Dufour et al. (1979) first established that the dark band crossing the elliptical galaxy is in fact the image of a highly inclined disk component consisting of a metal-rich population of stars, nebulae and dust clouds (Fig. 1). Metallicities are close to those in the Solar Neighbourhood (Dufour et al. 1979; Phillips 1981; Eckart et al. 1990a; Viegas & Prieto 1992). The disk is at position angle 122° ± 4° and star formation is rampant. The present burst of star formation apparently started 50 million years ago and created at least a hundred HII regions embedded in the disk (Dufour et al. 1979; Hodge & Kennicutt 1983). Remarkable concentrations of luminous and very blue stars can be seen at the northwestern and southeastern edges of the dark band; they must represent very large OB associations (Fig. 1). Recent HST observations allowed Schreier et al. (1996) to also find a large number of point-like sources embedded in the dust band with colours likewise suggestive of OB associations (see also Alonso & Minniti 1997). However, since the review by Ebneter & Balick (1983), little quantitative progress has been made on the issue of star formation. This may occur at rates ten times higher than in the Milky Way (Telesco 1978) although disk average UV energy densities close to Solar Neighbourhood values (Eckart et al. 1990a) suggest more moderate rates.
As NGC 5128 is no more distant than e.g. M 82, its brightest HII regions and supernova remnants should be detectable at centimetre radio wavelengths. Most of the published high-resolution radio maps lack the dynamic range to reach the low flux-density levels expected for HII regions or SNR's in the disk. The 1425 MHz map published by Condon et al. (1996) does, however, show an extension with peaks of 125 mJy in an 18" (295 pc) beam coinciding with the eastern half of the dark band, while the 43 GHz map by Tateyama & Strauss (1992) likewise seems to show weak emission from the eastern dark band. Further high-resolution observations of the disk at centimetre wavelengths are desirable as they provide one of the few means studying the star formation history of the disk.
As defined by its OB stars and HII regions, the disk extends out to a radius of 4 kpc. Molecular line and infrared continuum emission, further discussed in Sects. 4.2 and 4.3, is concentrated within 40% of this radius (Joy et al. 1988; Eckart et al. 1990a; Quillen et al. 1992). HI emission, however, extends much farther out, to radii of 7 kpc (cf. van Gorkom et al. 1990; Schiminovich et al. 1994). The outer parts of the disk, as traced by the dark band and the HI emission, show a pronounced warp to position angle 90° (cf. Fig. 7). The disk is in rapid rotation (Graham 1979; van Gorkom et al. 1990; Quillen et al. 1992). The tilted-ring modelling by Nicholson et al. (1992) shows that in spite of appearances, the distribution of dust in NGC 5128 is that of a warped thin disk of about 200 pc thickness (Sect. 6.1) along the minor axis. Deep images of NGC 5128 show the disk to be well inside the elliptical galaxy. Its inclination is a function of radius, but remains generally high with respect to the plane of the sky. The HII regions are distributed throughout the warped disk and embedded in diffuse ionized gas. Nicholson et al. (1992) showed that their warped disk model also quite naturally explains the various CaII and NaI velocity components seen in absorption against supernova 1986g by d'Odorico et al. (1989). In addition to the seven components associated with NGC 5128, these observationsy also showed three components with Galactic foreground gas, and two intermediate velocity components of unknown origin.
The inner part of NGC 5128 is associated with diffuse X-ray emission in the form of ridges along the dark band edges but also in more isotropically distributed form (Feigelson et al. 1981; Turner et al. 1997). Although the origin of this diffuse emission is not established unequivocally, among the most reasonable explanations for its existence are gas ejected from late-type stars dynamically heated to the required temperatures, X-ray binaries associated with the young stellar population, stellar winds in HII regions or combinations thereof (Feigelson et al. 1981; Turner et al. 1997). In any case, NGC 5128 is underluminous in diffuse X-ray emission as compared to other early-type galaxies (Döbereiner et al. 1996).
4.2. Atomic and molecular gas
The HI observations by van Gorkom et al. (1990) and Schiminovich et al. (1994) show the atomic hydrogen to follow the dust lane, including the warp (Fig. 7). It could, however, not be traced over the central 2.5 kpc because of strong absorption against the centre. Van Gorkom et al. (1990) found a total HI amount of about 3.3 × 108 M, but cautioned that they might have missed a significant amount because of limited sensitivity; nor does this estimate include the HI in the shells found by Schiminovich et al. 1994 (Fig. 7). Indeed, Richter, Sackett & Sparke (1994) find within the 21' (21 kpc) beam of the Green Bank 140 ft telescope a higher mass of 8.3 ± 2.5 × 108 M, still uncertain because of the strong central absorption (Sect. 7).
In the central part of the disk, molecular line emission from CO and its isotopes is found out to radii of about 2 kpc, but most of it is concentrated within a radius of 1 kpc (Phillips et al. 1987; Eckart et al. 1990a; Quillen et al. 1992; Rydbeck et al. 1993). Within R = 1 kpc, the area filling factor of the disk is of the order of 3-12%, its thickness is less than 35 pc and the velocity dispersion is about 10 km s-1 (Quillen et al. 1992). The J = 2-1/J=1-0 temperature ratios of about 0.9 for both 12CO and 13CO as well as the isotopic emission ratios 12CO / 13CO = 11 and 12CO / C18O = 75 are comparable to those of Milky Way giant molecular cloud complexes (Wild, Eckart & Wiklind 1997). Modelling the CO observations as tracer for the much more abundant H2 molecule, Eckart et al. (1990a) and Wild et al. (1997) estimate molecular hydrogen temperatures Tk = 10-15 K and densities of a few times 10 4 cm-3. Emission from other molecular species has also been detected in the disk (Whiteoak, Gardner & Höglund 1980; Seaquist & Bell 1988; d'Odorico et al. 1989; Israel 1992; Paglione, Jackson & Ishizuki 1997).
Total molecular hydrogen masses are probably about 4 × 108 M, but may be a factor of two higher depending on the CO-to-H2 conversion factor favoured. The vibrationally excited warm H2 (Tk 1000 K) detected by Israel et al. (1990) represents only a minute fraction of all molecular hydrogen and is associated with the circumnuclear disk (Sect. 5.2). The total gaseous mass of the disk, including helium, is thus of the order of 1.3 × 109 M, only about 2% of the dynamical mass contained in the elliptical component within the radius of the disk (R = 7 kpc). However, because of the pronounced concentration of interstellar gas at smaller radii, that fraction increases to about 8% at R = 2 kpc.
4.3. Dust emission
At far-infrared wavelengths, the disk of NGC 5128 stands out by its emission from warm dust. Dust temperatures are 30-40 K depending on the assumed dust emissivity Q100 or respectively (Joy et al. 1988; Marston & Dickens 1988; Eckart et al. 1990a). The overall distribution of far-infrared emission is very similar to that of the carbon monoxide. The present far-infrared information still leaves considerable room for improvement. The KAO scans presented by Joy et al. (1988) do not fully sample the galaxy, but show that 10% of the total far-infrared luminosity arises in central source which may be identified with the circumnuclear disk (see Sect. 5.2). IRAS survey observations cover the whole galaxy, but the resolution is limited. Several of the published fluxes refer to poorly calibrated data or underestimate the total flux from the extended galaxy. Best fluxes are probably the colour-corrected values S12 = 26.4 Jy, S25 = 25.7 Jy, S60 = 236 Jy and S100 = 520 Jy given by Rice et al. (1988). Use of IRAS non-survey data (DSD maps: Marston & Dickens 1988; CPC-maps: Eckart et al. 1990a; Marston 1992) provided some improvement in resolution, but at the cost of photometric accuracy. Because of unsolveable calibration problems (cf. van Driel et al. 1993), the image-sharpened CPC maps discussed by Eckart et al. (1990a) and by Marston (1992) must be considered as unreliable. The most recent image-sharpened IRAS maps (resolution 1-2') incorporating all survey-instrument data are shown in Fig. 9; they still lack the resolution to bring forth the full detail of the disk.
Good mid-infrared imaging of the dust disk itself is still lacking. The mean 12µm surface brightness of the disk as derived from IRAS data is about 25 MJy sr-1. In addition to the nucleus (Sect. 5.3), Telesco (1978) detected 10µm emission weaker by a factor of about five from a number of disk HII regions. Clearly, much more remains to be done.
Figure 9. IRAS image-sharpened maps of the NGC 5128 dusty disk at 12µm (Band 1) and 60µm (Band 3), showing the warped outer edges. Courtesy D. Kester, University of Groningen.
As much as 50% of the 100µm emission may be due to "cirrus" (Marston & Dickens 1988; Eckart et al. 1990a). The various far-infrared data indicate the presence of small amounts of dust outside the disk in the main elliptical galaxy as well as large amounts in the disk out to a radius of 3 kpc (Eckart et al. 1990a; Marston 1992). The total dust mass can be estimated from IRAS photometry as Md = 1-2 × 106 M (cf. Hunter et al. 1989) with a luminosity LFIR = 2 × 1010 L (Joy et al. 1988; Eckart et al. 1990a). With considerable uncertainty, the total gas-to-dust ratio within a radius of 2 kpc is Mgas / Mdust = 450. This is an upper limit if significant amounts of cold dust are present, imperfectly sampled by IRAS. Marston & Dickens (1988) explicitly modelled the IRAS emission in terms of large, cool grains and warm, small grains, and arrived at a (distance-corrected) dust mass an order of magnitude higher, implying a gas-to-dust ratio of 45, which seems rather low.
4.4. Polarization and extinction
At optical and infrared wavelengths, the disk is significantly polarized up to 6% parallel to the dust band (Elvius & Hall 1964; Hough et al. 1987; Scarrott et al. 1996). The observations by the latter show higher levels of polarization at the dust band extremities, with directions perpendicular to the dust band. Hough et al. (1987) concluded that the dust grains sampled are about 20% smaller than those in the Milky Way, implying an extinction law differing from that in the Solar Neighbourhood, with a total to selective extinction ratio of 2.4. The HST R- and I-band imaging polarimetry presented by Schreier et al. (1996) shows the polarization to reach a peak at a knot close to the nucleus, shining by reflected light. Like Hough et al. (1987), they assumed scattering to be negligible elsewhere, but that is inconsistent with the apparent importance of scattering in the central region and is unlikely in view of the conclusions reached by Packham et al. (1996 - see Section 5.3). The observations by Scarrott et al. (1996) can only be explained by assuming simultaneous operation of both dichroic extinction and scattering, with the latter dominating at the dust band extremities. This suggests that the very blue colours observed by van den Bergh (1976) just at the northern dark band edge are caused by the light of intrinsically blue objects (cf. Alonso & Minniti 1997 and references therein) enhanced by scattered light and suffering relatively little extinction. This is supported by the optical continuum observations of the central region by Storchi-Bergmann et al. (1997) who find that the major contribution comes from a metal-rich old bulge, but that there are also significant contributions from young stars and from scattered light especially at the dark band edges.
The optical colours suggests significant extinction in the dust band itself (E(B - V) 0.5 mag - van den Bergh 1976; Dufour et al. 1979). Indeed, the near-infrared estimates by Harding et al. (1981) yield AV = 3-6 mag, while HST observations indicate V-band extinctions ranging from 0.5 to 7 mag in the dark band (Schreier et al. 1996) and even reaching values in excess of 30 mag (AK 3 mag) just south of the optically invisible nucleus (Alonso & Minniti 1997). The R-band and I-band images by Schreier et al. (1996) clearly show how the dust band is seen nearly edge-on, but slightly tilted with the near side south of the centre so that we are looking from above. The glow of the nuclear region (but not the nucleus itself) on the north-side of the dust disk is strikingly apparent even through the high extinction it suffers.