Reconciling escape fractions and observed line emission in Lyman-continuum-leaking galaxies

Context. Finding and elucidating the properties of Lyman-continuum(LyC)-emitting galaxies is an important step in improving our understanding of cosmic reionization.Aims. Although the z similar to 0.3-0.4 LyC emitters found recently show strong optical emission lines, no consistent quantitative photoionization model taking into account the escape of ionizing photons and inhomogenous interstellar medium (ISM) geometry of these galaxies has yet been constructed. Furthermore, it is unclear to what extent these emission lines can be used to distinguish LyC emitters.Methods. To address these questions we construct one- and two-zone photoionization models accounting for the observed LyC escape, which we compare to the observed emission line measurements. The main diagnostics used include lines of [OIII], [OII], and [OI] plus sulfur lines ([SII], [SIII]) and a nitrogen line ([NII]), which probe regions of different ionization in the ISM.Results. We find that single (one-zone) density-bounded photoionization models cannot reproduce the emission lines of the LyC leakers, as pointed out by earlier studies, because they systematically underpredict the lines of species of low ionization potential, such as [OI] and [SII]. Introducing a two-zone model, with differing ionization parameter and a variable covering fraction and where one of the zones is density-bounded, we show that the observed emission line ratios of the LyC emitters are well reproduced. Furthermore, our model yields LyC escape fractions, which are in fair agreement with the observations and independent measurements. The [OI] lambda 6300 excess, which is observed in some LyC leakers, can be naturally explained in this model, for example by emission from low-ionization and low-filling-factor gas. LyC emitters with a high escape fraction (f(esc) greater than or similar to 38%) are deficient both in [OI] lambda 6300 and in [SII] lambda lambda 6716,6731. We also confirm that a [SII] lambda lambda 6716,6731 deficiency can be used to select LyC emitter candidates, as suggested earlier. Finally, we find indications for a possible dichotomy in terms of escape mechanisms for LyC photons between galaxies with relatively low (f(esc)less than or similar to 10%) and higher escape fractions.Conclusions. We conclude that two-zone photoionization models are sufficient and required to explain the observed emission line properties of z similar to 0.3-0.4 LyC emitters. This is in agreement with UV absorption line studies, which also show the co-existence of regions with high hydrogen column density (i.e., no escape of ionizing photons) and density-bounded or very low column density regions responsible for the observed escape of LyC radiation. These simple but consistent models provide a first step towards the use of optical emission lines and their ratios as quantitative diagnostics of LyC escape from galaxies.