H. A. Krimm (CRESST/GSFC/USRA), J. P. Norris (U. Denver), T. N. Ukwatta (GSFC/GWU),
S. D. Barthelmy (GSFC), P. A. Evans (U. Leicester), N. Gehrels (GSFC),
M. Stamatikos (GSFC/ORAU)
Using the Swift/BAT data we have completed spectral lag analysis for
the high redshift burst, GRB 090423 (Krimm et al., GCN 9198). Using
16-ms binning on the event data, for the "whole" burst (15 seconds),
chans 3->1: 0.046 +0.085 -0.058 seconds
chans 3->2: 0.044 +0.070 -0.052 seconds
For the peak 3 seconds of the burst we obtain:
chans 3->1: 0.021 +0.054 -0.032 seconds
chans 3->2: 0.006 +0.046 -0.071 seconds
where the channels are 15-25 keV, 25-50 keV, and 50-100 keV. The lag
in each case is consistent with zero, but the 1-sigma error bars are
roughly as large as the median lag values for long bursts. Thus this
burst is too dim for us to utilize lags as a discriminant for long vs.
Given the relatively short duration of this burst, it is informative to
look at other possible indicators of whether the burst is in fact a
short burst seen at a large distance.
The calculations below use z=8, which is roughly the average of the
photometric and spectroscopic redshifts determined for this burst
(Cucchiara et al., GCN 9213; Olivares et al., GCN 9215; Thoene et al.,
GCN 9216; Perley et al., GCN 9217; Tanvir et al., GCN 9219;
Fernandez-Soto et al., GRB 9222).
When converted to the rest frame, the T90 values (10.3 � 1.1 sec,
Swift/BAT 15-350 keV, Palmer et al., GCN 9204) and (12 sec, Fermi/GBM
8-1000 keV, Kienlin, GCN Circ. 9229) transform to 1.1 � 0.1 sec and
1.3 sec, respectively. However, one must be careful in comparing these
numbers to the BATSE short-hard burst divide (Kouveliotou et al.,
ApJ 413, L101, 1993). The BATSE duration distribution is in the observer frame. With a typical redshift of z = 1-2 for BATSE bursts, the dividing
line between long and short in the rest frame is 0.7 to 1.0 seconds.
Thus this burst is on the boundary and toward the long side.
While the duration and lag are consistent with short bursts (with large
errors), there are other observations which are more consistent with
GRB 090423 being a long burst in the source frame.
The spectral fits to the BAT data (Palmer et al., GCN 9204) are
inconclusive as to whether this is a long or short burst. Both the photon
index, alpha = 0.8 +/- 0.5, and Epeak = 440 keV in the source frame, from
a cut-off power-law model fit, are consistent with the results for other
short bursts detected by Swift/BAT. However, these numbers also fall
within the distributions for long bursts (Krimm et al., in preparation).
The main argument against this being a short burst is the isotopic energy.
The analysis performed by Amati et al. (GCN 9227) gives the burst
Eiso=1053 erg and shows that the burst is consistent with the Epeak-Eiso relation for long bursts. All previously known short bursts are outliers
to this relation. The calculated value of Eiso is a factor of > 50 greater than that for most other short bursts.
Furthermore it is questionable whether there would be sufficient time for
a binary neutron star system to form and inspiral to a merger given the
very early time interval since the Big Bang implied by z=8 (lookback time
of ~13 Gyr, corresponding to an age of the universe of 0.6 Gyr).
In conclusion, we can not say if GRB 090423 is short or long. Its
duration in the source frame is at the boundary between the two
classes, the lag analysis is inconclusive, the BAT spectral shape is
inconclusive, the Amati relationship favors long burst and the merger
time at such high redshift could be problematic for a short burst