4.1. Molecular gas in spiral galaxies from CO surveys of local galaxies
Because H2 is so difficult to detect in the normal interstellar medium (Section 10), most of our knowledge about the global molecular gas of other galaxies than the Milky Way also relies on CO observations, with a similar problematics to precisely infer the amount of H2. In parallel with mapping molecular gas in the Milky Way, since the first detection of extragalactic CO (Rickard et al. 1975), enormous efforts have been devoted in the last three decades to perform the inventory of molecular gas in galaxies, through systematic millimetre observations of CO in a large, diversified sample of local galaxies. This enterprise has first addressed the main reservoir of molecular gas, large spiral disk galaxies with their various classes, including that of the Milky Way. Single dish CO observations, with relatively low angular resolution, were generally first performed to derive global properties of the molecular gas, its total amount, proportion with respect to HI, relation with metallicity, morphology, star formation, etc. Detailed studies of fine details have then been undertaken, mostly with interferometers. In parallel, the studies have been extended to other classes and peculiar galaxies where the importance of molecular gas is more marginal.
The results are impressive and provide a remarkably complete view of molecular gas in (local) galaxies. The emerging picture is perhaps more complex than initially thought. Molecular gas is extremely intricated with atomic gas, and the most important quantity for the evolution of galaxies and even in some way star formation, is the total amount of gas HI + H2. Molecular gas is intricately woven with the atomic gas. Because of the constant exchanges between HI and H2, the gas should be considered globally with its dynamics, transport properties including exchanges with the intergalactic medium and the central regions, and condensation processes leading to star formation and its feedback action on the gas. We will defer most problems related to star formation, including starburst galaxies, to Section 5.
The total number of local (D 100 Mpc) galaxies where CO data has been detected, approaches 1000, whith more than half being `normal' spirals Sa-Sb (see, e.g. the global analysis by Casoli 1998, the catalogue for non interacting galaxies compiled by Bettoni et al. 2003, and more recent observations including Boselli et al. 2002, Sauty et al. 2003, Hafok & Stutsky 2003, Yao et al. 2003, Garland et al. 2005, Leroy et al. 2005). These data were mostly obtained in dedicated surveys of the 12CO(1-0) line with 12-15 m radiotelescopes whose intermediate size is well fitted to such programmes: the most organized, initial long-term effort was performed with the FCRAO 14 m telescope and achieved more than one third of all these detections (Young et al. 1995, 1986, 1989). Other important contributions came from various groups (see references in Bettoni et al. 2003 and recent results quoted above), using mainly the NRAO 12 m dish and SEST 15 m, as well as other telescopes such as BTL, KOSMA, JCMT, etc., and the larger dishes of Onsala 20 m, IRAM 30 m and Nobeyama 45 m.
Most of these observations, carried out with a limited angular resolution (e.g. 45" with 14 m FCRAO, ~ 2 kpc at 10 Mpc), were single pointings aimed at a global CO detection from the central regions in objects with small angular extension, or often a few pointings along the major axis in order to trace the large-scale radial distribution of molecular gas (e.g. Young et al. 1995). However, several tens of detailed maps have been carried out with the much better angular resolution of the largest single dishes (e.g. Nishiyama & Nakai 2001) or especially interferometers such as the BIMA survey of 44 nearby galaxies (Regan et al. 2001), and the high-resolution (0.5"-1") NUGA survey at IRAM of 12 low luminosity AGN (García-Burillo et al. 2003). Comprehensive high resolution maps were performed for most of the members of the Local Group: M 31 with FCRAO (Loinard et al. 1999) and IRAM-30m (Neininger et al. 1998, 2001, Nieten et al. 2006; see Fig. 1); M 33 (Engargiola et al. 2003), dwarf spheroidals (Blitz & Robishaw 2000) with BIMA; and the Magellanic Clouds with SEST and NANTEN (see below). See also the review by Blitz et al. (2007) of the properties of GMCs in the Local Group., and CO maps of M 51 by Scoville & Young (1983) and Schuster et al. (2007).
Figure 1. (reproduced from Fig. 1 of Nieten et al. 2006) Distribution of CO emission in M 31 (Andromeda). (a) (top) The velocity-integrated intensity distribution of the 12CO(1-0) spectrum, observed with the IRAM 30-m telescope. The X and Y coordinates are taken along the major and minor axis, respectively. The dashed line marks the border of the area surveyed which is about one degree squared. (b) (bottom) The velocity field as traced by the CO emission.
Typical massive spirals are the standard habitat of molecules in the Universe. Such a large sample of several hundreds provide a very comprehensive view of their molecular medium with its complexity. Various analyses of the properties of molecular gas in spiral galaxies have been published in the last two decades (e.g. Young & Knezek 1989, Young & Scoville 1991, Bregman et al. 1992, Braine et al. 1993, Latter et al. 1996, Casoli et al. 1998, Boselli et al. 2002, Wong & Blitz 2002, and references therein). The first striking observation is the enormous range of variation of the molecular content of galaxies, even within the relatively homogeneous class of massive spirals Sa-Sc (e.g. Bettoni et al. 2003). The ratio ICO / M(HI), approximately proportional to M(H2) / M(HI), spans two orders of magnitude for a given value of any galactic scaling parameter such as the blue or far-infrared luminosities Lb and LFIR. The average values of ICO / M(HI), however, clearly correlate with various quantities such as the metallicity, the dynamical mass, Lb or LFIR.
The difficulty of precisely estimating the conversion factor X = NH2 / ICO adds a major source of uncertainty in the values of M(H2) / M(HI). Its calibration for the various types of galaxies remains a very complex and uncertain process. There is presently no firm basis for a precise physical calibration because of the complexity of the interstellar medium, although better understanding the physics may provide the best constraints for its value. Its empirical calibration relies on the detailed information on the local ISM of the Milky Way where the gamma rays are the fundamental calibrator (Section 3.2), and on less accurate methods applied to a sample of various types of well studied galaxies, generally nearby (see e.g. Boselli et al. 2002). The most useful of these methods for estimating the mass of the molecular gas seems the virial equilibrium of giant molecular clouds (Young and Scoville 1991), or the determination of the mass of cold dust from millimetre/submillimetre observations assuming a metallicity-dependent gas-to-dust ratio (Guélin et al. 1995). However, even in nearby galaxies such determinations of X remain very uncertain. It should also be reminded that, inside a given object, X may change by a factor ~ 10 from the diffuse medium to the core of the GMCs (Polk et al. 1988). Despite such difficulties, the most recent analysis by Boselli et al. (2002) has confirmed that the most massive spiral neighbours of the Milky Way have comparable X factors, but with non negligible variations up to a factor ~ 2 of the best average MW estimate from ray emission (Eq.3), 1.56 ± 0.05 1020 cm-2 (K × km × s-1)-1 (Hunter et al. 1997). Many former studies use a previous MW calibration of X, 2.38 × 1020; they should thus be recalibrated by at least a factor ~ 1.5, maybe more, as estimated for the average of X with a larger sample by Boselli et al. (2002). Consequently, it seems now well established that the global amount of molecular gas is on average significantly smaller than the atomic one, with average values of M(H2) / M(HI) 0.2-0.4 for large spirals depending on the sample (see Fig. 2b and last discussions by Casoli et al. 1998, Bettoni et al. 2003 and Boselli et al. 2002), which is similar to the Milky Way value 0.2 (e.g. Blitz 1996).
Figure 2. (a) (left, reproduced from Fig. 2 of Boselli et al. 2002) The relationship for a template sample of nearby galaxies between the X conversion factor from CO line intensity to H2 column density (Eq. 1) and the metallicity index 12 + log(O/H). (b) (right, reproduced from Bettoni et al. 2003) The mean molecular to atomic gas content ratio as a function of galaxy Hubble type t. The open symbols are derived from the catalogue of Bettoni et al. (2003), while full symbols represent ratios published by Casoli et al. (1998).
It is also well established that the molecular gas radial distributions in most spirals galaxies are centrally peaked with exponential profiles, and thus markedly different from what is seen in HI (see e.g. Young 2000). Indeed, the Milky Way is one of the rare exceptions with its CO depression interior to the molecular ring. For the vertical distribution, the molecular gas is more concentrated in a thin disk than HI, as seen from the CO emission of edge-on galaxies. However, the arm-interarm contrast is generally not very strongly marked (but see Loinard et al. 1999 for M 31).
An easily accessible quantity, the isotopic line ratio I(12CO) / I(13CO), may give some indication about the value of X and its variation across a galaxy. However, this ratio is sensitive to variation in temperature and column density, as well as isotope fractionation and especially isotope-selective photo-dissociation. Mapping surveys of 12CO and 13CO emission in nearby galaxies have been used to trace the properties of their molecular gas and the variation of X, especially using the 4x4 receiver array of FCRAO by Paglione et al. (2001 and references therein). They have concluded that similar physical processes may affect the value of X and I(12CO) / I(13CO), and that X might decrease by factors of 2-5 from disks to nuclei.
Stellar bars drive gas into the circumnuclear region of galaxies. CO studies of the molecular gas in bars are important to investigate their structure and dynamics, and their influence on star formation in circumnuclear regions. Such CO observations of the bar and the inner region (about the central kiloparsec) have been carried out in a few tens of barred galaxies (e.g. Kenney et al. 1992, Regan et al. 2002, Lee et al. 2006, Jogee et al. 2005). It is found that the mean nuclear molecular gas surface density of barred spirals is significantly higher than that of unbarred spirals, explaining in part why powerful starbursts reside there.
4.2. Molecular gas in other types of galaxies
Even within the relatively homogeneous class of massive spirals, there are very significant variations of the relative amount of molecular material and the ability of CO to trace H2 through the X-factor, between different galaxies and within a given galaxy. The relationships between the H2 / CO conversion factor X and various galactic parameters - UV radiation field, metallicity, blue and near-IR luminosities - have been analysed by Boselli et al. (2002) and displayed in their figure 2. As expected, there is a strong dependence of X on these parameters, e.g. the CO abundance, and hence X-1, decrease with the UV strength and increase with the metallicity (see e.g. Fig. 2a for the metallicity). It is thus not surprising that the range of the variations of X is much increased when one encompasses the full variety of galaxies. Other galaxies are generally less favourable than spirals for the survival of molecules in general and CO in particular, their molecular gas is therefore less spectacular, and CO emission even relatively weaker. Even the latest-type spirals are in general not strong CO emitters (Böker et al. 2003, Young & Knezek 1989). Recent models of galaxy evolution, incorporating the formation of H2 out of HI gas, have also explored the possibility of a diffuse H2 gas phase outside star-forming regions (e.g. Pelupessy et al. 2006). The mass of such diffuse warm H2 should be significantly underestimated by CO observations in metal-poor regions (Papadopoulos et al. 2002).
The Magellanic Clouds are particularly interesting because of their proximity which allows us to check the molecular gas with much detail about conditions of star formation, UV intensity and metallicity very different from the Milky Way. A first full coverage in the 12CO(J = 1-0) emission line at 115 GHz, was performed at low angular resolution with the Columbia 1.2 m telescope (Cohen et al. 1988). Then both LMC and SMC were the object of systematic survey studies of 12CO and 13CO with the SEST 15 m telescope (Rubio et al. 1991, Israel 1997, Israel et al. 1993, 2003, and references therein), and with NANTEN (Mizuno et al. 2001a, b, Yamaguchi et al. 2001) (see also studies of higher-J lines of CO with AST/RO, Bolatto et al. 2005). At the distance of the Clouds, the SEST beam size, 13 pc, is smaller than the typical size, ~ 20 pc, of their giant molecular clouds, itself significantly smaller than GMCs in the Milky way, ~ 50 pc. Dwarf galaxies such as the Magellanic Clouds are generally poor CO emitters, so that most SEST studies were limited to regions with significant CO emission, mainly traced by far-infrared IRAS emission. CO was detected at strengths significantly smaller than those expected from Galactic sources at Magellanic Cloud distances, typically three times weaker in the LMC and an order of magnitude lower in the SMC (Israel et al. 1993). The ratio X = NH2 / ICO is also 2-3 times larger than in the Milky Way. Similarly, the emission of CO associated with well developed HII regions remains quite modest, and the lack of diffuse CO emission there (e.g., Lequeux et al. 1994) suggests that these molecular clouds are generally part of photo-dissociation regions (PDRs).
Large samples of Magellanic type dwarf galaxies and irregulars have been observed: for instance, Albrecht et al. (2004) detected 41 galaxies with the IRAM 30m-telescope, and Leroy et al. (2005) detected 28 with BIMA (see also Hoffman et al. 2003). Leroy et al. found that the CO luminosity is most strongly correlated with the K-band and the far-infrared luminosities. There are also strong correlations with the radio continuum and B-band luminosities and linear diameter. Conversely, they found that far-IR dust temperature is a poor predictor of CO emission within the dwarfs alone, although a good predictor of normalized CO content among a larger sample of galaxies.
Various studies of low surface brightness galaxies (LSBs), (e.g. Schombert et al. 1990, O'Neil & Schinnerer 2004, Matthews et al. 2005) have demonstrated that despite their typical low metallicities and low mean gas surface densities, some LSB galaxies contain a molecular medium that is traced by CO. MH2 and MH2 / MHI values fall within the ranges typically found for high surface brightness objects, albeit at the low end of the distribution.
CO was detected in several tens of elliptical galaxies (e.g. Wiklind et al. 1995, Knapp & Rupen 1996, Sofue & Wakamatsu 1993, and references therein). It was found that the CO-to-dust abundance ratio in elliptical galaxies is approximately the same as that for spirals and for local molecular clouds. The molecular gas masses range from 2 × 106 to 109 M, and appear to be unrelated to the underlying stellar population. This suggests an external origin of the gas. Low excitation temperatures for CO transitions in galaxies with cold dust could lead to an underestimate of the molecular gas mass by a factor of 5. The average MH2 / MHI ratio for the elliptical galaxies is 2-5 times lower than for normal spiral galaxies.
Millimetre CO emission has been detected in the cooling flows of a dozen central massive elliptical galaxies of clusters (e.g. Edge & Frayer 2003, Wilman et al. 2006, Salomé & Combes 2003, 2005, 2006 and references therein). It is important to understand the exact nature of their complex structures (bubbles, cavities, cold fronts) unveiled by X-ray data, which may contain huge optical nebula. It seems now established that cooling flows entertain some fueling of the AGN activity which reheats the intra-cluster gas. CO was also detected in the context of galaxy collisions in the tidal debris of violent galaxy-galaxy interactions (Braine et al. 2000), and particularly in the Stephan's Quintet group of galaxies (Lisenfeld et al. 2004). The cold gas is probably a mixture of gas falling down on the central galaxy and of uplifted gas dragged out by a rising bubble in the intracluster medium. Its peculiar morphology and kinematics argue for the picture of an intermittent cooling flow scenario where the central AGN plays an important role.
To sum up, the picture of the molecular medium coming out of CO observations of many hundreds of galaxies is complex, reflecting the variety of the history, mass, luminosity and metallicity of the host galaxies. However, as expected, definite correlations emerge between the last parameters and the amount of molecular material as well as with the `X conversion factor', NH2 / ICO. In spiral disks, the average fraction of the interstellar gas in molecules does not much depend on the precise morphological type of the galaxy and is comparable to the Milky Way value, ~ 0.2. But it significantly decreases in earlier and later types, i.e. ellipticals and irregulars, enhancing the dearth of molecular gas there where the total amount of gas is low. The increase of the X factor itself with decreasing metallicity renders more difficult the detection of CO in irregular dwarfs. However, the proximity of the Magellanic Clouds provide us with very sensitive benchmarks of the CO distribution in such cases. The detection of CO in a number of ellipticals shows that it can be an interesting tracer of gas when present, as well as in cooling flows of central massive ellipticals of clusters.