STAR FORMATION AND THE DISTRIBUTION OF H-I AND INFRARED-EMISSION IN M51

H I, infrared, CO, H-alpha, and B-band observations of M51, the prototypical grand design spiral galaxy, are used to study the consequences of star formation for the distribution of H I and dust. Using the new VLA map of 21 cm emission, our OVRO CO mosaic map, and an H-alpha image, we carry out new tests of the idea of Tilanus & Allen that the H I is largely a photodissociation product in star-forming regions. We confirm that the H I spiral arms are generally coincident with the H II region arms and offset downstream from the CO arms. The radial distributions of total gas, H-alpha, and H I surface density have a simple explanation in the dissociation picture. The distributions also demonstrate how the surface density of H I might be related to the star formation efficiency in molecule-rich galaxies. The large width of the H I regions along the arms compared to that of the giant H II regions can be understood in terms of a simple calculation of the expected size of an H I region associated with a typical giant H II region. The longer lifetime of the stars producing dissociating radiation versus those producing ionizing radiation and the relatively long molecule formation time scale will also contribute to the greater width of the H I arms if stars are continuously forming on the arms. The lack of detailed coincidence of the H I and H II regions along the inner arms has a variety of possible explanations within the dissociation scenario. We carry out two simple tests to probe the origins of the IRAS emission in M51. First, we find that the infrared excess (IRE) of M51 is 24, suggesting that a substantial fraction of the infrared emission arises from dust heated by photons which do not originate in massive star-forming regions. Second, radial cuts through the IRAS maps show that at 12, 25, and 60-mu-m, the arm-interarm contrast of the IRAS emission is substantially less than that of the H-alpha emission (convolved to matching resolutions), providing further evidence for the above explanation and for the existence of a broadly distributed dust component. Deconvolved IRAS maps have improved resolution but do not change this finding.