Identification of a Local Sample of Gamma-Ray Bursts Consistent with a Magnetar Giant Flare Origin

Abstract

Cosmological gamma-ray bursts (GRBs) are known to arise from distinct progenitor channels: short GRBs mostly from neutron star mergers and long GRBs from a rare type of core-collapse supernova (CCSN) called collapsars. Highly magnetized neutron stars called magnetars also generate energetic, short-duration gamma-ray transients called magnetar giant flares (MGFs). Three have been observed from the Milky Way and its satellite galaxies, and they have long been suspected to constitute a third class of extragalactic GRBs. We report the unambiguous identification of a distinct population of four local (<5 Mpc) short GRBs, adding GRB 070222 to previously discussed events. While identified solely based on alignment with nearby star-forming galaxies, their rise time and isotropic energy release are independently inconsistent with the larger short GRB population at >99.9% confidence. These properties, the host galaxies, and nondetection in gravitational waves all point to an extragalactic MGF origin. Despite the small sample, the inferred volumetric rates for events above 4 × 10⁴⁴ erg of R_(MGF) = 3.8^(+4.0)_(−3.1) × 10⁵ Gpc⁻³ yr⁻¹ make MGFs the dominant gamma-ray transient detected from extragalactic sources. As previously suggested, these rates imply that some magnetars produce multiple MGFs, providing a source of repeating GRBs. The rates and host galaxies favor common CCSN as key progenitors of magnetars.

Full text

Identification of a Local Sample of Gamma-Ray Bursts Consistent with a Magnetar
Giant Flare Origin
E. Burns
1
, D. Svinkin
2
, K. Hurley
3
, Z. Wadiasingh
4,5
, M. Negro
6
, G. Younes
7,8
, R. Hamburg
9
, A. Ridnaia
2
, D. Cook
10
,
S. B. Cenko
4,11
, R. Aloisi
12,13
, G. Ashton
14
, M. Baring
15
, M. S. Briggs
9
, N. Christensen
16
, D. Frederiks
2
,
A. Goldstein
17
, C. M. Hui
18
, D. L. Kaplan
12
, M. M. Kasliwal
19
, D. Kocevski
18
, O. J. Roberts
17
, V. Savchenko
20
,
A. Tohuvavohu
21
, P. Veres
9
, and C. A. Wilson-Hodge
18
1
Department of Physics & Astronomy, Louisiana State University, Baton Rouge, LA 70803, USA
2
Ioffe Physical-Technical Institute, Politekhnicheskaya 26, St. Petersburg, 194021, Russia
3
Space Sciences Laboratory, University of California, 7 Gauss Way, Berkeley, CA 94720-7450, USA
4
NASA Goddard Space Flight Center, 8800 Greenbelt Road, Greenbelt, MD 20771, USA
5
Universities Space Research Association, Columbia, MD 21046, USA
6
University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, MD 21250, USA
7
Department of Physics, The George Washington University, Washington, DC 20052, USA
8
Astronomy, Physics and Statistics Institute of Sciences (APSIS), The George Washington University, Washington, DC 20052, USA
9
Department of Space Science, University of Alabama in Huntsville, Huntsville, AL 35899, USA
10
IPAC/Caltech, 1200 East California Boulevard, Pasadena, CA 91125, USA
11
Joint Space-Science Institute, University of Maryland, College Park, MD 20742, USA
12
University of Wisconsin–Milwaukee, P.O. Box 413, Milwaukee, WI 53201, USA
13
Department of Astronomy, University of Wisconsin–Madison, 475 North Charter Street, Madison, WI 53706, USA
14
OzGrav: The ARC Centre of Excellence for Gravitational Wave Discovery, Clayton, VIC 3800, Australia
15
Department of Physics and Astronomy, Rice University, MS-108, P.O. Box 1892, Houston, TX 77251, USA
16
Artemis, Université Côte dAzur, Observatoire de la Côte dAzur, CNRS, Nice F-06300, France
17
Science and Technology Institute, Universities Space Research Association, Huntsville, AL 35805, USA
18
Astrophysics Office, ST12, NASA/Marshall Space Flight Center, Huntsville, AL 35812, USA
19
Division of Physics, Mathematics, and Astronomy, California Institute of Technology, Pasadena, CA 91125, USA
20
Department of Astronomy, University of Geneva, Ch. dEcogia 16, 1290, Versoix, Switzerland
21
Department of Astronomy & Astrophysics, University of Toronto, 50 St. George Street, Toronto, ON, M5S 3H4, Canada
Received 2020 December 1; revised 2021 January 5; accepted 2021 January 5; published 2021 January 28
Abstract
Cosmological gamma-ray bursts (GRBs)are known to arise from distinct progenitor channels: short GRBs mostly
from neutron star mergers and long GRBs from a rare type of core-collapse supernova (CCSN)called collapsars.
Highly magnetized neutron stars called magnetars also generate energetic, short-duration gamma-ray transients
called magnetar giant flares (MGFs). Three have been observed from the Milky Way and its satellite galaxies, and
they have long been suspected to constitute a third class of extragalactic GRBs. We report the unambiguous
identification of a distinct population of four local (<5 Mpc)short GRBs, adding GRB 070222 to previously
discussed events. While identified solely based on alignment with nearby star-forming galaxies, their rise time and
isotropic energy release are independently inconsistent with the larger short GRB population at >99.9%
confidence. These properties, the host galaxies, and nondetection in gravitational waves all point to an extragalactic
MGF origin. Despite the small sample, the inferred volumetric rates for events above 4 ×10
44
erg of
=´
-
+
R
3.8 10
MGF 3.1
4.0
5
Gpc
−3
yr
−1
make MGFs the dominant gamma-ray transient detected from extragalactic
sources. As previously suggested, these rates imply that some magnetars produce multiple MGFs, providing a
source of repeating GRBs. The rates and host galaxies favor common CCSN as key progenitors of magnetars.
Unified Astronomy Thesaurus concepts: Gamma-ray bursts (629);Magnetars (992);Soft gamma-ray
repeaters (1471)
1. Introduction
The histories of gamma-ray bursts (GRBs)and magnetars
are intertwined. Short bursts of gamma rays were recorded by
the Vela satellites beginning in 1967 (Klebesadel et al. 1973)
and were given the phenomenological name GRBs. The
InterPlanetary Network (IPN)localized GRB 790305B to the
Large Magellanic Cloud (Mazets et al. 1979; Evans et al.
1980). It was unique in being the brightest event seen at Earth,
the prompt emission had a long-lasting, exponentially decaying
periodic tail (Barat et al. 1979), and additional weaker bursts
were localized to the same source (Mazets et al. 1979).
Immediately, there were papers investigating whether the main
event shared a common origin with other GRBs (Cline et al.
1980; Mazets et al. 1982). It is now known to be the first signal
identified from a magnetar.
Key results on the nature of GRBs in the subsequent decades
were often proven by population-level statistical analysis
before direct “smoking-gun”proof. Perhaps the greatest debate
was whether these events had a galactic or an extragalactic
origin, with the latter initially disfavored, as it would require
intrinsic energetics beyond anything previously known. Proof
came first indirectly via statistical studies of the spatial
distribution of GRBs (Meegan et al. 1992)and then directly
from redshift measurements (Metzger et al. 1997).
Studies of the prompt GRB emission provided strong
evidence in favor of two populations (Kouveliotou et al.
1993), with short and long GRBs traditionally separated at 2 s,
as measured by the T
90
parameter. Long GRBs were tied to
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 https://doi.org/10.3847/2041-8213/abd8c8
© 2021. The American Astronomical Society. All rights reserved.
1

broad-line type Ic core-collapse supernovae (CCSNe)called
collapsars (Galama et al. 1998). The Neil Gehrels Swift
Observatory (Swift)mission enabled successful detections of
afterglow from a sample of short GRBs. Circumstantial
evidence pointed toward a neutron star merger origin (Eichler
et al. 1989; Fong et al. 2015), with direct confirmation that
some GRBs arise from binary neutron star mergers coming
with GW170817 and GRB 170817A (Abbott et al. 2017).
Yet another debate on the behavior of GRBs is whether or
not the sources repeated. This is best explained using modern
parlance. Soft gamma-ray repeaters (SGRs)are galactic
magnetars named phenomenologically for the weak, recurrent
short bursts that first identified them before their physical origin
was known. SGR flares are classified as distinct from GRBs
and have recently been tied to radio emission similar to the
cosmological fast radio bursts (Bochenek et al. 2020). The flare
on 1979 March 5 and the subsequent similar events
GRB 980827 (Hurley et al. 1999; Mazets et al. 1999b)and
GRB 041227 (Palmer et al. 2005; Frederiks et al. 2007a)from
magnetars in the Milky Way are referred to as magnetar giant
flares (MGFs). The designation for the prompt emission of
MGFs often carries the GRB designation, which we use here.
GRBs are not thought to repeat as collapsars and neutron star
mergers are cataclysmic events. While several galactic
magnetars have been observed to produce multiple SGR flares,
none have been observed to produce multiple giant flares
(though this is not surprising). The historic debate on potential
repeating GRBs was likely confounded by magnetar transients
before the separation of SGR flares from GRBs.
We here refer to GRBs 790305B, 980827, and 041227 as the
known MGF sample. The detection of three from the Milky
Way and its satellite galaxies implies a high intrinsic rate on a
per-galaxy or volumetric basis. These events should be
detectable to extragalactic distances by GRB monitors such
as Konus-Wind (Aptekar et al. 1995), Swift-BAT (Barthelmy
et al. 2005), and Fermi-GBM (Meegan et al. 2009). However,
at these distances, only the immediate bright spike would be
detectable, and the event should resemble a short GRB (Hurley
et al. 2005). There are two events discussed in previous
literature as extragalactic MGF candidates, GRB 051103 (Ofek
et al. 2006; Frederiks et al. 2007b; Hurley et al. 2010)and
GRB 070201 (Mazets et al. 2008; Ofek et al. 2008), whose
chance alignment coincidence was measured to be ∼1%
(Svinkin et al. 2015).
There have been population-level searches for additional
events, which identified no additional candidates (Popov &
Stern 2006; Ofek 2007; Svinkin et al. 2015). However, these
studies allow us to constrain the fraction of detected short
GRBs that have an MGF origin; Ofek (2007)showed that the
rate of galactic events requires this to be >1%, while the lack
of additional candidates found in several searches constrains
the upper bound to be <8% (Tikhomirova et al. 2010; Svinkin
et al. 2015; Mandhai et al. 2018). These studies and their
conclusions generally assumed that the brightest MGFs could
be detectable to tens of megaparsecs.
Recently, GRB 200415A was identified as the third and
likeliest extragalactic MGF (Svinkin et al. 2021). In this work,
we perform a new population-level search utilizing the largest
GRB sample and new galaxy catalogs that are both more
complete and provide additional information, and we develop a
new formalism to determine if we can prove that extragalactic
MGFs contribute to the observed GRB population. Section 2.4
details the search formalism that identifies four nearby events,
identifying an additional extragalactic candidate. The progeni-
tors of our identified sample are investigated in Section 3, the
implications of which are discussed in Section 4. We conclude
with discussions in Section 5.
2. Local GRBs
The “smoking-gun”evidence of an MGF is the long periodic
tails, which are modulated by the rotation period of the neutron
star (Hurley et al. 1999)and also show quasi-periodic
oscillations related to the modes of the neutron star itself (Barat
et al. 1983; Israel et al. 2005; Strohmayer & Watts 2005; Watts
& Strohmayer 2006). However, these signatures are not
unambiguously identifiable at extragalactic distances with
existing instruments. As such, we follow prior population-
level searches and focus on spatial information; if a well-
localized short GRB is an MGF, it should occur within
∼50 Mpc and be consistent with a cataloged galaxy. We
combine existing GRB and galaxy catalogs to build the most
complete set of information from existing literature. For each
individual burst, we quantify our belief that it is an MGF from a
known galaxy through comparison of two probability distribu-
tion functions (PDFs), which are discussed below. These PDFs
are generated in HEALPix (Gorski et al. 2005). The resolution
of the HEALPix maps is defined by the NSIDE parameter,
where the number of total pixels is equal to the square of the
NSIDE times 12. The maps were generated with
NSIDE =8192, corresponding to a pixel width of ∼05.
2.1. The GRB Sample
We utilize data from the BATSE instrument on board the
Compton Gamma-Ray Observatory (Fishman et al. 1989),
Konus-Wind (Aptekar et al. 1995), Swift-BAT (Barthelmy
et al. 2005), Fermi-GBM (Meegan et al. 2009), and additional
information from the IPN.
22
Triggers from the same events
were matched utilizing temporal information for all events and
spatial information (Ashton et al. 2018)when available. The
total sample contains more than 11,000 GRBs observed, with
>1200 short GRBs using the standard 2 s cutoff.
Our burst sample selection requires three things. First, we
consider only short GRBs (T
90
<2s), where the T
90
used is the
shortest reported by any triggering instrument. Second, we
require the bolometric fluence (1 keV–10 MeV)determined
from a broadband instrument (Konus, BATSE, or GBM),
converting from the instrument-specific ranges as necessary.
Intercalibration uncertainties are within 25%. For the trigger
times, duration, and spectral properties, we utilized the latest
catalog information (Paciesas et al. 1999; Lien et al. 2016;
Svinkin et al. 2016; von Kienlin et al. 2020), updated online
catalogs,
23
and GCN circulars and performed dedicated
analysis when necessary.
Lastly, we require well-localized GRBs constructed from all
available information. For BATSE localization, we utilize the
latest catalogs (Goldstein et al. 2013)and apply the largest
systematic error (Briggs et al. 1999). Swift-BAT positions are
taken from the updated Swift-BAT Catalog,
24
and Swift-XRT
localizations are utilized when available.
25
Fermi-GBM
22
ssl.berkeley.edu/ipn3/index.html
23
http://www.ioffe.ru/LEA/shortGRBs/Current/index.html
24
https://swift.gsfc.nasa.gov/results/batgrbcat/index.html
25
https://swift.gsfc.nasa.gov/archive/grb_table/
2
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

localizations are quasi-circular and were generated using the
latest methods (Goldstein et al. 2020)for all bursts.
Konus localizations are an ecliptic band, which are
summarized in the IPN catalogs. The IPN compiles localization
information for GRBs, including the timing annuli derived
from the relative arrival times of gamma rays at distant
spacecraft. The information used here is from the IPN
localizations of Konus short GRBs through 2020 (Pal’Shin
et al. 2013)and the IPN list kept up to date online.
26
Additional
IPN localizations were compiled for more than 100 additional
short GRBs for this work, which were added to the online table.
The location information, including systematic error, from the
autonomous localizations, timing annuli, and Earth occultation
selections are converted to the HEALpix format using the
GBM Data Tools.
27
These independent PDFs are combined
into a final PDF referred to as P
GRB
.
The localization threshold is set to a 90% confidence
area <4.125 deg
2
when including systematic error. This value
is chosen because it is 1/10,000 the area of the sky, comparable
to the sum of the angular size of galaxies (as defined in the
following section)within 200 Mpc, and between previously
used thresholds (Svinkin et al. 2015). With the bolometric
fluence measure requirement and the removal of bursts with
known redshift (Lien et al. 2016)beyond the distance where the
event may be a detected MGF, we are left with a sample of 250
short GRBs. We do not apply more stringent cuts on spectral or
temporal information at this stage, as the relevant parameters
are not uniformly reported in GRB catalogs.
2.2. The Galaxy Sample
For the galaxies considered in this work, we require the
position (R.A., decl., distance), angular extent (if nonnegligible
at our spatial resolution; they are represented here as ellipses),
and current star formation rate (SFR). The z=0 Multi-
wavelength Galaxy Synthesis (z0MGS)Catalog (Leroy et al.
2019)combines the ultraviolet observations from the Galaxy
Evolution Explorer (Morrissey et al. 2007)with the infrared
observations of the Wide-field Infrared Survey Explorer
(WISE; Wright et al. 2010)to uniformly measure gas and dust
for galaxies within approximately 50 Mpc. As a result, for
galaxies contained in this catalog, these measures of the
distance and SFR are our default values. The angular size of
galaxies is represented as an ellipse when the data allow or a
circle when the axial ratio is not known. Angular extent is taken
from the input catalogs but is generally the Holmberg isophote,
i.e., where the B-band brightness is 26.5 mag arcsec
2
.
The Census of the Local Universe (CLU)Catalog (Cook
et al. 2019)aims to provide the most complete catalog of
galaxies out to 200 Mpc. We use the CLU measures of distance
and SFR when they are not provided by z0MGS, and we use
the CLU measures for angular size (which are not provided by
z0MGS). When missing, we add position angle information
from HyperLEDA (Paturel et al. 2003). The SFR measures of
these two catalogs correct for internal extinction using WISE4/
far-UV luminosities. To ensure completeness within <10 Mpc,
we supplement these two catalogs with the Local Volume
Galaxy (LVG)Catalog (Karachentsev & Kaisina 2013). The
three catalogs are matched by name, with help from the
NASA/IPAC Extragalactic Database (NED),
28
and position
information.
We consider galaxies between 0.5 Mpc (excluding the Milky
Way and its satellite galaxies)and 200 Mpc (beyond where
MGFs can be detected), which leaves more than 100,000
galaxies. The SFR is a key parameter in our method, and our
inferences also rely on scaling the properties of our host galaxy.
The Milky Way SFR used here is 1.65 ±0.19 M
e
yr
−1
(Licquia & Newman 2015). We specify the SFR for
NGC 3256, which was identified in Popov & Stern (2006)as
being a likely source of detectable extragalactic MGFs. We
searched the literature for values of the active SFR in this
galaxy and took the value of ∼36 M
e
yr
−1
from Lehmer et al.
(2015), which is inferred using UV information and is among
the middle reported values.
2.3. MGF Spatial Distribution
We seek an all-sky PDF, P
MGF
, representing the probability
that a given position will produce a MGF with a particular
fluence at Earth. Note that this is determined by the fluence of
each burst considered but is constructed independently of the
location of the burst itself, P
GRB
. The comparison of the two
PDFs generated for each burst quantifies the likelihood that a
given short GRB has an MGF origin, which is performed in the
next section. This section details the burst-specific construction
of P
MGF
.
If a given burst has an MGF origin it should arise from a
cataloged galaxy and its intrinsic energetics should fall into the
expected range. To construct this, we compute a weight for
each galaxy representing how likely it is to have produced the
observed fluence for the burst under consideration. This weight
has two components: a linear weighting with SFR and a more
complex weighting that compares the inferred intrinsic
energetics (determined by the burst fluence and potential host
galaxy distance)against an assumed PDF.
Magnetars are expected to be able to produce MGFs only for
a short period of time (approximately 10 kyr; Beniamini et al.
2019), tying the predicted rate of MGFs to the rate of their
formation. The rate of CCSNe can be inferred from the SFR,
since the lifetimes of stars that undergo core collapse are much
shorter than the timescale probed by the SFR tracers (Botticella
et al. 2012). Under the assumption that the dominant formation
channel for magnetars is CCSNe (which is explored in
Section 4), we can infer the rate of MGFs from a galaxy from
its SFR. Thus, each galaxy is linearly weighted with SFR. We
use the far-UV measure of SFR (Lee et al. 2010)when
available, as it should track massive stars likely to undergo core
collapse; otherwise, we use the Hαmeasure (Kennicutt 1998)
scaled by the average difference from galaxies with both
measures to account for the lack of dust correction in the LVG
catalog.
Next, we can determine the total isotropic-equivalent
energetics of a potential burst–galaxy pair as E
iso
=4πd
2
S,
where Sis the burst fluence and dis the distance to the potential
host. This value can be compared to an assumed intrinsic
energetics PDF to determine how likely the event is to be an
MGF. For example, a particularly high fluence short GRB
spatially aligned with a distant galaxy would require an
intrinsic energetics far beyond what has been observed in the
galactic MGFs, excluding an MGF origin. We note that some
26
http://www.ssl.berkeley.edu/ipn3/
27
https://fermi.gsfc.nasa.gov/ssc/data/analysis/gbm/gbm_data_tools/
gdt-docs/
28
https://ned.ipac.caltech.edu/
3
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

studies utilize the peak luminosity
L
iso
Max , but we work with an
E
iso
distribution, as there is stronger theoretical guidance on the
maximum total energy that can be released (related to the
magnetic fields of the magnetar)than on the timescale on which
it is released.
We now construct an informed intrinsic energetics function,
assuming a power-law distribution with an assumed minimum
and maximum value, which is similar to the behavior of lower-
energy magnetar flares (Cheng et al. 1996). Our method
bypasses the need for an assumed detection threshold, which is
difficult to quantify when considering many instruments over
30 yr. The assumed and inferred values are reported below,
with the initially determined distribution shown in Figure 1.
The slope of a power law can be determined via maximum
likelihood, independent of an assumed maximum value, as
⎡
⎣
⎢
⎢
⎛
⎝
⎜⎞
⎠
⎟
⎤
⎦
⎥
⎥()()
å
as
a
=+ = -+
a
=
-
-
nE
En
On1ln , 1,1
i
n
1
iso, i
iso,min
1
1
where the sum is over the observed E
iso
and
E
iso,min is the
lowest considered value (Newman 2005; Bauke 2007). We set
E
iso,min as 1.0 ×10
44
erg, which is a factor of a few below the
lowest value measured in a known MGF, as shown in Table 1,
but above the brightest SGR flare that lacked the periodic tail
emission (Mazets et al. 1999a). Iterating over the E
iso
values of
the known MGFs (GRBs 790305B, 090827, and 041227)gives
α=1.3 ±0.9 at 90% confidence, where we have included the
O(n
−1
)error contribution. In order to minimize the required
computation, we assume the centroid (α=1.3)in what
follows; the effect of this assumption on our results is
discussed in the closing paragraph of this section.
There must be a physical maximum energy for an MGF,
which should be related to the total magnetic energy. This is
supported by the lack of detection of more energetic events
otherwise consistent with an MGF origin. The highest E
iso
observed for a known MGF is 2.3 ×10
46
erg, which comes
from the magnetar with the highest reported magnetic field at
the surface of 2.0 ×10
15
G(Olausen & Kaspi 2014). We note
that this reported value is approximately three times larger than
the dipolar spin-down inferred magnetic field value of
7×10
14
G(Younes et al. 2017), but we have confirmed that
this does not affect our results. To determine an
E
iso, max for our
search, we assume a dipole field, where the available energy
scales as B
2
, and a nominal maximum magnetic field strength
of ∼1.0 ×10
16
G. This gives
()=´ ´ ´ ´
E
2.3 10 erg 1.0 10 G 2.0 10 G
iso, max 46 16 15 2
=´5.75 1047 erg.
This allows us to determine the burst-specific two-comp-
onent weight for each of the >100,000 galaxies in our sample,
which are weighted linearly by SFR multiplied by the value of
the E
iso
PDF for the inferred energetics considering the burst
fluence and galaxy distance. The sum of the galaxy weights is
normalized to unity. Then, P
MGF
is built by placing the
calculated weights at the position of the host galaxy. If the
angular diameter of the galaxy is larger than the effective
resolution of our discrete sky representation (∼arcmin
2
), then
its weight is uniformly distributed over its angular extent.
2.4. The Search
For each of the 250 short GRBs in our sample, we generate
P
GRB
from the observations of the GRB and P
MGF
from
theoretically motivated expectations. We quantify the like-
lihood that a given GRB has an MGF origin using
pW= åPP A4iii
i
GRB MGF , where P
GRB
i
and P
MGF
i
indicate the
probability for each PDF in the ith sky region, which has area
A
i
(Ashton et al. 2018).
Significance is determined by the empirical false-alarm
method (e.g., Messick et al. 2017)with Ωas our ranking
statistic. Our backgrounds are generated by simulating different
galaxy distributions. Each iteration is generated by uniform
rotation of the 2D (R.A., decl.)positions of the galaxies in our
sample, which maintains the distance and SFR distributions, as
well as local structure. Population-level confidence intervals
created through comparison of each rotation against our full
GRB sample with results are shown in Figure 2. At three and
four events, the short GRB sample has an excess surpassing 5σ
discovery significance, with individual significance values of
the four bursts between 1.2 ×10
−4
and 4.9 ×10
−6
, as given in
Table 1.
Three of the four are discussed in the literature as
extragalatic MGF candidates. The Konus-Wind lightcurves
are shown in Figure 3. The GRB 070201 has the least robust
association with a nearby galaxy; however, the localization is
comparatively large (∼10×the other events), and M31 has the
largest angular size of any galaxy in our sample, together
lowering Ωeven for real associations. We confirm this by
checking GRB 790305B with the Large Magellanic Cloud
(Evans et al. 1980; Cline et al. 1982), which has an even larger
angular extent than M31, giving Ω=500.
We perform a number of sanity checks to ensure our
assumptions do not significantly affect our results. The search
we run assumes the centroid α=1.3 value; however, we have
confirmed that running the search at the 90% confidence
interval bounds (α=0.5, 2.2)identifies the same four bursts as
significant outliers and does not identify other candidates.
Running the search at greater NSIDE affects our Ωvalues by
<10%. Rerunning the search where the linear SFR weighting is
altered to the stellar mass results in identification of the same
galaxies but with generally lower Ωvalues. Running with a
specific SFR returns similar results. Together, these suggest a
progenitor that tracks SFR. Our results are insensitive to the
assumed
E
iso, mi
n
, so long as we do not exclude known events,
as events of this strength are not detected far into the universe.
There are a few events with Ω>1 that are either excluded as
Figure 1. Initial assumed MGF energetics distribution, with
E
iso, min and
E
iso, max set to the x-axis boundaries. The PDF form is
()(
)
/a--
aa a-- -
EE E1iso max
1min
1. As described in the text, α=1.3 ±0.9 (at
90% confidence). The three E
iso
values from the known MGFs used to
constrain the slope are shown as black vertical lines.
4
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

events of interest for our MGF search or insignificant given our
sample. Lastly, significantly raising the assumed
E
iso, max
marginally identifies GRB 100216A (Ω=10), which indeed
has a potential host galaxy within 200 Mpc (Perley et al. 2010),
which is inconsistent with expectations for MGFs.
3. Progenitor Investigations
To determine the origin of these four bursts, we first
determine if the known GRB progenitors are compatible.
Collapsars power long GRBs with durations 2 s and are
followed immediately by afterglow and then broad-line type Ic
supernovae. This origin is excluded, as all four events have
durations of 0.1 s or less. Additionally, no subsequent super-
novae were reported in any case (Li et al. 2011b; though see
Gehrels et al. 2006; Grupe et al. 2007). A neutron star merger
origin is excluded by LIGO nondetections in gravitational
waves for three of the four events (Abbott et al. 2008; Abadie
et al. 2012; Aasi et al. 2014), but observations are insufficiently
sensitive to inform on the origin of GRB 200415A. One may
consider whether off-axis GRBs could explain these events.
The best-known such event is GRB 170817A, where the
duration was longer and spectrum softer than the bulk of the
short GRB population, which is inconsistent with the prompt
emission from these four local events. Further, the rates of
these local events (discussed in the following section)are
orders of magnitude higher than cosmological GRBs (Siegel
et al. 2019), even considering events that are oriented away
from Earth.
To determine the progenitors of these events, we follow the
historical procedure, where we begin by population comparison
of prompt emission parameters. The only additional potential
progenitors for extragalatic GRBs commonly discussed in the
literature are MGFs, where, contrary to the works that
identified the two confirmed progenitors, we have the
advantage of observations of galactic events, which are
summarized in Table 1. The parameters relevant for only the
main peak of the flare that appear distinct from cosmological
GRBs are the rise time and intrinsic energetics. Figure 4
contains the population comparison of these parameters.
First, MGFs have rise times of order a few milliseconds, far
shorter than most cosmological short GRBs (Hakkila et al.
2018). Rise times are not reported in most GRB catalogs. As a
proxy for the rise time, we define the time to peak as the time
from the start of the emission to the beginning of the peak 2 ms
counts interval. An Anderson–Darling k-sample test against 75
bright Konus short GRBs (∼15% brightest bursts detected by
Konus between 1994 and 2020)rejects the null hypothesis that
Table 1
A Summary of the MGF Sample
Known Extragalactic
MGF Event 790305B 980827 041227 200415A 070222 051103 070201
Origin
False-alarm rate 0 0 0 4.9 ×10
−6
7.8 ×10
−6
1.5 ×10
−5
1.2 ×10
−4
BNS excl. (Mpc)6.7 5.2 3.5
Galaxy Properties
Catalog name LMC MW MW NGC 253 M83 M82 M31
Distance (Mpc)0.054 0.0125 0.0087 3.5 4.5 3.7 0.78
SFR (M
e
yr
−1
)0.56 1.65 1.65 4.9 4.2 7.1 0.4
GRB Properties
Duration (s)<0.25 <1.0 <0.2 0.100 0.038 0.138 0.010
Rise time (ms)∼2∼4∼12 4 2 24
L
iso
max (10
46
erg s
−1
)0.65 2.3 35 140 40 180 12
E
iso
(10
45
erg)0.7 0.43 23 13 6.2 53 1.6
Index −0.7 0.0 −1.0 −0.2 −0.6
E
peak
(keV)500 1200 850 1080 1290 2150 280
Note. The significance for extragalactic events is from this text. Here BNS excl. refers to the neutron star merger exclusion distances from LIGO, LMC refers to the
Large Magellanic Cloud, and MW refers to the Milky Way. Individual significance is determined by comparison of the individual Ωagainst the full background
sample. Distances for the known magnetars come from Olausen & Kaspi (2014); extragalactic distances are taken from the host galaxy values (which have minor
variations with our catalog values). The GRB parameters include E
peak
as the energy of peak output and index as the low-energy power law from the spectral fit, and
the rest are discussed in the text. The GRB measures for the galactic events are from the literature; GRB measures for extragalactic events are all measured from
Konus-Wind data.
Figure 2. Discovery of a local but extragalactic population of GRBs. Here Ωis
a statistic that ranks how believable it is that the event is an extragalactic MGF,
with values for the true population shown in orange. The background
confidence intervals at 1σ,3σ, and 5σare shown in blue. The four most
significant events together surpass 5σdiscovery significance.
5
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

they are drawn from the same population at >99.9%
confidence.
Second, MGF E
iso
values are orders of magnitude fainter
than cosmological GRBs, where only the unusual
GRB 170817A (Abbott et al. 2017)is comparable. This
parameter depends on the distance to the source, which is not
directly observable from prompt emission. For some cosmo-
logical GRBs, the direct distance (redshift)determination is
made from follow-up observations. However, for most short
GRBs, the distance is determined by first robustly associating
the short GRB with an aligned or nearly aligned host galaxy
and then determining the distance to the host (Fong et al. 2015).
We adapt this last approach for MGFs to enable the use of
larger prompt emission localizations and expected host galaxy
properties. For each GRB and potential host galaxy, we
calculate /pW=åPP A4iii ihost GRB host , with P
host
the weighted
spatial distribution of that galaxy. Each GRB has only a single
likely host, providing a robust association. In the literature,
GRB 051103 has been discussed as belonging to the M81
Group of galaxies (Frederiks et al. 2007b), which is dominated
by the interacting galaxies M81 and M82. Our galaxy catalog
selection and method assign the burst to M82.
The inferred E
iso
values for each extragalatic MGF candidate
are given in Table 1. For the population comparison, we add
the E
iso
distribution of GBM short GRBs (Abbott et al. 2017)to
the sample of Konus bursts with measured redshift (Tsvetkova
et al. 2017). Together, these give 23 short GRBs with E
iso
determined by a broadband instrument, which is the largest
such sample to date. The extragalactic MGFs are clearly
inconsistent with the broader population, rejecting the null
hypothesis at >99.9% confidence.
Host galaxy studies of GRBs have been key in determining
prior progenitor channels (e.g., Fong et al. 2015). As discussed
in the design of our method, MGFs are expected to arise in star-
forming galaxies or regions. Within our maximal detection
distance for these bright events, the galaxies with the highest
SFR are M82, M83, NGC 253, and NGC 4945 (Mattila et al.
2012). GRB 051103 is associated with M82 by our method or
consistent with star-forming knots on the outskirts of M81
(Ofek et al. 2006), GRB 070222 with M83, and GRB 200415A
with the star-forming core of NGC 253 (Svinkin et al. 2021).
GRB 790305B is associated with the star-forming Large
Magellanic Cloud. This is consistent with a massive-star
progenitor, as expected for an MGF origin.
Individually, GRBs 200415A and 051103 are the most
robust identifications of extragalactic MGFs based on our
significance assessment and the results of partner analyses
including lightcurve morphology and submillisecond variation
of the prompt emission (Roberts et al. 2021; Svinkin et al.
2021). Newly identified is GRB 070222, which is in-class with
key properties of MGFs. However, it has two distinct but
overlapping pulses, which is not known to occur from galactic
events. This requires either a broader morphology of MGFs, a
distinct and unknown origin, or a 1 in 100,000 chance
alignment (Table 1). However, given the range of (quasi)
periodic oscillations seen from magnetar emission, such a
morphology is not necessarily surprising.
To summarize the observational case for an MGF origin:
these events localize to the nearby universe, particularly to star-
forming regions or galaxies. The prompt emission is incon-
sistent with a collapsar origin, and gravitational wave
observations exclude a compact merger involving neutron
stars and/or black holes. The event rates, quantified below, are
in excess of the majority of energetic astrophysical transients
but consistent with predictions from the known MGFs. The
properties of the prompt emission are distinct from the larger
Figure 3. Lightcurves of the candidate extragalactic MGFs in order of significance from Table 1. These are from Konus-Wind and plotted with 2 ms resolution
(Frederiks et al. 2007b; Mazets et al. 2008; Svinkin et al. 2021), with GRB 070222 reported here for the first time. While GRBs 200415A and 051103 are strikingly
similar (Svinkin et al. 2021)and GRB 070201 is broadly consistent with a single emission episode, GRB 070222 has two temporally and spectrally distinct pulses (see
Appendix B), suggesting varied behavior.
6
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

short GRB population but again consistent with the properties
from the known MGFs. There is additional evidence for
individual events in partner analyses. We conclude that we
have confirmed a sample of extragalactic MGFs that match
prior predictions on detection rates and properties from both
theoretical and observational studies.
A remaining question is, why have we not identified MGFs
to greater distances? Previously, MGFs were thought to be
detectable to tens of megaparsecs. The spectra of the initial
pulse of GRBs 200415A, 051103, and GRB 070222 are
particularly spectrally hard, with a shallow spectral index and
high peak energies, which is consistent with GRB 041227
(Frederiks et al. 2007a). Assuming a cutoff power-law
spectrum for bright MGFs with a low-energy spectral index
of ≈0.0 and peak energies of ≈1.5 MeV produces only 15%–
20% of the photons in the nominal triggering energy range of
50–300 keV, as compared to a typical short GRB (assuming an
index of 0.4 and peak energy of 0.6 MeV; Goldstein et al.
2017). The GRB monitors are triggered by photon counts,
which suggests that the harder spectrum reduces the detectable
distance by a factor of ∼5 and therefore the volume by a factor
of more than 100. Instrument-specific comments are given in
Appendix A. Further, there is a local overdensity within
∼5 Mpc of CCSNe (Mattila et al. 2012), which provides an
additional explanation of detections within this range and the
lack of detections beyond it.
4. Inferences
We now proceed to make population-level inferences
utilizing the three known MGFs and treating all four of our
events as extragalatic MGFs.
4.1. Intrinsic Energetics Distribution
The power-law distribution of the energetics of normal SGR
flares gave hints of the physical process that produces them
(Cheng et al. 1996). Thus, it is interesting to measure the slope
of the E
iso
distribution for MGFs. We assign our search volume
and detection threshold by empirical means, selecting
2.0 ×10
−6
erg cm
−2
for the IPN and a maximal detection
distance of ∼5 Mpc. We further restrict our sample to the past
27 yr, where we have sufficient sensitivity to extragalactic
events, leaving the six most recent bursts (excluding
GRB 790305B).
We assume the same power-law functional form for the E
iso
PDF as our search method; however, we cannot utilize the
maximum-likelihood estimate because it requires the assump-
tion that the observed sample is complete, which is not true for
MGFs at extragalactic distances. Instead, we simulate a large
number of extragalactic MGFs by drawing E
iso
from PDFs over
a range of αvalues, assigning them to specific host galaxies
weighted by their SFR, and setting the event distance as the
host galaxy distance. Events that would be detected are those
where the sampled E
iso
and distance produce a flux greater than
our detection threshold. Here =´
E
3.7 10
iso, min 44 erg is
determined by sampling the Kolmogorov–Smirnov test statistic
value over a range of viable options (Bauke 2007). Then, we
calculate an Anderson–Darling k-sample value for a range of
potentially viable αvalues. We take the 5% rejection values as
the bounds on a 90% confidence interval and determine the
mean assuming a symmetric Gaussian distribution, giving
α=1.7 ±0.4. We note that this is consistent with the reported
slope values of 5/3(Cheng et al. 1996)and 1.9 (Götz et al.
2006)recurrent flares from galactic SGRs.
4.2. Rates
Utilizing the same sample and selection as above, we can
constrain the intrinsic volumetric rate of MGFs. The dominant
sources of uncertainty are the Poisson uncertainty and the
imprecisely known sample completeness. The latter is limited
by the uncertainty on the power-law index of the intrinsic
energetics function, where for a steep index, the majority of
events will be missed (with most events below 1.0 ×10
45
erg
missed in our sample volume), and for a shallow index, most
events are recovered. The αdistribution is taken as a Gaussian.
The SFR within 5 Mpc is 35.5 M
e
yr
−1
, which is scaled to a
volumetric rate by considering the total SFR within 50 Mpc,
which is ∼4000 M
e
yr
−1
from our galaxy sample. We infer a
volumetric rate of =´
-
+
R
3.8 10
MGF 3.1
4.0
5
Gpc
−3
yr
−1
.
4.3. Magnetar Formation Channel
Magnetars may be generated in a variety of events, including
common CCSNe, low-mass mergers (Price & Rosswog 2006),
a rare evolution of white dwarfs (Dessart et al. 2007), or a rare
subtype of CCSN such as collapsars or superluminous
supernovae (Nicholl et al. 2017). Each of these is consistent
with the observed association of magnetars with supernova
remnants (Beniamini et al. 2019). Low-mass merger events
Figure 4. Key parameter comparison of the extragalactic MGF candidates
against the wider short GRB population and known MGFs. The top panel
shows the time to peak Konus distributions, and the bottom panel shows the
E
iso
distributions. The only comparable E
iso
value for a burst from a neutron
star merger is the off-axis GRB 170817A.
7
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

have long inspiral times and should track total stellar mass
rather than the current SFR, which is disfavored given our
model preference for SFR over stellar mass and the discovery
of the first MGF from the Large Magellanic Cloud. A CCSN
origin would arise from regions with high rates of star
formation. This is consistent with our observations and
bolstered by both the lack of detections beyond 5 Mpc due to
the local SFR overdensity and the detection of GRB 790305B
from the low-mass, star-forming Large Magellanic Cloud. The
host galaxies of our extragalactic sample and the Milky Way
itself have larger mass and higher metallicity than is typically
seen in hosts of collapsars or superluminous supernovae
(Taggart & Perley 2019). Therefore, the types of host galaxies
favor common CCSNe as the dominant formation channel of
magnetars.
Additional support for this conclusion is provided from the
event rates. We can relate our inferred MGF rates to progenitor
formation rates as R
MGF
=R
event
f
M
τ
active
r
MGF/M
(Tendulkar
et al. 2016), where R
event
is the rate of events that may form
magnetars, f
M
is the fraction that successfully forms magnetars,
τ
active
is the timescale on which magnetars can produce MGFs,
and r
MGF/M
is the rate of MGFs per magnetar. We take
τ
active
≈10
4
yr, limited by the decay of the magnetic field
(Beniamini et al. 2019). Given the incompleteness of our
known magnetar sample and lack of understanding as to which
magnetars can produce MGFs, we use only the three known to
be capable of producing MGFs to estimate an upper bound of
r
MGF/M
<0.02 yr
−1
magnetar
−1
. We note that this is
significantly weaker than those reported in the literature that
consider all known SGRs, being ∼1×10
−4
yr SGR
−1
(e.g.,
Ofek 2007; Svinkin et al. 2015).
Of the discussed formation channels, only CCSNe are
expected to track star-forming regions and have a comparable
rate, being 7 ×10
4
Gpc
−3
yr
−1
in the local universe (Li et al.
2011a).Afiducial value of f
M
is 0.4 with a 2σconfidence
interval of 0.12–1.0 (Beniamini et al. 2019); other estimates
range between 0.01 and 0.1 (e.g., Woods & Thompson 2004;
Gullón et al. 2015). We require either that some magnetars
produce multiple MGFs or that both f
M
≈1 and the true rate of
R
MGF
are near our 95% lower bound. Alternatively, using the
CCSN rate and the 95% lower limit on R
MGF
, we can place
observational constraints using our results of f
M
>0.005,
further excluding particularly rare subtypes of and favoring
common CCSNe as the dominant formation channel of
magnetars.
5. Conclusions
We summarize our conclusions as follows.
1. We have shown that four short GRBs that occurred
within ∼5 Mpc are the closest events by an order of
magnitude in distance. Our analysis was the first to
identify GRB 070222 as a local event.
2. They are inconsistent with a collapsar or neutron star
merger origin.
3. Their prompt emission is inconsistent with the properties
of cosmological GRBs but consistent with the observa-
tions of the known MGFs.
4. They originate from star-forming regions or galaxies,
including those with metallicity that prevents collapsars
from occurring.
5. All together, this matches expectations for an MGF
origin, which appears to produce 4 out of 250 events.
This would be ∼2% of detected short GRBs (consistent
with the 1%–8% range from the literature; Ofek 2007;
Svinkin et al. 2015)or ∼0.3% of all detected GRBs.
6. Modeling the intrinsic energetics distribution of MGFs as
a power law constrains the index to be 1.7 ±0.4.
7. The volumetric rates are =´
-
+
R
3.8 10
MGF 3.1
4.0
5
Gpc
−3
yr
−1
.
8. The rates and host galaxies of these events favor CCSNe
as the dominant formation channel for magnetars,
requiring at least 0.5% of CCSNe to produce magnetars.
9. We estimate the rate of MGFs per magnetar to be 0.02
yr
−1
.
10. Our results suggest that some magnetars produce multiple
MGFs; this would be the first known source of
repeating GRBs.
11. GRB 070222 suggests that MGFs can have multiple
pulses.
12. The MGFs may not be detectable to tens of megaparsecs
with existing instruments due to their spectral hardness.
Our analysis suggests that additional extragalactic MGFs may
be identified with improved analysis, but “smoking-gun”
confirmation likely requires future instruments. The inferred
rates are sufficiently high that they may contribute to the
stochastic background of gravitational waves. This, along with
the recent observations of a fast radio burst to lower-energy
gamma-ray flares from magnetars (Bochenek et al. 2020;Li
et al. 2020; Marcote et al. 2020; Ridnaia et al. 2020), suggests
that the coming years will bring new insights into the physics
and emission of magnetars.
N.C. is supported by NSF grant PHY-1806990. The Fermi-
GBM Collaboration acknowledges the support of NASA in the
United States under grant NNM11AA01A and DRL in
Germany. The CLU galaxy list made use of the NASA/IPAC
Extragalactic Database (NED), which is funded by the National
Aeronautics and Space Administration and operated by the
California Institute of Technology, and was supported by the
Global Relay of Observatories Watching Transients Happen
(GROWTH)project funded by the National Science Founda-
tion under PIRE grant No. 1545949.
Appendix A
We present rough estimates for the maximal detection
distance of bright MGFs with representative active instruments.
Konus-Wind can detect bright MGFs to ∼13–16 Mpc, based
on GRBs 051103 and 200415A (Svinkin et al. 2021). This can
be taken as the approximate detection distance of the IPN
(Svinkin et al. 2015). The following investigations assume a
hard spectrum based on the time-integrated values for the most
energetic bursts, with a low-energy spectral index of ≈0.0 and
peak energies of ≈1.5 MeV. This has only 15% (20%)of the
number of photons over the 15–150 keV (50–300 keV)energy
range, reducing the detection distance by ∼5×and thus the
volume by >100.
The GBM GRB trigger algorithms cover 50–300 keV, where
the short GRB sensitivity is usually quoted over the 64 ms
timescale. With the assumed spectral and energetics values, the
GBM would have only triggered these onboard algorithms out
8
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

to ∼15–20 Mpc. At greater distances, only the peak flux
interval would be visible, which would be spectrally harder and
reduce this distance. The GBM localizations alone are
insufficient to associate events with any specific burst.
Ground-based searches for GRBs and terrestrial gamma-ray
flashes should be able to recover additional events but may
require confirmation on other GRB instruments
The INTEGRAL SPI-ACS and IBIS are especially sensitive
to hard and short bursts, and additional extragalactic MGFs
have likely triggered the SPI-ACS real-time pipeline in the
past. However, SPI-ACS and IBIS lack the capacity to
discriminate extragalactic MGFs from high cosmic-ray effects
that appear similar to real events in these instruments. The real-
time IBAS pipeline has not been tuned to favor short and hard
events. We estimate that SPI-ACS would record sufficient
signal from extragalactic MGFs for association with another
instrument up to 25–35 Mpc but would only independently
report much brighter events out to 15–20 Mpc. The sensitivity
of IBIS is close to or better than that of SPI-ACS in about 10%
of the sky, and in the majority of directions, IBIS would only
yield detectable signal for extragalactic MGF flares out to at
most 10 Mpc. However, PICsIT may often be more suitable for
triangulation, owing to better time resolution, and can provide
some spectral characterization.
Swift/BAT has >500 different rate trigger criteria running in
real time on board, continuously sampling and testing trigger
timescales from 4 ms up to 64 s, each of which is evaluated for
36 different combinations of energy ranges and focal plane
regions. While the BAT detector is sensitive to photons with
energies up to 500 keV, the transparency of the lead tiles in the
mask above 200 keV limits its imaging energy range (necessary
for a successful autonomous trigger)to 15–150 keV. This
narrow and low energy range limits the BAT’s sensitivity to
hard events, such as MGFs, despite its high effective area. Due
to the number and complexity of the onboard triggering
algorithms, the varying computer load on the BAT CPU, the
evolving state of the BAT detector array, and the changing
operational choices for trigger vetoes/thresholds, modeling the
likelihood of an onboard autonomous trigger is quite difficult.
In addition, due to the BAT’s high effective area, continuous
time-tagged event data cannot be downlinked, making it
difficult to assess the relative completeness of the triggering
algorithms versus ground searches, though this is partly
ameliorated by GUANO (Tohuvavohu et al. 2020). Under
the assumed energetics and spectral values, we estimate that as
of 2020 (averaging half of the original detector array online),
Swift/BAT should reliably trigger on MGFs out to ∼25 Mpc in
the highest coded region of its field of view. Ground analyses
in the downlinked BAT event data can extend this, but the
availability of these data will often depend on an external
trigger (e.g., GUANO). We note that operational changes to the
BAT onboard triggering thresholds with the goal of increasing
sensitivity to extragalactic MGFs and local low-luminosity
GRBs have been previously attempted. In 2012, the threshold
for a successful trigger from an image was lowered from the
usual value of 6.5–5.7, with the condition that triggers in this
range had to be localized to within 12′projected offset from a
local cataloged galaxy stored in the BAT onboard catalog. No
local GRB-like source was ever identified in this program.
Appendix B
As GRB 070222 has not been reported elsewhere, we
describe its basic analysis here. The event was detected by
Konus-Wind, HEND on Mars Odyssey, and both SPI-ACS and
PICsIT on INTEGRAL. Combination of the two best annuli
produces a localization with a 90% containment region of
0.004 deg
2
. This location and its consistency with M83 are
shown in Figure 5.
This burst is distinct from the separate candidates as having
two separate pulses. Time-resolved analysis of this burst is
summarized in Table 2, while time-integrated analysis is
reported in the Second Konus GRB Catalog (Svinkin et al.
2016).
ORCID iDs
E. Burns https://orcid.org/0000-0002-2942-3379
G. Younes https://orcid.org/0000-0002-7991-028X
A. Ridnaia https://orcid.org/0000-0001-9477-5437
D. Cook https://orcid.org/0000-0002-6877-7655
S. B. Cenko https://orcid.org/0000-0003-1673-970X
R. Aloisi https://orcid.org/0000-0003-2822-616X
G. Ashton https://orcid.org/0000-0001-7288-2231
M. Baring https://orcid.org/0000-0003-4433-1365
D. Frederiks https://orcid.org/0000-0002-1153-6340
A. Goldstein https://orcid.org/0000-0002-0587-7042
C. M. Hui https://orcid.org/0000-0002-0468-6025
Figure 5. Localization of GRB 070222 compared to the position and angular
size of M83.
Table 2
The Time-resolved Analysis of the Two Pulses of GRB 070222
T
start
T
stop
Index E
peak
Flux
(s)(s)(keV)(1×10
−6
erg s
−1
cm
−2
)
−0.006 0.012 -
+
0.14 0.24
0.2
8
-
+
7
33 99
138 -
+
1
53.4 16.5
21,2
0.026 0.038 --
+
0.27 0.36
0.48 -
+
1
93 14
25 -
+
2
4.5 3.0
3.0
Note. Errors are quoted at 68% confidence. The main pulse is spectrally hard,
similar to the time-integrated fits of GRB 200415A and GRB 051103.
9
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.

D. L. Kaplan https://orcid.org/0000-0001-6295-2881
M. M. Kasliwal https://orcid.org/0000-0002-5619-4938
D. Kocevski https://orcid.org/0000-0001-9201-4706
O. J. Roberts https://orcid.org/0000-0002-7150-9061
V. Savchenko https://orcid.org/0000-0001-6353-0808
A. Tohuvavohu https://orcid.org/0000-0002-2810-8764
P. Veres https://orcid.org/0000-0002-2149-9846
C. A. Wilson-Hodge https://orcid.org/0000-0002-
8585-0084
References
Aasi, J., Abbott, B., Abbott, R., et al. 2014, PhRvL,113, 011102
Abadie, J., Abbott, B., Abbott, T., et al. 2012, ApJ,755, 2
Abbott, B., Abbott, R., Adhikari, R., et al. 2008, ApJ,681, 1419
Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, ApJL,848, L13
Aptekar, R., Frederiks, D., Golenetskii, S., et al. 1995, SSRv,71, 265
Ashton, G., Burns, E., Dal Canton, T., et al. 2018, ApJ,860, 6
Barat, C., Chambon, G., Hurley, K., et al. 1979, A&A, 79, L24
Barat, C., Hayles, R. I., Hurley, K., et al. 1983, A&A, 126, 400
Barthelmy, S. D., Barbier, L. M., Cummings, J. R., et al. 2005, SSRv,120, 143
Bauke, H. 2007, EPJB,58, 167
Beniamini, P., Hotokezaka, K., van der Horst, A., & Kouveliotou, C. 2019,
MNRAS,487, 1426
Bochenek, C. D., Ravi, V., Belov, K. V., et al. 2020, arXiv:2005.10828
Botticella, M., Smartt, S., Kennicutt, R., et al. 2012, A&A,537, A132
Briggs, M. S., Pendleton, G. N., Kippen, R. M., et al. 1999, ApJS,122, 503
Cheng, B., Epstein, R. I., Guyer, R. A., & Young, A. C. 1996, Natur,382, 518
Cline, T., Desai, U., Pizzichini, G., et al. 1980, ApJL,237, L1
Cline, T., Desai, U., Teegarden, B., et al. 1982, ApJL,255, L45
Cook, D. O., Kasliwal, M. M., Van Sistine, A., et al. 2019, ApJ,880, 7
Dessart, L., Burrows, A., Livne, E., & Ott, C. D. 2007, ApJ,669, 585
Eichler, D., Livio, M., Piran, T., & Schramm, D. N. 1989, Natur,340, 126
Evans, W., Klebesadel, R., Laros, J., et al. 1980, ApJL,237, L7
Fishman, G., Meegan, C., Wilson, R., et al. 1989, in Proc. GRO Science
Workshop, GSFC 2
Fong, W.-f., Berger, E., Margutti, R., & Zauderer, B. A. 2015, ApJ,815, 102
Frederiks, D. D., Golenetskii, S. V., Palshin, V. D., et al. 2007a, AstL,33, 1
Frederiks, D. D., Palshin, V. D., Aptekar, R. L., et al. 2007b, AstL,33, 19
Galama, T. J., Vreeswijk, P., Van Paradijs, J., et al. 1998, Natur,395, 670
Gehrels, N., Norris, J., Barthelmy, S., et al. 2006, Natur,444, 1044
Goldstein, A., Fletcher, C., Veres, P., et al. 2020, ApJ,895, 40
Goldstein, A., Preece, R. D., Mallozzi, R. S., et al. 2013, ApJS,208, 21
Goldstein, A., Veres, P., Burns, E., et al. 2017, ApJL,848, L14
Gorski, K. M., Hivon, E., Banday, A. J., et al. 2005, ApJ,622, 759
Götz, D., Mereghetti, S., Molkov, S., et al. 2006, A&A,445, 313
Grupe, D., Gronwall, C., Wang, X.-Y., et al. 2007, ApJ,662, 443
Gullón, M., Pons, J. A., Miralles, J. A., et al. 2015, MNRAS,454, 615
Hakkila, J., Horváth, I., Hofesmann, E., & Lesage, S. 2018, ApJ,855, 101
Hurley, K., Boggs, S., Smith, D., et al. 2005, Natur,434, 1098
Hurley, K., Cline, T., Mazets, E., et al. 1999, Natur,397, 41
Hurley, K., Rowlinson, A., Bellm, E., et al. 2010, MNRAS,403, 342
Israel, G., Belloni, T., Stella, L., et al. 2005, ApJL,628, L53
Karachentsev, I. D., & Kaisina, E. I. 2013, AJ,146, 46
Kennicutt, R. C., Jr. 1998, ARA&A,36, 189
Klebesadel, R. W., Strong, I. B., & Olson, R. A. 1973, ApJL,182, L85
Kouveliotou, C., Meegan, C. A., Fishman, G. J., et al. 1993, ApJL,413, L101
Lee, J. C., De Paz, A. G., Kennicutt, R. C., Jr., et al. 2010, ApJS,192, 6
Lehmer, B., Tyler, J., Hornschemeier, A., et al. 2015, ApJ,806, 126
Leroy, A. K., Sandstrom, K. M., Lang, D., et al. 2019, ApJS,244, 24
Li, C., Lin, L., Xiong, S., et al. 2020, arXiv:2005.11071
Li, W., Chornock, R., Leaman, J., et al. 2011a, MNRAS,412, 1473
Li, W., Leaman, J., Chornock, R., et al. 2011b, MNRAS,412, 1441
Licquia, T. C., & Newman, J. A. 2015, ApJ,806, 96
Lien, A., Sakamoto, T., Barthelmy, S. D., et al. 2016, ApJ,829, 7
Mandhai, S., Tanvir, N., Lamb, G., Levan, A., & Tsang, D. 2018, Galax,6, 130
Marcote, B., Nimmo, K., Hessels, J., et al. 2020, Natur,577, 190
Mattila, S., Dahlén, T., Efstathiou, A., et al. 2012, ApJ,756, 111
Mazets, E., Aptekar, R., Butterworth, P., et al. 1999a, ApJL,519, L151
Mazets, E., Cline, T., Aptekar, R., et al. 1999b, arXiv:astro-ph/9905196
Mazets, E., Golenetskii, S., Gurian, I. A., & Ilinskii, V. 1982, Ap&SS,84, 173
Mazets, E., Golenetskii, S., Il’Inskii, V., Guryan, Y. A., et al. 1979, Natur,
282, 587
Mazets, E. P., Aptekar, R. L., Cline, T. L., et al. 2008, ApJ,680, 545
Meegan, C., Fishman, G., Wilson, R., et al. 1992, Natur,355, 143
Meegan, C., Lichti, G., Bhat, P. N., et al. 2009, ApJ,702, 791
Messick, C., Blackburn, K., Brady, P., et al. 2017, PhRvD,95, 042001
Metzger, M., Djorgovski, S., Kulkarni, S., et al. 1997, Natur,387, 878
Morrissey, P., Conrow, T., Barlow, T. A., et al. 2007, ApJS,173, 682
Newman, M. E. 2005, ConPh,46, 323
Nicholl, M., Guillochon, J., & Berger, E. 2017, ApJ,850, 55
Ofek, E., Muno, M., Quimby, R., et al. 2008, ApJ,681, 1464
Ofek, E. O. 2007, ApJ,659, 339
Ofek, E. O., Kulkarni, S., Nakar, E., et al. 2006, ApJ,652, 507
Olausen, S., & Kaspi, V. 2014, ApJS,212, 6
Paciesas, W. S., Meegan, C. A., Pendleton, G. N., et al. 1999, ApJS,122, 465
Palmer, D. M., Barthelmy, S., Gehrels, N., et al. 2005, Natur,434, 1107
Pal’Shin, V., Hurley, K., Svinkin, D., et al. 2013, ApJS,207, 38
Paturel, G., Petit, C., Prugniel, P., et al. 2003, A&A,412, 45
Perley, D. A., Meyers, J., Hsiao, E., et al. 2010, GCN, 10429, 1
Popov, S. B., & Stern, B. 2006, MNRAS,365, 885
Price, D. J., & Rosswog, S. 2006, Sci,312, 719
Ridnaia, A., Svinkin, D., Frederiks, D., et al. 2020, arXiv:2005.11178
Roberts, O. J., Veres, P., Baring, M. G., et al. 2021, Natur,589, 207
Siegel, D. M., Barnes, J., & Metzger, B. D. 2019, Natur,569, 241
Strohmayer, T. E., & Watts, A. L. 2005, ApJL,632, L111
Svinkin, D., Frederiks, D., Aptekar, R., et al. 2016, ApJS,224, 10
Svinkin, D. S., Hurley, K., Aptekar, R. L., Golenetskii, S. V., &
Frederiks, D. D. 2015, MNRAS,447, 1028
Svinkin, D., Frederiks, D., Hurley, K., et al. 2021, Natur,589, 211
Taggart, K., & Perley, D. 2019, arXiv:1911.09112
Tendulkar, S. P., Kaspi, V. M., & Patel, C. 2016, ApJ,827, 59
Tikhomirova, Y. Y., Pozanenko, A., & Hurley, K. 2010, AstL,36, 231
Tohuvavohu, A., Kennea, J. A., DeLaunay, J., et al. 2020, ApJ,900, 35
Tsvetkova, A., Frederiks, D., Golenetskii, S., et al. 2017, ApJ,850, 161
von Kienlin, A., Meegan, C., Paciesas, W., et al. 2020, ApJ,893, 46
Watts, A. L., & Strohmayer, T. E. 2006, ApJL,637, L117
Woods, P., & Thompson, C. 2004, arXiv:astro-ph/0406133
Wright, E. L., Eisenhardt, P. R., Mainzer, A. K., et al. 2010, AJ,140, 1868
Younes, G., Baring, M. G., Kouveliotou, C., et al. 2017, ApJ,851, 17
10
The Astrophysical Journal Letters, 907:L28 (10pp), 2021 February 1 Burns et al.