The concept of the spiral nebulae as ``island universes'' originally arose from the debate as to whether these were separate systems, distinct from our own Milky Way galaxy. Today, we might well ask the same question, but in relation to whether a galaxy ``communicates'' with its environment. That is, for any given ``isolated'' galaxy, whether or not it is surrounded by a dark matter halo, is there any significant connection between this system and the intergalactic medium (IGM) around it? An obvious connection would exist if there were some net mass flow from one of these environments to the other, for example, through gaseous inflows or outflows. Another might be through interconnecting magnetic field lines. Few galaxies yet have sufficient data to draw firm conclusions about these possibilities, but the cases for which data are accumulating are particularly interesting.
The most dramatic examples of inflows are the cooling flows associated with central galaxies in clusters. These hot (i.e. 107 -> 108 K) quasi-static flows are observed in X-rays where they emit by thermal Bremsstrahlung. The flows are massive, containing more mass than the visible stars, and the material is enriched, rather than primordial. A pressing problem is the fate of the cooled gas which cannot go into ``normal'' star formation since the central galaxies would then be much bluer than observed. Since mass deposition rates are estimated to be typically several hundred M yr>sup>-1 whereas optical observations suggest massive star formation rates of only a few M yr-1, it would appear that most of the mass must be going preferentially into low mass stars, ``dark objects'', or cool clouds. In the case of NGC 1275 (Fig. 1f) at the center of the Perseus Cluster cooling flow, a population of blue objects, ostensibly young globular clusters, have recently been detected with the HST (Holtzmann et al. 1992), at first glance, supporting the hypothesis of low-mass star formation from the cooling flow. However, the apparent colours of these clusters suggest that the constituent stars are not particularly low in mass and, if formed continuously from the flow, the colours would also not appear so uniform (unless the clusters are also continuously disrupted). Therefore, an alternative interpretation is that the young globulars may have formed from an interaction. NGC 1275 is also the only cooling flow galaxy in which cold molecular gas (i.e. CO) has been directly detected (Lazareff et al. 1989; Mirabel et al. 1989; Reuter et al. 1993; Bridges and Irwin 1995), though the presence of molecular gas in other cooling flow galaxies has been inferred from X-ray absorption data and has also been modeled as forming from the condensing cooling flow gas (Ferland et al. 1994). Future studies of the molecular component at high resolution and sensitivity may be fruitful in linking the large-scale cooling flow to star formation processes in the galaxy, and in identifying the missing reservoir of material. For a recent review of cooling flows, see Fabian (1994).
Related to hot gas in clusters of galaxies is the well-known Sunyaev-Zel'dovich (S-Z) effect (Sunyaev and Zel'dovich 1980) in which photons from the 2.7 K cosmic background are (inverse-Compton) up-scattered off of the hot electrons. The observed background spectrum is thereby distorted, the lower frequency emission being depressed and the high frequency emission boosted. A different distortion due to the peculiar velocity of the cluster also occurs which, in principle, is separable from the S-Z effect itself. In conjunction with X-ray observations, measurements of the S-Z effect can lead to a luminosity-independent estimate of the cluster distance and therefore of the Hubble constant. Detection of the S-Z effect requires observations at cm -> sub-mm wavelengths, with the higher frequency observations preferred because of fewer background confusing sources in this regime (Fischer et al. 1993). For reviews of the Sunyaev-Zel'dovich effect, see Sunyaev and Zel'dovich 1980, Raphaeli 1990, and Birkinshaw 1990.
There are a few cases in which inflows of cool material may be inferred, the most well-known being the high velocity clouds (HVCs) of our own Galaxy which are observed in neutral hydrogen. Many of these are associated with the Magellanic Stream which stretches over 180° in the sky from the Magellanic Clouds towards the Galaxy (see Mebold 1991), presumably due to a tidal interaction. There are a number of examples of possible impacts of HVCs with the Galaxy disk (see list in Mebold 1991 and Comerón and Torra 1994 and references therein), at least some of which are likely from HVCs originating in the Magellanic Clouds. Other origins for HVCs have also been proposed, however, including the condensation/cooling and subsequent infall of hot gas which has been expelled into the Galaxy halo from supernova explosions in the disk (Bregman 1980). For a recent summary of disk-halo interaction models for the Galaxy, see McKee 1993. There is no direct observational evidence yet for infalling cool clouds in external galaxies. However, impacting cool clouds or even gas rich dwarf galaxies have been suggested to explain the large HI holes and related high velocity features observed in some approximately face-on galaxies (e.g. van der Hulst and Kamphuis 1991) as well as the kpc scale HI arcs and extensions observed directly in some edge-on galaxies (see references in Irwin 1994). The inferred large energies ( 1053 erg) of some of these features are greater than expected from collective supernova explosions and stellar winds in typical OB associations (see Section 4.3 below), but might result from external clouds ``splashing'' onto the disk (see review by Tenorio-Tagle and Bodenheimer 1988). In these examples, however, the source of the infalling gas is not the intergalactic medium, but rather a galaxy, i.e. a smaller galaxy, a tidal shred of an external galaxy, or condensed gas which was expelled from the victim galaxy, itself.
Infalling gas has also been invoked to explain on-going galaxy formation and/or evolution. For example, accretion of smaller objects is required in ``bottom-up'' theories of galaxy formation (see Evrard 1993), a process which may be continuing to the present time, although Tóth and Ostriker 1992 argue for a rather limited accretion rate at the present epoch (i.e. Larson et al. 1980), i.e. the timescale for consuming all available disk gas by star formation is less than the timescale inferred from the star formation history of the galaxy. The suggested sources of infalling gas are large residual envelopes from galaxy formation, gas-rich companions, and tidal debris. In the model of Pfenniger et al. 1994 (see Section 3.2), infalling cold primordial gas (which is assumed to be the dark matter) contributes to on-going star formation as it accretes onto the disk. The infalling gas hypothesized in these scenarios has so far not been observed. Possible exceptions are the few cases in which large galaxies appear to be accreting smaller ones (``galactic cannibalism'') if this process can indeed be interpreted as a modern day extrapolation of bottom-up themes. Galaxy interactions and mergers will be discussed in more detail in Section 5.
In contrast to inflows, the evidence for outflows from galaxies, especially nuclear outflows, is abundant, not because the fraction of galaxies displaying nuclear outflows is large but because, observationally, such outflows can be spectacular. Nuclear outflows have been observed since the early days of radio astronomy and there is now a large body of literature describing them, from the powerful classical extragalactic radio sources associated with elliptical galaxies, to the weaker bipolar radio sources in some spirals with active galactic nuclei (AGN), to the bipolar ``superwinds'' associated with nuclear starbursts. For recent overviews of AGN as well as nuclear starbursting, see, e.g. the review by Antonucci 1993, Heckman et al. 1990, Suchkov et al. 1994, and papers in Courvoisier and Blecha 1994. This rather heterogeneous group of galaxies is interesting in its own right and the quantity and fate of the outflowing nuclear gas may have important implications for the galaxy - intergalactic medium connection. However, I wish here to concentrate on more ``normal'' (and more numerous) galaxies, specifically spirals, and the possibility of outflows from their disks. I will also focus on outflows which appear to be internally generated, in contrast to external processes like gas stripping due to the ram pressure of the IGM, as is occurring in the core of the Virgo Cluster (Chamaraux et al. 1980; van Gorkom et al. 1984; Cayatte et al. 1990).
Recent high resolution and high sensitivity observations of some spiral galaxy disks have revealed features which suggest that processes related to star formation can drive hot winds into galaxy halos. Among these are observations of our own Galaxy in which HI supershells and ``worms/chimneys'' have been identified extending above the disk (Heiles 1979, 1984). Related features have now been observed in other tracers (e.g. Koo et al. 1991; Waller and Boulanger 1993; Reach et al. 1993) and in some cases, associated active sites of star formation have also been identified (Reach et al. 1993). A number of theoretical advances have been made in terms of relating disk activity to these high latitude features (e.g. Norman and Ikeuchi 1989; Völk 1991; Heckman et al. 1993). However, a comprehensive model which can predict the structure and intensity of the disk-halo features over all observable tracers (e.g. radio continuum, H, HI, and X-ray) is still required. The situation is particularly challenging since both the ISM as well as star forming regions are ``clumpy'', and by amounts which (especially in external galaxies) are not yet well known.
In external galaxies of low inclination, e.g. M101, NGC 6946, M33, M31, IC 10, and the Small Magellanic Cloud (van der Hulst and Kamphuis 1991 and references therein), a disk-halo interaction is sometimes inferred by the presence of HI ``holes''. In M101, in particular, high velocity red and blue-shifted gas, the signature of an expanding shell, is clearly associated with one of the holes (Kamphuis et al. 1991). The implied kinetic energy of the shell is a few x 1053 erg, requiring the input of at least several hundred supernovae, a quantity which is consistent with the numbers of OB stars seen in nearby HII regions. HI holes as well as shells are also particularly obvious in some dwarf galaxies (e.g. Puche et al. 1992).
It is galaxies of high inclination, however, in which the disk-halo interaction can be studied most directly. Edge-on galaxies can be searched both for smooth, continuous halos which may be indicative of global processes, as well as any discrete extensions or filaments which may be associated with localized activity in the disk. In the former category is the ``diffuse ionized gas'' (DIG) or ``warm ionized medium'' (WIM), terms used interchangeably to describe the 1 kpc scale height smooth H emission seen in galaxies like NGC 891 (Dettmar 1992, 1993; Walterbos 1992) and known in our own Milky Way as the Reynolds layer (Reynolds 1991). Radio continuum halos of larger scale height have also been observed in a number of galaxies (e.g. Hummel et al. 1991; Carilli et al. 1992) and z extents of almost galaxy-sized proportion have been detected in NGC 4631 and NGC 5775 (Golla 1993; Hummel and Dettmar 1990; Duric et al. 1995; Fig. 4). ROSAT observations are now also revealing X-ray ``halos'' around some systems (e.g. Pietsch et al. 1994; Vogler et al. 1994). In the latter category, discrete z extensions in edge-on systems have been observed in a number of tracers, including H, neutral hydrogen, radio continuum, and X-rays. For example, kpc-scale HI features have been directly observed in NGC 3079, NGC 4565, NGC 891, NGC 4631 and NGC 5775 (see Irwin 1994). Very high kinetic energies (e.g. up to 1055 erg) are inferred for some features, based on rough estimates of expansion velocity and HI mass. This requires the correlated input of some 104 supernovae, if they are the source, leading to the energy problem alluded to above (Section 4.2). Fig. 4 displays HI arcs/loops in the edge-on galaxy, NGC 5775. The overlaid radio continuum contours (lower spatial resolution) reveal substructure in the halo out to at least 16 kpc from the plane (cf. NGC 4631, Golla 1993). Discrete radio continuum extensions closer to the plane can also be seen at spatial resolutions equivalent to the neutral hydrogen, but they are smoother and larger than the HI, likely due to cosmic ray diffusion. Consequently, the transition between discrete features and continuous ``halos'' (and therefore localized and global origins) can be rather fuzzy.
Figure 4. The edge-on galaxy, NGC 5775 and its interacting companion to the N-W, NGC 5774 (crosses mark the galaxy centers). The neutral hydrogen column density map is shown in greyscale (resolution = 13".5) and contours are of the 20 cm radio continuum emission from independent observations (resolution = 23".4). The peak HI column density is 1 x 1022 cm-2. The lowest contour level is 0.075 mJy beam-1 and the highest is 32 mJy beam-1. Note the ``bridges'' connecting the galaxies in both HI and continuum. HI arcs and extensions protrude away from the major axis of NGC 5775 and this galaxy is surrounded by a radio continuum halo extending to 16 kpc, containing considerable substructure (Duric et al.1995, and Irwin 1994).
In terms of the galaxy/IGM relationship, several questions might be raised. Firstly, does the outflowing gas leave the galaxy altogether (galactic winds), or is it confined to the halo region, eventually returning again to the disk (galactic fountains)? In other words, do galaxies ``evaporate'' or do galaxies ``percolate''? Secondly, what fraction of all galaxies experience such activity? And thirdly, could this activity be a significant driver of galaxy evolution? There are, as yet, no clear answers to these questions. Theoretically, some models predict that gas should indeed be leaving the galaxy provided certain conditions are met, e.g. if magnetic field lines are ``open'' (e.g. Völk 1991), or if the supernova clustering is high and ISM density is low (Norman and Ikeuchi 1989). Observationally, there is little evidence either way. Bregman and Pildis 1994 found that the hot, X-ray emitting gas around NGC 891 appears to be bound to the galaxy. On the other hand, the radio continuum morphology in galaxies like NGC 5775 (Fig.4), in which structure suggestive of dynamical processes far from the plane is observed, would suggest that at least some relativistic gas is flowing into the IGM. As for the second question, a recent survey by Sorathia and Irwin (1995) of 16 radio-selected edge-on galaxies suggests that almost every galaxy shows some radio emission extending beyond the optical disk. If significant numbers of star forming galaxies are evaporating at rates of order 1 M yr-1 as suggested for galaxies like the Milky Way (Völk 1991), or higher for starburst galaxies, then substantial depletion of the galaxy's ISM and enrichment of the IGM may occur over the lifetime of the galaxy. (Cooling flows, for example, may be supplied by such processes.) Thus galaxy outflows should have significant implications for the evolution of the galaxy itself, as well as its environment.
4.4 Magnetic Fields
Another way in which a galaxy might be considered ``open'', is if its internal magnetic field is connected to the field in the intergalactic medium, much as the sun's field was connected to the primordial interplanetary medium. It is now well established that disk galaxies as well as gas-rich dwarf galaxies contain magnetic fields which are of the order of a few microGauss (Sofue et al. 1986; Wielebinski 1990; Kronberg 1994). In contrast, intergalactic magnetic fields have been more elusive. Only a few clusters of galaxies display halos of synchrotron emission from which the presence of a magnetic field can be inferred, for example, Abell 2256 (Bridle et al. 1979), Abell 2319 (Harris & Miley 1978) and the Coma Cluster (Kim et al. 1990). Both the presence as well as the strength of a magnetic field in clusters can also be determined from observations of the Faraday rotation of polarized background sources through a cluster's hot X-ray emitting IGM. From a sample of 53 Abell clusters studied in this way by Kim et al. 1991, it was found that widespread fields of order a microGauss were typical, even in clusters in which a synchrotron halo had not been detected. In localized regions in clusters and in more unusual environments, like cooling flows or extragalactic radio lobes, the magnetic field can be higher still.
The origin of these fields, whether by amplification of seed fields, dynamo mechanisms, or turbulence, is currently the subject of some debate. However, the fact that galaxy and IGM magnetic field strengths are, in fact, comparable suggests a possible galaxy/IGM connection through these fields. Further evidence comes from observations of perpendicular magnetic field lines in galaxy halos (NGC 4631, Golla 1993; M82, Reuter et al. 1994), though few halo fields have yet been mapped in this way. Since the perpendicular fields tend to be associated with active star forming galaxies (and known winds as in the case of M82), this also argues in favour of gas outflows being associated with open field lines as discussed earlier. Indeed, wind and fountain activity provide a natural means by which magnetic fields, themselves, may outflow into the IGM, possibly being the source of the IGM fields to begin with (but see Tribble (1994) for arguments in favour of an origin from radio galaxies). For an excellent review of extragalactic magnetic fields, see Kronberg 1994.