Gas in Galaxies

The gaseous content of galaxies is a crucial counterpart to their stars. This is where stars come from, where some of their mater goes when they expire, and has intricate interactions with stars at many evolutionary phases. Furthermore, the spectra of gases afford emission features in the spectrum making them relatively easy to trace.

A remarkably useful tracer of overall gas content is the 21-cm H I line. Atomic hydrogen has two states in its ground level, split by spin-orbit coupling. Parallel spins give a slightly higher energy than antiparallel, and decays by emission of a photon at 1420.406 MHz. The upper level is populated by collisions, and has a decay half-life of about 3.5 x 1014 seconds = 1.1 x 107 years; this is detectable only because we are dealing with vast numbers of H atoms. Under some non-restrictive assumptions (spin temperature, optical depth), the total H I mass can be derived from the integrated observed line profile. The intensity unit is frequently taken as brightness temperature TB for historical reasons, though more recent VLA observations are expressed in Jy (so that the integrated line intensity is conventionally in Jy km/s).

Details are given by, for example, Kerr 1968 (in Nebulae and Interstellar Matter, Chicago, p. 575). The number density of H0 atoms in the antenna beam is N(H0) = 1.82 x 1018 Bint in atoms/cm2, where Bint is given in K km/s. This equation applies transparently to a cloud filling the beam (not a typical galaxy). In the more realistic case of an object smaller than the beam, one must multiply by Ω / 4 π D2, where Ω is the solid angle of the measurement beam and D is the galaxy distance. Following Roberts (SSS vol 9, p. 310) a more useful form is

where Sν is in Jy, Vr is in km/s, and D is in Mpc. One frequently uses distance-independent combinations such as M(H I)/LB to sidestep distance uncertainties altogether.

H I masses for single galaxies range up to a few 1010 solar masses. The absolute mass, and mass normalized to optical luminosity, vary systematically along the Hubble sequence as follows: Ellipticals generally have less than 107 solar masses, and that is frequently in large configurations possibly acquired from outside. There have been reports of optically thick H I in large amounts in cooling-flow galaxies such as M87, though more recent observations leave this conclusion in serious doubt. S0s may have a large range in H I mass, up to 109 solar masses. Sc's typically have several 109 solar masses. The very late types Sm,Im often have more mass in H I alone than in the visible stellar population; in this sense they may still be thought of as young galaxies. The trend with Hubble type is illustrated by Fig. 16 from Roberts 1975. Detailed aperture-synthesis maps of H I are becoming available for numerous galaxies, carrying rich information on kinematics as well as the H I distribution (see for example Cayatte et al. 1990, AJ 100, 604 on the Virgo cluster). As a sample, here are H I column density maps of three galaxies in the Sculptor group (NGC 247, 300, and 7793) as observed with the VLA by Puche, Carignan, and van Gorkom (1991 AJ 101, 456), who generously provided these images for the NRAO ``Radio Universe" CD-ROM.

In some spirals (the best-known example is probably NGC 3198), H I is detected well beyond the optical disk, and similar relative extents are not uncommon for very late Hubble types. This makes H I a powerful dynamical probe of dark-matter halos. It is common to have the edge of the H I disk abruptly cut off at column densities of a few times 1019 atoms/cm2, which may be due to the disk becoming ionized by the general radiation field of AGN and starburst galaxies just below the Lyman limit (see Maloney 1993 ApJ 414, 41). At these low densities, recombination is very slow so that ionization is a good way to hide the gas, since recombination is a collisional process whose rate scales as (density)2.

As technology allowed measurements in the millimeter regime, where strong transitions of astrophysically important molecules are found, it became clear that molecular gas is as important as atomic (H I) gas in many systems. Molecular hydrogen (H2) is the most abundant molecule, but since this molecule is a symmetric rotor, it has no permanent dipole moment and thus none of the mm-wave rotation and vibrational transitions that make other molecules easy to detect. H2 in the ground state can be detected only via absorption, in the Lyman and Werner bands deep in the UV (below 1108 Angstroms), and then only in the presence of a suitable background source (which rules out looking in most areas of high column density. Such absorption is ubiquitous in the Milky Way, as seen by virtually all FUSE spectra (Shull et al. 2000, ApJL 538, L73; this stuff is a positive nuisance if you are interested in the background sources themselves). H2 can also produce emission in the near-IR when appropriately excited, such as by either shocks or fluorescence, as seen in many IR-bright galaxies in the S(1) systems near 2.2 microns.

Because of the limitations in what kinds of H2 can be observed, and where, the most useful tracer of molecular gas has long been the trace molecule CO, which has a "fundamental" J=0-->1 transition at 2.6 mm (115 GHz) and higher-level transitions at higher frequencies. For the most abundant isotopic form 12C16O, most individual clouds are optically thick (note that much of a galaxy's molecular gas comes in giant clouds, the GMCs, with masses of order 106 solar masses). As it's sometimes phrased, the physics of CO emission is so complicated that it's simple (yeah, right!). It's commonly taken that a CO measurement more nearly counts clouds or sums surface area than directly measures gas mass. In a useful approximation, the luminosity in the CO line from a single cloud is L(CO) = ρ R2 TCO ΔV for cloud radius R and velocity dispersion ΔV, where TCO is the brightness temperature (i.e. the cloud acts as an opaque blackbody emitter across the optically thick part of the line profile). This means that in measuring the total CO line intensity from an ensemble of clouds, we are counting clouds (or totalling surface area) rather than getting a total mass. To get the typical mass per cloud, we rely on nearby clouds (Milky Way and Magellanic Clouds) in which we can resolve the linear extent as well as velocity dispersion, and assume the cloud is virially bound. This gives a mean conversion from CO to H2, 3.6 x 1020 cm-2 (K km s-1)-1, which is believed to be accurate to ~50% for global mass estimates in spirals -- see, for example, Young's 1990 review in The Interstellar Medium in Galaxies (Kluwer, p. 67). A check on this can be obtained from using less abundant isotopic forms (with 13C, for example). However, there remain problems with knowing how well the observed CO emission traces the mass distribution of a medium vastly dominated by H2 in the presence of both abundance and excitation effects. In particular, most of the molecular emission may arise in the thin skins of photodissociation regions (PDRs) where UV radiation is breaking down the molecules; to the extent this is true, our measures are biased toward star-forning regions.

As for H I, there are strong trends in CO luminosity with galaxy type. Controversy remains about what the basic trends are along the Hubble sequence, and in arm/interam comparisons, mostly because of different choices of normalization for the quantities. Molecular gas is especially important because these clouds are the immediate progenitors of star formation, and so there is a strong link between molecular mass and star-formation indicators. Interferometry is starting to trace the behavior of molecular clouds in disks, to see how it interacts with H I. As in the Milky Way, the most prominent molecular matter in spiral disks is in giant molecular clouds with estimated masses 106-7 solar masses, though a smattering of smaller ones may well be present and not be negligible in its contribution. In typical disks, the CO is strongly confined to clumpy regions tracing the ridges of the spiral patterm, although nuclear concentrations in early-type spirals and starburst systems are common enough as well. The disk pattern shows very well in this OVRO CO map of M51 overlaid on an HST image, from Aalto et al. 1999 (ApJ 522, 165, reproduced by permission of the AAS):

Contrasts between disk- and nuclear-dominated CO distributions are seen in comparing NGC 4414 and NGC 6574 from the survey by Sakamoto et al. (1999 ApJS 124, 403, again courtesy of the AAS):

NGC 4414 in CO NGC 6574 in CO

Even at the global level, there remain major unknowns in interpreting CO data. At lower abundances, characteristic of very late-type and irregular galaxies, the molecular clouds may be undetectable - there's not enough C and O present to give a signal, though the H2 should still form clouds (which can sometimes actually be seen by their effects on the neighboring H II regions; McCall, Hill, and English 1990 AJ 100, 193). In very dense regions, the molecular balance can shift in favor of more astronomically exotic species (from CS and HCN on up), so that we need to look at different transitions and consider a wide mix of cloud properties. The whole relationship between atomic and molecular gas has been re-evaluated by several studies of nearby galaxies (such as Allen et al. 1997 ApJ 487, 171) which suggest that molecular gas is the more basic form, with H I now being generated by photodissociation of H2 rather than H I being dominant and H2 being created transiently from association on grain surfaces. The grains are important, because the association rate for most astrophysical molecules in a pure gas phase is vanishingly small, but this is catalyzed by grains; atoms can be adsorbed onto grain surfaces, where the grain can act as a sink for momentum during molecule formation. In fact, there is some evidence (for example, Smith et al. 2000 ApJ 538, 608) that the change in grain population with metal abundance drives the overall change from atomic to molecular gas going inwards in luminous disk systems. It has become clear that much of the CO emission from galaxies is associated specifically with photodissociation regions (PDRs), interfaces where the molecular material is being dissociated by UV radiation from nearby hot stars (the most photogenic example probably being the famous Eagle Nebula). In these structures, there is a thin layer of heated, but still molecular, material, emitting more strongly than the colder, still undisturbed cloud interior.

A few molecular transitions can be pumped, either collisionally or by ambient ultraviolet radiation, to give astrophysical masers. These are found in star-forming regions or near active nuclei, and can be interested dynamical tracers (for galaxy rotation or central mass). OH emission at 18 cm (1665 and 1667 MHz, additional "satellite" transitions radiate at 1612 and 1720 MHz) is especially important for galaxies, sometimes producing "megamasers" in IR-bright systems and sometimes pumped by a central AGN. This was the tracer used to examine the dynamics of the inner gas disk in NGC 4258, suggesting a large central mass (Nakai, Inoue, & Myoshi 1993 Nature 361, 45). Water vapor also appears as a maser, with cameo roles played by H2CO and CH. Beyond the interesting physics of pumping and amplification, masers are observationally useful because they are so confined both spatially and spectrally - making them ideal tracers for line-of-sight or proper-motion velocity studies. For example, the VLBA is not far from being able to measure geometric distances to the Virgo cluster via proper motions of masing region in the disks of spirals.

The first ISM phase to be amenable to study is often the least important by mass, yet remains wonderfully informative - ionized gas. Electron transitions give rise to highly diagnostic sets of emission and absorption lines. Emission is especially useful, since it can be traced spatially using narrow-band techniques, and line ratios can tell use characteristic abundances, temperatures, and densities much more directly then other approaches such as modelling stellar atmospheres. We'll have more to say on these issues when dealing with star-formation rates and AGN. For now, I'll note that the major processes are ionization, recombination, and collisional cooling. For either recombination or collisional excitation, line ratios can tell us abundances of the relevant ions (which sounds more obvious than it really is - the ratios can go backwards in some regimes), densities (from comparing lines with very different cross-sections for collisional de-excitation before emitting) and temperatures (comparing transitions which require substantially different energies for collisional excitation in the first place, so they sample different pieces of the electron energy distribution). In interpretation, we must watch closely where the gas came from - an H II region quickly becomes enriched by products of the same burst of star formation that lights it up for us to see, and in chemically young systems, the H II regions may give a very biased view of the overall composition. There has been recent progress using far-UV absorption lines from the surrounding ISM, which may give very different chemistry than the emission lines.

As is set out in the star-formation lecture, the luminosity in recombination lines is related to the ionizing luminosity and spectrum. This works well because it is such a global bit of bookkeeping - it has become clear that H II regions are seldom much like Stromgren spheres, and more often have much of the line radiation coming from an ionization front which is being driven into a surrounding dense cloud. O'Dell in particular has made this case, most clearly for the Orion Nebula.

Narrow-band imaging can quickly trace the location of ionized gas, though extinction is often a concern. NGC 5427 in Hα illustrates how organized regions of star formation can be, traced now by the OB stars which are hot enough to ionize hydrogen into H II regions:

Because of dust extinction, one always wants to use lines at the longest possible wavelength. As Susan Kleinmann once said, "I don't care if it's plastic, as long as it's in the infrared and you can see it". For hydrogen, this has driven work into the Paschen and Brackett lines in the near-IR, although instrumental and background limitations have made this slow going. A NICMOS filter survey was especially fruitful in unveiling star formation deep in the dusty centers of galaxies. Seeing line emission does require a source of energy input, so that hydrogen is ionized only where 13.6 eV per atom is readily available. This may come from OB stars, from hot evolved stars (as in planetary nebulae), or from interstellar shocks (as in supernova remnants). A few ions, like S+, have lower ionization potentials, and since radiation longward of the Lyman limit at 912 Angstroms isn't absorbed by neutral hydrogen, there's plenty of starlight available to ionize these species. Also, collision excitation of neutral atoms by electrons released during hydrogen ionization can be important, so that emission from neutral O and N occurs in H II regions. [Note that the ionization stage is denoted as N0, N+,N++, while the associated spectral lines would be N I, N II, N III, along with brackets if they're so-called forbidden transitions from metastable upper levels. The numbering schemes differ in being zero - and one-indexed; by now we seem to be stuck with this nomenclature. To make it worse. for recombination lines the atom starts in the next ionization stage up.]

Some heavy ions have fine-structure lines, transitions between very low-lying states near the ground level, lying in the mid- and far-infrared. These can be important cooling terms in the overall gas energy budget, since some of these ions have ionization potentials for formation less than one Rydberg (13.6 eV) so that there's lots of radiation available to keep them ionized. A particularly important example is the [C II] line at 157.75 microns, which is ubiquitous both in the Milky Way and in other galaxies observed globally (see the Nov. 1996 special ISO issue of A&A). This single line may carry 0.1% of a galaxy's total luminosity. It originates in ``warm" gas outside H II regions, typically at 300 K and 1000 particles/cm3.

The overall interstellar medium of galaxies is foamy (and, as argued by Elmegreen, maybe even approximately fractal), with the various molecular, neutral atomic, and ionized phases often thought to be close to pressure balance (but note that EUVE results on the local ISM seem to rule this out). The interstices between clouds are more or less filled with hot gas at ~10 7 K, detected most often by absorption lines from highly ionized species such as O VI seen against distant stars, QSOs, or (if you look quickly) supernovae. The heating source is usually supernova explosions; in starburst galaxies they may merge to create a global wind or superwind, visible through both line emission from shocked gas and thermal X-rays. The ISM is clearly dynamic, like so much of the Universe once we learn to see the snapshots as part of a movie.

« Systematics and galaxy statistics | Dust in galaxies »

Course Home | Bill Keel's Home Page | Image Usage and Copyright Info | UA Astronomy
Last changes: 8/2009	  © 2000-9