences of supernova heating for the chemical enrichment of disk galaxies. Similarly, Larson & Dinerstein (1975), Vader (1986), Franx & Illingworth (1990), Lynden-Bell (1992) and others have proposed that superwinds are the mechanism that underlies the correlation between the metal abundance of the stellar population and the local escape velocity in elliptical galaxies (i.e., both the mass vs. metallicity relation between galaxies and the metallicity vs. radius relation within galaxies). That is, the deeper the local gravitational potential, the more difficult to drive a wind and hence the greater the ability of that locale to retain metal-rich processed material.
Mathews (1989) has explored the consequences of superwinds in normal ellipticals. He argues that the observed mass-to-light ratio of the stellar population is set by superwind-driven mass loss. Superwinds might also play an important role in the evolution of the star-bursting product of the merger of two major disk galaxies into a giant elliptical (see Hernquist's contribution to this volume): superwinds could shock-heat the cool ISM up to X-ray temperatures or even eject much of the ISM from the galaxy (e.g., Mathews & Baker 1971).
It has even been suggested that the superwinds ('explosions') associated with the formation of galaxies may have played some role in the formation of structure in the universe (Ostriker & Cowie 1981). They propose that superwinds swept up the IGM into large shells of dense radiatively-cooled material that then fragmented into Jeans-unstable lumps that became new galaxies. If the process of amplification via 'stimulated galaxy formation' is efficient enough, then superwinds may have helped induce structure in the early universe.
5.3. QSO Absorption Lines
The nature of the intervening clouds responsible for producing the absorption-lines in the spectra of QSO's remains unclear (cf. the collection of reviews in the volume compiled by Blades, Turnshek, & Norman 1988 - see also the reviews by Steidel, Lanzetta, Bechtold, Weymann, and Bajtlik in this volume). The sharp metal lines and Lyman-limit systems are believed to arise in the gaseous halos of intervening galaxies, and the implied cross-sectional area of absorbing material per galaxy significantly exceeds the traditional 'optical' dimensions of galaxies (even if every galaxy in the universe has such an extended gaseous halo).
While it is by no means obvious that superwinds are consistent with the myriad properties observed for the various classes of QSO absorption-line systems, at the very least, they offer a novel way to transport metals out into a galactic halo (or even out into the general IGM), and could perhaps help us understand some of the most salient properties of the absorption-line systems:
One obvious test of these ideas is to obtain far-UV spectra of QSO's viewed through the halos of known starburst/superwind galaxies, and then to compare these spectra to spectra of typical high-redshift QSO sharp metal-line systems. Some of us have recently embarked on an HST program to do that for a sample of QSOs seen behind the halos of the nearby starburst galaxies NGC 253, NGC 520, and NGC 3079.
5.4. The X-Ray background
Another enduring mystery concerns the origin of the cosmic X-ray background (CXRB). Bookbinder et al. (1980) first suggested that X-ray emission from the hot thermal plasma in superwinds might make an important contribution to the CXRB. More recently, Weedman (1987) and Griffiths & Padovani (1990) have shown that the X-ray luminosity function of local starburst galaxies could provide a significant fraction of the CXRB if they evolve by only a modest amount with cosmic epoch.
It is worth emphasizing however that even if superwinds do make a non-negligible contribution to the CXRB, this will occur only at relatively low energies. More specifically, superwinds are not expected to radiate significantly above about 10 keV (see Section 2 above), while most of the overall energy in the CXRB is in the range 10-40 keV. If the superwinds contributing to the CXRB are located at a significant mean redshift this will only exacerbate this spectral deficiency at high energies.
5.5. The Disk-Halo Connection
One of the principal themes in the study of the interstellar medium (ISM) is the relationship between massive stars and the thermal and dynamical state of the ISM. Recently, there has been an explosive growth in the number of papers on this topic, spurred in part by the discovery of expanding HI 'supershells' (Heiles 1990) and of a warm, ionized medium which has a relatively large scale-height and contains a substantial fraction of the Milky Way's total gas mass (Reynolds 1991). As reviewed by Dettmar (1993), there also mounting evidence of a 'disk-halo connection' in external galaxies by which star-formation in the disk ISM can propel mass and thermal/kinetic energy up into the galactic halo.
Starbursts offer wonderful laboratories for studying these processes, with the advantage that conditions are so extreme in these systems that the number of relevant physical processes may be fewer than in a normal disk-halo system. Even if it does not prove to be easier to understand the disk-halo connection in starbursts, the spectacular superwind phenomena we have described above demonstrate that is clearly easier to observe the disk-halo connection there!
The ultimate goal of course will be to develop a comprehensive theory of the feedback effect of the formation of massive stars that explains a whole continuum of disk-halo systems from normal star-forming disks with their galactic fountains or chimneys to powerful starbursts with global superwinds. Some important steps in this direction have already been taken (cf. Norman & Ikeuchi 1989).
5.6. The Starburst-AGN Connection & Ultraluminous Galaxies
One of the most fascinating issues in extragalactic astronomy concerns the relationship between AGNs and starbursts (cf. Heckman 1991; Blandford 1992; Filippenko 1993). Roberto Terlevich and his colleagues have argued that ultra-compact starbursts occurring in conditions of very high gas density can reproduce all the phenomena associated with classical radio-quiet AGNs (e.g., Terlevich 1992). Our contention here is more limited in scope, but similar in spirit: superwinds driven by starbursts can significantly blur the phenomenological distinction between starbursts and AGNs.
First, superwinds are estimated to have terminal velocities of several thousand km s-1, and so (at least in principle) can accelerate gas up to velocities well in excess of normal galaxian orbital velocities. Thus, the mere presence of broad emission-lines does not necessarily imply that a galaxy harbors an AGN.
Second, ambient material that has been shock-heated by the superwind will produce LINER-type line ratios (which again are often taken as firm evidence for the presence of an AGN). This classification problem is exacerbated by the much heavier obscuration suffered by the optical line emission produced within the central starburst compared to the emission originating in the superwind (thereby making it harder to clearly recognize the signature of the OB-star-photoionized gas in the integrated optical spectra of distant and powerful starbursts).
Third, superwinds are potential sources of X-rays (at least up to 10 keV or so). Now X-rays comprise an energetically trivial fraction of the bolometric luminosity (typically few times 10-4) in IR-selected starburst galaxies (e.g., Griffiths & Padovani 1990; Green, Anderson, & Ward 1992). Since the fractional contribution of the X-ray luminosity is several orders-of-magnitude larger than this in type 1 Seyfert nuclei and QSOs (e.g., Sanders et al. 1989), there is no difficulty in using X-ray data to differentiate between these AGN classes and starbursts (cf. Rieke 1988). On the other hand, there is a considerable amount of overlap in the distribution of LX / Lbol) between starbursts and type 2 Seyfert galaxies (e.g., Green, Anderson, & Ward 1992). Indeed, Wilson et al. (1992) have suggested that much of the hard X-ray emission from the type 2 Seyfert NGC 1068 is associated with the circumnuclear starburst rather than with the AGN itself.
Finally, superwinds are also capable of producing large-scale radio sources that extend at least 10 kpc out into the halo of the starburst galaxy. Such radio sources are very weak (the total core-plus-halo radio luminosity of M 82 is only about 10-5 of the bolometric luminosity of the starburst), and show only a weakly bipolar morphology. Thus, we suggest that spatially-extended (>kpc-scale) radio emission should be taken as unambiguous evidence for the presence of an AGN only if the radio source provides a much larger fraction than this of the bolometric luminosity and/or exhibits a highly-collimated, jet-type morphology.
Taking these points together, and noting that the observational manifestations of superwinds become more spectacular with increasing starburst luminosity, we suggest that the evidence that most 'ultraluminous' IRAS galaxies are powered by buried quasars (e.g., Sanders et al. 1988) is not entirely persuasive, Indeed - based especially on the near-IR colors and the IR spectral energy distributions of the ultraluminous galaxies (see Sanders et al 1988 figures 14 & 15) - we suggest that only three of the ten ultraluminous galaxies (namely IRAS 05189-2524, IRAS 08572+3915, and Mrk 231) clearly have ultraluminous AGNs in their cores (see also Majewski et al. 1993).