Discovery of a New Soft Gamma Repeater, SGR 1627-41

Abstract

We report the discovery of a new soft gamma repeater (SGR), SGR 1627-41, and present BATSE observations of the burst emission and BeppoSAX NFI observations of the probable persistent X-ray counterpart to this SGR. All but one burst spectrum are well fit by an optically thin thermal bremsstrahlung (OTTB) model with kT values between 25 and 35 keV. The spectrum of the X-ray counterpart, SAX J1635.8-4736, is similar to that of other persistent SGR X-ray counterparts. We find weak evidence for a periodic signal at 6.41 s in the light curve for this source. Like other SGRs, this source appears to be associated with a young supernova remnant G337.0-0.1. Based upon the peak luminosities of bursts observed from this SGR, we find a lower limit on the dipole magnetic field of the neutron star B_dipole > 5 * 10^14 Gauss. Comment: 5 pages, 4 figures, submitted to ApJ Letters

Full text

Page 1

arXiv:astro-ph/9903267v1 17 Mar 1999
submitted to The Astrophysical Journal Letters
DISCOVERY OF A NEW SOFT GAMMA REPEATER, SGR 1627–41
Peter M. Woods1, Chryssa Kouveliotou2,3, Jan van Paradijs1,4, Kevin Hurley5, R. Marc Kippen3,6,
Mark H. Finger2,3, Michael S. Briggs1,3, Stefan Dieters1,3and Gerald J. Fishman3
ABSTRACT
We report the discovery of a new soft gamma repeater (SGR), SGR 1627–41, and present BATSE
observations of the burst emission and BeppoSAX NFI observations of the probable persistent X-
ray counterpart to this SGR. All but one burst spectrum are well fit by an optically thin thermal
bremsstrahlung (OTTB) model with kT values between 25 and 35 keV. The spectrum of the X-ray
counterpart, SAX J1635.8–4736, is similar to that of other persistent SGR X-ray counterparts. We find
weak evidence for a periodic signal at 6.41 s in the light curve for this source. Like other SGRs, this
source appears to be associated with a young supernova remnant G337.0–0.1. Based upon the peak
luminosities of bursts observed from this SGR, we find a lower limit on the dipole magnetic field of the
neutron star Bdipole∼> 5 × 1014Gauss.
1. INTRODUCTION
Soft gamma repeaters (SGRs) are a rare type of stellar
object characterized by their transient emission of bursts
of hard X-rays and soft γ-rays. Bursts have been detected
from three such sources from 1979 (Mazets et al. 1981) un-
til early 1998; two are in the galactic plane (SGR 1806–20,
SGR 1900+14) and one is in the Large Magellanic Cloud
(SGR 0526–66). One of the first SGR bursts detected, the
famous 1979 March 5 burst from SGR 0526–66 (Mazets et
al. 1979), provided a wealth of information about these
sources (Thompson & Duncan 1995). This flare started
with a short initial spike followed by a 3 minute train of
coherent 8 s pulsations. A precise location of this burst was
consistent with a young (∼ 104year) supernova remnant
(SNR) N49 (Cline et al. 1982). The train of pulsations and
the positional coincidence with the SNR indicated that the
burst source is a young, magnetized neutron star.
Pointed X-ray observations of SGR burst location re-
gions have shown that each SGR has associated with it
a persistent X-ray source (Murakami et al. 1994, Hurley
et al. 1999a, Rothschild et al. 1994) within or near a
young SNR (Kulkarni & Frail 1993, Hurley et al. 1999a,
Cline et al. 1982). Furthermore, the persistent sources
associated with the two galactic SGRs are X-ray pulsars
(Kouveliotou et al. 1998a, Hurley et al. 1999a) which show
secular spin down at a rate ∼ 10−10s s−1(Kouveliotou et
al. 1998a, 1999). As argued by Kouveliotou et al. (1998a),
this spin down is likely caused by magnetic dipole radia-
tion which implies a neutron star dipole magnetic field of
∼ 1014−15Gauss. These results have provided strong ob-
servational evidence in support of the idea that SGRs are
strongly magnetized neutron stars or ‘magnetars’ (Duncan
& Thompson 1995).
The majority of SGR bursts have durations less than 200
msec and are well characterized by optically thin thermal
bremsstrahlung (OTTB) spectra with temperatures kT ∼
30 – 40 keV (Kouveliotou 1995). With the exception of the
much more luminous March 5thevent and a similar bright
flare detected recently from SGR 1900+14 (Hurley et al.
1999c), SGR bursts reach peak luminosities up to ∼ 1042
ergs sec−1, far exceeding the Eddington luminosity for a
1.4 M⊙ neutron star. A statistical study of bursts from
SGR 1806–20 has shown that no correlation exists between
the energy released in a burst and the time until the next
burst (Laros et al. 1987). Also, it was found that both
burst peak fluxes and time intervals between bursts re-
semble truncated log-normal and log-normal distributions,
respectively (Laros et al. 1987, Hurley et al. 1994). A dif-
ferential energy distribution of events follows a Gutenberg-
Richter power law (– 1.66 exponent; Gutenberg & Richter
1956) with a maximum energy Emax
(Cheng et al. 1995). Each of these statistical properties
are consistent with characteristics of earthquakes, which
suggests the SGR bursts may be triggered by neutron star
crustquakes (Thompson & Duncan 1995).
Here, we report the discovery of SGR 1627–41, the first
new SGR to be detected since 1979. We provide infor-
mation on general burst characteristics and the persistent
X-ray emission and draw comparisons to other SGRs. Like
the other SGRs, this SGR is associated with a persistent
X-ray source and the young SNR, G337.0–0.1.
≈ 5 × 1041ergs
2.
BATSE OBSERVATIONS
During
SGR 1900+14, the Burst and Transient Source Experi-
aperiodof intenseburstactivity from
1Deptartment of Physics, University of Alabama in Huntsville, Huntsville, AL 35899; peter.woods@msfc.nasa.gov
2Universities Space Research Association
3NASA Marshall Space Flight Center, ES–84, Huntsville, AL 35812
4Astronomical Institute “Anton Pannekoek”, University of Amsterdam, 403 Kruislaan, 1098 SJ Amsterdam, NL
5University of California, Berkeley, Space Sciences Laboratory, Berkeley, CA 94720-7450
6CSPAR (Center for Space Plasma, Aeronomic and Astrophysics Research), University of Alabama in Huntsville, Huntsville, AL 35899
1

Page 2

Discovery of SGR 1627–412
ment (BATSE; Fishman et al. 1989) trigger criteria were
optimized to detect SGR burst events. On 1998 June 15,
three consecutive BATSE triggered bursts short in du-
ration and having soft spectra, originated from a region
of the sky which was inconsistent with the three known
SGR locations. Two days later, BATSE detected another
17 events from the same region, confirming the existence
of a new SGR, SGR 1627–41 (Kouveliotou et al. 1998b;
source name based upon initial BATSE location). Over
the course of the next month and a half, a total of 99
bursts from this source were detected with BATSE, 39 of
which triggered the instrument. Figure 1 shows the ob-
served burst rate as seen with BATSE.
The bursts from SGR 1627–41 last between 25 msec
and 1.8 sec with most burst durations clustering near 100
msec. In Figure 2, we show some representative burst
profiles.This small sample shows the diverse tempo-
ral variability observed in bursts from this source. The
longest event (Figure 2d) is similar to two bursts seen from
SGR 1900+14 with respect to both spectrum and tempo-
ral structure. These bursts are much longer than typical
SGR events and have very smooth temporal profiles with
abrupt γ-ray emission start and end points.
Due to the rapid succession of bursts on June 17 and
18, only limited fine spectral data for a given trigger were
read out from the spacecraft before the next trigger, mak-
ing detailed spectral reconstruction impossible for most
bursts. Of the 39 triggered bursts, fine spectral resolution
data were available for only 8 events, including two very
bright bursts (Figures 2c and 2d). In fact, these two bright
bursts have the two highest peak count rates of any extra-
Solar event ever observed with BATSE. We fit power law,
blackbody and OTTB spectral models to those bursts, and
find the OTTB model best represents the time-integrated
burst spectra of the six dim events and one of the two
bright bursts (Figure 2d).
the six dim events range between 25 – 35 keV; having a
weighted mean of 27 keV. The spectral form of these six
events agrees well with previous modeling of burst spectra
from the other three SGRs (see e.g. Fenimore et al. 1994).
Due to dead-time problems at the peak of the two bright
bursts, we fit spectra taken at the tail of each event. For
the longest event (Figure 2d), we find the tail spectrum
is best fit by an OTTB model with a kT = 27.0 ± 0.4
keV. Two spectra taken from the tail of the brightest
event (Figure 2c) are significantly harder than any other
SGR 1627–41 burst spectrum measured with BATSE. Fur-
thermore, they are not consistent with one another, which
shows spectral evolution exists for this event. The OTTB
and power law spectral models cannot fit the first spec-
trum separately, but a combination of the two yields an
acceptable fit. For this fit, the power law (photon) index
α = −2.07 ± 0.13 and the OTTB kT = 32 ± 1 keV. The
following spectrum taken is at a lower flux level and is
much harder. We find this spectrum can be fit by a simple
power law with an index α = −1.86 ± 0.07. Evidence for
similar hard burst emission from SGR 1806–20 was found
by Strohmayer & Ibrahim (1997). A detailed discussion of
this topic is beyond the scope of this Letter, but it will be
presented elsewhere (Woods et al. 1999a).
In order to estimate the peak fluxes of a larger sample of
bursts, we applied the OTTB model to bursts with coarse
The measured kT values of
spectral resolution (4 channels) and fair temporal resolu-
tion (64 msec). We assumed a fixed kT corresponding to
the measured weighted mean value for the five dim events
(27 keV) and allowed only the normalization (energy flux)
to vary. One drawback to this method is that many bursts
reach their peak flux for only a short time, less than 64
msec, so these peak flux measurements will underestimate
the true peak flux for some events. Given the limited data
availability, however, this time scale provided the largest
sample of events.Figure 3 shows the cumulative peak
flux distribution on the 64 msec time scale for 57 events.
The observed peak fluxes range over 3 orders of magnitude
between 9 × 10−8and 1.1 × 10−4ergs cm−2sec−1. Dead-
time effects for the two brightest events were excessive, so
these peak flux measurements (1.1 and 0.51 × 10−4ergs
cm−2sec−1) can be treated as lower limits. The dashed
line represents a power law fit to this distribution which
has an exponent γ = −0.6 ± 0.1. No turnover is seen
for this distribution out to 1.1 × 10−4ergs cm−2sec−1.
We also constructed a cumulative burst fluence distribu-
tion for these events, which has a slightly flatter slope of
γ = −0.5 ± 0.1. The differential fluence (energy) distri-
bution then has an exponent equal to – 1.5, which agrees
well with the the Gutenberg-Richter power law index (–
1.66).
Using the BATSE triggers, the burst source was coarsely
located (Kouveliotou et al. 1998b) at α = 16h27mand δ =
– 41◦(J2000) with an error circle of radius 2◦. Detection
of SGR events by both BATSE and the Ulysses spacecraft
provided a narrow location annulus 1.7′wide (Hurley et
al. 1998a). Using BATSE Earth occultation constraints,
we limited the allowable range along the annulus to 1.5◦
(Woods et al. 1998; Figure 4). A more detailed account
of the localization of this SGR is reported in Hurley et al.
(1999d) and Smith et al. (1999). In view of the association
of SGRs with young SNRs, we searched the Whiteoak &
Green (1996) catalogue of SNRs near the refined error box.
A single SNR, G337.0–0.1 (Sarma et al. 1997), was found
(Woods et al. 1998) within the 1.5◦×1.7′error box. With
hopes of detecting an X-ray counterpart for this SGR, a
ToO observation of this SNR was initiated using the Bep-
poSAX (Boella et al. 1997a) Narrow Field Instruments
(NFIs).
3. BEPPOSAX OBSERVATIONS
Two observations of SNR G337.0–0.1 were performed
on 1998 August 7 and again on September 16. These ob-
servations revealed a previously undetected X-ray source
(SAX J1635.8–4736) at α = 16h35m49.8sand δ = –47◦
35′44′′(J2000) with an error circle of radius 1′(95% con-
fidence; Figure 4), consistent with the SNR location. A
known source, 4U1630–472, is also seen near the edge of
the field of view for each observation. A light curve of
the new source for each observation does not show any
burst activity which is consistent with BATSE observa-
tions for those time periods (see Figure 1). Using the Low
Energy Concentrator Spectrometer (LECS; Parmar et al.
1997) and two Medium Energy Concentrator Spectrom-
eters (MECS; Boella et al. 1997b), we fit the spectrum
of SAX J1635.8–4736 from 0.1 – 10 keV. The spectrum
is well represented by a power law with interstellar ab-
sorption. Under the assumption that the spectral form

Page 3

3Woods et al.
(i.e. the power law index and Hydrogen column density)
remains constant between observations, we fit the obser-
vations simultaneously allowing only the normalization to
vary between the two. We get an acceptable fit with a
reduced χ2value of 0.92 for 160 degrees of freedom and
find a power law (photon) index α = – 2.5 ± 0.2 and a
column density NH= (7.7 ± 0.8) × 1022cm−2. The unab-
sorbed flux (2 – 10 keV) declines between the observations
(40.3 days) from (6.7 ± 0.3) × 10−12ergs cm−2sec−1
to (5.2 ± 0.4) × 10−12ergs cm−2sec−1. Assuming
this source is located within the SNR, the distance is 11
kpc (Sarma et al. 1997). The source luminosity is then
9.7 × 1034ergs sec−1and 7.6 × 1034ergs sec−1for the
two observations.
Using standard SAX analysis techniques, source counts
were extracted from the combined MECS units for
SAX J1635.8–4736 and binned at 0.5 sec time resolution.
We then performed a Fast Fourier transform (FFT) of the
1998 August light curve searching frequencies from 0 – 1
Hz and found the largest value in the power density spec-
trum was at 0.156 Hz. Although not very significant by
itself, the corresponding period falls within the tight range
of observed periods (5 – 8 sec) for the other SGRs. Us-
ing the barycenter corrected time tags, we ran an epoch
fold search about the period corresponding to the high-
est power, which revealed a marginally significant peak
(6 × 10−3chance probability taking into account the num-
ber of trials; 1500 between 6.38 and 6.44 sec) at 6.41318(3)
sec (JD = 2451032.5). The nearly sinusoidal pulse profile
has an r.m.s. pulse fraction of 10.0 ± 2.6 %. We performed
the same analysis on the 1998 September observation, but
did not find any significant peak in the power density spec-
trum near 0.156 Hz or anywhere else between 0 – 1 Hz.
However, given the weak signal found in August and the
fact that 45% fewer source counts were recorded in the
MECS units during the September observation, we would
not expect to find this pulsed signal.
4. DISCUSSION
We propose that SAX J1635.8–4736 is the X-ray coun-
terpart to SGR 1627–41. There are a number of obser-
vations discussed here which support this claim. First,
the position of SAX J1635.8–4736 is mutually consistent
with the narrow error box for SGR 1627–41 (Hurley et
al. 1999d, Smith et al. 1999) and the SNR G337.0–0.1.
Second, its spectrum is very similar to those found for
the other SGR X-ray counterparts (Hurley et al. 1999a).
Also, this X-ray source is variable near a burst active pe-
riod for SGR 1627–41 which has also been found for the
X-ray counterpart of SGR 1900+14 (Hurley et al. 1999a,
Murakami et al. 1999, Woods et al. 1999b). Finally, the
marginal detection of pulsations at 6.4 sec, if confirmed,
would agree well with the known spin periods of the three
other SGRs which fall within a tight range of 5 – 8 sec
(Kouveliotou et al. 1998a, Hurley et al. 1999a, Mazets et
al. 1979).
For the distance of 11 kpc, G337.0–0.1 has a small di-
ameter of ∼ 5 pc (Sarma et al. 1997). There are only
a few SNRs this small (see Case & Bhattacharya 1998),
and a large fraction of them are very young (e.g. Tycho,
Kepler, Cas A). This suggests that G337.0–0.1 is also very
young.The association of SGR 1627–41 with G337.0–
0.1 strengthens the connection between SGRs and young
SNRs.
Given the distance to the SNR G337.0–0.1 and assuming
isotropic emission, the burst peak luminosities (> 25 keV)
vary from 1039to 1042ergs sec−1(Figure 3). Paczynski
(1992) suggested that SGR 0526-66 may have a critical lu-
minosity ∼ 2 × 1042ergs sec−1. He calculated the relation
between the dipole magnetic field (Bdipole) of a neutron
star and the critical luminosity (Lcrit) the magnetosphere
will allow to escape in the limit where Lcrit ≫ LEdd(LEdd
is the standard Eddington luminosity for a 1.4 M⊙neutron
star). This relation is given by
Lcrit
LEdd
≈ 2
?
Bdipole
1012Gauss
?4/3?
g
2 × 1014cm sec−1
?−1/3
where g is the surface acceleration due to gravity. For
SGR 0526–66, Paczynskifound a dipole magnetic field of ∼
6 × 1014Gauss, which agrees with independent estimates
made by Duncan & Thompson (1992). For SGR 1627–41,
we do not detect a turnover in the cumulative peak lumi-
nosity distribution, but we can place a lower limit on this
value of 1.6 × 1042ergs sec−1based upon the highest ob-
served peak luminosity. This, in turn, places a lower limit
on the magnetic field of the presumed magnetar associated
with SGR 1627–41 of Bdipole∼> 5 × 1014Gauss. This es-
timate may be confirmed with the definitive measurement
of a pulse period and its derivative for SGR 1627–41.
Acknowledgements – We thank other BATSE team mem-
bers for useful discussions of dead-time effects of the in-
strument. This research has made use of SAXDAS lin-
earized and cleaned event files produced at the BeppoSAX
Science Data Center. We would also like to thank Lorella
Angelini for her assistance with analyzing the SAX data.
The MOST is operated by the University of Sydney with
support from the Australian Research Council and the Sci-
ence Foundation for Physics within the University of Syd-
ney. P.M.W. acknowledges support from NASA through
grant NAG 5-3674.J.v.P. acknowledges support from
NASA through grants NAG 5-3764, NAG 5-7060 and NAG
5-7808.
REFERENCES
Boella, G., et al. 1997a, A&AS, 122, 299
Boella, G., et al. 1997b, A&AS, 122, 327
Case, G. & Bhattacharya 1998, ApJ, 504, 761
Cheng, B., et al. 1995, Nature, 382, 518
Cline, T.L., et al. 1982, ApJ, 255, L45
Duncan, R. & Thompson, C. 1992, ApJ, 392, L9
Fenimore et al. 1994, ApJ, 432, 742
Fishman, G.J., et al. 1989, Compton Observatory Science Workshop,
ed., W.N. Johnson, NASA Conference Publication, 2
Gutenberg, B. & Richter, C.F. 1956, Bull. seism. Soc. Am., 46, 105
Hurley, K., Kouveliotou, C., Kippen, M. & Woods, P. 1998a, IAU
Circ 6948
Hurley, K., et al. 1998b, IAU Circ 6966
Hurley, K., Li, P., Kouveliotou, C., Murakami, T., Ando, M.,
Strohmayer, T., van Paradijs, J., Vrba, F, Luginbuhl, C., Yoshida,
A. & Smith, I. 1999a, ApJ, 510, L111

Page 4

Discovery of SGR 1627–414
Hurley, K., Kouveliotou, C., Woods, P., Cline, T., Butterworth, P.,
Mazets, E., Golenetski, S. & Frederics, D. 1999b, ApJ, 510, L107
Hurley, K., et al. 1999c, Nature, 397, 41
Hurley, K., Kouveliotou, C., Woods, P., Mazets, E., Golenetskii,
Cline, T. & van Paradijs, J. 1999d, ApJ submitted
Hurley, K.J., et al. 1994, A&A, 288, L49
Kouveliotou, C. 1995, Ap&SS, 231, 49
Kouveliotou, C., et al. 1998a, Nature, 393, 235
Kouveliotou, C., Kippen, M., Woods, P., Richardson, G., Con-
naughton, V. & McCollough, M. 1998b, IAU Circ 6944
Kouveliotou, C., Strohmayer, T., Hurley, K., van Paradijs, J., Finger,
M., Dieters, S., Woods, P., Thompson, C. & Duncan, R. 1999,
ApJ, 510, L115
Laros, J., et al. 1987, ApJ, 320, L111
Kulkarni, S., & Frail, D. 1993, Nature, 365, 33
Mazets, E.P. 1979, Nature, 282, 587
Mazets, E.P. 1981, Ap&SS, 80, 3
Murakami, T., et al. 1994, Nature, 368, 127
Murakami, T., et al. 1999, ApJ, 510, L119-L122
Paczynski, B., 1992, Acta Astronomica, 42, 145
Parmar, A.N., et al. 1997, A&AS, 122, 309
Rothschild, R., Kulkarni, S. & Lingenfelter, R. 1994, Nature, 368,
432
Sarma, A. et al. 1997, ApJ, 483, 335-339
Smith, D., Bradt, H. & Levine, A. 1999, ApJ submitted
Strohmayer, T. & Ibrahim, A. 1997, Gamma-Ray Bursts:
Huntsville Symposium, AIP Proceeding 428, ed. Meegan, Preece
& Koshut
Thompson, C., & Duncan, R. 1995, MNRAS, 275, 255-300
Whiteoak, J. & Green, A. 1996, A&AS, 118, 329
Woods, P., Kippen, M., van Paradijs, J., Kouveliotou, C., McCol-
lough, M. & Hurley, K. 1998, IAU Circ 6948
Woods, P., et al. 1999a, in preparation
Woods, P., Kouveliotou, C., van Paradijs, J., Finger, M. & Thomp-
son, C. 1999b, ApJ submitted
4th

Page 5

5Woods et al.
Fig. 1.— Burst rate of SGR 1627–41 (triggered and untriggered) as observed with BATSE. In total, 99 bursts were detected within 60 days.
Arrows indicate times of BeppoSAX NFI observations of G337.0–0.1.
Fig. 2.— A sample of six bursts of SGR 1627–41 observed with the BATSE Large Area Detectors (LADs). The upper and lower panels are
Time-Tagged Event (TTE) data (> 25 keV) accumulated with 4 msec (a), 2 msec (b), 2 msec (e) and 1 msec (f) time resolution. The middle
panels are DISCriminator SCience (DISCSC) data (> 25 keV) accumulated with 64 msec time resolution. Trigger times for these six events
are 50981.87322 (a), 50982.03545 (b), 50982.07120 (c), 50982.16969 (d), 50993.30911 (e) and 51006.91017 (f) in MJD (UT). BATSE trigger
numbers are given in the upper left corner of each panel.
Fig. 3.— Cumulative peak flux (0.064 sec) distribution for 57 events. Dashed line is a power law fit to these data having an exponent equal
to – 0.6. Bottom horizontal axis labels the peak flux and the top horizontal axis gives the peak luminosity assuming isotropic emission and an
11 kpc distance to the source.
Fig. 4.— Localization of SGR 1627–41 and SAX J1635.8–4736. Panel (a) shows the Interplanetary Network (IPN) arc (solid lines) and Earth
occultation constraints (shaded region). Panel (b) is a magnification of region near G337.0–0.1 (radio contours; Whiteoak & Green 1996). Solid
straight lines represent BATSE-Ulysses IPN arc. Dotted lines are the Konus-Ulysses IPN arc (Hurley et al. 1998b). Solid, bold circle represents
error region for SAX J1635.8-4736.

Page 6

Page 7

Page 8

Page 9

-49
-48
-47
-46
δ2000 (deg) →
252250
← α2000 (deg)
248246
-47.7
-47.6
-47.5
249.1249.0248.9248.8
(a)
(b)