A method for deriving accurate gas-phase abundances for the multiphase interstellar galactic halo

We describe a new method for accurately determining total gas- phase abundances for the Galactic halo interstellar medium with minimal ionization uncertainties. For sight lines toward globular clusters containing both ultravioletbright stars and radio pulsars, it is possible to measure column densities of H I and several ionization states of selected metals using ultraviolet absorption line measurements and of H ii using radio dispersion measurements. By measuring the ionized hydrogen column, we minimize ionization uncertainties that plague abundance measurements of Galactic halo gas. We apply this method for the first time to the sight line toward the globular cluster Messier 3 [(l,b) (42.degrees 2, + 78.degrees 7), d 10.2 kpc, z 10.0 kpc] using Far Ultraviolet Spectroscopic Explorer and Hubble Space Telescope ultraviolet spectroscopy of the post - asymptotic giant branch star von Zeipel 1128 and radio observations by Ransom et al. of recently discovered millisecond pulsars. The fraction of hydrogen associated with ionized gas along this sight line is 45% +/- 5%, with the warm (T similar to 104 K) and hot (T >= 10(5) K) ionized phases present in roughly a 5: 1 ratio. This is the highest measured fraction of ionized hydrogen along a high-latitude pulsar sight line. We derive total gas-phase abundances log N( S)/N( H) -4.87 +/- 0.03 and log N( Fe)/N( H) similar to 5.27 +/- 0.05. Our derived sulfur abundance is in excellent agreement with recent solar system determinations of Asplund, Grevesse, & Sauval. However, it is - 0.14 dex below the solar system abundance typically adopted in studies of the interstellar medium. The iron abundance is similar to-0.7 dex below the solar system abundance, consistent with the significant incorporation of iron into interstellar grains. Abundance estimates derived by simply comparing S II and Fe II to H I are + 0.17 and + 0.11 dex higher, respectively, than the abundance estimates derived from our refined approach. Ionization corrections to the gas- phase abundances measured in the standard way are, therefore, significant compared with the measurement uncertainties along this sight line. The systematic uncertainties associated with the uncertain contribution to the electron column density from ionized helium could raise these abundances by less than or similar to 0.03 dex (+ 7%). Uncertainties in the amount of very hot gas (T similar to 10(6) K) along the line of sight could also affect these determinations.