2.4. The gaseous ISM
2.4.1. The atomic component
Most ISO studies of the ISM in normal galaxies focussed on observations with LWS of the [C II] 158 µm fine structure transition, which is the main channel for cooling of the diffuse ISM. Prior to ISO, knowledge of [C II] emission from external galaxies was limited to measurements of a total of 20 strongly star-forming galaxies sufficiently bright to be accessible to the Kuiper Airborne Observatory (Crawford et al. 1985; Stacey et al. 1991; Madden et al. 1993). The main conclusion of these early investigations was that the [C II] line emission arose from PDRs, where the UV light from young stars impinged on molecular clouds in star-forming regions. The only exception to this was Madden et al's resolved study of NGC 6946, who showed that the line could also be emitted from the diffuse disk.
The principal contributions of ISO to the knowledge of the neutral gaseous ISM have been twofold: Firstly, knowledge of the [C II] line emission has been extended to quiescent spiral galaxies and dwarf galaxies. These targets are difficult to access from beneath the atmosphere on account of their low luminosities and brightnesses. Secondly, ISO has tripled the number of higher luminosity systems with [C II] measurements, and has also observed a whole range of other fine structure lines in these systems. These lines have been used to probe in detail the identity of the emitting regions and their physical parameters.
Fundamental to the studies of the quiescent spiral galaxies were the measurements by Leech et al. (1999), who detected 14 out of a sample of 19 Virgo cluster spirals (a subset of the IVCDS sample - see Sect. 2.2) in the [C II] line using LWS. These galaxies are the faintest normal galaxies ever measured in the [C II] line, and can be regarded as quiescent in star-forming activity (H EW is less than 10Å for 8 galaxies). In a series of papers Pierini & collaborators (Pierini et al. 1999; Pierini et al. 2001, & Pierini et al. 2003a) used these LWS data, in combination with the existing KAO measurements of more actively star-forming galaxies, to investigate the physical origin of the line emission in normal galaxies as a function of star-formation activity. The main result was that the [C II] emission of quiescent systems predominantly arises from the diffuse ISM (mainly the cold neutral medium; CNM), as evidenced from the fact that these systems occupy a space in the L[C II]/LCO versus L[C II] / LFIR diagram devoid of individual star-formation regions or giant molecular clouds (Pierini et al. 1999). An origin of the [C II] emission from the CNM had earlier been proposed by Lord et al. (1996) as an alternative to an origin in localised PDRs, on the basis of a full LWS grating spectrum of the spiral galaxy NGC 5713.
The luminosity ratio of [C II] line emission to FIR dust emission, L[C II] / LFIR, was found to be in the range 0.1 to 0.8% for the quiescent spirals. This is consistent with the basic physical interpretation, originally established for the more active galaxies, of a balance between gas cooling (mainly through the [C II] line) and gas heating through photoelectric heating from grains 2. However, the range of a factor of almost an order of magnitude in the observed values of L[C II] / LFIR also suggests that the relation between the [C II] line emission and the SFR may be quite complex. This impression is reinforced by the large scatter and non-linearities found by Boselli et al. (2002) in the relations between the [C II] luminosity and SFRs derived from H measurements for a sample principally composed of the Virgo galaxies measured by Leech et al. 1999] with the LWS.
One consequence of the physical association of the [C II] emission with the diffuse dusty ISM in quiescent galaxies is that the observed L[C II] / LFIR ratio will be reduced due to an optically- (rather than UV-) heated component of the FIR dust emission. A model quantifying this effect was developed by Pierini et al. 2003a in which L[C II] / LFIR depends on the fractional amount of the non-ionising UV light in the interstellar radiation field in normal galaxies. Overall, systematic variations in the L[C II] / LFIR ratio for star-forming galaxies can be summarised as follows: In progressing from low to high star-formation activities, L[C II] / LFIR first increases in systems in which the [C II] emission is mainly from the diffuse medium, due to a decrease in importance of optical photons in heating the diffuse dust. After reaching a maximum, L[C II] / LFIR then decreases as the star-formation activity is further increased to starburst levels, due to an increase in the fraction of [C II] emission arising from localised PDRs associated with star-forming regions, coupled with the quenching of the [C II] emission through the decreased efficiency of photoelectric heating of the gas in high radiation fields. Thus, the dominance of the [CII] emission from PDRs turned out to be the asymptotic limit for high SFR in gas-rich galaxies.
The LWS observations of Virgo spirals also showed that the linear relation between L[C II] and LCO previously established for starburst systems extends into the domain of the quiescent spirals, though with an increased scatter (see Fig. 14). The tight relation between L[C II] / LCO and the H equivalent width (EW) found by Pierini et al. (1999) indicated that this scatter may be induced by different strengths of the far-UV radiation field in galaxies. Alternatively, as discussed by Smith & Madden (1997) in their study of five Virgo spirals, the fluctuations in L[C II]/LCO from one galaxy to the next might reflect changes in LCO from PDRs caused by variations in metallicity (and dust abundance), as well as varying fractions of [C II] emission arising from PDRs and the cold neutral medium. The measurements of the [C II] / CO line ratios derived from ISO observations have also thrown new light on the much debated relation between the strength of the CO line emission and the mass of molecular hydrogen in galaxies. Bergvall et al. (2000) emphasise how the low metallicity, the intense radiation field and the low column density in the dwarf starburst galaxy Haro 11 can explain the extremely high observed [C II] / CO flux ratio, indicating that CO may be a poor indicator of the H2 mass in such systems. Similarly, the enhanced [C II] / CO ratios found in the Virgo spirals observed by Smith & Madden (1997) were attributed to the low metallicities in these galaxies, although, as also pointed out by these authors, an alternative explanation could be that radiation from diffuse HI may dominate the [C II] emission.
Figure 14. The relationship between the observed central [CII] line intensity, I[CII], and the central CO line intensity, ICO from Pierini et al. (2001). Asterisks and filled triangles denote starburst galaxies and gas-rich galaxies of the KAO sample, while open circles identify spiral galaxies of the ISO sample. The long- and short-dashed lines show the average ratios of the two observables obtained by Stacey et al. (1991) for the starburst galaxies and the gas-rich galaxies, respectively.
Mapping observations of the [C II] line using the LWS provide a means to directly probe the relative amounts of [C II] arising from localised PDRs associated with star-forming regions in the spiral arms and the diffuse disk, as well to investigate the relation between the [C II] emission and the neutral and molecular components of the gas. They have also allowed the nuclear emission to be separately studied. Contursi et al. (2002) mapped the two nearby late-type galaxies NGC 1313 and NGC 6946 in the [C II] line, finding that the diffuse HI disk contributes ~ 40% and ~ 30% of the integrated [C II] emission in NGC 6946 and NGC 1313, respectively. CO(1-0) and [C II] were also found to be well correlated in the spiral arms in NGC 6946, but less well so in NGC 1313. Stacey et al. (1999) mapped the barred spiral M 83 in a variety of fine structure lines in addition to the [C II] line - the [O I] 63 & 145 µm lines, the [N II] 122 µm line and the [O III] 88 µm line, and obtained a full grating scan of the nucleus. At the nucleus, the line ratios indicate a strong starburst headed by O9 stars. Substantial [N II] emission from low density HII regions was found in addition to the anticipated component from PDRs. A further resolved galaxy studied with the LWS is M 33, for which Higdon et al. (2003) made measurements of fine structure lines and dust continuum emission towards the nucleus and six giant HII regions. Overall, the picture presented by these investigations of resolved galaxies is broadly in accordance with the theory that the integrated [C II] emission from spirals galaxies is comprised of the sum of components from the diffuse disk and from localised PDRs and diffuse ionised gas associated with star-formation regions in the spiral arms.
Fundamental to the statistical investigations of gas in more active galaxies are the full grating spectra of 60 normal galaxies from the ISO key-project sample of Helou et al. (1996), which were obtained by Malhotra et al. (1997, 2001) with the LWS. One of the main results was the discovery of a smooth but drastic decline in L[C II] / LFIR as a function of increasing star-formation activity (as traced by the L60 / L100 IRAS colour ratio or LFIR / LB) for the most luminous third of the sample. This trend was accompanied by a increase in the ratio of the luminosity of the [O I] 63 µm line to that of the [C II] line. This is readily explained in PDR models in terms of a decreased efficiency of photoelectric heating of the gas as the intensity of the UV radiation field is increased, coupled with an increased importance of the [O I] cooling line as the gas temperature is increased. Both Malhotra et al. (2001) and Negishi et al. (2001) present a comprehensive analysis of the ratios of the most prominent lines ([C II] 158 µm; [O I] 63 & 145 µm; [N II] 122 µm; [O III] 52 & 88 µm and [N III] 57 µm) in terms of the densities of gas and radiation fields in PDRs, as well as of the fraction of emission that comes from the diffuse ionised medium (as traced by [N II] 122 µm). Malhotra et al (2001) found that the radiation intensities G0 increases with density n as G0 n, with being ~ 1.4. They interpret this result, together with the high PDR temperatures and pressures needed to fit the data, as being consistent with the hypothesis that most of the line and continuum luminosity of relatively active galaxies arises from the immediate proximity of the star-forming regions. By contrast, on the basis of a comparison of the [C II] line with the [N II] 122 µm line, Negishi et al. (2001) argue that a substantial amount (of order 50%) of the [C II] line emission might arise not from PDRs but instead from low density diffuse ionised gas. Negishi et al. (2001) also found a linear (rather than non-linear) relation between G0 and n, which would argue against a decreased photoelectric heating efficiency being responsible for the decline in L[C II] / LFIR with star-formation activity. Instead, Negishi et al. invoke either an increase in the collisional de-excitation of the line with increasing density, or a decrease in the diffuse ionised component with increasing star-formation activity as being responsible for this effect.
An explanation for the decrease in L[C II] / LFIR in terms of a decreased efficiency of photoelectric heating of the gas (Malhotra et al. 2001) or due to a collisional de-excitation of the [C II] line (Negishi et al. 2001) would predict that the L[C II] will not be a good probe of high redshift starburst galaxies. However, an alternative (or complementary) explanation for the decrease in L[C II] / LFIR with star-formation activity, proposed by Bergvall et al. (2000), is that the line becomes self-absorbed in local universe starbursts with high metallicity. This scenario would also explain why the metal poor luminous blue compact dwarf galaxy Haro 11 was found by Bergvall et al. to have a high ratio of L[C II] / LFIR, despite the intense radiation fields present.
2.4.2. The molecular component
The lowest energy transitions of the most abundant molecule in the Universe, H2, are located in the MIR, typically between 3 an 30 µm. These pure rotational lines potentially offer a new access to the molecular component of the ISM, to be compared with the less direct CO tracer. One caveat though is that the lines can be excited by different mechanisms, typically shocks or UV-pumping, which render their interpretation difficult. Furthermore, the emission is very rapidly dominated by that from the warmest fraction of the gas, and therefore this new tracer offers only a partial and complex access to the molecular phase.
Most of the extragalactic H2 detections were made with ISOSWS (de Graauw et al. 1996; Leech et al. 2003), since high spectral resolution is necessary to isolate the very thin molecular lines from the broader features that abound in the MIR regime (although a tentative detection in NGC 7714 made with ISOCAM is presented by O'Halloran et al. 2000). For sensitivity reasons, the galaxies that were observed in these lines tended to be starburst or active galaxies and are discussed elsewhere in this volume.
Nevertheless, the S(0) and S(1) transitions were observed in the central region and in the disk of NGC 6946 (Valentijn et al. 1996 and Valentijn & van der Werf 1999a) and all along the disk of NGC 891 Valentijn & van der Werf 1999b. In the central region of NGC 6946, the molecular component detected with ISOSWS is clearly the warm fraction (of order 5-10%) of the molecular gas present in the region and is raised to the observed temperature of 170K by the enhanced star-formation activity in the center of the galaxy. H2 is also detected further out (4kpc) in the disk, with a cooler temperature, though still relatively warm.
The observations of H2 in the disk of NGC 891 reach much larger distances, extending about 10kpc on both sides of the nucleus. Valentijn & van der Werf 1999b argue that the observed line ratios indicate the presence of both warm (150-230K) molecular gas, identical as that observed in NGC 6946, and cool (80-90K) molecular gas. This cooler gas needs to have a very large column density to explain the observed surface brightness and the masses derived by Valentijn & van der Werf 1999b are in fact so large that they would resolve the missing mass problem in the optical disk of NGC 891. This highly controversial conclusion however rests on a rather uncertain basis dealing with the actual spatial distribution of the H2-emitting clouds within the ISOSWS beam. Upcoming observations with the IRS on Spitzer should remove the remaining uncertainties regarding the presence of large amounts of cold H2 in the disks of galaxies.
2 The exception is the Virgo cluster galaxy NGC 4522, which was found by Pierini et al. (1999) to have an abnormally high value for L[C II] / LFIR, possibly indicating mechanical heating of the interstellar gas as the galaxy interacts with the intracluster medium. Back.