At z < 1 a large fraction of the baryons is thought to reside in diffuse gas that has been shock-heated to high temperatures ( 10 ^ { 5 } -10 ^ { 6 } \leavevmode \nobreak { K } ) .
Absorption by the 770.41 , 780.32 Å doublet of NeVIII in quasar spectra represents a unique tool to study this elusive warm-hot phase .
We have developed an analytic model for the properties of NeVIII absorbers that allows for an inhomogeneous metal distribution .
Our model agrees with the predictions of a simulation from the OWLS project indicating that the average line-of-sight metal-filling fraction within the absorbing gas is low ( c _ { L } \sim 0.1 ) .
Most of the NeVIII in our model is produced in low-density , collisionally ionised gas ( { { n } _ { H } } = 10 ^ { -6 } -10 ^ { -4 } \leavevmode \nobreak { cm ^ { -3 } } , T = 10 ^ { 5 } -10 ^ { 6 } \leavevmode \nobreak { K } ) .
Strong NeVIII absorbers ( \log _ { 10 } ( { N _ { \hbox { \scriptsize Ne { \tiny VIII } } } } / { cm ^ { -2 } } ) \gtrsim 14 ) , like those recently detected by HST/COS , are found to arise in higher density gas ( { { n } _ { H } } \gtrsim 10 ^ { -4 } { cm ^ { -3 } } , T \approx 5 \times 10 ^ { 5 } \leavevmode \nobreak { K } ) .
NeVIII cloudlets harbour only 1 per cent of the cosmic baryon budget .
The baryon content of the surrounding gas ( which has similar densities and temperatures as the NeVIII cloudlets ) is a factor c _ { L } ^ { -1 } higher .
We conclude that NeVIII absorbers are robust probes of shock-heated diffuse gas , but that spectra with signal-to-noise ratios { \leavevmode \nobreak S / N } > 100 would be required to detect the bulk of the baryons in warm-hot gas .