|Annu. Rev. Astron. Astrophys. 1994. 32:
Copyright © 1994 by . All rights reserved
Verter (1985, 1990) and Young & Scoville (1991) have summarized extragalactic CO observations. More recent survey results are given by Braine et al. (1993), Sage (1993), and Ohta et al. (1993). There is a strong correlation between FIR and CO fluxes. However, a significant fraction of the data in the literature refer to galaxies chosen by some IR criterion which in turn reflects active star formation, frequently triggered by interactions with companion galaxies. For such interacting systems Braine & Combes (1993) and Sage (1993) find a higher mass fraction of molecular hydrogen, MH2/MT than for more isolated galaxies. Sage derives a mean increase of about a factor of two. Thus some of the statistical studies have samples rich in CO because active star-forming galaxies were preferentially chosen. Reviews and conference proceedings include those by Young & Scoville (1991), Combs (1992), and Combs & Casoli (1991).
Working with a volume-limited sample Sage (1993) found that for galaxies of type Sa-Sc, MH2 is approximately 2% of the total mass within the region where the H2 is measured. H2 appears to be a smaller fraction in Sd-type galaxies, significantly so if the upper limits on MH2 in his data set are considered. To derive MH2 Sage assumes that the conversion factor for CO to H2 is the same in all galaxies and equal to the Galactic value. Essentially all values of MH2 in the literature make this assumption. This appears to be a reasonable approach for M31 and M33 but not so in later-type galaxies (Cohen et al. 1988; Rubio et al. 1991; Ohta et al. 1993). What is evident for these late-type systems is the general faintness of the CO radiation, even though current star formation is ongoing. We must conclude that values of the molecular hydrogen content in late-type systems (Sd, Sm, Ir) derived in this manner are uncertain and possibly too low by up to an order of magnitude.
Uncertainty of a different sort also exists for CO results in early-type systems. For ellipticals there is the well recognized problem of classification uncertainty (Buta 1992a, b; Roberts et al. 1991; Bregman et al. 1992; Hogg et al. 1993). If we wish to speak of the gas content of different sorts of early-type galaxies, e.g., E or S0, it is important that we minimize the classification uncertainty. Unfortunately, this is difficult, and the literature is correspondingly confusing. As an example, Buta (1992a) calls attention to a well recognized spiral, NGC 3928, which because of its compactness is typed as an E0 on the Palomar Sky Survey prints. This ``elliptical'' stands out in the IRAS lists and was successfully searched for CO (Gordon 1990). This is an extreme case. Generally the disagreement exists among the similar-in-appearance types of E, E/S0, S0, and Sa. But it is just in this range of galaxy type that the question is raised of bifurcation or continuity in various properties. Until quantitive measurement of type becomes possible, uncertainty will remain.
Weak CO emission has been detected in only two of over a dozen of the RSA catalog ellipticals that have been searched in depth (Sofue & Wakamatsu 1993; Roberts et al. 1991; and Bregman et al. 1992). Only three (out of 64 searched) isolated RSA ellipticals have HI detections. Two of these appear to be examples of capture and the third, NGC 2974, shows spiral features on deep imaging; it is an Sa with low surface brightness arms. Thus cool gas, CO and HI, is a rarity in the earliest of galaxies, the ellipticals, while hot, X-ray emitting gas is quite common in these systems. Cool gas is more frequently found in the early galaxies of type S0, S0/Sa, and Sa, the frequency of detection increasing in this sense. There are also instances of stringent upper limits for these galaxy types for CO and for HI. Thus, unless survival analysis is used to derive average values of MH2 and MHI, trends regarding these quantities for the early-type categories will be too high.
It is both of these effects - the conversion constant for late-type galaxies and inclusion of upper limits for early-type systems - that could well alter any trend with morphological type. With this caution in mind we consider the various relationships for the molecular gas content with type. These are described in the review by Young and Scoville (1991). Those appropriate to our discussion are:
|1.||MH2/LB. This ratio is essentially constant for types Sa-Sc and then decreases for later types, Scd-Sdm; the scatter is large. A similar trend in terms of fractional mass content, MH2/MT is found by Sage (1993) using a differently selected sample. This near constancy over Sa-Sc followed by a decrease for later types is also seen in the intrinsic quantities of radius, blue luminosity, and total mass. However, the decrease in MH2/LB could also reflect a changing conversion constant. We conclude that MH2/LB and MH2/MT are near constant over types Sa, Sb, and Sc. All we can say for later types is that the appropriately normalized CO luminosity is less.|
|2.||MH2/MHI. This ratio decreases with type S0/Sa-Sd/Sm; the scatter is large. Sage (1993) finds the same trend for a different sample. He attributes it to the long recognized increase of MHI/MT with type since MH2/MT is essentially constant (for Sa-Sc). The range and details of the MH2/MT variation with type are obviously dependent on the conversion factor and the statistical treatment of the upper limits. He further notes, as do others (e.g., Devereux & Young 1990), the spatial anticorrelation of these two forms of hydrogen in galaxies, including our own; H2 is more centrally concentrated than HI. We conclude that the global decrease in MH2/MHI is in the expected sense and reflects the increase in MHI/MT with type.|
|3.||(MH2 + MHI) / Area. This quantity, the total cool gas surface density (note that the area is based on the optical diameter D25), increases with later type. This is the sense of the trend for HI, as well as for H2 (Young & Knezek 1989). For the latter, constancy of H2 over a limited mid-type range is also permissible from the data since there is the usual uncertainty at the extreme types. We conclude that the surface density of cool gas increases with later type. This is an important systematic trend, similar to that displayed by H II regions within galaxies of different type.|