Hot ionized gas is an essential component of the ISM. However, it is difficult to study the distribution and physical properties of hot gas in the Galaxy because of confusion and obscuration along the Galactic plane. The Large Magellanic Cloud (LMC), owing to its close proximity, nearly face-on view, and small extinction, presents an ideal site to study hot gas and its role in the interstellar matrix.
Soft X-ray images of the LMC show bright diffuse emission, indicating the existence of hot (106-107 K) ionized gas. Clues to the origin of the hot gas in the LMC are provided by comparisons between X-ray and Halpha images. Halpha images of the LMC show a variety of interstellar shells, ranging from small supernova remnants a few pc in diameter to large supergiant shells more than 1000 pc across. Some of the diffuse X-ray emission is associated with shell structures, but there is also large-scale diffuse emission that does not appear to be confined by any cooler interstellar structures. The origins of hot gas in these different types of diffuse X-ray emission regions are individually discussed.
A hot (105 K) gas halo of the LMC was suggested in the early 1980's. As most of the previously reported interstellar C IV absorption occurs in large shell structures where supernovae and fast stellar winds are rampant, the evidence for a global hot gas halo around the LMC has been weak. Recent HST GHRS observations of five carefully chosen probe stars show clear detections of the interstellar C IV, providing unambiguous evidence for a hot gas halo of the LMC. Three sightlines have C IV velocities blue shifted from the Halpha by 25-60 km s-1, indicating that the hot gas halo may be expanding away from the disk gas. The hot gas halo of the LMC is most likely patchy, as implied by a stringent nondetection by the IUE along one sightline.
Spitzer (1956) proposed a hot gas halo around the Galaxy to confine the H I clouds observed at high Galactic latitudes. Following his suggestion, interstellar N V and C IV absorption lines in the UV have been used to search and confirm the existence of 105 K coronal gas in the Galactic halo (Savage & de Boer 1979; Sembach & Savage 1992). The existence of 106 K coronal gas in the halo of the Galaxy has also been demonstrated by the X-ray shadow of the Draco H I cloud at a high Galactic latitude (Snowden et al. 1991). In the meantime, however, it has been observed that the cooler Galactic disk gas actually has a low-density extension to more than 1 kpc above the disk plane (Lockman 1984; Reynolds 1989). The gaseous halo of the Galaxy is thus not as simple as Spitzer's model envisioned. What is the relationship among the coronal gas, the cooler ionized gas, and the neutral gas components?
The multi-phase ISM was initially introduced by Field, Goldsmith, & Habing (1969). Assuming that the ISM was heated by low-energy cosmic rays and that a pressure balance existed in the ISM, they found two thermally stable gas phases, one at 104 K and one at <300 K. The Copernicus satellite detected interstellar O VI absorption, indicating a widespread coronal component of the ISM (Jenkins & Meloy 1974; Jenkins 1978a,b). The heating of the ISM by supernova remnants (SNRs) was also gaining recognition (Cox & Smith 1974). Incorporating these concepts, McKee & Ostriker (1977) developed a self-consistent, three-phase model of the ISM. This model successfully explains many observations of the ISM and predicts that the hot 106 K gas component, the third phase, is pervasive in the Galaxy. However, McKee & Ostriker's three-phase model has been criticized by Cox (1995a), who challenges the three tenets of this model: high supernova rate, pressure balance in the ISM, and cloud evaporation in a hot medium. Debates between McKee and Cox on the three-phase model raged at Elba in 1994, at Paris in 1995, and at Madison in 1996 (Cox 1995b, 1996a,b; McKee 1995, 1996a,b). Is the hot ionized medium really pervasive in the Galaxy? Is there convincing evidence for a pervasive hot medium? Do evaporating clouds exist?
These questions on physical structures of the Galactic ISM are difficult to answer because of our disadvantageous ``edge-on'' view of the Galactic disk and the uncertainty in distances. The kinematic distances of the 104 K ionized gas and the H I gas can be used to determine their distribution in the Galaxy, but large errors may arise if the gas motion deviates from the Galactic rotation. The distance to the 106 K gas is particularly uncertain, as X-ray shadows of neutral clouds provide the only means to determine whether the X-ray emitting gas is on the near side or far side of a neutral cloud. Without knowing distances, it is impossible to determine the 3-D distribution of interstellar gas components in the Galaxy, hence the above posed questions cannot be answered. It is thus necessary to use a different galaxy to study the physical structure of the ISM.
Optical emission-line images of ionized gas in the LMC reveal an assortment of shell structures. Since the stars encompassed within the shells can be resolved and spectroscopically classified, the nature of the shells and the relationship between the shells and the underlying stars can be easily assessed. It is quite clear that the smallest shells, a few pc in diameter, are ``bubbles'' blown by single massive stars or are simply SNRs. Ionized interstellar gas shells of diameters a few tens to ∼200 pc may be formed by different mechanisms. The faintest of such shells could be ``bubbles'' blown by one or a small number of massive stars or evolved SNRs in a very low-density medium, but generally such shells are ``superbubbles'' blown by OB associations in a normal medium. The largest interstellar shell structures are ``supergiant shells'' with diameters reaching or exceeding ∼1000 pc. Supergiant shells usually contain stellar populations resultant from multiple episodes of star formation distributed over an extended period of time and surface area. The relative sizes of bubbles, SNRs, superbubbles, and supergiant shells are illustrated in Fig. 1, an Halpha image of the supergiant shell LMC 4. Bubbles, SNRs, and superbubbles along the periphery of LMC 4 are marked. It is conceivable that these varied interstellar shells may produce different amounts of hot gas and have different physical structures.
The 104 K ionized gas is most easily observed at optical wavelengths in emission lines. The Curtis Schmidt Telescope at Cerro Tololo Inter-American Observatory (CTIO) has been used to make an emission-line survey of the Magellanic Clouds; CCD images are taken with filters centered on Halpha, [O III], [S II], red continuum, and green continuum (Smith et al. 1998). For selected regions in the LMC, we have used the CCD camera on the CTIO 0.9 m telescope to take higher resolution emission-line images. For a small number of LMC H II regions, we have obtained Hubble Space Telescope (HST) WFPC2 images. We have also used the echelle spectrograph on the CTIO 4 m telescope to study the kinematics of the ionized 104 K gas.
The 105 K ionized gas is commonly observed with the interstellar absorption lines of C IV, Si IV, and N V in the UV. We have used the high-dispersion spectrograph on the International Ultraviolet Explorer (IUE), and the Goddard High Resolution Spectrograph (GHRS) on the HST for these observations. The recently commenced Space Telescope Imaging Spectrograph (STIS) can be used to continue such observations. The to-be-launched Far Ultraviolet Spectroscopy Explorer (FUSE) provides an opportunity to observe the O VI lines, which indicate unambiguously the existence of shock-heated gas.
The 106 K ionized gas is best observed in soft X-rays. The ROSAT X-ray satellite has provided two types of detectors, the Position Sensitive Proportional Counter (PSPC) and the High Resolution Imager (HRI) in the energy range of ∼0.1-2.4 keV. The PSPC has an on-axis angular resolution of ∼30" and a spectral resolution of 45% at 1 keV; the HRI has an on-axis angular resolution of ∼5", but essentially no spectral capability. Both PSPC and HRI have been used to survey hot gas in the LMC. The ASCA X-ray satellite provides the Solid-state Imaging Spectrometer (SIS) and the Gas Imaging Spectrometer (GIS). The SIS is useful for obtaining X-ray spectra at a high spectral resolution, 2% at 5.9 keV. Several diffuse X-ray sources in the LMC have ASCA observations available. In the near future, the Advanced X-ray Astrophysics Facility (AXAF) will provide the AXAF CCD Imaging Spectrometer (ACIS) and High Resolution Camera (HRC) for X-ray imaging at 1" resolution and a 2% spectral resolution.
The X-ray mosaic of the LMC in Fig. 2 shows a variety of discrete and diffuse sources. Some of the bright discrete sources are point sources that have been identified to be X-ray binaries in the LMC, foreground G, K, and dMe stars in the Galaxy, or background AGNs (Cowley et al. 1984, 1997; Schmidtke et al. 1994), while some are actually SNRs and superbubbles in the LMC (Mathewson et al. 1983, 1984, 1985; Chu & Mac Low 1990). In addition to the discrete sources, there are diffuse sources with spatial extent of 100 pc to greater than 1000 pc. Some of the large diffuse sources are within ionized shell structures or active star formation regions, but others do not seem to be associated with any interstellar structures or star formation activity. These sources will be discussed in more detail in the next section.
Snowden & Petre's (1994) PSPC mosaic of the LMC is sensitive to diffuse X-ray emission, but its angular resolution is not ideal. It is difficult to distinguish between a point source and a small diffuse source, especially at large off-axis angles. The poor angular resolution hampers the optical identification of X-ray sources. Therefore, a ROSAT HRI survey of the LMC has been undertaken in ROSAT AO5 - AO8 (Chu & Snowden 1998). The goal of this HRI survey is to map a 2.5° × 2.5° area centered on the 30 Dor complex and a 1° × 2° area along the LMC bar, with 20' separation between adjacent pointings and at least 20 ks exposure at each pointing. The typical angular resolution is 5"; the worst-case resolution at 14' off-axis is 15".
The high quality of the survey is illustrated in Fig. 4. The upper panel shows an HRI mosaic of a 140' × 110' area centered on 30 Dor and the supergiant shells LMC 2 and LMC 3. For comparison, the lower panel shows an Halpha image of the same field. Identifications of conspicuous sources are marked. The HRI survey has detected not only bright discrete sources but also faint diffuse emission, most notably within the supergiant shell LMC 2. This HRI survey will greatly aid in the incision of point sources and the analysis of diffuse X-ray emission from the LMC.
Diffuse X-ray emission is indicative of the existence of hot gas. However, the spectral and morphological properties of the X-ray emission alone do not provide sufficient information to determine the production mechanism of the hot gas. Furthermore, the amount of hot gas cannot be derived form the X-ray observations alone, because the volume of the X-ray emitting gas is unknown. It is necessary to resort to the ambient warm, 104 K, gas for further constraints. Morphology and kinematics of the associated 104 K gas are particularly useful. If the excitation mechanism is known, the X-ray-emitting volume can then be better assessed. The X-ray and Halpha images of the LMC (Figs. 2 and 3) show hot ionized gas associated with a range of interstellar structures. Some of these structures are identified in Fig. 4. Below I discuss hot gas in SNRs, superbubbles, supergiant shells, the 30 Dor giant H II region, and in the field.
Collision between two SNRs has been considered rare because it is statistically unlikely to have two successive supernova events within a small distance, <50 pc, and within a short time interval, < a few ×104 yr. However, the LMC remnant DEM L 316 has been shown to contain a pair of colliding SNRs. The Halpha image of DEM L 316 shows two connecting shells, while the X-ray image shows two detached volumes of hot gas (Fig. 5). The two SNR shells of DEM L 316 have similar pressures in the hot gas, and both are an order of magnitude higher than that of the warm 104 K gas shells (Williams et al. 1997).
The Honeycomb Nebula near the supernova SN 1987A (Wang 1992) has been identified as a SNR (Chu et al. 1995) based on its bright X-ray emission, nonthermal radio emission, enhanced [S II]/Halpha ratio, and violent internal motion. The Honeycomb Nebula's irregular morphology is unlike any other known SNR. This morphology can be explained only if the supernova explosion occurred in a hot medium and the Honeycomb Nebula is the first dense interstellar material the SNR shock encounters. The Honeycomb SNR thus answers Cox's (1996) question 5.9 ``Are there examples of SNRs outside of superbubbles which appear to have exploded in empty regions?''
The LMC SNR N63A has been puzzling because its X-ray and radio sizes are much larger than its optical size (Mathewson et al. 1983). Optical long-slit echelle observations in the Halpha line did not detect any post-shock interstellar material outside the optical remnant (Chu 1997), which had been identified based on its high [S II]/Halpha ratio (Mathewson et al. 1983). This puzzle is finally solved by HST WFPC2 images in Halpha and [S II] lines (Chu et al. 1998). Figure 6 shows the WFPC2 image in the Halpha line overlaid by X-ray contours. The X-ray boundary delineates the location of the shock front. The previously identified optical remnant is composed of a shocked dense cloud, about 5 pc (20") across. The supernova must have exploded inside a wind-blown bubble in a clumpy ISM. The SNR shock propagates rapidly through the low-density bubble interior, leaving behind the dense cloudlets that were engulfed in the bubble interior. The shocked cloudlets evaporate and transfer mass into the hot SNR interior, enhancing the X-ray emission. As shown in Fig. 6, the X-ray peak in N63A is coincident with two evaporating cloudlets. Thus, N63A answers Cox's (1996) questions 5.1 ``Do evaporating clouds exist?'' and 5.2 ``Do SNRs acquire significant amounts of mass from clouds via thermal evaporation?''
It has been found that the X-ray-brightest superbubbles have X-ray luminosities more than an order of magnitude higher than expected, and that these superbubbles must contain interior SNRs shocking the inner shell walls (Chu & Mac Low 1990). A detailed analysis has been carried out for N44, the brightest among all LMC superbubbles. Shown in Fig. 7 are Halpha and X-ray images of N44. The N44 complex contains three large shells that are detected in both optical and X-ray wavelengths: Shell 1, the bright superbubble near the center; Shell 2, the faint superbubble to the west; and Shell 3, a SNR at ∼6' NE of Shell 1. The X-ray emission from Shell 1 extends beyond the southern shell boundary, suggesting a break-out structure (Chu et al. 1993). This break-out has been confirmed by the kinematic properties of the 104 K gas and the temperature variation in the 106 K gas from the shell interior to the outflow region (Magnier et al. 1996).
The X-ray-faintest superbubbles were not detected by the ROSAT PSPC even in long (10-15 ks) exposures. The 3σ upper limits of the X-ray emission from four X-ray-dim superbubbles, DEM31, DEM105, DEM106, and DEM137, are comparable to those expected in pressure-driven bubble models (Chu et al. 1995). Deeper observations are needed to determine whether these superbubbles are actually fainter than what models predict. Many LMC superbubbles have X-ray emission only slightly higher than expected, such as N11 (Mac Low et al. 1998). The variations of X-ray surface brightness in superbubbles are mostly caused by SNR activity rather than large-scale interstellar density variations.
Two supergiant shells, LMC 2 and LMC 3, can be seen in Fig. 4. Diffuse X-ray emission is detected in both LMC 2 and LMC 3. The X-ray surface brightness is not uniform. The brightest X-ray emission from LMC 2 is projected within the shell interior, while the brightest X-ray emission from LMC 3 occurs along the eastern rim. To determine the origin of hot gas in LMC 2, the kinematics of the 104 K gas and the optical and X-ray morphologies have been analyzed in detail by Points et al. (1998). They find that some low-surface-brightness regions are caused by X-ray shadows of neutral clouds on the front side, and that some high-surface-brightness regions are caused by additional heating. The northeastern corner of LMC 2 is apparently heated locally by SNRs, while the southwestern corner is heated by an outflow from the star-forming region N158/N160/N159. The X-ray surface brightness of a supergiant shell should be interpreted with caution; one should not assume a unity filling factor for the entire supergiant shell.
The hot gas corona around the LMC was first reported by de Boer & Savage (1980), based on the interstellar C IV and Si IV absorption in the IUE spectra of early-type stars in the LMC. It was realized later that the early-type stars themselves could photoionize C and Si to C+3 and Si+3, hence the interstellar C IV and Si IV absorption might take place locally in the surrounding H II region, instead of remotely in the LMC's halo. To eliminate this possibility, de Boer & Nash (1982) used the detection of N V absorption and the blue extension of the C IV and Si IV lines in the spectrum of HD 36402 to argue for a corona around the LMC, as N+4 cannot be produced by photoionization. Savage (1984) used the detection of C IV absorption at LMC velocities in the spectrum of HDE 269599 to further argue for a corona around the LMC, as HDE 269599 is not inside any H II region and it is a B0.5 supergiant too cool to photoionize C to C+3. However, it is pointed out that HD 36402 is inside the X-ray-bright superbubble N51D containing SNRs that might be responsible for the N V C IV and Si IV absorption, and that HDE 269599 is inside the supergiant shell LMC 3, which is also likely to contain SNR activity (Chu et al. 1994). Therefore, neither HD 36402 nor HDE 269599 provides unambiguous evidence for hot halo gas around the LMC.
To convincingly demonstrate the existence of a hot gas halo around the LMC, the probe stars have to meet two criteria.
The search for a hot gas halo around the LMC was carried over into the HST era (Wakker et al. 1998). The GHRS has been used to observe five probe stars selected on the basis of the above-mentioned criteria. Interstellar C IV absorption at LMC velocities is detected in all five cases, although two stars have stellar features confusing the interstellar components. In the three cases that the interstellar C IV lines can be measured well, the derived C+3 column densities are in the range (8-12)·1013 cm-2 and their velocities are blue-shifted by -25 to -60 km s-1 from those of the Halpha line along the same lines of sight. These C+3 column densities and the 3σ upper limit of <2·1013 cm-2 for Sk -67 05, measured from IUE spectra, imply that the hot gas halo of the LMC must be patchy. The velocity offsets between the Halpha emission and the C IV absorption indicate that the 105 K ionized gas and the 104 K ionized gas are not co-spatial. If the 104 K gas represents the warm disk gas, then the 105 K gas appears to be expanding away from the warm disk gas.
The HST GHRS observations show the first convincing evidence of a patchy hot gas halo of the LMC (Wakker et al. 1998). The LMC's hot gas halo may be studied further using new UV instruments. The HST STIS, having a larger wavelength coverage than the GHRS, is more efficient in observing multiple spectral lines. More lines of sight should be observed in order to determine the spatial distribution, kinematic properties, and physical conditions of the hot halo gas. It would be most interesting to determine the relationship between the large-scale diffuse X-ray emission and the interstellar absorption properties. This is the only way to establish the location of the X-ray-emitting gas relative to the disk plane. The FUSE can be used to observe the O VI absorption and probe the hot gas at a higher temperature, 3·105 K. It is expected that HST STIS and FUSE observations would produce exciting new insights into the physical structure of the LMC's hot gas halo.