Information about elemental abundances in halo gas can provide important clues about the origin of the gas. For a review of the abundances in neutral halo gas see Jenkins (1986). If the halo gas is supplied from the disk, its abundances should approximate those found in disk gas provided allowances are made for the effects of heavy-element depletion onto dust grains. If the halo gas is supplied from intergalactic space, its abundances may reflect those found in relatively unprocessed matter. Type I supernova explosions occurring in the halo might provide for the selective enrichment of particular elements in the gas. The amount of mixing between disk and halo gas will influence the elemental abundances in halo gas. This mixing involves not only the vertical mixing, i.e. mixing in the z direction, but also the radial mixing (if it occurs) since galaxies are observed to have radial abundance gradients. The possible presence of dust introduces a major complication in interpreting abundance data for elements measured in the gas phase. For elements such as Fe, Ti, Ca and Al, the gas phase depletion for normal diffuse cloud matter which contains dust ranges from factors of 50 to 1000. Therefore, even a very modest amount of grain destruction can have an enormous effect on the measured gas phase abundances.
TiII is the best ion available to ground based optical telescopes for the study of the abundance of a normally highly depleted element. This is because the TiII lines are usually weak and therefore suitable for deriving accurate column densities provided the lines can be detected. Furthermore, TiII is the dominant ionization state of Ti in neutral hydrogen clouds and therefore the gas phase abundance relative to HI follows directly from a comparison of the column density ratio N(TiII) / N(HI). The TiII studies of Stokes (1978) and Albert (1983) reveal that although the gas phase TiII abundance is typically 10-2 solar in low velocity neutral disk gas, its abundance often equals and sometimes exceeds 10-1 solar in high velocity clouds which are found either in the galactic disk or in the galactic halo. Similar studies with the CaII lines also reveals a shift toward more gaseous CaII in high velocity and/or halo clouds [see Section 2.2, Albert (1983) and Figure 4]. However, it is more difficult to convert the CaII data to abundance relative to HI because of uncertainties associated with the ionization equilibrium between CaII and CaIII.
The study of elemental abundances in halo gas in the ultraviolet with the IUE satellite has been limited by the low signal to noise of typical data and by the low resolution of the IUE spectrograph for interstellar studies. For the intermediate velocity gas toward the LMC (which may or may not be associated with the Milky Way) Savage and de Boer (1981) performed a curve of growth study and obtained column density estimates or limits for neutral O and singly ionized Mg, Al, Si and Fe. When these results are combined with high sensitivity 21 cm radio data of HI emission at similar velocities from McGee, Newton and Morton (1983), the data indicate that the composition is within a factor of three of solar. The most detailed halo gas abundance study based on ultraviolet data has been for the high velocity gas toward the bright halo star HD93521 observed by the Copernicus satellite (Caldwell 1979). Although this star is in the low halo (|z| 0.7 kpc), the elemental abundance results are informative. In the clouds at vLSR < -10 km s-1, the abundances of Fe, Si, Ar, and P are found to approach those found in the sun. This result for HD93521 is also confirmed for Ti in the ground based TiII measurements of Albert (1983).
Collectively, the available gas phase abundance information suggests that the gas in the low and perhaps distant halo has abundances which in some cases approach those found in the sun. This is in marked contrast to the highly depleted gas phase abundances usually found for disk gas. The most likely explanation for these results is that the neutral gas in the halo exists in regions relatively devoid of solid matter. The process(es) which inject matter into the halo therefore seem capable of destroying dust or of preventing dust from forming. Alternately it may be that dust destruction proceeds much more rapidly in the lower density gas of the halo. Finally, because of the additional complications involved with understanding the effects of the gas/dust interactions the existing data do not rule out the possibility that certain elements (e.g. Fe) are selectively enriched in the halo through Type I supernova explosions (Jenkins 1983).