On the spectral lags of the short gamma-ray bursts

We present a detailed analysis on the spectral lags of the short gamma-ray bursts (GRBs) and compare them with that of the long GRBs by using the CGRO (Compton Gamma-ray Observatory)/BATSE GRB catalogue. Our sample includes 308 short GRBs and 1008 long GRBs. The light curves of these GRBs are in 64-ms time bin and they have at least one long and intense pulse, which satisfies delta T >= 0.512 s at c= 1 sigma and c(max)>= 6 sigma, where delta T is the pulse duration, c is the photon counts and sigma is the standard error of the background. We calculate the cross-correlation function (CCF) for the light curves in 25-55 and 110-300 keV bands and derive the spectral lag by fitting the CCF with the Gaussian model. Our results are as follows. (i) The spectral lag distribution of the short GRBs is significantly different from that of the long GRBs. Excluding the statistical fluctuation effect, a proportion of similar to 17 per cent of the short GRBs has a negative spectral lag, i.e. the hard photons are being lag behind the soft photons. We do not find any peculiar features from their light curves to distinguish these bursts from those with a positive spectral lag. We argue that a more physical mechanism dominated the hard lag may be hid behind the morphological features of the light curves. This should be a great challenge to the current GRB models. We note that this proportion is consistent with the proportion of short GRBs correlated with nearby galaxies newly discovered by Tanvir et al., although it is unclear if these short GRBs are indeed associated with the sources originated at low redshift. (ii) While the spectral lags of the long GRBs are strongly correlated with the pulse durations, they are not for the short GRBs. However, the ratios of the spectral lag to the pulse duration for the short and long GRBs are normal distributions at 0.023 and 0.046, respectively, with the sample width, indicating that the curvature effect alone could not explain the difference of the spectral lags between the two types of GRBs. The hydrodynamic time-scales of the outflows and the radiative processes at work in GRBs might also play an important role as suggested by Daigne and Mochkovitch.