The Astrophysical Journal, 749:146 (9pp), 2012 April 20 doi:10.

1088/0004-637X/749/2/146

C 2012. The American Astronomical Society. All rights reserved. Printed in the U.S.A.

CHANDRA OBSERVATIONS OF THE OLD PULSAR PSR B1451−68

B. Posselt1 , G. G. Pavlov1,4 , R. N. Manchester2 , O. Kargaltsev3 , and G. P. Garmire1
1 Department of Astronomy & Astrophysics, Pennsylvania State University, 525 Davey Lab, University Park, PA 16802, USA;
[email protected], [email protected]
2 CSIRO Astronomy and Space Science, Australia Telescope National Facility, P.O. Box 76, Epping, NSW 1710, Australia
3 Department of Astronomy, University of Florida, 205 Bryant Space Center, Gainesville, FL 32611, USA
Received 2012 January 10; accepted 2012 February 23; published 2012 April 2

ABSTRACT
We present 35 ks Chandra ACIS observations of the 42 Myr old radio pulsar PSR B1451−68. A point source is
detected 0. 32 ± 0. 73 from the expected radio pulsar position. It has ∼200 counts in the 0.3–8 keV energy range.
We identify this point source as the X-ray counterpart of the radio pulsar. PSR B1451−68 is located close to a
Two Micron All Sky Survey point source for which we derive 7% as the upper limit on the flux contribution to the
measured pulsar X-ray flux. The pulsar spectrum can be described by either a power-law model with photon index
−2
Γ = 2.4+0.4−0.3 and a unrealistically high absorbing column density NH = (2.5−1.3 ) × 10 cm , or by a combination
+1.2 21

of a kT = 0.35−0.07 keV blackbody and a Γ = 1.4 ± 0.5 power-law component for NH = 2.6 × 1020 cm−2 ,
+0.12 DM

estimated from the pulsar dispersion measure. At the parallactic, a Lutz–Kelker bias corrected distance of 480 pc,
−1
the non-thermal X-ray luminosities in the 0.3–8 keV energy band are either Lnonth0.3–8 keV = (11.3 ± 1.7) × 10 erg s
29
−1
or L0.3–8 keV = (5.9−5.0 ) × 10 erg s , respectively. This corresponds to non-thermal X-ray efficiencies of either
nonth +4.9 29
−3 −3
–8 keV = L0.3–8 keV /Ė ∼ 5 × 10 or 3 × 10 , respectively.
nonth nonth
η0.3
Key words: pulsars: individual (PSR B1451−68) – stars: neutron – X-rays: stars
Online-only material: color figures

1. INTRODUCTION to understand the quantitative contributions of magnetospheric

and thermal components.
Old (>1 Myr), rotational-powered pulsars have lost a signif- PSR B1451−68 is an old, nearby pulsar, previously unex-
icant amount of their initial rotation energy. Their spin-down plored in X-rays. The 0.263 s radio pulsar has a characteris-
power, Ė  1034 erg s−1 , is usually several orders of magnitude tic spin-down age τ = P (2Ṗ )−1 = 4.25 × 107 yr, a dipole
lower than those of younger pulsars. The lower energy budget magnetic field strength of 1.63 × 1011 G, and a spin-down
translates into fainter high-energy emission. Only the closest power Ė = 2.1 × 1032 erg s−1 . The proper motion compo-
of the old rotational-powered pulsars can be detected in X-rays. nents of PSR B1451−68 were measured by Bailes et al. (1990):
Since the bulk of the neutron star surface is too cold to be visible μα = −39.5 ± 0.7 mas yr−1 and μδ = −12.3 ± 0.6 mas yr−1 .
in X-rays (Yakovlev & Pethick 2004), one expects to see mainly Unlike most other pulsars, the proper motion is directed to-
the X-rays from the non-thermal magnetospheric emission with ward the Galactic plane. Bailes et al. (1990) measured an an-
a possible small thermal contribution from polar caps (PCs) nual parallax of π = 2.2 ± 0.3 mas and reported a distance of
heated by infalling, accelerated particles (Harding & Muslimov d = 450 ± 60 pc. According to Verbiest et al. (2010) the dis-
2001, 2002). So far, nine old pulsars have been detected at tance corrected for the Lutz–Kelker bias is d = 480+80
−60 pc. The
X-ray wavelengths (e.g., Becker et al. 2004, 2006; Zavlin & spin-down power of PSR B1451−68 is comparable to those of
Pavlov 2004; Tepedelenlioǧlu & Ögelman 2005; Zhang et al. other nearby old pulsars which have been detected in X-rays.
2005; Kargaltsev et al. 2006; Misanovic et al. 2008; Hui & Aiming for constraints on the X-ray spectral properties of
Becker 2008; Gil et al. 2008; Pavlov et al. 2009), several of PSR B1451−68, we report here on Chandra observations of
them having only a few dozen counts. Most of the X-ray spec- this pulsar.
tra can be described by relatively soft power laws (PLs) with
photon indices 2  Γ  4. Some of these spectra can be fitted 2. OBSERVATIONS AND DATA REDUCTION
with blackbody (BB) models or by a combination of BB and
PL components. The BB models usually indicate a rather small PSR B1451−68 was observed with the Chandra X-Ray
emission area compared to the conventional PC sizes whose Observatory on 2010 May 16 using the ACIS-S detector in
radii are calculated as RPC = [2π RNS 3
/(cP )]1/2 , assuming a the VFAINT imaging mode. The nominal exposure time was
magnetic dipole field (Harding & Muslimov 2001, 2002). Here, 35.1 ks. The data reduction was done using CIAO (version 4.3).
RNS is the radius of the neutron star, P is the pulsar period, and c We reprocessed the data in order to apply the VFAINT back-
is the speed of light. There has been a debate as to how significant ground cleaning. We investigated event files with and without
the thermal contributions are (e.g., Becker et al. 2004; Zavlin & event position randomization using the Energy-Dependent Sub-
Pavlov 2004; Misanovic et al. 2008; Hui & Becker 2008). Bet- pixel Event Repositioning (EDSER) algorithm by Li et al. (2004),
ter count statistics, the consideration of new information from as well as the subpixel algorithm by Mori et al. (2001). If not
other wavelengths as well as a larger object sample are required stated otherwise, we will discuss in the following the events
with the pixel position randomization turned off. No large back-
4 Also at St.-Petersburg State Polytechnic University, Polytekhnicheskaya ul. ground flares occurred during the observation, and therefore we
29, 195251, Russia. filtered the reprocessed event file for good time intervals (GTIs)

1
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

Table 1 correspondence with the energy distribution of the ∼200 source

Chandra Source Positions in the Reprojected Event Files counts within an energy range of 0.3–8 keV. The fit results
Method R.A. Decl. are presented in Table 2. We checked whether the VFAINT
(◦ ) (◦ ) background cleaning influenced the spectral properties of the
Event file without randomization
source. Without this correction there are two more source counts
(1%), and the corresponding spectral model fit results are the
wavdetect 224.00006(2) −68.72757(1) same within 1σ .
Centroid, r = 0. 6 224.00007(2) −68.72757(1)
Centroid, r = 2 224.00002(2) −68.72759(1)
3. RESULTS AND DISCUSSION
Event file with EDEF subpixel algorithm employed
3.1. Position and Possible Source Blending
Centroid, r = 0. 6 224.00007(2) −68.72757(1)
Centroid, r = 2 224.00002(2) −68.72758(1) In order to determine the pulsar coordinates, we combined
the Molonglo data used by Siegman et al. (1993) and Parkes
Notes. Count-weighted variances of the centroid positions are listed in brackets data from 1991 to 2006. First, we obtained the dispersion
and refer to the last respective digits. These errors do not include astrometric measure, DM = 8.557 ± 0.014(2σ ) cm−3 pc, using the Parkes
uncertainties.
data which have 400, 600, and 1400 MHz times of arrivals.
This value improves the accuracy of the previously reported
which have less a 5σ deviation from the overall light curve DM = 8.6 ± 0.2(2σ ) cm−3 pc by D’Alessandro et al. (1993).
mean rate. After the GTI filtering the actual exposure time was Fixing the proper motion at the interferometric values of Bailes
34.7 ks. Due to the ACIS frame time of 3.2 s, it is not possible et al. (1990), and using the Molonglo as well as the Parkes data,
to search for X-ray pulsations corresponding to the P = 0.263 s we derive the position of PSR B1451−68 as R.A.(J2000) =
of PSR B1451–68. The wavdetect task was applied to obtain 14:56:00.071(6); decl.(J2000) = −68:43:39.25(5) at epoch
an X-ray source list. A source was detected at the pulsar po- MJD 50135. The quoted uncertainties are twice those given
sition on the back-illuminated chip S3. Using the CIAO task by the least-squares solution. The expected pulsar coordinates
arfcorr we estimate that 95% of the source counts are in- at the epoch of the Chandra observations, MJD 55332.7,
cluded in a r = 2 aperture in an energy range from 0.3 keV are: R.A.PSR (J2000) = 14:55:59.968(8); decl.PSR (J2000) =
to 5 keV. The aperture-corrected source count rate is 0.0059 ± −68:43:39.43(6). The errors reflect 2σ uncertainties including
0.0004 counts s−1 . those due to the proper motion uncertainties.
Since there is a stellar neighbor with small angular separation The measured X-ray source position slightly depends on
(0. 7) from the initial X-ray centroid position, we aimed the method applied, see Table 1. We will use in the fol-
for improving the astrometric solution of the event file. As lowing the wavdetect position in the reprojected event file
reference frames, we used the European Southern Observatory without pixel position randomization. The measured X-ray
(ESO) New Technology Telescope (NTT) SUSI2 observations source position is then R.A.CXO = 14:56:0.015(7) and
of PSR B1451−68 (PI: Bucciantini, obtained as part of their decl.CXO = −68:43:39.25(4). The errors reflect 2σ variances
program 68.D-0249) and the Two Micron All Sky Survey of the wavdetect position.
(2MASS) point-source catalog (PSC; Skrutskie et al. 2006). Close to this position there is a 2MASS point source
The details of the reprojection process are given in Appendix A. which has also been detected in the ESO SUSI2 WB655
Our derived overall absolute astrometric uncertainty of the image as well as other optical surveys. The listed J2000
AM
reprojected Chandra event file is 3σabs = 0. 73, and the relative position of 2MASS 14560002−6843400 is R.A.2MASS =
astrometric uncertainty of Chandra with respect to the 2MASS 14:56:00.028; decl.2MASS = −68:43:40.02 (Cutri et al. 2003).
PSC is 3σrelAM
= 0. 66. 2MASS 14560002−6843400 is also listed in the USNO B1 cat-
We obtained updated wavdetect positions for all X-ray alog (Monet et al. 2003). Roeser et al. (2010) determined the
sources on the S3 chip. The X-ray source at the pulsar location apparent proper motion of this source to be −0.6 ± 8.2 mas yr−1
appears to be point-like. For this source, we also measured the in right ascension and −1.9 ± 8.2 mas yr−1 in declination. Thus,
centroid positions of the flux distribution in r = 2 and r = 0. 6 we can neglect movement of the 2MASS source, and the cal-
apertures for both the event file with just the pixel position culated separation between the expected pulsar position and the
randomization removed and the one with the EDSER subpixel 2MASS point source is 0. 68.
algorithm applied in addition. Statistical centroid position errors The X-ray source position is offset by 0. 26 ± 0. 06 in right
are calculated as the standard error of the mean in the pixel ascension and 0. 17 ± 0. 06 in declination from the expected
coordinates X and Y in the respective aperture and transformed pulsar position. Here, the errors reflect only the 3σ wavdetect
to the world coordinate system. There are no statistically positional variances. In addition, systematic astrometric errors
significant differences between the different positions listed in apply. Our absolute astrometric accuracy is better than 3σabs AM
=
Table 1. 0. 73 (Section 2 and Appendix A). Thus, the Chandra X-ray
We extracted a source spectrum within 2 of the pulsar position is coincident with the expected pulsar position within
centroid position. A background spectrum was extracted from astrometric uncertainties.5
an annulus of 5 to 10 around the same position. The extracted The X-ray source position is offset by 0. 07 ± 0. 06 in right
source spectrum counts have been combined to energy groups ascension and 0. 76 ± 0. 06 in declination from the catalog posi-
of 15 counts each. Applying XSPEC (version 12.6.0) to analyze tion of 2MASS 14560002−6843400. To discuss the position of
the spectrum, we used χ 2 -fitting with standard weighting, the this 2MASS source, only the 3σrel AM
= 0. 66 relative astrometric
tbabs model for the X-ray absorption and abundance tables by 5 The separation between the position of 2MASS 14560002−6843400 and
Wilms et al. (2000) as well as photoelectric absorption cross the optical source detected in the ESO SUSI2 image is only 46 mas,
sections from Balucinska-Church & McCammon (1992) and underlining the tight relation between the SUSI2 and 2MASS (and
Yan et al. (1998). Different spectral models were checked for subsequently the Chandra) astrometry.

2
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

Table 2
XSPEC Fit Results
Model NH Γ PL Norm at 1 keV kT RBB EM Red. χ 2 /dof Funabs
(1020 cm−2 ) (10−6 photons keV−1 cm−2 s−1 ) (keV) (m) (1038 cm−3 ) (10−14 erg cm2 s−1 )
PL 25+12
−13 2.4+0.4
−0.3 13+6
−4 ··· ··· ··· 0.4/10 6.0+2.1
−1.4
BB 12 ··· ··· 0.42+0.06
−0.05 13.2+11.2
−7.2 ··· 1.4/10 2.4+0.4
−0.3
BB+PL 17+26
−17 2.2 ± 1.0 8.2+10
−8.2 0.3 ± 0.2 9.9+38.2
−9.9 ··· 0.5/8 3.8+5.2
−2.8 (PL) / 0.7+1.4
−0.7 (BB)
APEC 6.5+7.9
−4.7 ··· ··· 3.1+1.3
−0.8 ··· 6.6 ± 1.0 0.8/10 3.8+0.7
−0.6

frozen
PL 2.6 1.7 ± 0.2 6.6 ± 0.9 ··· ··· ··· 1.6/11 4.1 ± 0.6
BB 2.6 ··· ··· 0.40+0.06
−0.05 14.1+12.3
−8.3 ··· 1.4/10 2.4+0.4
−0.3
BB+PL 2.6 1.4 ± 0.5 2.7+2.1
−2.7 0.35+0.12
−0.07 13.8+24.2
−12.3 ··· 0.5/9 2.1 ± 1.1 (PL) / 1.4+0.9
−0.7 (BB)

Notes. All errors indicate 90% confidence intervals, the BB emitting area radius errors take the distance error into account. The parallactic distance, corrected
for the Lutz–Kelker bias, D = 480+80
−60 pc, is used. The unabsorbed fluxes, Funabs , are given for the energy range from 0.3 keV to 8 keV.

precision between Chandra and 2MASS has to be considered 0.01

(Section 2 and Appendix A). Considering the X-ray statistical
(wavdetect) positional errors as well as the relative astromet-
ric precision between Chandra and 2MASS, it appears unlikely

Counts s−1 keV−1

that the Chandra X-ray point source with 0. 77 separation is
the counterpart of the 2MASS point source. We discuss spec- 10−3
tral constraints on the nature of 2MASS 14560002−6843400 in
Appendix B.1. To estimate a limit of the possible X-ray flux con-
tribution of 2MASS 14560002−6843400 to the measured X-ray
emission of the pulsar we use different approaches, which are
discussed in detail in Appendix B.2. In short, we measure flux 10−4
percentages in small apertures centered on the X-ray source and 1
Δ in σ

the 2MASS source positions, deconvolve the image applying

the Chandra Ray Tracer (ChaRT), MARX, and the CIAO task 0
arestore, and investigate the count energy distribution of the
−1
X-ray source. Considering (1) the flux percentage measurements 0.5 1 2
in the reprojected event file with or without subpixel algorithm Energy (keV)
applied, (2) the comparison of the deconvolved image with a
deconvolved simulated double source image, (3) the location of
the point-spread function (PSF) asymmetry region, and (4) the
4

hardness of the detected counts toward the 2MASS star, we con-

clude that any contribution of 2MASS 14560002−6843400 to
the measured X-ray flux of the pulsar must be <7%.
Photon Index

3.2. The X-Ray Spectrum of the Pulsar

3

As described in Section 2, we used XSPEC to check different

spectral models for the extracted X-ray source spectrum. As
discussed above, at least 93% of the X-ray source flux is +

attributed to the pulsar. We employ in the following the counts

from the r = 2 extraction region for the pulsar spectrum
2

(198 counts in an energy range from 0.3 keV to 5 keV)

and neglect the possible (small) contribution of the 2MASS
star. The fit results are presented in Table 2. The derived 0.2 0.4 0.6
NH [ 1022 cm−2 ]
absorbed X-ray fluxes for the different spectral models range
from 2.4 × 10−14 erg cm2 s−1 to 2.9 × 10−14 erg cm2 s−1 for Figure 1. X-ray spectrum of PSR B1451−68 and the best-fit-absorbed PL model
energies between 0.3 keV and 5 keV. Allowing all parameters if NH is a free-fit parameter; see the text and Table 2. The lower panel of the
to vary, we tested simple single-component models like a BB upper plot shows the residuals in units of σ . The lower plot shows 68%, 90%,
and 99% confidence contours in the NH − Γ plane of the PL model for the pulsar
or a PL spectrum. An absorbed, NH = 2.5+1.2 −1.3 × 10
21
cm−2 , spectral data.
PL model with photon index Γ = 2.4−0.3 fits the spectrum well
+0.4

(Figure 1), while the obtained BB fit is statistically unacceptable.

In addition, we checked whether the X-ray spectrum can be 2MASS 14560002−6843400 in more detail in Appendix B.1.
described by an optically thin thermal plasma model as one For the pulsar counterpart of the X-ray source, the thin plasma
could expect from a stellar corona. The apec plasma model model is physically irrelevant and not further discussed.
fit is acceptable. It is slightly worse than the PL, but better Among the X-ray spectral fits in Table 2, the PL fit and the
than the BB fit. We discuss the apec parameters with respect to BB+PL fit show the lowest χ 2 values, but the uncertainties

3
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

0.01 case of the PL-only fit is a factor of 10 higher than one

would expect from the dispersion measure. Even accounting
for the fit parameter error and a possible uncertainty factor
Counts s−1 keV−1

of three for the DM-based NH , these values do not overlap.

Therefore, we regard the spectral fits, in particular the BB+PL
10−3
fit, with fixed NH more realistic. In the combined model, the
photon index of Γ = 1.4 ± 0.4 is comparable to the typical
values for younger pulsars (0.6 < Γ < 2.1; Kargaltsev & Pavlov
2008; Gotthelf 2003), but the inferred BB emission area radius,
10−4 13.8+24.2
−12.3 m, is rather small. The conventional PC radius is
1 RPC  280 m, assuming a dipolar magnetic field for this pulsar
and RNS = 10 km.
Δ in σ

0 Similar discrepancies have been found for BB components

of other old pulsars (Pavlov et al. 2009; Misanovic et al. 2008;
−1 Kargaltsev et al. 2006; Zhang et al. 2005). Apart from geomet-
0.5 1 2 rical projection effects, the following explanations have been
Energy (keV) discussed: only a small fraction of the PC is heated by inflowing
Figure 2. X-ray spectrum of PSR B1451−68 and the best-fit-absorbed BB+PL particles at the footpoints of “spark discharges” created above
model in case of frozen NH = 2.6 × 1020 cm−2 . The red solid and blue the PC (Zhang et al. 2005; Ruderman & Sutherland 1975); the
dashed lines visualize the BB model component and PL model component PCs are covered by hydrogen or helium atmosphere resulting in
contributions, respectively. The lower panel shows the residuals in units of σ .
For more details see the text and Table 2.
effective temperatures of a factor two less and a radius a fac-
tor of 3–10 larger than the BB fit (Zavlin & Pavlov 2004); the
(A color version of this figure is available in the online journal.)
nondipolar component of the magnetic field is much stronger
than the dipolar component, causing partially screened acceler-
are very large for the two-component model. Both fits show ation regions and smaller PCs as consequences (Gil et al. 2003).
suspiciously high best-fit values of the hydrogen column density,
3.3. The X-Ray Efficiency of Old Rotation-powered Pulsars
NH ∼ 2 × 1021 cm−2 . This value is close to the Galactic
value for this line of sight (l = 313.◦ 87, b = −8.◦ 54): For PSR B1451–68, the luminosity of the non-thermal (PL)
the LAB Survey of Galactic neutral hydrogen reports 1.6 × X-ray component in the 0.3–8 keV energy band is Lnonth X =
1021 cm−2 toward this direction (Kalberla et al. 2005), the PL
4π d 2 , FX, = 5.9+4.9
× 1029
erg s −1
, where the error is cal-
unabs −5.0
Dickey & Lockman (1990) neutral hydrogen survey reports culated from the 90% confidence error of the unabsorbed flux
2.1 × 1021 cm−2 . The pulsar has a parallactic distance of and the 80 pc uncertainty of the Lutz–Kelker corrected distance.
only 450 pc and a low DM value. Assuming 10 H atoms for Such a luminosity corresponds to a non-thermal X-ray efficiency
each e− , we derive an expected NHDM = 2.6 × 1020 cm−2 ηnonth = Lnonth /Ė ∼ 3 × 10−3 . The X-ray luminosities versus
X
from the pulsar DM = 8.557 ± 0.014 pc cm−3 (Section 3.1). characteristic ages of old pulsars, including PSR B1451–68, are
Freezing NH at = 2.6 × 1020 cm−2 , we obtain new X-ray plotted in Figure 3. There is no obvious characteristic age effect
spectral fits, presented in Table 2. The PL fit is now slightly visible in this plot. We also include PSR B1451−68 in the plot
worse than the BB fit, both are unacceptable. The combined showing the X-ray luminosities versus spin-down energies of
BB+PL model (Figure 2) fits the data best. According to old pulsars (Figure 4). Distances to other old pulsars are up-
the F-test, the combined model fits the data better than the dated according to Deller et al. (2009) and corrected for the
PL-only fit with a probability of 99.4%, and better than the Lutz–Kelker bias (Verbiest et al. 2010). The non-thermal X-ray
BB-only fit with a probability of 99.1%. The inferred BB efficiency of PSR B1451−68 is larger than those of young pul-
emitting area radius is of a similar size as in the case of the sars, most of which have ηnonth values smaller than 10−3 , see
BB-only fit, albeit with larger error. The individual contributions Figure 5. The efficiency of PSR B1451−68 is, however, com-
to the unabsorbed fluxes in the energy range from 0.3 keV parable to those of other old pulsars (Figure 4). While it is
−14
to 8 keV are FX, BB
unabs = (1.4−0.7 ) × 10
+0.9
erg cm2 s−1 and possible that old pulsar spectra with too few counts may have
−14
FX,PL
unabs = (2.1 ± 1.1) × 10 erg cm2 s−1 for the BB and the an undetected thermal contribution (e.g., from heated PCs),
PL components, respectively. Flux errors are large because the Figure 4 shows only little influence of such a thermal com-
fitted component normalization values have high uncertainties ponent on the inferred non-thermal X-ray efficiencies in the
due to the low number of counts. energy range 1–10 keV for the three pulsars with sufficiently
We also checked neutron star (NSA) models (Zavlin et al. large number of counts. This supports the hypothesis that the
1996; Pavlov et al. 1995) with fixed parameters for neutron star conversion of spin-down power into X-ray emission becomes
mass and radius (MNS = 1.4 M , RNS = 12 km). The fits are more efficient as pulsars get less powerful (Zharikov et al. 2006;
marginally acceptable, and the non-magnetized NSA model fit Kargaltsev et al. 2006). There is, however, an observational bias
achieves the lowest χ 2 . However, the fits are worse than the in favor of brighter X-ray sources which can lead to the detection
BB+PL fit. Allowing both the mass and radius to vary results in of only the most efficient old neutron stars.
implausibly low masses and radii. A look on the increasingly complete observational statistics
The inferred X-ray spectral fit parameters are similar to of the closest pulsars can help to minimize the effect of the
those found for other old pulsars (e.g., Pavlov et al. 2009; observational bias in favor of brighter X-ray sources. Instead
Kargaltsev et al. 2006; Becker et al. 2004). From a statistical of the characteristic ages, which may be very different from
viewpoint alone we are unable to differentiate between a the true ages, we prefer to use the directly measurable Ė as
purely magnetospheric emission model and a combination of parameter to differentiate between powerful, young and less
magnetospheric and thermal emission. But the derived NH in powerful, old pulsars. We choose Ė = 9 × 1033 erg s−1 as

4
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

Figure 3. X-ray luminosities and upper limits of 12 old pulsars vs. their Figure 5. Non-thermal X-ray pulsar luminosities and upper limits vs. the pulsar
characteristic ages. In case of a sufficient number of counts, the diamonds spin-down power. This plot is an update of Figure 5 by Kargaltsev & Pavlov
and asterisks show the non-thermal and thermal X-ray luminosities, respec- (2008) including the old pulsars from the previous figure, recycled X-ray pulsars,
tively; otherwise the luminosities were obtained from PL fits. The arrows and new X-ray detected Fermi pulsars. The latter will be presented in detail in
mark upper limits derived from X-ray non-detections. This figure is an up- Kargaltsev et al. (2012). As in the previous figure, parallactic distances were
date of the one presented earlier by Kargaltsev et al. (2006) using new dis- corrected for the Lutz–Kelker bias, where applicable.
tances by Deller et al. (2009), the Lutz–Kelker bias corrections by Verbiest (A color version of this figure is available in the online journal.)
et al. (2010), the detection of PSR B0834+06 by Gil et al. 2008, and the
upper limit result for PSR J2144−3933 by Tiengo et al. (2011). Note that
the large X-ray luminosity of the extremely fast moving PSR B2224+65 may be most radio pulsars have characteristic ages τ  1 Myr for
overestimated due to a suspected, unresolved shocked pulsar wind component
(e.g., Johnson & Wang 2010). Ė < 9 × 1033 erg s−1 while younger radio pulsars usually
(A color version of this figure is available in the online journal.)
have higher Ė. The close γ -ray pulsar PSR J1741−2054 with
τ = 386 kyr, but modest Ė = 9.5 × 1033 erg s−1 influenced our
choice of the particular boundary value.
According to the ATNF pulsar catalog, PSR B1451–68 with
its Lutz–Kelker corrected distance of 480+80 −60 pc is the 11th
closest of the isolated radio pulsars with Ė < 9 × 1033 erg s−1 .
Six of these eleven pulsars7 are now detected in X-rays, all six
having parallactic distances. Two of the X-ray detected pulsars,
PSR B1929+10 and PSR B0823+26, have X-ray efficiencies
10−4 < ηnonth < 10−3 ; the other four X-ray detected pulsars have
ηnonth > 10−3 . Thus, at least 36% of all the closest old pulsars
with Ė < 9 × 1033 ergs s−1 show ηnonth > 10−3 .
For comparison, we now consider the 11 closest, isolated,
younger pulsars with Ė > 9 × 1033 ergs s−1 . Nine of them are
detected in X-rays, and the two others were not probed deep
enough. Four of these nine detected pulsars have parallactic
distances. Based on information of the ATNF pulsar catalog,
X-ray flux investigations by Marelli et al. (2011), Camilo et al.
(2009), Kargaltsev & Pavlov (2008), and Tepedelenlioǧlu &
Figure 4. X-ray luminosities and upper limits of 12 old pulsars vs. their spin- Ögelman (2007), and distance corrections according to Verbiest
down power. In case of a sufficient number of counts, the diamonds and asterisks et al. (2010) and Mignani et al. (2010), we find that all nine X-
show the non-thermal and thermal X-ray luminosities, respectively; otherwise ray detected younger pulsars have ηnonth < 10−3 . Thus, at most
the luminosities were obtained from PL fits. The arrows mark upper limits
derived from X-ray non-detections. This figure is an update of those presented 18% of all these 11 pulsars with Ė > 9 × 1033 ergs s−1 could
earlier by Kargaltsev et al. (2006) and Pavlov et al. (2009) using new distances have ηnonth > 10−3 .
by Deller et al. (2009), the Lutz–Kelker bias corrections by Verbiest et al. (2010), Thus, looking only at the 11 closest sources in each case, it is
the detection of PSR B0834+06 by Gil et al. 2008, and the upper limit result for suggestive that at least one-third of old pulsars have efficiencies
PSR J2144−3933 by Tiengo et al. (2011). Note that the large X-ray luminosity
of the extremely fast-moving PSR B2224+65 may be overestimated due to a
η > 0.001, while none of the detected closest, younger pulsars
suspected, unresolved shocked pulsar wind component (e.g., Johnson & Wang do. Recently, Marelli et al. (2011) reported the following LX –Ė
2010). relation for 29 Fermi pulsars with available distance estimates:
(A color version of this figure is available in the online journal.) log10 LX, 29 = (1.11+0.21
−0.30 ) + (1.04 ± 0.09) log10 Ė34 , which is in
agreement with older estimates for X-ray pulsars by Kargaltsev
a rough, admittedly arbitrary boundary between young and & Pavlov (2008) and Possenti et al. (2002). From this, one
old pulsars. This boundary was motivated by pulsar statistics
in the ATNF pulsar catalog6 (Manchester et al. 2005), where 7 PSR B2224+65 with its potentially overestimated X-ray luminosity due to
an unresolved shocked pulsar wind component is not among the 11 closest old
6 http://www.atnf.csiro.au/research/pulsar/psrcat/ pulsars.

5
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

would indeed expect η < 10−3 for most X-ray pulsars, and the
>36% deviation of the closest, low Ė pulsars appears even
more puzzling. Unfortunately, an insufficient number of probed
objects, inhomogeneous depth of the X-ray observations, low
signal-to-noise ratios in many cases, potentially unresolved
compact pulsar wind nebulae, as well as large uncertainties of
DM-based distances, prohibit currently any further conclusive
statistics, for a larger sample of old pulsars in particular. Whether
there is indeed a higher X-ray efficiency for pulsars with low Ė
can only be checked by increasing the sample of such pulsars
with sufficiently sensitive X-ray observations.

4. SUMMARY
We investigated Chandra observations of PSR B1451−68.
Through various image analysis techniques we estimate the
possible contribution from a nearby 2MASS star to the pulsar’s
X-ray flux to be less than 7%. No significant extended emission
is seen. The pulsar has a soft X-ray spectrum similar to those Figure 6. Left: the NTT SUSI2 WB655 image is shown for the CCD chip
covering the target PSR B1451−68. Right: the reprojected X-ray image is
of other old pulsars. Nominally, the spectrum is best fit with plotted for the same coordinate range. The approximate size of each image is
a power law, having a photon index of Γ = 2.4+0.4 −0.3 . However, 2. 5 × 3. 8. The brightness scale is stretched to highlight the respective sources.
the inferred hydrogen column density, NH , is uncomfortably The X-ray source position of PSR B1451−68 is marked by the largest box in
close to the Galactic value of H i in this direction. Fixing NH to the upper part of the images. Smaller boxes in the SUSI2 image indicate optical
the DM-derived 2.6 × 1020 cm−2 , a combination of a thermal sources used for the reprojection. The circles in the X-ray image indicate the
corresponding X-ray sources.
and a non-thermal component fits the data best. The inferred
radius of a heated PC is small, a factor of 20 less than expected
conventionally. The non-thermal X-ray efficiency, ∼3 × 10−3 , of the ESO SUSI2 image with the help of the 2MASS PSC
is high in comparison to younger pulsars, but comparable to (Skrutskie et al. 2006) using GAIA (Graphical Astronomy and
estimates for other old pulsars. Image Analysis; Draper et al. 2007). Nominally, an absolute
positional accuracy of 3σ2MASS = 300 mas (Skrutskie et al.
We thank P. Broos, K. Getman, and L. Townsley for en- 2006) can be achieved for high-quality 2MASS point sources.
lightening discussions regarding Chandra data reduction. We The rms of the Gaia astrometric fit to the AAA-2MASS source
also thank Elizabeth Galle and her colleagues from the CXC positions was determined to be σSUSI2 = 178 mas. Thus, we
helpdesk for helpful suggestions regarding the reprojection of expect the source positions in the ESO SUSI2 image to have an
the Chandra event file. absolute astrometric 3σ error of 612 mas.
Support for this work was provided by the ACIS Instru- Nine optical sources in the SUSI2 image were found to cor-
ment Team contract SV4-74018 (PI: G. Garmire) issued by the respond to wavdetect X-ray sources on the ACIS S3 chip.
Chandra X-ray Observatory Center, which is operated by the Five of them are also 2MASS point sources. We excluded one
Smithsonian Astrophysical Observatory for and on behalf of source close to the expected pulsar position and proceeded with
the National Aeronautics Space Administration under contract eight optical detections. The sources are indicated in Figure 6.
NAS8-03060. This work was partly supported by NASA grant The X-ray and optical sources were matched and the Chandra
NNX09AC84G, NSF grant AST09-08611, and by the Min- image astrometry updated by applying the CIAO tasks
istry of Education and Science of the Russian Federation (con- reproject_aspect and wcs_match. After removing one poor
tract 11.G34.310001). Based on observations made with ESO match, seven sources remained. For the seven sources the aver-
Telescopes at the La Silla Paranal Observatory under program age residual after reprojecting was σCXO = 130 mas in com-
ID 68.D-0249. parison to 340 mas for the original event file (all without pixel
This research has made use of SAOImage DS9, developed position randomization). Thus, our overall absolute astrometric
by SAO; the SIMBAD and VizieR databases, operated at
2
uncertainty is 3 (σSUSI2 2
+ σ2MASS 2
+ σCXO )1/2 = 3σabsAM
=

CDS, Strasbourg, France; and SAO/NASA’s Astrophysics Data 0. 73, while the relative astrometric uncertainty of Chandra
System Bibliographic Services. with respect to the 2MASS PSC is 3 (σSUSI2 2 2
+ σCXO )1/2 =

Facility: CXO (ACIS) 3σrel = 0. 66.
AM

APPENDIX B
APPENDIX A
2MASS 14560002−6843400
REPROJECTION OF THE X-RAY DATA AND
X-RAY SOURCE POSITION B.1. Spectral Constraints on the Nature of
2MASS 14560002−6843400
For the optical reference frame, we used the ESO NTT
observations of PSR B1451−68. The NTT was equipped with Here, we want to test if—neglecting positional arguments—
the SUSI2 imager. We chose an image which was obtained with 2MASS 14560002−6843400 could be in principle the counter-
the WB655, a wide R-band filter in 2002 February. The field of part of the X-ray source. As described in Section 2, we checked
view of the CCD chip with the target on it is 2. 8 × 5. 5, and there different spectral models for correspondence with the energy
are plenty of 2MASS point sources in this field, e.g., 46 with distribution of the ∼200 source counts within an energy range
the highest quality flag AAA. We obtain the absolute astrometry of 0.3–8 keV; the fit results are listed in Table 2. Among others,

6
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

we checked whether the X-ray spectrum can be described by an

optically thin thermal plasma model as one could expect from
a stellar corona. In Table 2, we list the obtained apec model
parameter values; the mekal or raymond models gave similar
values. The apec plasma model fit is acceptable. The derived
plasma temperature is high, kT = 3.1+1.3 −0.8 keV. Such high tem-
peratures can be reached in young stars (e.g., Preibisch et al.
2005; Getman et al. 2005).
As a reminder, the derived absorbed X-ray fluxes for the dif-
ferent spectral models range from 2.4 × 10−14 erg cm2 s−1
to 2.9 × 10−14 erg cm2 s−1 for energies between 0.3 keV
and 5 keV. We use the formula by Maccacaro et al. (1988),
log(fX /fV ) = log(fX ) + 0.4mV + 5.37, to estimate the optical
to X-ray flux ratio. 2MASS 14560002−6843400 has an ap-
parent V-band magnitude of mV = 16.2 mag in the General
Guide Star Catalogue version 2.3.2 (Lasker et al. 2008), and
mV = 15.7 mag in the NOMAD catalog (Zacharias et al. 2004).
Assuming that the detected X-ray source is the counterpart of
2MASS 14560002−6843400, the log(fX /fV ) range is accord-
ingly from −1.7 to −2.0. Such flux ratios are typical for K or
M stars (e.g., Agüeros et al. 2009).
2MASS 14560002−6843400 has the following magnitude
measurements: J = 14.50 ± 0.04 mag, H = 13.94 ± 0.04 mag,
K = 13.83 ± 0.06 mag, B2 = 16.83 mag (USNO B1), R2 =
15.05 mag (USNO B1), and I = 15.11 mag (USNO B1). The
Figure 7. This RGB image shows the reprojected event file with the CIAO
obtained colors, B2 − V and V − I, are consistent with a late EDSER subpixel algorithm applied. A 0.25 subpixel binning, as well as slight
G to an early K star, the colors J − H, H − K, V − J, V − H, smoothing (Gauss kernel 2 subpixel) was used in each band. The red band shows
and V − K indicate a K0 to K2 star following intrinsic stellar the events from 0.3 keV to 1 keV, the green band shows the events from 1 keV
colors listed by Currie et al. (2010) and Covey et al. (2007). A to 2 keV, and the blue band shows the events from 2 keV to 8 keV. The scale is
K2 dwarf would be consistent with all colors without requiring linear and the same for all bands. North is up and east is to the left. The expected
pulsar position is marked with a black cross, the 2MASS catalog position of
significant extinction. From the NH = 6.5+7.5 −4.7 × 10
20
cm−2 a nearby star is marked by a red diamond, and the position of the star in the
obtained for the apec model fit we estimate the optical extinction SUSI2 image is marked with the red square. The big cyan circle has a radius of
AV = 0.36+0.44
−0.26 using the relation between the hydrogen column 0. 6 and marks the centroid position of the X-ray source. The other cyan region
density and extinction by Predehl & Schmitt (1995). Using the indicates the area of the Chandra PSF asymmetry as inferred by applying the
CIAO task make_psf_asymmetry_region. See the text for the discussion.
extinction relations by Cardelli et al. (1989) and AV = 0.36 we
estimate color corrections and find that a G9 dwarf is the most (A color version of this figure is available in the online journal.)
likely extinction-corrected counterpart for the cataloged near-IR
and optical magnitudes. Assuming MJ = 4.2 mag according to the CIAO EDSER subpixel algorithm (Li et al. 2004) applied.
the Padova tracks (Bertelli et al. 2008), the reported apparent Apparently, there is emission near the 2MASS source. However,
J magnitude translates into a distance of 1.1 kpc. For a K2 the location of the 2MASS source is very close to a region
dwarf without extinction, assuming MJ = 4.8 mag, the reported where enhanced emission is expected due to the recently found
apparent J magnitude translates into a distance of 0.9 kpc. asymmetry in the Chandra point-spread function8 (PSF). For
A typical K or late G dwarf main-sequence star is expected to our observation we obtained the inflicted region by applying the
have its X-ray peak emission below 1 keV and shows nearly CIAO task make_psf_asymmetry_region; it is marked in the
no emission above 1 keV. Young dwarf stars, on the other figure in cyan and the artifact can constitute ∼5% of the total
hand, could produce the observed X-ray spectrum. There is no brightness. If a normal K star at the position of the 2MASS object
obvious star formation region within 1◦ of the target position, contributes to the X-ray flux, one would expect an enhancement
and as noted above, 2MASS 14560002−6843400 has a very in soft counts toward this position. However, there appear to be
slow proper motion (<12 mas yr−1 ). While one cannot exclude slightly more hard counts in the direction of the 2MASS source.
an in situ young stellar object, there is no indication for this from A hard spectrum is unusual for a normal star, less so for a young
extinction, outflows, or flares, and a late G or an early K dwarf stellar object. As discussed in Appendix B.1, the 2MASS source
main-sequence star seems to be the most likely counterpart of is unlikely to be a young stellar object.
2MASS 14560002−6843400. From this in turn, we conclude Next, we measure the number of counts in similar regions
that the contribution of 2MASS 14560002−6843400 to the around the center of the X-ray source to quantify the overabun-
X-ray emission is small, and that the X-ray source is not the dance of counts toward the 2MASS star. Figure 8 shows the
counterpart of 2MASS 14560002−6843400, but of the pulsar. chosen counting regions: The “star” circle is centered on the
2MASS source’s 2002 optical SUSI2 position. Circles 1–4 have
B.2. Investigation of Possible Source Blending similar separations from the X-ray centroid and the same size as
We wish to obtain an upper limit for any potential X-ray flux the “star” region. They are used to obtain the average expected
contribution of the 2MASS star to the X-ray source detected number of counts from the PSF wings for this particular sepa-
at the pulsar position. First, we look at the spatial distribution ration and area. This assumes a symmetric PSF (see comments
of the dominant count energies in the X-ray image. Figure 7
8
shows a color-coded image of the reprojected event file with http://cxc.harvard.edu/ciao/caveats/psf_artifact.html

7
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

Figure 8. This image shows same-size source regions on the reprojected event
Figure 9. On the left is the deconvolved image produced from the event file
file with the CIAO EDSER subpixel algorithm applied. The binning is 0.25
without randomization, obtained by applying the CIAO task arestore. The
Chandra ACIS sky pixels. The “star” circle is centered on the SUSI2 opti-
right image is the deconvolved image, which was produced from a simulated
cal position of the 2MASS point source—its center position is marked by the
event file with two sources—one with 95% flux and the other with 5% flux of
black box. Similarly, the white box marks the centroid position of the X-ray
the detected X-ray source in our data set. The dashed circle with r = 0. 6 is
source. All circles have radii of 0. 4. In the southwest, the region of the
centered at the centroid position of the original event file of our observation.
Chandra PSF asymmetry is indicated as obtained with the CIAO task
The pulsar position is marked with a cross, the 2MASS point-source position
make_psf_asymmetry_region. See the text for the discussion.
with a diamond, and the 2002 SUSI2 optical position of this 2MASS source
with a small box. The region of the PSF asymmetry is also indicated. One of
the two solid, r = 0. 4, circles in each image is centered on the corresponding
below). To relate the “star” counts, Cstar , to the pulsar counts, centroid position of the deconvolved main source, the other is centered on the
CPSR , we use a circle of the same size at the position of the X-ray SUSI2 position. Both images are in log brightness scale to emphasize the low
centroid. We estimate the ratio between the extra “star” counts count numbers. See the text for the discussion.
and the pulsar counts as (Cstar − (C1 + C2 + C3 + C4 ) 0.25)/CPSR .
Of course, this is a very crude estimation neglecting a detailed an energy range from 0.3 to 8 keV. From the centroid source
Chandra PSF model in general and the outer PSF wings in par- position we obtain the off-axis angle, θ = 18 .6, and the
ticular. In the ideal case of well-separated point sources with mirror spherical coordinate azimuth, φ = 327.◦ 9.10 We use
similar spectra and using the same aperture for each source, a long exposure time (180 ks) in order to generate a high
their flux ratios should be constant for different aperture sizes. signal-to-noise-ratio PSF image which is advantageous for the
In our case, we need to account for a flux contribution due to following deconvolution and the source extent estimate. The
the overlapping of the individual PSFs. This contribution is sig- raytrace file is then used within MARX11 to produce a PSF
nificant for the potential “star” source; hence the subtraction image with 0.25 pixel binning. We set the DitherBlur parameter
of 0.25 (C1 + C2 + C3 + C4 ) is an approximation of this con- to 0. 2 as recommended in the MARX manual for ACIS data
tribution. The flux contribution of the “star” PSF to CPSR is without pixel randomization. Using the PSF image, the CIAO
negligible as seen from Chandra MARX simulations. For cir- task srcextent reports the X-ray source to be not extended at
cles with r = 0. 4 we infer “star” percentages of 6.3% for the 90% confidence. We then deconvolved the event file applying
event file with the EDSER subpixel algorithm applied, and 0% the task arestore, which is based on the Lucy–Richardson
(actually −1.3%) for the event file without randomization. For deconvolution algorithm (Lucy 1974). The image is shown in
circles with r = 0. 3 we infer “star” percentages of 7.1% for the Figure 9. The new centroid position of the main source in the
event file with the EDSER subpixel algorithm applied, and 3.1% deconvolved image has a small offset of ∼0. 1 with respect
for the event file without randomization. Note that the ACIS to the centroid position in the original image. Comparing the
encircled energy is 50% for an aperture radius of r = 0. 418, count numbers in an r = 0. 4 circle centered on the former
while it is 90% for r = 2 , the radius used in Sections 2 and 3.2 and a circle of the same size centered on the SUSI2 optical
(Chandra Proposers’ Observatory Guide, Section 6.69 ). Due to position of the 2MASS star, we derive a flux percentage of
the Poisson noise character of the very few counts we are deal- 6.5% for the counts around the star with respect to those of the
ing with, it is difficult to obtain a rigorous error estimate for pulsar. Using the original centroid position for the main source
the above formula. Changing slightly the positions of the indi- decreases this percentage to 6.1%. The deconvolution produces
vidual counting regions, we approximate the error to be around “extra” counts not only toward the direction of the 2MASS
1% of the pulsar counts. The known PSF asymmetry region of source, which is located at one edge of the PSF asymmetry
enhanced flux is actually partly overlapping our “star” circle region, but also southwest of the pulsar where the other edge of
(see Figure 8). Since part of the enhanced flux in the “star” the PSF asymmetry region lies. Since neither ChaRT, MARX,
circle is likely due to the asymmetry “leakage,” our estimated nor arestore account for this asymmetry, some artifacts in the
percentages are indeed conservative upper limits. south–southwest direction can be expected after deconvolution.
We also deconvolved the image applying the Chandra Ray For comparison, we also used MARX to simulate two sources
Tracer (ChaRT), MARX (v. 4.4 and 4.5), and the CIAO task having combined as many counts as the r = 2 main-source
arestore following standard CIAO threads. Since the subpixel region—one source with 95% of the flux at the main-source
algorithm is not yet implemented in the simulators, we discuss centroid position and the other with 5% flux at the SUSI2
here only the event file without pixel randomization. We apply optical position of the 2MASS star. Again, we set the DitherBlur
ChaRT (Carter et al. 2003) to construct the PSF at the source parameter to 0. 2 and proceeded with the processing as described
position. As the input spectrum, we used the parameters derived
from the PL fit of the source spectrum (see Section 3.2) in 10 For details on the Chandra coordinate systems see
http://cxc.harvard.edu/contrib/jcm/ncoords.ps.
9 http://cxc.harvard.edu/proposer/POG/html/index.html 11 http://space/mit/edu/CXC/MARX/index.html

8
The Astrophysical Journal, 749:146 (9pp), 2012 April 20 Posselt et al.

above. Again we find an ∼0. 1 shift (but in another direction) Hui, C. Y., & Becker, W. 2008, A&A, 486, 485
in the centroid position of the deconvolved image comparing Johnson, S. P., & Wang, Q. D. 2010, MNRAS, 408, 1216
Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005, A&A, 440, 775
to the original input coordinates. While these shifts are all Kargaltsev, O., Durant, M., Pavlov, G. G., & Garmire, G. P. 2012, ApJ, submitted
within the nominal Chandra astrometric accuracy, we do not (arXiv:1202.3838)
utilize the arestore positions for further astrometric purposes. Kargaltsev, O., & Pavlov, G. G. 2008, in AIP Conf. Proc. 983, 40 Years
Using again the r = 0. 4 circles as indicated in Figure 9, we of Pulsars: Millisecond Pulsars, Magnetars and More, ed. C. Bassa et al.
(Melville, NY: AIP), 171
find 4.0% of the simulated main-source counts in the circle Kargaltsev, O., Pavlov, G. G., & Garmire, G. P. 2006, ApJ, 636, 406
around the simulated 2MASS source. From this we infer that, Lasker, B. M., Lattanzi, M. G., McLean, B. J., et al. 2008, AJ, 136, 735
in principle, it is possible to recover a ∼5% X-ray source at a Li, J., Kastner, J. H., Prigozhin, G. Y., et al. 2004, ApJ, 610, 1204
0. 77 separation, which the 2MASS source has from the main Lucy, L. B. 1974, AJ, 79, 745
X-ray source centroid. Maccacaro, T., Gioia, I. M., Wolter, A., Zamorani, G., & Stocke, J. T. 1988, ApJ,
326, 680
Manchester, R. N., Hobbs, G. B., Teoh, A., & Hobbs, M. 2005, AJ, 129, 1993
REFERENCES Marelli, M., De Luca, A., & Caraveo, P. A. 2011, ApJ, 733, 82
Mignani, R. P., Pavlov, G. G., & Kargaltsev, O. 2010, ApJ, 720, 1635
Agüeros, M. A., Anderson, S. F., Covey, K. R., et al. 2009, ApJS, 181, 444 Misanovic, Z., Pavlov, G. G., & Garmire, G. P. 2008, ApJ, 685, 1129
Bailes, M., Manchester, R. N., Kesteven, M. J., Norris, R. P., & Reynolds, J. E. Monet, D. G., Levine, S. E., Canzian, B., et al. 2003, AJ, 125, 984
1990, Nature, 343, 240 Mori, K., Tsunemi, H., Miyata, E., et al. 2001, in ASP Conf. Ser. 251, New
Balucinska-Church, M., & McCammon, D. 1992, ApJ, 400, 699 Century of X-ray Astronomy, ed. H. Inoue & H. Kunieda (San Francisco,
Becker, W., Kramer, M., Jessner, A., et al. 2006, ApJ, 645, 1421 CA: ASP), 576
Becker, W., Weisskopf, M. C., Tennant, A. F., et al. 2004, ApJ, 615, 908 Pavlov, G. G., Kargaltsev, O., Wong, J. A., & Garmire, G. P. 2009, ApJ, 691,
Bertelli, G., Girardi, L., Marigo, P., & Nasi, E. 2008, A&A, 484, 815 458
Camilo, F., Ray, P. S., Ransom, S. M., et al. 2009, ApJ, 705, 1 Pavlov, G. G., Shibanov, Y. A., Zavlin, V. E., & Meyer, R. D. 1995, in Proc.
Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 NATO Advanced Study Inst. on the Lives of the Neutron Stars, ed. M. A.
Carter, C., Karovska, M., Jerius, D., Glotfelty, K., & Beikman, S. 2003, in ASP Alpar, U. Kiziloglu, & J. van Paradijs (Dordrecht: Kluwer), 71
Conf. Ser. 295, Astronomical Data Analysis Software and Systems XII, ed. Possenti, A., Cerutti, R., Colpi, M., & Mereghetti, S. 2002, A&A, 387, 993
H. E. Payne, R. I. Jedrzejewski, & R. N. Hook (San Francisco, CA: ASP), Predehl, P., & Schmitt, J. H. M. M. 1995, A&A, 293, 889
477 Preibisch, T., Kim, Y.-C., Favata, F., et al. 2005, ApJS, 160, 401
Covey, K. R., Ivezić, Ž., Schlegel, D., et al. 2007, AJ, 134, 2398 Roeser, S., Demleitner, M., & Schilbach, E. 2010, AJ, 139, 2440
Currie, T., Hernandez, J., Irwin, J., et al. 2010, ApJS, 186, 191 Ruderman, M. A., & Sutherland, P. G. 1975, ApJ, 196, 51
Cutri, R. M., Skrutskie, M. F., van Dyk, S., et al. 2003, The IRSA 2MASS Siegman, B. C., Manchester, R. N., & Durdin, J. M. 1993, MNRAS, 262,
All-Sky Point Source Catalog, NASA/IPAC Infrared Science Archive 449
(http://irsa.ipac.caltech.edu/applications/Gator/) Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163
D’Alessandro, F., McCulloch, P. M., King, E. A., Hamilton, P. A., & McConnell, Tepedelenlioǧlu, E., & Ögelman, H. 2005, ApJ, 630, L57
D. 1993, MNRAS, 261, 883 Tepedelenlioǧlu, E., & Ögelman, H. 2007, ApJ, 658, 1183
Deller, A. T., Tingay, S. J., Bailes, M., & Reynolds, J. E. 2009, ApJ, 701, 1243 Tiengo, A., Mignani, R. P., de Luca, A., et al. 2011, MNRAS, 412, L73
Dickey, J. M., & Lockman, F. J. 1990, ARA&A, 28, 215 Verbiest, J. P. W., Lorimer, D. R., & McLaughlin, M. A. 2010, MNRAS, 405,
Draper, P., Gray, N., Berry, D. S., & Taylor, M. 2007, GAIA—Graph- 564
ical Astronomy and Image Analysis Tool Starlink User Note 214, Wilms, J., Allen, A., & McCray, R. 2000, ApJ, 542, 914
http://star-www.rl.ac.uk/docs/sun214.htx/sun214.html Yakovlev, D. G., & Pethick, C. J. 2004, ARA&A, 42, 169
Getman, K. V., Flaccomio, E., Broos, P. S., et al. 2005, ApJS, 160, 319 Yan, M., Sadeghpour, H. R., & Dalgarno, A. 1998, ApJ, 496, 1044
Gil, J., Haberl, F., Melikidze, G., et al. 2008, ApJ, 686, 497 Zacharias, N., Monet, D. G., Levine, S. E., et al. 2004, BAAS, 36, 1418
Gil, J., Melikidze, G. I., & Geppert, U. 2003, A&A, 407, 315 Zavlin, V. E., & Pavlov, G. G. 2004, ApJ, 616, 452
Gotthelf, E. V. 2003, ApJ, 591, 361 Zavlin, V. E., Pavlov, G. G., & Shibanov, Y. A. 1996, A&A, 315, 141
Harding, A. K., & Muslimov, A. G. 2001, ApJ, 556, 987 Zhang, B., Sanwal, D., & Pavlov, G. G. 2005, ApJ, 624, L109
Harding, A. K., & Muslimov, A. G. 2002, ApJ, 568, 862 Zharikov, S., Shibanov, Y., & Komarova, V. 2006, Adv. Space Res., 37, 1979