The X-ray Emission and Population of Highly Magnetized Neutron Stars

Scott A. Olausen

Doctor of Philosophy

Physics

McGill University Montreal, Quebec June 18, 2014

A thesis submitted to McGill University in partial fulfillment of the requirements for the degree of Doctor of Philosophy c Scott A. Olausen, 2014 TABLE OF CONTENTS

TABLE OF CONTENTS...... ii LIST OF FIGURES...... iv LIST OF TABLES...... vi ABSTRACT...... vii RESUM´ E´...... x ACKNOWLEDGMENTS...... xiii PREFACE...... xiv 1 Introduction...... 1 1.1 Pulsars...... 1 1.1.1 Neutron Star Structure...... 6 1.1.2 Neutron Star Magnetosphere...... 8 1.1.3 Pulsar Wind Nebulae...... 11 1.1.4 Neutron Star Cooling...... 12 1.2 Magnetars...... 14 1.2.1 History of AXPs and SGRs...... 14 1.2.2 The Magnetar Model...... 17 1.3 High-B Pulsars and Magnetars...... 21 2 X-ray Astronomy and Instrumentation...... 25 2.1 History of X-ray Astronomy...... 25 2.2 Wolter Mirrors...... 29 2.3 CCDs...... 31 2.4 Description of the XMM-Newton Observatory...... 32 3 X-ray Detection and Temperature of the High-B PSR J1734–3333 37 3.1 Introduction...... 37 3.2 Observations and Results...... 38 3.2.1 Imaging...... 39 3.2.2 Timing Analysis...... 41 3.2.3 Spectral Analysis...... 42 3.3 Discussion...... 46

ii 3.3.1 Associating the X-ray Source with the Radio Pulsar 46 3.3.2 The Temperature of PSR J1734–3333...... 48 3.3.3 X-ray Luminosities of High-B Pulsars...... 52 3.4 Conclusions...... 54 4 The Extended Emission Around the Magnetar 1E 1547.0–5408.. 55 4.1 Introduction...... 55 4.2 Observations...... 57 4.3 Imaging Analysis and Results...... 58 4.3.1 Spectral Analysis...... 63 4.4 Discussion...... 65 4.4.1 A Pulsar Wind Nebula?...... 65 4.4.2 Dust-scattering Halo...... 67 4.5 Conclusions...... 71 5 The First Catalog of the Magnetar Population...... 74 5.1 Introduction...... 74 5.2 Data Tables...... 75 5.2.1 Table 1: Positions and Proper Motions...... 75 5.2.2 Table 2: Timing Properties...... 77 5.2.3 Table 3: Quiescent Soft X-ray Properties...... 79 5.2.4 Table 4: Optical and Near-Infrared Observations.. 82 5.2.5 Table 5: Radio and Mid-Infrared Observations... 85 5.2.6 Table 6: Hard X-ray and Gamma-ray Observations. 85 5.2.7 Table 7: Associations and Distances...... 89 5.3 Discussion...... 91 5.3.1 Spatial Properties...... 94 5.3.2 Timing Properties...... 101 5.3.3 X-ray Properties...... 107 5.3.4 Multiwavelength Properties...... 112 5.4 Conclusions...... 116 6 Conclusions...... 118 6.1 High-B Pulsars...... 118 6.2 Magnetar Winds...... 119 6.3 The Magnetar Population...... 121 6.4 Concluding Remarks...... 124 References...... 126

iii LIST OF FIGURES Figure page 1–1 The P –P˙ diagram...... 2 1–2 Cross section of a neutron star...... 7 1–3 Diagram of the magnetosphere of a pulsar...... 10 2–1 Diagram of Wolter X-ray optics...... 30 2–2 The XMM-Newton observatory...... 33 2–3 Layout of the CCDs in the MOS and pn cameras...... 34 3–1 XMM image of PSR J1734−3333...... 39 3–2 Radial profile of PSR J1734−3333...... 40 3–3 XMM spectrum of PSR J1734−3333...... 44

3–4 Confidence contours for kT and NH...... 45 3–5 Blackbody temperature vs. characteristic age for normal pul- sars, high-B pulsars, XINSs, and magnetars...... 49 3–6 X-ray luminosity vs. magnetic field strength for X-ray observed high-B radio pulsars...... 53 4–1 Radial profiles of 1E 1547.0−5408 at 1–6 keV and 6–12 keV... 60 4–2 Extended emission count rate vs. point source count rate... 61 4–3 Fractional intensity of the extended emission...... 62 4–4 Hardness ratio for the extended emission in 2006...... 64 5–1 Number of confirmed magnetars discovered over time...... 91 5–2 Top-down view of the Galaxy...... 94 5–3 Distribution in Galactic longitude of all Galactic disc pulsars.. 96 5–4 Distribution in Galactic latitude of all Galactic disc pulsars.. 97 5–5 Cumulative distribution function and histogram of the height above the Galactic plane for the Galactic magnetars..... 98

iv 5–6 Distribution in pulse period of pulsars and magnetars...... 102 5–7 Distribution in magnetic field of pulsars and magnetars..... 103 5–8 Distribution in spin-down luminosity of pulsars and magnetars. 104 5–9 Distribution in characteristic age of pulsars and magnetars... 105 5–10 P –P˙ diagram for radio pulsars, XINSs, and magnetars..... 106 5–11 Photon index and blackbody temperature vs. magnetic field.. 107 5–12 Quiescent X-ray luminosity vs. photon index and blackbody temperature...... 109 5–13 Quiescent X-ray luminosity vs. magnetic field, spin-down lu- minosity, and characteristic age...... 110 5–14 P –P˙ diagram of the magnetars, showing those detected in ra- dio and hard X-ray...... 112 5–15 Magnetar detections as a function of quiescent X-ray flux.... 114

v LIST OF TABLES Table page 3–1 Summary of XMM Observations of PSR 1734−3333...... 38 3–2 Spectral Models for PSR J1734−3333...... 43 3–3 High-magnetic-field Radio Pulsars...... 51 4–1 Summary of XMM Observations of 1E 1547.0−5408...... 58 4–2 Hardness Ratios for 1E 1547.0−5408 and the Surrounding Ex- tended Emission...... 63 4–3 Contribution to Extended Emission Not from Dust Scattering. 70 5–1 Magnetar Positions and Proper Motions...... 76 5–2 Magnetar Timing Properties...... 78 5–3 Soft X-Ray Properties of Magnetars in Quiescence...... 80 5–4 Optical and Near-Infrared Counterparts of Magnetars..... 83 5–5 Radio and Mid-Infrared Observations of Magnetars...... 86 5–6 Hard X-Ray and Gamma-Ray Observations of Magnetars... 87 5–7 Magnetar Associations and Distances...... 90 5–8 Magnetar Names...... 92

vi ABSTRACT

Over the past few decades, advances in X-ray and gamma-ray astron- omy have greatly expanded our knowledge of the neutron-star family. One important recent discovery has been that of the “magnetars,” isolated neu- tron stars whose radiation and occasional bursting activity is thought to be powered by their very high magnetic fields (1014–1015 G as inferred from timing), unlike ordinary pulsars that are powered by their rotational energy. There do, however, exist rotation-powered pulsars with inferred magnetic fields that approach those of the magnetars (∼1013 G). These two groups might therefore be expected to show some similarities in their properties or behaviour. Careful study of both the high-magnetic-field pulsars and magne- tars, then, may help us to understand magnetar physics and determine their relations and connections with the rest of the pulsar population. In Chapter3, I present the results of two XMM-Newton observations of the high-magnetic-field radio pulsar PSR J1734−3333. We successfully detect the X-ray counterpart of the pulsar. Its spectrum fits well to a blackbody with temperature 300 ± 60 eV, and its bolometric luminosity

+2.2 32 −1 is Lbb = 2.0−0.7 × 10 erg s , or ∼0.4% of its spin-down power, for a distance of 6.1 kpc. We detect no X-ray pulsations from the source, setting a 1σ upper limit on the pulsed fraction of 60% in the 0.5–3 keV band. We compare PSR J1734−3333 to other rotation-powered pulsars of similar age and find that it is significantly hotter, supporting the hypothesis that the magnetic field affects the observed thermal properties of pulsars. We also tabulate the properties of this and all other known high-B radio pulsars with measured thermal X-ray luminosities or luminosity upper limits, and speculate on a possible correlation between LX and B.

vii In Chapter4, I present an analysis of the extended emission around the magnetar 1E 1547.0−5408. Based on four XMM-Newton observations taken with the source in various stages from outburst to quiescence, we find that the extended emission flux is highly variable and strongly correlated with the flux of the magnetar. From this result, as well as spectral and energetic considerations, we conclude that the extended emission is dominated by a dust-scattering halo and not a pulsar wind nebula (PWN), as has been previously argued. We obtain an upper limit on the 2–10 keV flux of a possible PWN of 4.7 × 10−14 erg s−1 cm−2, three times less than the previously claimed value. We do, however, find strong evidence for X-ray emission from a supernova remnant surrounding the pulsar, as previously reported. Finally, I present a study of the magnetar population as a whole in Chapter5, with a catalog of the 26 currently known magnetars and magnetar candidates. Tables are provided of astrometric and timing data for all catalog sources, as well as of their observed radiative properties, particularly the spectral parameters of the quiescent X-ray emission. We show histograms of the spatial and timing properties of the magnetars and compare them with the known pulsar population. We measure the scale height of magnetars to be in the range of 20–31 pc, assuming they are exponentially distributed. This range is smaller than that measured for OB stars, providing evidence that magnetars are born from the most massive O stars. From the same fits, we find that the Sun lies ∼13–22 pc above the Galactic plane, consistent with previous measurements. We confirm previously identified correlations between quiescent X-ray luminosity, LX, and magnetic field, B, as well as X-ray spectral power-law indexes, Γ and

B, and show evidence for an excluded region in a plot of LX versus Γ. We

viii observe that while there is a clear correlation between the hard and soft X-ray fluxes in magnetars, the radio-detected magnetars all have low, soft X-ray flux, suggesting, if anything, that the two bands are anti-correlated.

ix RESUM´ E´

Au cours du dernier demi-si`ecle,les avanc´eesinstrumentales en astronomie `arayons X et gammas ont grandement am´elior´enotre compr´ehensiondes ´etoiles`aneutrons. Une r´ecente d´ecouverte importante est celle d’une classe d’´etoiles`aneutrons isol´ees,les “magn´etars”,dont la lu- minosit´eet des ´episodes occasionnels de sursauts sont attribu´es`aun champ magn´etique´elev´e(1014–1015 G, tel que determin´epar le chronom´etrage), contrairement aux pulsars ordinaires dont l’´energieprovient du ralentisse- ment de leur rotation. Cependant, il existe des pulsars aliment´espar la rotation pour lesquels le champ magn´etiqued´eduitapproche celui des magn´etars.Il serait donc plausible que ces deux classes d’objets montrent quelques similarit´esdans leurs propri´et´esou dans leur comportement. Une ´etudeapprofondie, autant des pulsars `afort champ magn´etiqueque des magn´etars,pourrait donc aider `ala compr´ehensionde la physique des magn´etarset permettre de d´eterminerleur relation avec le reste de la population des pulsars. Dans le Chapitre3, je pr´esente les r´esultatsde deux observations, obtenues avec l’observatoire XMM-Newton, du pulsar `ahaut champ magn´etiquePSR J1734−3333. Nous identifions le pulsar dans les rayons X. Son spectre est ad´equatement d´ecrispar un corps noir d’une temp´erature

+2.2 32 −1 de 300 ± 60 eV, et sa luminosit´ebolom´etriqueest Lbb = 2.0−0.7 × 10 erg s , ou ∼0.4% de sa puissance de ralentissement, pour une distance de 6.1 kpc. Nous ne d´etectonsaucune pulsation de la source, posant une limite sup´erieure`a1σ de 60% sur la fraction de pulsation dans la bande de 0.5 `a3.0 keV. Nous comparons PSR J1734−3333 `ad’autres pulsars aliment´es par la rotation et d’ˆagesimilaire et trouvons qu’il est significativement plus

x chaud, supportant l’hypoth`eseselon laquelle le champ magnetique affecte les propri´et´esthermales observ´eesdes pulsars. Nous recen¸cons´egalement les propri´et´esde tous les pulsars `ahaut champ magn´etiquedont la luminosit´e thermale a ´et´emesur´edans les rayons X et sp´eculonssur une corr´elation possible entre LX et le champ magn´etique. Dans le Chapitre4, je pr´esente une analyse de l’´emission´etendue autour du magn´etar1E 1547.0−5408. A` partir de quatre observations obtenues avec l’observatoire XMM-Newton o`ula luminosite de la source varie pendant que cette derni`erela source transitionne d’une p´eriode active `ala quiescence, nous mesurons une ´emission´etenduetr`esvariable et hautement corr´el´eeavec le flux du magn´etar. De ce r´esultat,ainsi que d’une analyse spectrale et de consid´erationsenerg´etiques, nous concluons que l’´emission´etendueest domin´eepar un halo de diffusion par la poussi`ere et non par une n´ebuleuse`avent de pulsar, hypoth`ese ayant ´et´eavanc´ee dans le pass´e.Nous obtenons une limite sup´erieure,dans la bande spectrale de 2 `a10 keV, pour le flux d’une n´ebuleuse `avent de pulsar de 4.7 × 10−14 erg s−1 cm−2, soit trois fois moins que pr´ec´edemment rapport´e.Nous d´etectonscependant un restant de supernova autours du pulsar, tel que rapport´edans la litt´erature. Finalement, je pr´esente une ´etude exhaustive de la population des magn´etarsdans le Chapitre5, r´esultant en la publication d’un catalogue des 26 magn´etarspresentement connus ou de candidats. Des tableaux des posi- tions astrom´etriqueset des propri´et´esde chronom´etragesont donn´espour toutes les sources, ainsi que leurs propri´etesradiatives, particuli`erement les param`etresspectraux en rayons X des sources en quiescence. Nous con- struisons des histogrammes de la distribution spatiale et des param`etresde chronom´etragedes magn´etarset les comparons `aceux des pulsars ordinaires

xi connus. Nous mesurons une hauteur caract´eristiquepour les magn´etars dans l’intervalle de 20 `a31 pc, assumant qu’ils sont distribu´esexponen- tiellement. Cette valeur est plus petite que celle mesur´eepour les ´etoiles OB, indiquant que les magn´etarssont issus des prog´eniteursO les plus massifs. De la mˆemeanalyse, nous mesurons que le soleil se situe ∼13– 22 pc au dessus du plan Galactique, ce qui est en accord avec des mesures pr´ec´edentes. Nous confirmons les corr´elationspr´ec´edemment identifi´eeentre la luminosit´edans les rayons X en quiescence, LX, et le champ magn´etique, B, de mˆemequ’entre l’index de la loi de puissance `arayons X, Γ, et le champ magn´etique,et montrons qu’il existe une zone interdite dans le di- agramme LX versus Γ. Nous obtenons une corr´elation claire entre les flux dans les rayons X doux et durs parmi les magn´etars,et que les magn´etars d´etect´esdans les ondes radio ont tous de faibles flux dans les rayons X doux, sugg´erant que l’´emissiondans les deux bandes est anti-corr´el´ee.

xii ACKNOWLEDGMENTS

I first want to thank my supervisor, Vicky Kaspi, for taking me on as a student, assisting me when needed, and showing me so much support and patience. I also thank all the current and former members of the McGill Pulsar Group. They provided a great environment and plenty of lively discourse during group meetings and neutron star discussions. Thanks go out to all of the co-authors and referees whose comments and suggestions helped shape each of the papers that went into this thesis. In addition, I thank Bob Rutledge and Maggie Livingstone for their helpful comments on the PSR J1734 detection paper that makes up part of Chapter3. For their help with Chapter5, the magnetar catalog, I also thank Joe Lazio, who assisted in producing Figure 5–2, as well as Norm Murray, Laurent Drissen, Jes´usMa´ızApell´aniz,Chris Thompson, Bryan Gaensler, Jules Halpern, Shriharsh Tendulkar, Robert Duncan, and the entire McGill Pulsar Group, who all provided helpful discussions. I also thank Kristen Boydstun and Cindy Tam for their work on early versions of the online magnetar catalog. Thanks to Fran¸coisDufour for translating the thesis abstract into French. Finally, I want to thank my parents for the regular phone calls that kept me connected and for all their support and love.

xiii PREFACE Statement of Originality and Contribution of Authors

This thesis is a collection of papers published in the Astrophysical Journal (Chapters3 and4) and the Astrophysical Journal Supplement Series (Chapter5), of which I am the first author. Each paper reports new and original results based on X-ray observations of a high-magnetic-field pulsar or magnetar, or based on statistical analysis of magnetar properties reported in the literature. Here we summarize the main results of each paper and list the contributions of the co-authors. Chapter 3: X-ray Detection and Temperature of the High-B PSR J1734−3333

The content of this chapter originally appeared in two papers: Olausen, Scott A.; Kaspi, Victoria M.; Lyne, Andrew G.; Kramer, Michael. XMM-Newton X-Ray Observation of the High-Magnetic-Field Radio Pulsar PSR J1734−3333. ApJ, Volume 725, Issue 1, Pages 985–989 (2010). Olausen, Scott A.; Zhu, Weiwei; Vogel, Julia K.; Kaspi, Victoria M.; Lyne, Andrew G.; Espinoza, Crist´obal M.; Stappers, Ben W.; Manchester, Richard M.; McLaughlin, Maura A. X-Ray Observations of High-B Radio Pulsars. ApJ, Volume 764, Issue 1, Page 1 (2013). In this chapter, we report on the first X-ray detection of the high- magnetic-field radio pulsar PSR J1734−3333 as well as a deep follow-up observation. We find that the pulsar’s spectrum is thermal with a blackbody temperature significantly higher than almost all other rotation-powered pulsars for which that quantity has been measured. We detect no X-ray

xiv pulsations but do not set a very tight constraint on the pulsed fraction. We compiled X-ray luminosities for all high-magnetic-field pulsars which have been observed in X-rays and find evidence of a possible correlation between that quantity and the magnetic field. We also update the plot of blackbody temperature versus characteristic age from Zhu et al.(2011) and confirm the trend that for pulsars of similar age, those with higher magnetic fields tend to be hotter, first noted in that paper. The contributions of the co-authors are as follows: Prof. Kaspi was the primary investigator on the proposals for both XMM observations reported on in this chapter. Drs. Lyne and Kramer were co-investigators on both proposals, and they also provided the radio timing ephemeris that we used to search for pulsations for the 2010 paper. Drs. Lyne, Espinoza, and Stappers provided an updated timing ephemeris used for the 2013 paper. Dr. Zhu performed the analysis for an XMM observation of another high-magnetic-field pulsar, PSR B1845−19, and wrote most of the text for the corresponding section of the 2013 paper. Likewise, Dr. Vogel did the same for a Chandra observation of the high-magnetic-field pulsar PSR J1001−5939 that was also covered in the 2013 paper. Drs. Manchester and McLaughlin were co-investigators on the Chandra proposal for PSR J1001−5939. The analysis and discussion of these two pulsars does not appear in this thesis. I performed the imaging, timing, and spectral analysis for the two XMM observations of PSR J1734−3333. I updated the data for the plot of blackbody temperature versus characteristic age, and I compiled the data for the table of high-magnetic-field pulsars and the associated plot of X-ray luminosity versus magnetic field. The text of this chapter was written by me, with all of the co-authors providing comments and suggestions on each draft of the manuscript.

xv Chapter 4: The Nature of the Extended Emission Around the Magnetar 1E 1547.0−5408

The content of this chapter originally appeared as: Olausen, Scott A.; Kaspi, Victoria M.; Ng, Chi Y.; Zhu, Weiwei; Dib, Rim; Gavriil, Fotos P.; Woods, Peter M. On the Extended Emission Around the Anomalous X-ray Pulsar 1E 1547.0−5408. ApJ, Volume 742, Issue 1, Page 4 (2011). In this chapter, we analyze four XMM-Newton observations of the magnetar 1E 1547.0−5408 taken between 2006 and 2010 with the source in various states of outburst and quiescence, with the aim of characterising the extended emission around the magnetar. We find that the flux of the extended emission is highly correlated with the flux of the point source and conclude that the extended source is dominated by a dust-scattering halo and not a pulsar wind nebula as previously claimed. In particular, we set an upper limit on the luminosity of a pulsar wind nebula that is a factor of three lower than that of the previously claimed detection. The contributions of the co-authors are as follows: Prof. Kaspi was the primary investigator on the proposal for the 2010 XMM observation, and Drs. Zhu, Dib, Gavriil, and Woods were co-investigators. I performed the spatial and spectral analysis of the four XMM data sets. I also wrote the text of the chapter, with the exceptions that Prof. Kaspi provided contributions to the introduction and Dr. Ng wrote most of section 4.4.1 of the discussion. In addition, all of the co-authors provided comments and suggestions on each draft of the manuscript. Chapter 5: The First Catalog of the Magnetar Population

The content of this chapter originally appeared as:

xvi Olausen, Scott A.; Kaspi, Victoria M. The McGill Magnetar Catalog. ApJS, Volume 212, Issue 1, Page 6 (2014). In this chapter, we present the first catalog of all known magnetars and magnetar candidates. We present seven tables containing their spatial, timing, and radiative properties across the electromagnetic spectrum, collected and sourced from the literature. We compare the spatial and timing properties of magnetars with those of the rest of the known pulsars. We measure the scale height of magnetars in the Galaxy and find it to be smaller than that of OB stars, implying that magnetar progenitors are among the most massive of these stars. We confirm previously reported correlations between the quiescent X-ray luminosity and magnetic field, and between the X-ray power law index and magnetic field. Finally, we find that magnetars with higher quiescent soft X-ray flux are more likely to be detected in the hard X-ray, but, if anything, the opposite is true for detecting them in radio. The contributions of the co-authors are as follows: I collected all of the results from the literature and constructed the data tables. I also performed the analysis and constructed all of the histograms and plots, though Prof. Kaspi provided guidance throughout this process. The text of this chapter was written by both myself and Prof. Kaspi.

xvii CHAPTER 1 Introduction 1.1 Pulsars

Baade & Zwicky(1934) were the first to predict the existence of neu- tron stars — extremely compact stars formed in supernovae and supported against gravitational collapse by neutron degeneracy pressure. At the time they were expected to be too small and dim to ever be observed, and it wouldn’t be until over 30 years later that the prediction would be confirmed with the discovery of pulsars. The first neutron star to be observed was the bright X-ray source Scorpius X-1 in 1962 (Giacconi et al., 1962), but it took five years until Shklovsky(1967) correctly identified it as a neutron star accreting from its binary companion. That same year Hewish and his student Jocelyn Bell discovered the first radio pulsar, now known as PSR B1919+21 (Hewish et al., 1968). Although most initial explanations for the pulsar phenomenon favoured white dwarf stars, Gold(1968) and Pacini (1968) suggested that pulsars originated from rapidly rotating neutron stars. The subsequent discoveries of the Crab and Vela pulsars (Large et al., 1968; Staelin & Reifenstein, 1968; Richards & Comella, 1969) settled the matter as their short periods (33 ms and 89 ms, respectively) could not be explained by the white dwarf models. The most important pulsar observables are the spin period, P , and period derivative, P˙ . Observed pulsar periods range from slightly over one millisecond (the shortest is P = 1.4 ms; Hessels et al., 2006) to several seconds (the longest seen in a radio pulsar is P = 8.5 s; Young et al., 1999; the longest overall is P = 11.8 s; Vasisht & Gotthelf, 1997), while P˙ can fall

1 10−9 Pulsars XINSs Magnetars

10−12 ) 1 - s

s (

e v i t

a −15 v

i 10 r e D

d o i r e P

10−18

10−21 10−3 0.01 0.1 1 10 Period (s)

Figure 1–1: The P –P˙ diagram. See also Figure 5–10.

2 between ∼10−10 s s−1 and ∼10−21 s s−1. The distribution of the known pulsar population1 in terms of these two quantities is shown in Figure 1–1, and from them several further pulsar parameters can be inferred. A rotating body with moment of inertia I and angular spin frequency

2π 1 2 Ω = P has angular kinetic energy E = 2 IΩ . If the body is spinning down it loses energy at a rate

d 1 IΩ2
E˙ = − 2 = −IΩΩ˙ = 4π2IPP˙ −3, (1.1) dt which for pulsars is called the spin-down luminosity. The moment of inertia, I = kMR2, can be estimated by taking k = 0.4 (for a sphere of uniform density; more realistic models predict k closer to 0.2, e.g., Lattimer & Prakash, 2001) and adopting the canonical values for the mass and radius of

45 2 a neutron star, M = 1.4M and R = 10 km. The result is I = 10 g cm , and given this value, E˙ varies between ∼1029 and ∼1038 erg s−1. The spin- down luminosity is particularly important because ordinary radio pulsars are thought to be powered by their rotation (hence they are also known as rotation-powered pulsars), and E˙ represents the total such rotation power available to a pulsar. According to classical electrodynamics (Jackson, 1999), a rotating magnetic dipole with magnetic moment M radiates away energy at a rate

2 E˙ = M 2Ω4 sin2 α, (1.2) 3c3 where α is the angle between the spin axis and magnetic dipole. Assuming that a pulsar spins down as a magnetic dipole, we can solve for M by

1 As sourced from the ATNF pulsar database, version 1.50. http://www.atnf.csiro.au/research/pulsar/psrcat/

3 combining Equations 1.1 and 1.2. For pulsars, however, the typically desired value is the strength of the magnetic field at the surface of the star or surface dipolar magnetic field, B = M/R3, where R is the star’s radius. Solving for B gives r 3c3I p B = P P˙ = 3.2 × 1019 G P P,˙ (1.3) 8π2R6 sin2 α for I = 1045 g cm2, R = 10 km, and α = 90◦. Note that this equation gives the magnetic field strength at the equator of the star; at the magnetic poles B should be two times larger. Typical pulsars have B ∼ 1012 G, but values can range from as low as 108 G to upwards of 1014–1015 G. Solving Equations 1.1 and 1.2 for P˙ gives

8π2M 2 sin2 α P˙ = P −1 = KP −1, (1.4) 3c3I which may be integrated to find the age of the pulsar assuming K is constant. Doing so gives " # P P 2 T = 1 − 0 , (1.5) 2P˙ P where P0 is the period of the pulsar at birth. Taking P0  P gives us the characteristic age P τc = . (1.6) 2P˙ In some cases the characteristic age provides a good estimate of the true age, as with the Crab pulsar whose true age of 0.96 kyr is closely reflected by its characteristic age of 1 kyr, but in others there can be significant discrepancies. For example, the pulsar PSR J0205+6449 has τc = 5.4 kyr which is almost 7 times larger than its true age of 0.83 kyr. In general, the assumptions of pure magnetic dipole braking and of a significantly smaller

4 birth period are not necessarily good ones, so the characteristic age should not be considered a universally reliable age estimate. If we express Equation 1.4 in terms of the rotational frequency ν = 1/P , we getν ˙ = −K ν3. More generally,

ν˙ = −K νn, (1.7) where K is a constant and n is called the braking index, with n = 3 for pure magnetic dipole braking. By differentiating Equation 1.7 and solving for n while eliminating K we find n = νν/¨ ν˙ 2. Therefore, the braking index can be determined directly if the second derivative of the spin frequencyν ¨ can be measured. This measurement has been made for seven pulsars with results for n ranging from 0.9 to 2.91 (Lyne et al., 1993, 1996; Livingstone et al., 2007, 2011; Weltevrede et al., 2011; Espinoza et al., 2011a), showing that in general it is incorrect to assume that n = 3! The observed deviations from n = 3 could be explained if K is not constant, as would be the case if one or more of the moment of inertia, I, the magnetic field strength, B, or the angle between the spin and magnetic axes, α, have been changing over the lifetime of the pulsar. Alternatively, the spin-down might be due to a combination of magnetic dipole braking and braking from another source such as an outflowing stellar wind (which on its own would give n = 1). Although pulsars are normally seen to spin down in a steady fashion, some of them have exhibited sudden timing irregularities known as glitches. A glitch is an instantaneous increase in the spin frequency of the star by a small amount, on the order of ∆ν/ν ∼ 10−9–10−6. This change in ν may be permanent or it may be followed by an exponential recovery to the pre-glitch value, and it is frequently accompanied by a simultaneous change inν ˙. The first glitch ever observed was in the Vela pulsar (Radhakrishnan

5 & Manchester, 1969), and since then there have been over 300 glitches observed in over 100 different sources, including radio pulsars and magnetars (Espinoza et al., 2011b; Dib & Kaspi, 2014). Glitches are more frequent in younger pulsars, with 16 having been observed in Vela and 24 in the Crab pulsar, while glitches from the oldest pulsars are rare or nonexistent. 1.1.1 Neutron Star Structure

The canonical values for a neutron star’s mass and radius are 1.4M and 10 km, but their precise relation to each other and allowed limits depend on the star’s equation of state (EOS). The EOS gives the relation between pressure and density inside the star and also governs its internal behaviour and composition. Unfortunately, the EOS of matter at the extreme densities seen inside a neutron star is not well understood (the canonical values above imply an average density of 6.7 × 1014 g cm−3, greater even than that of an atomic nucleus), so likewise our knowledge of the neutron star interior structure is limited. Nevertheless, we provide here a description of the standard picture of what we do understand. The interior of a neutron star can be divided into several regions, as shown in Figure 1–2 (see, e.g., Shapiro & Teukolsky, 1983 or Lattimer & Prakash, 2004 for more details): Above the surface is the neutron star atmosphere, no more than a few centimeters thick with density <100 g cm−3, generally assumed to be mostly hydrogen and helium. The outermost layer of the star itself is the outer crust, composed of a solid, crystalline lattice of ionized atoms and free, degenerate electrons. Near the surface, the structure is dominated by 56Fe nuclei, but as the density increases the nuclei begin capturing free electrons and becoming more neutron rich. The outer crust ends when the density reaches 4 × 1011 g cm−3, the neutron drip point, so called because the density is so great that neutrons start to leak

6 Figure 1–2: Cross section of a neutron star. http://heasarc.gsfc.nasa. gov/docs/objects/binaries/neutron_star_structure.html

7 or ‘drip’ out of nuclei. Thus the next region, the inner crust, contains two components: the solid lattice also present in the outer crust and a sea of free, superfluid neutrons. The inner crust continues until the depth at which nuclear density, ∼2 × 1014 g cm−3, is achieved. Here, the nuclei dissolve completely and all that remains in the neutron star core is the neutron superfluid along with a small amount (∼5%) of free protons and electrons. Beyond nuclear density we reach the point at which the EOS of matter is poorly known. Therefore, so is the exact behaviour and composition of the neutron superfluid in the core, especially at the centre where the density is expected to reach 1015 g cm−3. At these extreme densities it has even been suggested that the neutrons may form into pion or kaon condensates or even dissolve into some sort of quark matter, resulting in a possibly solid inner core of exotic matter (Lattimer & Prakash, 2004). 1.1.2 Neutron Star Magnetosphere

Previously, in Section 1.1, we inferred several properties of pulsars using a model of a rotating magnetic dipole in a vacuum. For 45 years, however, the latter assumption — that pulsars rotate in a vacuum — has been known to be untrue. Goldreich & Julian(1969) determined that a neutron star’s rotating magnetic field should induce an electric field that, in turn, is balanced by a build-up of charge on the surface of the star. In the presence of a vacuum the induced external electric fields exert a force on the surface charges that is ∼1010–1012 times greater than gravity (Lorimer & Kramer, 2005), easily stripping them from the star into the surrounding vacuum. Equilibrium will be reached when the charge density external to the neutron star hits a critical value known as the Goldreich-Julian density,

ΩB B ρGJ = ce = ceP . Under these circumstances the external plasma co-rotates with the neutron star, forming a region called the magnetosphere. The

8 magnetosphere cannot extend to infinity, however; it is bounded by the

c cP light cylinder, RLC = Ω = 2π , the distance at which something would have to travel at the speed of light in order to co-rotate with the neutron star. As shown in Figure 1–3, then, the magnetic field lines surrounding the neutron star can be divided into two groups: the closed field lines, which are contained entirely within the light cylinder; and the open field lines, which extend outside the light cylinder and therefore cannot close back in on the neutron star. Charged particles in the closed field line region are contained within the magnetosphere and co-rotate with the neutron star, but those in the open field line region are able to travel along the field lines to leave the magnetosphere. Determining the mechanism by which pulsars emit radiation has been an ongoing field of research since their discovery. Generally, in order to produce emission in the magnetosphere, the plasma density must be orders of magnitude above the Goldreich-Julian density. This condition is thought to be fulfilled by plasma multiplication in magnetospheric “gap” regions, where the Goldreich-Julian equilibrium is not maintained and induced electric fields exist (Lorimer & Kramer, 2005). A basic picture of pulsar emission, from Ruderman & Sutherland(1975), proposes that such a gap is located above the magnetic pole (or polar cap) of the neutron star. As particles are pulled from the surface of the star to replenish those that escape the magnetosphere along open field lines, they are accelerated to relativistic speeds in the gap. Moving along the strong magnetic field lines, these relativistic particles emit curvature radiation that in turn induces pair cascades, creating a secondary plasma with much higher density that is believed to produce the radio emission. Polar gap models can also explain the high-energy X-ray and gamma-ray emission from pulsars (Daugherty

9 Figure 1–3: Diagram of the magnetosphere of a pulsar. Figure adopted from Lorimer & Kramer(2005).

10 & Harding, 1982, 1996), but recent results from the Fermi Gamma-Ray Space Telescope favour outer gap models (Abdo et al., 2010b), which posit that the high-energy emission is produced in gaps occurring in the outer magnetosphere between the open and closed field lines, as shown in Figure 1–3(Cheng et al., 1986; Romani, 1996). Nevertheless, despite many years of theoretical work and observations, we still do not fully understand the pulsar emission mechanism. 1.1.3 Pulsar Wind Nebulae

A feature commonly observed in energetic young pulsars is the “pulsar wind nebula” (PWN), the most classic example of which being the Crab Nebula. Such nebulae, evident particularly at radio and X-ray energies, are produced when relativistic electrons and positrons from the pulsar’s stellar wind emit synchrotron emission as they spiral in the ambient magnetic field and when they collide and interact with the surrounding medium (Kaspi et al., 2006). PWNe are generally asymmetric and frequently show complex features such as tori, jets, and wisps. Rapidly moving pulsars can also produce bow shock features as they move through the interstellar medium at supersonic speeds. The spectrum of PWNe is broadband and non-thermal. Their radio emission can be characterised by a power law,

α Sν ∝ ν , where Sν is the flux density at frequency ν, and α is the spectral index, typically falling between −0.3 and 0. Likewise, their X-ray emission

−Γ can also be described by a power law, NE ∝ E , where NE is the number of photons with energy between E and E + dE, and Γ ≡ 1 − α is the photon index, usually with Γ ≈ 2 (Gaensler & Slane, 2006). PWNe are common only to the most energetic pulsars with E˙ > 1036 erg s−1; very few are seen around pulsars with E˙ < 1035 erg s−1. In particular, the X-ray efficiency ˙ of PWNe, η = LX,pwn/E, where LX is X-ray luminosity, is observed to

11 decrease from η ≈ 10−1–10−2 for E˙ ∼ 1038 erg s−1 down to η ≈ 10−3–10−5 for E˙ ∼ 1035 erg s−1 (Kargaltsev & Pavlov, 2008). 1.1.4 Neutron Star Cooling

Soon after Baade & Zwicky(1934) first proposed their existence, it was pointed out that neutron stars, being born in supernovae, should initially be very hot, with surface temperatures above 106 K(Zwicky, 1938). The resulting thermal emission would be potentially detectable in X-rays, so research into cooling models began in earnest with the discovery of extrasolar X-ray sources in the 1960s (e.g., Morton, 1964; Chiu & Salpeter, 1964). However, the earliest neutron stars to be identified as X-ray sources were in binary systems where the X-ray emission was produced by material from the companion accreting onto the neutron star. It wasn’t until two decades later that detections of thermal emission from cooling neutron stars were made (Cheng & Helfand, 1983; Harnden et al., 1985). In particular, one unforeseen difficulty was that many of the youngest and brightest isolated neutron stars, such as the Crab pulsar, have bright magnetospheric emission that swamps out any thermal emission and renders it undetectable.

11 At birth, neutron stars have central temperatures of Tc ∼ 10 K, but

9 10 within a matter of hours to days Tc drops to ∼10 –10 K, after which it continues to cool more gradually (Shapiro & Teukolsky, 1983). There are two mechanisms by which the core of a neutron star cools: emission of neutrinos and heat transport to the surface followed by thermal photon emission. At high temperatures, i.e. for young neutron stars, neutrino emis-

5 6 7 8 sion dominates. After ∼10 –10 yr, Tc drops below 10 –10 K (depending on the exact neutrino emission process) and neutrino emission dies off, leaving photon emission as the dominant cooling mechanism. Cooling models are primarily concerned with the neutrino emission mechanism, since all cooling

12 neutron stars whose thermal radiation is observed are young enough that neutrino emission either still is or has only recently stopped being the dom- inant process (Yakovlev et al., 2008). The main mechanism for cooling via neutrino emission is known as the Urca process, which in its simplest form, called the direct Urca process (Lattimer et al., 1991), consists of beta decay followed by electron capture:

− − n → p + e +ν ¯e, p + e → n + νe.

This process, however, is forbidden in the outer core of a neutron star because momentum cannot be conserved. Instead, the involvement of a spectator neutron or proton is required to conserve momentum, but the resulting modified Urca process (Chiu & Salpeter, 1964) is less efficient by upwards of seven orders of magnitude! The dominant neutrino emission process in the inner core depends greatly on its composition and equation of state. If the proton fraction is sufficiently high (&11%), the direct Urca process is viable, and the presence of more exotic material, such as pion condensates or quark matter, facilitates other Urca processes more efficient than the modified Urca process (Pethick, 1992). For any given neutron star cooling model, a corresponding plot of temperature versus age, or cooling curve, can be constructed, although first the temperature of the core, Tc, must be related to an observable quantity, the surface temperature, Ts. Typically, the core is taken to be isothermal

10 −3 out to a density of ρb ≈ 10 g cm . The region with ρ < ρb is called the stellar envelope and is on the order of a hundred meters thick (Page et al., 2006). Neutrino emission is assumed to be negligible there, allowing the heat transport equations to be solved easily to compute a temperature gradient that relates Tc to Ts. This relation can be greatly affected by both

13 the composition of the stellar envelope and the configuration of the star’s magnetic field, since heat is transported much more easily along magnetic field lines than against them (Geppert et al., 2004). Additionally, the neutron star atmosphere must also be considered because it can significantly affect the shape of the emitted thermal spectrum. In particular, models of hydrogen atmospheres show a high-energy tail in the spectrum such that a naive fit to a simple blackbody overestimates the surface temperature by up to a factor of 2–3 (Romani, 1987; Zavlin et al., 1996). 1.2 Magnetars

1.2.1 History of AXPs and SGRs

The first detection of a member of the class of neutron stars today identified as “magnetars” was of a single burst on 1979 January 5 from the source known today as SGR 1806−20 (Mazets & Golenetskii, 1981). This was followed on 1979 March 5 by an enormous flare from the direction of the star-forming Dorado region in the Large Magellanic Cloud (LMC) (Mazets et al., 1979b), consisting of an initial bright spike followed by a 3-minute long tail with 8 s pulsations visible in the declining flux. These events, along with further bursts from the LMC and repeated bursts from what today is known to be magnetar SGR 1900+14 (Mazets et al., 1979a; Mazets & Golenetskii, 1981) were originally classified as a subtype of classical gamma- ray bursts (GRBs) due to their shorter durations and somewhat softer spectra than most GRBs. Likewise, the sources of the repeated bursts were designated as “Soft Gamma Repeaters” (SGRs). The pulsations seen in the tail following the large flare were strongly suggestive of a neutron-star origin, but that these sources truly represented a distinct class of gamma-ray bursters was not fully recognized until 1983 when SGR 1806−20 underwent a major burst episode, also known as an outburst (Laros et al., 1987).

14 Both Galactic sources were noted to be very close to the Galactic Plane, suggesting youth, a conclusion supported by the coincidence of the LMC source with the supernova remnant (SNR) N49 (Cline et al., 1982). Meanwhile, Fahlman & Gregory(1981) reported an unusual 7-s X-ray pulsar, 1E 2259+586, in the Galactic supernova remnant CTB 109. It was originally thought to be a low-mass X-ray binary albeit without any obvious companion, but later observations revealed it to be spinning down steadily with no signs of orbital modulation. In the ensuing years the source was recognized as being similar to a handful of other ‘anomalous’ sources such as 4U 0142+61 and 1E 1048.1−5937 (see Hellier, 1994; Duncan & Thompson, 1996; van Paradijs et al., 1995; Mereghetti & Stella, 1995), distinguished by their bright X-ray pulsations at few-second periods, X-ray luminosities far greater than could be explained via rotation power, but no apparent companions from which to accrete. These distinctions led to the sources being termed “Anomalous X-ray Pulsars” (AXPs) and this descriptor has stuck. Duncan & Thompson(1992) proposed that very strongly magnetized neutron stars could be the origin of SGR emission, thereby coining the term “magnetar.” Thompson & Duncan(1995) demonstrated that many SGR phenomena are readily explained by a model in which spontaneous magnetic field decay serves as an energy source for both the bursts and any persistent emission. They cited not only energetics arguments but also the need for a high B field to spin down a young neutron star from tens to hundreds of ms (thought to be the typical birth spin period range) to several seconds, within a supernova remnant lifetime. Thompson & Duncan(1996) further argued that AXPs are also magnetars, with their X-ray luminosities powered by magnetic field decay. Shortly thereafter,

15 Kouveliotou et al.(1998) observed X-ray pulsations from SGR 1806 −20 and measured it to be spinning down at a rate that implied its magnetic field to be B = 8 × 1014 G, consistent with the model prediction and a powerful confirmation of the magnetar picture. Following this, 2 SGR-like bursts were detected from the AXP 1E 1048.1−5937 (Gavriil et al., 2002), and an SGR-like outburst involving over 80 bursts in a few hours was seen from 1E 2259+586 (Kaspi et al., 2003), thereby unifying AXPs and SGRs observationally as predicted by Thompson & Duncan(1996). Since then, the distinction between AXPs and SGRs has been further blurred, with practically all sources having shown characteristics of both: bursting has now been shown to be a generic behaviour of so-called AXPs (e.g. Gavriil et al., 2004; Woods et al., 2005; Kaneko et al., 2010; Scholz & Kaspi, 2011) and AXP-like behaviour (namely, absence of bursts for long periods) has been seen in objects originally deemed SGRs, including the original LMC source, SGR 0526−66 (Kulkarni et al., 2003). It is clear that there exists a continuous spectrum of behaviour, ranging from anomalously high quiescent X-ray luminosity to occasional bursting and major flaring, in the single class of objects we now call magnetars. At the time of this writing, there are 21 confirmed magnetars. Ten of them were discovered as SGRs; that is, they were first detected by their gamma-ray bursts. These sources are SGRs 1806−20 (Mazets & Golenetskii, 1981), 0526−66 (Mazets et al., 1979b), 1900+14 (Mazets et al., 1979a; Kouveliotou et al., 1993), 1627−41 (Woods et al., 1999), 0501+4516 (Rea et al., 2009b), 0418+5729 (van der Horst et al., 2010), 1833−0832 (G¨o˘g¨u¸s et al., 2010b), Swift J1822.3−1606 (Scholz et al., 2012; Rea et al., 2012b), Swift J1834.9−0846 (Kargaltsev et al., 2012), and SGR J1745−2900 (Mori et al., 2013). The remaining 11 magnetars were first discovered by as AXPs

16 by detection of their persistent X-ray emission (with the exception of PSR J1622−4950, which was first discovered in radio); they are 1E 2259+586 (Fahlman & Gregory, 1981), 1E 1048.1−5937 (Seward et al., 1986), 4U 0142+61 (Israel et al., 1994), 1E 1841−045 (Vasisht & Gotthelf, 1997), 1RXS J170849.0−400910 (Israel et al., 1999), XTE J1810−197 (Israel et al., 2004a), CXOU J010043.1−721134 (McGarry et al., 2005), CXOU J164710.2−455216 (Israel et al., 2007), 1E 1547.0−5408 (Camilo et al., 2007a), PSR J1622−4950 (Levin et al., 2010), and CXOU J171405.7−381031 (Sato et al., 2010; Halpern & Gotthelf, 2010a). In addition, there are 5 candidate magnetars: AX J1845.0−0258 (Torii et al., 1998), SGRs 1801−23 (Cline et al., 2000), 1808−20 (Lamb et al., 2003) and 2013+34 (Sakamoto et al., 2011), and AX J1818.8−1559 (Mereghetti et al., 2012). In Chapter5, we present a detailed summary of the observed properties of all known magnetars and magnetar candidates, but the most up-to-date information can be found in the McGill Online Magnetar Catalog2. 1.2.2 The Magnetar Model

Thompson & Duncan(1993) were the first to propose a mechanism for the formation of very high magnetic fields in newborn neutron stars. They suggested that a neutron star born with a sufficiently short spin period,

P . 10 ms, would undergo a turbulent dynamo process that, while lasting for mere seconds, could amplify the initial magnetic field to strengths as high as 1016 G. This dynamo process was also predicted to impart a large kick velocity to the resulting magnetar and to dump large amounts of energy into the surrounding supernova remnant (Duncan & Thompson, 1992). However, measurements have so far found no strong evidence that

2 http://www.physics.mcgill.ca/~pulsar/magnetar/main.html

17 the proper motions of magnetars are any higher than those of the standard pulsar population (Tendulkar et al., 2013), and estimates of the explosion energies of SNRs containing magnetars are close to the canonical value of 1051 erg (Vink & Kuiper, 2006). An alternate formation scenario supposes that the conventional “fossil field” theory, i.e., that neutron star magnetic fields are the result of magnetic flux conservation during core collapse, can produce magnetar-strength fields. In this case, magnetars would simply be the product of massive progenitor stars with the highest magnetic fields (Ferrario & Wickramasinghe, 2006), a claim bolstered by the evidence that the magnetar CXOU J164710.2−455216 must have a massive progenitor (Muno et al., 2006). Many magnetars are observed to have persistent X-ray emission with luminosity on the order of ∼1035 erg s−1, a property which, according to Thompson & Duncan(1996), arises due to the decay and diffusion of their strong magnetic fields. In neutron stars with fields stronger than 1014–1015 G, the process of ambipolar diffusion leads to significant heating of the core, although this effect only begins to dominate over standard cooling rates at an age of ∼103 yr. Therefore, magnetars between that age and ∼104 yr, the timescale of the ambipolar diffusion process, should be significantly hotter and more luminous than a cooling neutron star of similar age but with a smaller magnetic field. Likewise, the process of ohmic dissipation is thought to be a significant source of heating in the crusts of neutron stars with fields stronger than 1013 G(Pons et al., 2007). It is also thought to become dominant at an age of ∼103–104 yr for magnetars, providing an alternative or additional means for their high temperatures and luminosities at those ages.

18 As the strong magnetic field of a magnetar diffuses out of the core, it is expected to meet resistance from the star’s rigid crust. Thompson & Duncan(1995) explained that eventually the built-up stress will overwhelm the shear strength of the crust and cause it to crack, prompting a sudden release and rearrangement of the magnetic field outside the star. This interchange and reconnection of magnetic fields in the magnetosphere produces fireballs that are observed as magnetar bursts and flares. Small- scale cracking of the crust would create similarly small-scale fireballs that radiate away quickly, corresponding to the smaller SGR-like bursts, whereas a large enough fracture could produce a fireball that engulfs the entire magnetosphere. Such a phenomenon would be energetic enough to produce a giant flare much like the 1979 March 5 event from SGR 0526−66. The magnetar model was further developed by Thompson et al.(2002) to include large-scale twisting of magnetic field lines in the magnetosphere. They found that such twists support strong electrical currents that have two major effects. First, charged particles flowing along closed field lines and falling back upon the star could provide an additional source of heating of the surface comparable to that produced by ambipolar diffusion in the core. Second, the currents are capable of upscattering soft thermal photons to produce the harder, non-thermal component often present in the persistent X-ray emission of magnetars. Furthermore, they contend that the variations in the spin-down rate that are observed in some sources are due to torques induced by the twisting or untwisting of the external magnetic field even as the strength of the field remains fixed. In particular, the enhanced spin- down seen in SGR 1900+14 after its 1998 giant flare and SGR 1806−20 in the years preceding its 2004 giant flare are consistent with this twisted magnetosphere model. The spectral hardening and increased burst rate that

19 also preceded the giant flare of SGR 1806−20 are consistent, as well. In fact, under this model, the SGR giant flares should be caused by or associated with a sudden relaxation or untwisting of the field. A recent challenge for the magnetar model has been the discovery of so-called low-field magnetars. These sources were discovered by the detection of multiple bursts and, except for their low spin-down rates and accordingly low inferred B fields, they display otherwise typical magnetar properties. The two such identified magnetars are SGR 0418+5729, which has an inferred magnetic field of B = 6 × 1012 G(Rea et al., 2013) that would not be unusual for a normal pulsar, and Swift J1822.3−1606, with B = 1.4 × 1013 G(Scholz et al., 2014), a value below those inferred for a not-insignificant number of rotation-powered pulsars. Given these results, it seems that a large dipolar magnetic field is not necessary for magnetar activity. Indeed, what is really important in the magnetar model is the strength of the total field, including the multipolar or toroidal components that do not contribute to the spin-down. SGR 0418+5729, in particular, has been suggested to be an old magnetar with a much stronger toroidal than dipolar magnetic field that has already undergone substantial decay (Turolla et al., 2011). Overall, the magnetar model has been very successful in explaining most observations, though there is yet much research to be done, both the- oretically and in monitoring and observation. The origins and mechanisms for the persistent X-ray emission and bursts from magnetars are not yet fully understood, and we have only begun to study their radiative properties outside of the soft X-ray band. The recently discovered low-field magnetars, too, demand further study as to where in the picture they fit. Finally, we note that some alternative models for AXPs and SGRs have been proposed,

20 including a fall-back disk model that has the sources accreting from sur- rounding debris (e.g., Ertan et al., 2007, 2009), a massive white dwarf model (e.g., Malheiro et al., 2012), and also a quark nova model (Ouyed et al., 2007a,b). Although these models are interesting and have their merits, the current evidence to support these pictures for the overall magnetar population is weak. We consider them no further in this thesis but refer the interested reader to the above references. 1.3 High-B Pulsars and Magnetars

Although the magnetars can be generally distinguished from the rest of the pulsar population by their high inferred surface dipolar magnetic fields, considerable overlap between the populations is evident in Figure 1–1. In particular, there are a half-dozen otherwise ordinary radio pulsars found

13 with B greater than the quantum critical field, BQED = 4.4 × 10 G, defined as the magnetic field in which an electron has cyclotron energy equal to its rest mass. Some models predict that given such strong magnetic fields, conventional radio emission from pulsars should be suppressed (Baring & Harding, 1998), yet nearly all of these high-B pulsars are detected normally in radio. Likewise, none of them show evidence for conventional magnetar- like emission. Some, such as PSRs J1847−0130 and J1718−3718, have B greater than that measured for the bona fide magnetar 1E 2259+586, yet no anomalous X-ray emission (McLaughlin et al., 2003; Kaspi & McLaughlin, 2005). Uncertainties in inferred B from spin-down can be substantial (e.g. Harding et al., 1999; Spitkovsky, 2006), but still, in the magnetar picture, some evidence for anomalous X-ray emission is reasonably expected in some high-B radio pulsars. The idea that high-B pulsars might exhibit anomalous X-ray emission was confirmed several years ago by the discovery of SGR-like X-ray bursts

21 and a long-lived X-ray flux enhancement from what was previously thought to be a purely rotation-powered pulsar. Gavriil et al.(2008) found that the high-B pulsar PSR J1846−0258 at the centre of the SNR Kes 75 emitted several SGR-like bursts in 2006, contemporaneous with a flux enhancement and a rotational glitch (see also Kumar & Safi-Harb, 2008; Ng et al., 2008; Livingstone et al., 2010). This is the first pulsar known to have quiescent X-ray luminosity that could be rotation-powered, and indeed has many properties of rotation-powered pulsars, while showing obvious magnetar-like behaviour. Its small characteristic age of 884 yr lends further credence to the idea that PSR J1846−0258 could be a very young magnetar, and one of the “missing links” in the hypothesized high-B pulsar/magnetar evolutionary chain. The past several years have also seen unusual magnetar discoveries that may provide clues to the connections between the two populations. Most notable are the two low-field (B < BQED) magnetars, SGR 0418+5729 (Rea et al., 2010) and Swift J1822.3−1606 (Rea et al., 2012b; Scholz et al., 2014), discussed in the previous section. Had they been found by something other than the detection of SGR-like bursts they may not have been identified as magnetars. Another example is PSR J1622−4950, a radio pulsar discovered in the High Time Resolution Universe survey whose rotational and X-ray properties identified it as a magnetar (Levin et al., 2010), making it the only magnetar not to be discovered at X-ray or soft gamma wavelengths. Another notable group of high-magnetic-field neutron stars are the X-ray-isolated neutron stars (XINSs3). The XINSs are a group of nearby (distance ≤ 500 pc) neutron stars characterized by soft thermal X-ray

3 Also known as X-ray dim isolated neutron stars (XDINSs).

22 spectra and no detected radio emission (see Haberl, 2007 and Turolla, 2009 for reviews). Although radio quiet, it is unclear whether they are intrinsically so, or whether it is merely the case that their radio emission does not cross our line of sight (Kondratiev et al., 2009). Those XINSs with detailed timing measurements have spin periods in the range of 3–11 s, high inferred magnetic fields (B ∼ 1–3 × 1013 G), and characteristic ages of ∼106 yr (Kaplan & van Kerkwijk, 2009, 2011; van Kerkwijk & Kaplan, 2008). The XINSs are considerably hotter and more X-ray luminous than similarly aged ordinary rotation-powered pulsars (Kaplan & van Kerkwijk, 2009), although they show no bursting activity and are less luminous than the magnetars. From an analysis of many isolated neutron stars, including magnetars, XINSs, and ordinary rotation-powered pulsars, Pons et al.(2007) saw evidence for a correlation between the blackbody temperature, T , and inferred magnetic field, B, with T ∝ B1/2 seeming to hold over three orders of magnitude. They suggested that such a correlation could be explained by magnetic-field decay heating the crusts of neutron stars with B ≥ 1013 G. Further work on this subject has led to the development of a model for the magnetothermal evolution of neutron stars (Aguilera et al., 2008; Pons et al., 2009; Popov et al., 2010; Vigano et al., 2013), incorporating the effects of high magnetic fields where standard cooling models do not. This model attempts to unite all of the high-magnetic-field neutron stars, with the different populations arising due to differences in age or in initial magnetic field strength. For example, the model finds that the bulk of the magnetar population will evolve into the region of the P –P˙ diagram inhabited by the XINSs and low-field magnetars. The magnetothermal evolution model predicts that neutron stars with B ≥ 1013 G should have

23 a higher temperature than would be predicted by standard cooling models and, indeed, should be hotter than neutron stars of similar age but lower magnetic field. Evidence supporting the latter prediction was shown in Zhu et al.(2011), and in Chapter3 we present updated results that support the same conclusion.

24 CHAPTER 2 X-ray Astronomy and Instrumentation 2.1 History of X-ray Astronomy

The X-ray band is the part of the electromagnetic spectrum with en- ergies of about 0.1 keV to 500 keV. With such high energies, X-ray photons are easily able to ionize any atoms they run into, but in doing so they will lose energy and be absorbed in a process known as the photoelectric effect. In particular, they are easily absorbed by the Earth’s atmosphere, meaning that ground-based observatories have no hope of detecting X-rays from extraterrestrial sources. Thus, the field of X-ray astronomy did not begin until after World War II when we started to develop space technology to bring objects above and beyond the atmosphere of the planet. The first attempts to detect X-rays from space were made in the late 1940’s using detectors attached to rockets that flew above the atmosphere and then parachuted back to the ground. The first successful detection, of X-ray emission from the Sun, was made on 1948 August 5 (Keller, 1995). At the time, astronomers expected that extrasolar objects would not produce enough X-rays to be detected, but in 1962 a rocket sent up with the purpose of detecting solar X-rays reflecting off of the Moon discovered a new X-ray source even brighter than the Sun or Moon. This source, labelled Scorpius X-1 and later determined to be an accreting neutron star (Giacconi et al., 1962; Shklovsky, 1967), proved that there could be many objects in the sky that were able to be studied in X-rays. The first dedicated X-ray satellite mission was the UHURU space observatory, launched in 1970 December. Its main goals were to perform an

25 X-ray survey of the sky and monitor variable X-ray sources. Many of these variable sources were discovered to be binary star systems with accreting neutron stars, though a few were determined to have masses greater than the allowed upper limit for neutron star mass, making them black hole candidates. UHURU also discovered X-ray emission from active galactic nuclei and from hot gas in clusters of galaxies. The final UHURU (4U) X- ray catalog (Forman et al., 1978) contained positions and 2–6 keV intensities for 339 Galactic and extragalactic sources. The next major step in X-ray astronomy was the launch in 1978 November of the first focusing X-ray telescope, the Einstein observatory (or HEAO-2, the High Energy Astronomy Observatory 2) (Giacconi et al., 1979). This satellite was equipped with Wolter type I X-ray optics (see Section 2.2) that allowed for angular resolution on the order of a few arcseconds, meaning extended objects could be imaged and resolved for the first time. Additionally, its increased sensitivity and ability to make deep surveys of small areas of the sky meant that it was the first telescope to detect X-rays from many different kinds of astronomical objects, including ordinary stars and galaxies. Launched in 1990 June, the ROSAT satellite (Truemper, 1993) was the first imaging X-ray telescope to perform an all-sky survey. With a similar angular resolution as that of the Einstein observatory and over four times the collecting area, the ROSAT All-Sky Survey detected almost 19,000 X-ray sources in the Bright Source Catalogue (Voges et al., 1999) alone, and well over 100,000 X-ray sources total. The first X-ray telescope to use CCD detectors was the ASCA obser- vatory (Tanaka et al., 1994), launched in 1993 February. These detectors, known as the Solid-state Imaging Spectrometers (Burke et al., 1994), were

26 located at the focus of two of the observatory’s four telescopes and provided spectral energy resolution 5–10 times better than previous X-ray detectors. ASCA was also the first imaging satellite able to cover a wide range of X-ray energies up to 10 keV. Its primary scientific purpose was X-ray spectroscopy of astrophysical plasmas. In 1995 December, the Rossi X-ray Timing Explorer (RXTE; Jahoda et al., 2006) was launched. Unlike most of the other missions mentioned here, RXTE had no focusing telescopes onboard and was not intended for imaging. Rather, it was notable for the 1-µs time resolution on its primary instrument, the Proportional Counter Array, allowing for very precise timing of even the fastest millisecond pulsars. Other instruments included a hard X-ray (15–250 keV) detector and an all-sky monitor that covered 80% of the sky in each of the spacecraft’s 90-minute orbits, the latter of which was designed to monitor the variability of bright X-ray sources and report the appearance of transient events. In 1999, the XMM-Newton and Chandra observatories were launched. The telescopes on XMM consist of 58 nested Wolter type I mirrors, result- ing in a total collecting area of over 4000 cm3, an order of magnitude larger than that of ROSAT and more than any other imaging X-ray telescope ever made (Jansen et al., 2001). Conversely Chandra’s telescope has a signifi- cantly smaller collecting area with only 4 Wolter mirrors, but the mirrors are so well manufactured that they provide an unprecedented sub-arcsecond angular resolution (Weisskopf et al., 2000). As such, the two telescopes complement each other well. Like ASCA, both XMM and Chandra are out- fitted with CCD detectors, providing much better image quality and energy resolution than was previously possible. Furthermore, both spacecraft are also equipped with grating instruments for high resolution spectroscopy, the

27 High and Low Energy Transmission Grating Spectrometers on Chandra and the Reflection Grating Spectrometer on XMM (den Herder et al., 2001). XMM is also outfitted with another telescope, the Optical Monitor, which simultaneously observes X-ray targets in the ultraviolet and optical bands. For a more detailed description of the XMM-Newton spacecraft and X-ray telescopes, see Section 2.4. Although primarily a gamma-ray observatory, the INTEGRALspacecraft, launched in 2002 October and still ongoing, is also a sensitive hard X-ray telescope. Its primary instruments are the imaging telescope IBIS (Uber- tini et al., 2003), which operates between 15 keV and 10 MeV, and the spectrometer SPI (Vedrenne et al., 2003), which is sensitive to the energy range 20 keV–8 MeV. The spacecraft is also equipped with the JEM-X X-ray monitor (3–35 keV) and the OMC optical monitor (500–600 nm) to allow simultaneous observation of targets in those two energy bands. Besides observing targeted sources, INTEGRAL is also an effective instrument for detecting gamma-ray bursts (GRBs) due to the large field of view of the IBIS instrument. The ongoing, multiwavelength Swift Gamma-Ray Burst Mission was launched in 2004 November to study GRB science. The instruments onboard are the Burst Alert Telescope (BAT; Barthelmy et al., 2005), which mon- itors a large fraction of the sky in the energy range 15–150 keV for GRBs, and the X-ray Telescope (Burrows et al., 2005) and Ultraviolet/Optical Telescope, intended to quickly follow up on GRBs detected by the BAT and observe their afterglows in soft X-ray, ultraviolet, and optical bands. As well as gamma-ray bursts, Swift is an excellent telescope for detecting and monitoring other transient events in gamma and X-rays such as magnetar bursts and outbursts.

28 Another satellite intended to be capable of high resolution spectroscopy was Suzaku (Mitsuda et al., 2007), launched in 2005 July. Its primary instrument, the X-ray Spectrometer (XRS; Kelley et al., 2007), consists of an array of cryogenically-cooled microcalorimeters, the first such instrument to be successfully launched. Unfortunately, within a few weeks of launch the XRS failed when the coolant was lost. Suzaku’s other instruments, the four X-ray Imaging Spectrometers (Koyama et al., 2007) and the Hard X-ray Detector (10–600 keV; Takahashi et al., 2007), remain operational to the current day. Finally, the most recent X-ray mission to be launched is NuSTAR (Harrison et al., 2013) in 2013 June. NuSTAR is a focusing telescope that operates from 3 to 79 keV, making it the first telescope capable of focusing hard X-rays above ∼10 keV. 2.2 Wolter Mirrors

In standard optics, images are made using lenses and mirrors to focus light onto a detector. At X-ray energies, however, these focusing techniques do not work; instead the photons are simply absorbed by or pass through the lens or mirror. Sufficiently soft X-rays (<20 keV), though, can be reflected if they strike a mirror at a very shallow angle, i.e. with an angle of incidence less than 1◦. Based on this fact, Hans Wolter (1952) introduced three designs for focusing X-ray optics, originally intended for X-ray microscopy but quickly recognized as applying to X-ray telescopes as that branch of astronomy developed. The three configurations of Wolter mirrors, distinguished here as type I, type II, and type III, are shown in Figure 2–1. In all three of them, X-rays reflect twice, off of two differently shaped sections of mirror, because under the restriction of grazing incidence, a single mirror cannot properly focus light. The type I configuration consists

29 Figure 2–1: Diagram of the three types of Wolter X-ray optics. http://www.ess.sunysb.edu/fwalter/AST443/xrga.html

30 of one mirror in paraboloid shape and a second one in a hyperboloid shape and is by far the most important to X-ray astronomy because, unlike the other designs, multiple sets of type I Wolter mirrors can be nested together. The effective collecting area of a single set of mirrors is much smaller than its physical area (reduced by a factor roughly proportional to the sine of the grazing angle), but nesting many of them in a coaxial arrangement with identical focal lengths can greatly increase the overall effective area of the telescope. For example, the telescopes on ASCA were made up of 120 nested sets of mirrors. The type II Wolter mirror design also consists of paraboloid and hyperboloid mirrors; with its longer focal length, it is preferred for spectroscopy and finds use in some solar X-ray observatories, but since it cannot be nested this is not common. Finally, the type III configuration is composed of paraboloid and ellipsoid mirrors and has not been used for X-ray astronomy. 2.3 CCDs

Charge-coupled devices (CCDs) are nowadays the most common type of detector in X-ray astronomy, found at the focus of imaging X-ray telescopes such as XMM, Chandra, Swift, and Suzaku. They are able to record the position, energy, and arrival time of X-rays between 0.1 and 15 keV. CCDs operate based on the photoelectric effect. An X-ray photon that strikes the CCD can get photoelectrically absorbed by the material, which ionizes atoms in it to create electron-hole pairs. Normally these electrons and holes would recombine, but by applying an electric field to the material, they can be separated and the electrons led to a readout circuit where the resulting current can be recorded and measured. The material in a CCD cannot be conductive because that would allow leakage currents that would act as significant noise in measurement, but typical insulators have many

31 imperfections with excess electrons or holes that can affect measurements. For these reasons, semiconductors, which can be manufactured to be extremely pure, are used in CCDs. Furthermore, semiconductors can be doped with impurities to alter their properties, allowing for CCDs to be more complex than the basic picture above. By selectively doping the material in a CCD, narrow channels are formed and connected to separate readout circuits. A series of potential wells are then set up in each channel to trap electrons freed by X-ray photons, and by moving the potential wells the electrons can be transported to the readout electronics. In this way the CCD is divided into pixels, which allows one to record the location on the detector of incident X-rays. In addition to the location, X-ray CCDs are also capable of measuring the energy of incident photons. Because the ionization energy of silicon in a semiconductor is ∼3.7 eV, a single X-ray can create dozens to thousands of electron-hole pairs, proportionate to its energy. Therefore, as long as only one X-ray photon has struck a CCD pixel in a single readout cycle, its energy can be reconstructed. Similarly, under these circumstances it is a simple matter to record the time of arrival of each photon as the pixels are read out, in which case the time resolution is equal to the length of one readout cycle. For more information see Longair (1992) and Arnaud et al.(2011). 2.4 Description of the XMM-Newton Observatory

In Chapters 3 and 4 we present analyses of X-ray observations taken with the European Photon Imaging Camera (EPIC) onboard the XMM- Newton observatory. Therefore, we provide here a description of the EPIC instruments as well as the satellite itself. On 1999 December 10, the XMM-Newton space observatory (Jansen et al., 2001) was launched into Earth orbit by the European Space Agency.

32 Figure 2–2: View of the XMM-Newton observatory. Figure adopted from Lumb et al.(2012).

The spacecraft has a mass of 3800 kg and is 10 m long; its orbit has a 48 hour period and is highly eccentric, with a perigee of 7000 km and apogee of 114,000 km. On board are three X-ray telescopes, each comprised of 58 Wolter type I mirrors nested in a coaxial and confocal configuration (see Figure 2–2). The mirrors have a focal length of 7.5 m and the diameter of the largest one is 70 cm, resulting in an effective area (at 1 keV) of about 1500 cm2 for each telescope. At the end of the telescopes lie the three X-ray CCD cameras which make up the EPIC instrument. One camera contains twelve pn CCDs and is called the pn camera, while the other two consist of seven MOS CCDs and are known as the MOS cameras. The telescopes leading to the MOS cameras are equipped with gratings that divert about half of the incoming X-ray flux to another instrument, the Reflection Grating Spectrometer (RGS), such that only 44% of the original flux reaches the MOS CCDs; the pn camera sees no such obstructions.

33 Figure 2–3: Layout of the CCDs in the MOS and pn cameras. http://xmm.esac.esa.int/external/xmm_user_support/documentation/ sas_usg/USG.pdf

The pn camera’s twelve 3×1 cm2 CCDs are laid out as shown in Figure 2–3 to form a 6×6 cm2 square that covers about 97% of the 300 diameter field of view of the telescope. The pixel size of each CCD chip is 150 µm × 150 µm, which corresponds to 400.1 × 400.1 on the sky, chosen to be slightly better than the angular resolution of the telescope (its point spread function has a 600.6 FWHM). The quantum efficiency of the pn camera is greater than 0.5 between 0.15 and 15 keV, and within a narrower energy range of 0.4 to 10 keV it is above 0.9. As previously discussed, X-ray CCDs provide data on the spectrum of incident X-rays as long as no more than one photon hits each pixel per readout cycle. However, if the readout cycle is too long or the source is too bright and two or more photons hit a pixel before it is read out, the electronics will assume that the multiple photons were instead a single

34 photon with higher energy. This phenomenon is known as “pile-up,” and a large amount of pile-up events will produce an incorrect observed source spectrum from which it may be difficult or impossible to recover the original spectrum. The pn camera, therefore, is designed to be operated in one of several different readout modes depending on the brightness of the source being observed. In the ‘extended full frame’ and ‘full frame’ modes the entirety of the CCD chips are in operation with readout cycles that are 199.1 ms and 73.4 ms long, respectively, intended for observing faint extended and point sources. The ‘large window’ mode is used for point sources of medium brightness; in this mode only half of each chip is active to facilitate a 47.7-ms long readout cycle. Likewise, the ‘small window’ mode for observing bright point sources restricts the field of view even further, down to a small section of one chip, to reduce to readout cycle to 5.7 ms. Finally, in the ‘timing’ and ‘burst’ modes, not only is only part of one chip active, but the position data along one axis is given up to attain even better time resolution (30 µs and 7 µs, respectively). The two MOS cameras each consist of 7 CCDs arranged as in Figure 2– 3, slightly overlapping each other to minimize the dead zones on the edge of each chip. The pixel size on the MOS CCDs is 40 µm × 40 µm, or 100.1 × 100.1 on the sky, smaller than that on the pn camera to go with the slightly better angular resolution of the MOS telescopes (FWHM of 600.0 for MOS1 and 400.5 for MOS2). The MOS CCDs have worse quantum efficiency than the pn; in the 0.3–10 keV range it varies from as low as 0.2 up to 0.9, and it is consistently above 0.5 only between 1 and 6 keV. Additionally, because half of the incident flux is diverted for the MOS to the RGS, the two MOS cameras collectively receive fewer photons than the pn camera to begin with.

35 The MOS cameras are, like the pn, equipped with several different readout modes for sources of differing brightness. As with the pn, the entire CCD is in operation in full frame mode, while in large window and small window modes only part of it is active (there is no extended full frame mode). These three modes have readout cycles that are 2.6 s, 0.9 s, and 0.3 s long, respectively. The MOS cameras also have a timing (but not burst) mode that reduces the field of view and discards position data in one dimension for a time resolution of 1.75 ms. When operating in imaging modes, the EPIC CCDs will register photons not only during the integration time, but also when charges are being transported along the chip to the readout electronics. A photon that strikes during this interval will be assigned the wrong position and its energy information will be improperly calibrated. Such events are called “out-of-time” events, and they are visible in images as vertical streaks overlapping any sufficiently bright source. The fraction of out-of-time events differs for each camera and mode and is most severe for the pn full frame mode (6.3%) and least severe for the pn large window mode (0.16%) and MOS full frame mode (0.35%). Finally, all three EPIC cameras are outfitted with four different filters to use for observing: two thin filters, one medium filter, and one thick filter. In addition to X-rays, the pn and MOS CCDs are also sensitive to UV and optical light, so the filters are designed to block these potential contaminants. If a bright optical source is present in the field of view, one of the thicker filters should be used. On the other hand, the filters also block low-energy X-rays, especially below 1 keV, so a balance must be struck when observing very soft X-ray sources.

36 CHAPTER 3 X-ray Detection and Temperature of the High-B PSR J1734–3333 The contents of this chapter are based on two papers published in the Astrophysical Journal,“XMM-Newton X-ray Observation of the High- magnetic-field Radio Pulsar PSR J1734−3333” (Olausen et al., 2010) and “X-Ray Observations of High-B Radio Pulsars” (Olausen et al., 2013). Where necessary, analysis and discussion from Olausen et al.(2010) has been updated in this chapter to include the additional observations reported on in Olausen et al.(2013). 3.1 Introduction

PSR J1734−3333 is a radio pulsar with period P = 1.17 s and period derivative P˙ = 2.3 × 10−12 that was discovered in the Parkes Multibeam Survey (Morris et al., 2002). Its spin parameters imply a spin-down lumi-

˙ 34 −1 nosity of E = 5.6 × 10 erg s , characteristic age of τc = 8.1 kyr, and an inferred surface dipolar magnetic field of B = 5.2 × 1013 G, which is among the highest of all known radio pulsars and similar to those of bona fide magnetars such as 1E 2259+586 (B = 5.9 × 1013 G; Kaspi et al., 1999). It has a radio dispersion measure (DM) of 578 pc cm−3 which, based on the NE2001 model for Galactic free electron density (Cordes & Lazio, 2002), gives a best-estimated distance to the pulsar of 6.1 kpc (although these dis- tance estimates typically have large uncertainties of 25% or more). Based on its unusually low braking index (n = 0.9 ± 0.2), Espinoza et al.(2011a) suggested that the pulsar’s magnetic field is growing, i.e., its trajectory on a conventional P/P˙ diagram (see Figure 1–1 or 5–10) is up and to the right, toward the region occupied by the magnetars. This makes PSR J1734−3333

37 Table 3–1: Summary of XMM-Newton Observations of PSR 1734−3333

ObsID Date Detector Time Resolution Exposurea (s) (ks) 0553850101 2009 Mar 9 pn 0.048 8.7 MOS1/2 2.6 10.6 0653320101 2011 Mar 11 pn 0.048 42.8 MOS1 2.6 59.8 MOS2 2.6 64.5

a The exposure time is dead-time corrected and has intervals of high back- ground flaring removed. a good candidate for exhibiting magnetar-like anomalous X-ray emission. For this reason, we obtained XMM-Newton observations of this source which we report on here. 3.2 Observations and Results

We analyzed two observations of PSR J1734−3333 taken with the XMM-Newton observatory (Section 2.4): a short, 10 ks observation carried out on 2009 March 9–10, and a deep, 125 ks one performed two years later on 2011 March 11–12. In both observations, the EPIC pn camera was operating in large-window mode and the EPIC MOS cameras in full- window mode; the medium filter was in use for all three cameras. Details of the two observations are summarized in Table 3–1. The data from both observations were analyzed with the XMM Science Analysis System (SAS) version 11.0.01. To search for times of high background flaring that are known to sometimes affect XMM-Newton data, we extracted light curves from over the entire field of view of all three cameras. The 2011 observation was heavily affected, with over half the exposure length contaminated by background flares. The 2009 observation showed no such problems.

1 See http://xmm.esac.esa.int/sas/

38 32:50.0

-33:33:00.0

10.0

20.0

30.0 Declination (J2000)

40.0

50.0 17:34:30.0 29.0 28.0 27.0 26.0 25.0 Right Ascension (J2000)

Figure 3–1: XMM-Newton image of the PSR J1734−3333 field in the 0.5– 3.0 keV band, smoothed by a Gaussian kernel with σ = 300. The radio timing position is shown by the ellipse, and the crosses denote the positions of op- tical sources from the USNO-B1.0 catalog. The optical source closest to the X-ray source (NOMAD Catalogue ID 0564-0621454) is represented by the cross marked with a box.

3.2.1 Imaging

In order to find a possible X-ray counterpart of PSR J1734−3333, we performed a blind search for point sources using the SAS tool edetect chain. In both the 2009 and 2011 observations, a faint X-ray source was detected near the radio position of the pulsar by edetect chain in all three cameras, with the reported count rates being consistent between the two observations. Figure 3–1 shows the X-ray emission near the radio position of the pulsar, made by combining the 2009 and 2011 pn, MOS 1, and MOS 2 images into a mosaic and smoothing with a Gaussian kernel of radius σ = 300. The best-fit position of the X-ray source as reported by edetect chain for the 2009 observation is (J2000) R.A. = 17h34m27s.19 ± 0s.24, decl. = −33◦3302200.0 ± 300.0, although a slightly more precise position

39 Observation Model PSF ) 2 - Background c 2×10−5 e s c r a 1 - s

−5

s 10 t n u o c (

y

t −6 i 5×10 s n e t n I

1 10 100 Radius (arcsec)

Figure 3–2: Radial profile of PSR J1734−3333 in 2011. is reported for the 2011 observation: (J2000) R.A. = 17h34m27s.03 ± 0s.20, decl. = −33◦3302200.4 ± 200.5. Here, the reported uncertainties consist of both the statistical error and XMM-Newton’s absolute pointing uncertainty2 of 200. In principle it is possible to reduce the pointing uncertainty by matching at least two bright X-ray sources in the field to known optical counterparts; unfortunately, the field of view of this observation contained only one such source. The most up to date radio timing position for PSR J1734−3333 is R.A. = 17h34m26s.9 ± 0s.2, decl. = −33◦3302000 ± 1000 (Espinoza et al., 2011a). The 2011 X-ray position is consistent with the radio timing position, with an offset of 0s.13 or 0.46σ in R.A. and 200.4 or 0.23σ in declination (for comparison, the centroid of the 2009 position lies slightly outside the radio error ellipse, offset by 0.93σ in R.A. and 0.19σ in declination). Finally, we also produced a radial profile of the X-ray source, shown in Figure 3–2, using only the 2011 data because there were too few source

2 See http://xmm2.esac.esa.int/docs/documents/CAL-TN-0018.pdf

40 counts in the 2009 observation, and we found it to be consistent with the XMM point-spread function (PSF). Therefore, there is no evidence of extended emission. 3.2.2 Timing Analysis

To search for X-ray pulsations from PSR J1734−3333, we extracted counts in the 0.5–3 keV energy band from a region of 3000 radius centred on the source. The MOS cameras, operating with a time resolution of 2.6 s, were unsuitable for this analysis, so only the pn data were used. Within the 2009 source region, we found 150 counts, and by extracting counts in the same energy range from source-free regions on the same CCDs, we determined that 100 ± 6 of those counts were from the background. Following the same procedure for the 2011 data, we determined that of the 576 total counts in the source region, 352 ± 12 were from background photons. After barycentring the source region events with the SAS tool barycen, we folded them into eight phase bins based on a radio timing ephemeris obtained using the Jodrell Bank 76-m Lovell telescope (Espinoza et al., 2011a). Since the time span covered by the radio ephemeris included the epochs of both the 2009 and 2011 observations, we could produce a single summed light curve from both sets of data. Fitting the folded curve to a straight line gave a best-fit χ2 of 11.1 for 7 degrees of freedom, or a 13% chance that the folded curve could result with no signal in the data. Therefore, no significant pulsations were detected. We also performed additional searches, attempting to improve the signal- to-noise ratio by using an energy range of 1–2 keV or a source region of 2000 radius, as well as searching around the predicted period, but in all cases there were again no detections of pulsations.

41 To find an upper limit for the pulsed fraction, we simulated event lists with the same number of total counts as found in the source region. The simulated signal had a sinusoidal profile with a random phase and had a user-specified area pulsed fraction, defined as the fraction of the pulse profile that lies above the off-pulse minimum. We found that a signal with an area pulsed fraction of ∼0.23 would be detected with >3σ significance 68% of the time. Therefore, because 62% of the counts in the source region are from the background, we estimate that the 1σ upper limit on the area pulsed fraction of PSR J1734−3333 is 0.6 in the 0.5–3 keV energy range. 3.2.3 Spectral Analysis

We extracted the spectrum of PSR J1734−3333 from the 2009 pn and 2011 pn and MOS data using a region of 3000 radius centred on the X-ray source position. For the 2009 data, a background spectrum was taken from an elliptical annulus surrounding but excluding the source region, while for the 2011 data, different background regions were chosen, away from the source region and any other sources but still on the same CCD. Response and ancillary response files were generated with the SAS tasks rmfgen and arfgen. The 2009 spectrum was grouped to have a minimum of 20 counts per bin and the 2011 spectra, with their greater amount of source counts, were grouped to have a minimum of 25 counts per bin. We did not extract spectra from the 2009 MOS images due to the low number of source counts in those cameras. These four spectra were then jointly fit in XSPEC 12.7.0 using simple power-law and blackbody models, as well as with nsa (Pavlov et al., 1995) and nsmax (Ho et al., 2008) neutron-star atmosphere models, all using phabs to model interstellar absorption. The absorbed power-law model provided a statistically acceptable fit to the data (χ2 = 34.6 for 44 degrees

42 Table 3–2: Spectral Models for PSR J1734−3333

Parameter Blackbody nsaa nsmaxb 22 −2 +0.35 +0.41 +0.40 NH (10 cm ) 0.67−0.25 0.83−0.29 0.88−0.28 ∞ +0.05 kT (keV) 0.30 ± 0.06 0.14 ± 0.05 0.13−0.04 ∞ +0.55 +8.9 R (km) 0.45−0.20 13 (fixed) 3.2−2.0 c +50 Distance (kpc) ··· 27−20 ··· χ2 (dof) 34.7(44) 34.6(44) 34.3(44) −15 −1 −2 d +1.1 +1.2 fabs (10 erg s cm ) 9.9 ± 1.2 9.9−1.2 9.8−1.1 −14 −1 −2 d +3.7 +9.9 +11.6 funabs (10 erg s cm ) 3.6−1.4 5.6−2.7 6.6−3.4 a The nsa model for a pulsar with B = 1013 G. b The nsmax model for a pulsar with B = 2 × 1013 G. c The distance to the pulsar is fit for in the nsa model, while in the other two models the DM distance of 6.1kpc is used to determine the radius of emission. d Absorbed and unabsorbed fluxes are given in the 0.5–2.0 keV band. of freedom), but as it gave an unphysically steep photon index of 5.2, we do not consider it any further. Best-fit parameters for the other three models are shown in Table 3–2. The best-fit blackbody model is plotted in Figure 3–3. We found a best-fit3 temperature of kT = 0.30 ± 0.06 keV, corresponding to a

+0.36 22 −2 column density of NH = 0.67−0.24 × 10 cm . The blackbody radius is

+0.55 +2.2 Rbb = 0.45−0.20 d6.1 km, giving a bolometric luminosity of Lbb = 2.0−0.7 ×

32 2 −1 10 d6.1 erg s , where d6.1 is the distance to the pulsar in units of 6.1 kpc. In order to better explore the confidence range of kT and NH for the blackbody model, we plotted their confidence contours in Figure 3–4. The plot shows that the 3σ lower limit on kT is 0.18 keV and that lower values of kT require higher values of NH. We note that the maximum column density along the

4 22 −2 line of sight to the pulsar is NH = 1.1–1.4 × 10 cm , and because the

3 All errors in this section are 90% confidence intervals.

4 http://heasarc.nasa.gov/cgi-bin/Tools/w3nh/w3nh.pl

43 1 - 2011 pn V

e 2011 MOS1

k −3 6×10 2011 MOS2 1 -

s 2009 pn

s t n

u −3

o 4×10 c

d e z i l −3

a 2×10 m r o n 0 2×10−3 s l a u d

i 0 s e r −2×10−3 0.5 1 2 Energy (keV)

Figure 3–3: XMM-Newton spectrum of PSR J1734−3333 with the best- fit blackbody model. The bottom panel shows the residuals in units of counts s−1 keV−1. Error bars are at the 1σ confidence level.

DM distance places the pulsar less than halfway through the Galaxy, the expected column density is significantly lower than that. As a result, models

22 −2 with kT < 0.2 keV which require NH > 1 × 10 cm are further disfavoured. Both of the neutron-star atmosphere models provide good fits to the data and yield much lower best-fit temperatures than the blackbody model, as is typical. The nsa model gives kT ∞ = 0.14 ± 0.05 keV, but because it assumes that the emission is from the entire surface of the star, it implies an unreasonably large distance to the pulsar of ∼27 kpc. Fixing the distance at 6.1 kpc results in an even lower temperature of 90 ± 3 eV (χ2 = 37.6 for 45 degrees of freedom), but the required column density

+0.2 22 −2 rises to NH = 1.2−0.1 × 10 cm , which is not favoured as described above. Given these difficulties, we also fit the spectrum to the nsmax model for a highly magnetized neutron-star atmosphere. This model gives

44 0.45

0.40 LAB DL

0.35 ) V e k ( 0.30 T k

0.25

0.20

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 22 -2 NH (10 cm )

Figure 3–4: 1σ, 2σ, and 3σ confidence contours for kT and NH from fitting an absorbed blackbody model to the X-ray spectrum of PSR J1734−3333. The cross marks the best-fit value, and the dashed lines mark the total NH along the line of sight to the pulsar according to the Leiden/Argentine/Bonn (labeled LAB) and Dickey & Lockman (labeled DL) Galactic H I surveys.

45 ∞ +0.05 +0.40 22 −2 kT = 0.13−0.04 keV, NH = 0.88−0.28 × 10 cm , and an emission radius

+8.9 of R = 3.2−2.0 d6.1 km. Finally, we note that the nsa and nsmax models were fitted using the highest available values for the magnetic field (in XSPEC), 1013 and 2 × 1013 G, respectively. Since these values are below the 5.2 × 1013 G field of PSR J1734−3333, the results are not completely reliable. 3.3 Discussion

3.3.1 Associating the X-ray Source with the Radio Pulsar

Although the location of the detected X-ray source is consistent with the radio timing position of PSR J1734−3333, given that no X-ray pulsations were detected from it at the radio period to provide unam- biguous proof of association, it is reasonable to question whether the X-ray source really is associated with the radio pulsar. We can estimate the probability of a chance superposition from the log N–log S curves for XMM-Newton Galactic-plane sources in the 0.5–2.0 keV band (Motch, 2006). Given the lower limit for the source’s absorbed flux in that band, 8.7 × 10−15 erg s−1 cm−2, the log N–log S curves predict ∼110 sources per square degree at this flux or higher. The probability of a random X-ray source lying within the radio error ellipse is then only 0.05–0.1%. On the other hand, looking at an optical image of the field of the XMM-Newton observation, there are several sources near the X-ray source. In fact, the closest of these optical sources (NOMAD Catalog ID 0564- 0621454) lies within the 2009 X-ray error circle at coordinates R.A. = 17h34m27s.319 ± 0s.008, decl. = −33◦3302200.61 ± 000.05 (See Figure 3–1), although it is outside the 2011 error circle. We shall nevertheless consider the possibility that the X-ray source is associated with this optical source in addition to or instead of the radio pulsar.

46 Assuming that the X-ray source is associated with 0564-0621454, its X-ray to optical flux ratio can be estimated using the following formula from Georgakakis et al.(2004):

log (fX/fopt) = log f (0.5–8 keV) + 0.4B + 4.89.

Here f(0.5–8 keV) is the unabsorbed X-ray flux in that energy band and B = 18.79 is the optical magnitude of 0564-0621454 taken from the USNO-B1.0 catalog. Although the unabsorbed X-ray flux is not very well constrained, we can calculate a rough upper limit of log (fX/fopt) < 0. PSR J1734−3333 is an isolated neutron star, however, which typically has log (fX/fopt) ∼ 5 (Treves et al., 2000; Kaspi et al., 2006). Therefore, we conclude that if our X-ray source is associated with 0564-0621454, it cannot also be associated with PSR J1734−3333, and vice versa. Finally, just as the probability of a chance superposition of an X- ray source with the pulsar’s radio position was estimated above, we can calculate the probability of an optical source coinciding with the X-ray position. A query of the USNO-B1.0 catalog returns almost 5500 sources within a 100 radius of the centre of the XMM-Newton field. The probability of one of these sources lying within 3.600 of the 2011 X-ray position is ∼18%, over two orders of magnitude higher than the probability of a chance X-ray/radio alignment. Even considering only the 2009 X-ray position, from which 0564-0621454 is merely 1.700 away, the probability of a chance superposition is still 4%–5%. We conclude that it is most likely that the X-ray source is associated with PSR J1734−3333, and highly improbable that it is associated with the optical source 0564-0621454.

47 3.3.2 The Temperature of PSR J1734–3333

PSR J1734−3333 has a thermal spectrum, well fit by a blackbody model with an unusually high temperature, kT = 0.30 ± 0.06 keV, and corre-

+0.55 sponding blackbody radius of 0.45−0.20 d6.1 km. The bolometric luminosity is 32 2 −1 ˙ 2 Lbb = 2.0 × 10 d6.1 erg s , giving an X-ray efficiency of Lbb/E = 0.0036d6.1. The pulsar’s small emission radius is suggestive of thermal emission from heated polar caps, but such models predict X-ray efficiencies less than 10−3. For example, the polar cap reheating model of Harding & Muslimov(2001)

˙ −4 predicts Lbb/E ≈ 3 × 10 for PSR J1734−3333, an order of magnitude below what we observe. Of course, if the DM distance is incorrect and the true distance to the pulsar is within 2 kpc (d6.1 . 0.3), then the X-ray effi- ciency drops to roughly what is predicted for polar cap reheating. However, Harding & Muslimov(2001) warn that their model may not apply to pulsars

12 with B & 4 × 10 G, so these predictions are unreliable. The pulsar’s blackbody temperature of 0.30 keV is almost three times as high as expected from a minimal cooling model (0.07–0.11 keV; Page et al., 2006) given its age. Even the neutron-star atmosphere models give best-fit temperatures that are too high, kT ∞ = 0.13–0.14 keV, though their uncertainties do overlap with the range of predicted cooling temperatures. For instance, the nsa model with distance fixed at 6.1 kpc gives kT ∞ = 0.09 keV. However, note that as with the simple blackbody, lower temperatures are correlated with higher NH in the atmosphere models too, which can disfavour that region of parameter space as explained in Section 3.2.3. Additionally, as mentioned above, the atmosphere models assume magnetic fields below that of the pulsar; indeed, given the high magnetic field of PSR J1734−3333, the pulsar may not be able to support a

48 0.7 Magnetars XINSs High-B PSRs 0.6 PSRs

0.5 ) V

e 0.4 k (

T k 0.3 J1734−3333

0.2

0.1

0 1000 104 105 106 107 Age (yr)

Figure 3–5: Blackbody temperatures vs. characteristic ages for normal pul- sars (blue triangles), high-B pulsars (B ≥ 1013 G; red diamonds), XINSs (red squares), and magnetars (yellow stars). Data for the magnetars are taken from Chapter5 and references therein. The data for the other sources are taken from Zhu et al.(2011) with the addition of PSRs B1845 −19 and J1001−5939 (Olausen et al., 2013), PSR J0726−2612 (Speagle et al., 2011), an updated temperature for PSRs J1734−3333 (this work) and J1119−6127 (Ng et al., 2012), and an updated timing solution for RX J0420.0−5022 (Kaplan & van Kerkwijk, 2011). hydrogen atmosphere, as has been argued to be the case for the magnetars (Chang et al., 2004). Although our ignorance regarding the correct atmosphere model for PSR J1734−3333 precludes meaningful comparisons of its X-ray emission with models of neutron-star cooling, we can nevertheless compare our results with those of other pulsars of comparable age. We do this in Figure 3–5, where we plot blackbody temperature versus characteristic age for rotation- powered pulsars, XINSs, and magnetars, with high-B (≥ 1013 G) sources

49 shown in red and yellow (with the exception that the low-field magnetar, SGR 0418+5729, see Chapter5, is also shown in yellow). Although a blackbody model neglects atmospheric effects and hence is almost certainly incorrect for these sources, it provides a simple spectral parameterization by which to describe all source spectra and allows a consistent comparison of spectral properties, even if ultimately not fully physical. Figure 3–5 shows that PSR J1734−3333 has a much higher blackbody temperature than nearly all other radio pulsars for which this quantity has been measured, with the possible exception of PSR J1119−6127, another high-B radio pulsar (Gonzalez et al., 2005; Ng et al., 2012). It is even hotter than the magnetars XTE J1810−197 and Swift J1822.3−1606, although its temperature is well below those of the other magnetars. In particular, however, it is far hotter than multiple other rotation-powered pulsars of comparable age but smaller magnetic field. Indeed, as noted by Zhu et al. (2011), high-B pulsars in general show a trend toward higher blackbody temperatures when compared with lower-B pulsars of similar age. This suggests that the magnetic field affects the observed thermal properties of pulsars, as is expected if the star is actively heated by field decay. However, enhanced thermal emission could also be a result of passive effects such as magnetically altered thermal conductivity (Geppert et al., 2006; Page et al., 2007) without any active field decay. Distinguishing between active and passive explanations for the enhanced thermal emission in high-B pulsars could be achieved via detailed modeling, or perhaps by observing other signatures of active field decay, such as magnetar-like bursts in high-B rotation-powered pulsars. Indeed, this has been observed in one source already (Gavriil et al., 2008).

50 Table 3–3: High-magnetic-field Radio Pulsars ˙ a b Name PB E τ d LX Energy Range Reference (s) (1013 G) (erg s−1) (kyr) (kpc) (erg s−1) (keV) J1847−0130 6.71 9.4 1.7 × 1032 83 8.4 <5 × 1033 2–10 McLaughlin et al.(2003) J1718−3718 3.38 7.4 1.6 × 1033 34 4.5 2–9 × 1032 Bolometric Zhu et al.(2011) J1814−1744 3.98 5.5 4.7 × 1032 85 10 <6.3 × 1035 0.1–2.4 Pivovaroff et al.(2000) <4.3 × 1033 2–10 J1734−3333 1.17 5.2 5.6 × 1034 8.1 6.1 1.0–3.3 × 1032 0.5–2 This work J1819−1458c 4.26 5.0 2.9 × 1032 117 3.6 2.8–4.3 × 1033 0.3–5 McLaughlin et al.(2007) J1846−0258 0.33 4.9 8.1 × 1036 0.9 6.0d 2.5–2.8 × 1034f 0.5–10 Ng et al.(2008) 1.2–1.7 × 1035g 0.5–10 Ng et al.(2008) J1119−6127 0.41 4.1 2.3 × 1036 1.7 8.4e 1.3–3.6 × 1033 0.5–8 Ng et al.(2012) J0726−2612 3.44 3.2 2.8 × 1032 186 3.0 ∼1.4 × 1033 0.32–1.1 Speagle et al.(2011) 51 J0847−4316c 5.98 2.7 2.2 × 1031 790 3.4 <1 × 1032 0.3–8 Kaplan et al.(2009c) J1846−0257c 4.48 2.7 7.1 × 1031 442 5.2 <3 × 1032 0.3–8 Kaplan et al.(2009c) J1001−5939 7.73 2.2 5.1 × 1030 2100 2.7 <1.3 × 1032 Bolometric Olausen et al.(2013) B0154+61 2.35 2.1 5.7 × 1032 197 1.7 <8 × 1031 0.3–10 Gonzalez et al.(2004) B1916+14 1.18 1.6 5.1 × 1033 88 2.1 ∼3 × 1031 Bolometric Zhu et al.(2009) B1845−19 4.31 1.0 1.1 × 1031 2900 0.75 <6.8 × 1031 Bolometric Olausen et al.(2013)

a Unless otherwise noted, all distances were estimated from the dispersion measure of the source. b The ranges for PSRs J1718−3718 and J1819−1458 indicate 68% confidence intervals, while the ranges for PSRs J1734−3333, J1846−0258, and J1119−6127 indicate 90% confidence intervals. c Pulsar classified as a rotating radio transient (RRAT). d The distance to PSR J1846−0258 was found from H I and 13CO spectral measurements (Leahy & Tian, 2008a). e The distance to PSR J1119−6127 was found from H I absorption measurements (Caswell et al., 2004). f This value is from 2000, prior to this pulsar’s 2006 outburst. g This value is from 2006, during this pulsar’s 2006 outburst. 3.3.3 X-ray Luminosities of High-B Pulsars

In Table 3–3, we list all high-magnetic-field radio pulsars with inferred magnetic field B > BQED, as well as all other radio pulsars with B > 1.5 × 1013 G, that have measured thermal X-ray luminosities or luminosity upper limits. Of these stars only one, PSR J1846−0258, has exhibited clear magnetar-like behaviour (Gavriil et al., 2008). Other than that, the high-B pulsars show X-ray spectral properties that are very different from those of active magnetars, being much softer and fainter X-ray sources. They do, however, show some similarities with transient magnetars (defined as magnetars whose quiescent flux is multiple orders of magnitude lower than their flux in outburst), which in quiescence are also softer, fainter X-ray sources. One transient magnetar in particular, XTE J1810−197, displays spectral properties in quiescence that are consistent with those of the

33 −1 detected high-B pulsars (kT ≈ 0.18 keV and 0.5–10 keV LX ∼ 10 erg s ) (Gotthelf et al., 2004). Additionally, we took the reported X-ray luminosities for the pulsars in Table 3–3, converted them to the 0.5–10 keV energy band, and plotted this value versus inferred magnetic field in Figure 3–6. The resulting plot shows possible evidence of a correlation between the two quantities, with higher

B-field stars having higher 0.5–10 keV LX; certainly the pulsars here with

13 13 B & 4 × 10 G are brighter in X-rays than those with B . 4 × 10 G. There are important caveats that should be noted, however. In particular, the four youngest pulsars in Table 3–3 also comprise four out of the five brightest X- ray detected pulsars in the Table, suggesting that the perceived correlation may be due entirely or in part to age instead of B. On the other hand, one of the brightest of these pulsars, PSR J1819−1458, has a considerably higher characteristic age; it is higher even than that of the much dimmer

52 1035 J1814−1744 J1847−0130 J1846−0258

1034

) J1119−6127 J1819−1458 V e k

0 33 1 10 – J1718−3718 5 . 0 (

) J1734−3333 1

- J1846−0257 s

g 32

r 10 J0847−4316 e (

X J0726−2612 L B1916+14 B0154+61 1031

J1001−5939

1030 1013 4×1013 1014 B (G)

Figure 3–6: X-ray luminosity (0.5–10 keV) vs. inferred surface dipolar mag- netic field strength for X-ray observed high-B radio pulsars. Luminosity val- ues from the original references were extrapolated to this energy band based on the fit or assumed spectral model. Error bars are at the 1σ confidence level and do not include uncertainties in the distance measurements. Note that the value plotted for PSR J1846−0258 is its quiescent X-ray luminosity. 13 The dashed line represents the quantum critical field, BQED = 4.4 × 10 G. See Table 3–3 for references.

53 PSR B1916+14. Additionally, of the four pulsars with the highest B fields, two have only X-ray luminosity upper limits. If future observations were to show one or both of these pulsars to be substantially more X-ray dim than PSR J1819−1458 or PSR J1718−3718, that would considerably weaken the possible correlation. 3.4 Conclusions

In summary, we have presented results from two XMM-Newton obser- vations of the young, high-B radio pulsar PSR J1734−3333. We successfully detected the X-ray counterpart at a position that lies within the error ellipse of the pulsar’s radio timing position. We were unable to detect X-ray pul- sations, although the 1σ upper limit of 60% (0.5–3 keV) that we set on the pulsed fraction is not very constraining. We also find that it has a thermal spectrum with kT = 300 ± 60 eV, significantly higher than predicted by cooling models, though atmosphere models (nsa and nsmax) yield lower temperatures that are consistent with standard cooling. Comparing PSR J1734−3333 with other rotation-powered pulsars, we find that it has a significantly higher blackbody temperature than nearly every other radio pulsar for which such a measurement has been made, particularly other pulsars of similar age but lower magnetic field. This result suggests that a high magnetic field can affect the observed thermal properties of pulsars, for example by heating of the surface due to magnetic-field decay or by passive atmospheric effects.

54 CHAPTER 4 The Extended Emission Around the Magnetar 1E 1547.0–5408 The contents of this chapter are reported in the paper “On the Ex- tended Emission around the Anomalous X-Ray Pulsar 1E 1547.0−5408” published in the Astrophysical Journal (Olausen et al., 2011). 4.1 Introduction

One important open question regarding magnetars is whether they pro- duce particle outflows akin to those seen in conventional rotation-powered pulsars. The latter are well known to produce often spectacular “pulsar wind nebulae” (PWNe; see Section 1.1.3), the classic example of which is the Crab Nebula. Such nebulae, evident particularly at radio and X-ray energies, are the result of synchrotron emission due to pulsar-produced rela- tivistic electrons and positrons, as they spiral in the ambient magnetic field. Although the low spin-down luminosities typical to magnetars (see Chap- ter5, in particular Table 5–2) mean that they are not expected to harbour conventional, rotation-powered PWNe, particle outflows still seem reason- able to consider given the large magnetic energy reservoir hypothesized to exist in them. Indeed, they have been suggested to be present ubiquitously in magnetars, either in continuous or sporadic forms (Thompson & Blaes, 1998; Harding et al., 1999). This idea was initially buoyed by the claim of an apparent wind nebula associated with SGR 1806−20 (Murakami et al., 1994). Although the latter association was later disproved (Hurley et al., 1999), the possibility of a nebula-producing magnetar wind has not been. Extended radio emission was unambiguously identified following one SGR giant flare (Gaensler et al., 2005b; Taylor et al., 2005; Gelfand et al.,

55 2005; Fender et al., 2006), but is thought to be from relativistic, weakly baryon-loaded magnetic clouds (Lyutikov, 2006) or from a baryonic outflow (Gelfand et al., 2005; Gelfand, 2007), and associated exclusively with the flare. Recently Rea et al.(2009a) have suggested that an unusual X-ray nebula surrounding the relatively high-B Rotating Radio Transient (RRAT) J1819−1458 is magnetically powered, though the mechanism for this is unclear. The X-ray source 1E 1547.0−5408 was discovered with the Einstein X-ray satellite in 1980 by Lamb & Markert(1981), but only recently was it suggested to be a magnetar candidate on the basis of its spectrum, variable nature, and likely association with the supernova remnant (SNR) shell G327.24−0.13 (Gelfand & Gaensler, 2007). Radio observations of the source by Camilo et al.(2007a) revealed pulsations at a period of 2.1 s and the measured spin-down rate implied a surface magnetic field strength of B = 3.2 × 1014 G and a spin-down luminosity of E˙ = 2.1 × 1035 erg s−1. 1E 1547.0−5408 is thus the fastest-rotating and highest-E˙ magnetar yet known. Although observations in 2006 showed the source to be in quiescence, a 2007 XMM-Newton observation showed it to be in a high, apparently post- outburst state (Halpern et al., 2008). In 2008 October and again on 2009 January 29, 1E 1547.0−5408 underwent strong outburst events, experiencing dramatic increases in its X-ray luminosity and exhibiting many SGR-like bursts within a few hours. For further details on these outbursts and the source’s history, see Israel et al.(2010), Kaneko et al.(2010), Ng et al. (2011), Bernardini et al.(2011), Safi-Harb & Kumar(2008), and Dib et al. (2012).

56 In observations of 1E 1547.0−5408 taken with Swift and XMM-Newton following the 2009 outburst, Tiengo et al.(2010) observed dust-scattering X- ray rings centred on the magnetar and derived from them a source distance of ∼4–5 kpc. They also found evidence of time variable diffuse emission around the source which they attributed to dust scattering of the bursts and of the persistent X-ray emission. Meanwhile, Vink & Bamba, 2009, hereafter VB09, analyzing Chandra and XMM-Newton observations of the source taken in 2006 when it was in quiescence, detected extended emission and characterised it as the result of a PWN, in analogy with those seen around rotation-powered pulsars. They argued for a PWN based on the high flux level of the extended emission, and because it appeared to have a harder spectrum than the point source. This would make 1E 1547.0−5408 unusual among the known magnetars, as no other such source has been shown to power such emission. They also showed the presence of extended X-ray emission coincident with the SNR shell. Here, we present an analysis of the extended emission around 1E 1547.0−5408 using multi-epoch XMM data, in which the source flux varies strongly, as does the putative nebular emission. We show conclusively that the putative PWN reported by VB09 is in fact dominated by dust scattering, rather than by emission from any pulsar outflow. 4.2 Observations

We obtained an observation of 1E 1547.0−5408 with the XMM-Newton Observatory (Section 2.4) on 2010 February 10 in order to track the star’s properties as it decayed to quiescence after the 2009 outburst. We also reanalyzed three archival XMM-Newton observations of the source: one in 2006 with the source in quiescence, the 2007 post-outburst observation,

57 Table 4–1: Summary of XMM-Newton Observations of 1E 1547.0−5408

Date ObsID Exposure Count Ratea Mode/Filterb (ks) (cnt s−1) 2006 Aug 21 0402910101 38.7 0.074 FF/medium 2007 Aug 9 0410581901 11.6 0.59 LW/medium 2009 Feb 3 0560181101 48.9 4.6 FF/thick 2010 Feb 10 0604880101 35.7 1.9 LW/medium

a Background-subtracted count rate of the point source in the 1–6 keV energy range. b The time resolutions of the operating modes of the EPIC pn camera are full frame (FF): 73.4 ms and large window (LW): 47.7 ms. and one taken two weeks after the 2009 outburst. For each observation, the data from the two EPIC MOS cameras were not suitable for our analysis, either because the operating mode provided too small a field of view or because the data were highly piled-up. We therefore restricted our analysis to data from the EPIC pn camera, which had no such issues. The data from all four observations were analyzed using the XMM-Newton Science Analysis System (SAS) version 10.0.21 with calibrations updated 2010 July 29. Each observation was filtered for times of strong background flaring that sometimes occur in XMM-Newton data, and two bursts were removed from the 2009 data. Details of each of the four observations, including the total pn exposure time after removing the bad time intervals, are listed in Table 4–1. 4.3 Imaging Analysis and Results

In order to search for and characterise any extended emission around 1E 1547.0−5408, we began by removing all point sources detected in the

1 See http://xmm.esac.esa.int/sas/

58 field of each observation other than the magnetar. We also removed the out- of-time events that were present in the 2009 observation by subtracting an image of simulated out-of-time events generated by the SAS task epchain. To construct a radial profile, we extracted events from concentric annuli having width 200 centred on the position of the star, as determined by a standard centroid search algorithm. The number of counts in each bin was divided by both the geometric area of each extraction region and the mean exposure time therein as determined from the unvignetted exposure map. This procedure corrects for the chip gaps, dead pixels, and removed regions on the detector. The radial profile of a point source in XMM-Newton is given by the energy-dependent, radially averaged point-spread function (PSF). Using the SAS task eradial we extracted from instrument calibration files the theoretical PSF at 1 keV energy intervals from 1 to 12 keV for each observation. These component PSFs were weighted such that the PSF at

N10(|E−Ei|<0.5 keV) energy Ei was given the weight W (Ei) = , where N10(E) N10(E) is the total number of counts within 1000 of the source position. They were then summed to produce a weighted PSF, S(r). Finally, by scaling S(r) using the formula P (r) = a · S(r) + b, we created an expected point-source radial profile, P (r), for each obser- vation. Here, b is a spatially uniform background count rate found by averaging the count rate in all the bins of the observed profile a sufficient distance away from the source. The normalization factor a was derived via a least-squares fit of P (r) to the first five bins (1000) of the observed radial profile, under the assumption that the contribution of any extended emission to the profile in that region would be minimal. We found, however, that

59 10−3 10−2 Observation Observation Model PSF Model PSF ) ) 2 2 - Background - Background c c e e s s

c −3 c r 10 r −4

a a 10 1 1 - - s s s s t t

n −4 n

u 10 u o o

c c −5 ( (

10 y y t t i i s s

n 10−5 n e e t t n n

I I 6–12 keV 1–6 keV 10−6 10−6 1 10 100 1 10 100 Radius (arcsec) Radius (arcsec)

Figure 4–1: Left: 2010 February radial profile of 1E 1547.0−5408 in the 1–6 keV energy band. Right: same but for 6–12 keV. the least-squares fit tended to be poor, especially for the observations with higher count rate and thus better statistics. This suggested that the uncer- tainty on a based on the χ2 was not reliable because of additional systematic errors affecting the fit. For example, in addition to possible contamination by extended emission, the 200 size chosen for the radial bins oversamples the 400.1 square pixels of the pn detector. Therefore, in order to better estimate the uncertainty, we obtained a range of possible values for a by making a least-squares fit of P (r) to any four of the first five and any five of the first six bins of the observed profile. Our uncertainty estimate was then given by

δa = (amax − amin)/2. The 2010 radial profiles of 1E 1547.0−5408 in the 1–6 keV and 6– 12 keV energy bands are shown in Figure 4–1. Below 6 keV the observed profile has a significant excess of counts over the expected point-source profile, extending out to r ≈ 4.05; conversely, above 6 keV no excess is detected and the observed profile is consistent with a point source. Radial profiles constructed from the three archival data sets display similar results, although the shape and extent of the excess vary among the observations,

60 B n

o B i 0.04 s 1 s i 2006

m 2009

E A

d 0.02 e d

n A 2010 e t x 0 E

0 0.04 0.08 f 0.1 o

2007 e t a R

t n

u 2006 o C 0.01

0.1 1 Total Count Rate of Point Source

Figure 4–2: 1–6 keV count rate of the extended emission integrated over annular regions with 2000 < r < 4000 (lower set of points, labeled A) and 4000 < r < 15000 (upper set of points, labeled B) vs. the total background- subtracted 1–6 keV count rate of the point source for each of the four XMM- Newton observations of 1E 1547.0−5408. The dotted lines are a linear fit to the leftmost three points. Inset: blow-up of the region near the origin covering the 2006 data points. and the three dust-scattering rings reported in Tiengo et al.(2010) are visible in the 2009 data. We therefore confirm the presence of extended emission around 1E 1547.0−5408 below 6 keV as previously reported by VB09 and Tiengo et al.(2010). We also find that the extended emission is brighter at low energies (<3 keV), in line with the soft spectrum reported by VB09. In Figure 4–2, we plot the 1–6 keV count rate of the extended emission,

Iext, as a function of the total background-subtracted point-source count rate, Ips, for two regions: region A, an annulus centred on the source with radius 2000 < r < 4000; and region B, a similar annulus but with radius

61 0.14 A B 0.8 0.12

0.1 0.6

0.08 c a r f I 0.06 0.4

0.04 0.2 0.02

0 0 2006 2007 2009 2010 2006 2007 2009 2010 Year Year

Figure 4–3: Fractional intensity of the extended emission, Ifrac = Iext/Ips, in the 1–6 keV energy band for regions A and B of all four XMM-Newton observations of 1E 1547.0−5408.

4000 < r < 15000. In both regions we find a tight correlation between the two quantities; in fact, the first three points (2006, 2007, and 2010) fit well to a straight line, although the fourth point (2009) lies above the extrapolated linear fit in both regions A and B (but see Section 4.4.2). The extended emission flux varies wildly between observations, increasing and decreasing along with the flux of the pulsar. For example, in region A the extended emission brightened by a factor of nearly 50 between the 2006 and 2009 observations, and by the following year it had faded by almost a factor of three. Figure 4–3 shows the fractional intensity of the extended emis- sion, Ifrac = Iext/Ips in regions A and B for all four observations of 1E

1547.0−5408. The most prominent feature in both regions is that Ifrac in 2006 is notably higher than in the other three observations, particularly in region B.

62 Table 4–2: Hardness Ratios for 1E 1547.0−5408 and the Surrounding Ex- tended Emission

Observation Hardness Ratioa Point Source Region A Region B 2006 0.306(18) 0.21(6) 0.46(5) 2007 0.482(16) 0.18(5) 0.32(4) 2009 1.014(6) 0.55(4) 0.45(1) 2010 0.814(8) 0.34(4) 0.37(2)

a The hardness ratio is defined as I (3–6 keV) /I (1–3 keV), where I is the background-subtracted count rate of the point source or extended emission. Numbers in parenthe- ses are 1σ uncertainties.

4.3.1 Spectral Analysis

Spectral analysis of the extended emission in regions A and B was com- plicated by contamination from the broad wings of the XMM-Newton PSF. For example, in each observation, less than half of the total background- subtracted counts found in region A were contributed by the extended source. In principle, extracted spectra could be fit to a three-component model, with power law and blackbody components for the point source, and another power law component for the extended emission. However, the statistics in the 2006 and 2007 observations were too poor to reliably fit so many parameters. Therefore, we instead computed a simple hardness ratio for the extended emission, HRext ≡ Iext (3–6 keV) /Iext (1–3 keV), and we list in Table 4–2, for all four observations, these hardness ratios in regions A and B. For comparison, the table also lists the hardness ratio of the point source, HRps ≡ Itot (3–6 keV) /Itot (1–3 keV), where Itot is the background-subtracted count rate within 1000 of the source position.

We find that in region A, HRext < HRps for all four observations, meaning that the extended emission has a softer spectrum than the point source, although in 2006 the hardness ratio is smaller by only 1.5σ. In

63 A B 0.8 2006 o i

t 0.6 a R

s s e n d r

a 0.4 H

0.2

0 20 40 60 80 100 120 140 160 Radius (arcsec)

Figure 4–4: Hardness ratio, HR ≡ Iext (3–6 keV) /Iext (1–3 keV), for the ex- tended emission of 1E 1547.0−5408 in 2006. The first bin gives the hardness ratio of the point source, whereas all subsequent bins are for the extended emission only. Regions A and B are labeled. region B, the results are the same as above for all but the 2006 observation, where the extended emission spectrum is instead harder than the source spectrum. This behaviour is further illustrated in Figure 4–4, which shows the 2006 hardness ratio of the extended emission as a function of distance from the source, beginning at 2000 (the first bin provides the hardness ratio of the point source itself). Finally, we used the WEBPIMMS tool2 to estimate the photon index, Γ, and flux of the extended emission, assuming an absorbed power-law spectrum, based on the count rates in Figure 4–2 and the hardness ratios

22 −2 in Table 4–2. In particular, taking NH = 2.75 × 10 cm as in VB09, we find that the 2–10 keV unabsorbed flux of the extended emission in region A

2 http://heasarc.gsfc.nasa.gov/Tools/w3pimms.html

64 varied from a minimum of ∼4 × 10−14 erg s−1 cm−2 (for Γ ≈ 4) in 2006 to a maximum of ∼3 × 10−12 erg s−1 cm−2 (for Γ ≈ 3) in 2009. 4.4 Discussion

Our analysis has confirmed the presence of extended emission around 1E 1547.0−5408, visible in four observations taken at very different stages of its flux history. VB09 previously detected extended emission around this source based primarily on a 2006 Chandra observation and interpreted it as a PWN and, farther out, the X-ray counterpart of SNR G327.24−0.13. Here we re-examine this interpretation in light of our new data. 4.4.1 A Pulsar Wind Nebula?

In the context of conventional rotation-powered pulsars, given the low spin-down power of 1E 1547.0−5408, we generally would not expect it to harbor a bright PWN in X-rays. For E˙ = 1035 erg s−1, the typical X-ray effi- ciency of a PWN is about 10−4 (Kargaltsev & Pavlov, 2008). Even allowing for the X-ray efficiency to be up to an order of magnitude greater, this pre-

−14 −2 −1 −2 dicts an unabsorbed PWN flux of .5 × 10 (d/4 kpc) erg s cm in the 0.5–8 keV range for 1E 1547.0−5408. On the other hand, the putative PWN as suggested by VB09 has an unabsorbed flux of ∼1 × 10−12 erg s−1 cm−2 in the same band. Thus, if the extended emission is entirely rotation-powered, this would require an unusually high X-ray efficiency of over 1%. This problem could be alleviated, however, by hypothesizing that magnetic power could be contributing to the nebula, as has been suggested by Rea et al. (2009a) for RRAT 1819−1458. However, another issue is the soft spectrum of the putative PWN. The hardness ratios in Section 4.3.1 suggest a power-law spectrum with photon index Γ ≈ 3–4, a range that is consistent with the value reported in VB09 (Γ = 3.4 ± 0.4). This is much softer than previously reported PWNe, which

65 typically have Γ ≈ 1.5–2 (Kargaltsev & Pavlov, 2008), although VB09 proposed that the discrepancy could be explained as being somehow a result of the magnetar nature of 1E 1547.0−5408. A more pressing problem with the PWN interpretation is its failure to explain the strong flux correlation seen in Figure 4–2. First, PWNe in rotation-powered pulsars have not been seen to have large luminosity variations as are observed for 1E 1547.0−5408. Even if energy injection due to outbursts played a role, we show here that the observed fading time is incompatible with a synchrotron origin. First, to estimate the magnetic field strength in the putative PWN, we consider as an analogy the 2004 flare of SGR 1806−20, which was accompanied by nebular radio emission. In that event, a total energy of 2 × 1046 erg was released (Palmer et al., 2005). Lyutikov(2006) proposed that the event could produce relativistic, weakly baryon-loaded magnetic clouds analogous to a solar coronal mass ejection, and deduced a total energy of 8 × 1044 erg for the relativistic electrons plus magnetic field, with an average magnetic field B of 0.1 G within a radius ∼1.5 × 1016 cm. As the nebula expands, the field strength decays with the volume V as V −1/2, or V −2/3 if the field is tangled (Gaensler et al., 2005b). Assuming the former case, we have scaled these values to those appropriate for the properties of 1E 1547.0−5408 and find a conservative B-field estimate of <80 µG at 3000, which corresponds to 1.8 × 1018 cm from the source. We note that the B-field is much lower for the latter field decay case. The synchrotron cooling timescale is then

−3/2 1/2 τsyn = 37(B/1 µG) (εγ/1 keV) kyr & 120 yr for particles emitting at

εγ = 6 keV. This is incompatible with the flux decay timescale observed and shown in Figure 4–2, in particular between 2009 and 2010.

66 The above theory suggests that the particle energy was about 4% of the burst fluence for SGR 1806−20. The total burst fluence for the 2009

43 2 event of 1E 1547.0−5408 was measured to be &5 × 10 (d/10 kpc) erg by Mereghetti et al.(2009) and estimated to be in the range 10 44–1045 erg by Tiengo et al.(2010). Taking the uppermost value of 10 45 erg would give an injected particle energy of 4 × 1043 erg. Assuming this results in synchrotron emission from the radio regime up to 6 keV, with a typical photon index of 1.5, we find a power of 1.1 × 1032 erg s−1, corresponding to a flux of 6 × 10−14(d/4 kpc)−2 erg s−1 cm−2 between 1 and 6 keV. This is well below the observed extended emission flux in 2009. For completeness, we note that there are alternative models proposing baryonic outflows for the magnetar outbursts (Gelfand et al., 2005; Granot et al., 2006). If this is the case, then no detectable synchrotron X-rays are expected. 4.4.2 Dust-scattering Halo

Extended emission around an X-ray source can be produced by the scattering of X-rays off dust particles between the source and observer. The flux of such a dust-scattering halo is expected to be proportional to the source flux (Mathis & Lee, 1991). Returning to Figure 4–2, then, for a dust-scattering halo, all of the points should fit well to a straight line (allowing for some scatter because the source spectrum did not remain constant). This is indeed the case for the first three points, although the 2009 point lies above the linear fit. In order to fit with the linear trend, the 2009 source flux would have to be 15%–20% higher than what we observed. However, because the scattered photons in a dust halo travel a longer path than photons observed directly from the source, the halo flux depends in a complicated manner on the recent history of the source flux over a period

67 of hours or days (Mauche & Gorenstein, 1986). The 2009 XMM-Newton observation of 1E 1547.0−5408 was taken only 13 days after its January outburst, at which time the magnetar’s flux was decaying following a power law of index α = −0.34 to −3.1 (Bernardini et al., 2011; Scholz & Kaspi, 2011). As a result, the source would have been bright enough to produce the observed halo 4.5–5.5 days prior to the observation, not an unreasonable timescale for the evolution of a dust halo. We therefore conclude that the observed variability in the extended emission flux is entirely consistent with a dust-scattering halo. VB09 rejected the dust-scattering halo interpretation of the extended emission based on two arguments. The first one is that the extended emission around 1E 1547.0−5408 had a harder spectrum than the source itself. A dust halo, on the other hand, is expected to have a softer spectrum than the source because the scattering cross section has an inverse-square dependence on energy. From the hardness ratios in Table 4–2 we find that, contrary to the claim by VB09, the extended source in region A has an unambiguously softer spectrum than the magnetar in 2007, 2009, and 2010, supporting the interpretation of dust scattering. In 2006, the hardness ratios suggest a softer spectrum too, although the difference is not statistically significant. For comparison, though, VB09 reported that the photon index Γ of the ‘PWN’ differed from that of the point source by only 1σ, which is not statistically significant either. We cannot, therefore, conclude that the spectrum of the extended emission in region A supports either interpretation in 2006. It should be noted, however, that be it harder or softer than the point source, the extended emission in 2006 still has a much softer spectrum than any previously reported PWN, as discussed above.

68 Table 4–2 also indicates similar results for region B as in region A. The extended emission shows a softer spectrum than the magnetar in all observations except 2006. Again, we note that although the extended emission spectrum is harder than the point source in 2006, it is still very soft overall. Our investigations so far strongly support that the extended emission observed around 1E 1547.0−5408 in regions A and B is dominated by a dust-scattering halo, at least in 2007, 2009, and 2010, although this interpretation is less clear in 2006. We now examine the other argument given in VB09 against dust scattering: that the extended emission was too bright, especially above 3 keV, to be a dust halo. They estimated the expected fractional halo intensity and the dust-scattering optical depth τsca based on models by Predehl & Schmitt(1995) and Draine(2003), which depend on the absorption column NH, the X-ray energy E, and a parameter β describing the distribution of the dust between source and observer. Following a similar procedure, we assumed an effective energy E = 2 keV for the 1–6 keV photons based on the spectrum of the magnetar, took β = 1

(meaning most of the dust is close to the source) and τsca = 1.5, as in VB09, and calculated Ifrac for regions A and B. This gives Ifrac = 0.08 for region A

3 and Ifrac = 0.18 for region B, which are in good agreement with the 2010 values in Figure 4–3, suggesting that dust scattering is adequate to explain the brightness of the extended emission, at least in 2010. For pure dust scattering we expect Ifrac in each region to be the same for each observation,

3 In region B, the calculated Ifrac is actually below the observed value by 3σ. However, if we subtract the non-dust contribution to the extended emis- sion from the SNR (see below and Table 4–3) then the two values are in full agreement.

69 Table 4–3: Contribution to Extended Emission Not from Dust Scattering

Region Count Rate of Extended Emissiona Fraction Not Attributable (cnt s−1) to Dust Scatteringa Extrapolated y-interceptb Observed Value in 2006 A 0.003(1) 0.0085(4) 0.36(12) B 0.040(3) 0.053(4) 0.75(6) a Numbers in parentheses are 1σ uncertainties. b Extrapolated y-intercept for the linear fits in Figure 4–2. except in 2009 where it should be higher due to source variability as discussed above. Indeed, there is good agreement between the 2007, 2009, and 2010 observations, but the 2006 value of Ifrac stands out. In region A, it is 3σ higher than what is expected from the 2007 and 2010 data, and in region B an even larger increase is evident, with Ifrac being >10σ higher in 2006 than in any subsequent observation. The best explanation for all of our results is that the extended emission around 1E 1547.0−5408 consists of a dust-scattering halo plus an additional component independent of the source flux. This secondary component is significant mainly in region B and becomes noticeable only when the halo is faint, as is the case for the 2006 observation. In order to better quantify it we return to Figure 4–2. For pure dust scattering, the linear fits in the diagram should pass through the origin. As seen from the inset, however, both fits have a positive y-intercept, suggesting that some of the extended emission does not come from the dust halo. In Table 4–3, we list the values of the y-intercepts for regions A and B and the extended emission count rate in 2006, and we calculate the fraction of the latter that was not contributed by dust scattering. We find that in region B, 75% ± 6% of the extended emission in 2006 is not from the dust halo. Since our region B mostly corresponds to the SNR region (4500 < r < 17400) from VB09 and therefore to the location of the radio

70 SNR shell, we conclude that the X-ray counterpart of this shell is the source of the non-dust extended emission here. Region A corresponds roughly to the ‘PWN’ region (400 < r < 4500) from VB09, noting that the broad PSF of XMM-Newton restricts us to r > 2000. Unlike farther out, the 2006 extended emission in region A is still dominated by the dust halo; only 36% ± 12% of it comes from another source. In fact, since the significance is only 3σ above zero, it is possible that dust scattering alone is sufficient to explain all of the extended emission in region A. Nevertheless, we can use the parameters in Table 4–3 to estimate an upper limit on the flux of a possible PWN, assuming an absorbed power-

22 −2 law spectrum with NH = 2.75 × 10 cm , as in VB09, and a photon index of Γ = 2, as is typical of PWNe. We find a 3σ upper limit4 on the

−14 −1 −2 2–10 keV unabsorbed flux of .4.7 × 10 erg s cm , corresponding to a luminosity of 9 × 1031(d/4 kpc)2 erg s−1. This implies an X-ray efficiency of

˙ −4 2 LX/E . 9 × 10 (d/4 kpc) . We also repeated our estimation for a softer PWN spectrum of Γ = 3, but found that the 2–10 keV upper limit was largely insensitive to changes in Γ. 4.5 Conclusions

In this chapter, we have examined multi-epoch XMM-Newton data for the magnetar candidate 1E 1547.0−5408 and we show that the observed extended emission surrounding the source is dominated by dust-scattered magnetar emission. Specifically we find that the luminosity of the nebular emission is proportional to the source flux, as expected for dust scattering,

4 In this case, the most stringent upper limit found was based on the non-detection of extended emission above 6 keV, not the parameters from Table 4–3.

71 but not seen in any known PWN or other magnetar candidate. Additional strong evidence for dust-scattering comes from spectral and energetics arguments, as well as from the disagreement between the observed nebular variability timescale and the expected synchrotron loss time in the PWN interpretation. We note that contrary to a previous claim (VB09), even in 2006 when the source was relatively faint, 64% ± 12% of the nebular emission is from dust scattering. We cannot, however, rule out the presence

31 2 −1 of a faint PWN with luminosity .9 × 10 (d/4 kpc) erg s in the 2–10 keV band, a limit three times lower than the previously claimed detection (∼2.9 × 1032 erg s−1 from VB09). Deep observations of this source when the magnetar is in quiescence are necessary to test this hypothesis. We do, on the other hand, find strong evidence for non-dust-scattered extended X-ray emission at angular distance ∼4000–15000, which we argue is from the SNR shell surrounding the pulsar, as previously reported by VB09. With the absence of evidence for a PWN surrounding any AXP or SGR, now including 1E 1547.0−5408, previous models for the production of PWNe by magnetars (Thompson & Blaes, 1998; Harding et al., 1999) remain unsupported. On the other hand, with clear evidence for the existence of PWNe surrounding many high-magnetic-field radio pulsars (e.g., PSR J1846−0258: Helfand et al., 2003; PSR J1119−6127: Gonzalez & Safi-Harb, 2003) which generally have substantially higher spin-down luminosities than any known AXP or SGR, the production of the relativistic particle wind necessary to generate an observable PWN seems intimately tied to the rotation-derived power, rather than that from magnetic-field decay. This then makes the detection of the surprisingly bright X-ray PWN surrounding the presumably rotation-powered but relatively high-B RRAT

72 PSR J1819−1458 (Rea et al., 2009a) particularly interesting and worthy of follow-up.

73 CHAPTER 5 The First Catalog of the Magnetar Population The contents of this chapter are reported in the paper “The McGill Magnetar Catalog,” published in the Astrophysical Journal Supplement Series (Olausen & Kaspi, 2014). Some of the values in Tables 5–2 and 5–3 have been updated with results from papers that were published after Olausen & Kaspi(2014) was submitted for publication, and the figures in Section 5.3 were changed to reflect the new data. None of these changes, however, significantly affect the paper’s analysis or conclusions. 5.1 Introduction

Now that the number of identified magnetars (see Section 1.2) and magnetar candidates have grown to over two dozen, the time is ripe for a systematic compilation of these objects in the form of the first magnetar catalog, presented in this chapter. Specifically, we have collected and com- piled a wide variety of information on the 21 confirmed and 5 unconfirmed magnetars, including their spatial, spin, and radiative properties across the electromagnetic spectrum. Our hope is that this catalog serves as a useful resource to the magnetar-interested community, and ultimately helps to identify and highlight important population properties that could help an- swer some of the outstanding questions in magnetar physics. Accompanying this catalog is a fully referenced and linked online version1 that is regularly maintained. We note that Manchester et al.(2005) include magnetars in

1 http://www.physics.mcgill.ca/pulsar/magnetar/main.html

74 their online and published radio pulsar catalog2, however the information compiled there is basic and restricted for the most part to spatial, spin, and radio properties. In Section 5.2, we present the catalog in the form of seven data tables separated by topic. In Section 5.3, we provide analysis and discussion of the magnetar population based on our catalogued data. Finally, concluding remarks are given in Section 5.4. 5.2 Data Tables

5.2.1 Table 1: Positions and Proper Motions

In Table 5–1, we list the astrometric parameters of the catalogued magnetars. These include the right ascension and declination (J2000.0 epoch), the Galactic longitude, l, and latitude, b, and the proper motion, µ, in R.A. and decl. Measurements of distances to the magnetars are listed in Table 5–7. The positions listed in this table are generally those from the literature with the smallest reported uncertainties. The uncertainties are unchanged from the original papers and typically, but not necessarily, represent 90% confidence intervals. In most cases, the listed position is from a Chandra observation of the persistent X-ray source, or Swift/X-ray Telescope (XRT) in the case of Swift J1822.3−1606. The exceptions are 4U 0142+61 and SGR 1806−20, where the position is of an optical counterpart, and 1E 1547.0−5408, SGR J1745−2900, XTE J1810−197, and SGR 1900+14, whose listed positions are of radio counterparts. Finally, the five candidate mag- netars have no confirmed counterparts at any wavelength, so we list either the best position of the observed bursts or, in the case of AX J1818.8−1559

2 http://www.atnf.csiro.au/research/pulsar/psrcat/

75 Table 5–1: Magnetar Positions and Proper Motions a a b b Name Right Ascension Declination l b µRA µDec References (J2000) (J2000) (◦)(◦) (mas yr−1) (mas yr−1) CXOU J010043.1−721134 01 00 43.14(13) −72 11 33.8(6) 301.93 −44.92 ······ 1 4U 0142+61 01 46 22.407(28)c +61 45 03.19(20)c 129.38 −0.43 −5.6(1.3) 2.9(1.3) 2, 3 SGR 0418+5729 04 18 33.867(43) +57 32 22.91(35) 147.98 +5.12 ······ 4 SGR 0501+4516 05 01 06.76(1) +45 16 33.92(11) 161.55 +1.95 ······ 5 SGR 0526−66 05 26 00.89(10) −66 04 36.3(6) 276.09 −33.25 ······ 6 1E 1048.1−5937 10 50 07.14(8) −59 53 21.4(6) 288.26 −0.52 ······ 7 1E 1547.0−5408 15 50 54.12386(64)d −54 18 24.1141(20)d 327.24 −0.13 4.8(5)f −7.9(3)f 8 PSR J1622−4950 16 22 44.89(8) −49 50 52.7(8) 333.85 −0.10 ······ 9 SGR 1627−41 16 35 51.844(20) −47 35 23.31(20) 336.98 −0.11 ······ 10 CXOU J164710.2−455216 16 47 10.20(3) −45 52 16.90(30) 339.55 −0.43 ······ 11 1RXS J170849.0−400910 17 08 46.87(6) −40 08 52.44(70) 346.48 +0.04 ······ 12 CXOU J171405.7−381031 17 14 05.74(5) −38 10 30.9(6) 348.68 +0.37 ······ 13 SGR J1745−2900 17 45 40.164(2)d −29 00 29.818(90)d 359.94 −0.05 ······ 14 SGR 1806−20 18 08 39.337(4)c −20 24 39.85(6)c 10.00 −0.24 −4.5(1.4) −6.9(2.0) 15, 16 XTE J1810−197 18 09 51.08696(28)d −19 43 51.9315(40)d 10.73 −0.16 −6.60(6)f −11.72(1.03)f 17 Swift J1822.3−1606 18 22 18.00(12) −16 04 26.8(1.8) 15.35 −1.02 ······ 18 SGR 1833−0832 18 33 44.37(3) −08 31 07.5(4) 23.34 +0.02 ······ 19 Swift J1834.9−0846 18 34 52.118(40) −08 45 56.02(60) 23.25 −0.34 ······ 20 76 1E 1841−045 18 41 19.343(20) −04 56 11.16(30) 27.39 −0.01 <4 <4 10, 21 SGR 1900+14 19 07 14.33(1)d +09 19 20.1(2)d 43.02 +0.77 −2.1(4) −0.6(5) 22, 16 1E 2259+586 23 01 08.295(77) +58 52 44.45(60) 109.09 −1.00 −9.9(1.1) −3.0(1.1) 23, 3 SGR 1801−23 18 00 59e −22 56 48e 6.91 +0.07 ······ 24 SGR 1808−20 18 08 11.2(29.5) −20 38 49(414) 9.74 −0.26 ······ 25 AX J1818.8−1559 18 18 51.38(4) −15 59 22.62(60) 15.04 −0.25 ······ 26 AX 1845.0−0258 18 44 54.68(4) −02 56 53.1(6) 29.56 +0.11 ······ 27 SGR 2013+34 20 13 56.9(7.3) +34 19 48(90) 72.32 −0.10 ······ 28 Note. In this and all subsequent tables, the unconfirmed candidate magnetars are separated from the confirmed magnetars by a horizontal line. a Positions are of the X-ray source unless otherwise specified. b Proper motions have been corrected for Galactic rotation unless otherwise specified. c Position of the near-infrared counterpart. d Position of the radio counterpart. e See reference for the size and shape of the error box. f Proper motion in the sky frame. References. (1) Lamb et al.(2002); (2) Hulleman et al.(2004); (3) Tendulkar et al.(2013); (4) van der Horst et al.(2010); (5) G¨o˘g¨u¸set al.(2010a); (6) Kulkarni et al.(2003); (7) Wang & Chakrabarty(2002); (8) Deller et al.(2012); (9) Anderson et al. (2012); (10) Wachter et al.(2004); (11) Muno et al.(2006); (12) Israel et al.(2003); (13) Halpern & Gotthelf(2010a); (14) Shan- non & Johnston(2013); (15) Israel et al.(2005); (16) Tendulkar et al.(2012); (17) Helfand et al.(2007); (18) Pagani et al.(2011); (19) G¨o˘g¨u¸set al.(2010b); (20) Kargaltsev et al.(2012); (21) Tendulkar(2013); (22) Frail et al.(1999); (23) Hulleman et al. (2001); (24) Cline et al.(2000); (25) Lamb et al.(2003); (26) Mereghetti et al.(2012); (27) Tam et al.(2006); (28) Sakamoto et al.(2011). and AX J1845.0−0258, the Chandra position of the unconfirmed, persistent X-ray counterpart. Unlike positions, all of the tabulated proper motion measurements or upper limits were found in the radio (1E 1547.0−5408 and XTE J1810−197) and optical (4U 0142+61, SGR 1806−20, 1E 1841−045, SGR 1900+14, and 1E 2259+586) bands (but see Kaplan et al., 2009b for proper motion upper limits found in X-ray with Chandra). The optical measurements are all corrected for Galactic rotation, whereas the radio ones are not. We also caution that the proper motion measurement of SGR 1900+14 is of its unconfirmed optical counterpart (see Table 5–4). 5.2.2 Table 2: Timing Properties

Table 5–2 contains timing parameters for all catalogued magnetars for which they are available. Specifically, we tabulate the period, P , and the epoch at which it was measured, the period derivative, P˙ , and the range over which it was measured, the method of measuring P˙ (see below), and three physical properties inferred from P and P˙ , namely, the surface dipolar magnetic field strength, B, the spin-down luminosity, E˙ , and the characteristic age, τc (see Section 1.1 for the definitions of these quantities). Note that the expression for B assumes simple vacuum dipole radiation and ignores the potentially important torques due to magnetospheric variability and the internal superfluid, both of which have been proposed to be relevant to magnetars (Kaspi et al., 2003; Dib et al., 2009; Archibald et al., 2013; Thompson et al., 2002; Beloborodov, 2009). The values of P˙ were found using one of two methods. In the first case (denoted in Table 5–2 by A), P˙ is a long-term average, calculated by fitting a slope to two or more individual measurements of the period. This was done for sources with only sparse timing data or, in the cases

77 Table 5–2: Magnetar Timing Properties a Name P Epoch P˙ P˙ Range Method B E˙ τc References (s) (MJD) (10−11 s s−1) (MJD) (1014 G) (1033 erg s−1) (kyr) CXOU J010043.1−721134 8.020392(9) 53032 1.88(8) 52044–53033 A 3.9 1.4 6.8 1 4U 0142+61 8.68869249(5) 53800 0.2022(4) 51610–53800 ED 1.3 0.12 68 2 SGR 0418+5729 9.07838822(5) 54993 0.0004(1) 54993–56164 E 0.061 0.00021 36000 3 SGR 0501+4516 5.7620695(1) 54701 0.594(2) 54701–55248 ED 1.9 1.2 15 4 SGR 0526−66 8.0544(2) 54414 3.8(1) 52152–54414 A 5.6 2.9 3.4 5 1E 1048.1−5937 6.4578754(25) 54185.9 ∼2.25 50473–54474 A 3.9 3.3 4.5 6 1E 1547.0−5408 2.0721255(1) 54854 ∼4.77 54743–55191 A 3.2 210 0.69 7 PSR J1622−4950 4.3261(1) 55080 1.7(1) 54939–55214 A 2.7 8.3 4.0 8 SGR 1627−41 2.594578(6) 54734 1.9(4) 54620–54736 A 2.2 43 2.2 9, 10 CXOU J164710.2−455216 10.610644(17) 53999.1 <0.04 53513–55857 A <0.66 <0.013 >420 11 1RXS J170849.0−400910 11.00502461(17) 54836 1.9455(13) 54836–55516 ED 4.7 0.58 9.0 2 CXOU J171405.7−381031 3.825352(4) 55272 6.40(5) 54856–55272 A 5.0 45 0.95 12 SGR J1745−2900 3.76363824(13) 56513 1.385(15) 56457–56519 ED 2.3 10 4.3 13

78 SGR 1806−20 7.547728(17) 53097.5 ∼49.5 52021–53098 A 20 45 0.24 14 XTE J1810−197 5.5403537(2) 54000 0.777(3) 53850–54127 E 2.1 1.8 11 15 Swift J1822.3−1606 8.43772106(6) 55761 0.00214(21) 55758–56600 E 0.14 0.0014 6300 16 SGR 1833−0832 7.5654084(4) 55274 0.35(3) 55274–55499 ED 1.6 0.32 34 17 Swift J1834.9−0846 2.4823018(1) 55783 0.796(12) 55782–55812 E 1.4 21 4.9 18 1E 1841−045 11.788978(1) 55585 4.092(15) 55615–55903 ED 7.0 0.99 4.6 2 SGR 1900+14 5.19987(7) 53826 9.2(4) 53634–53826 A 7.0 26 0.90 19 1E 2259+586 6.9790427250(15) 54187 0.048369(6) 54194–54852 E 0.59 0.056 230 2 SGR 1801−23 ··························· SGR 1808−20 ··························· AX J1818.8−1559 ··························· AX 1845.0−0258 6.97127(28) 49272 ·················· 20 SGR 2013+34 ···························

a Method by which P˙ was measured. A: long-term average, E: phase-coherent timing ephemeris. ED: phase-coherent timing ephemeris with addi- tional higher derivatives. References. (1) McGarry et al.(2005); (2) Dib & Kaspi(2014); (3) Rea et al.(2013); (4) Camero et al.(2014); (5) Tiengo et al.(2009); (6) Dib et al.(2009); (7) Dib et al.(2012); (8) Levin et al.(2010); (9) Esposito et al.(2009b); (10) Esposito et al.(2009a); (11) An et al.(2013a); (12) Sato et al.(2010); (13) Kaspi et al.(2014); (14) Woods et al.(2007); (15) Camilo et al.(2007b); (16) Scholz et al.(2014); (17) Esposito et al.(2011); (18) Kargaltsev et al.(2012); (19) Mereghetti et al.(2006); (20) Torii et al.(1998). of 1E 1048.1−5937, 1E 1547.0−5408, and SGR 1806−20, for sources with large variations in P˙ . In the second case (E, ED), P˙ was taken from a phase-coherent timing ephemeris that spans the specified range. If the ephemeris has higher-order derivatives (denoted by ED), then the listed value of P˙ is only accurate at the period epoch; otherwise P˙ is valid over the entire range. For sources where multiple phase-coherent timing solutions were found in the literature, we generally chose the solution from the most recent refereed publication that covered the most recent glitch-free interval of time, preferring solutions that covered at least several months. If a publication presented multiple timing solutions covering the same interval, we selected the solution that was preferred by the authors. In all cases, see the references provided for details. 5.2.3 Table 3: Quiescent Soft X-ray Properties

This table contains the soft X-ray properties of catalog magnetars in quiescence. To facilitate cross-source comparisons, we generally report only the phenomenological parameters of an absorbed blackbody plus power-law model, though in several cases only one of these two components is required.

The columns provided are the neutral hydrogen column density, NH, spectral photon index, Γ, blackbody temperature, kT , a second blackbody temperature, kT2 (only used for CXOU J010043.1−721134, for which a blackbody plus power law was a poor fit to the data), and the absorbed and unabsorbed fluxes as well as the energy range over which they were derived. We also include a column for the 2–10 keV unabsorbed flux, which was estimated with the WebPIMMS tool3 in cases where the reference gave

3 http://heasarc.gsfc.nasa.gov/Tools/w3pimms.html

79 Table 5–3: Soft X-Ray Properties of Magnetars in Quiescence a a a Name NH Γ kT kT2 Abs. Flux Unabs. Flux Energy Range References Unabs. Flux (1022 cm−2) (keV) (keV) (keV) (2–10 keV)

+0.020 +0.09 CXOU J010043.1−721134 0.063−0.016 ··· 0.30(2) 0.68−0.07 0.14 0.14 2–10 1 0.14 +0.004 4U 0142+61 1.00(1) 3.88(1) 0.410−0.002 ··· 58(1) ··· 2–10 2 67.9 +0.0014 SGR 0418+5729 0.115(6) ··· 0.32(5) ··· 0.012(1) ··· 0.5–10 3 0.0020−0.0010 SGR 0501+4516 0.88(1) 3.84(6) 0.50(2) ··· 3.0(2) 19(1) 0.5–10 4 1.7 +0.058 +0.11 +0.08 +0.13 SGR 0526−66 0.604−0.059 2.50−0.12 0.44(2) ··· 1.01−0.13 1.58−0.20 0.5–10 5 0.55 1E 1048.1−5937 0.97(1) 3.14(11) 0.56(1) ······ 5.1(1) 2–10 6 5.1(1) c +0.01 1E 1547.0−5408 3.2(2) 4.0(2) 0.43(3) ··· 0.37−0.03 ··· 0.5–10 7 0.54 b +1.6 +0.008 +0.09 +0.063 PSR J1622−4950 5.4−1.4 ··· 0.5(1) ··· 0.030−0.006 0.11−0.04 0.3–10 8 0.045−0.028 c +0.03 +0.17 SGR 1627−41 10(2) 2.9(8) ······ 0.10−0.02 ··· 2–10 9, 10 0.25−0.10 CXOU J164710.2−455216 2.39(5)b 3.86(22) 0.59(6) ······ 0.25(4) 2–10 11 0.25(4) +0.008 +0.007 +0.4 1RXS J170849.0−400910 1.36(4) 2.792−0.012 0.456−0.004 ······ 87.0−0.2 0.5–10 12 24.3 +0.15 +0.09 CXOU J171405.7−381031 3.95−0.14 3.45−0.08 ······ 1.51(3) 2.68(9) 2–10 13 2.68(9) SGR J1745−2900 ··············· <0.013 2–10 14 <0.013 SGR 1806−20 6.9(4) 1.6(1) 0.55(7) ······ 18(1) 2–10 15 18(1) XTE J1810−197 0.63(5)c ··· 0.18(2) ··· 0.75 ··· 0.5–10 16 0.029 c +0.20 d Swift J1822.3−1606 0.453(8) ··· 0.12(2) ··· 0.09−0.09 ··· 0.1–2.4 17 <0.0013 80 SGR 1833−0832 ············ <0.02 <0.2 2–10 18 <0.2 Swift J1834.9−0846 ··············· <0.004 2–10 19 <0.004 +9 1E 1841−045 2.2(1) 1.9(2) 0.45(3) ······ 43−12 0.5–10 20 21.3 SGR 1900+14 2.12(8) 1.9(1) 0.47(2) ······ 4.8(2) 2–10 21 4.8(2) 1E 2259+586 1.012(7) 3.75(4) 0.37(1) ··· 11.5(2) 14.1(3) 2–10 22 14.1(3) SGR 1801−23 ··························· SGR 1808−20 ··························· +0.16 AX J1818.8−1559 3.6(5) 1.17(17) ······ 1.37(7) ··· 2–10 23 1.68−0.15 1.6(3) ··· 1.87(12) ··· 1.26(7) ··· 2–10 23 1.37(10) +2.3 +0.5 +0.07 +0.07 AX 1845.0−0258 7.8−1.8 1.0−0.3 ······ 0.28(2) 0.33−0.08 2–10 24 0.33−0.08 +1.6 +0.4 +0.10 +0.10 5.6−1.2 ··· 2.0−0.3 ··· 0.26(2) 0.40−0.11 2–10 24 0.40−0.11 SGR 2013+34 ··························· a Fluxes are listed in units of 10−12 erg s−1 cm−2. b The flux of this source was fading and may not yet have reached quiescence during the observation used. c NH was fixed at the best-fit value when fitting the quiescent spectrum. d Elsewhere in this paper, we use the more conservative flux upper limit of 2.5×10−14 erg s−1 cm−2 for Swift J1822.3−1606, derived from the quies- cent parameters given in Rea et al.(2012b). References. (1) Tiengo et al.(2008); (2) Rea et al.(2007a); (3) Rea et al.(2013); (4) Camero et al.(2014); (5) Park et al.(2012); (6) Tam et al. (2008); (7) Bernardini et al.(2011); (8) Anderson et al.(2012); (9) Esposito et al.(2008); (10) An et al.(2012); (11) An et al.(2013a); (12) Rea et al.(2007b); (13) Sato et al.(2010); (14) Mori et al.(2013); (15) Esposito et al.(2007); (16) Gotthelf et al.(2004); (17) Scholz et al.(2012); (18) Esposito et al.(2011); (19) Younes et al.(2012); (20) Kumar & Safi-Harb(2010); (21) Mereghetti et al.(2006); (22) Zhu et al.(2008); (23) Mereghetti et al.(2012); (24) Tam et al.(2006). only absorbed flux or flux in a different energy range. X-ray luminosities are reported in Table 5–7. The tabulated parameters generally differ in various papers in the literature for any given source, therefore, the following explains our proce- dure in selecting which properties to catalog. We selected parameters from publications in which the reported source flux was historically lowest, in order to ensure as accurately as possible that the source was truly in quies- cence. In cases where there were multiple publications with equivalently low flux, we report the model parameters that had the smallest uncertainties, unless more recent observations appeared to be more reliable, e.g., were able to better disentangle potentially contaminating supernova remnant emission. For a majority of the sources, this resulted in the use of spectral parameters obtained from XMM-Newton data, although in several cases the results are taken from Chandra (1E 1048.1−5937, Swift J1834.9−0846, AX J1845.0−0258, and SGRs 0526−66, 1627−41, and J1745−2900) or archival ROSAT (SGR 0501+4516, XTE J1810−197, and Swift J1822.3−1606) data instead. The only other exception is SGR 1806−20 for which we use a model fit derived from simultaneous Suzaku and XMM observations. We caution that in general the stated uncertainties, statistical in nature, may be smaller than the systematic uncertainties due to calibration and cross-calibration issues; for this reason, reported parameters may not be optimal when considering data from a different telescope even in the absence of source variability. There are a few caveats we must make with regards to the flux values listed in Table 5–3. First, although we do list the lowest reported flux for PSR J1622−4950, it is not clear whether the source had reached quiescence during that observation or whether it was still fading. Hence, the value

81 we report may be an overestimation of its true quiescent flux. Also, note the upper limit for the 2–10 keV flux of Swift J1822.3−1606 even though it was detected in quiescence. The reasons for this are that Scholz et al. (2012) reported the lower bound for the 0.1–2.4 keV flux to be zero (likely due to rounding since they do not claim their result is consistent with a non-detection) and that varying the spectral parameters within their reported uncertainties changed the estimated 2–10 keV flux by over an order of magnitude. We therefore decided to report the highest such estimated flux as an upper limit. Additionally, Rea et al.(2012b) reported somewhat different spectral parameters for the same observation that gave a 2–10 keV flux an order of magnitude greater than the one in the table; it is this more conservative value that we use as an upper limit in calculations (including that of the luminosity in Table 5–7) and figures presented in this chapter. Finally, for the candidate magnetars AX J1818.8−1559 and AX J1845.0−0258, we provide separate spectral parameters and fluxes for single power-law and single blackbody models, but these results are for unconfirmed quiescent X-ray counterparts that may not be correctly identified. The parameters in Table 5–3 are identical to what is provided in the main table of our online catalog. However, we also provide a table of alternative values online, including model parameter results from other observations (e.g., from different telescopes) which may also be of interest. 5.2.4 Table 4: Optical and Near-Infrared Observations

In Table 5–4, we summarize measurements of catalog magnetars made in the optical and near-infrared bands. Because magnetars are typically variable sources at these wavelengths, we list the range of magnitudes over which they have been detected in the Ks, H, J, I, R, V , B, and U bands.

82 Table 5–4: Optical and Near-Infrared Counterparts of Magnetars

Name Ks HJIRVBU References CXOU J010043.1−721134a ········· >25.9 ··· 24.2–>26.2 >25.6 >24.2 1, 2 4U 0142+61 19.7–20.8 20.5–20.9 22.0–22.2 23.4–24.0c 24.9–25.6 25.3–26.1 27.2–28.1 >25.8 3–6 SGR 0418+5729 >19.6 ··· >27.4 >25.1 >24 >28.6 ······ 7–10 SGR 0501+4516 18.6–19.7c ······ 23.3–24.4c >23.0 ··· >26.9 >24.7 11–14 SGR 0526−66 ········· >26.7 ··· >26.6 >24.7 >25.0 15 1E 1048.1−5937 19.4–21.5 20.8–>23.3 21.7–>25.0 24.9–26.2c >26.0 >25.5 >27.6 >25.7 16–21 1E 1547.0−5408a 18.5–>21.7 ············ >20.4 >20.7 >20.3 22–24 PSR J1622−4950 >20.7 ····················· 25 SGR 1627−41 ≥19.1* >19.5 >21.5 ··············· 26, 27 CXOU J164710.2−455216 >21 ····················· 28 1RXS J170849.0−400910b ≥18.9* ≥20.0* ≥21.9* >25.1 >26.5 ········· 29–31 CXOU J171405.7−381031 ··························· SGR J1745−2900 ··························· SGR 1806−20 19.3–21.9 >19.5 >21.2 ··· >21.5 ········· 32–34 XTE J1810−197 20.8–21.9 21.5–22.7 22.9–23.9 >24.3 >21.5 >22.5 ······ 31, 34–38 Swift J1822.3−1606 >17.3 >18.3 >19.3 >22.2 ············ 39 SGR 1833−0832 >22.4 ······ >24.9 ··· >21.4 >21.3 >22.3 40, 41 Swift J1834.9−0846 >19.5 ······ >21.6 ············ 42, 43 1E 1841−045a 19.6–20.5 20.8–>21.5 >22.1 ··············· 31, 44 83 SGR 1900+14a 19.2–19.7 ······ >21 ············ 31, 45 1E 2259+586 20.4–21.7 ··· >23.8 >25.6 >26.4 ········· 46–48 SGR 1801−23 ··························· SGR 1808−20 ··························· AX J1818.8−1559 >17 ····················· 49 AX 1845.0−0258 ··· >21 ·················· 50 SGR 2013+34 >18.3 >18.5 >19.3 >20.6 >19 >20.2 >21.8 >21.2 51–54 Note. We do not distinguish between the standard filters listed and any other ones such as K, K0, z0, r0, etc. See Table 3 of the online catalog or the original references for further information. a Counterpart is unconfirmed. b The originally proposed counterpart has been disputed by Testa et al.(2008). c Pulsations have been detected in this waveband. References. (1) Durant & van Kerkwijk(2005a); (2) Durant & van Kerkwijk(2008); (3) Hulleman et al.(2004); (4) Dhillon et al.(2005); (5) Morii et al.(2005); (6) Durant & van Kerkwijk(2006b); (7) van der Horst et al.(2010); (8) Esposito et al.(2010); (9) Durant et al. (2011); (10) Rea et al.(2013); (11) Tanvir & Varricatt(2008); (12) Halpern(2008); (13) Fatkhullin et al.(2008); (14) Dhillon et al.(2011); (15) Kaplan et al.(2001); (16) Israel et al.(2002); (17) Wang & Chakrabarty(2002); (18) Durant & van Kerkwijk(2005b); (19) Tam et al. (2008); (20) Wang et al.(2008a); (21) Dhillon et al.(2009); (22) Holland & Krimm(2008); (23) Mignani et al.(2009); (24) Israel et al. (2009); (25) Anderson et al.(2012); (26) Wachter et al.(2004); (27) de Ugarte Postigo et al.(2009); (28) Wang et al.(2006); (29) Israel et al.(2003); (30) Durant & van Kerkwijk(2006c); (31) Testa et al.(2008); (32) Kosugi et al.(2005); (33) Israel et al.(2005); (34) Bal- man et al.(2003); (35) Gotthelf et al.(2004); (36) Israel et al.(2004a); (37) Rea et al.(2004); (38) Camilo et al.(2007c); (39) Rea et al. (2012b); (40) Marshall & Gelbord(2010); (41) G¨o˘g¨u¸set al.(2010b); (42) Tello et al.(2011); (43) Kargaltsev et al.(2012); (44) Durant (2005); (45) Klose et al.(2001); (46) Hulleman et al.(2001); (47) Kaspi et al.(2003); (48) Tam et al.(2004); (49) Mereghetti et al.(2012); (50) Israel et al.(2004b); (51) Guidorzi et al.(2005); (52) Qiu et al.(2005); (53) Rosen et al.(2005); (54) Bloom(2005). We also provide the limiting magnitudes (usually 3σ upper limits, but occasionally 2σ or 5σ) in cases where observations failed to detect them. Since this table provides merely a range of values, we reference only the detections with the lowest and highest reported magnitudes and/or the non-detection with the highest reported limiting magnitude. In cases where the same observation was analyzed in both non-refereed and refereed publications, we considered only the latter for inclusion. Finally, we must caution that any ‘non-standard’ filter (that is, any filter other than the eight listed above, such as K, K0, z0, a Hubble Space Telescope filter, etc.) was assumed to be identical to whichever standard filter it most closely approximated, with no effort made to properly convert the magnitude. Therefore, please check the original references or the online catalog to confirm the filter used. Seven magnetars have confirmed counterparts in the optical or near- infrared: 4U 0142+61, SGR 0501+4516, 1E 1048.1−5937, 1E 1547.0−5408, SGR 1806−20, XTE J1810−197, and 1E 2259+586. Of these, optical pulsations have been detected from 4U 0142+61, 1E 1048.1−5937, and SGR 0501+4516, of which the latter also shows good evidence for pulsations in the near-infrared band. There are also suggested counterparts for CXOU J010043.1−721134, 1E 1841−045, and SGR 1900+14, but these are unconfirmed. There was a near-infrared counterpart proposed for 1RXS J170849.0−400910, but Testa et al.(2008) disputed the association when they found multiple fainter sources within the error circle of its X-ray position. To denote this ambiguity, we report the detected magnitude of the originally proposed candidate (Star 3 in Testa et al., 2008) as an upper limit marked with an asterisk. Similarly, Ks-band observations of SGR 1627−41 reveal multiple sources that may be the counterparts, therefore, as

84 an upper limit, we list the detected magnitude of the brightest one (Source C in de Ugarte Postigo et al., 2009). For more information, the online version of this catalog contains a more comprehensive table of optical and near-infrared counterparts. It tabulates individual observations of each magnetar, listing the date of observation, the detected (or limiting) magnitude, and any non-standard filters that were used. 5.2.5 Table 5: Radio and Mid-Infrared Observations

Table 5–5 contains information regarding radio and mid-infrared observations of catalogued magnetars. For radio observations, we list all radio frequency ranges in which detections of pulsations have been reported, as well as the reported dispersion measure (DM). We also list, where available, the range of detected flux densities in the 1.4 and 2.0 GHz bands; for sources that have never been detected at these wavelengths we provide an upper limit. Note that the transient radio counterparts of SGRs 1806−20 and 1900+14 were detected following giant flares (Cameron et al., 2005; Frail et al., 1999), but because no pulsations were ever detected they are not included in this table. For mid-infrared observations, we list the reported fluxes or flux upper limits for catalogued magnetars at three wavelengths: 4.5 µm, 8.0 µm, and 24 µm. Note that we are only concerned with the flux of the point source, so phenomena such as the infrared ring seen around SGR 1900+14 (Wachter et al., 2008) are not included. 5.2.6 Table 6: Hard X-ray and Gamma-ray Observations

This table contains the spectral properties of catalog magnetars in the hard (>10 keV) X-ray and gamma-ray range. The persistent hard X-ray emission from magnetars can typically be characterised by a power law,

85 Table 5–5: Radio and Mid-Infrared Observations of Magnetars Name Radio Mid-Infrared

Detection Frequencies DM S1.4 GHz S2.0 GHz References F4.5 µm F8.0 µm F24 µm References (GHz) (cm−3 pc) (µJy) (µJy) (µJy) (µJy) (µJy) CXOU J010043.1−721134 ··························· 4U 0142+61 0.11 27 <46 <4.5 1–3 32.1(2.0) 59.8(8.5) <38 22, 23 SGR 0418+5729 ··························· SGR 0501+4516 ········· <40 4 ············ SGR 0526−66 ··························· 1E 1048.1−5937 ······ <20 ··· 5 <5.2 <21.8 <39 24, 25 1E 1547.0−5408 1.4–8.6, 18.5, 43, 45 830(50) <500 – 4400a ··· 6, 7 ············ PSR J1622−4950 1.4–9.0, 17, 24 820(30) <1200 – 16500a ··· 8–10 ············ SGR 1627−41 ······ <80 ··· 11 ············ CXOU J164710.2−455216 ······ <40 ··· 12 ············ 1RXS J170849.0−400910 ······ <20 ··· 5 <120 <170 <590 24 CXOU J171405.7−381031 ··························· SGR J1745−2900 1.2–8.9, 14.6–20, 22 1778(3) ∼90 ∼200 13–16 ············ 86 SGR 1806−20 ········· <6.9 2 ············ XTE J1810−197 0.06, 0.35–19, 42, 88.5, 144 178(5) <150 – 13600a ··· 17, 18, 3 <23 <130 <880 24 Swift J1822.3−1606 ········· <50 19 ············ SGR 1833−0832 ······ <90 ··· 20 ············ Swift J1834.9−0846 ······ <220 <50 21 ············ 1E 1841−045 ······ <20 <10.2 5, 2 ············ SGR 1900+14 ········· <7.1 2 ············ 1E 2259+586 0.06, 0.11 79 ··· <10.8 2, 3 6.3(1.0) <20 ··· 26 SGR 1801−23 ··························· SGR 1808−20 ··························· AX J1818.8−1559 ··························· AX 1845.0−0258 ······ <20 <9.2 5, 2 ············ SGR 2013+34 ········· <9.7 2 ············ a Since these sources are not always visible in radio, the flux densities here range from the lowest reported upper limit for a non-detection to the highest detected value. References. (1) den Hartog et al.(2007); (2) Lazarus et al.(2012); (3) Malofeev et al.(2012); (4) Hessels et al.(2008); (5) Crawford et al.(2007); (6) Camilo et al.(2007a); (7) Camilo et al.(2008); (8) Levin et al.(2010); (9) Keith et al.(2011); (10) Anderson et al.(2012); (11) Esposito et al.(2009a); (12) Burgay et al.(2006a); (13) Shannon & Johnston(2013); (14) Eatough et al.(2013); (15) Spitler et al.(2014); (16) Palaniswamy et al.(2013); (17) Camilo et al.(2006); (18) Camilo et al.(2007c); (19) Rea et al.(2012b); (20) Esposito et al.(2011); (21) Esposito et al.(2013); (22) Wang & Kaspi (2008); (23) Wang et al.(2008b); (24) Wang et al.(2007); (25) Wang et al.(2008a); (26) Kaplan et al.(2009a). Table 5–6: Hard X-Ray and Gamma-Ray Observations of Magnetars Name Hard X-ray Spectral Parameters Gamma-Raya

b Telescope Pulsed Emission Total Emission Ecut References p p c t t c c Γ F20–150 keV Γ F20–150 keV (keV) F0.1–10 GeV CXOU J010043.1−721134 ························ +65 4U 0142+61 R, I 0.40(15) 2.68(1.34) 0.93(6) 9.09(35) 279−41 1 <0.9 +0.11 S ······ 0.89−0.10 ∼10.3 2 SGR 0418+5729 ····················· <0.4 +0 .20 +2 .0 SGR 0501+4516 I, S ······ 0 .79−0 .16 <3.5, 8 .4−1 .5 >100 3 <1.9 SGR 0526−66 ························ 1E 1048.1−5937d ····················· <5.3  +0 .28 +0 .42  +0 .9 1E 1547.0−5408 R, I − 0 .37−0 .20 – 1 .55−0 .26 4 .1 (9 )–7 .5−1 .0 0 .87 (7 )–1 .45 (4 ) <1.5, 8 .0 (2 .2 )–25 .2 (3 .7 ) 5 <10.0 +0 .06 +1 .4 S ······ 1 .54−0 .05 17 .4−1 .8 >200 6 PSR J1622−4950 ························ SGR 1627−41 ····················· <20.0 CXOU J164710.2−455216 ····················· <10.0 1RXS J170849.0−400910 R, I 0.86(16) 2.60(35) 1.13(6), 1.46(21) 5.2(1.0), 6.61(23) >300 7, 8 <10.0 CXOU J171405.7−381031 ························ +0 .46 +0 .20 SGR J1745−2900 N ······ 1 .47−0 .37 0 .67−0 .27 >50 9 ··· 87 SGR 1806−20 I ······ 1.5(3), 1.9(2) 6.0(9), 11(2) >160 10, 11 <0.6 S ······ 1.2(1)–1.7(1) ∼3.8–9.9 12 XTE J1810−197 ····················· <5.0 Swift J1822.3−1606 ························ SGR 1833−0832 ························ Swift J1834.9−0846 ························ 1E 1841−045 I 0.72(15) ∼4.0 1.32(11) ∼6.9 >140 13 <3.0 +0.30 +0.21 S 1.35−0.25 ∼2.7 1.62−0.22 ∼4.6 14 N 0.99(36) ∼3.0 1.33(3) ∼8.0 15 SGR 1900+14 I ······ 3.1(5) 1.6(4) >100 16 <0.4 S ······ 1.2(5)–1.4(3) ∼1.4–3.2 12 1E 2259+586 R, S −1.02(24) ∼5.9e ··· <2.0 ··· 13, 12 <1.7

SGR 1801−23 ························ SGR 1808−20 ························ AX J1818.8−1559 ························ AX 1845.0−0258 ························ SGR 2013+34 ························

Note. Values in italics were measured when the source was in outburst. a Gamma-ray flux upper limits are taken from Abdo et al.(2010a). b R: RXTE, I: Integral, S: Suzaku, N: NuSTAR. c Hard X-ray and gamma-ray fluxes are in units of 10−11 erg s−1 cm−2. d 1E 1048.1−5937 was detected in hard X-rays with INTEGRAL by Leyder et al.(2008), but no spectral information was given. e Pulsed emission from 1E 2259+586 was only observed by RXTE up to ∼25 keV, so the extrapolated 20–150 keV pulsed flux should not be considered reliable. References. (1) den Hartog et al.(2008b); (2) Enoto et al.(2011); (3) Rea et al.(2009b); (4) Enoto et al.(2010c); (5) Kuiper et al.(2012); (6) Enoto et al.(2010b); (7) G¨otz et al.(2007); (8) den Hartog et al.(2008a); (9) Mori et al.(2013); (10) Mereghetti et al.(2005); (11) Molkov et al.(2005); (12) Enoto et al.(2010a); (13) Kuiper et al.(2006); (14) Morii et al.(2010); (15) An et al.(2013b); (16) G¨otzet al.(2006). therefore we report the photon index Γ and the unabsorbed 20–150 keV flux (as in Section 5.2.3, this was estimated using WebPIMMS if flux was given for a different energy range) for both the pulsed and total emission, as denoted, respectively, by superscripts p and t. Additionally, because the hard X-ray spectrum is expected to break or turn over at some point, we also list the cutoff energy, Ecut, though, except for the case of 4U 0142+61, only lower limits are available. Most of the hard X-ray data in this table comes from the INTEGRAL and Suzaku telescopes, and we generally tried to include results from both instruments (in that order) for each source where available. For results from INTEGRAL, we preferred the parameters derived using the longest integration time, though if it was clear that the parameters differed between two different time spans we included both results. Additionally, in cases where one publication gave multiple parameters for the same Suzaku observation, we chose the one preferred by the authors. Apart from those two telescopes, RXTE data were used for the pulsed emission from some sources, and the results for SGR J1745−2900 were found with NuSTAR. Italicized values in the table, seen for SGR 0501+4516, 1E 1547.0−5408, and SGR J1745−2900, were taken when the source was in outburst, and, here, multiple values of the photon index and flux represent the source fading back into quiescence. Finally, we must clarify that the inconsistency seen for 1E 2259+586, where the pulsed flux is three times higher than the upper limit for the total flux, is due to pulsed emission only being seen by RXTE up to ∼25 keV, meaning the extrapolated flux value reported in the table must be greatly overestimated.

88 Unlike at lower energies, no magnetars have yet been detected in gamma rays. We therefore provide only upper limits on their 0.1–10 GeV flux, taken from Table 1 of Abdo et al.(2010a). 5.2.7 Table 7: Associations and Distances

In Table 5–7, we tabulate distances to catalogued magnetars and related information. In particular, for each source, we list any objects (e.g., supernova remnants, star clusters, etc.) that are proposed to be associated with it, the age of the supernova remnant (where applicable and available), the distance measurement, and specifically to which object the distance is measured (be it the magnetar itself or an associated object). Associations whose validity has been disputed are noted. We also tabulate two parameters calculated using the distance, d: the height above the Galactic plane, z, defined as z = d sin (b) where b is the Galactic latitude

(see Table 5–1); and the quiescent 2–10 keV X-ray luminosity, LX, defined

2 as LX = 4πd FX where FX is the unabsorbed 2–10 keV flux (see Table 5– 3). For sources with no distance measurements, these derived parameters were estimated assuming a distance of 10 kpc. Additionally, since CXOU J010043.1−721134 and SGR 0526−66 are extragalactic magnetars located in the Magellanic Clouds, we do not calculate z for them. In cases where multiple distances to the same source exist in the literature, we chose the most recently measured value. Usually, this distance was either consistent with earlier measurements or generally accepted over them among the literature, but for 1E 1048.1−5937 and 1E 2259+586 there is some disagreement in the literature between multiple incompatible distance measurements. For more details, see the table of alternate values in our online catalog, which lists these other distance measurements with references, or see the discussion in the papers cited in Table 5–7.

89 Table 5–7: Magnetar Associations and Distances a Name Proposed Associations SNR Age References Distance Measured To Reference z LX (kyr) (kpc) (pc) CXOU J010043.1−721134 SMC ··· 1 62.4(1.6) SMC 28 ··· 65 4U 0142+61 ········· 3.6(4) 0142+61 29 −27(3) 105 SGR 0418+5729 · · · · · · · · · ∼2 Perseus Arm 30 ∼180 0.00096 SGR 0501+4516 SNR HB 9b 4–7 2, 3 ∼2 Perseus Arm 31 ∼68 0.40 SGR 0526−66 LMC, SNR N49b, SL 463 ∼4.8 4–6 53.6(1.2) LMC 32 ··· 189 1E 1048.1−5937 GSH 288.3−0.5−28b ··· 7 9.0(1.7) 1048.1−5937 29 −82(15) 49 1E 1547.0−5408 SNR G327.24−0.13 ··· 8 4.5(5) 1547.0−5408 33 −10.3(1.1) 1.3 PSR J1622−4950 SNR G333.9+0.0 <6 9 ∼9 J1622−4950 34 ∼−16 0.44 SGR 1627−41 CTB 33, MC −71, SNR G337.0−0.1 ··· 10, 11 11.0(3) G337.0−0.1 11 −21.4(6) 3.6 CXOU J164710.2−455216 Westerlund 1 ··· 12 3.9(7) Westerlund 1 35 −29(5) 0.45 1RXS J170849.0−400910 ········· 3.8(5) J170849.0−400910 29 2.4(3) 42 +2.50 CXOU J171405.7−381031 SNR CTB 37B 0.65−0.30 13, 14 ∼13.2 CTB 37B 36 ∼86 56 SGR J1745−2900 Galactic Centre ··· 15 ∼8.5 Galactic Centre 37 ∼−7.0 <0.11 +1.8 +6.3 SGR 1806−20 W31, MC 13A, Star cluster ··· 16, 17 8.7−1.5 Star cluster 38 −36.7−7.6 163 +0.5 +1.1 XTE J1810−197 ········· 3.5−0.4 J1810−197 39 −9.7−1.4 0.043 Swift J1822.3−1606 M17 ··· 18 1.6(3) M17 18 −28.5(5.3) <0.0077 90 SGR 1833−0832 ··· ··· ··· ··· ··· ··· ∼3.6 <2.4 Swift J1834.9−0846 SNR W41 ∼100 19, 20 4.2(3) W41 40 −25(2) <0.0084 +1.3 +0.11 1E 1841−045 SNR Kes 73 0.5–1 21, 22 8.5−1.0 Kes 73 22 −0.97−0.15 184 SGR 1900+14 Star cluster ··· 23 12.5(1.7) Star cluster 41 167(23) 90 1E 2259+586 SNR CTB 109 14(2) 24, 25 3.2(2) CTB 109 42 −55.6(3.5) 17 SGR 1801−23 ··· ··· ··· ··· ··· ··· ∼12 ··· SGR 1808−20 · · · · · · · · · · · · · · · · · · ∼−45 ··· AX J1818.8−1559 · · · · · · · · · · · · · · · · · · ∼−44 20 AX 1845.0−0258 SNR G29.6+0.1 <8 26 ∼8.5 Scutum Arm 43 ∼16 2.9 SGR 2013+34 W58 ··· 27 ∼8.8 W58 27 ∼−16 ··· a 2–10 keV X-ray luminosity in units of 1033 erg s−1. No uncertainties have been included. b The proposed association with this source has been disputed. References. (1) Lamb et al.(2002); (2) Gaensler & Chatterjee(2008); (3) Leahy & Tian(2007); (4) Cline et al.(1982); (5) Klose et al.(2004); (6) Park et al.(2012); (7) Gaensler et al.(2005a); (8) Gelfand & Gaensler(2007); (9) Anderson et al.(2012); (10) Woods et al.(1999); (11) Cor- bel et al.(1999); (12) Muno et al.(2006); (13) Nakamura et al.(2009); (14) Halpern & Gotthelf(2010b); (15) Mori et al.(2013); (16) Fuchs et al. (1999); (17) Corbel & Eikenberry(2004); (18) Scholz et al.(2012); (19) Tian et al.(2007); (20) Kargaltsev et al.(2012); (21) Vasisht & Gotthelf (1997); (22) Tian & Leahy(2008); (23) Vrba et al.(2000); (24) Fahlman & Gregory(1981); (25) Sasaki et al.(2013); (26) Gaensler et al.(1999); (27) Sakamoto et al.(2011); (28) Haschke et al.(2012b); (29) Durant & van Kerkwijk(2006a); (30) van der Horst et al.(2010); (31) Lin et al. (2011); (32) Haschke et al.(2012a); (33) Tiengo et al.(2010); (34) Levin et al.(2010); (35) Kothes & Dougherty(2007); (36) Tian & Leahy(2012); (37) Shannon & Johnston(2013); (38) Bibby et al.(2008); (39) Minter et al.(2008); (40) Leahy & Tian(2008b); (41) Davies et al.(2009); (42) Kothes & Foster(2012); (43) Torii et al.(1998). 1745 20 1834 1822 s

r 1714 a t 1622 e

n 1833

g 15

a 0418 M

0501 d

e 1547

m 1647 r fi 0100 n 10 o 1810 C

f 1708 o

r 1627 e

b 1841

m 0142 u 5

N 1048 2259 1806 1900 0526 0 1980 1985 1990 1995 2000 2005 2010 Year

Figure 5–1: Number of confirmed magnetars discovered over time. Labels in boldface indicate the source was either discovered or later detected by an all-sky X-ray/soft gamma-ray burst monitor. The dashed and dot-dashed lines mark the launches of Swift in 2005 and Fermi in 2008, respectively.

5.3 Discussion

Figure 5–1 shows the accumulated number of known confirmed mag- netars as a function of year up to the present day. The vertical dashed line shows the launch date of Swift with its Burst Alert Telescope (BAT) on board (Barthelmy et al., 2005), and the dot-dashed line shows the launch date of the Fermi mission and its Gamma-ray Burst Monitor (GBM; Mee- gan et al., 2009). It is no coincidence that the slope of the accumulation increased significantly when BAT became active and again when GBM turned on, since they are extremely well designed to detect bright magnetar bursts. In fact, they, and previous all-sky X-ray/soft-gamma ray monitors,

91 Table 5–8: Magnetar Names

Current Name Alternate Current Name MG Name ATNF (PSR) Name CXOU J010043.1−721134 ··· MG J0100−7211 PSR J0100−7211 4U 0142+61 ··· MG J0146+6145 PSR J0146+6145 SGR 0418+5729 ··· MG J0418+5732 PSR J0418+5732 SGR 0501+4516 ··· MG J0501+4516 PSR J0501+4516 SGR 0526−66 ··· MG J0526−6604 PSR J0525−6607 1E 1048.1−5937 ··· MG J1050−5953 PSR J1050−5953 1E 1547.0−5408 SGR J1550−5418 MG J1550−5418 PSR J1550−5418 PSR J1622−4950 ··· MG J1622−4950 PSR J1622−4950 SGR 1627−41 ··· MG J1635−4735 PSR J1635−4735 CXOU J164710.2−455216 ··· MG J1647−4552 PSR J1647−4552 1RXS J170849.0−400910 ··· MG J1708−4008 PSR J1708−4008 CXOU J171405.7−381031 ··· MG J1714−3810 PSR J1714−3810 SGR J1745−2900 SGR J1745−29 MG J1745−2900 PSR J1745−2900 SGR 1806−20 ··· MG J1808−2024 PSR J1808−2024 XTE J1810−197 ··· MG J1809−1943 PSR J1809−1943 Swift J1822.3−1606 ··· MG J1822−1604 PSR J1822−1604 SGR 1833−0832 ··· MG J1833−0831 PSR J1833−0831 Swift J1834.9−0846 ··· MG J1834−0845 PSR J1834−0845 1E 1841−045 ··· MG J1841−0456 PSR J1841−0456 SGR 1900+14 ··· MG J1907+0919 PSR J1907+0919 1E 2259+586 ··· MG J2301+5852 PSR J2301+5852 SGR 1801−23 ········· SGR 1808−20 ········· AX J1818.8−1559 GRB 071017 ······ AX 1845.0−0258 ······ PSR J1844−0256 SGR 2013+34 GRB 050925 ······ were designed to detect GRBs, which are one-time bursters of cosmological origin. Hence, these monitors are specifically designed to view the entire sky in an unbiased fashion, and are therefore sensitive to Galactic, repeating bursters regardless of their location in the Galaxy. Thus, they have yielded a directionally unbiased sample of magnetars, selected only for their magnetar activity, namely bursting. In Figure 5–1, sources that were either discovered by an all-sky X-ray/soft gamma-ray monitor, or were later detected (and therefore could have been discovered) by one, are highlighted in bold, for this reason. Many known magnetars have thus been found via their bursting be- haviour, which raises an important point regarding how they are named. Because they tend to be found by burst monitors, magnetars have of- ten been named with the designation “SGR” in recent years (e.g., SGR 1833−0832 and SGR J1745−2900). We strongly argue that this naming

92 convention requires amendment because, as discussed in this work and extensively elsewhere (e.g., Gavriil et al., 2002; Kaspi et al., 2003; Woods & Thompson, 2006; Mereghetti, 2008, 2013; Kaspi, 2010; Rea & Esposito, 2011), the distinction between sources designated as “AXP” and “SGRs” has been largely erased via the discovery of objects that have properties previously ascribed to both categories. Today, it is very difficult to classify some sources as one or the other; rather it has become clear that there is a continuous spectrum of magnetar-type activity which can even include some high-B rotation-powered pulsars (e.g., PSR J1846−0258; Gavriil et al., 2008). Sources discovered via bursting seem to be SGRs, but they may later lie dormant and burstless for decades and seem to be AXPs (e.g., SGR 0526−66; Kulkarni et al., 2003). Meanwhile, sources discovered in quiescence and showing no bursts, and therefore initially classified as AXPs, may later begin bursting (e.g., 1E 1547.0−5408; Gelfand & Gaensler, 2007; Israel et al., 2010; Kaneko et al., 2010). A source’s fixed designation clearly cannot depend on behaviour that is constantly evolving. We instead propose a naming scheme that designates magnetars by the acronym ‘MG,’ analogous to ‘PSR’ as used for pulsars. In Table 5–8, we list the commonly used names (as well as a few less-common alternatives) of all catalogued magnetars along with our proposed new names, which follow the convention MG JHHMM+/−DDMM, and the PSR names used by the ATNF catalog. Note that this proposed naming scheme is only used for confirmed magnetars, though for completeness we also list the five magnetar candi- dates in the table. Another possibility would be to designate names such as those of other X-ray sources, for which the initial prefix is informative re- garding the discovery telescope, as for, e.g., XTE J1810−197, discovered by

93 15

10

5 ) c p

k 0 (

y

−5

−10 Magnetars XINSs Pulsars −15 −15 −10 −5 0 5 10 15 x (kpc)

Figure 5–2: Top-down view of the Galaxy, with the Galactic centre at co- ordinates (0, 0) and the location of the Sun marked by a cyan arrow at coordinates (0, 8.5). The greyscale shows the distribution of free electrons given by the model of Cordes & Lazio(2002). The magnetars are denoted by red circles with distance uncertainties indicated by the lines, the X-ray Isolated Neutron Stars (XINSs) are shown by the yellow circles near the Sun, and the locations of all other pulsars are given by the blue dots.

RXTE. We suggest that these, and other possible alternatives, be seriously discussed by the community. 5.3.1 Spatial Properties

Figure 5–2 shows a top-down view of the Galactic plane with the Galactic centre at coordinate (0, 0). The greyscale is the distribution of free electrons from the model of Cordes & Lazio(2002) and delineates the approximate locations of the spiral arms. Galactic disc radio pulsars from

94 the ATNF catalog4 are denoted with blue dots. The X-ray isolated neutron stars (XINSs) are shown in yellow and are, without exception, very close to the Sun. The magnetars are shown as red circles, with their estimated distance uncertainties indicated. Note the magnetar SGR J1745−2900, whose location is consistent with the Galactic centre. This plot clearly indicates the preponderance of magnetars in the direction of the inner Galaxy, but with several notable exceptions in the outer Galaxy. The lack of clustering around the solar system of magnetars, particularly compared with the known radio pulsar population, suggests that fewer selection effects exist in the known magnetar population, apart from selection for bursting, particularly in the Swift and Fermi eras. Figure 5–3 presents histograms of the distribution of ATNF Galactic radio pulsars and magnetars in Galactic longitude, l. The radio pulsars are colour-coded for age as indicated, and the magnetars are indicated by the hatched red region. As surmised from Figure 5–2, the known Galactic magnetars are more concentrated in the inner Galaxy, which is not a mere selection effect, again given the all-sky nature of the burst detectors. While selection effects in radio pulsar surveys may hinder the detection of the youngest objects in the very inner Galaxy where multipath scattering is important (Rickett, 1990), we can nevertheless compare the l distributions of the magnetars and young radio pulsars using a Kolmogorov-Smirnov (K-S) test to see if they are consistent with having been drawn from the same distribution. For radio pulsars having τ < 10 kyr, we find a K-S probability of the null hypothesis of p = 0.14, and likewise we also find p = 0.14

4 http://www.atnf.csiro.au/research/pulsar/psrcat/, version 1.50.

95 300 Binaries Isolated PSRs PSRs (<1 Myr) PSRs (<100 kyr) 250 PSRs (<10 kyr) Magnetars

r 200 e b m

u 150 N

100

50

0 20

15 r e b

m 10 u N 5

0 −180 −120 −60 0 60 120 180 Galactic Longitude (degrees)

Figure 5–3: Top panel: distribution in Galactic longitude, l, of all Galac- tic disc pulsars. Young, isolated pulsars are indicated by the various blue regions (<10 kyr: cyan; <100 kyr: blue; and <1 Myr: dark blue), with the remaining isolated pulsars and pulsars in binary systems shown respectively by the grey and light grey regions. Bottom panel: zoom-in to better show the distribution of the magnetars, given by the hatched red region, and the youngest pulsars.

96 250 30 Binaries Isolated PSRs PSRs (<1 Myr) 200 PSRs (<100 kyr) PSRs (<10 kyr)

r 20 Magnetars e b m u

r 150 N e

b 10 m u

N 100 0 0 2 4 6 8 10 Galactic Latitude (degrees) 50

0 0 5 10 15 20 25 30 Galactic Latitude (degrees)

Figure 5–4: Distribution in Galactic latitude, b, of all Galactic disc pulsars (colours as in Figure 5–3). Inset: zoom-in near the origin with the magne- tars shown by the hatched red region. for τ < 100 kyr. Hence, we cannot exclude that the two distributions are consistent with being drawn from the same underlying distribution. Figure 5–4 presents histograms of the distribution of ATNF Galactic disc radio pulsars and magnetars in Galactic latitude, b, in degrees, with a zoom-in to the most populated region in order to better highlight the magnetars that are relatively few in number. Note that with the exception of just one magnetar (SGR 0418+5729, but see Section 5.3.2), all known Galactic magnetars lie within 2◦ of the Galactic plane, consistent with their interpretation as a population of young objects. The physical scale height in parsecs, however, is more relevant to understanding the Galactic distribution, which we discuss below.

Magnetar Scale Height

In Figure 5–5 (bottom panel), we plot a histogram of the distribution of magnetars as a function of their height above the Galactic plane, z ≡ d sin(b), in parsecs, where d is the distance to the object in parsecs. It is

97 1

0.8

n 0.6 o i t c a r

F 0.4

0.2

0 6 r e

b 4 m u

N 2

0 −200 −100 0 100 200 Galactic Height, z (pc)

Figure 5–5: Top panel: cumulative distribution function of the height, z, above the Galactic plane for the 19 magnetars located in the Milky Way. Data are fit to an exponential model (solid line) and a self-gravitating, isothermal disc model (dashed line). See the text for details. Bottom panel: histogram of the distribution in z of the Galactic magnetars. Lines are as above.

98 evident that the distribution does not peak at z = 0, meaning that simply fitting the distribution to exp (− |z| /h), as is typically done for pulsars, will not give an accurate result. The Sun does not lie in the Galactic plane as defined by the magnetars. We therefore used two models that included a term for the height of the Sun: an exponential model and a self-gravitating, isothermal disc model (e.g., Bahcall, 1984):     |z + z0| 2 |z + z0| n(z) = n0 exp − , n(z) = n0 sech , he 2hs where he and hs are, respectively, the scale heights for the exponential and self-gravitating models, and z0 is the height of the Sun above the Galactic plane. Due to the small number of sources we can work with, as well as the significant distance uncertainties involved, we constructed and fit our models to the unbinned cumulative distribution function (top panel of Figure 5–5) rather than fitting them directly to the histogram. The resulting best-fit values were he = 30.7 ± 5.9 pc and z0 = 13.5 ± 2.6 pc for the exponential model, and hs = 17.9 ± 3.3 pc and z0 = 13.9 ± 2.5 pc for the self-gravitating model. Note that the listed 1σ uncertainties include both the statistical uncertainty from fitting as well as the 1σ uncertainty obtained from a Monte Carlo analysis in which we randomly varied the distance (and therefore z) to each magnetar within their uncertainties. In an effort to check the stability of our results, we also repeated this procedure for a few different subsets of the magnetar population. In particular, we tried fitting the two models to only the 14 Galactic magnetars that have been detected by all-sky monitors (see Figure 5–1) since those sources do not have any sort of directional selection effects. Additionally, since the bottom panel of Figure 5–5 suggests that fitting to the cumulative distribution weights the outlying points more

99 heavily if they were fit to the histogram, we also tested fits excluding the two sources with |z| > 100 pc (SGR 0418+5729 and SGR 1900+14). We found that these changes tended to decrease he and hs and increase z0. Overall, the best-fit values for the scale height varied within the range of

∼20–31 pc for he and ∼13–18 pc for hs, and the best-fit values for the height of the Sun z0 ranged from ∼13–22 pc for both models. For comparison, we repeated the same procedure for all ATNF pulsars with characteristic ages less than 100 kyr (excluding magnetars) and found scale heights of he = 61 ± 5 pc and hs = 39 ± 3 pc, approximately twice as large as our results for the magnetars. However, note that, unlike the magnetars, strong selection effects are at work in shaping the known population of radio pulsars (see, e.g., Faucher-Gigu`ere& Kaspi, 2006 for a detailed discussion). Indeed, it is generally more difficult to find faster — hence typically younger — radio pulsars closer to the Galactic plane because of the deleterious effects of dispersion smearing and scattering, though recent pulsar surveys of the radio sky are improving the situation (Manchester et al., 2001; Lazarus, 2013). Hence, we may easily have over-estimated the scale height of young radio pulsars. Regardless, it is unsurprising that the scale height of magnetars is smaller or similar to that of young radio pulsars, given that magnetars are believed to be young neutron stars. We can also compare our results with measurements in the literature of the scale heights of OB stars, the progenitors of neutron stars. In particular,

Reed(2000) and Elias et al.(2006) derived values of he (45 ± 20 pc and 34 ± 2 pc, respectively) which overlap with the upper end of our own range, but other measurements by Joshi(2007; he = 61.4 ± 2.6 pc) and Ma´ız-

Apell´aniz(2001; hs = 34.2 pc) are significantly greater. This discrepancy

100 may argue in favour of the hypothesis that magnetars are born from massive progenitors (Figer et al., 2005; Muno et al., 2006) if the OB star scale height depends on stellar mass such that more massive O stars have a scale height that agrees with that of the magnetars. Unfortunately, there is no compelling evidence for such a dependence on stellar mass via spectral type (Ma´ızApell´anizet al., 2008), though it cannot yet be claimed disproven either. Nevertheless, we argue that the observed magnetar scale height favours massive progenitors. In particular, 9 M stars have an expected lifetime of about 20 Myr (Milhalas & Binney, 1981), therefore, assuming a peculiar velocity of ∼5–10 km s−1 (Gies, 1987), they will have travelled ∼70– 140 pc in the direction perpendicular to the plane by the end of their lives, significantly greater than the ∼20–30 pc magnetar scale height. Conversely,

40 M stars live for approximately 1 Myr, so given the same velocity they will travel only ∼3–7 pc during their life span, a much smaller value that is consistent with the observed distribution of magnetars. Finally, we find that our measurement of the height of the Sun above the Galactic plane, z0, agrees well with previous measurements, which generally all fall within the range of 10–30 pc (e.g., ∼10–12 pc, Reed, 1997; 15 ± 3 pc, Conti & Vacca, 1990; 16 ± 5 pc, Elias et al., 2006; 24.2 ± 2.1 pc, Ma´ız-Apell´aniz, 2001). 5.3.2 Timing Properties

In Figures 5–6 to 5–9, we show histograms of pulse periods and properties inferred from timing of the radio pulsar population (not including those found in globular clusters), the XINSs, and the magnetars. Figure 5–6 shows the periods, and it is clear that magnetars have longer spin periods than the vast majority of the radio pulsars, although there is overlap with the long-period tail of the radio pulsar distribution. Additionally, the spin

101 120 20 Binaries Isolated PSRs 100 15 PSRs (<1 Myr) r PSRs (<100 kyr) e

b PSRs (<10 kyr) 10 m XINSs u Magnetars 80 N 5 r

e 0 b 1 5 10 m 60

u Period N

40

20

0 10−3 0.01 0.1 1 10 Period (s)

Figure 5–6: Histogram showing the distribution in pulse period of all known radio pulsars (colours as in Figure 5–3, globular cluster pulsars are not in- cluded), XINSs (yellow), and magnetars (red). Inset: zoom-in on P > 1 s, where the magnetars are all located. periods of the magnetars are very similar to those of the XINSs. Indeed, models of magnetic and thermal evolution in neutron stars are suggestive of an evolutionary relationship between magnetars and XINSs, with the latter descendants of the former (Vigano et al., 2013; Popov et al., 2010). Also notable is the small range of magnetar periods, especially compared with those of radio pulsars. The paucity at shorter periods is understood as being a result of their rapid spin-down due to their high-B fields. On the other hand, the reason for the lack of magnetar spin periods longer than 12 s is not well established; one possibility is that by the time objects reach so long a period, their fields have decayed so much that the hallmark activity and X-ray emission has ceased (e.g., Colpi et al., 2000). On the other hand, the longest period magnetar yet known (1E 1841−045) also has the highest persistent 2–10 keV luminosity (Table 5–3). This suggests that even longer-period magnetars are yet to be found.

102 150 15 Binaries Isolated PSRs PSRs (<1 Myr)

r PSRs (<100 kyr) 125 e 10 b PSRs (<10 kyr)

m XINSs u Magnetars

N 5 100 r

e 0 b 1013 1014 1015 m

u 75 Magnetic Field (G) N

50

25

0 108 109 1010 1011 1012 1013 1014 1015 Magnetic Field (G)

Figure 5–7: Histogram showing the distribution in magnetic field, B, of all known radio pulsars, XINSs, and magnetars for which P˙ has been measured (colours as in Figure 5–6). Inset: zoom-in on B > 5 × 1012 G to better show the distribution of the magnetars.

In Figure 5–7, distributions of the spin-inferred surface dipolar magnetic field, B, are shown. Again, it is clear that the typical magnetar field is two to three orders of magnitude greater than that of the typical radio pulsar, and indeed the overlap of the magnetar field distribution with the high-B tail of the radio pulsar distribution is relatively small, restricted to three objects (SGR 0418+5729, Swift J1822.3−1606, and 1E 2259+586). Indeed, the magnetars largely stand alone on this plot, with the XINSs having intermediate field values. Much has been made of the discovery of SGR 0418+5729 (Rea et al., 2010) given its low spin-inferred B strength; however, Figure 5–7 makes clear that when viewing the overall known magnetar population, which is largely selected in an unbiased fashion based on burst activity, low-B objects are the exception. Interestingly, this figure also shows that the younger known radio pulsars tend to have B fields higher than the field of the typical known radio

103 Binaries 100 Isolated PSRs PSRs (<1 Myr) PSRs (<100 kyr) PSRs (<10 kyr) 80 XINSs Magnetars r

e 60 b m u N 40

20

0 1030 1033 1036 1039 Spin-down Luminosity (erg s-1)

Figure 5–8: Same as Figure 5–7 but for the spin-down luminosity, E˙ . pulsar. This might naively suggest that radio pulsar magnetic fields decay with time. On the other hand, higher-field sources spin down more rapidly, reaching the death line sooner, so the most common radio pulsar found is likelier to have lower B since it has a longer lifetime. The small scale height for magnetars described in Section 5.3.1 then is consistent with the relative numbers of high-B and low-B magnetars: the objects with the highest fields have the smallest lifetimes, hence, they have little time to leave their birthplace. Indeed, it is unsurprising that the source with the lowest known B field, SGR J0418+5729, is also the magnetar furthest from the Galactic plane (see Table 5–7). Figure 5–8 shows a histogram of the spin-down luminosity, E˙ . In this plot, the magnetars are distributed fairly uniformly but broadly, spanning a full five orders of magnitude. Below, we consider correlations between E˙ and radiative properties, but for the moment we note that the broad range of E˙ — in contrast to the far narrower and more distinctive range in B —

104 80 Binaries Isolated PSRs PSRs (<1 Myr) PSRs (<100 kyr) 60 PSRs (<10 kyr) XINSs Magnetars r e b

m 40 u N

20

0 103 106 109 Characteristic Age (yr)

Figure 5–9: Same as Figure 5–7 but for the characteristic age. suggests that the former does not play a dominant role in the high-energy emission from magnetars. ˙ Figure 5–9 shows distributions of characteristic age, τc. As with E, magnetar ages are uniformly but broadly distributed. The breadth is inter- estingly at odds with their very small Galactic scale height (Section 5.3.1), even given magnetars’ relatively low mean velocity (Tendulkar et al., 2013). This indicates that the characteristic ages of magnetars are poor proxies for their true ages. Independent evidence for this is already clear from the disparity in the characteristic age of 1E 2259+586 (230 kyr; see Table 5–2) compared with the estimated age of its host supernova remnant CTB 109 (14 kyr; see Table 7). Note, however, that the latter example is extreme; in contrast stands 1E 1841−045 whose characteristic age, 4.8 kyr, is much closer (though still larger) than the estimated age of its host remnant, Kes 73 (0.5–1 kyr). The primary reason for the breadth in characteristic age is unclear. In some cases it may be at least partially due to fluctuations in P˙ (as in 1E 1048.1−5937; Gavriil & Kaspi, 2004; Dib & Kaspi, 2014) which

105 10−9 Magnetars XINSs PSRs (<10 kyr) PSRs (<100 kyr) PSRs (<1 Myr) Isolated PSRs Binaries −12 10 14 10 G )

1 r - 1 ky s

s (

e v i t

a −15 v i 10 r r 0 ky e 10 D 10 12 d G o i r e P yr 10 M 10−18

yr 1 G 10 10 G

10−21 0.01 0.1 1 10 Period (s)

Figure 5–10: P –P˙ diagram for all known radio pulsars (grey or blue dots as indicated, globular cluster pulsars are not included), XINSs (yellow squares), and magnetars (red stars). could bias a short-term measurement. Alternatively, torque decay as the magnetic field decays is also a likely factor (e.g., Thompson et al., 2002). A third possibility is simply that some magnetars are born with spin periods similar to their current values (e.g., the age disparity seen in 1E 2259+586 could be explained with a birth spin period of P0 = 6.76 s; see Equation 1.5). In Figure 5–10, we present a P –P˙ diagram which includes all cata- logued magnetars, XINSs, and radio pulsars having measured P and P˙ . This presentation re-emphasises the relatively long periods and large spin- down rates of the magnetar population. Also made clear by this diagram is the overlap in P –P˙ space between magnetars and some radio pulsars. This is suggestive of potential magnetar activity from these apparently

106 r = −0.79 r = 0.46 4 p = 0.0020 0.7 p = 0.076

0.6 3.5 G

,

x 0.5 ) e V d

3 e n I k

(

n 0.4 T o t k o

h 2.5

P 0.3

2 0.2

1.5 0.1 1014 1015 1013 1014 1015 Magnetic Field, B (G) Magnetic Field, B (G)

Figure 5–11: Photon index, Γ, (left) and blackbody temperature, kT , (right) vs. magnetic field, B. The correlation coefficient, r, and associated null- hypothesis probability, p, are shown in the upper right of each plot. The open circle represents SGR 1627−41, which was excluded from the calcula- tion of r due to its large uncertainty in Γ. high-B radio pulsars. The observed short-lived magnetar activity from the rotation-powered pulsar PSR J1846−0258 supports this idea (Gavriil et al., 2008), as does apparently enhanced thermal X-ray emission from high-B radio pulsars compared with that from lower-B radio pulsars of comparable age (Kaspi & McLaughlin, 2005; Olausen et al., 2010, 2013; Zhu et al., 2011). Figure 5–10 also makes clearer that XINS spin properties do not fully overlap with those of magnetars; the former have smaller spin-down rates, hence smaller inferred B. These objects are thus evidence for torque decay in high-B neutron stars and suggest that XINS could be descendants of magnetars as mentioned above. 5.3.3 X-ray Properties

Figure 5–11 plots photon index, Γ, and blackbody temperature, kT , versus spin-inferred magnetic field, B, for those sources that have a power-law or blackbody component in their quiescent X-ray spectrum (see Table 5–3). The left graph shows evidence of a trend where Γ decreases

107 as B increases, previously identified in Kaspi & Boydstun(2010) and in a different but analogous form by Enoto et al.(2010a). Following the example of Kaspi & Boydstun, we attempt to quantify the trend by calculating Pearson’s correlation coefficient, finding r = −0.79 (upper limits were included in the calculation of r by assuming a value of half of the upper limit). For a sample size of N = 12, this result gives a (two- tailed) probability for the null hypothesis of p = 0.0020, similar to the result obtained by Kaspi & Boydstun and significant at better than the 3σ level. Conversely, examination of the plot on the right for evidence of a correlation between kT and B revealed none; in particular, we obtained r = 0.46 for N = 16, giving p = 0.076, which does not exclude the null hypothesis. Overall, these results support the “twisted magnetosphere” model of Thompson et al.(2002), further developed by Beloborodov(2009), which predicts that a higher B field drives stronger currents in the star’s magnetosphere which in turn produces brighter and harder non-thermal X-ray emission.

In Figure 5–12, we plot LX, the quiescent X-ray luminosity in the 2–10 keV energy band, against Γ and kT for the same sources as above. We again calculate the correlation coefficient, r, but in both cases we derive a null-hypothesis probability of 0.01–0.03, not low enough to comfortably reject. Certainly a correlation between LX and kT is not evident; notice how the luminosity spans five orders of magnitude at kT ≈ 0.3 keV. Likewise,

LX spans more than two orders of magnitude at Γ ≈ 3.8. On the other hand, there does appear to be an excluded region in the LX versus Γ graph, where one would find lower-luminosity sources with hard power laws (though given the large uncertainty in Γ, SGR 1627−41 cannot be excluded from encroaching into this region). This cannot simply be due to a selection

108 r = −0.70 r = 0.53 p = 0.011 1035 p = 0.035 1035

1034 ) ) 1 1 - - 33 s s

10 34 g 10 g r r e e ( (

X X 32

L L 10

1031 1033

1030 1.5 2 2.5 3 3.5 4 0.1 0.2 0.3 0.4 0.5 0.6 Photon Index, G kT (keV)

Figure 5–12: Quiescent 2–10 keV X-ray luminosity, LX, vs. Γ (left) and kT (right). The correlation coefficient, r, and null-hypothesis probability, p, are shown in the upper right or left of each plot and the open circle is the same as in Figure 5–11. effect, because given the same luminosity a harder source will produce less flux at energies prone to Galactic absorption than a softer one and should therefore be easier to detect. As indicated above, a harder spectrum is associated with greater X-ray luminosity in the twisted magnetosphere model, so such a gap is consistent with that. However, the model also implies that we should not expect to see high-luminosity sources with soft power laws. We do note that a calculation of r excluding the upper- rightmost point (4U 0142+61) drops the probability of the null hypothesis below 1% (r = −0.82 for N = 11, p = 0.0020), though there is no compelling reason to ignore or discard it. In the leftmost panel of Figure 5–13, we show the quiescent 2–10 keV luminosity LX as a function of B. This plot is an update of Figure 4 from An et al.(2012), though, rather than assuming 50% uncertainties as in that paper, we calculate error bars for B by propagating the uncertainties on P ˙ and P in Table 5–2 and do not estimate error bars for LX. The solid and

109 r = 0.62 r = 0.16 r = −0.37 35 10 p = 0.0045 p = 0.51 p = 0.12

1034 ) 1

- 33

s 10

g r e (

X 32 L 10

1031 B4.4 B2 1030 1013 1014 1015 1029 1030 1031 1032 1033 1034 1035 102 103 104 105 106 107 Magnetic Field (G) Spin-down Luminosity (erg s-1) Characteristic Age (yr)

Figure 5–13: Left panel: quiescent 2–10 keV X-ray luminosity LX vs. B for the magnetars (solid and open circles) and select high-B radio pulsars (open diamonds). Data for the radio pulsars was taken from Table 3 in An et al. (2012). The solid and dashed lines show fits to the data for the relations 4.4 2 ˙ LX ∝ B and LX ∝ B , respectively. Middle panel: LX vs. E. The dotted ˙ line marks LX/E = 1. Right panel: LX vs. τc. All panels: the open circles mark SGRs 0418+5729 and 1806−20. Because these two magnetars lie at opposite corners of each graph, they were excluded from the calculation of the correlation coefficient, r, shown together with the null-hypothesis prob- ability, p, in the upper left or right of each plot, to ensure that a correlation did not depend on their presence. open circles denote the magnetars and the open diamonds represent the five high-B radio pulsars also considered by An et al. A possible correlation can be seen in the data, therefore, as above, we investigated it by calculating Pearson’s correlation coefficient and found that it strongly supports the existence of such a correlation (r = 0.74 for N = 21, p = 1.4 × 10−4). We noticed, however, that there were points in the upper right and lower left corners of the graph, marked by the open circles, that could have had a significant impact on the calculation of r. Removal of these extreme points, SGRs 0418+5729 and 1806−20, still resulted in rejection of the null hypothesis (r = 0.62 for N = 19, p = 0.0045). Furthermore, as in An et al., the inclusion of high-B radio pulsars only strengthened the relation (r = 0.73 for N = 24, p = 5 × 10−5). Therefore, it appears that there could be a genuine correlation between LX and B in high-magnetic-field neutron

110 stars. There are two other magnetars, 4U 0142+61 and 1E 2259+586, that stand out in the plot with unusually high luminosities given their lower magnetic fields. This may suggest that their magnetic fields have strong nondipolar components, not seen in the spin-inferred field, B, that would bring the total field strength in line with the other magnetars of similar LX. Overall, however, these results support the idea that there is a continuum in the X-ray luminosities of high-B radio pulsars and magnetars (see An et al., 2012 for further discussion) as expected on physical grounds based on magnetic dissipation and expected magnetothermal evolution (Thompson & Duncan, 1996; Pons et al., 2009).

In the middle panel of Figure 5–13, we show a plot of LX versus spin- down luminosity, E˙ . The panel shows little more than a scatter plot, as borne out by the correlation coefficient (r = 0.41 for N = 21, p = 0.064; r = 0.16, p = 0.51 with SGRs 0418+5729 and 1806−20 removed). This result is expected in the magnetar model, since the X-ray emission is not powered by the rotational energy. The rightmost panel of Figure 5–13 presents LX versus characteristic age, τc, and like the previous graph there is no visual sign of a strong trend. Naively calculating the correlation coefficient, however, does show evidence for a relation (r = −0.58 for N = 21, p = 0.0057), but it is carried entirely by SGRs 0418+5729 and 1806−20 (r = −0.37, p = 0.12 with those two points removed). Again, this is unsurprising because not only are the characteristic ages of magnetars not necessarily good measures of their true ages as discussed above, but the 2–10 keV luminosity is dominated by the non-thermal emission so we do not expect to detect a cooling trend anyway. Finally, we refer the reader back to Figure 3–5 from Chapter3, the plot of kT versus τc for magnetars, XINSs, and select radio pulsars. It shows

111 10−9 X-ray 1806 Radio 1015 G 3 yr Hard X-ray 10

−10 10 1900 1714 1547 0526 1841 1048 1627 1622 0100 1708 1745 10−11 )

1 1834

- 1810 14 s 10 0501

G s

( 1833 5

e 10 yr

v 0142 i t

a −12 4 13 v .4×1 i 10 0 G r e

D 2259 1647 d o i r e P 10−13 1013 G 7 yr 1822 10

10−14

0418

10−15 2 5 10 Period (s)

Figure 5–14: P –P˙ diagram showing radio pulsars (crosses), XINSs (aster- isks), and magnetars (circles). Radio-detected magnetars are marked with red squares. Blue circles denote magnetars that have been detected in the hard X-ray band (>10 keV), with a dotted circle indicating that it has been so detected only in outburst. a general trend for higher-B sources — of course the magnetars, but also high-B radio pulsars — to display greater blackbody temperatures than low-B pulsars of similar age, suggesting that the magnetic field plays a role in the observed thermal properties of pulsars. For more details, see the discussion in Section 3.3, though we also note here that SGR 0418+5729, despite having B < 1013 G, is set apart from the other low-B sources by its much greater kT . 5.3.4 Multiwavelength Properties

Figure 5–14 shows a P –P˙ diagram with radio pulsars, XINSs, and the magnetars indicated, as well as their detection status in soft X-rays,

112 hard X-rays, and the radio band. From the plot, it is clear that sources detected persistently in hard X-rays tend to be those with the highest B fields (1014.5–1015 G and above) unless they are particularly distant, e.g., in the Magellanic Clouds. 4U 0142+61 and 1E 2259+586 are detected in hard X-rays but have somewhat lower B fields; this further emphasises their apparently outlier nature (noted above). Alternatively, from Figure 5–15, where the multiwavelength detections of the catalogued magnetars are shown as a function of quiescent 2–10 keV X-ray flux FX, it is clear that only the sources with the highest FX are detected persistently in hard X-rays. Moreover, hard X-rays are generally detected in sources in outburst, i.e., when the soft X-ray flux is anomalously high. These facts suggest that all magnetars produce hard X-rays but that current hard X-ray missions do not have the sensitivity to detect them. NASA’s NuSTAR mission (Harrison et al., 2013), the first focusing hard X-ray telescope, may help in this regard. The radio emission observed from magnetars is strikingly different from the hard X-ray behaviour. As is clear from Figure 5–15, radio emission has only been seen in sources with low FX when in outburst in spite of extensive radio observations and stringent upper limits (see Table 5–5) of most of the objects catalogued including those with the largest values of FX (Burgay et al., 2006b; Crawford et al., 2007; Lazarus et al., 2012; Tong et al., 2013; Archibald et al., 2013).5 Although beaming may play a role (see discussion in Lazarus et al., 2012), with increasing statistics, the segregation of radio detections in the P –P˙ diagram (Figure 5–14) is interesting.

5 We note the claimed radio detections of 4U 0142+61, 1E 2259+586, and XTE J1810−197 (Malofeev et al., 2012); however, the detections have not yet been confirmed using another observatory.

113 2013 X-ray 1845 Radio 1818 Hard X-ray 1808 1801 2259 1900 1841 1834 1833 1822 1810 1806 1745 1714 1708 1647 1627 1622 1547 1048 0526 0501 0418 0142 0100

10−15 10−14 10−13 10−12 10−11 10−10 Quiescent 2-10 keV X-ray Flux (erg s-1 cm-2)

Figure 5–15: Magnetar detections as a function of quiescent 2–10 keV X-ray flux FX. Radio and hard X-ray detections are marked by red squares and blue circles, as described for Figure 5–14.

114 Rea et al.(2012a) have suggested that there exists a ‘fundamental plane’ in magnetar spin and radiative phase space that distinguishes sources of different radio emission properties. Specifically, they argue on physical ˙ grounds that magnetars with high E and low LX should be radio bright, ˙ while low E, high LX should not be radio detected. This is, in principle, an explanation for the striking asymmetry in the P –P˙ distribution we see for radio-emitting sources. On the other hand, the recent non-detection of ˙ magnetar Swift J1834.9−0846, which has E/LX where Rea et al.(2012a) would predict radio emission, argues against this picture (Tong et al., 2013; Esposito et al., 2013). Moreover, the ‘fundamental plane’ picture also predicts radio emission from the high-B rotation-powered pulsar PSR J1846−0258, which in fact has been shown to be radio quiet (Archibald et al., 2008). Rea et al.(2012a) argue that a previously reported very large distance (21 kpc) to the source (Becker & Helfand, 1984) together with its presence in supernova remnant Kes 75 could somehow hinder a radio detection, perhaps because of a high DM. However, up-to-date distance estimates for this pulsar (Leahy & Tian, 2008a; Su et al., 2009) place it significantly closer (5.1–10.9 kpc), and the system’s hydrogen column

22 −2 density as measured with X-ray observations, NH = 2–4 × 10 cm predicts, on the basis of an empirical DM versus NH law (He et al., 2013), DM ' 600–1300 pc cm−3. This is well within the range of observed DMs for radio pulsars, particularly those with only moderately fast rotation periods like the 0.326 s period of PSR J1846−0258. Hence, we disagree with the conclusion of Rea et al.(2012a) that a radio detection of PSR J1846 −0258 is difficult due to its environment. On the other hand, unfortunate radio beaming, as well as the episodic nature of radio emission from magnetars, may play a role for this pulsar and Swift J1834.9−0846. Continued radio

115 observations of magnetars in outburst to increase statistics for radio ˙ emission in the population will be helpful for deciding whether E/LX plays a meaningful role in radio detectability of magnetars. 5.4 Conclusions

We have compiled the first catalog of all currently known magnetars, including 21 confirmed sources and 5 candidates. Where available from the literature, we have provided spatial properties (coordinates, proper motion, distance, and proposed associations), timing data (period, period derivative, and derived parameters), spectral parameters for the quiescent soft X-ray emission, and observed properties or upper limits in the radio, infrared, optical, hard X-ray, and gamma-ray bands. We note that the known magnetar population is relatively free from selection for location in the Galaxy thanks to the all-sky X-ray monitors that have found so many of these objects in recent years. We constructed histograms in Galactic longitude and latitude, spin period, P , spin-inferred magnetic field, B, spin- ˙ down luminosity, E, and characteristic age, τc, to compare the magnetar distributions with the distributions of the known pulsar population. We measure the scale height of magnetars for the first time and find it to be smaller than that of OB stars, supporting the hypothesis that the most massive O stars are magnetar progenitors. We note the relatively narrow ranges in P and B observed for the magnetars, which stand in contrast ˙ to the far wider ranges in E and τc. The fact that the characteristic age range is so broad, in spite of so small a scale height for these objects, argues that the former is generally a poor age indicator. We confirm correlations between Γ and B, previously identified by Kaspi & Boydstun(2010) and

Enoto et al.(2010a), and LX and B, previously noted by An et al.(2012), and observe an excluded region in the plot of LX versus Γ. Finally, we find

116 that detections of magnetars in the hard X-ray band seem to be correlated with soft X-ray flux and B, while radio detections show, if anything, the opposite trend. A regularly maintained online version of the catalog has been made available, with one main table focused on the timing and X- ray data, and two additional tables for alternative values and detailed records of optical and near-infrared observations. We plan to maintain this with regular updates as new magnetar results appear and encourage the community to provide feedback and suggestions for improvement on this constantly evolving initiative.

117 CHAPTER 6 Conclusions 6.1 High-B Pulsars

As I introduced in Section 1.3, an important challenge for the magnetar picture is to determine their connection with the high-magnetic-field rotation-powered pulsars. The recently developed models of magnetothermal evolution (Vigano et al., 2013), which propose that the behaviour and evolution of a neutron star are greatly determined by the strength of its magnetic field at birth, hold the promise of unifying the seemingly disparate families of isolated pulsars (Kaspi, 2010). If such an idea of unification is correct, one would expect to find a continuum of magnetar-like behaviour, particularly among neutron stars with ∼1013–1014 G fields. Indeed, there does exist evidence of this, such as the low-field (B = 1.4 × 1013 G) magnetar Swift J1822.3−1606 (Scholz et al., 2014) or the magnetar-like outburst of the rotation-powered PSR J1846−0258 (Gavriil et al., 2008). In Chapter3, I report on the first X-ray observations of one such high-B pulsar, PSR J1734−3333 (Olausen et al., 2010, 2013). This pulsar is notable because its unusually low braking index (n = 0.9 ± 0.2; Espinoza et al., 2011a) may indicate that its magnetic field is increasing, and that it may therefore be evolving into a magnetar. Although I was able to successfully detect the neutron star, I did not detect X-ray pulsations from it and set a 1σ upper limit of 0.6 on the pulsed fraction in the 0.5– 3 keV energy range. I found that the pulsar has a thermal spectrum with a blackbody temperature of 0.30 ± 0.06 keV, implying a bolometric luminosity

+2.2 32 −1 of Lbb = 2.0−0.7 × 10 erg s for a distance of 6.1 kpc, and that fits to

118 neutron star atmosphere models gave temperatures of 0.14 ± 0.05 keV (nsa)

+0.05 and 0.13−0.04 keV (nsmax). The blackbody temperature is significantly higher than predicted for standard neutron star cooling models, and so are the best-fit temperatures given by the neutron star atmosphere models, though within the uncertainties they are consistent with cooling models. I compiled a list of X-ray luminosities and upper limits for all high-B pulsars that have been observed in X-rays and find evidence for a correlation between that quantity and the magnetic field, with the caveat that age may bear some or all of the responsibility for the observed pattern. I also produced an updated version of the blackbody temperature versus characteristic age plot from Zhu et al.(2011), adding in the magnetars as well as new or updated results for several high-B pulsars. The updated plot continues to support the finding of Zhu et al.(2011) that neutron stars with high magnetic fields tend to be hotter than those of similar age but with lower magnetic fields. This result, in turn, suggests that a sufficiently strong magnetic field can delay the cooling of a neutron star, either by passively altering the thermal conductivity of the crust (Geppert et al., 2006; Page et al., 2007), or by heating it via active field decay as predicted in magnetothermal evolution. Ultimately, deep X-ray observations of more high-B pulsars and monitoring of sources considered likely to exhibit magnetar-like activity are recommended to investigate all of these connections further. 6.2 Magnetar Winds

In Section 1.1.3, I introduced the pulsar wind nebula (PWN), a phenomenon produced from synchrotron emission radiated by relativistic electrons and positrons in the particle wind generated by a pulsar. Although commonly seen around young, energetic rotation-powered pulsars, it is

119 currently unknown whether magnetars produce similar particle outflows that are capable of generating PWNe. Indeed, given the large reservoirs of magnetic energy available to magnetars, the presence of magnetically powered particle outflows has already been proposed (Thompson & Blaes, 1998; Harding et al., 1999), making plausible the existence of wind nebulae produced by these outflows. The first claim of a wind nebula surrounding a magnetar was made by Murakami et al.(1994) regarding SGR 1806 −20, but this was later refuted as being associated with an unrelated nearby star by Hurley et al. (1999). More recently, Vink & Bamba(2009) claimed to detect extended emission around the magnetar 1E 1547.0−5408 and characterised it as being a rotation-powered PWN and, farther out, emission from the supernova remnant (SNR) G327.24−0.13. Chapter4 reports on my analysis of four XMM-Newton observations of this source (Olausen et al., 2011) that was undertaken to investigate these claims. The flux of the magnetar varied significantly among the observations, and I found that so did the flux of the extended emission. In fact, the two were tightly correlated, a property that is expected for emission from a dust-scattering halo but not a PWN. I concluded that the extended emission is dominated by dust scattering and, though I confirmed the existence of SNR emission farther away from the magnetar as seen by Vink & Bamba(2009), I found no strong evidence for a PWN. I calculated a 3σ upper limit on the 2–10 keV luminosity of a possible PWN of 9 × 1031 erg s−1 for a distance of 4 kpc, three times lower than that of the previously claimed detection. Since the publication of Olausen et al.(2011), there has been one more claimed detection of a wind nebula around a magnetar. Younes et al. (2012) identified a region of extended emission around the magnetar Swift

120 J1834.9−0846 that, unlike the previously identified dust-scattering halo (Kargaltsev et al., 2012), was asymmetric and present in an observation of the source in quiescence. They characterised the emission, which was significantly brighter than expected for a rotation-powered PWN, as from a magnetically powered “magnetar wind nebula.” This interpretation was challenged by Esposito et al.(2013), who contended that the nebula could be better explained by dust scattering. As a result, there remains no unambiguous proof of the existence of magnetar wind nebulae, and therefore relativistic particle outflows from magnetars. Further deep observations of magnetars such as 1E 1547.0−5408 in quiescence are recommended in order to search for possible PWNe and set constraining upper limits on their luminosities. Additionally, there is one high-B pulsar, PSR J1819−1458, surrounded by an X-ray nebula that, being brighter than expected for a rotation-powered PWN, has been suggested to be magnetically powered (Rea et al., 2009a; Camero-Arranz et al., 2013). This, too, warrants further study. 6.3 The Magnetar Population

In Section 1.2.1, I discussed the history of the study of magnetars and their original classifications as SGRs and AXPs. SGRs were first discovered in 1979 with the detection of a giant flare and a few repeated bursts, but for much of the next two decades, only three of these sources had been seen, with a fourth appearing in 1998. Likewise, the group of pulsars that came to be called AXPs in the early to mid-90s contained only 3–4 known sources until the end of that decade. In the past 15 years, however, there has been incredible progress in the field. Already predicted by the magnetar model, the observation of bursts from AXPs and long periods of quiescence from

121 SGRs unified the two groups into one, and their ranks have swollen to over 20 confirmed members. Because the number of identified magnetars has grown so much, it was felt that a systematic compilation of these sources and their properties was needed. Therefore, I have put together in Chapter5 the first catalog of all 26 currently known magnetars (but see below), of which 21 are confirmed and 5 are candidates (Olausen & Kaspi, 2014). I gathered results from the literature about their spatial and timing properties, the spectral parameters of their quiescent soft X-ray emission, and their observed properties in the radio, infrared, optical, hard X-ray, and gamma ray bands. This information was collected into seven tables in this thesis and into the three tables that comprise the online version of the catalog. I also made histograms of Galactic longitude and latitude, period, and the inferred timing parameters ˙ B, E, and τc in order to present these quantities in relation to the rest of the pulsar population. I measured, for the first time, the scale height of the magnetars and found it to be in the range ∼20–31 pc assuming they are exponentially distributed. This value is smaller than the scale height of OB stars and, more specifically, smaller than the distance away from the plane that less massive B stars are expected to travel during their lives, thus supporting the hypothesis that magnetars have massive progenitors (Figer et al., 2005; Muno et al., 2006). I also confirmed previously reported correlations between the X-ray power law and magnetic field (Kaspi & Boydstun, 2010; Enoto et al., 2010a), and between the X-ray luminosity and magnetic field An et al.(2012). Finally, I observed that a high quiescent soft X-ray flux seemed to be correlated with detecting a magnetar in the hard X-ray, but that it was, if anything, anti-correlated with detecting it in the radio.

122 Not long after my paper was first submitted to the Astrophysical Journal, a new magnetar candidate, 3XMM J185246.6+003317, was discovered in archival XMM data (Zhou et al., 2014; Rea et al., 2014). This discovery came out too late to add to the paper, and although I was willing to update a few results for this thesis, there was insufficient time to fully include its properties and implications. This means that the paper and thesis are already out of date (not surprising in this very active field) in that there are actually 27 currently known confirmed and candidate magnetars. The online version of the catalog, however, has been updated to include this new source. While writing the manuscript, there was also some discussion regarding whether PSR J1846−0258 should be included in the catalog in some capacity. On the one hand, the source was seen to undergo a magnetar-like outburst in 2006 (Gavriil et al., 2008), but on the other hand, except for those few weeks, its properties and behaviour during the 14 years over which it has been monitored have been typical of a young, rotation-powered pulsar. In the end, I did not include it, but since then it has also been added to the online catalog, albeit set apart to note its nature as a transition object between the rotation-powered pulsars and the magnetars. Magnetars are complicated and ever-surprising objects, and there is much about them that is not covered in this catalog. These include the properties of their bursts and flares, their X-ray spectra and flux during outburst and the way the flux decays post-outburst, and the wander and anomalies, including glitches, seen in their timing histories (for a list of magnetar glitches, see Dib & Kaspi, 2014). Notably, the first anti-glitch ever detected in a neutron star was seen in the magnetar 1E 2259+586 (Archibald et al., 2013). The tables in this catalog also do not capture the

123 variability seen in their optical and radio counterparts, nor do they touch upon the radio spectra or pulse profile shapes that, for most radio-detected magnetars, are so unusual compared to standard radio pulsars. There is still much that can be done, even simply in the realm of compiling and analyzing properties of magnetars from the existing literature. 6.4 Concluding Remarks

The past decade has been an exciting time for studying magnetars and other high-B pulsars. There have been 12 new magnetars discovered in that time (see Figure 5–1), and a nominally rotation-powered pulsar was even seen to briefly turn into a magnetar. This rate of discovery is more than triple that of the previous 25 years and has more than doubled the number of confirmed magnetars. Many of these new sources were found by the all-sky monitors on board the Swift and Fermi satellites, which have been a boon to this area of research. They have also facilitated the discovery of more transient magnetars, including the low-field ones and the Galactic centre one, which in quiescence often appear unremarkable if they are even detected at all. Still, the rate of discovery is low compared to that of ordinary pulsars, and we have likely already seen the most dramatic increase in the known population. Assuming the current rate holds up, it will take another two decades to once again double the number of known magnetars. Meanwhile, unprecedented events like the outburst of PSR J1846−0258 or the anti-glitch of 1E 2259+586 would not have been identified were they not the subject of monitoring campaigns with RXTE or Swift. Therefore, although detecting new sources with Swift, Fermi, and perhaps the forthcoming AstroSAT mission will still be very useful, continued monitoring of known magnetars will be just as or more

124 important in furthering our understanding and discovering new facets of these remarkable objects.

125 References Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010a, ApJL, 725, L73 —. 2010b, ApJS, 187, 460 Aguilera, D. N., Pons, J. A., & Miralles, J. A. 2008, ApJ, 673, L167 An, H., Kaspi, V. M., Archibald, R., & Cumming, A. 2013a, ApJ, 763, 82 An, H., Kaspi, V. M., Tomsick, J. A., et al. 2012, ApJ, 757, 68 An, H., Hascoet, R., Kaspi, V. M., et al. 2013b, ApJ, 779, 163 Anderson, G. E., Gaensler, B. M., Slane, P. O., et al. 2012, ApJ, 751, 53 Archibald, A. M., Kaspi, V. M., Livingstone, M. A., & McLaughlin, M. A. 2008, ApJ, 688, 550 Archibald, R. F., Kaspi, V. M., Ng, C.-Y., et al. 2013, Natur, 497, 591 Arnaud, K., Smith, R., & Siemiginowska, A., eds. 2011, Handbook of X-ray Astronomy, Cambridge Observing Handbook for Research Astronomy (Cambridge: Cambridge University Press) Baade, W., & Zwicky, F. 1934, PNAS, 20, 254 Bahcall, J. N. 1984, ApJ, 276, 156 Balman, S., Safi-Harb, S., Ibrahim, A. I., Swank, J. H., & Markwardt, C. B. 2003, ATel, 195 Baring, M. G., & Harding, A. K. 1998, ApJL, 507, L55 Barthelmy, S. D., Barbier, L. M., Cummings, J. R., et al. 2005, SSRv, 120, 143 Becker, R. H., & Helfand, D. J. 1984, ApJ, 283, 154 Beloborodov, A. M. 2009, ApJ, 703, 1044 Bernardini, F., Israel, G., Stella, L., et al. 2011, A&A, 529, 19

126 Bibby, J. L., Crowther, P. A., Furness, J. P., & Clark, J. S. 2008, MNRAS, 386, L23 Bloom, J. S. 2005, GCN, 4042 Burgay, M., Rea, N., Israel, G., & Possenti, A. 2006a, ATel, 903 Burgay, M., Rea, N., Israel, G., et al. 2006b, MNRAS, 372, 410 Burke, B. E., Mountain, R. W., Daniels, P. J., Cooper, M. J., & Dolat, V. S. 1994, ITNS, 41, 375 Burrows, D. N., Hill, J. E., Nousek, J. A., et al. 2005, SSRv, 120, 165 Camero, A., Papitto, A., Rea, N., et al. 2014, MNRAS, 438, 3291 Camero-Arranz, A., Rea, N., Bucciantini, N., et al. 2013, MNRAS, 429, 2493 Cameron, P. B., Chandra, P., Ray, A., et al. 2005, Natur, 434, 1112 Camilo, F., Ransom, S. M., Halpern, J. P., & Reynolds, J. E. 2007a, ApJL, 666, L93 Camilo, F., Ransom, S. M., Halpern, J. P., et al. 2006, Natur, 442, 892 Camilo, F., Reynolds, J. E., Johnston, S., Halpern, J. P., & Ransom, S. M. 2008, ApJ, 679, 681 Camilo, F., Cognard, I., Ransom, S. M., et al. 2007b, ApJ, 663, 497 Camilo, F., Ransom, S. M., Pe˜nalver, J., et al. 2007c, ApJ, 669, 561 Caswell, J. L., McClure-Griffiths, N. M., & Cheung, M. C. M. 2004, MN- RAS, 352, 1405 Chang, P., Arras, P., & Bildsten, L. 2004, ApJ, 616, L147 Cheng, A. F., & Helfand, D. J. 1983, ApJ, 271, 271 Cheng, K. S., Ho, C., & Ruderman, M. 1986, ApJ, 300, 500 Chiu, H.-Y., & Salpeter, E. E. 1964, PhRvL, 12, 413 Cline, T. L., Frederiks, D. D., Golenetskii, S., et al. 2000, ApJ, 531, 407 Cline, T. L., Desai, U. D., Teegarden, B. J., et al. 1982, ApJL, 255, L45 Colpi, M., Geppert, U., & Page, D. 2000, ApJL, 529, L29

127 Conti, P. S., & Vacca, W. D. 1990, AJ, 100, 431 Corbel, S., Chapuis, C., Dame, T. M., & Durouchoux, P. 1999, ApJL, 526, L29 Corbel, S., & Eikenberry, S. S. 2004, A&A, 419, 191 Cordes, J. M., & Lazio, T. J. W. 2002, arXiv, arXiv:astro-ph/0207156 Crawford, F., Hessels, J. W. T., & Kaspi, V. M. 2007, ApJ, 662, 1183 Daugherty, J. K., & Harding, A. K. 1982, ApJ, 252, 337 —. 1996, ApJ, 458, 278 Davies, B., Figer, D. F., Kudritzki, R.-P., et al. 2009, ApJ, 707, 844 de Ugarte Postigo, A., Castro-Tirado, A. J., Covino, S., et al. 2009, A&A, 500, 1157 Deller, A. T., Camilo, F., Reynolds, J. E., & Halpern, J. P. 2012, ApJL, 748, L1 den Hartog, P. R., Kuiper, L., & Hermsen, W. 2008a, A&A, 489, 263 den Hartog, P. R., Kuiper, L., Hermsen, W., et al. 2008b, A&A, 489, 245 —. 2007, Ap&SS, 308, 647 den Herder, J. W., Brinkman, A. C., Kahn, S. M., et al. 2001, A&A, 365, L7 Dhillon, V. S., Marsh, T. R., Hulleman, F., et al. 2005, MNRAS, 363, 609 Dhillon, V. S., Marsh, T. R., Littlefair, S. P., et al. 2009, MNRAS, 394, L112 —. 2011, MNRAS, 416, L16 Dib, R., & Kaspi, V. M. 2014, ApJ, 784, 37 Dib, R., Kaspi, V. M., & Gavriil, F. P. 2009, ApJ, 702, 614 Dib, R., Kaspi, V. M., Scholz, P., & Gavriil, F. P. 2012, ApJ, 748, 3 Draine, B. T. 2003, ApJ, 598, 1026 Duncan, R. C., & Thompson, C. 1992, ApJL, 392, L9

128 Duncan, R. C., & Thompson, C. 1996, in AIP Conference Proceedings, Vol. 366, High Velocity Neutron Stars and Gamma-ray Bursts, ed. R. E. Rothschild & R. E. Lingenfelter (Melville, NY: AIP), 111 Durant, M. 2005, ApJ, 632, 563 Durant, M., Kargaltsev, O., & Pavlov, G. G. 2011, ApJ, 742, 77 Durant, M., & van Kerkwijk, M. H. 2005a, ApJL, 628, L135 —. 2005b, ApJ, 627, 376 —. 2006a, ApJ, 650, 1070 —. 2006b, ApJ, 652, 576 —. 2006c, ApJ, 648, 534 —. 2008, ApJ, 680, 1394 Eatough, R. P., Falcke, H., Karuppusamy, R., et al. 2013, Natur, 501, 391 Elias, F., Cabrera-Ca˜no,J., & Alfaro, E. J. 2006, AJ, 131, 2700 Enoto, T., Makishima, K., Nakazawa, K., et al. 2011, PASJ, 63, 387 Enoto, T., Nakazawa, K., Makishima, K., et al. 2010a, ApJL, 722, L162 —. 2010b, PASJ, 62, 475 Enoto, T., Rea, N., Nakagawa, Y. E., et al. 2010c, ApJ, 715, 665 Ertan, U.,¨ Alpar, M. A., Erkut, M. H., Ek¸si,K. Y., & C¸alı¸skan, S¸. 2007, Ap&SS, 308, 73 Ertan, U.,¨ Ek¸si,K. Y., Erkut, M. H., & Alpar, M. A. 2009, ApJ, 702, 1309 Espinoza, C. M., Lyne, A. G., Kramer, M., Manchester, R. N., & Kaspi, V. M. 2011a, ApJL, 741, L13 Espinoza, C. M., Lyne, A. G., Stappers, B. W., & Kramer, M. 2011b, MNRAS, 414, 1679 Esposito, P., Mereghetti, S., Tiengo, A., et al. 2007, A&A, 476, 321 Esposito, P., Israel, G., Zane, S., et al. 2008, MNRAS, 390, L34 Esposito, P., Burgay, M., Possenti, A., et al. 2009a, MNRAS, 399, L44

129 Esposito, P., Tiengo, A., Mereghetti, S., et al. 2009b, ApJL, 690, L105 Esposito, P., Israel, G., Turolla, R., et al. 2010, MNRAS, 405, 1787 —. 2011, MNRAS, 416, 205 Esposito, P., Tiengo, A., Rea, N., et al. 2013, MNRAS, 429, 3123 Fahlman, G. G., & Gregory, P. C. 1981, Natur, 293, 202 Fatkhullin, T., de Ugarte Postigo, A., Castro-Tirado, A. J., et al. 2008, GCN, 8160 Faucher-Gigu`ere,C.-A., & Kaspi, V. M. 2006, ApJ, 643, 332 Fender, R. P., Muxlow, T. W. B., Garrett, M. A., et al. 2006, MNRAS, 367, L6 Ferrario, L., & Wickramasinghe, D. 2006, MNRAS, 367, 1323 Figer, D. F., Najarro, F., Geballe, T. R., Blum, R. D., & Kudritzki, R. P. 2005, ApJL, 622, L49 Forman, W., Jones, C., Cominsky, L., et al. 1978, ApJ, 38, 357 Frail, D. A., Kulkarni, S. R., & Bloom, J. S. 1999, Natur, 398, 127 Fuchs, Y., Mirabel, F., Chaty, S., et al. 1999, A&A, 350, 891 Gaensler, B. M., & Chatterjee, S. 2008, GCN, 8149 Gaensler, B. M., Gotthelf, E. V., & Vasisht, G. 1999, ApJL, 526, L37 Gaensler, B. M., McClure-Griffiths, N. M., Oey, M. S., et al. 2005a, ApJL, 620, L95 Gaensler, B. M., & Slane, P. O. 2006, ARA&A, 44, 17 Gaensler, B. M., Kouveliotou, C., Gelfand, J. D., et al. 2005b, Natur, 434, 1104 Gavriil, F. P., Gonzalez, M. E., Gotthelf, E. V., et al. 2008, Sci, 319, 1802 Gavriil, F. P., & Kaspi, V. M. 2004, ApJL, 609, L67 Gavriil, F. P., Kaspi, V. M., & Woods, P. M. 2002, Natur, 419, 142 —. 2004, ApJ, 607, 959

130 Gelfand, J. D. 2007, Ap&SS, 308, 39 Gelfand, J. D., & Gaensler, B. M. 2007, ApJ, 667, 1111 Gelfand, J. D., Lyubarsky, Y. E., Eichler, D., et al. 2005, ApJL, 634, L89 Georgakakis, A., Georgantopoulos, I., Vallb´e,M., et al. 2004, MNRAS, 349, 135 Geppert, U., K¨uker, M., & Page, D. 2004, A&A, 426, 267 —. 2006, A&A, 457, 937 Giacconi, R., Gursky, H., Paolini, F. R., & Rossi, B. B. 1962, PhRvL, 9, 439 Giacconi, R., Branduardi, G., Briel, U., et al. 1979, ApJ, 230, 540 Gies, D. R. 1987, ApJS, 64, 545 G¨o˘g¨u¸s,E., Woods, P. M., Kouveliotou, C., et al. 2010a, ApJ, 722, 899 G¨o˘g¨u¸s,E., Cusumano, G., Levan, A. J., et al. 2010b, ApJ, 718, 331 Gold, T. 1968, Natur, 218, 731 Goldreich, P., & Julian, W. H. 1969, ApJ, 157, 869 Gonzalez, M. E., Kaspi, V. M., Camilo, F., Gaensler, B. M., & Pivovaroff, M. J. 2005, ApJ, 630, 489 Gonzalez, M. E., Kaspi, V. M., Lyne, A. G., & Pivovaroff, M. J. 2004, ApJL, 610, L37 Gonzalez, M. E., & Safi-Harb, S. 2003, ApJL, 591, L143 Gotthelf, E. V., Halpern, J. P., Buxton, M., & Bailyn, C. 2004, ApJ, 605, 368 G¨otz,D., Mereghetti, S., Tiengo, A., & Esposito, P. 2006, A&A, 449, L31 G¨otz,D., Rea, N., Israel, G., et al. 2007, A&A, 475, 317 Granot, J., Ramirez-Ruiz, E., Taylor, G. B., et al. 2006, ApJ, 638, 391 Guidorzi, C., Mundell, C. G., Gomboc, A., et al. 2005, GCN, 4035 Haberl, F. 2007, Ap&SS, 308, 181 Halpern, J. P. 2008, GCN, 8129

131 Halpern, J. P., & Gotthelf, E. V. 2010a, ApJ, 725, 1384 —. 2010b, ApJ, 710, 941 Halpern, J. P., Gotthelf, E. V., Reynolds, J. E., Ransom, S. M., & Camilo, F. 2008, ApJ, 676, 1178 Harding, A. K., Contopoulos, I., & Kazanas, D. 1999, ApJL, 525, L125 Harding, A. K., & Muslimov, A. G. 2001, ApJ, 556, 987 Harnden, F. R. J., Grant, P. D., Seward, F. D., & Kahn, S. M. 1985, ApJ, 299, 828 Harrison, F. A., Craig, W. W., Christensen, F. E., et al. 2013, ApJ, 770, 103 Haschke, R., Grebel, E. K., & Duffau, S. 2012a, AJ, 144, 106 —. 2012b, AJ, 144, 107 He, C., Ng, C.-Y., & Kaspi, V. M. 2013, ApJ, 768, 64 Helfand, D. J., Chatterjee, S., Brisken, W. F., et al. 2007, ApJ, 662, 1198 Helfand, D. J., Collins, B. F., & Gotthelf, E. V. 2003, ApJ, 582, 783 Hellier, C. 1994, MNRAS, 271, L21 Hessels, J. W. T., Ransom, S. M., Stairs, I. H., et al. 2006, Sci, 311, 1901 Hessels, J. W. T., Rea, N., Ransom, S. M., & Stappers, B. W. 2008, GCN, 8134 Hewish, A., Bell, S. J., Pilkington, J. D. H., Scott, P. F., & Collins, R. A. 1968, Natur, 217, 709 Ho, W. C. G., Potekhin, A. Y., & Chabrier, G. 2008, ApJS, 178, 102 Holland, S. T., & Krimm, H. A. 2008, GCN, 8325 Hulleman, F., Tennant, A. F., van Kerkwijk, M. H., et al. 2001, ApJL, 563, L49 Hulleman, F., van Kerkwijk, M. H., & Kulkarni, S. R. 2004, A&A, 416, 1037 Hurley, K., Kouveliotou, C., Cline, T. L., et al. 1999, ApJL, 523, L37 Israel, G., Campana, S., Dall’Osso, S., et al. 2007, ApJ, 664, 448

132 Israel, G., Covino, S., Stella, L., et al. 1999, ApJL, 518, L107 Israel, G., Mereghetti, S., & Stella, L. 1994, ApJL, 433, L25 Israel, G., Covino, S., Stella, L., et al. 2002, ApJL, 580, L143 Israel, G., Covino, S., Perna, R., et al. 2003, ApJL, 589, L93 Israel, G., Rea, N., Mangano, V., et al. 2004a, ApJL, 603, L97 Israel, G., Stella, L., Covino, S., et al. 2004b, in IAU Symposium, Vol. 218, Young Neutron Stars and Their Environments, ed. F. Camilo & B. M. Gaensler (San Francisco: ASP), 247 Israel, G., Covino, S., Mignani, R. P., et al. 2005, A&A, 438, L1 Israel, G., Rea, N., Rol, E., et al. 2009, ATel, 1909 Israel, G., Esposito, P., Rea, N., et al. 2010, MNRAS, 408, 1387 Jackson, J. D. 1999, Classical Electrodynamics, 3rd edn. (New York: Wiley) Jahoda, K., Markwardt, C. B., Radeva, Y., et al. 2006, ApJS, 163, 401 Jansen, F., Lumb, D., Altieri, B., et al. 2001, A&A, 365, L1 Joshi, Y. C. 2007, MNRAS, 378, 768 Kaneko, Y., G¨o˘g¨u¸s,E., Kouveliotou, C., et al. 2010, ApJ, 710, 1335 Kaplan, D. L., Chakrabarty, D., Wang, Z., & Wachter, S. 2009a, ApJ, 700, 149 Kaplan, D. L., Chatterjee, S., Hales, C. A., Gaensler, B. M., & Slane, P. O. 2009b, AJ, 137, 354 Kaplan, D. L., Esposito, P., Chatterjee, S., et al. 2009c, MNRAS, 400, 1445 Kaplan, D. L., Kulkarni, S. R., van Kerkwijk, M. H., et al. 2001, ApJ, 556, 399 Kaplan, D. L., & van Kerkwijk, M. H. 2009, ApJ, 705, 798 —. 2011, ApJL, 740, L30 Kargaltsev, O., & Pavlov, G. G. 2008, in AIP Conference Proceedings, Vol. 983, 40 Years of Pulsars: Millisecond Pulsars, Magnetars, and More, ed.

133 C. G. Bassa, Z. Wang, A. Cumming, & V. M. Kaspi (Melville, NY: AIP), 171 Kargaltsev, O., Kouveliotou, C., Pavlov, G. G., et al. 2012, ApJ, 748, 26 Kaspi, V. M. 2010, PNAS, 107, 7147 Kaspi, V. M., & Boydstun, K. 2010, ApJL, 710, L115 Kaspi, V. M., Chakrabarty, D., & Steinberger, J. 1999, ApJL, 525, L33 Kaspi, V. M., Gavriil, F. P., Woods, P. M., et al. 2003, ApJL, 588, L93 Kaspi, V. M., & McLaughlin, M. A. 2005, ApJL, 618, L41 Kaspi, V. M., Roberts, M. S. E., & Harding, A. K. 2006, in Compact Stellar X-ray Sources, ed. W. H. G. Lewin & M. van der Klis (Cambridge: Cambridge University Press), 279 Kaspi, V. M., Archibald, R. F., Bhalerao, V., et al. 2014, ApJ, 786, 84 Keith, M. J., Johnston, S., Levin, L., & Bailes, M. 2011, MNRAS, 416, 346 Keller, C. U. 1995, Cell Mol Life Sci, 51, 710 Kelley, R. L., Mitsuda, K., Allen, C. A., et al. 2007, PASJ, 59, 77 Klose, S., Stecklum, B., & Laux, U. 2001, GCN, 1044 Klose, S., Henden, A. A., Geppert, U., et al. 2004, ApJL, 609, L13 Kondratiev, V. I., McLaughlin, M. A., Lorimer, D. R., et al. 2009, ApJ, 702, 692 Kosugi, G., Ogasawara, R., & Terada, H. 2005, ApJL, 623, L125 Kothes, R., & Dougherty, S. M. 2007, A&A, 468, 993 Kothes, R., & Foster, T. 2012, ApJL, 746, L4 Kouveliotou, C., Fishman, G. J., Meegan, C. A., et al. 1993, Natur, 362, 728 Kouveliotou, C., Dieters, S., Strohmayer, T., et al. 1998, Natur, 393, 235 Koyama, K., Tsunemi, H., Dotani, T., et al. 2007, PASJ, 59, 23 Kuiper, L., Hermsen, W., den Hartog, P. R., & Collmar, W. 2006, ApJ, 645, 556

134 Kuiper, L., Hermsen, W., den Hartog, P. R., & Urama, J. O. 2012, ApJ, 748, 133 Kulkarni, S. R., Kaplan, D. L., Marshall, H. L., et al. 2003, ApJ, 585, 948 Kumar, H. S., & Safi-Harb, S. 2008, ApJL, 678, L43 —. 2010, ApJL, 725, L191 Lamb, D., Graziani, C., Shirasaki, Y., et al. 2003, GCN, 2351 Lamb, R. C., Fox, D. W., Macomb, D. J., & Prince, T. A. 2002, ApJL, 574, L29 Lamb, R. C., & Markert, T. H. 1981, ApJ, 244, 94 Large, M. I., Vaughan, A. E., & Mills, B. Y. 1968, Natur, 220, 340 Laros, J. G., Fenimore, E. E., Klebesadel, R. W., et al. 1987, ApJL, 320, L111 Lattimer, J. M., & Prakash, M. 2001, ApJ, 550, 426 —. 2004, Sci, 304, 536 Lattimer, J. M., Prakash, M., Pethick, C. J., & Haensel, P. 1991, PhRvL, 66, 2701 Lazarus, P. 2013, in IAU Symposium, Vol. 291, Neutron Stars and Pul- sars: Challenges and Opportunities after 80 Years, ed. J. van Leeuwen (Cambridge University Press), 35 Lazarus, P., Kaspi, V. M., Champion, D. J., Hessels, J. W. T., & Dib, R. 2012, ApJ, 744, 97 Leahy, D. A., & Tian, W. W. 2007, A&A, 461, 1013 —. 2008a, A&A, 480, L25 —. 2008b, AJ, 135, 167 Levin, L., Bailes, M., Bates, S. D., et al. 2010, ApJL, 721, L33 Leyder, J. C., Walter, R., & Rauw, G. 2008, A&A, 477, L29 Lin, L., Kouveliotou, C., Baring, M. G., et al. 2011, ApJ, 739, 87

135 Livingstone, M. A., Kaspi, V. M., & Gavriil, F. P. 2010, ApJ, 710, 1710 Livingstone, M. A., Kaspi, V. M., Gavriil, F. P., et al. 2007, Ap&SS, 308, 317 Livingstone, M. A., Ng, C.-Y., Kaspi, V. M., Gavriil, F. P., & Gotthelf, E. V. 2011, ApJ, 730, 66 Longair, M. S. 1992, High Energy Astrophysics, 2nd edn. (Cambridge: Cambridge University Press) Lorimer, D., & Kramer, M. 2005, Handbook of Pulsar Astronomy, Cam- bridge Observing Handbook for Research Astronomy (Cambridge: Cambridge University Press) Lumb, D. H., Schartel, N., & Jansen, F. A. 2012, OptEn, 51, 1009 Lyne, A. G., Pritchard, R. S., & Graham-Smith, F. 1993, MNRAS, 265, 1003 Lyne, A. G., Pritchard, R. S., Graham-Smith, F., & Camilo, F. 1996, Natur, 381, 497 Lyutikov, M. 2006, MNRAS, 367, 1594 Ma´ız-Apell´aniz,J. 2001, AJ, 121, 2737 Ma´ızApell´aniz,J., Alfaro, E. J., & Sota, A. 2008, arXiv, arXiv:0804.2553 Malheiro, M., Rueda, J. A., & Ruffini, R. 2012, PASJ, 64, 56 Malofeev, V. M., Teplykh, D. A., & Logvinenko, S. V. 2012, ARep, 56, 35 Manchester, R. N., Hobbs, G. B., Teoh, A., & Hobbs, M. 2005, AJ, 129, 1993 Manchester, R. N., Lyne, A. G., Camilo, F., et al. 2001, MNRAS, 328, 17 Marshall, F. E., & Gelbord, J. M. 2010, GCN, 10540 Mathis, J. S., & Lee, C.-W. 1991, ApJ, 376, 490 Mauche, C. W., & Gorenstein, P. 1986, ApJ, 302, 371 Mazets, E. P., & Golenetskii, S. V. 1981, Ap&SS, 75, 47

136 Mazets, E. P., Golenetskij, S. V., & Guryan, Y. A. 1979a, SvAL, 5, 343 Mazets, E. P., Golentskii, S. V., Il’Inskii, V. N., Aptekar, R. L., & Guryan, I. A. 1979b, Natur, 282, 587 McGarry, M. B., Gaensler, B. M., Ransom, S. M., Kaspi, V. M., & Veljkovik, S. 2005, ApJL, 627, L137 McLaughlin, M. A., Stairs, I. H., Kaspi, V. M., et al. 2003, ApJL, 591, L135 McLaughlin, M. A., Rea, N., Gaensler, B. M., et al. 2007, ApJ, 670, 1307 Meegan, C., Lichti, G., Bhat, N. P., et al. 2009, ApJ, 702, 791 Mereghetti, S. 2008, A&ARv, 15, 225 —. 2013, BrJPh, 43, 356 Mereghetti, S., Esposito, P., Tiengo, A., et al. 2012, A&A, 546, 30 Mereghetti, S., G¨otz,D., Mirabel, I. F., & Hurley, K. 2005, A&A, 433, L9 Mereghetti, S., & Stella, L. 1995, ApJL, 442, L17 Mereghetti, S., Esposito, P., Tiengo, A., et al. 2006, ApJ, 653, 1423 Mereghetti, S., G¨otz,D., Weidenspointner, G., et al. 2009, ApJL, 696, L74 Mignani, R. P., Rea, N., Testa, V., et al. 2009, A&A, 497, 451 Milhalas, D., & Binney, J. 1981, Galactic Astronomy: Structure and Kinematics, 2nd edn. (San Francisco: Freeman) Minter, A. H., Camilo, F., Ransom, S. M., Halpern, J. P., & Zimmerman, N. 2008, ApJ, 676, 1189 Mitsuda, K., Bautz, M., Inoue, H., et al. 2007, PASJ, 59, 1 Molkov, S. V., Hurley, K., Sunyaev, R. A., et al. 2005, A&A, 433, L13 Mori, K., Gotthelf, E. V., Zhang, S., et al. 2013, ApJL, 770, L23 Morii, M., Kawai, N., Kataoka, J., et al. 2005, AdSpR, 35, 1177 Morii, M., Kitamoto, S., Shibazaki, N., et al. 2010, PASJ, 62, 1249 Morris, D. J., Hobbs, G. B., Lyne, A. G., et al. 2002, MNRAS, 335, 275 Morton, D. C. 1964, ApJ, 140, 460

137 Motch, C. 2006, in ESA Special Publication, Vol. 604, The X-ray Universe 2005, ed. A. Wilson (Noordwijk: ESA Publications Division), 383 Muno, M. P., Clark, J. S., Crowther, P. A., et al. 2006, ApJL, 636, L41 Murakami, T., Tanaka, Y., Kulkarni, S. R., et al. 1994, Natur, 368, 127 Nakamura, R., Bamba, A., Ishida, M., et al. 2009, PASJ, 61, 197 Ng, C.-Y., Kaspi, V. M., Ho, W. C. G., et al. 2012, ApJ, 761, 65 Ng, C.-Y., Slane, P. O., Gaensler, B. M., & Hughes, J. P. 2008, ApJ, 686, 508 Ng, C.-Y., Kaspi, V. M., Dib, R., et al. 2011, ApJ, 729, 131 Olausen, S. A., & Kaspi, V. M. 2014, ApJS, 212, 6 Olausen, S. A., Kaspi, V. M., Lyne, A. G., & Kramer, M. 2010, ApJ, 725, 985 Olausen, S. A., Kaspi, V. M., Ng, C.-Y., et al. 2011, ApJ, 742, 4 Olausen, S. A., Zhu, W., Vogel, J. K., et al. 2013, ApJ, 764, 1 Ouyed, R., Leahy, D., & Niebergal, B. 2007a, A&A, 473, 357 —. 2007b, A&A, 475, 63 Pacini, F. 1968, Natur, 219, 145 Pagani, C., Beardmore, A. P., & Kennea, J. A. 2011, ATel, 3493 Page, D., Geppert, U., & K¨uker, M. 2007, Ap&SS, 308, 403 Page, D., Geppert, U., & Weber, F. 2006, NuPhA, 777, 497 Palaniswamy, D., Bhat, N. D. R., Tingay, S. J., et al. 2013, ATel, 5076 Palmer, D. M., Barthelmy, S. D., Gehrels, N., et al. 2005, Natur, 434, 1107 Park, S., Hughes, J. P., Slane, P. O., et al. 2012, ApJ, 748, 117 Pavlov, G. G., Shibanov, Y. A., Zavlin, V. E., & Meyer, R. D. 1995, in NATO ASI Series C, Vol. 450, The Lives of the Neutron Stars, ed. M. A. Alpar, U.¨ Kizilo˘glu,& J. van Paradijs (Dordrecht: Kluwer), 71 Pethick, C. J. 1992, RvMP, 64, 1133

138 Pivovaroff, M. J., Kaspi, V. M., & Camilo, F. 2000, ApJ, 535, 379 Pons, J. A., Link, B., Miralles, J. A., & Geppert, U. 2007, PhRvL, 98, 071101 Pons, J. A., Miralles, J. A., & Geppert, U. 2009, A&A, 496, 207 Popov, S. B., Pons, J. A., Miralles, J. A., Boldin, P. A., & Posselt, B. 2010, MNRAS, 401, 2675 Predehl, P., & Schmitt, J. H. M. M. 1995, A&A, 293, 889 Qiu, Y., Lu, C. L., Lou, Y. Q., Urata, Y., & Huang, K. Y. 2005, GCN, 4036 Radhakrishnan, V., & Manchester, R. N. 1969, Natur, 222, 228 Rea, N., & Esposito, P. 2011, in High-Energy Emission from Pulsars and their Systems, ed. D. F. Torres & N. Rea, Astrophysics and Space Science Proceedings (Berlin: Springer), 247 Rea, N., Pons, J. A., Torres, D. F., & Turolla, R. 2012a, ApJL, 748, L12 Rea, N., Vigano, D., Israel, G., Pons, J. A., & Torres, D. F. 2014, ApJL, 781, L17 Rea, N., Testa, V., Israel, G., et al. 2004, A&A, 425, L5 Rea, N., Nichelli, E., Israel, G., et al. 2007a, MNRAS, 381, 293 Rea, N., Israel, G., Oosterbroek, T., et al. 2007b, Ap&SS, 308, 505 Rea, N., McLaughlin, M. A., Gaensler, B. M., et al. 2009a, ApJL, 703, L41 Rea, N., Israel, G., Turolla, R., et al. 2009b, MNRAS, 396, 2419 Rea, N., Esposito, P., Turolla, R., et al. 2010, Sci, 330, 944 Rea, N., Israel, G., Esposito, P., et al. 2012b, ApJ, 754, 27 Rea, N., Israel, G., Pons, J. A., et al. 2013, ApJ, 770, 65 Reed, B. C. 1997, PASP, 109, 1145 —. 2000, AJ, 120, 314 Richards, D. W., & Comella, J. M. 1969, Natur, 222, 551 Rickett, B. J. 1990, ARA&A, 28, 561

139 Romani, R. W. 1987, ApJ, 313, 718 —. 1996, ApJ, 470, 469 Rosen, S., Holland, S. T., Marshall, F. E., Boyd, P., & Gehrels, N. 2005, GCN, 4038 Ruderman, M. A., & Sutherland, P. G. 1975, ApJ, 196, 51 Safi-Harb, S., & Kumar, H. S. 2008, ApJ, 684, 532 Sakamoto, T., Barbier, L., Barthelmy, S. D., et al. 2011, AdSpR, 47, 1346 Sasaki, M., Plucinsky, P. P., Gaetz, T. J., & Bocchino, F. 2013, A&A, 552, 45 Sato, T., Bamba, A., Nakamura, R., & Ishida, M. 2010, PASJ, 62, L33 Scholz, P., & Kaspi, V. M. 2011, ApJ, 739, 94 Scholz, P., Kaspi, V. M., & Cumming, A. 2014, ApJ, 786, 62 Scholz, P., Ng, C.-Y., Livingstone, M. A., et al. 2012, ApJ, 761, 66 Seward, F. D., Charles, P. A., & Smale, A. P. 1986, ApJ, 305, 814 Shannon, R. M., & Johnston, S. 2013, MNRAS, 435, L29 Shapiro, S. L., & Teukolsky, S. A. 1983, Black Holes, White Dwarfs, and Neutron Stars: The Physics of Compact Objects (New York: Wiley- Interscience) Shklovsky, I. S. 1967, ApJL, 148, L1 Speagle, J. S., Kaplan, D. L., & van Kerkwijk, M. H. 2011, ApJ, 743, 183 Spitkovsky, A. 2006, ApJL, 648, L51 Spitler, L. G., Lee, K. J., Eatough, R. P., et al. 2014, ApJL, 780, L3 Staelin, D. H., & Reifenstein, E. C. I. 1968, Sci, 162, 1481 Su, Y., Chen, Y., Yang, J., et al. 2009, ApJ, 694, 376 Takahashi, T., Abe, K., Endo, M., et al. 2007, PASJ, 59, 35 Tam, C. R., Gavriil, F. P., Dib, R., et al. 2008, ApJ, 677, 503

140 Tam, C. R., Kaspi, V. M., Gaensler, B. M., & Gotthelf, E. V. 2006, ApJ, 652, 548 Tam, C. R., Kaspi, V. M., van Kerkwijk, M. H., & Durant, M. 2004, ApJL, 617, L53 Tanaka, Y., Inoue, H., & Holt, S. S. 1994, PASJ, 46, L37 Tanvir, N. R., & Varricatt, W. 2008, GCN, 8126 Taylor, G. B., Gelfand, J. D., Gaensler, B. M., et al. 2005, ApJL, 634, L93 Tello, J. C., Sota, A., & Castro-Tirado, A. J. 2011, GCN, 12272 Tendulkar, S. P. 2013, in IAU Symposium, Vol. 291, Proceedings of the International Astronomical Union, ed. J. van Leeuwen (Cambridge University Press), 514 Tendulkar, S. P., Cameron, P. B., & Kulkarni, S. R. 2012, ApJ, 761, 76 —. 2013, ApJ, 772, 31 Testa, V., Rea, N., Mignani, R. P., et al. 2008, A&A, 482, 607 Thompson, C., & Blaes, O. 1998, PhRvD, 57, 3219 Thompson, C., & Duncan, R. C. 1993, ApJ, 408, 194 —. 1995, MNRAS, 275, 255 —. 1996, ApJ, 473, 322 Thompson, C., Lyutikov, M., & Kulkarni, S. R. 2002, ApJ, 574, 332 Tian, W. W., & Leahy, D. A. 2008, ApJ, 677, 292 —. 2012, MNRAS, 421, 2593 Tian, W. W., Li, Z., Leahy, D. A., & Wang, Q. D. 2007, ApJL, 657, L25 Tiengo, A., Esposito, P., & Mereghetti, S. 2008, ApJL, 680, L133 Tiengo, A., Esposito, P., Mereghetti, S., et al. 2009, MNRAS, 399, L74 Tiengo, A., Vianello, G., Esposito, P., et al. 2010, ApJ, 710, 227 Tong, H., Yuan, J.-P., & Liu, Z.-Y. 2013, RAA, 13, 835

141 Torii, K., Kinugasa, K., Katayama, K., Tsunemi, H., & Yamauchi, S. 1998, ApJ, 503, 843 Treves, A., Turolla, R., Zane, S., & Colpi, M. 2000, PASP, 112, 297 Truemper, J. 1993, Sci, 260, 1769 Turolla, R. 2009, in Astrophysics and Space Science Library, Vol. 357, Neutron Stars and Pulsars, ed. W. Becker (Berlin: Springer), 141 Turolla, R., Zane, S., Pons, J. A., Esposito, P., & Rea, N. 2011, ApJ, 740, 105 Ubertini, P., Lebrun, F., Di Cocco, G., et al. 2003, A&A, 411, L131 van der Horst, A. J., Connaughton, V., Kouveliotou, C., et al. 2010, ApJL, 711, L1 van Kerkwijk, M. H., & Kaplan, D. L. 2008, ApJL, 673, L163 van Paradijs, J., Taam, R. E., & van den Heuvel, E. P. J. 1995, A&A, 299, L41 Vasisht, G., & Gotthelf, E. V. 1997, ApJL, 486, L129 Vedrenne, G., Roques, J. P., Sch¨onfelder,V., et al. 2003, A&A, 411, L63 Vigano, D., Rea, N., Pons, J. A., et al. 2013, MNRAS, 434, 123 Vink, J., & Bamba, A. 2009, ApJL, 707, L148 Vink, J., & Kuiper, L. 2006, MNRAS, 370, L14 Voges, W., Aschenbach, B., Boller, T., et al. 1999, A&A, 349, 389 Vrba, F. J., Henden, A. A., Luginbuhl, C. B., et al. 2000, ApJL, 533, L17 Wachter, S., Ramirez-Ruiz, E., Dwarkadas, V. V., et al. 2008, Natur, 453, 626 Wachter, S., Patel, S. K., Kouveliotou, C., et al. 2004, ApJ, 615, 887 Wang, Z., Bassa, C. G., Kaspi, V. M., Bryant, J. J., & Morrell, N. 2008a, ApJ, 679, 1443 Wang, Z., & Chakrabarty, D. 2002, ApJL, 579, L33

142 Wang, Z., Chakrabarty, D., & Kaplan, D. L. 2008b, in AIP Conference Proceedings, Vol. 983, 40 Years of Pulsars: Millisecond Pulsars, Magne- tars, and More, ed. C. G. Bassa, Z. Wang, A. Cumming, & V. M. Kaspi (Melville, NY: AIP), 274 Wang, Z., & Kaspi, V. M. 2008, ApJ, 675, 695 Wang, Z., Kaspi, V. M., & Higdon, S. J. U. 2007, ApJ, 665, 1292 Wang, Z., Kaspi, V. M., Osip, D., et al. 2006, ATel, 910 Weisskopf, M. C., Tananbaum, H. D., Van Speybroeck, L. P., & O’Dell, S. L. 2000, in Proceedings of the SPIE, Vol. 4012, X-ray Optics, Instru- ments, and Missions III, ed. J. E. Tr¨umper & B. Aschenbach, 2 Weltevrede, P., Johnston, S., & Espinoza, C. M. 2011, MNRAS, 411, 1917 Wolter, H. 1952, AnP, 445, 94 Woods, P. M., Kouveliotou, C., Finger, M. H., et al. 2007, ApJ, 654, 470 Woods, P. M., & Thompson, C. 2006, in Compact Stellar X-ray Sources, ed. W. H. G. Lewin & M. van der Klis (Cambridge: Cambridge University Press), 547 Woods, P. M., Kouveliotou, C., van Paradijs, J., et al. 1999, ApJL, 519, L139 Woods, P. M., Kouveliotou, C., Gavriil, F. P., et al. 2005, ApJ, 629, 985 Yakovlev, D. G., Gnedin, O. Y., Kaminker, A. D., & Potekhin, A. Y. 2008, in AIP Conference Proceedings, Vol. 983, 40 Years of Pulsars: Millisecond Pulsars, Magnetars, and More, ed. C. G. Bassa, Z. Wang, A. Cumming, & V. M. Kaspi (Melville, NY: AIP), 379 Younes, G., Kouveliotou, C., Kargaltsev, O., et al. 2012, ApJ, 757, 39 Young, M. D., Manchester, R. N., & Johnston, S. 1999, Natur, 400, 848 Zavlin, V. E., Pavlov, G. G., & Shibanov, Y. A. 1996, A&A, 315, 141 Zhou, P., Chen, Y., LI, X.-D., et al. 2014, ApJL, 781, L16

143 Zhu, W., Kaspi, V. M., Dib, R., et al. 2008, ApJ, 686, 520 Zhu, W., Kaspi, V. M., Gonzalez, M. E., & Lyne, A. G. 2009, ApJ, 704, 1321 Zhu, W., Kaspi, V. M., McLaughlin, M. A., et al. 2011, ApJ, 734, 44 Zwicky, F. 1938, ApJ, 88, 522

144