Annu. Rev. Astron. Astrophys. 1982. 20: 517-45 Copyright © 1982 by Annual Reviews. All rights reserved

3. THE DISTRIBUTION OF MOLECULAR MATERIAL WITHIN GALAXIES

The galaxies for which maps have been made in the ¹²CO, J = 1-0 line at 2.6 mm are listed in Table 2. A typical map (for M51) is displayed in Figure 1, which shows contours of integrated intensity under the CO line profile, P(CO) = T*_A (CO) dv [K km s^-1]. According to convention, T*_A is not corrected for forward coupling efficiency of the telescope.

The integrated intensity, P(CO), can be converted to a surface mass density of molecular hydrogen, _H2, but only by applying sets of assumptions that are individually ill-justified. In one procedure (Encrenaz et al. 1979, Rickard et al., in preparation), a given kinetic temperature is adopted (e.g. 25 K), and it is assumed that the lowest levels of CO are thermalized at this temperature. One then corrects for optical depth by assuming [¹²CO] / [¹³CO] = 50 and applying in general the mean value of P(¹²CO) / P(¹³CO) = 13.5 observed by Encrenaz et al. (1979). Finally, one assumes [¹²C0] / [H₂] = 5 x 10^-5 (Leung & Liszt 1976) and thus obtains the relation _H2 [M pc^-2] = 1.8 P(¹²CO). An alternative path starts on the assumption that the molecular medium is comprised of "standard clouds" having specified parameters (Solomon & de Zafra 1975, Morris & Lo 1978). One then finds for the surface density that _H2 = (M/WA) P(CO), where M is the mass, W the undiluted integrated intensity, and A the cross-sectional area of a standard cloud. For M = 10⁵ M, W = 20 K km s^-1, and A = 2 x 10³ pc², one finds _H2 = 2.5 P(¹²CO). The two methods are in reasonable agreement, reflecting the fact that the mass of a standard cloud is ultimately derived on the basis of assumptions similar to those of the first procedure. Although these methods are crude, some comfort may be taken from the agreement of the resulting gas masses with those derived by quite different methods. For example, Elias et al. (1978) derive H₂ masses for the central sources in M82 and NGC 253 from 1-mm continuum data, agreeing with the Rickard et al. (1977a) CO estimates to within a factor of two. For consistency, we adopt _H2 = 1.8 P(CO) throughout this review. The principal uncertainties lie in the assumed values of P(¹²CO) / P(¹³CO). [¹²CO] / [¹³CO], and [¹²CO] / [H2]; we estimate the overall uncertainties to be a factor of three in the inferred CO column densities, and a factor of ten in the surface mass densities.

Some of the more obvious errors have a systematic character. In the two cases for which P(¹²CO) / P(¹³CO) has been measured away from the nucleus - our Galaxy (Solomon et al. 1979) and M31 (Encrenaz et al. 1979) - the ratio is about three times less than the average value for the central sources of all galaxies measured. Also, if [¹²CO] / H₂] mirrors the galactocentric variation of [C]/[H], the outer disk value may be about three times less than the central value (Talbot 1980). A decrease in cosmic-ray heating in the outer disk max result in lower values of P(¹²CO) for clouds of comparable mass to those closer to the center (Kutner & Mead 1981). All these effects would imply underestimates of _H2, in the outer disk, relative to the center.

GALAXIES WITH STRONG CENTRAL CO PEAKS     The majority of galaxies mapped in CO have pronounced intensity maxima centered on the galactic nuclei. Among these galaxies are M51, NGC 6946, IC 342, Maffei 2, and M83, P(CO) is four to six times greater toward the centers than in the surrounding disks. (The variation in peak T*_A between center and disk is seldom more than a factor of two; the remaining difference is due to the larger velocity extent of the central emission.) In most galaxies with central CO peaks, the radial distribution of CO emission, expressed as the variation with radius of the azimuthally averaged value of P(CO), is either flat or slowly decreasing with galactocentric radius over most of the disk, in some cases being detectable out to radii that are as large as 10 kpc. The disks are far from uniform, however - there is considerable fluctuation about the azimuthal mean at any given radius. At the edge of the bright stellar disks (e.g. as seen for M51 in Sandage 1961), there is a distinct drop-off in P(CO). At present, it is unclear whether this is sudden or (following the distribution of optical luminosity) is exponentially declining (Lord et al. 1981, Tacconi et al. 1981, Young & Scoville 1981). For example, a chi-square analysis of the M51 radial distribution cannot distinguish between a disk of constant flux out to some critical radius and an exponentially declining flux (Rickard & Palmer 1981b). The basic reason for the uncertainty is that the fluctuations in the disk emission are of comparable size to the azimuthal mean values. In either case, isolated CO clouds do appear beyond the main body of molecular material, such as toward specific H II complexes in the faint outer regions of IC 342 (Morris & Lo 1978). For all well-observed galaxies, the atomic hydrogen distributions extend well beyond he rough limits of the CO disks (e.g. Shane 1975, Rogstad et al. 1973, for M51, NGC 6946, and IC 342).

Figure 1. Map of integrated 12Co emission at 115 GHz from M51, taken from Rickard & Palmer (1981b). The contour levels are spaced by 2 K km s-1. The crosses mark positions where data were taken; the shaded circle is the telescope beam size.

The peak values of P(CO) observed toward the central regions typically correspond to values of _H2 in the range 40-60 M pc^-2, and thus to total H₂ masses of 2 x 10⁸ M to 2 x 10⁹ M. As noted by Morris & Lo (1978; see also Rickard et al. 1977c), these central molecular concentrations often contain enough gas mass to "fill-in" central depressions in the HI distributions, so that in many cases the total proton density is a slowly declining function of galactocentric radius. This behavior is illustrated for the case of NGC 6946 in Figure 2. To some extent, this claim is predicated on the present spatial resolution of the CO data. It is possible that higher resolution data will show that the central H₂ maxima are much more confined than the HI holes, and thus that the total proton densities have an annular minimum at 1-2 kpc radius. Also, there are a few galaxies with central HI minima that are not compensated by H₂, such as M31. In addition, M51 shows no central HI deficiency (Shane 1975), so the presence of a central CO peak does not necessarily imply an HI hole.

Away from the central peaks, the inferred values of _H2 are typically about 7 M pc^-2 and the total H₂ masses in the observed disks are generally several times 10⁹ M. By comparison, a recent estimate for our galactic disk is 14 M pc^-2 (Liszt et al. 1981). The two are thus in good agreement, especially considering the uncertainty of the extragalactic value. Furthermore, _H2 ~ _HI in the disks at ~ 5 kpc (Rickard & Palmer 1981b), compared with _H2 ~ 6_HI for the disk of our Galaxy (Liszt et al. 1981). This apparent discrepancy may be due to a deficiency of atomic gas in our Galaxy compared to spirals of similar type (Liszt 1980), or it may reflect an underestimate of the opacity in the 21-cm HI line. (Indeed, our perspective on the Milky Way maximizes this opacity.) The _H2 / _HI ratio in the disks of spiral galaxies deserves further study, since models for the formation and destruction of molecular clouds depend sensitively on this quantity (see Section 5).

Of course, one must keep in mind the possible errors in the conversion from P(CO) to _H2. Looking at Figure 2, one can see that a small change in that calculation will drastically shift the location of the radius at which _H2 = _HI and thus whether molecular gas dominates only within the inner galaxy or over the entire disk. The shape of the total proton distribution in the outer disk depends sensitively on this conversion, a point that must be borne in mind when, for example, analyzing the dependence of the star formation rate on the density.

In all the radial distributions thus compiled, there is no sign of an annular minimum at galactocentric radii of 2-4 kpc. Such a minimum was found in our Galaxy by Scoville & Solomon (1975) and Gordon & Burton (1976), and was identified by Gordon (1978) with the inner Lindblad resonance. By the same token, Young & Scoville (1981) argue that the smooth radial distributions of CO in NGC 6946 and IC 342 are consistent with absences of inner Lindblad resonances. It is possible that in some galaxies this feature might be concealed by inadequate angular resolution, but in most cases it is definitely not present.

Figure 2. Radial distributions of atomic and molecular gas in NGC 6946. The atomic mass surface density, HI, is taken from Rogstad et al. (1973); the molecular mass surface density, H2, is taken from Rickard & Palmer (1981b). The vertical bars represent the scatter about the azimuthal averages. The dashed line is the distribution of the total proton mass surface density.

The maps of the galaxies with strong central peaks show structure on a variety of scales in their disks. Peaks and depressions as small as the beam size can be seen in Figure 1. Individual peaks can be identified with specific HII complexes; for example, the plateau in M51 at (+ 10^S, + 2') seems to be associated with the bright HII regions of the northeast arm. But there is nothing apparent in the form of a coherent, large-scale, nonaxisymmetric structure (i.e. a spiral pattern). For M51 and NGC 6946, low angular resolution could be responsible for washing out details on the scale of the optical spiral structure, at least in the inner regions, but an extreme contrast (such as in the HII region distribution) would still be detectable. In 45" resolution observations of NGC 6946 and IC 342, Young & Scoville (1981) find no arm-interarm contrasts larger than a factor of two. For M51, Rickard et al. (in preparation) constructed model spiral patterns based on the Tully (1974) H maps and fit them to the CO data. They found that the arm-interarm contrast is probably less than a factor of six. However, the arm-interarm contrast over the bright disk of M51 does not exceed a factor of three even for HI. The HI arms emerge only beyond the 3' radius at which the CO emission has fallen below the detection limit.

GALAXIES WiTHOUT STRONG CENTRAL CO PEAKS     M31 clearly differs from the centrally concentrated galaxies. There is no evidence in M31 for a central source, despite several deep searches (Rickard et al. 1977a, Blitz, private communication). However, CO emission has been measured over much of the optical disk at a somewhat reduced level compared with the disks of galaxies having strong central sources. It would seem that the central sources represent a distinct molecular component not present in M31.

In the disk component, there are additional distinctions between M31 and the galaxies with strong central components. Stark et al. (1981) find that the average integrated CO intensity is less than 20% of the value in similar regions of our Galaxy, and the inferred _H2 / _HI is less than 0.2. They have mapped out a portion of the southwest spiral arm segments, finding considerable structure (Figure 3). The peak CO emission is excellently correlated with the HI arm segment and its associated HII regions and dust clouds. The contrast between arm and interarm regions is as much as a factor of ten, comparable to that in the HI (Unwin 1980). Curiously, the CO structures show velocity coherence over a scale of ~ 1 kpc, with no indication of the streaming motions expected from density-wave theory.

Figure 3. Map of integrated 12CO emission from the southwest spiral arm segments in M31, taken from Stark et al. (1981). The contours are in units of K km s-1; the axes are offsets from she nucleus in a coordinate system aligned with the major and minor axes.

M81 also appears to be a galaxy with CO in the disk but not the center (Combes et al. 1977a). If the relative weakness of the emission from such galaxies is common, it is not surprising that few have been detected.