2 Observation and analysis

An ASCA observation of GRB 970828 in the 0.5-10 keV range as a Target of Opportunity Observation (TOO) was conducted soon after the alerts from the RXTE team based on the ASM/RXTE detection. The observation with net exposure time of $\sim 36$ ksec was performed during a period of August 29.91 - August 30.85 UT beginning at 1.17 days after the GRB. Both two scientific instruments on ASCA, the SIS and the GIS detector, saw an X-ray source of an average 2-10 keV flux of $F_{\rm X}\sim4\ 10^{-13}$ erg s^-1 cm^-2 within the combined source error region given by ASM/RXTE and IPN (Smith et al. 1997; Hurley et al. 1997; Murakami et al. 1997). The position of the X-ray source was determined at $\rm R.A.= 18^h08^m32\hbox{$.\!\!^{\rm s}$}2$ and $\rm Decl.= +59^{\circ}18'54''$ (J2000) with a 90% error radius of $0\hbox{$.\mkern-4mu^\prime$}5$ by an image analysis on the datasets.

2.1 X-ray flux variability

The X-ray source clearly displayed a fading behavior during the observation. The flux in 2-10 keV range estimated assuming a power-law spectrum with the photon index of -2 decayed from $\sim 1\ 10^{-11}$ ergs cm^-2 s^-1 (the PCA observation at $t \sim 1.4 \ 10^4$ s) to $\sim 3\ 10^{-13}$ ergs cm^-2 s^-1 (at the end of the ASCA observation at $t \sim 1.7 \ 10^5$ s) where t is an elapsed time since the burst. However,the ASCA data indicate a variability of the source rather than a simple power-law like decay. Figure 1 shows the SIS count rate together with the "best-fit'' power-law fading model: $t^{-1.44 \pm 0.07}$ (solid line) based on the PCA/RXTE (Marshall et al. 1997) and the ASCA observation. A peak structure appears around $t=1.25\ 10^5$ s.

  

Figure 1: SIS light curve with time bins of 4000 s. It displays the time variation of the source brightness. A simple power-law decay model cannot fit the data; the solid line shows the "best-fit'' model. Dashed lines labeled "A'', "B'' and "C'' indicate the intervals for which spectral studies are done

2.2 Afterglow spectra

Spectra of the source were made for three periods during the observation indicated by dashed lines labeled "A'', "B'' and "C'' in Fig. 1. We made fittings with a model "absorbed power-law'' ($\exp(- N_{\rm H} \sigma(E)) \ast E^{-\alpha}$), and found a large absorption column density, $N_{\rm H}$ in the best fit for the period "C''. This is $4.1^{\,+2.1}_{\,-1.6}\ 10^{21}$ cm^-2 which is larger than the galactic value towards the source direction, $3.4\ 10^{20}$ cm^-2, estimated by COLDEN at The Einstein On-Line Service, Smithsonian Astrophysical Observatory. For other periods, however, derived column densities are consistent with the galactic values.

Another remarkable result is an excess feature around 5 keV in the spectrum of "B'' which can be represented by a gaussian line (see Fig. 2). Introducing a line improves the fitting by $\Delta \chi^2 = 11.15$ with three additional line parameters, and results in $\chi^2$ of 58.85 for 58 dof. An F-test gives the confidence of 98.3%. The best fit line centroid is $E_{\rm line} = 5.04^{\,+0.23}_{\,-0.31}$keV, the line width $\sigma = 0.31^{\,+0.38}_{\,-0.31}$keV, and the line flux $F_{\rm Fe} = (1.5\pm0.8)\ 10^{-13}$ ergs cm^-2 s^-1. While, for the spectra of "A'' and "C'', the model including a line results in its normalization consistent with zero. The upper-bounds (90% confidence) of line flux are $9.5\ 10^{-14}$ ergs cm^-2 s^-1 and $5.6\ 10^{-14}$ ergs cm^-2 s^-1 for "A'' and "C''respectively. This may imply that the line intensity changed during the observation.

  

Figure 2: The X-ray spectrum for the period (B) shown in Fig. 1 fitted with the model of "absorbed power-law'' ($\exp(- N_{\rm H} \sigma(E)) \ast E^{-\alpha}$)