The metallicity dependence of long-duration gamma-ray bursts

Context . Theoretical models and observations of collapsar-created gamma-ray bursts, typically long-duration gamma-ray bursts (LGRBs), both suggest that these transients cannot occur at high metallicity, likely due to angular momentum losses via stellar winds for potential progenitor stars. However, the precise metallicity threshold (if it is a hard threshold) above which the formation of LGRBs is suppressed is still a topic of discussion. Aims . We investigated observed LGRBs and the properties of their host galaxies to constrain this metallicity dependence. Methods . In order to compute LGRB rates we modelled the cosmic history of star formation as a function of host galaxy metallicity and stellar mass, and added a LGRB efficiency function that can include various shapes including abrupt cutoffs and more gradual variations in the GRB yield with metallicity. In contrast to previous work, this model includes scatters in the relations between mass, metallicity, and star formation rate, as well as a scatter in the metallicity distribution inside galaxies. We then varied both the threshold value and the shape, and compared the results of our model to observed LGRBs and the properties of their host galaxies. Results . In our model a sharp cutoff at an oxygen abundance Z O/H = 12 + log(O/H) = 8.6 ± 0.1 (corresponding to ~0.6 Z ⊙ ) provides the best explanation for the observed LGRB data. In contrast, a lower threshold proposed in the literature (i.e. at Z O/H = 8.3 or ~0.3 Z ⊙ ) fits the observations poorly. Conclusions . We therefore conclude that, in contrast to most theoretical LGRB models, a relatively high metallicity threshold at near solar values provides the best match between our model and observed LGRBs.