HETE-2 Localizations and Observations of Four Short Gamma-Ray Bursts: GRBs 010326B, 040802, 051211 and 060121

Article (PDF Available) · June 2006with25 Reads
Source: arXiv
Abstract
Here we report the localizations and properties of four short-duration GRBs localized by the High Energy Transient Explorer 2 satellite (HETE-2): GRBs 010326B, 040802, 051211 and 060121, all of which were detected by the French Gamma Telescope (Fregate) and localized with the Wide-field X-ray Monitor (WXM) and/or Soft X-ray Camera (SXC) instruments. We discuss eight possible criteria for determining whether these GRBs are "short population bursts" (SPBs) or "long population bursts" (LPBs). These criteria are (1) duration, (2) pulse widths, (3) spectral hardness, (4) spectral lag, (5) energy Egamma radiated in gamma rays (or equivalently, the kinetic energy E_KE of the GRB jet), (6) existence of a long, soft bump following the burst, (7) location of the burst in the host galaxy, and (8) type of host galaxy. In particular, we have developed a likelihood method for determining the probability that a burst is an SPB or a LPB on the basis of its T90 duration alone. A striking feature of the resulting probability distribution is that the T90 duration at which a burst has an equal probability of being a SPB or a LPB is T90 = 5 s, not T90 = 2 s, as is often used. All four short-duration bursts discussed in detail in this paper have T90 durations in the Fregate 30-400 keV energy band of 1.90, 2.31, 4.25, and 1.97 sec, respectively, yielding probabilities P(S|T90) = 0.97, 0.91, 0.60, and 0.95 that these bursts are SPBs on the basis of their T90 durations alone. All four bursts also have spectral lags consistent with zero. These results provide strong evidence that all four GRBs are SPBs (abstract continues).
arXiv:astro-ph/0605570v2 3 Jun 2006
HETE-2 Localizations and Observations of Four Short
Gamma-Ray Bursts: GRBs 010326B, 040802, 051211 and 060121
T. Q. Donaghy,
1
D. Q. Lamb,
1
T. Sakamoto,
2
J. P. No r ris,
2
Y. Nakagawa,
3
J. Villasenor,
4
J.-L. Atteia,
5
R. Vanderspek,
4
C. Graziani,
1
N. Kawai,
6,7
G. R. Ricker,
4
G. B. Crew,
4
J. Doty,
18,4
G. Prigozhin,
4
J. G. Jernigan,
8
Y. Shirasaki,
9,7
M. Suzuki,
6
N. Butler,
8,4
K. Hurley,
8
T. Tamagawa,
7
A. Yoshida,
3,7
M. Matsuoka,
11
E. E. Fenimore,
10
M. Galassi,
10
M. Boer,
12,21
J.-P. Dezalay,
12
J.-F. Olive,
12
A. Levine,
4
F. Martel,
19,4
E. Morgan,
4
R. Sato,
6
S. E. Woosley,
13
J. Braga,
14
R. Manchanda,
15
G. Pizzichini,
16
K. Takag ishi,
17
and M. Yamauchi
17
2
ABSTRACT
Here we report the localizations and properties of four short-duration GRBs
localized by the High Energy Transient Explorer 2 satellite (HETE-2): GRBs
1
Department of Astronomy and Astrophysics, University of Chicago, 5640 South Ellis Avenue, Chicago,
IL 60637.
4
MIT Kavli Institute, Massachusetts Institute of Technology, 70 Vassar Street, Cambridge, MA, 02139 .
3
Department of Physics, Aoyama Gakuin Univers ity, Chitosedai 6-16-1 Setagaya-ku, Tokyo 157-8572,
Japan.
5
Laboratoire d’Astrophysique, Observatoire Midi-Pyr´en´ees, 14 Ave. E. Belin, 31400 Toulouse, France.
6
Department of Physics, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551,
Japan.
7
RIKEN (Institute of Phys ic al and Chemical Research), 2-1 Hirosawa, Wako, Saitama 351-0 198, Japan.
8
University of California at Berkeley, Space Sciences L aborato ry, Berkeley, CA, 94720-7450.
9
National Astronomical Observatory, Osawa 2-21-1, Mitaka, Tokyo 181-8588 Ja pan.
10
Los Alamos National Laboratory, P.O. Box 1663, Los Alamos, NM, 87545.
11
Tsukuba Space Center, National Space Development Agency of Japan, Tsukuba, Ibaraki, 305-8505,
Japan.
12
Centre d’Etude Spatiale des Rayonnements, Observatoire Midi-Pyr´en´ees , 9 Ave. de Colonel Roche,
31028 Toulouse Cedex 4, France.
13
Department of Astronomy and Astrophysics, University of California at Santa Cruz, 477 Clark Kerr
Hall, Santa Cruz, CA 95064.
2
NASA Goddard Space Flight Center, Greenbelt, MD, 207 71.
14
Instituto Nacional de Pesquisas Espaciais, Avenida Dos Astronautas 1758, ao J os´e dos Campos 12227-
010, Brazil.
15
Department of Astronomy and Astrophysics , Tata Institute of Fundamental Research, Homi Bhabha
Road, Mumbai, 400 005, India.
16
INAF/IASF Bo logna, via Gobetti 10 1, 40129 Bologna, Italy.
17
Faculty of engineering, Miyazaki University, Gakuen Kibanada i Nishi, Miyazaki 889-2192, Japan.
18
Noqsi Aerospace, Ltd., 2822 South Nova Road, Pine, CO 80 470.
19
Espace Inc., 30 Lynn Avenue, Hull, MA 02045.
20
Department of Earth and Space Science, Graduate School of Science, Osaka University, 1-1
Machikaneyama-cho, Toyonaka, Osaka, 56 0-0043, Japan.
21
Observatoire de Haute Provence, 04870 St. Michel l’Observatoire, France .
3
010326B, 040802, 051211 and 060121, all of which were detected by the French
Gamma Telescope (Fregate) and localized with the Wide-field X-ray Monitor
(WXM) and/or Soft X-ray Camera (SXC) instruments. We discuss ten possi-
ble criteria for determining whether these GRBs are “short population bursts”
(SPBs) or “long p opulation bursts” (LPBs). These criteria are (1) duration, (2)
pulse widths, (3) spectral hardness, (4) spectral lag, (5) energy E
γ
radiated in
gamma rays (or equivalent ly, the kinetic energy E
KE
of the GRB j et), (6 ) exis-
tence of a long, soft bump following the burst, (7) location of the burst in the
host galaxy, (8) lack of detection of a supernova component to deep limits, (9)
type of host galaxy and (10) detection of gravitational waves. In particular, we
have developed a likelihood method for determining the probability that a burst
is an SPB or a LPB on t he basis of its T
90
duration alone. A striking feature of
the resulting probability distribution is that the T
90
duration at which a burst
has an equal proba bility of being a SPB or a LPB is T
90
= 5 s, not T
90
= 2 s,
which is the criterion that is often used to separate the two populations. The
four short-duration bursts discussed in detail in this paper have T
90
durations in
the Fregate 30-400 keV energy band of 1.90, 2.31, 4.25, and 1.97 sec, respectively,
yielding probabilities P (S|T
90
) = 0.97, 0.91, 0.60, and 0.95 that these bursts are
SPBs on the basis of their T
90
durations alone. All four bursts also have spectral
lags consistent with zero. These results provide strong evidence that all four
GRBs are SPBs.
Focusing further on the remarkable properties of GRB 060121, we present the
results of a detailed analysis of the light curve and time-resolved spectroscopy
of GRB 060121. The former reveals the presence of a long, soft bump typical of
those seen in the light curves of SPBs. This provides additional strong evidence
that GRB 060121 is an SPB. The latter reveals the existence of dramatic spec-
tral evolution during the burst, making this burst one of only a few SPBs for
which strong spectral evolution has been demonstrated. We find that the spec-
tral evolution exhibited by GRB 060121 obeys the Amati et al. (20 02) relation
internally.
GRB 060 121 is also the first SPB for which it has been possible to obtain
a photometric redshift fro m the optical and NIR afterglow of the burst. The
result provides strong evidence that GRB 060121 lies at a redshift z > 1.5, and
most likely at a redshift z = 4.6, making this the first short burst for which a
high redshift has been securely determined. At either redshift, its E
iso
and E
obs
peak
values are consistent with the Amati et al. (2002) relation. However, adopting
the jet o pening angles derived from modeling of its afterglow the values of E
γ
are
3.0×10
49
ergs if z = 1.5 and 1.3×10
49
ergs if z = 4.6. These values are similar to
4
those of the SPBs GRB 050709 and GRB 050724 and 10 0 times smaller than
those of almost all other hard GRBs. They therefore provide additional evidence
that GRB 060121 is a SPB. HST observations have shown that the pro bable host
galaxy of GR B 06 0121 is irregularly shaped and undergoing star formation. The
location of GRB 060121 appears not to be coincident with the strongest star
forming regions in the galaxy, which provides additional evidence that it is a n
SPB. Thus, all of the attributes of GRB 060121, when taken together, make a
strong, although not conclusive case, that GRB 060121 is an SPB. If GRB 060121
is due to the merger of a compact binary, its high redshift and probable origin in
a star-forming galaxy argue for a progenitor population for SPBs that is diverse
in terms of merger times and locations.
Subject headings: gamma rays: bursts (GRB 010326B, GRB 040802, GRB 051211,
GRB 060121) binaries: close stars: neutron black hole physics
1. Introduction
Gamma-Ray Bursts (GRBs) are thought to belong to two populations: short bursts
and long bursts (Mazets & Golenetskii 198 1; Hurley 1992; Lamb, Graziani & Smith 1993;
Kouveliotou et al. 1992). The localizations by HETE-2 (Ricker et al. 2003) and Swift of
three “short population bursts” ( SPBs) during the summer of 20 05 have solved in large part
the mystery of SPBs. The localization of GRB 050509B by Swift led to the first detection of
the X-ray afterglow of a short GRB, which was found to lie in the vicinity of a large elliptical
galaxy at redshift z = 0.225 (Gehrels et al. 2005). The first detection of an afterglow for
a SPB implied that such bursts also have detectable optical afterglows, and held out the
promise that the precise localization of the optical afterglow of a short GRB would lead to
the identification of the host and a secure measurement of the redshift of a short GRB.
It was not long before this promise was fulfilled. The localization of GRB 050709 by
HETE-2 (Villasenor et al. 2005) led to several firsts for a SPB: (1 ) the first detection of an
optical afterglow (Hjorth et al. 2005; Fox et al. 2005; Covino et al. 2006), (2) the first secure
identification of the host galaxy, (3 ) the first secure measurement of the redshift (z = 0.16)
(Fox et al. 2005; Covino et al. 2006), a nd (4) the first determination of where in the host
galaxy the burst occurred ( Fox et al. 2005). No supernova light curve was detected in the
case of either burst down to very sensitive limits (Fox et al. 20 05).
The localization of GRB 050724 by Swift (Barthelmy et al. 2005a) also led to the
detection of the X-ray (Barthelmy et al. 20 05a) and optical (Berger et al. 2005a) a fterglows
5
of the burst, a secure identification of the host galaxy, a determination o f where in the galaxy
the burst occurred, and a secure measurement of the redshift (z = 0.25). The peak fluxes
and fluences of GRBs 050709 and 050724, together with their redshifts, imply that these
SPBs are a thousand times less luminous and energetic than are typical long GRBs.
Both bursts occurred in the outskirts of their host galaxies, implying that they come from
very old systems, as do the facts that the host ga la xy of GRB 050724 is an elliptical galaxy
in which star formation ceased long ago and that no supernova light curve was detected in
either case down to very sensitive limits. These results strongly support the int erpretation
that many SPBs are due to the mergers of neutron star-neutron star or neutron star-black
hole binaries (Eichler et al. 19 89; Narayan, Paczy´nski & Piran 1992), and are therefore likely
associated with the emission of strong bursts of gravitational waves.
In contrast, “lo ng population bursts (LPBs) a re known to have X- r ay (Costa et al. 1997)
and optical afterglows (van Para dijs 1997), to occur at cosmological distances (Metzger 1997)
in star-forming galaxies (Castander & Lamb 1999), and to be associated with the explosion
of massive stars (Stanek et al. 2003; Hjorth et al. 2003) .
HETE-2 has localized six short-duration GRBs so far. Observational results for HETE-
2–localized short-duration bursts GRBs 020531 and 050709 have been reported previously
(Lamb et al. 2004, 2 006; Villasenor et al. 2005). In this paper, we report t he r esults of HETE-
2 observations of fo ur other short-duration bursts localized by HETE-2: GR Bs 0103 26B,
040802, 051 211, and 060121. These four bursts have T
90
durations in the Fregate 3 0-400 keV
energy band of 1.90, 2.31, 4.25, and 1.97 sec, respectively, yielding probabilities P (S|T
90
) =
0.97, 0.91, 0.60, and 0.95 that these bursts are SPBs on the basis of their T
90
durations alone.
All four bursts also have spectral lags consistent with zero. These results provide strong
evidence that all four of these GRBs are SPBs.
We focus particular attention on GRB 06 0121, a bright short-duration burst fo r which
both X-ray (Mangano et al. 2006a,b) and optical (Levan et al. 2006b; Postigo et al. 2006)
afterglows were detected, and a probable host galaxy identified ( Levan et al. 2006 b). The
light curve of GRB 060121 consists of a hard spike followed by a lo ng, soft bump features
that are similar to those of the light curves o f the short bursts GRBs 050709 (Villasenor et
al. 2005) and 050724 (Barthelmy et al. 2005a), and are characteristic of many perhaps all
SPBs, as analysis of BATSE (Lazzati, Ramirez-Ruiz & Ghisellini 2001; Connaughton et
al. 2002; Norris & Bonnell 2006) and Ko nus (Frederiks et al. 2004) short bursts have shown.
This provides additional strong evidence that GRB 060121 is an SPB. GRB 06012 1 exhibits
strong sp ectral evolution in both the value of the low-energy sp ectral index α and in the
peak energy E
obs
peak
of the spectrum in νF
ν
, and we find that this spectral evolution obeys the
Amati et al. (2002) relation internally.
6
GRB 060121 is the first short GRB for which it has been possible to obtain a photometric
redshift from the optical and NIR afterglow of the burst (Postigo et al. 2006). The results
provide strong evidence that GRB 060121 lies at a redshift z > 1.5 and most likely at
a redshift z = 4.6 (Postigo et a l. 2006) [see also Levan et al. (2006b)], making this the
first short burst for which a high redshift has been securely determined [the short burst
GRB 050813 may also lie at high redshift (Berger 2005b)]. At either redshift, the inferred
luminosity L, and isotropic-equivalent energy E
iso
, are 100 times larger than those inferred
for GRBs 0 50709 and 050724, and probably for GRB 050 509B, and are similar to those of
long GRBs; and the values of E
iso
and E
obs
peak
are consistent with the Amati et al. (2002)
relation. However, adopting the jet opening angles derived from modeling of its afterglow
the values of E
γ
are 3.0 × 10
49
ergs if z = 1.5 and 1.3 × 10
49
ergs if z = 4.6. These values are
similar to those of the SPBs GR B 050709 and GRB 050724 and 100 times smaller than
those of almost all other hard GRBs. They therefore provide additional evidence that GRB
060121 is a SPB.
HST observations (Levan et al. 2006b) have shown that the probable host galaxy of GRB
060121 is irregularly shaped and undergoing star formation. The location of G RB 060121
appears not to be coincident with the strongest star forming regions in the galaxy, which
provides a dditional evidence that it is an SPB. Thus, when taken together, the properties of
GRB 060121 make a very strong, although not conclusive case, that GRB 060121 is an SPB.
In §2, 3, 4 and 5, we describe the HETE-2 observations of the short-duration bursts,
GRBs 010326B, 040802, 051211, and 060121, respectively. For each burst we report their
localizations, temporal properties and spectral analyses, including time-resolved spectroscopy
of GRB 060121. We used XSPEC version 11.3.2 (Arnaud 1 996) for all spectral analyses
presented here. In §6 we discuss ten criteria fo r determining whether a particular burst is
an SPB or an LPB, consider the properties of the four short-duration bursts in the light of
these criteria, and discuss the implications for the nature of these four bursts especially
GRB 060121. In §7 we present our conclusions.
2. Observations of GRB 010326B
GRB 010 326B (trigger H1 496) was one of the very first GRBs detected by HETE-2.
The burst wa s detected by both Fregate ( Atteia et al. 2003) and the WXM (Kawai et al.
2003), but it occurred before the availability of real-time optical aspect data. Consequently,
analysis of the burst was carried out o n the ground and a WXM localization was circulated
about 5 hours af ter the trigger (Ricker et al. 2005). Table 1 details the localization time line
and Figure 1a shows a skymap. No optical transient was detected in the WXM error box
7
(Price et al. 2005).
The initial GCN reported a duration of about 4 seconds” for this burst (Ricker et al.
2005), but recent analysis finds a T
90
duration in the 85-400 keV energy band of 2.05 ± 0.65.
The duration of the burst increases at lower energies, reaching 5.44 ± 1.70 in the 2-10 keV
band (see Figure 6a and Table 2). An analysis of the spectral lag for this burst finds lag
values of -4
+24
32
msec between the 40-80 keV and 80-400 keV bands, and -2
+16
20
msec between
the 6-40 keV and 80-400 keV bands. Figure 2 shows the lightcurve of this burst in various
energy bands.
Table 4 lists the results o f the spectral analysis of this burst, which were first reported
in Sakamoto et al. (2005a). The burst-average spectrum is well-fit by a power-law times
an exponential (PLE)
1
model, with spectral index α = 1.08
+0.25
0.22
and peak energy E
obs
peak
=
51.8
+18.6
11.3
keV. A simple PL model is strongly disfavored. Fitting to a Band
2
model does not
yield any decrease in χ
2
for the extra degree of freedom, and the high-energy PL index β is
unconstrained by the fit. Table 6 gives the photon number and photon energy fluences, and
the photon number and photon energy peak fluxes, in various energy bands for this burst.
Figure 9 shows the best-fit PLE model and residuals f or this burst.
As can be seen in Figure 17, the spectral properties of GRB 010326B make it the softest
short event seen by HETE-2. However, this burst can be classified as a short G RB, based
on its T
90
duration and a spectral lag consistent with zero.
3. Observations of GRB 040802
GRB 040802 (trigger H34 85) was a bright, short burst that was detected by Fregate
but was only seen in the X-detector of the WXM. As is usual in such cases, we were able
to obtain a nar row localization in the X-direction and a much wider localization in the Y-
direction, derived from the exposure pattern in the X-detector. This resulted in a long,
narrow localization that was reported in a series of GCN Notices. Table 1 details the local-
ization t ime line a nd Figure 1b shows the skymap. The burst was a lso detected by the Mars
Odyssey (HEND), Konus-Wind and INTEGRAL (SPI-ACS) instruments, and the IPN was
1
Also known as a cutoff power-law, or CPL, model. It is defined by f (E) = A(E/E
scale
)
α
exp(E/E
0
),
where E
0
= E
obs
peak
/(2 + α). We take E
scale
= 15 keV for this work.
2
The Band model (Band et a l. 1993) is defined by f(E) = A(E/E
scale
)
α
exp(E/E
0
) for E < E
break
and f (E) = A(E
break
/E
scale
)
αβ
exp(β α)(E/E
scale
)
β
for E E
break
, where E
0
= E
obs
peak
/(2 + α) and
E
break
= E
0
(α β). We take E
scale
= 15 keV for this work.
8
able to derive an annulus that intersected the WXM error box. The 80 square arcminute
intersection was reported by Hurley et a l. (2004). No afterglow has been reported for this
burst.
GRB 040 802 has a T
90
duration in the 85-400 keV energy band of 1.35 ± 0.34. The
duration of the burst increases at lower energies, reaching 3.44 ± 0.76 in the 6-15 keV band
(see Figure 6b and Table 2). An analysis of the spectral lag for this burst finds lag values of
29
+32
30
msec between the 40-80 keV and 80-400 keV bands, and -6
+15
16
msec between the 6-40
keV and 80-40 0 keV bands. Figure 3 shows the lightcurve of this burst in various energy
bands.
Table 4 lists t he results of our spectral analysis of this burst. The burst is well-fit
by a PLE model, with spectral index α = 0.85
+0.23
0.20
and peak energy E
obs
peak
= 92.2
+18.8
13
keV. A fit t o a simple PL model is strongly disfavored. Fitting to a Band model does not
yield any decrease in χ
2
for the extra degree of freedom, and the high-energy PL index β is
unconstrained by the fit. Table 6 gives the photon number and photon energy fluences, and
the photon number and photon energy peak fluxes, in various energy bands for this burst.
Figure 10 shows the best-fit PLE model and residuals for this burst.
GRB 040802 can be classified as a short GRB, based on its T
90
duration and a spectral
lag consistent with zero. Its spectral properties are typical of those of HETE-2 short GRBs.
4. Observations of GRB 051211
The short GRB 051211 was detected by the Fregate, WXM and SXC (Villasenor et al.
2003) instruments. A WXM flight localization with a correct X-location but an incorrect
Y-location was sent out in near real time. Ground analysis of the WXM data confirmed
the flight X position, but due to low signal-to-noise in the Y-detector, yielded three roughly
equivalent Y po sition candidates. The SXC was able to localize soft emission occurring
35 seconds after the harder emission that triggered Fregate and the WXM. This emission
yielded a SXC X position that matched the WXM X localization and a SXC Y position that
matched one of the WXM Y candidates. This SXC localization was reported by Atteia et
al. (2005) and confirmed by Kawai et al. (200 5). Multiple follow-up observations yielded a
possible optical counterpart (Guidorzi et al. 2005), that was later found to b e more likely
a star and not a n afterglow (Halpern et al. 2005). Table 1 details the localization time line
and Figure 1c provides a skymap.
GRB 051211 has a T
90
duration in the 85-400 keV energy band of 4.02 ± 1.28 seconds.
The duration of the burst increases slightly at lower energies, reaching 4.82±0.79 in the 6-40
9
keV band (see Figure 6c and Table 2). A spectral lag of 0 ± 24 msec between the 30-85 keV
and 85-400 keV bands was first reported by Norris et al. (2005a). We have further calculated
a spectral lag of -2 ± 23 msec between the 40-80 keV and 80-400 keV bands. Figure 4 shows
the light curve of this burst in va r io us energy bands.
Table 4 lists the results of our spectral analysis of this burst. The burst is well-fit by
a PLE model, with spectral index α = 0.07
+0.50
0.41
and peak energy E
obs
peak
= 121
+33.0
20.3
keV. A
fit to a simple PL model is strongly disfavored. Fitting to a Band model spectrum does not
yield any decrease in χ
2
for the extra degree of freedom, and the high-energy PL index β is
unconstrained by the fit. Table 6 gives the photon number and photon energy fluences, and
the photon number and photon energy peak fluxes, in various energy bands for this burst.
Figure 11 shows the best-fit PLE model and residuals for this burst.
GRB 051211 can be classified as a short GRB, based on its T
90
duration and a spectral
lag consistent with zero. Its spectral properties are typical of those of HETE-2 short GRBs.
5. Observations of GRB 060121
On January 21 2006, at 22:24:54.5 UTC (80694.5 SOD), HETE-2 detected a short GRB
with Fregate. GRB 0 60121 was localized correctly in flight by the WXM and the position
was relayed to the GCN burst alert network within 13 seconds after the start of the burst.
The burst was a lso detected by the SXC, whose smaller error region was distributed after
analysis of the data on the ground. Optical and X-ray tra nsients were discovered in the SXC
error box, thus placing this burst on t he short list of short GRBs with observed afterglows.
This burst has provided a wealth of new results about short GRBs which we outline in this
section.
5.1. Localization
The WXM flight location o f GRB 060121 (with the standard 14
error radius) was
relayed to the ground via the burst alert network 13 seconds after the start of the burst.
After reviewing the data on the ground, a revised localization (Arimoto et al. 2006) with
an 8
radius was distributed 48 minutes after t he trigger. The SXC position with a 90 %
confidence error radius of 80
′′
was distributed 90 minutes aft er the trigger (Prigozhin et al.
2006).
The Swift satellite perfo r med a 5 ksec ToO observation of the HETE-2 error box be-
ginning on 22 January 20 06 at 01:21:37 UTC, or 2hr 56min 42.5s after the HETE trig ger.
10
Mangano et al. (2006a) reported a bright source inside the SXC error circle, located at R.A.
+09
h
09
m
52.13
s
, Dec +45
39
44.9
′′
, that was seen to fade in later o bservations ( Manga no
et al. 2006b). Early optical and infrared observations of the SXC error circle did not reveal
any optical detections, however a fter the discovery of the bright X-Ray transient, two g r oups
(Malesani et al. 2006; Levan et al. 2006a) reported detections of a very faint, variable optical
source at the position of the XRT source in the previously reported observations. Detection
of the near infrared (NIR) afterglow was reported by Hearty et al. (2006a,b), and further
observations were reported for the afterglow (Postigo et al. 2006; Levan et al. 2006b) and
the host galaxy (Leva n et al. 2006b).
Table 1 details the time line of localizations by the WXM and SXC instruments, as well
as the X-ray and optical followups. Figure 1d shows the relative sizes o f the error regions
for the WXM Flight, WXM Ground, and SXC Ground localizations, and the position of the
optical and X-ray counterparts.
5.2. Temporal Properties
Figure 5 shows the light curve of GRB 060121 in vario us energy bands. The burst
structure shows two peaks at 2 and 3 seconds after the trigger. GRB 0601 21 has a
T
90
duration in the 85-400 keV energy band of 1.60 ± 0.0 7 seconds. Figure 6 and Table 2
show the dependence of T
90
and T
50
on energy. The duration of GRB 06012 1 is shorter at
higher energies than at lower (T
90
10 sec in the 2-10 keV band), as is the case for the short
burst GRB 020531 (Lamb et al. 2004, 2006), as well as most long bursts. The discrepancy
between WXM and Fregate T
90
durations in similar bands in Table 2 is due to the different
background levels and sensitivities of the two instruments, and to the fact that T
90
is highly
sensitive to the choice of background. An analysis of the spectral lag for this burst finds lag
values of 2
+29
14
msec between the 40-80 keV and 80-400 keV bands, and 17 ± 9 msec between
the 6-40 keV and 80-400 keV bands.
Figures 7 and 8 show the light curves of GRB 060121 in the WXM 2-5 keV and 2-10
keV energy bands and the SXC 2-14 keV energy band from 50 s before the trigger until 300
s af t er the trigger, binned at 1 and 3 seconds respectively. Visual inspection of these
light curves reveals evidence f or a long, soft bump beginning about 70 s after the trigg er
and extending to about 120 s after the trigger in the WXM, and 150 s after the trigger in
the SXC. To assess the significance of this soft bump, we compare two models using the
likelihood ratio test, one assuming only a flat background is present and one assuming a flat
background plus a constant emission lasting from t
1
to t
2
are present.
11
We find evidence fo r the presence of soft emission in t he interval from t
1
= 66.87 sec to
t
2
= 155.43 sec in the SXC 2-14 keV energy band at a significance level of 4.4 × 10
4
. We
also find evidence for the soft emission from t
1
= 70 sec to t
2
= 122 sec in the WXM 2-5
keV and 2-10 keV energy bands at significance levels of 0.016 and 0.009, resp ectively. Thus
the light curve of GRB 060121 consists of a spike plus a long, soft bump beginning about 70
seconds after the spike and lasting about 50-90 seconds.
5.3. Spectrum
Table 5 lists the results of our spectral analysis of this burst. The burst-average spectrum
(t=0-10 sec) o f GRB 060121 is adequately fit by a PLE model, with spectral index α =
0.79
+0.12
0.11
and peak energy E
obs
peak
= 114
+14
11
keV (see the second-to-last set of entries in Table
5). A fit to a simple PL model is strongly disfavored. A fit to a Band model spectrum does
not yield any increase in χ
2
for t he extra degree of freedom, a nd the high-energy powerlaw
index β is unconstrained by the fit. Figure 12 shows the comparison of the burst-average
observed and predicted spectrum in count space. The best-fit parameter va lues that we find
for a PLE model are consistent with the values of α = 0.51
+0.55
0.60
, peak energy E
obs
peak
= 134
+32
17
keV, and β = 2.39
+0.27
1.41.
reported by (Golenetskii et al. 2006a) for a Band model from a
preliminary analysis of KONUS-WIND spectral data for the burst.
GRB 060121 was bright enough for us to perform a time-resolved spectral analysis of
the burst. Preliminary results of our joint WXM and Fregate spectral a na lysis were reported
by ( Boer et a l. 2006); the final results are summarized in Table 5.
Figure 13 shows the background and 5 foreground regions that we used for the time-
resolved spectral analysis. We selected the following five time intervals for our spectral
analysis: t = 0.0 1 .7 5, 1.75 2.7, 2.7 3.64, 3.64 5.186 and 5.18 6 10.0 seconds, as
measured from the trigger time. The spectral data for each of the five time intervals are
well fit by a PLE model (as were each of the 3 short GRBs considered above). In each case,
a simple PL model is strongly disfavored and fitting to a Band model spectrum does not
yield a ny decrease in χ
2
for the extra degree of freedom, nor is the high-energy PL index
β constrained by t he fit. Table 5 lists the results of our spectral analysis and Figure 14
shows the best-fit PLE model and residuals fo r each of the five time intervals. Table 5 also
lists the results of our spectral analysis of the time intervals t = 1.75 3.64, 0.0 3.64 and
0.05.186 sec. The Band spectral model is favored over the PLE model for the time interval
t = 1.75 3.64 sec as a consequence of the rapid sp ectral evolution that is occurring within
it.
12
We have calculated the 68% confidence region in the [α,E
obs
peak
]-plane for each time in-
terval. Figure 15 shows t he dramatic spectral evolution of the burst from a soft spectrum
with a low E
obs
peak
during the rise of the first peak, to a quite hard spectrum with a high E
obs
peak
during the first peak, fo llowed by softening and a decrease in E
obs
peak
during the second peak
and into the tail. Liang, Da i & Wu (2004) showed that the time-resolved spectra of bright
BATSE long bursts obeys internally the E
peak
L relation found by (Yonetoku, et al. 2 004)
[see also La mb, Donag hy, & Graziani (2005) ]. Following Liang, Dai & Wu (2004), in Figure
16, we plot E
obs
peak
against the average energy flux in each time interval and find that the four
points are consistent with a slope of +2, as is the case for long GRBs.
We have also analyzed the spectrum of the long, soft bump seen in the WXM 2-5 keV
and 2-10 keV time history data. The WXM 3-2 5 keV spectral data in the time interval
t = 71.2 121.6 sec as measured from the trigger time (which matches as closely as we can
the time interval during which soft emission is present, identified above) is adequately fit by
a simple PL spectrum with a PL index α = 2.81
+1.14
2.11
(see Table 5). The lower right panel
of Figure 14 shows the count sp ectrum and the residuals for the fit to the WXM data.
6. Discussion
Observations of short-duration GRBs, especially G RBs 050709 and 050724, made last
summer by HETE-2 and Swift provide strong evidence that some short-duration GRBs come
from merging neutron star-neutron star or neutron star-black hole binaries, whereas it has
been known for some time that most long-duration GRBs come from the collapse of massive
stars. However, as we will discuss below, there are clearly “short” GRBs (i.e., bursts that
almost certainly come from the merger of compact binaries) with durations at least as long
as 8 s, and “long” GRBs (i.e., bursts that come from the collapse of massive stars) with
durations at least as short as 2 s. Thus, a given “short” burst can be longer than many
“long” bursts, and a given long” burst can be shorter than many “short” bursts, making
this nomenclature quite awkward.
Another possibility might be to classify bursts as merger,” “magnetar,” or “collapsar”
GRBs, since the nature of the central engine that produces each kind of burst is key. However,
it seems premature to try to assign bursts to these three classes at this time.
Therefore, in this paper, we adopt the terms “short population burst” (SPB) and “long
population burst” (LPB). These terms have t he advantage of being closely related to the
often-used terms “short burst” and “long burst,” while emphasizing that r eference is being
made to two different populations of bursts, many o f whose attributes overlap. We note that
13
there is some evidence from the distribution of BATSE bursts in duration and hardness for a
third population o f soft bursts with durations int ermediate between those of SPB and LPB
bursts (Horv´ath 1998, 2002; Horv´ath et al. 200 6). How this third population of bursts, if it
exists, relates to the GRBs produced by the merger of compact binaries or t he collapse of
massive stars, is unknown.
In this section, we first discuss ten criteria for determining whether a particular burst
is an SPB or a LPB. We then consider the temporal and spectral properties of t hree of the
HETE-2 short-duration bursts described in detail in this paper, in the light of four of these
ten criteria. We then discuss in detail the properties of the fourth burst, GRB 0601 21, and
consider these properties in the light of all ten criteria. Finally, we consider the properties
of eight HETE-2 short-duration GRBs and twelve Swift short-duration GRBs observed to
date in the light of these ten criteria.
6.1. Criteria for Distinguishing Between SPBs and LPBs
We consider ten criteria for determining whether a particular burst is an SPB or a LPB.
These criteria are (1) duration, (2) pulse widths, (3) spectral hardness, (4) spectral lag, (5)
energy E
γ
radiated in gamma rays (or equivalently, the kinetic energy E
KE
of the GRB jet),
(6) existence of a long, soft bump following the burst, ( 7) location of the burst in the host
galaxy, (8) lack of detection of a supernova component to deep limits, (9) type of ho st galaxy
and (10) detection of gravitational waves.
The redshift distribution of the SPBs observed by HETE-2 and Swift is uncertain, but
possibly broad; the redshift distribution of LPBs is certainly broad. This broadens the
distributions of the properties of the bursts themselves, weakening the power of the first
four above criteria to distinguish between SPBs and LPBs. It would therefore be preferable
from both an empirical and a theoretical point of view to apply these criteria to the
properties of the bursts themselves, as measured in the rest frame of the burst. However,
this would limit the application of these criteria to bursts whose redshifts are known, which
is a small fraction of both populations of bursts. Worse, it is difficult to determine the
necessary burst properties in the rest frame of the burst, and in the case of some properties,
it is impossible to do so, as we discuss b elow. Therefore, in this paper, we apply the criteria
to the properties of the bursts themselves as measured in the observer frame.
The temporal properties o f SPBs differ from those of LPBs in two ways: SPBs generally
have shorter durations than do LPBs and the time histories of SPBs g enerally have much
narrower pulses than do LPBs (Lamb, Graziani & Smith 1 993; Norris et al. 1994, 1996), as
14
befits both, given the nomenclature for the two populations that we use in t his paper. These
two differences in temporal properties can potentially be used to distinguish between SPBs
and LPBs.
The T
50
duration distribution of GRBs exhibits two peaks that strongly overlap. These
two peaks are more evident in the T
90
duration distribution (Hurley 1992; Lamb, Graziani
& Smith 1993; Kouveliotou et al. 1992), but T
90
is more difficult to measure. The peaks in
the T
90
distribution lie at T
90
0.3 s and T
90
30 s, and the minimum in between lies at
T
90
2 s. The double-peaked T
90
duration distribution can be well fit by two (or three)
lognormal distributions [see, e.g., (Horv´ath 19 98, 2002; Horv´ath et a l. 2006)]. However,
observational selection effects may affect both the T
50
and the T
90
duration distributions. As
one example, the short end of the duration distribution very likely reflects the fact that the
shortest BATSE trigg er was 64 ms, rather than the intrinsic duration distribution of SPBs
(Lee & Petrosian 1996).
Determining the T
50
and T
90
durations of GRBs in t he rest frame of the burst would re-
quire taking into account three factors: (1) cosmological time dilation, which is proportional
to 1 + z; (2) the dependence of the duration on the energy band in which it is measured,
which is approximately proportional to (1 + z)
0.4
, and (3) the fact that T
50
and T
90
depend
on the background level. The first factor is straightforward to account for. The second can
be accounted for either approximately by using the factor (1 + z)
0.4
, which is correct on
average, or by attempting to measure the T
50
and T
90
durations in a fixed energy band in the
rest frame of the burst something that is difficult to do, given the possible broad redshift
range of SPBs and the known broad redshift range of LPBs. The third is impossible to
account fo r, since it would require knowing the light curve of the burst far below the level
of the background (i.e., with essentially infinite accuracy). In another paper, we explore the
use of the distribution of the “emission duration” introduced by Reichart et al. (2001) as a
possible criterion for distinguishing between SPBs and LPBs (Donaghy, Gra ziani, & Lamb
2006). Unlike T
50
or T
90
durations, the “emission duration” can be defined in the rest frame
of the burst.
In the present paper, we restrict ourselves to the use of T
90
as a criterion for distinguish-
ing between SPBs a nd LPBs. Using the best-fit parameters for the fit to the T
90
duration
distribution carried out by Horv´ath (2002 ) for two lognormal distributions, we have devel-
oped a likelihood method for determining the probability that a burst is an SPB or a LPB on
the basis of its T
90
duration alone (Donaghy, Graziani, & Lamb 2 006). Figure 18 shows the
resulting probability distribution. A striking feature of the resulting probability distribution
is that the T
90
duration at which a burst has an equal probability of being a SPB or a LPB is
T
90
= 5 s. The reason is t hat the duration distribution of SPBs is very bro ad (log σ = 0.61) .
15
Thus, the appropriate duration to use in dividing bursts into SPBs and LPBs is T
90
= 5 s,
not T
90
= 2 s, which is the criterion that is often used to separate the two populations.
Pulse widths in SPBs ar e much smaller than those in LPBs (Lamb, Graziani & Smith
1993; No rr is et al. 1994, 1996). The pulse width distribution in SPBs has a mean of 60
ms, while that for LPBs ha s a mean of 600 ms (Norris et al. 1994, 1996). However,
both distributions are bro ad. Therefore, pulse widths provides a good, but not conclusive,
criterion for distinguishing whether a particular burst is an SPB or a LPB.
The spectral properties of SPBs also differ from those of LPBs in two ways: the distribu-
tion of hardness ratios for SPBs may be somewhat harder than the distribution of hardness
ratios for LPBs; and, at high energies, SPBs exhibit zero spectral lag (Norr is & Bo nnell
2006), whereas all LPBs for which a spectral lag has been measured exhibit non-zero spec-
tral lag (No r ris 2002). These two differences in spectral properties can potentially be used
to distinguish between SPBs and LPBs.
SPBs have often been said to be harder than LPBs [see, e.g., (Kouveliotou et al. 1992)].
However, the distributions of the har dness ra tios for the two populations are broad and
overlap greatly, making it difficult to use this criterion to distinguish between SPBs and
LPBs. In addition, Sakamoto et a l. (2006) has recently shown that the short and long GRBs
detected by KONUS-WIND appear to have very similar hardness ratios. Figure 17 shows
that the hardness ratios for SPBs and for the hardest LPBs localized by HETE-2 and by
Swift are also very similar (Sakamoto et al. 2006). Furthermore, it is clear from Figure 17
that observa t io nal selection effects can strongly affect the hardness ratio distributions, since
the BATSE sample o f GRBs is missing many of the X-ray-rich GRBs and all of the XRFs
localized by HETE-2. All of these properties make the hardness ra t io a difficult criterion to
use to distinguish between SPBs and LPBs. Therefore, we do not use this criterion.
Extensive studies have shown that, at high energies, SPBs exhibit zero spectral lag
(Norris & Bonnell 2006), whereas all LPBs for which a spectral lag has been measured
exhibit non-zero spectral lag (Norris 2002)
3
. We t herefore consider an accurate measurement
of spectral lag, if it can be done, to be one of the best criteria for distinguishing between
SPBs and LPBs.
The spectral lag measurements for a number of the HETE-2 short-duration GRBs have
relatively large uncertainties, making it difficult to reach definite conclusions about whether
3
We no te that zero spectral la g does not preclude strong spectral evolution during a burst, if the burst
consists of more than one peak and the peaks have different spectral properties. What it does preclude is
strong spectral evolution within each peak.
16
these bursts are SPBs or LPBs on the basis of spectral lag alone. In another paper, we
explore the use of spectral lag measurements as a means of distinguishing between SPBs
and LPBs, using the model of the spectral lag distribution for LPBs developed by Norris
(2002) and taking into account rigorously the uncertainty in the measurement of spectral
lag (Donaghy, Graziani, & Lamb 2006).
The energy E
γ
radiated by a burst in gamma rays is E
γ
= 10
48
10
49
ergs for the SPBs
GRB 050709, and 050724, and E
γ
10
50
10
51
ergs for hard LPBs. We therefore consider
E
γ
to be a good, but not conclusive, criterion for distinguishing between SPBs and LPBs.
Long, soft bumps were seen in the light curves of the short bursts GRBs 050709 (Vil-
lasenor et al. 2005) and 050724 (Barthelmy et al. 2005a) , and appear to b e characteristic of
many perhaps all SPBs, as analysis of BATSE (Lazzati, Ramirez-Ruiz & Ghisellini 2001;
Connaughton et al. 2002; Norris & Bonnell 2006) and Konus (Frederiks et al. 2004) short
bursts have shown. We therefore consider the existence of a long, soft bump to be one of
the best criteria for distinguishing between SPBs and LPBs.
The precise localizations made possible by detection of the o ptical afterglows (Hjorth
et al. 2005; Fox et al. 2005; Covino et al. 2006; Berger et al. 2005a) of the short bursts
GRBs 050709 (Villasenor et al. 2005) and 050724 (Barthelmy et al. 2005a) revealed that
both bursts occurred in the outskirts of their host galaxies. In addition, no supernova light
curve was detected in either case down to very sensitive limits (Fox et al. 2005; Berger
et a l. 2 005a). These results provide strong support for the interpretation that many short
GRBs are due to the mergers of neutron star-neutron star or neutron star-black hole binaries
(Eichler et al. 1989; Narayan, Pa czy´nski & Piran 1992). In contrast, every LPB for which
an accurate location has been determined is coincident with a bright star-forming region in
the host galaxy (Fruchter et al. 2006); indeed, the locations of LPBs are much more tightly
concentrated in these star-forming regions than is the blue light from the host galaxy. We
therefore consider the location of the burst to be one of the best criteria for distinguishing
between SPBs and LPBs.
Supernova components have been detected in the optical afterglow light curves of many
LPBs, and ar e a common feature in the o ptical afterglow light curves of LPBs tha t lie at
redshifts z < 1. In contra st, supernova light curves are not expected for SPBs that come
from the mergers of compact binaries, and none were detected to very deep limits for GRB
050509b and GRB 050709 (Fox et al. 2005). We therefore consider the presence or absence
of a supernova comp onent in the optical afterglow light curve to be one of the best criteria
for distinguishing between SPBs a nd LPBs. In particular, we regard the clear detection of
a supernova component to be very strong evidence that the burst is an LPB, and the lack
of detection of a supernova component down to deep limits to be very strong evidence that
17
the burst is an SPB, provided that the burst lies at a redshift z < 1.
The SPB GRB 050724 occurred in an elliptical galaxy (Barthelmy et al. 2005a; Berger
et al. 2005a) in which star formation ceased long ago, as probably did GRB 0505 09b Gehrels
et al. (2005). However, two other SPBs [GRB 05 0709 (Villasenor et al. 2005) and G RB
051221A (Berger & Soderb erg 2005)] occurred in star-forming galaxies (Hjorth et al. 2005;
Fox et al. 2005; Covino et a l. 2006; Soderberg et al. 2006 ) . Indeed, even in some cases where
the host galaxy cannot be identified, it may be sufficient t o demonstrate that the stellar
population is old, as Gorosabel et al. (2006) argue for the case of 050813. We t herefore
consider that, if a particular burst occurs in an elliptical galaxy, this is conclusive evidence
that it is an SPB, while if the burst occurs in a star-forming galaxy, the kind of host ga la xy
provides no information about whether the burst is an SPB or an LPB.
If most SPBs are indeed due to mergers of neutron star-neutron star or neutron star-
black hole binaries, these events produce powerful bursts of gravitational radiation (Eichler
et al. 1989 ; Narayan, Paczy´nski & Piran 1 992) that should be detectable by the second-
generation Laser Interferometry Gravitational- wave Observatory (Thorne & Cutler 200 2;
Belczynski et al. 200 6). The detection of gravitational waves from a short-duration GR B
would therefore provide conclusive evidence that the event is an SPB. While it is unlikely
that gravitational waves will be detected fro m a short-duration GRB a nytime soon, the
detection of such waves will eventually be the gold standard for determining whether a burst
is an SPB, and we therefore include it here.
In summary, we have discussed ten possible criteria for determining whether a particular
burst is an SPB o r a LPB. Based on a careful consideration of the strengths a nd weaknesses
of each of the criteria, we rate spectral lag; long, soft bump; location in the host galaxy;
type of host galaxy; and detection of gravitational waves a s “g old” criteria (i.e., the best
criteria); duration, pulse width, and E
γ
as “silver” criteria (i.e., good criteria), and spectral
hardness as a “bronze” criterion (i.e., a poor criterion).
6.2. Temporal Properties
In this section, we consider the temporal properties of the four HETE-2 short-duration
bursts discussed in detail in this paper in t he light of criteria (1-2) discussed above.
Table 8 shows that three of the four bursts have probabilities P (S|T
90
) > 0.9 of being
SPBs, based on their T
90
durations a lone, and that in the cases of GRBs 010326B and
060121, the probability is 0.95 or greater. The table shows that P (S|T
90
) > 0.85 fo r four
HETE-2 short bursts (GRBs 020531, 021211, 040924, and 0507 09) whose properties have
18
been reported elsewhere (Lamb et al. 20 04, 2006; Crew, et al. 2003; Fenimore, et al. 2004;
Villasenor et al. 2005), and GRBs 020531 and 050709 have probabilities 0.995 and 1.000 of
being SPBs, based on their T
90
durations alone.
A detailed analysis of the pulse widths of the four HETE-2 short GRBs discussed in
detail in this paper and a quantitative comparison of these widths with the pulse width
distributions of SPBs and LPBs lies beyo nd the scope of the present paper. However, we
are par t icularly interested in GRB 0601 21. We have therefore made very r ough estimates
of the widths of the three pulses that are most clearly evident in this burst. These three
pulses have F WHMs of 600-800 ms, 300-400 ms, and 300-50 0 ms in the observer frame.
These pulse widths are larger than is typical of SPBs and somewhat smaller than is typical
of LPBs; consequently, the widths of the pulses in GRB 06012 1 considered in the o bserver
frame provide no clear evidence one way or another about whether the burst is an SPB
or a LPB. However, pulse widths (unlike T
50
and T
90
durations) can be calculated in the
rest fr ame of the burst, and we revisit these pulse widths below, after having discussed the
redshift of GRB 060121.
6.3. Spectral Properties
In this section, we consider the spectral properties of the four HETE-2 short-duration
bursts discussed in detail in this paper in light of criteria (3-4) discussed above.
The burst-average spectra of the four HETE-2 short GRBs discussed in this paper ar e
chara cterized by low-energy spectral indices α = 0.07
+0.50
0.41
1.08
+0.25
0.22
and peak energies
E
obs
peak
= 51.8
+18.6
11.3
121
+33.0
20.3
. Similar values were found for the two HETE-2 short GRBs
whose spectral properties were reported elsewhere (Lamb et al. 2004, 2006; Villasenor et al.
2005). The values of α and E
obs
peak
for the six HETE-2 short GRBs are typical of those for
bright short GRBs detected by BATSE (Ghirlanda et al. 2004) and for the three short GRBs
localized by Swift for which a PLE model is requested by the Swift BAT spectral data (see
Table 8). We note that in none of t he six HETE-2 short GRBs do the spectral data request
a Band model; this is also the case for bright short GRBs detected by BATSE (Ghirlanda
et al. 2004).
As already reported, we have calculated the spectral lag for the four HETE-2 short-
duration GRBs discussed in detail in this paper. We have also calculated the spectral lag
for four other HETE-2 short-duration GRBs. Table 3 shows that t he spectral lags measured
for six of these eight HETE-2 short GRBs are consistent with zero, taking into account
the uncertainty in the measurement, and that in two cases (GRBs 020531 and 050709) the
19
upper limit on any sp ectral lag is very small. However, Table 3 also shows that the remaining
two HETE-2 short-duration bursts (GRBs 0212 11 and 040924) exhibit definite spectral lags.
These two bursts are therefore LPBs, despite the fa ct that their T
90
durations are only 2.7
s and 2.4 s and the probabilities that t hey are SPBs are 0.8 7 and 0.90, respectively, on the
basis of their T
90
durations alone (see Table 8). The results for these two bursts illustrate
the difficulty in determining whether a given burst belongs to either the short or the long
classes of GRBs, using solely its T
90
duration, and the ability of a spectral lag analysis to do
so (Norris & Bonnell 2006).
6.4. Properties of GRB 060121
The light curve of GRB 060121 consists of a hard spike followed by a long, soft bump,
and in this way, it is similar to those of the SPBs GRB 050709 (Villasenor et al. 2005) and
GRB 050 724 ( Ba r thelmy et al. 2005a) . These general features may well be typical of all
SPBs seen by BATSE (Lazzati, R amirez-Ruiz & Ghisellini 2001; Connaughton et al. 2 002)
and Konus (Frederiks et al. 2004). However, the ratio of the fluence in the burst itself and
the fluence in the long, soft bump spans a range of at least 10
4
(Norris & Bonnell 2006).
The photon number and photon energy fluences of GRB 060121 are more than twice
those of any of the other four HETE-2 short GRBs discussed in this paper or of the other
two HETE-2 short GRBs whose properties were reported elsewhere (Lamb et al. 2 004, 2006;
Villasenor et al. 2005). The large photon number and photon energy fluences of GRB
060121 have a llowed us to perform time-resolved spectroscopy of this burst. We find that
the spectrum of GRB 060121 exhibits dramatic spectral evolution in both the value of the
low-energy spectral index α and the value of the peak energy E
obs
peak
of the spectrum in
νF
ν
(see Table 5 and Figure 14). GRB 020 531 also showed evidence for modest spectral
evolution, but only in the value of its low-energy power-law index α (Lamb et al. 2004,
2006). GRB 060121 is one of only a few short GRBs for which strong spectral evolution has
been established.
Both the X-ray aft erglow (Mangano et al. 2006a,b) and near infrared (NIR) af t erglow
(Hearty et al. 20 06a,b) of GRB 060121 were bright, but the optical afterglow was faint
(Malesani et al. 2006; Levan et al. 2006a; Breeveld et al. 2006). Nevertheless, at early times
the afterglow was much brighter than the probable host galaxy (Levan et al. 2006b) in both
the optical and the NIR. GRB 060121 is consequently the first short GR B for which it has
been possible to obtain a photometric redshift from the optical and NIR afterglow of the
burst (Postigo et al. 2006). Observations of the afterglow in the I-, R-, and K-bands and
the upper limits derived for the U-, B-, a nd V-bands indicate that the burst occurred at a
20
redshift z = 4.6 ± 0.6, or less probably, at a redshift z = 1.5 ±0.2 and with a large extinction
(A
V
= 1.4 ± 0.4) (Postigo et al. 2006).
Further support for a high redshift comes fro m the unusually red color of the probable
host galaxy and the presence nearby on the sky of five extremely red objects (EROs), which
are exceptionally faint (and therefore have low surface brightnesses) or are undetected in
the Hubble Space Telescope (HST) ACS/F606W filter, but are relatively bright in the HST
NICMOS F16 0W filter (Levan et al. 2006b). The five nearby EROs correspond to an over-
density on the sky of a factor of 20 (Levan et al. 2006b), suggesting that the host galaxy of
GRB 060121 may belong to a cluster.
These results provide strong evidence that GRB 060121 lies at a redshift z > 1.5, and
most likely at a redshift z = 4.6 (Postigo et al. 2006) [see also Levan et al. ( 2006b)], making
this the first SPB for which a high redshift has been securely determined.
4
The T
90
duration of the spike in the light curve of GRB 060121 is 1 .9 7 s in the 30-400
keV energy band, which gives a probability of P (S|T
90
) = 0.95 o f GRB 060121 being a SPB,
based o n its T
90
duration alone. The spectral lag measurement for the spike in the time
history o f GRB 060121 is 2
+29
14
ms between the 40-80 keV and 80-400 keV bands, which
is consistent with zero spectral lag . These results provide very strong evidence that GRB
060121 is a classical short GRB.
The inferred peak luminosity L
iso
and isotropic-equivalent energy E
iso
of GRB 06012 1
are 4.2 × 10
52
ergs s
1
and 3.7 × 10
52
ergs (assuming z = 1.5 ) , or 6.9 × 10
53
ergs s
1
and
2.4×1 0
53
ergs (assuming z = 4.6). These values are 10-100 times larger than those inferred
for the short GRBs 050709 and 050724, and probably G RB 050509B, and ar e similar to those
of long GRBs. Modeling of the afterglow gives o pening angles of 2.3
if z = 1.5 and 0.6
if z = 4.6, although the uncertainties in the opening angles are substantial (Postigo et al.
2006). Taking these opening angles at face value implies that the energy E
γ
in gamma-rays
emitted by GRB 060121 is 3.0 × 10
49
ergs if z = 1.5 and 1.3 × 10
49
ergs if z = 4.6. These
values of E
γ
are similar to those of the SPBs GRB 050709 and GRB 050724.
Figure 19 shows that the location of GRB 060121 in the (E
iso
, E
peak
)-plane is consistent
with the Amati et al. (2002) relation. In contrast, GRB 050709 lies well away from the
Amati et al. (2002) relation, as do the trajectories of three of the other four HETE-2 short
GRBs. However, with so f ew SPBs having measured spectral properties and redshifts, it is
impossible to know whether this is evidence that GRB 060121 is not a SPB, or that SPBs
4
The short burst GRB 050813 may also lie at high r e ds hift, but no photo metric or spectroscopic redshift
of the host galaxy has been reported as yet (Berger 2005b).
21
form a broad swath in the (E
iso
, E
obs
peak
)-plane, the upper end of which is consistent with the
Amati et al. (2002) relation.
6.5. Nature of GRB 060121
In this section, we consider the properties of GRB 060121 in the light of nine of the
ten criteria for determining whether a particular burst is an SPB or a LPB discussed above.
Excepting the detection of gravitational waves, the nine criteria are (1) duration, (2) pulse
widths, (3) spectral hardness, (4) spectral lag, (5) energy E
γ
radiated in gamma-rays (or
equivalently, the kinetic energy E
KE
of the GRB jet), (6) existence of a long, soft bump
following the burst, (7 ) location of the burst in the host galaxy, (8) la ck of detection of a
sup ernova component to deep limits and (9) type of host galaxy.
Duration. The T
90
duration of GRB 060121 in the 30-4 00 keV energy band is 1.97 s,
which gives a probability o f P (S|T
90
) = 0.95 of GRB 060121 being a SPB, based on its T
90
duration alone. Furthermore, assuming that the light curve of the burst has no low-level
peaks that ar e masked by the background, the T
90
duration of the burst would be 0.8 s (if
z = 1.5) and 0.3 s (if z = 4.6) if it had occurred at z = 1.2, a redshift similar to those
at which GRBs 050709 (z = 1.17) and 0507024 (z = 1.25) occurred, where we have ta ken
into account cosmological time dilation but neglected the dependence of burst duration on
the energy band, since the burst is comprised of at least three peaks. Thus, criterion (a)
provides strong evidence that G RB 060121 is a SPB.
Pulse Width. The widths of the three pulses visible in the time history of GRB 060121
are roughly 600-800 ms, 300-400 ms, and 300-500 ms in the observer frame. These pulse
widths would be 300-400 ms, 150-20 0 ms, and 150-250 ms (assuming z = 1.5) and 130-170
ms, 60- 90 ms, and 60-110 ms (assuming z = 4.6) if GRB 0 60121 had occurred at z = 1.2,
a redshift similar to those at which GRBs 050709 and 0 507024 occurred. Thus, if z = 1.5,
the pulse widths are a factor of a few larger than those typical of SPBs and a factor of a few
smaller than those typical of LPBs; the application of criterion (2) is therefore inconclusive.
However, if z = 4 .6 , the pulse widths are similar to those of SPBs and much smaller than
those of LPBs; the application of criterion (1) then provides evidence that GRB 0 60121 is a
SPB.
Spectral Hardness. As discussed above, it is very difficult to use spectral hardness as a
criterion for distinguishing between SPBs and LPBs. We therefore do not use this criterion.
Spectral Lag. The spectral lag measurement fo r the spike in the time history of GRB
060121 is 2
+29
14
ms between the 40-80 keV and 80-400 keV bands, which is consistent with
22
zero spectral lag, taking into account the uncertainty in the measurement. However, the
uncertainty in the measurement is relatively large. Consequently, although the spectral lag
measured for GRB 060121 is fully consistent with its being a SPB, the measurement provides
only modest evidence t hat it is one.
Energy Radiated in Gamma Rays. Adopting the jet opening angles derived from fits
to the afterglow light curve (Postigo et al. 2006) implies that the energy E
γ
in gamma-rays
emitted by GRB 060121 is 3.0 × 10
49
ergs if z = 1.5 and 1.3 × 10
49
ergs if z = 4.6. These
values of E
γ
are similar to those of the SPBs GRB 0507 09 and GRB 050724 and much
smaller than almost all hard GRBs. Thus, this criterion (5 ) provides strong evidence that
GRB 060121 is a SPB.
Existence of Long, Soft Bump. GR B 060 121 clearly exhibits a long, soft bump. Such a
feature appears to be characteristic of all SPBs, although the ratio of the fluence in the long,
soft bump and that in the sharp spike ranges over a factor of at least 10
4
(Norris & Bonnell
2006). Thus, this criterion (6) also provides strong evidence that GRB 060121 is a SPB.
Location of Burst in Host Galaxy. While the HST images o f the probable host galaxy
of GR B 060121 are noisy, the location of the burst appears not to be coincident with the
strongest star forming regions in the galaxy, which provides evidence that it is an SPB.
Supernova Light Curve. No supernova component was detected in the optical afterglow
light curve of GR B 060121; however, the detection of such a component is not expected,
given that the burst lies at a redshift z > 1.5. Therefore, t he failure to detect a supernova
component provides no evidence about whether the burst is an SPB or an LPB.
Type of Host Galaxy. The probably ho st galaxy of GRB 060 121 is a star-forming galaxy,
and thus provides no evidence about whether the burst is an SPB or an LPB.
In summary, all of the properties of GRB 060121 are consistent with its being a SPB.
Two criteria (pulse width and location of burst in host galaxy) provide modest evidence that
it is a SPB, while three criteria (T
90
duration; E
γ
; presence of a long, soft bump) provide
strong evidence that it is a SPB. These results are summarized in Table 9 . These results,
taken together, provide very strong, but not conclusive, evidence that GRB 060121 is an
SPB.
23
6.6. HETE-2 and Swift Short-Duration GRBs in Light of Ten Criteria for
Distinguishing Between SPBs and LPBs
Table 8 lists some temporal and spectral properties of twenty short-duration bursts
(eight HETE-2 short-duration bursts and the twelve Swift short-duration bursts observed so
far), while Table 9 summarizes the evidence that these bursts are SPBs or LPBs in light of
the nine of the ten criteria discussed above.
Table 8 shows that there is compelling evidence that GRBs 020531 and 050709 are
SPBs on the basis o f t heir t
90
duration and their lack of any spectral lag. In the case of
GRB 050709, the additional criteria involving its pulse width; the presence of a long, soft
bump; its E
γ
; its location in the outskirts of its ho st galaxy; and the la ck of a supernova
component in its optical afterglow light curve, together with its t
90
duration and the lack of
any spectral lag, provide overwhelming evidence that this burst is an SPB.
Table 8 provides strong evidence that three of the four HETE-2 short-duration bursts
discussed in this paper (i.e., GRBs 010326B, 040802, and 060121) are SPBs on the basis of
their t
90
duration alone, and that there is evidence that GRB 060121 is an SPB, if its redshift
is z = 4.6. GRB 010326B is more likely to be an LPB than an SPB on the basis of its pulse
widths. Table 9 shows that, in the cases of the short-duration bursts GRBs 010326B, 040802 ,
and 051211, the information needed to apply the o ther six criteria is lacking. However, it is
also impor t ant to note that in none of these three cases is there any evidence that supports
their being LPBs.
Table 3 also shows that GRBs 021211 and 040924 exhibit definite spectral lags. These
two bursts are therefore LPBs, despite the f act that their T
90
durations are only 2.7 s and
2.4 s and the probabilities that they are SPBs are 0.87 and 0.90, respectively, on the basis
of their T
90
durations alone (see Table 8). As already commented above, he results for these
two bursts illustrate the difficulty in determining whether a given burst belongs to either
the short or the long classes of GRBs, using solely its T
90
duration, and the ability of a
spectral lag analysis to do so (Norris & Bonnell 2006 ) . The existence of two LPBs whose
T
90
durations are 1 s in the rest frame of the burst would also appear to impose a severe
constraint on the collapsar model of SPBs Woosley (1993); Zhang, Woosley, & MacFadyen
(2003).
Table 8 shows that nine of the twelve short bursts localized by Swift so far have prob-
abilities > 0.95 of being SPBs, based on their T
90
durations alone (the exceptions being
GRBs 050724, 05122 7, and 06050 5). The t able also shows that there is compelling evidence
that nine of the twelve Swift short-duration bursts observed so far are SPBs on the basis of
their lack of any spectral lag (the exceptions being GRB 051210 for which the uncertainty
24
in the spectral lag measurement is relatively large, and GRBs 050202 and 060505 for which
spectral lag measurements have not yet been reported). There is thus compelling evidence
that at least ten of the twelve Swift short-duration bursts observed so far are SPBs, the two
exceptions being GRB0 50202 and GRB 0 60505 for which spectral lag measurements have
not yet been r eported.
Table 9 shows that, in the cases of the Swift short-duration bursts GR Bs 05 0202,
050509b, 050813, 050906, 051105a, and 060502b, there is additional strong evidence that
they are SPBs o n the basis of their pulse widths. The table also shows that, in the cases
of GRBs 050724 and 051227, there is additional compelling evidence that they are SPBs on
the basis of the presence of a long, soft bump; in the cases of GRBs 050 724 and 05122 1a,
there is additional compelling evidence that they are SPBs on the basis of the location in the
burst in its host galaxy; a nd in the cases of GRBs 050509b and 060505, there is additional
compelling evidence that they are SPBs on the basis of the lack of a supernova component
in its optical afterglow, down to deep limits.
Considering all of the evidence together, we consider that of the eight HETE-2 short-
duration bursts discussed in this paper there is conclusive evidence that two are SPBs
(GRBs 020531 and 0507 09), there is very strong but not conclusive evidence that a third
is an SPB (GRB 060121), and there is moderately strong evidence that three others are
SPBs ( GRBs 010326b, 04 0802, and 051211). Finally, there is conclusive evidence that the
remaining two bursts are LPBs (GR Bs 021211 and 040924). We also consider that of
the twelve Swift short-duration bursts discussed in this paper there is conclusive evidence
that three are SPBs (GRBs 050509b, 050 724, 051227) ; compelling evidence t hat seven of
the remaining nine bursts are SPBs; and moderately strong evidence that one of the two
remaining bursts is an SPB, the sole exception being GRB 050202 for which very little
information has been reported.
7. Conclusions
In this paper, we have reported the localizations a nd observations by HETE-2 of four
short bursts: GRBs 01 0326b, 04080 2, 051 211, and 060121. The durations and absence of
spectral lags for the four bursts provides strong evidence that all four bursts are SPBs.
Of the four short bursts, GRB 060121 is the most fascinating. It is one of only a few
short GRBs for which strong spectral evolution has been demonstrated. It is also the first
short GRB for which it has been possible to obtain a photometric redshift from the optical
and NIR afterglow of the burst. The result provides strong evidence that GRB 060121 lies at
25
a redshift z > 1.5, and more likely at a redshift z = 4.6, making this the first short burst for
which a high redshift has been securely determined. The properties of GRB 060121, when
taken together, provide very strong, but not conclusive, evidence that it is an SPB. If GRB
060121 is due to the merger of a compact binary, its high redshift and probable origin in a
star-forming disk galaxy argue for a progenitor population that is diverse in terms of merger
times and locations.
Acknowledgments
The HETE-2 mission is supported in the US by NASA contract NASW-4690; in Japan,
in part by the Ministry of Education, Culture, Sports, Science, and Technology Grant-in-
Aid 13440063; and in France, by CNES contract 79 3-01-8479. KH is grateful f or HETE-2
suppo r t under Contract MIT-SC-R-293291, for Mars Odyssey Support under NASA g r ant
FDNAG5-11451.
26
REFERENCES
Aloy, M. A., Ja nka, H.-Th., & Mueller, E. 2005, A&A, 436, 273
Amati, L. et al. 2002, A&A, 390, 81
Arimoto, M. et al. 2006, GCN Circ. 4550
Arnaud, K. A., 1996, Astronomical Data Analysis Software and Systems V, Eds. Jacoby G
and Barnes J., p. 17, ASP Conf. Series Volume 101.
At teia, J-L, et al. 2003, in Gamma-Ray Burst and Afterglow Astronomy 2001 , AIP Conf.
Proceedings 662, ed. G. R. Ricker & R. K. Vanderspek (New York: AIP), 17
At teia, J.-L. et al. 2005, GCN Circ. 4324
Balastegui, A., Ruiz-Lauente, P. & Canal, R. 2001, MNRAS 238, 283
Band, D. et al. 1993, ApJ, 413, 281
Barbier, L. et al. 2005, GCN Circ. 4194
Barthelmy, S. et al. 2005a, Nature, 438, 994
Barthelmy, S. et al. 2005b, GCN Circ. 3385
Barthelmy, S. et al. 2005c, GCN Circ. 4401
Barthelmy, S. et al. 2006, GCN Circ. 4879
Belczynski, K. et al. 2006, ApJ, in press
Berger, E., et al. 2005a, Nature, 438, 988
Berger, E., 2005b, in proc. of Swift GRB Symposium
Berger, E. & Soderberg, A. M. 2005, GCN Circ. 4384
Bloom, J. S. et al. 2006, ApJ, 638, 354
Boer, M. et al. 2006, GCN Circ. 4552
Breeveld, A. A. et al. 2006, GCN Circ. 4567
Bulik, T. Belczy´nski, K., & Zbijewski, W. 1999, A&A Suppl. 138, 483
Castander, F. & Lamb, D. Q. 1999, ApJ, 523, 593
27
Connaughton, V. 2002, ApJ, 567, 1028
Costa, E., et al. 1997, Nature 387, 783-784
Covino, S., et al. 2006, A&A, 447, L5
Crew, G., et al. 2003, ApJ, 599, 387
Cummings, J. et al. 2005a, GCN Circ. 4190
Cummings, J. et al. 2005b, GCN Circ. 4365
Donaghy, T. Q., Graziani, C., Lamb, D. Q., & Norris, J. P. 20 06, in preparation
Eichler, D., Livio, M., Piran, T., & Schra mm, D. N. 1989, Nature 340, 126
Fenimore, E. E., et al. 2004, GCN Circ. 2735
Fox, D. B. & Moon, D.-S. 2004, GCN Circ. 2734
Fox, D. B., et al. 200 5, Nature, 437, 845
Frederiks, D. D., et al. 2004, in Gamma-Ray Bursts in the Afterglow Era, ed. M. Feroci, F.
Frontera, N. Masetti, and L. Piro (ASP, San Francisco) 197
Fruchter, A., et al. 2002, GCN Circ. 1781
Fruchter, A., et al. 2006, Nature, in press, (astro-ph/0603537 )
Fryer, C. L., Woosley, S. E. & Hartmann, D. H. 1999, 526, 152
Gal-Yam, A., et al. 2006, ApJ, in press (astro-ph/0509891)
Gehrels, N., et al. 2005, Nature, 437, 851
Ghirlanda, G., et al. 2004, A&A, 422, L55
Golenetskii, S. et al. 2005, GCN Circ. 4394
Golenetskii, S. et al. 2006a, GCN Circ. 4564
Golenetskii, S. et al. 2006b, GCN Circ. 4881
Gorosabel, J. et al. 2006, A&A, 450, 87
Guidorzi C. et al. 2005, GCN Circ. 4356
28
Halpern, J. P. et al. 2 005, GCN Circ. 4623
Hearty, F. et al. 2 006a, GCN Circ. 4604
Hearty, F. et al. 2 006b, GCN Circ. 4611
Hjorth, J. et al. 2003, Nature, 423, 847
Hjorth, J., et al. 2005, Nature, 437, 859
Horv´ath, I. 1998, ApJ, 508, 757
Horv´ath, I. 2002, A&A, 392, 791
Horv´ath, I., et al. 2006, A&A, 447, 23
Hullinger, D. et al. 2005, GCN Circ. 4400
Hullinger, D. et al. 2006, GCN Circ. 5142
Hurley, K. 1992, in Gamma-Ray Bursts, ed. W. Paciesas & G. Fishman ( AIP, New York) 3
Hurley, K. et al. 2004, GCN Circ. 2637
Hurley, K., et al. 2005, Nature, 434, 1098
Kawai, N., et al. 2003, in Gamma-Ray Burst and Afterglow Astronomy 2001, AIP Conf.
Proceedings 662, ed. G. R. Ricker & R. K. Vanderspek (New York: AIP), 25
Kawai, N. et al. 2005, GCN Circ. 4359
Kouveliotou, C., et al. 1993, ApJ, 413, L101
Krimm, H. et al. 2005, GCN Circ. 3667
Krimm, H., et al. 2006, ApJ, submitted
Lamb, D. Q., Graziani, C. & Smith, I. 1993, ApJ, 413, L11
Lamb, D. Q., Donaghy, T. Q., & Graziani, C. 2005, ApJ, 520, 335
Lamb, D. Q., et al. 2004, in Gamma-Ray Bursts in the Afterglow Era, ed. M. Feroci, F.
Frontera, N. Masetti, and L. Piro (ASP, San Francisco) 94
Lamb, D. Q., et al. 2006, ApJ, submitted (astro-ph/0206151)
Lazzati, D., Ramirez-Ruiz, E. & Ghisellini, G. 2001, A&A 379, L39
29
Levan, A. J. et al. 2006a, GCN Circ. 4562
Levan, A. J., et al. 2006b, ApJ, submitted (astro-ph/0603282)
Liang, E. W., Dai, Z. G. & Wu, X. F. 2004, ApJ, 606, L29
Malesani, D. et al. 2006, GCN Circ. 4561
Mangano, V. et al. 2006a, GCN Circ. 4560
Mangano, V. et al. 2006b, GCN Circ. 4565
Markwardt, C. et al. 2006, GCN Circ. 4873
Mazets, E. P. & Golenetskii, S. V. 1981, Ap&SS, 75 , 47
Metzger, M R. et al. 1997, Nature 387, 878
Naka r, E., Gal-Yam, & Fox, D. B. 2006, ApJ, submitted (astro-ph/0511254)
Narayan, R., Paczy´nski, B., & Piran, T. 1992 , ApJ, 395 , L83
Norris, J. P., Nemiroff, R. J. & Davis, S. P. 1994, in AIP Conf. Proc. 307, Gamma- Ray
Bursts, ed. G. J. Fishman, J. J. Brainerd, & K. Hurley (New York: AIP), 172
Norris, J. P., et al. 1996, ApJ, 459, 393
Norris, J. P. 2002, ApJ, 579, 386
Norris, J. P. & Bonnell, J. T. 2006, ApJ, in press (astro-ph/0601190)
Norris, J. P. et al. 2005a, GCN Circ. 4377
Norris, J. P. et al. 2005b, GCN Circ. 4388
Oechslin, R., & Janka, H.-Th. 2006, A&A, submitted (astro-ph/0507099)
Ofek, E. O. et al. 2006, GCN Circ. 5123
Palmer, D. M., et al. 2005, Nature, 434, 1107
Par sons, A. et al. 2005, GCN Circ. 3935
van Paradijs, J. et al. 1997, Nature, 386, 686
Lee, T. T., & Petrosian, V. 1996, ApJ, 470, 479
30
Postigo, A., et al. 2006, Nature, submitted
Price, P. et al. 2001, GCN Circ. 1020
Prigozhin, G. et al. 2006, GCN Circ. 4551
Prochaska, J. X. et al. 2006, ApJ, 642, 989
Reichart, D. E., et al. 2001, ApJ, 552, 57
Ricker, G.R. et al. 2003, in Gamma-Ray Burst and Afterglow Astronomy 2001, AIP Conf.
Proceedings 662, ed. G. R. Ricker & R. K. Vanderspek (New York: AIP), 3
Ricker, G. R. et al. 2001, GCN Circ. 1018
Rosswog, S., Ramirez-Ruiz, E., & Davies, M. B. 2003, MNRAS, 345, 1077
Sakamoto, T. et al. 2 005a, ApJ, 629, 311
Sakamoto, T. et al. 2 005b, GCN Circ. 301 0
Sakamoto, T. et al. 2 005c, GCN Circ. 4403
Sakamoto, T. et al. 2 006, in proc. Swift GRB Symposium
Sato, G. et al. 2005a, GCN Circ. 3793
Sato, G. et al. 2005b, GCN Circ. 4318
Sato, G. et al. 2006, GCN Circ. 5064
Schady, P. et al. 2006, GCN Circ. 4877
Soderberg, A. M., et al. 2006, ApJ, submitted (astro-ph/0601455)
Stanek, C., et al. 2003, ApJ, 591, L17
Thorne, K. S., & Cutler, C. 2002, General R elativity and Gravitation, ed. N. Bishop & S.
D. Maharaj (Singapore: World Scientific), 72
Villasenor, J.N., et al. 2003, in G amma-Ray Burst and Af t erglow Astronomy 2001, AIP
Conf. Pro ceedings 662, ed. G. R. Ricker & R. K. Vanderspek (New Yo rk: AIP), 33
Villasenor, J., et al. 2005, Nature, 437, 855