3 Highlight results

Some relevant results of our study are given below. An exhaustive description of our investigation along with complete results will be published elsewhere (Frontera et al. 1999).

  

Figure: Behaviour of the measured peak energy of the $\nu F(\nu)$spectrum as a function of the time from the onset of GRB 970111. Also shown is the expected behaviour of $E_{\rm p}$, normalized to the first value firmly established, in the case of an adiabatic (continuous line) and a radiative (dashed line) cooling of the shock, according to the model by Sari et al. (1998). The dashed verical line corresponds to the time at which the afterglow is expected to start (see text)

3.1 Test of the synchrotron shock model

The GRB spectra in our sample can be described by the Band's law (Band et al. 1993), that is a good approximation of the spectrum expected by a synchrotron shock model (Tavani 1996). We do not find low energy excesses that are not in agreement with the Band's law. The low energy photon index $\alpha$, below the peak energy $E_{\rm p}$ of the $\nu F(\nu)$spectrum, obtained from the spectra integrated over the rise time of the GRB in our sample, is below the limit photon index expected (-2/3) by the optically thin synchrotron shock model for 50% of the GRBs. For the other GRBs, this limit index is exceeded in the first few seconds from the GRB onset. In the case of GRB 970111, a higher low energy photon index is found for the entire GRB duration. From the high energy photon indices measured, we derive a spectral indices p of the energy distribution of the electrons accelerated in the shock $(N(E_{\rm e})\propto E_{\rm e}^{-p})$ that generally increase with time. The greatest increase is observed in the case of GRB 970111.

3.2 Correlation between GRB emission and X-ray afterglow

By extrapolating the X-ray afterglow fading law $(F(t) \propto t^{-\delta})$back to the time of the burst, we have already demonstrated that at least in two cases, GRB 970228 (Costa et al. 1997) and GRB 970508 (Piro et al. 1998), the expected 2-10 keV flux is consistent with that measured in the same energy band during the GRB tail. For all GRBs in our sample with well detected X-ray afterglow emission, we find that the 2-10 keV fluence of the GRB tail is consistent with that expected in the same energy band if the fluence is due to afterglow emission. The early afterglow appears to start at about 60% of the GRB duration.

3.3 Hydrodynamical evolution of a fireball shock

We find a continuous decrease with time of the peak energy $E_{\rm p}$ of the $\nu F(\nu)$ spectrum. At very early times from the GRB onset the decrease is faster than that expected in the case of a radiative or adiabatic cooling of an external shock (Sari et al. 1998). This fact could imply that the process that gives rise to the GRB or, at least, to the early phase of the GRB, is different from that that gives rise to the afterglow emission, as discussed by Sari (1997). For GRB 970111, at about 60% of the GRB duration (see dashed vertical line), when the afterglow is expected to start, a very slow decrease of $E_{\rm p}$ with time is observed (see Fig. 1). We find (see continuous line with lower slope) that the latter data points are consistent with a slow adiabatic cooling of an external shock of a fireball with the interstellar medium (Sari et al. 1998). For GRB 970508, combining the $E_{\rm p}$ values measured during the GRB event with that obtained by Galama et al. (1998) 12.1 days after the primary event, a transition from fast cooling to slow cooling is inferred. Assuming an adiabatic cooling, this transition is expected to occur at about 1000 s from the GRB onset. This time value is consistent with that (700 s) estimated by Galama et al. (1998).

Acknowledgements

This research is supported by the Italian Space Agency ASI.