The nature of H i absorbers in gamma‐ray burst afterglows: clues from hydrodynamic simulations

In recent work, we have shown that it is possible to link quantitatively many aspects of damped Lyman α (DLA) absorbers in the spectra of quasars to high‐resolution simulations of galaxy formation. Using runs from the same series of hydrodynamic numerical studies, we consider the expected properties of intrinsic Lyman α absorbers seen in the spectra of high‐redshift (z > 2) gamma‐ray burst afterglows (GRB–DLAs). If GRBs are associated with the death of massive stars, their afterglows provide insights into otherwise unprobed regions of protogalactic objects, but detailed physical interpretations are currently embryonic. We find that median impact parameters (measured from the potential minimum) are approximately 1 kpc for GRBs compared with 4 kpc for quasi‐stellar object–DLA (QSO–DLA). However, an equally important difference is that GRB–DLAs are predominantly associated with haloes of mass 1010 < Mvir/M⊙ < 1012, an order of magnitude larger than the hosts of QSO–DLAs. Accordingly, there are differences in the stellar properties of hosts. For instance, mean star formation rates are higher: for GRB–DLAs compared with for QSO–DLAs. Our simulations accurately predict the form of the GRB–DLA H i column density distribution, producing quantitative agreement for NH i > 1019 cm−2, but they somewhat underpredict the incidence of low column densities NH i < 1019 cm−2. This is reflected in our estimate of the ionizing photon escape fraction, fesc≃ 1 per cent, which is lower than the observational GRB‐derived escape fraction (2 per cent). Line‐of‐sight neutral gas metallicities predicted by our simulations (10−2 < Z/Z⊙ < 1) are consistent with the modest observational constraints. Because of large internal dispersions in gas metallicities, this agreement is not significantly compromised by imposing a cut‐off on the metallicity of stars able to launch GRBs (Z★ < Z⊙/3), confounding claims that the observed metallicity of GRB–DLAs poses a severe challenge to current GRB models.