We investigate the diagnostic capabilities of the iron lines for tracing the physical conditions of the shock-excited gas in jets driven by pre-main sequence stars .
We have analyzed the 3 000-25 000 Å , X-shooter spectra of two jets driven by the pre-main sequence stars ESO-H \alpha 574 and Par-Lup 3-4 .
Both spectra are very rich in [ Fe ii ] lines over the whole spectral range ; in addition , lines from [ Fe iii ] are detected in the ESO-H \alpha 574 spectrum .
NLTE codes solving the equations of the statistical equilibrium along with codes for the ionization equilibrium are used to derive the gas excitation conditions of electron temperature and density , and fractional ionization .
An estimate of the iron gas-phase abundance is provided by comparing the iron lines emissivity with that of neutral oxygen at 6300 Å .
The [ Fe ii ] line analysis indicates that the jet driven by ESO-H \alpha 574 is , on average , colder ( T _ { e } \sim 9 000 K ) , less dense ( n _ { e } \sim 2 10 ^ { 4 } cm ^ { -3 } ) and more ionized ( x _ { e } \sim 0.7 ) than the Par-Lup 3-4 jet ( T _ { e } \sim 13 000 K , n _ { e } \sim 6 10 ^ { 4 } cm ^ { -3 } , x _ { e } < 0.4 ) , even if the existence of a higher density component ( n _ { e } \sim 2 10 ^ { 5 } cm ^ { -3 } ) is probed by the [ Fe iii ] and [ Fe ii ] ultra-violet lines .
The physical conditions derived from the iron lines are compared with shock models suggesting that the shock at work in ESO-H \alpha 574 is faster and likely more energetic than the Par-Lup 3-4 shock .
This latter feature is confirmed by the high percentage of gas-phase iron measured in ESO-H \alpha 574 ( 50-60 % of its solar abundance in comparison with less than 30 % in Par-Lup 3-4 ) , which testifies that the ESO-H \alpha 574 shock is powerful enough to partially destroy the dust present inside the jet .
This work demonstrates that a multiline Fe analysis can be effectively used to probe the excitation and ionization conditions of the gas in a jet without any assumption on ionic abundances .
The main limitation on the diagnostics resides in the large uncertainties of the atomic data , which , however , can be overcome through a statistical approach involving many lines .