This conference is mainly concerned with the results of observing HI in galaxies, a subject which owes its advancement to the great strides we have made in radiometers and radio telescopes since the first detections in the early-1950s. These advances have included the development of large filled apertures such as the NRAO 300-foot telescope and the Arecibo Observatory, and especially the great synthesis-imaging instruments of the world, the Westerbork Synthesis Radio Telescope and the Very Large Array.
We have been mapping galaxies in HI at ever-increasing resolution and sensitivity for nearly 50 years, but as far as I know no one has seriously asked the question: Where does the HI come from? This question is likely to seem naive whether you are a theorist or an observer, but probably for diametrically-opposed reasons! If you are an observer who studies HI in galaxies with the VLA, as most of us here today are, then you probably assume that the HI is in some sense primordial, i.e. it was present in the galaxy before any stars were formed, and the stars formed from it, and anyone who would question that point of view must be out of touch with the mainstream. On the other hand, if you are a theorist who studies the physical state of the ISM, you know that the time scales for the formation and destruction of H2 in the ISM are so short that essentially all of the HI in a galaxy has been in the form of H2 many times during a galaxy lifetime, and to ask where the HI comes from in a galaxy is a naive question similar to ``which comes first, the chicken or the egg?''
From the point of view of the basic physics, two HI atoms will have a lower energy state if they can get together and make an H2 molecule. A physicist would therefore expect that, if there is a channel for this interaction to get rid of its binding energy and angular momentum, and if there is enough time, in the absence of a source of continual energy input the gas would all be in the form of H2. In pure HI gas, there is a small but non-zero probability that a collision of two HI atoms will indeed form an H2 molecule by emission of quadrupole radiation. This reaction is favored at low temperatures and high volume densities. The re-formation rate can be increased by many orders of magnitude if dust grains are present in the gas. H2 molecules are destroyed by UV photons, creating two HI atoms by photodissociation. The equilibrium state of the gas then depends on the balance between dissociation and re-formation, and the relative amounts of N(H2) / N(HI) may vary considerably over a galaxy. Indeed, we may expect to find both H2 and HI in various amounts near any and all sources of far-UV photons. What observational evidence is there for this association in galaxies?
Photodissociation regions (PDRs) have been identified in the Galactic ISM both in the general diffuse gas and in dense regions near hot young stars. Theoretical work on PDRs has been stimulated by recent satellite observations (e.g. COBE, ISO, and SWAS), and detailed models exist especially for the dense surfaces of molecular clouds (102 - 107 cm-3) which are illuminated by intense far-UV fluxes (100 - 106 times the local average interstellar radiation field (ISRF) near the Sun). Hollenbach & Tielens (1999) have recently written an excellent review on this subject with many references. By now we can say that the basic physics is well understood, and calculations of infrared line ratios and even line intensities are quite successful at accounting for the observations under a wide range of physical conditions in the Galactic ISM.
Perhaps less well known is the fact that the same physics used to calculate the mid-IR lines of H2 from the ISM also provides a straightforward way to calculate the amount of HI resulting from dissociation of the H2 molecules under the action of the same UV photons which provide the mid-IR excitation. However, an important difference between the mid-IR and 21-cm radio observing techniques now comes into play; in the mid-IR, observational selection favors high-density, high-UV-flux situations which lead to high surface brightness. But measurements of the 21-cm HI line favor the lower-density, lower-UV-flux situations, since in these cases the HI is spatially more extended and the observational beam filling factors are therefore larger, making the emission easier to detect with radio telescopes.