Large O vi columns are observed around star-forming , low-redshift \sim L ^ { * } galaxies , with a dependence on impact parameter indicating that most { O } ^ { 5 + } particles reside beyond half the halo virial radius ( \gtrsim 100 { kpc } ) .
In order to constrain the nature of the gas traced by O vi , we analyze additional observables of the outer halo , namely H i to O vi column ratios of 1 - 10 , an absence of low-ion absorption , a mean differential extinction of E _ { B - V } \approx 10 ^ { -3 } , and a linear relation between O vi column and velocity width .
We contrast these observations with two physical scenarios : ( 1 ) O vi traces high-pressure ( \sim 30 { cm } ^ { -3 } { K } ) collisionally-ionized gas cooling from a virially-shocked phase , and ( 2 ) O vi traces low-pressure ( \lesssim 1 { cm } ^ { -3 } { K } ) gas beyond the accretion shock , where the gas is in ionization and thermal equilibrium with the UV background .
We demonstrate that the high-pressure scenario requires multiple gas phases to explain the observations , and a large deposition of energy at \gtrsim 100 { kpc } to offset the energy radiated by the cooling gas .
In contrast , the low-pressure scenario can explain all considered observations with a single gas phase in thermal equilibrium , provided that the baryon overdensity is comparable to the dark-matter overdensity , and that the gas is enriched to \gtrsim { Z _ { \odot } } / 3 with an ISM-like dust-to-metal ratio .
The low-pressure scenario implies that O vi traces a cool flow with mass flow rate of \sim 5 { M _ { \odot } } { yr } ^ { -1 } , comparable to the star formation rate of the central galaxies .
The O vi line widths are consistent with the velocity shear expected within this flow .
The low-pressure scenario predicts a bimodality in absorption line ratios at \sim 100 { kpc } , due to the pressure jump across the accretion shock .