Long gamma-ray bursts (LGRBs) are associated to the deaths of massive stars and could thus be used as a potentially powerful tool to trace cosmic star formation. However the conditions needed to produce a LGRBs may introduce a bias in the LGRB rate versus star formation rate (SFR) relation (called LGRB efficiency hereafter).
We have undertaken a study of the properties of the host galaxies of the BAT6 complete sample of LGRB to improve our knowledge on the LGRB efficiency, its redshift evolution, and the factor affecting it. This is the base to properly use LGRBs as SFR tracers.
We show that at z < 1 LGRBs are not direct SFR tracers because they tend to avoid high-metallicity galaxies. The use of the BAT6 complete sample keeps this result from being affected by possible biases that could have influenced past results based on incomplete samples. The preference for low (but not extremely low) metallicities can be a consequence of the particular conditions needed for the progenitor star to produce a GRB.