Discussion

Our results show a complex behavior of SGR 1900+14 in outburst that can be resumed in the following points.

(1) A clear determination of the 5.16 s periodicity from the onset of the outburst. We note that such a feature is not detected by the Ulysses experiment suggesting that the capability to detect the periodicity during this part of the outburst is related to the bandpass, similarly to the fact that GRBM did not detect the precursor, while Ulysses did. It is also interesting that we detect a period consistent with the period measured with ASCA within the uncertainty in the period determination (0.02 s), implying that such a large outburst caused a glitch with $\Delta P/P < 3.1 \times 10^{-3}$.

(2) We see a clear transition after $\sim$35 s in the shape of the 5.16-s pulse profile. At the beginning the pulse is composed of two broad pulses, with relative intensity gradually changing from one period to the next one. At the 6-th pulse from the onset of the outburst the two peaks become sharper and gradually two additional narrow peaks appear. A striking feature of these 4 pulses is their equal mutual distance in time and their phase stability within the 5.16-s period. This feature is apparent in the PSD as a net enhancement of the n=5 harmonic with respect to the others, corresponding to a periodicity of $(1.03\pm0.03)$ s. It looks like the 5.16-s periodicity in this part of the outburst is basically due to the lack/occultation of the 5-th peak that one would have expected in correspondence of the 5.16-s dip. The occurrence of such a 5-th peak would have resulted in the disappearance of the 5.16-s periodicity.

(3) The determination of a complex spectral evolution showing a very hard initial outburst pulse followed by softer emission (parameterized with an OTTB with kT$\simeq 30$ keV) that is strongly modulated in agreement with the periodic pulse structure. We notice a `see-saw' behavior of the HR, with the highest hardness in correspondence of the dips of the light curve. If the dip would be due to an occultation this would imply that a persistent hard component is present in the emitted spectrum. A hard component is also suggested by the persistence for hundreds of seconds of a PL spectral component additional to the single OTTB (see Fig. 5.21, top and middle panel).

(4) Finally, we note that the decay of the event light curve as detected by the GRBM is well approximated by a double exponential law and cannot be described by a power law $I(t) \propto t^{-\alpha}$as derived for the Ulysses light curve. Also this difference is likely to be ascribed to the different Ulysses-BeppoSAX/GRBM band passes.

The emerging picture is that of a complex phenomenon, involving a highly non-trivial response of a compact object, most likely a highly magnetized neutron star with a structured emitting region, to a major explosive event. Our point no. 2 is worth of special attention. The emitting region surrounding the compact object is first subject to a violent readjustment following the initial strongly super-Eddington outburst (possibly related to a relativistic ejection producing transient radio emission). Subsequent evolution and settling of the 1-s periodic feature reveals a highly structured emitting region with the excitation of higher order, possibly resonant oscillating modes. In particular, the strong enhancement of the fifth harmonic that we observed is unprecedented among SGR strong outburst detections. Trapping in a magnetosphere of emitting `blobs' co-rotating with the surface of a neutron star, or having their proper oscillation mode, might explain the harmonic content of the time variable power spectrum of SGR 1900+14.

  

Figure 5.20: Top: 40-700 keV 1-s background subtracted light curve of the event (a value of 1000 counts/s has been added for display purposes). The dashed vertical lines define the intervals A, B and C for which we have the time averaged energy spectra. Bottom: (a) high resolution light curve (rebinned at 31.25 ms) of the available portion of the event ($\sim$100 s); (b) 1-s resolved spectral evolution described in terms of the simple hardness ratio HR=(100-700 keV)/(40-100 keV) and (c) in terms of an equivalent kT of an optically thin thermal bremsstrahlung. 1-$\sigma$ errors are shown. The vertical dotted lines are spaced by one spin period of the neutron star

  

Figure 5.21: Fit to the GRBM spectrum (the individual spectral components are shown as dashed lines) in the energy range 70-650 keV (A) and 70-400 keV (B and C) with a model including: (A) An Optically Thin Thermal Bremsstrahlung (OTTB, kT=31 keV) plus a Power Law (PL, $\alpha$=1.47) for the time interval from 0 to 68 s with respect to the trigger time; (B) OTTB plus PL with kT=28 keV and $\alpha$=4.5 for the time interval from 68 to 195 s; (C) OTTB with kT=29 keV for the time interval from 196 to 323 s.