General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
You may not further distribute the material or use it for any profit-making activity or commercial gain
You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from orbit.dtu.dk on: Aug 10, 2022
NuSTAR observations of magnetar 1e 1841-045
An, Hongjun; Hascoet, Romain; Kaspi, Victoria M.; Beloborodov, Andrei M.; Dufour, Francois; Gotthelf,
Eric V.; Archibald, Robert; Bachetti, Matteo; Boggs, Steven E.; Christensen, Finn Erland
Total number of authors:
21
Published in:
Astrophysical Journal
Link to article, DOI:
10.1088/0004-637X/779/2/163
Publication date:
2013
Document Version
Publisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA):
An, H., Hascoet, R., Kaspi, V. M., Beloborodov, A. M., Dufour, F., Gotthelf, E. V., Archibald, R., Bachetti, M.,
Boggs, S. E., Christensen, F. E., Craig, W. W., Greffenstette, B. W., Hailey, C. J., Harrison, F. A., Kitaguchi, T.,
Kouveliotou, C., Madsen, K. K., Markwardt, C. B., Stern, D., ... Zhang, W. W. (2013). NuSTAR observations of
magnetar 1e 1841-045. Astrophysical Journal, 779(2). https://doi.org/10.1088/0004-637X/779/2/163
The Astrophysical Journal, 779:163 (9pp), 2013 December 20 doi:10.1088/0004-637X/779/2/163
C
2013. The American Astronomical Society. All rights reserved. Printed in the U.S.A.
NuSTAR OBSERVATIONS OF MAGNETAR 1E 1841−045
Hongjun An
1
, Romain Hasco
¨
et
2
, Victoria M. Kaspi
1,13
, Andrei M. Beloborodov
2
, Fran¸cois Dufour
1
,
Eric V. Gotthelf
2
, Robert Archibald
1
, Matteo Bachetti
3,4
, Steven E. Boggs
5
, Finn E. Christensen
6
,
William W. Craig
5,7
, Brian W. Greffenstette
8
, Charles J. Hailey
2
, Fiona A. Harrison
8
, Takao Kitaguchi
9
,
Chryssa Kouveliotou
10
, Kristin K. Madsen
8
, Craig B. Markwardt
11
, Daniel Stern
12
,
Julia K. Vogel
7
, and William W. Zhang
11
1
Department of Physics, McGill University, Montreal, Quebec H3A 2T8, Canada
2
Columbia Astrophysics Laboratory, Columbia University, New York, NY 10027, USA
3
Universit
´
e de Toulouse, UPS-OMP, IRAP, Toulouse, France
4
CNRS, Institut de Recherche en Astrophysique et Plan
´
etologie, 9 Av. colonel Roche, BP 44346, F-31028 Toulouse Cedex 4, France
5
Space Sciences Laboratory, University of California, Berkeley, CA 94720, USA
6
DTU Space, National Space Institute, Technical University of Denmark, Elektrovej 327, DK-2800 Lyngby, Denmark
7
Lawrence Livermore National Laboratory, Livermore, CA 94550, USA
8
Cahill Center for Astronomy and Astrophysics, California Institute of Technology, Pasadena, CA 91125, USA
9
RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
10
Space Science Office, ZP12, NASA Marshall Space Flight Center, Huntsville, AL 35812, USA
11
Goddard Space Flight Center, Greenbelt, MD 20771, USA
12
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA
Received 2013 August 12; accepted 2013 October 22; published 2013 December 3
ABSTRACT
We report new spectral and temporal observations of the magnetar 1E 1841−045 in the Kes 73 supernova remnant
obtained with the Nuclear Spectroscopic Telescope Array. Combined with new Swift and archival XMM-Newton
and Chandra observations, the phase-averaged spectrum is well characterized by a blackbody plus double power
law, in agreement with previous multimission X-ray results. However, we are unable to reproduce the spectral
results reported based on Suzaku observations. The pulsed fraction of the source is found to increase with photon
energy. The measured rms pulsed fractions are ∼12% and ∼17% at ∼20 and ∼50 keV, respectively. We detect a
new feature in the 24–35 keV band pulse profile that is uniquely double peaked. This feature may be associated
with a possible absorption or emission feature in the phase-resolved spectrum. We fit the X-ray data using the
recently developed electron–positron outflow model by Beloborodov for the hard X-ray emission from magnetars.
This produces a satisfactory fit, allowing a constraint on the angle between the rotation and magnetic axes of the
neutron star of ∼20
◦
and on the angle between the rotation axis and line of sight of ∼50
◦
. In this model, the soft
X-ray component is inconsistent with a single blackbody; adding a second blackbody or a power-law component
fits the data. The two-blackbody interpretation suggests a hot spot of temperature kT ≈ 0.9 keV occupying ∼1%
of the stellar surface.
Key words: pulsars: individual (1E 1841−045) – stars: magnetars – stars: neutron
Online-only material: color figures
1. INTRODUCTION
Magnetars are isolated neutron stars whose X-ray luminosi-
ties are thought to be powered by the decay of their intense
magnetic fields (Duncan & Thompson 1992; Thompson &
Duncan 1996). They are observed as pulsating X-ray sources
that occasionally produce bright bursts on timescales as short
as 10 ms, as well as major enhancements in the persistent emis-
sion lasting days to months (for reviews, see Woods & Thomp-
son 2006; Mereghetti 2008; Rea & Esposito 2011). Magnetic
fields inferred from magnetar spin-down rates in many cases
exceed 10
14
G (e.g., 1E 1841−045, SGR 1806−20; Vasisht &
Gotthelf 1997; Kouveliotou et al. 1998), although weaker fields
are suggested by recent observations of several magnetars (e.g.,
SGR 0418 + 5729, Swift J1822.3−1606; Rea et al. 2010, 2012;
Livingstone et al. 2011; Scholz et al. 2012). There are 26 mag-
netars that have been discovered to date, including candidates
(see Olausen & Kaspi 2013).
14
13
Lorne Trottier Chair; Canada Research Chair.
14
See the McGill SGR/AXP Online Catalog for a compilation of known
magnetar properties:
http://www.physics.mcgill.ca/∼pulsar/magnetar/main.html.
The X-ray spectra of magnetars often require two or more
components. The soft X-ray component (which has a peak at
∼1 keV) is thought to be dominated by surface emission from
the neutron star and is likely modified by resonant scattering
in the magnetosphere (Thompson et al. 2002). It can be fitted
by an absorbed blackbody plus a power law or sometimes by
a two-blackbody model. The hard X-ray component (which
peaks in a νF
ν
spectral representation above 100 keV; Kuiper
et al. 2006; Enoto et al. 2010) is believed to be generated in the
magnetosphere. Its origin has been discussed by several authors
(Thompson & Beloborodov 2005; Heyl & Hernquist 2005;
Baring & Harding 2007; Beloborodov & Thompson 2007).
Recently, Beloborodov (2013) proposed a detailed model of
hard X-ray emission from the relativistic outflow created by e
±
discharge near the neutron star.
The Galactic magnetar 1E 1841−045 is located at the center
of the shell-type X-ray and radio supernova remnant (SNR)
Kes 73 and was first identified as an anomalous X-ray pulsar
by Vasisht & Gotthelf (1997). Its slow 11.8 s spin period
and rapid spin-down rate imply an extreme magnetic field of
B ≡ 3.2 × 10
19
(P
˙
P )
1/2
G = 6.9 × 10
14
G, assuming the
dipole spin-down model. Hard X-ray emission was detected by
Molkov et al. (2004) and reported by Kuiper et al. (2004, 2006)
1
The Astrophysical Journal, 779:163 (9pp), 2013 December 20 An et al.
Figure 1. NuSTAR images of 1E 1841−045 in the 3–7 keV (left) and 7–79 keV
(right) bands in logarithmic scale. Circles of 1
radius are shown in white. Energy
bands were chosen such that the two images have similar number of events in
the 1
circle; the scale beneath the plots shows the number of events per pixel.
Note that the diffuse Kes 73 emission (R ∼ 2
) is visible in the low-energy
image but not in the high-energy one.
(A color version of this figure is available in the online journal.)
to be highly pulsed, approaching 100% from 15 to 200 keV.
Spectral studies by these authors measured a hard power-law
photon index of ∼1.3 in the ∼20–300 keV band using the Rossi
X-Ray Timing Explorer (RXTE) and International Gamma-Ray
Astrophysics Laboratory (INTEGRAL). However, Morii et al.
(2010) modeled the spectrum obtained with Suzaku with an
absorbed blackbody plus two power laws and found results only
marginally consistent with those of Kuiper et al. (2006).
In this paper, we report on the spectral and temporal properties
of 1E 1841−045 in the 0.5–79 keV band, measured with
the Nuclear Spectroscopic Telescope Array (NuSTAR), the
Swift X-Ray Telescope (XRT), XMM-Newton, and Chandra.
In Section 2, we describe the observations used in this paper,
and in Section 3 we present the results of our data analysis.
In Section 3.4, we apply the model of Beloborodov (2013)
to our measurements of the hard X-ray emission from 1E
1841−045. We show that our spectral fitting yields results
consistent with the expectations from the Beloborodov model.
Section 4 presents our discussion and conclusions. These are
summarized in Section 5.
2. OBSERVATIONS
NuSTAR is the first satellite mission to have focusing capa-
bility above ∼10 keV (Harrison et al. 2013). It is composed
of two sets of focusing optics (Hailey et al. 2010) and two
CdZnTe focal-plane modules (Harrison et al. 2010; each focal-
plane module has four detectors). The observatory operates in
the 3–79 keV band with FWHM energy resolution of 400 eV
at 10 keV, angular resolution of 58
(half-power diameter; 18
FWHM), and temporal resolution of 2 μs (see Harrison et al.
2013 for more details).
We began observing 1E 1841−045 with NuSTAR on 2012
November 9 at UT 22:00:02.184, with a total net exposure of
48.6 ks. Although NuSTAR is extremely sensitive in the hard X-
ray band, a simultaneous 18 ks Swift XRT observation (photon-
counting mode) was conducted at UT 21:49:38.742 on 2012
November 9 to extend the spectral coverage down to ∼0.5 keV,
where the thermal component is dominant.
The NuSTAR data were processed with nupipeline 1.1.1
along with CALDB version 20130509, and the Swift data with
xrtpipeline along with the HEASARC remote CALDB
15
15
See http://heasarc.nasa.gov/docs/heasarc/caldb/caldb_remote_access.html.
Table 1
Summary of Observations
Observatory Obs. ID Obs. Date Exposure Mode
a
(MJD) (ks)
Chandra
b
730 51754 10.5 CC
XMM-Newton 0013340101 52552 3.9 FW/LW
XMM-Newton 0013340102 52554 4.4 FW/LW
Chandra
c
6732 53946 24.9 TE
Swift 00080220003 56240 17.9 PC
NuSTAR 30001025002 56240 48.6 ···
Notes.
a
(PC) Photon counting; (TE) timed exposure; (FW) full window; (LW) large
window; (CC) continuous clocking. MOS1,2/pn data are reported for XMM-
Newton.
b
Used only for 1E 1841−045.
c
Used only for Kes 73, because of pileup.
using the standard filtering procedure (Capalbi et al. 2005)to
produce cleaned event files. We then further processed these
files as described below. We also analyzed archival Chandra and
XMM-Newton observations to have better spectral sensitivity at
low energies (3keV).TheChandra data were reprocessed
using chandra_repro of CIAO 4.4 along with CALDB 4.5.3,
and the XMM-Newton data were processed with the Science
Analysis System (SAS), version 12.0.1. See Figure 1 for
NuSTAR images and Table 1 for a summary of the observations
on which we report.
3. DATA ANALYSIS AND RESULTS
3.1. Timing Analysis
We extracted source events in the 3–79 and 0.5–10 keV bands
within circular regions of radius 60
and 20
for NuSTAR and
Swift, respectively, and applied a barycentric correction to the
event lists using the barycorr tool with the orbit files and the
clock correction files using the position reported by Wachter
et al. (2004). We then used the H-test (de Jager et al. 1989)
to search for pulsations and to measure the period. Pulsations
were detected with very high significance, and the best measured
periods P were 11.79130(2) and 11.7914(2) s for NuSTAR and
Swift, respectively. The uncertainties were estimated using the
method given by Ransom et al. (2002). The periods we measured
are consistent with those predicted on the basis of the ephemeris
obtained with the Swift monitoring program, which will be
described elsewhere (R. Archibald et al. 2013, in preparation).
Since we are measuring the properties of 1E 1841−045, the
Kes 73 background must be subtracted; to do this optimally,
the background region should be within the remnant, which
extends out to 120
in radius from the neutron star. Extracting
backgrounds from a magnetar-free region within Kes 73 was
straightforward in the Swift data, thanks to the XRT’s good
angular resolution; the backgrounds were extracted from an
annular region with inner radius 60
and outer radius 85
.
However, extracting backgrounds was not easy for the NuSTAR
data, since the point-spread function (PSF) is broad, and finding
a source-free region within the remnant was not possible.
Therefore, we extracted the background with inner and outer
radii of 60
and 100
and then corrected for the source
contamination in the background region (see Wang & Gotthelf
1998). The correction factor was calculated with NuSTAR’s
measured instrumental PSF and estimated to be ∼10% (Harrison
et al. 2013).
We also analyzed archival Chandra and XMM-Newton ob-
servations. For these data, source events were extracted from a
2
The Astrophysical Journal, 779:163 (9pp), 2013 December 20 An et al.
Figure 2. Pulse profiles for 1E 1841−045 from NuSTAR data in various energy bands. Note that the y-axis scales differ in the plots.
circle with radius 16
and a ∼3
× 10
rectangle (continuous-
clocking mode, 3
along the event distribution), respectively.
XMM-Newton backgrounds were extracted from events within
an annulus with inner radius 48
and outer radius 80
centered
at the source region, and Chandra backgrounds were extracted
from two ∼5
× 10
rectangular regions offset 5
to each side
from the source. We then applied barycentric corrections to all
the event lists for the temporal analysis below.
We folded the source-event time series at the best measured
period to produce pulse profiles for multiple energy bands.
The background level was subtracted from these pulse profiles.
The background-subtracted pulse profiles obtained with NuS-
TAR are plotted in Figure 2. The energy bands were chosen to
enable comparison with those reported by Kuiper et al. (2004).
For each energy band, the significance of pulsation was greater
than 99%.
The pulse profiles in Figure 2 qualitatively agree well with
those reported by Kuiper et al. (2004). However, we see a
double-peaked pulse profile in the 24–35 keV band. The profile
in this band has not been previously reported. To see if the
apparent double peak was a chance occurrence due to binning
effects, we tried 250 different binnings by varying the zero
phase. For each trial binning, we fitted the profile to two
Gaussian functions and measured the significance of each peak.
In all 250 cases, the significance was greater than 3 σ for both
peaks. Moreover, the two peaks did not disappear when the
energy range was adjusted slightly (e.g., 25–40 keV). Therefore,
we conclude that they are genuine features in the light curve in
this energy band.
In order to better constrain the transition energies of the
feature, we produced pulsed profiles for smaller energy bins
(2 keV). The double-peaked structure is visible to the eye from
∼26 to ∼34 keV, although the structure seen in these individual
profiles may not be statistically significant.
We calculated the rms pulsed fraction, defined as
PF
rms
=
2
4
k=1
a
2
k
+ b
2
k
−
σ
2
a
k
+ σ
2
b
k
a
0
,
Figure 3. Pulsed fraction (rms) in several energy bands measured with the
four X-ray telescopes. Note that the data point at ∼30 keV corresponds to the
double-peaked structure in the pulse profile.
(A color version of this figure is available in the online journal.)
where a
k
= (1/N)
N
i=1
p
i
cos(2πki/N), σ
a
k
is the uncertainty
in a
k
, b
k
= (1/N)
N
i=1
p
i
sin(2πki/N), σ
b
k
is the uncertainty
in b
k
, p
i
is the number of counts in the ith bin, N is the total
number of bins, and n is the number of Fourier harmonics
included, in this case n = 4 (see Gonzalez et al. 2010 for
more details). We also performed similar analyses for the
XMM-Newton and Chandra data, and the measured rms pulsed
fractions are shown in Figure 3. We find that the rms pulsed
fraction exhibits somewhat complicated behavior with energy;
it is ∼20% around 50 keV but increases overall with energy.
We also searched for aperiodic variability with the NuSTAR
data in the energy band from 3 to 79 keV. In particular,
we searched for bursting activity in any energy band during
the observations. We produced light curves with various time
resolutions (0.1–1000 s) for several energy bands (e.g., 3–79,
15–79, 24–35 keV). We then searched for time bins having
significantly larger than average numbers of events, accounting
3
The Astrophysical Journal, 779:163 (9pp), 2013 December 20 An et al.
Table 2
Phenomenological Spectral Fit Results for 1E 1841−045
Phase Data
a
Energy Model
b
N
H
kT Γ
c
s
E
break
/F
s
d
Γ
e
h
F
h
f
L
BB
g
χ
2
/dof
(keV) (10
22
cm
−2
)(keV) (keV)
0.0–1.0 S 0.5–10 BB+PL 2.23(25) 0.46(5) 1.76(39) 1.73(19) ··· ··· 1.58(29) 177/182
0.0–1.0 X, C 0.5–10 BB+PL 2.26(5) 0.42(1) 2.07(7) 1.74(5) ··· ··· 1.61(8) 1866/1849
0.0–1.0 N, S, X, C 0.5–79 BB+BP 2.24(4) 0.44(1) 2.09(4) 10.7(4) 1.33(3) 6.84(6) 1.91(8) 2440/2371
0.0–1.0 N, S, X, C 0.5–79 BB+2PL 2.58(10) 0.42(1) 2.96(18) 1.55(2) 1.06(9) 5.70(9) 1.24(21) 2427/2371
0.15–0.5 N, X, C 0.5–79 BB+BP 2.24
h
0.44(1) 1.98(4) 12.4(9) 1.35(6) 7.50(7) 2.34(11) 819/797
0.5–0.8 N, X, C 0.5–79 BB+BP 2.24 0.44(1) 1.99(5) 12.6(8) 1.18(7) 7.78(9) 1.95(11) 687/652
0.8–1.15 N, X, C 0.5–79 BB+BP 2.24 0.45(1) 2.15(6) 10.0(5) 1.27(5) 5.79(7) 1.69(13) 606/633
0.15–0.5 N, X, C 0.5–79 BB+2PL 2.58 0.42(2) 2.99(13) 1.68(4) 1.19(10) 5.87(14) 1.51(26) 816/797
0.5–0.8 N, X, C 0.5–79 BB+2PL 2.58 0.45(4) 3.04(11) 1.51(4) 1.05(9) 6.45(16) 0.86(23) 680/652
0.8–1.15 N, X, C 0.5–79 BB+2PL 2.58 0.44(2) 2.91(11) 1.37(4) 0.91(11) 4.89(14) 1.33(22) 607/633
Pulsed
i
X, C 0.5–10 PL 2.24 ··· ··· ··· 2.40(15) 0.43(4) ··· 172/299
Pulsed N, X, C 0.5–79 PL 2.24 ··· ··· ··· 1.98(7) 1.31(6) ··· 429/640
Pulsed N 5–79 PL 2.24 ··· ··· ··· 1.70(12) 1.58(15) ··· 163/238
Pulsed N 10–79 PL 2.24 ··· ··· ··· 1.36(23) 1.72(22) ··· 79/114
Pulsed N 15–79 PL 2.24 ··· ··· ··· 0.99(36) 1.76(27) ··· 45/64
Notes. Uncertainties are at the 1 σ confidence level. When combining data from different observatories, cross-normalization factors were used. These were set
to 0.9 for module A of NuSTAR (see Section 3.2), or 1 for XMM-Newton if NuSTAR data were not included. Fluxes were absorption-corrected and measured
using the cflux model in XSPEC.
a
(N) NuSTAR;(S)Swift;(X)XMM-Newton;(C)Chandra.
b
(BB) Blackbody; (PL) power law; (BP) broken power law.
c
Photon index for the soft power law component.
d
Break energy for the broken power law (BB+BP) fit or soft power law flux (BB+2PL) in the 3–79 keV band if NuSTAR data were included. Otherwise,
power-law flux in the 2–10 keV band is listed, in units of 10
−11
erg cm
−2
s
−1
.
e
Photon index for the hard power law component.
f
Flux in units of 10
−11
erg cm
−2
s
−1
. The values are only the (hard) power-law flux in the 3–79 keV band for the BP (PL, 2PL) model when the NuSTAR data
are included; otherwise, power-law flux in the 2–10 keV band is listed.
g
Blackbody luminosity in units of 10
35
erg s
−1
for an assumed distance of 8.5 kpc.
h
N
H
for the phase-resolved and pulsed spectral analysis was frozen.
i
The lstat method in XSPEC was used for fitting pulsed spectra, and we report L-statistic/dof instead of χ
2
/dof.
for the number of trials, but found none. Therefore, we conclude
that there was no bursting activity on timescales of 0.1–1000 s
during the observations.
3.2. Phase-averaged Spectral Analysis
We extracted the source and background events using the
same regions defined in Section 3.1. To see if the Swift
observation was piled up, we measured the count rate within a
circle of radius 20 pixels (∼47
).
19
The count rate was ∼0.4 s
−1
.
Although the count rate was not high enough to produce pileup,
we verified by removing the bright core (2
–4
in radius)
and found that there was no significant spectral change and
thus no pileup. We also analyzed archival Chandra and XMM-
Newton observations to see whether there is long-term spectral
variability in the soft band and to combine with the NuSTAR
observation, to have better spectral sensitivity in the soft band.
We first fitted the Swift data alone to see if there was any
spectral change in the soft band (0.5–10 keV), since the last
Chandra or XMM-Newton measurements were made ∼12 yr
ago (e.g., Morii et al. 2003). We grouped the spectrum to have
at least 20 counts per bin for the fit. We first fitted the data with an
absorbed blackbody plus power law to compare with the archival
XMM-Newton and Chandra data. The spectrum is a little harder
than, but consistent with, previous results (Morii et al. 2003), as
well as with our reanalysis of XMM-Newton plus Chandra data
(see Table 2). We also tried to fit the Swift data with the same
model using the best-fit parameters obtained from modeling
19
See http://www.swift.ac.uk/analysis/xrt/pileup.php.
of the XMM-Newton and Chandra data and found that the
Swift spectrum is consistent with the model. Therefore, all four
observations can be combined with the NuSTAR observation,
and we report fit results for the combined data. Note that for
the blackbody luminosities reported in Table 2,weassumeda
distance of 8.5 kpc based on the H i absorption measurements
of Tian & Leahy (2008).
We then tied all the model parameters between
NuSTAR, Swift, XMM-Newton, and Chandra except for the
cross-normalization factors (set to 0.9 for NuSTAR; the PSF
correction factor) to fit the broadband spectrum (0.5–79 keV).
To fit the data, we grouped the spectra to have at least 100 and
20 counts per bin for NuSTAR and the soft-band instruments
(Swift, XMM-Newton, and Chandra), respectively. We tried to fit
the data simultaneously with a blackbody plus power law. In the
fitting, we used the 0.5–10 and 3–79 keV data for the soft-band
instruments and for NuSTAR, respectively. The model was un-
acceptable, with χ
2
per degrees of freedom (dof) of 2634/2373,
and adding one more component improved the fit significantly.
Therefore, we fitted the data to an absorbed blackbody plus
broken power law, tbabs*(bbody+bknpow), or an absorbed
blackbody plus two power laws, tbabs*(bbody+pow+pow) in
XSPEC 12.7.1. The former is to be compared with the results
of Kuiper et al. (2006) and the latter with those of Morii et al.
(2010). We note that the blackbody component was required in
both models. We show the spectra in Figure 4 and summarize
the results in Table 2.
We studied the effects of nonuniformity in the Kes 73 SNR
background, because the fit results may change depending on
4
