2021A&A...653A..83V

Aims. We present the updated census and statistics of Lyman-α emitting long gamma-ray burst host galaxies (LAE-LGRBs). We investigate the properties of a subsample of LAE-LGRBs and test the shell model that is commonly used to fit Lyman-α (Lyα) emission line spectra.
Methods. We considered all LGRB host galaxies with relevant publicly available information. We defined a golden sample of four LAE-LGRBs (GRBs 011211, 021004, 060926, and 070110) with afterglow and host galaxy observations allowing us to constrain the shell modeling of the Lyα line.
Results. The census results in 29 detected LAE-LGRBs. We present 5 new Lyα emission detections in host-galaxy spectra and the corresponding unpublished VLT/X-shooter data (GRBs 060926, 070110, 081121, 081222, and 100424A). From the comparison of the statistics and properties of LAE-LGRBs to those of LAE samples in the literature, we find evidence of Lyα suppression in dusty systems. The fraction of LAE-LGRBs in the overall LGRB hosts is lower than that found for Lyman-break galaxy (LBG) samples at similar redshift range. This result can arise because the selection criteria of the parent samples are different and the spectral observations of LGRB samples are shallower than those of LBG. However, we find that LAE-LGRBs are representative of Lyα emission from the bulk of UV-selected galaxies at z∼2. We find that the golden sample of LAE-LGRBs we studied consists of complex systems characterized by multiple emission blobs and by signs of possible galaxy interactions. The fitting procedure recovers the HI column densities (N_HI) measured from the afterglow spectra and the other properties described by the shell-model parameters in the two low-N_HI cases, but it fails to do this in the other two cases with high N_HI. The afterglows of most LGRBs and LAE-LGRBs show log(N_HI/cm^–2)>20.3, implying that statistically, the bulk of Lyα photons that is expected to be produced by massive stars in the star-forming region hosting the GRB will be surrounded by these opaque lines of sight. We therefore interpret our results in the context of more sophisticated models and of different dominant Lyα-emitting regions. We also compare LAE-LGRBs to LAE Lyman continuum (LyC) leakers in the literature in terms of the properties that are identified as possible indirect indicators of LyC leakage. We find that only one LGRB (GRB 021004) would likely be a strong LyC leaker and discuss the validity of these indicators at high redshift. While our work shows that LGRBs are useful tools for probing LAEs and radiative transfer models, larger statistics are required to strengthen our findings.