We have argued that the dust-laden clouds identified in the thick disks of spiral galaxies through direct optical imaging represent a cold, dense phase of the thick disk ISM (Howk & Savage 1999, 2000). The derived column densities and particle densities are consistent with this picture. The high column densities we derive for many individual structures are high enough that they would have molecular fractions of > 25% if the physical conditions (i.e., dust content and radiation fields) are similar to those found in the disk of the Milky Way. There are other observational indications that, sometimes indirectly, lead us to believe the thick disks of galaxies contain a CNM that is traceable through its extinction.
First, it is clear that the interstellar thick disks of spiral galaxies are multiphase media. The observations of gas at temperatures that span several orders of magnitude (e.g., through observations of H and X-ray emission) in the thick disks of galaxies imply the presence of distinct interstellar phases. Howk & Savage (2000) have compared high resolution (sub-arcsecond) imaging of H from the extraplanar DIG and of absorption due to the extraplanar dust in NGC 891. This comparison revealed not only that the structures seen in dust and H are not physically related, but that the WIM is much more smoothly distributed than the material traced by dust extinction. This almost certainly implies the volume filling factor of the material traced by extraplanar dust is much smaller than that of the DIG. Rossa and collaborators (this proceedings; Rossa et al. 2004) have recently presented even higher-resolution images of NGC 891 acquired with HST that lead to the same conclusion: the dust and the ionized gas occupy separate regions of space with very little correspondence and, therefore, represent distinct phases of the multiphase thick disk in this galaxy.
There is also some evidence for CO emission, a direct tracer of CNM material, in the thick disks of spiral galaxies. García-Burillo et al. (1999) have presented CO observations of the edge-on galaxy NGC 4013. Their interferometric maps reveal the presence of CO-bearing extraplanar filaments, some of which are coincident with extraplanar dust seen by Howk & Savage (1999). This is a direct indication of the presence of a CNM in the thick disk of this galaxy. Unfortunately, few galaxies have been observed with the sensitivity and resolution required to detect these structures. NGC 891 has been mapped in CO, although the picture in this galaxy is far from clear: single-dish (García-Burillo et al. 1992) and interferometric (Scoville et al. 1993) observations give conflicting results. We are pursuing deep interferometric CO mapping of this galaxy to limit the amount of CO in the thick disk of this system.
Perhaps the least direct, but most interesting, indicator for the presence of a CNM in the thick disks of galaxies is the recent evidence for thick disk star formation in spiral galaxies. Several authors have noted the presence of extraplanar HII regions in the thick disks of galaxies (Walterbos 1991; Ferguson, Wyse, & Gallagher 1996; Howk & Savage 1997, 2000). These regions are too far from the midplanes of the host galaxies for the OB associations required to ionize the gas to have been born in the disk and subsequently ejected. The study of the abundances and stellar content of such regions has just begun (Tüllman et al. 2003) and may give us important information on the circulation of metals within the thick disks of galaxies. Indirectly, the presence of newly-formed hot stars in the thick disks implies the presence of a CNM, since the latter is a crucial ingredient for star formation. We consider it likely that the young stars have formed from clouds similar to those seen in our images.