The Epeak-Eiso plane of long gamma-ray bursts and selection effects

We study the distribution of long gamma-ray bursts in the E-peak-E-iso and in the E-peak(obs)-fluence planes through an updated sample of 76 bursts, with measured redshift and spectral parameters, detected up to 2007 September. We confirm the existence of a strong rest-frame correlation E-peak proportional to E-iso(0.54 +/- 0.01). Contrary to previous studies, no sign of evolution with redshift of the E-peak-E-iso correlation (either its slope and its normalization) is found. The 76 bursts define a strong E-peak(obs)-fluence correlation in the observer frame (E-peak(obs) proportional to F-bol(0.32 +/- 0.05)) with redshifts evenly distributed along this correlation. We study possible instrumental selection effects in the observer frame E-peak(obs)-fluence plane. In particular, we concentrate on the minimum peak flux necessary to trigger a given gamma-ray burst (GRB) detector (trigger threshold) and the minimum fluence a burst must have to determine the value of E-peak(obs) (spectral analysis threshold). We find that the latter dominates in the E-peak(obs)-fluence plane over the former. Our analysis shows, however, that these instrumental selection effects do not dominate for bursts detected before the launch of the Swift satellite, while the spectral analysis threshold is the dominant truncation effect of the Swift GRB sample (27 out of 76 events). This suggests that the E-peak(obs)-fluence correlation defined by the pre-Swift sample could be affected by other, still not understood, selection effects. Besides, we caution about the conclusions on the existence of the E-peak(obs)-fluence correlation based on our Swift sample alone.