The abundance of oxygen in galaxies is widely used in furthering our understanding of galaxy formation and evolution .
Unfortunately , direct measurements of O/H in the neutral gas are extremely difficult to obtain as the only O i line available within the HST UV wavelength range ( 1150–3200 Å ) is often saturated .
As such , proxies for oxygen are needed to indirectly derive an O/H via the assumption that solar ratios based on local Milky Way sight lines hold in different environments .
In this paper we assess the validity of using two such proxies , P ii and S ii , within more typical star-forming environments .
Using HST-COS FUV spectra of a sample of nearby star-forming galaxies and the oxygen abundances in their ionized gas , we demonstrate that both P and S are mildly depleted with respect to O and follow a trend , \log ( P ii /S ii ) = -1.73 \pm 0.18 , in excellent agreement with the solar ratio of \log ( P/S ) _ { \odot } = -1.71 \pm 0.04 over the large range of metallicities ( 0.03–3.2 Z _ { \odot } ) and H i column densities ( \log [ N ( H i ) /cm ^ { -2 } ] = 18.44–21.28 ) spanned by the sample .
From literature data we show evidence that both elements individually trace oxygen according to their respective solar ratios across a wide range of environments .
Our findings demonstrate that the solar ratios of \log ( P/O ) _ { \odot } = -3.28 \pm 0.06 and \log ( S/O ) _ { \odot } = -1.57 \pm 0.06 can both be used to derive reliable O/H abundances in the neutral gas of local and high-redshift star-forming galaxies .
The trends in the difference between O/H in the ionized- and neutral-gas phases versus metallicity and H i content are consistent with primordial gas accretion , with galactic outflows and/or star-formation inefficiency also playing a role in the most metal-poor galaxies .