next previous
Up: GRANAT/SIGMA observation of GRB


3 Results and discussion


Figure 1 shows the burst light curve in 35-300 keV band with the time resolution $\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$<$}}}0.1$ s. The zero time is at the burst trigger. Figures 2 and 3 show the burst light curve in logarithmic coordinates of both time and flux. The light curve shape in the logarithmic coordinates strongly depends on the choice of the reference time. In Fig. 2, the reference time is chosen at the moment of the burst trigger (Fig. 1). The horizontal dotted line represents a $95\%$ upper limit on the possible internal background variations on the 300 s time scale. There appears to be a power-law decay of flux starting after 10-20 s after the trigger. This behavior is consistent with GRB entering the stage of self-similar fireball expansion soon after the main burst.

The solid line in Fig. 2 shows the power law fit in the time interval 20-1000 s, and the dash-dotted line shows the exponential fit. Power law provides a better description of the data than the exponential decay -- $\chi^2$ is 1.5 and 6.5 per 4 dof, respectively. The best fit power law index is $-0.69\pm0.17$ ($\Delta\chi^2=2.7$ and 5.7 for the indices -1 and -1.2) which is considerably flatter than BeppoSAX slopes (e.g. Costa et al. 1998). This power law tail contains at least ${\sim}20\%$ of the main burst fluence. Interestingly, the extrapolation of the power law (shown by the dotted line in the Fig. 3) points to the small peak near the beginning of the main burst (see also Fig. 1).

Figure 3 shows the burst light curve with the reference time chosen at the moment when the burst flux began the gradual decline, approximately 6 s after the trigger (Fig. 1). With this choice of zero time, the data in the 0.01-20 s time interval lies on the extrapolation of the above power law fit; adding these data to the fit results in the power law index $-0.70\pm0.03$. Note, in this case, the power law flux decay lasts over approximately four orders of magnitude of time.

  
\begin{figure}

\includegraphics [width=8.8cm,clip]{R93f3.ps}\end{figure} Figure 3: Same as Fig. 2, but the reference time was set at 6 s after the trigger. The main burst is not shown here because it is at t<0 with this choice of reference time

The spectral evolution of the burst flux can be characterized by the ratio of WATCH flux in the 8-20 keV band (Terekhov et al. 1993) and SIGMA flux in the 75-200 keV band[*]. During the main burst (0-6 s after the trigger), the WATCH/SIGMA flux ratio corresponds to the power law energy spectral index $0.05\pm0.04$. After 6s, we observe a much softer spectrum -- spectral index is $1.06\pm0.08$ for $6<t<7\,$s, $1.08\pm0.23$ for $8<t<16\,$s, and $1.01\pm0.25$ for $16<t<32\,$s. This drastic spectral change supports the idea that afterglow starts at $t\simeq 6$s after the trigger.

SIGMA data provides the first convincing observation of the power law afterglow in the soft gamma-rays and immediately after the burst. It fills the gap between BeppoSAX WFC and NFI observations. A very important issue is whether such afterglows are common to GRB. A preliminary analysis of other SIGMA bursts revealed no other convincing afterglows, primarily because of the faintness of other bursts; on the basis of SIGMA data alone, we cannot rule out that the soft gamma-ray afterglow is a common phenomenon. A preliminary analysis of the PHEBUS data confirms the detection of the afterglow in GRB 920723 and reveals a similar afterglow in GRB 910402 (Tkachenko et al. 1998). The results of our systematic search for soft gamma-ray afterglows in the GRANAT data will be presented in the future.

Acknowledgements

This work was supported by RBRF grants 96-02-18458 and 96-15-96930.


next previous
Up: GRANAT/SIGMA observation of GRB

Copyright The European Southern Observatory (ESO)