1: \documentclass[twocolumn,runningheads,natbib]{svjour2}
2: \journalname{Astrophysics and Space Science}
3:
4: %
5: \smartqed % flush right qed marks, e.g. at end of proof
6: %
7: \usepackage{graphicx}
8: %
9: % \usepackage{mathptmx} % use Times fonts if available on your TeX system
10: %
11: % insert here the call for the packages your document requires
12: %\usepackage{latexsym}
13: % etc.
14: %
15: % please place your own definitions here and don't use \def but
16: % \newcommand{}{}
17: %
18: %
19: \begin{document}
20:
21: \title{Recent Progress on Anomalous X-ray Pulsars%\thanks{Grants or other notes
22: }
23: %\subtitle{Do you have a subtitle?\\ If so, write it here}
24:
25: %\titlerunning{Short form of title} % if too long for running head
26:
27: \author{Victoria M. Kaspi}
28:
29: \authorrunning{V. M. Kaspi} % if too long for running head
30:
31: \institute{V. Kaspi \at
32: Physics Department, McGill University, Montreal, Canada, H3A 2T8\\
33: \email{vkaspi@physics.mcgill.ca}
34: }
35:
36: \date{Received: date / Accepted: date}
37: % The correct dates will be entered by the editor
38:
39:
40: \maketitle
41:
42: \begin{abstract}
43: I review recent observational progress on \\ Anomalous X-ray Pulsars,
44: with an emphasis on timing, variability, and spectra. Highlighted
45: results include the recent timing and flux stabilization of
46: the notoriously unstable AXP 1E 1048.1$-$5937, the remarkable
47: glitches seen in two AXPs, the newly recognized variety of AXP variability
48: types, including outbursts, bursts, flares, and pulse profile changes,
49: as well as recent discoveries regarding AXP spectra, including their
50: surprising hard X-ray and far-infrared emission, as well as the pulsed radio
51: emission seen in one source. Much has been
52: learned about these enigmatic objects over the past few years, with the pace
53: of discoveries remaining steady.
54: However additional work on both observational and theoretical fronts
55: is needed before we have a comprehensive understanding of AXPs and their
56: place in the zoo of manifestations of young neutron stars.
57: \keywords{pulsars \and magnetars \and variability \and X-ray spectra \and timing}
58: \PACS{97.60Jd \and 97.60.Gb \and 98.70.Qy}
59: \end{abstract}
60:
61: \section{Introduction}
62: \label{intro}
63:
64: Although very few in number, the seven, and possibly nine, known
65: so-called ``Anomalous X-ray Pulsars'' (AXPs; see Table~\ref{ta:axps})
66: are potentially very powerful for making progress on the physics of
67: neutron stars. AXPs embody properties that are highly reminiscent of
68: two other, very different classes of neutron star: the spectacular Soft
69: Gamma Repeaters (SGRs; see contributions by \linebreak S. Mereghetti and others in
70: this volume), and conventional radio pulsars. The great similarity
71: of AXPs to SGRs is what makes the case for AXPs being magnetars
72: so compelling and also offers the hope of constraining magnetar physics. The
73: intriguing similarities with radio pulsars offer the promise of solving
74: long-standing problems in our theoretical understanding of the latter.
75:
76: In this review, I describe the most recent observational progress on
77: AXPs. The review will be divided into sections
78: on timing (\S\ref{sec:timing}), variability (\S\ref{sec:variability}),
79: and spectra (\S\ref{sec:spectra}), choices that, unfortunately, may
80: exhibit some personal bias, necessary given the limited space available.
81: I hope to show that recent AXP progress has been significant, however
82: ultimately much observational and theoretical work remains to be done
83: before a complete picture of AXPs and their place in the neutron-star
84: zoo becomes clear.
85:
86: Note that the most comprehensive and recent review of AXPs and
87: magnetars in general is that by \citet{wt04}. In this review, I make
88: use of the detailed, online summary of magnetar properties and references maintained at
89: McGill University by C. Tam\\ (www.physics.mcgill.ca/$\sim$pulsar/magnetar/main.html).
90:
91:
92: \begin{table*}[t]
93: \caption{Properties of Known and Candidate AXPs.}
94: %\centering
95: \begin{center}
96: \label{ta:axps}
97: \begin{tabular}{llllllll}
98: \hline\noalign{\smallskip}
99: Name & $P$ & $\dot{P}^a$ & $B$ & Timing$^b$ & X-ray Variability$^c$ & Waveband$^d$ & Association \\[3pt]
100: & (s) &($\times 10^{-11}$)& ($\times 10^{14}$~G)& Properties & Properties & Detections & \\
101: \tableheadseprule\noalign{\smallskip}
102: CXOU J010043.1$-$721134 & 8.02 & 1.9 & 3.9 & S? & S & X O? & SMC \\
103: 4U 0142+61 & 8.69 & 0.2 & 1.3 & S G? & M P & H X O I & ... \\
104: 1E 1048.1$-$5937 & 6.45 & 2.7 & 4.2 & N & S F F B & H X I? & ... \\
105: CXOU J164710.2$-$455216 & 10.61 & ... & ... & ... & ... & X & Westerlund 1 \\
106: 1RXS J170849.0$-$400910 & 11.00 &1.9 & 4.7 & S G G & S & H X I & ... \\
107: XTE J1810$-$197 & 5.54 & 0.5 & 1.7 & N & O B & X I R & ... \\
108: 1E 1841$-$045 & 11.78 & 4.2 & 7.1 & N & S & H X I? & SNR Kes 73 \\
109: AX J1845$-$0258 & 6.97 & ... & ... & ... & O & X & SNR G29.6+0.1 \\
110: 1E 2259+586 & 6.98 & 0.048 & 0.59& S G & S O B P & H X I & SNR CTB 109\\
111: \noalign{\smallskip}\hline
112: \end{tabular}
113: \end{center}
114: $^a$Long-term average value.\\
115:
116: $^b$S=stable (i.e. can be phase-connected over many-month intervals, generally); N=noisy (i.e. generally difficult to phase-connect over many-month intervals); G= one glitch. \\
117:
118: $^c$S=stable (i.e. no variability, generally); M=modest variability; F=one flare; O=one outburst; B=bursts; P=pulse profile changes. \\
119:
120: $^d$H=hard X-ray; X=X-ray; O=optical; I=infrared; R=radio. \\
121:
122: \end{table*}
123:
124:
125:
126: \section{Timing}
127: \label{sec:timing}
128:
129: Timing observations of AXPs hold considerable information about both their
130: surroundings via the external torques they feel, as well as potentially about their
131: internal structure, via the ``glitches'' they experience. Here we summarize what
132: is known regarding AXP timing properties, highlighting the most interesting issues.
133:
134: \subsection{Stability and Phase-Coherent Timing}
135: \label{sec:stability}
136:
137: Studies of the timing properties of AXPs can reasonably be categorized
138: as pre- and post-Rossi X-ray Timing Explorer (RXTE), launched in
139: late 1995. Prior to RXTE, timing studies were limited to occasional
140: observations spread over many years. These characterized the overall
141: spin-down behavior of several AXPs, and also suggested some
142: interesting deviations from simple spin-down \citep[e.g.][]{bs96,cm97,bss+00,pkdn00}.
143: However the nature of these deviations could not be determined, because of
144: the paucity of observations. Careful
145: searches for Doppler shifts of the observed periodicities had been
146: done \citep[e.g.][]{ikh92,bs96}; typical upper limits on $a\sin i$
147: were $\sim$0.1 lt-sec for a variety of orbital periods.
148:
149: RXTE and its Proportional Counter Array (PCA) revolutionized
150: the timing of AXPs. The first PCA studies of AXPs reduced the limits
151: on $a\sin i$ to 0.03 lt-s for 1E~2259+586 and 0.06 lt-s for 1E
152: 1048.1$-$5937, effectively ruling out any main-sequence star companion
153: and rendering binary accretion models highly problematic
154: \citep{mis98}.
155: %Similar limits were obtained for 4U 0142+61 \citep{wdf+98}.
156:
157: Subsequently, a program of regular monthly monitoring of the AXPs with the PCA
158: on RXTE was approved and showed that fully phase-coherent timing
159: of AXPs could be done over periods of years, assuming spin-down models consisting
160: of very few parameters \citep{kcs99}. For example, \citet{gk02} showed that
161: in 4.5~yr of RXTE monitoring, pulse times of arrival for 1E~2259+586
162: could be predicted to within 1\% of the pulse period using $\nu$, $\dot{\nu}$ and
163: $\ddot{\nu}$ only. Such stability is comparable to that of conventional young radio
164: pulsars and very much unlike the large amplitude torque noise commonly seen
165: in accreting neutron stars. Long-term, regular monthly (or even bi-monthly) observations
166: of AXPs continue today, with four of the five persistent confirmed Galactic sources
167: (4U 0142+61, 1RXS J170849.0$-$400910, 1E 1841$-$045 and 1E 2259+586) generally
168: exhibiting stability that allows phase coherence with few parameters over years
169: \citep{klc00,gk02,ggk+02,kg03,dkg06}. A summary of the timing properties of the known
170: AXPs is given in Table~\ref{ta:axps}.
171:
172: \begin{figure}[h]
173: \centering
174: \includegraphics[width=0.45\textwidth]{1048freq.eps}
175: \caption{Spin-frequency history of 1E 1048.1$-$5937 from
176: RXTE monitoring (work in preparation). Note the
177: simple spin down in the last $1.9$~yr.}
178: \label{fig:1048freq}
179: \end{figure}
180:
181:
182: One of the established persistent Galactic AXPs, 1E 1048.1$-$5937,
183: has been much less stable than the others, such that phase-coherent timing
184: with unambiguous pulse
185: counting over more than a few weeks or months has been difficult
186: to achieve \citep{kgc+01,gk04}. This poor stability is apparent in
187: the source's frequency evolution (Fig. 1). Recently, however,
188: during an extended period of pulsed flux stability following two
189: long-lived X-ray flares (see \S\ref{sec:flares} below), the timing
190: has also become quite stable, such that unambiguous phase coherence
191: could be maintained over a nearly 2-yr interval from MJD 53158 to
192: 53858 using 4 spin parameters, although significant residuals remain
193: (Fig. \ref{fig:1048res}). Details of these results will be described
194: elsewhere. It remains to be seen if an end to this
195: stability will be accompanied by additional radiative events, a result
196: that would hopefully be useful for strongly constraining models for the
197: torque noise and flares.
198:
199:
200: \begin{figure}[h]
201: \centering
202: \includegraphics[width=0.35\textwidth,angle=270]{1048res.eps}
203: \caption{Phase residuals following removal of four frequency derivatives
204: for the last $1.9$~yr of data for AXP 1E 1048.1$-$5937.
205: The features are roughly 300~d apart and are presently consistent with
206: being due to random timing noise (work in preparation).}
207: \label{fig:1048res}
208: \end{figure}
209:
210:
211: \subsection{Glitches}
212: \label{sec:glitches}
213:
214: The impressive timing stability seen in most AXPs, in which pulse periods
215: could easily be determined to better than a part in a billion,
216: permitted for the first time the unambiguous detection
217: of sudden spin-up glitches in AXPs. 1RXS J170849.0$-$400910 was the first
218: pulsar seen to glitch \citep{klc00}. This glitch
219: had a fractional frequency increase of $6\times 10^{-7}$,
220: very similar to what is seen for the Vela pulsar and other
221: comparably young objects. Continued monitoring of the same AXP revealed
222: a second, larger glitch 1.5~yr later \citep{kg03,dis+03}, with a nearly complete
223: recovery of the frequency jump, unusual by pulsar standards (see Fig. \ref{fig:1708_glitches}).
224:
225: \begin{figure}
226: \centering
227: \includegraphics[width=0.4\textwidth]{1708_glitches.eps}
228: \caption{Long-term RXTE timing data for 1RXS J170849.0$-$400910 showing
229: the two glitches observed thus far \citep[from][]{kg03}.}
230: \label{fig:1708_glitches}
231: \end{figure}
232:
233: In June 2002, at the time of a major radiative outburst (see
234: \S\ref{sec:variability} below), the otherwise stable AXP \linebreak 1E 2259+586
235: exhibited a large glitch (fractional frequency increase of $4 \times 10^{-6}$)
236: which was accompanied by significant and truly remarkable recovery in which roughly
237: a quarter of the frequency jump relaxed on a
238: $\sim$40-day scale. As part of that recovery, the measured spin-down rate of
239: the pulsar was temporarily larger than the long-term pre-burst average by
240: a factor of 2! This recovery was somewhat similar to, though better sampled than,
241: the second glitch seen in
242: 1RXS J170849.0$-$400910. The 1E~2259+586 event was the first (and still the only
243: unambiguous, but see \S\ref{sec:flares})
244: time a spin-up glitch in a neutron star was accompanied by any form of
245: radiative change. This event suggests that large glitches in AXPs are
246: generally associated with radiative events; perhaps such an event occured
247: at the time of the second glitch in 1RXS J170849.0$-$400910
248: but went unnoticed due to sparse monitoring.
249: The very interesting recoveries of the 1E~2259+586 and second
250: 1RXS J170849.1$-$400910 glitches seem likely to be telling us something
251: interesting about the interior of a magnetar and how it is different
252: from that of a conventional, low-field neutron star. This seems worth
253: more attention than it has received in the literature thus far.
254:
255: Recently a timing anomaly was reported in AXP \linebreak 4U~0142+61 \citep{dkgw06}.
256: Although at first the anomaly seemed consistent with a spin-up glitch
257: following a burst observed on April 6, 2006 \citep{kdg06}, the subsequent
258: data do not support a simple glitch interpretation. This is work in progress and
259: will be reported on elsewhere. However, \citet{mks05} and \citet{dkg06} suggest that
260: this pulsar may have glitched during a large observing gap in 1998-99, a possibility
261: supported by apparent changes in the pulse profile that seem
262: to have occured in the same interval \citep{dkg06}.
263:
264:
265: \section{AXP Variability}
266: \label{sec:variability}
267:
268: Arguably one of the most interesting recent discoveries in the study of
269: AXPs is the range and diversity of their X-ray variability properties.
270: Pre-RXTE, there was variability reported \citep[e.g.][]{bs96,cm97,opmi98,pkdn00},
271: however those relatively sparse observations were made using different
272: instruments aboard different observatories, some imaging, some not, and
273: were often presented without uncertainties, making their interpretation difficult.
274: Moreover, given the sparseness of the observations, identifying a time scale
275: for the variations, or being certain the full dynamic range was being sampled,
276: was not possible.
277:
278: Post-RXTE, and, additionally, with contemporaneous observations made
279: using Chandra and XMM, the picture has become much clearer.
280: Presently there appear to be at least four types of X-ray variability
281: in AXP pulsed and persistent emission: outbursts (sudden large increases in the
282: pulsed and persistent flux, accompanied by bursts and other radiative and
283: timing anomalies, which decay on time scales of weeks or months), bursts (sudden
284: events, lasting milliseconds to seconds), long-term flux changes (time scales of
285: years), and pulse profile changes (on time scales of days to years).
286: We examine each of these phenomena in turn.
287:
288: \subsection{Outbursts and Transients}
289: \label{sec:outbursts_transients}
290:
291: The current best example of an AXP outburst is that seen
292: in June 2002 from 1E~2259+586 \citep{kgw+03,wtk+04}. This outburst,
293: which lasted only a few hours, fortuitously occured during a few-hour
294: monthly RXTE monitoring observation. During the outburst,
295: the pulsed and persistent fluxes increased by a factor of $\sim$20 (see
296: Fig. \ref{fig:2259_outburst}), there were over 80 short SGR-like bursts
297: in a few-hour period (see \S\ref{sec:bursts}), there were substantial
298: pulse profile changes (see \S\ref{sec:profile_changes} below), there
299: was a short-term decrease in the pulsed fraction, the spectrum hardened
300: dramatically, there was a large glitch (see \S\ref{sec:glitches} above),
301: and there was an infrared enhancement (see \S\ref{sec:ir_variability}).
302: All this came after over 4~yr of
303: otherwise uneventful behavior \citep{gk02}. Note that had RXTE
304: not been observing the source during the outburst, the entire event
305: would have appeared, from the monitoring data, to consist principally
306: of a glitch. With only monthly monitoring, all but the longest-term
307: radiative changes would have been missed. Interestingly, this outburst
308: was notably different from SGR outbursts; for the AXP, the energy
309: in the outburst ``afterglow'' ($\sim 10^{41}$~erg; see Fig. \ref{fig:2259_outburst}) greatly
310: exceeded that in the bursts ($\sim 10^{37}$~erg). This is in direct contrast to the giant
311: flares of SGRs. The reason for this difference is unknown.
312:
313: \begin{figure}
314: \centering
315: \includegraphics[width=0.5\textwidth]{uflux_2259_log.ps}
316: \caption{Flux time series plotted logarithmically following the 2002 1E 2259+586
317: outburst \citep{wtk+04}. Diamonds denote inferred unabsorbed
318: flux values calculated from RXTE PCA pulsed-flux measurements. Asterisks
319: and squares mark independent phase-averaged unabsorbed flux values from
320: RXTE and XMM, respectively. The dotted line denotes the flux level
321: measured using XMM 1 fortuitously week prior to the glitch. The dashed line
322: is a power-law fit to the PCA flux measurements during the observations
323: containing the burst activity ($<$1 day).}
324: \label{fig:2259_outburst}
325: \end{figure}
326:
327: The outburst of 1E~2259+586 seems likely to be a good model for the
328: behavior of the ``transient'' AXP XTE J1810$-$197. This source,
329: unknown prior to 2003, was discovered as a bright 5.5~s X-ray pulsar,
330: upon emerging from behind the Sun in that year \citep{ims+04}, and
331: has been fading ever since. The source's spin down and spectrum are
332: consistent with it being an AXP. \citet{ghbb04} showed from archival
333: ROSAT observations that in the past, the source, in quiescence,
334: was nearly two orders of magnitude fainter than in outburst. See the
335: contribution by Gotthelf et al. in this volume for more details.
336: An important question raised by the discovery of XTE J1810$-$917 is
337: how many more such objects exist in our Galaxy? This has important
338: implications for AXP birthrates.
339:
340: Such an outburst may also explain the behavior of the unconfirmed
341: transient AXP AX J1845$-$0258 (see Table~\ref{ta:axps}). Pulsations with
342: a period of 7~s were observed in an archival 1993 ASCA observation
343: \citep{tkk+98,gv98}, however subsequent observations of the target
344: revealed a large drop in flux, and pulsations have not been redetected.
345: Although no $\dot{\nu}$ has been measured for this source, it seems
346: likely to be an AXP given its period and location at the center of a supernova
347: remnant \citep{ggv99}. It thus is plausible that the 1993 outburst was
348: similar to that of 1E~2259+586 or XTE J1810$-$197, although the likely dynamic
349: range for this source is unprecedented for an AXP. See the contribution
350: to these proceedings by Tam et al. and \citet{tkgg06} for more details.
351:
352: One of the most puzzling aspects of transient AXPs is why the
353: quiescent source is so faint. In the standard magnetar model, the
354: requisite magnetic field decay energy input is persistent, as is the
355: magnetospheric twist for a fixed magnetar-strength magnetic field
356: \citep{td96a,tlk02}. Although much attention has been paid to why
357: a magnetar's flux might skyrocket suddenly -- a sudden reconfiguration
358: of the surface following a crustal yield -- relatively little attention
359: has been paid to how to stop or hide the bright X-ray emission from a
360: neutron star presumably harboring a magnetar-strength field.
361:
362: \subsection{Bursts}
363: \label{sec:bursts}
364:
365: During the 2002 1E~2259+586 outburst, over 80 short, SGR-like X-ray
366: bursts were seen superimposed on the overall flux trend over the course
367: of the few-hour RXTE observation \citep{kgw+03}. Some were \linebreak
368: super-Eddington, though
369: only on very short (few ms) time scales. In a detailed analysis of these
370: bursts, \citet{gkw04} found that in most respects, they are identical
371: to SGR bursts. Specifically, the durations, differential fluence
372: distribution, the burst morphologies, the wait-time distribution,
373: and the correlation between fluence and duration, are all SGR-like.
374: However a few of the burst properties were different from those of SGR
375: bursts: for example, the AXP bursts had a wider range of durations, the
376: AXP bursts were on average less energetic than in SGRs, and the more
377: energetic AXP bursts have the hardest spectra -- the opposite of what
378: is seen for SGRs. Unlike in SGRs, the AXP bursts were correlated with
379: pulsed intensity.
380:
381: Bursts in AXPs were first reported by \citet{gkw02} who discovered two such
382: events in archival RXTE data from the direction of 1E~1048.1$-$5937.
383: A third burst, found nearly 3 yr later \citep{gkw06}, unambiguously
384: identified the AXP as the origin thanks to a simultaneous pulsed flux increase.
385: The bursting behavior in this source is not obviously correlated with any other property,
386: although we note tentatively that no bursting has been seen during the most recent
387: 2~yr, when the pulsar has been exhibiting much improved timing
388: (see \S\ref{sec:timing}, Fig.~\ref{fig:1048freq}) and also pulsed flux stability.
389: Perhaps the bursting was symptomatic of whatever activity also caused the
390: timing instability and pulsed flux flares (see \S\ref{sec:flares} below).
391:
392: Four bursts have also been seen from the transient AXP XTE J1810$-$197
393: \citep{wkg+05}. These events consisted of a $\sim$1~s spike followed by
394: an extended tail in which the pulsed flux was enhanced, similar to the
395: third burst of 1E 1048.1$-$5937 and a handful from 1E~2259+586.
396: The XTE J1810$-$197 and 1E 1048.1$-$5937
397: bursts also showed a correlation with pulsed intensity, as did some of the 1E
398: 2259+586 bursts.
399:
400: These observations led \citet{wkg+05} to hypothesize
401: that there are in fact two distinct classes of bursts, which they
402: named Type A and B. Type A bursts are similar
403: to SGR bursts, in that they are uncorrelated with pulse phase and have
404: no extended tails. Type B bursts, on the other hand, thus far observed
405: exclusively in AXPs, are correlated with pulsed intensity, generally
406: have long extended tails, and those tails tend to contain more energy
407: than the burst itself. As both types of burst were seen in 1E~2259+586,
408: clearly they are not mutually exclusive, even during the same event.
409: \citet{wkg+05} speculate that Type A bursts have a magnetospheric trigger,
410: whereas Type B bursts originate from crust fractures.
411:
412: Most recently, a total of five small, sub-Eddington X-ray bursts
413: have been seen between April and June 2006 from 4U 0142+61 \citep[][work in
414: preparation]{kdg06,dkgw06}. The latter four, all within a single RXTE
415: orbit, were clearly accompanied by a pulsed flux increase (by a factor
416: of $\sim$4 relative to the long-term average) which decayed on
417: a time scale of minutes. This suggests, as does the accompanying pulse
418: profile change and timing anomaly, that this source may have entered an
419: extended active phase.
420:
421: With at least half of the known AXPs now established as capable of bursting, it
422: is clear that the production of occasional though clustered short SGR-like bursts is
423: a generic AXP phenomenon.
424:
425: \subsection{Long-Term Flux Variations}
426: \label{sec:flares}
427:
428: Variability in AXP 1E~1048.1$-$5937 had been reported for
429: years \citep[e.g.][]{cm97,opmi98,bss+00,mts+04}. RXTE monitoring determined
430: the time scale of the changes and the morphology of
431: the pulsed light curve with far superior time sampling than in previous
432: studies \citep{gk04}. Specifically, the RXTE monitoring
433: showed that over $\sim$7~yr, the source exhibited
434: two extended pulsed flux ``flares,'' the first lasting $\sim$100~days and the
435: second lasting over a year, each with rise times of several weeks.
436: Assuming a distance to the source of 5~kpc (which is not especially
437: well established), \citet{gk04} estimated the total energy released
438: in the pulsed components of the first and second flares to be 3 and 30
439: $\times10^{40}$~erg, respectively. Subsequently, \citet{tmt+05}, using
440: XMM, which (unlike RXTE) is sensitive to pulsed fraction,
441: showed that in fact the pulsed fraction is anti-correlated with the
442: phase-averaged flux, suggesting the total energy released was at least
443: twice that in the pulsed component.
444:
445: During these flares, the spin-down rate fluctuated by at least a factor
446: of 10 \citep{gk04}. However, there was no obvious correlation between
447: the detailed evolution of the spin-down rate and flux. Recently, while its flux
448: has been stable, 1E~1048.1$-$5937 has shown much greater timing stability
449: (see \S\ref{sec:timing} above). This suggests that the large torque noise
450: and flux flaring were causally related; we must await another such event
451: to confirm this.
452:
453: The flaring observed in 1E 1048.1$-$5937, a new phenomenon not yet
454: observed in any other AXP, is well understood in the context of the
455: twisted magnetosphere model \citep{tlk02}. The flux enhancements can
456: be seen as being due to increased twisting of the magnetosphere by
457: currents originating from the stressed crust. If so, a harder spectrum
458: is expected when the pulsar is brighter. Unfortunately the existing
459: data cannot confirm this important prediction for this source.
460: Decoupling of the torque from the pulsed flux can occur
461: in this model depending on the location of the enhanced magnetospheric
462: current; a current near the pole will have a disproportionate impact
463: on the spin-down. \citet{gk04} argued that the absence of predicted
464: torque--luminosity variations in this source are problematic for models
465: in which the X-ray emission originates from accretion from a fossil disk
466: \citep[e.g.][but see \"U. Ertan et al.'s contribution]{chn00,alp01}.
467:
468: The twisted magnetosphere model prediction that flux should be correlated
469: with hardness, though unconfirmed in 1E 1048.1$-$5937, does seem to be
470: borne out in observations of 1RXS J170849.0$-$400910 \citep[][see contribution
471: by N. Rea et al., this volume]{roz+05}. Moreover, those authors suggest
472: that the epochs of greatest hardness occur near those in which glitches
473: were detected in this source (see \S\ref{sec:glitches}), with subsequent
474: softening post-glitch. At least one additional glitch needs to occur, with
475: better observational coverage, before this conclusion is firm.
476:
477: Recently, much longer-term radiative variations have been identified in
478: AXP 4U 0142+61, in which the pulsed flux appears to be increasing steadily
479: since 2000, such that there was a $\sim$20\% change by early 2006,
480: just prior to its exhibiting a sudden pulse profile change and bursts.
481: Interestingly, this variation does not seem consistent with magnetospheric
482: changes given the predictions of the twi- sted magnetosphere model.
483: This behavior is described in more detail in the contribution by R. Dib
484: et al. to this volume as well as in
485: \citet{dkg06}. One particularly interesting implication of the increase
486: in X-ray flux from this source is that the phenomenon provides a simple
487: test of the irradiated fall-back disk model for the near-IR emission
488: (see \S\ref{sec:spectra_optir}). If the source of irradiation, the AXP, increases in brightness,
489: the disk ought to as well.
490:
491: \subsection{Pulse Profile Changes}
492: \label{sec:profile_changes}
493:
494: The first observed pulse profile change in an AXP was reported by
495: \citet{ikh92} using {\it GINGA} data obtained in 1989. They witnessed
496: a large change in the ratio of the amplitudes of the two peaks in
497: the pulse profile of 1E~2259+586, namely from near unity to over two.
498: They also reported a contemporaneous timing anomaly which, in hindsight,
499: is consistent with a spin-up glitch.
500:
501: A very similar pulse profile change was witnessed during and immediately
502: following the 2002 outburst of 1E~2259+586 \citep{kgw+03,wtk+04}. Here,
503: the ratio of the amplitudes of the two pulses in the profile went from
504: unity pre-outburst to roughly two mid-outburst, relaxing back to normal
505: on a time scale of a few weeks (Fig. \ref{fig:2259profile}). Curiously,
506: the temporarily larger peak
507: in the 2002 outburst appeared to be the temporarily smaller peak in 1989, suggesting
508: that even if the physical origin of the events is the same, the details of the geometry
509: were different. Given the nature of this event, a likely explanation for
510: the pulse profile change is a magnetospheric reconfiguration following
511: a crustal fracture that simultaneously affected the inside and
512: outside of the star. This very strongly suggests that the \citet{ikh92}
513: pulse profile change was observed not long after a similar event; this is
514: consistent with their reported timing anomaly, and suggests such events
515: occur roughly every 1-2 decades in this source.
516:
517: \begin{figure}
518: \centering
519: \includegraphics[width=0.5\textwidth]{2259profile.ps}
520: \caption{Pulse profile changes in 1E 2259+586 following its 2002 outburst
521: \citep{wtk+04}.}
522: \label{fig:2259profile}
523: \end{figure}
524:
525: Most recently, long-term (i.e. time scale of several years) pulse
526: profile variations have been reported for AXP 4U~0142+61 \citep[][see
527: contribution by R. Dib et al, these proceedings]{dkg06}. These accompany
528: a long-term pulsed flux increase (see \S\ref{sec:flares}).
529: The profile changes are consistent with a significant event having
530: occured some time between mid-1998 and 2000, in which the second and
531: higher harmonics of the profile increased dramatically, and have been
532: returning to their pre-event level ever since. This gradual evolution
533: was, however, interrupted by an apparent sudden activity episode,
534: in which an SGR-like burst was detected along with a timing anomaly in
535: April 2006 \citep[see \S\ref{sec:bursts};][]{kdg06}. The analysis of the
536: latter data are in progress.
537:
538: \subsection{Near-IR Variability}
539: \label{sec:ir_variability}
540:
541: The large number of recent near-IR detections of AXPs has revealed an
542: interesting new variability phenomenon in these sources. Following the
543: 2002 outburst of \linebreak 1E~2259+586, there was an infrared (K$_s$) enhancement
544: by a factor of $\sim$3, 3 days post-outburst. This then decayed together
545: with the X-ray pulsed flux, with identical power-law indices \citep[see
546: Fig.~\ref{fig:2259_outburst}; \S\ref{sec:outbursts_transients}][]{tkvd04}. Those authors
547: concluded that the origins of both flux increases were magnetospheric
548: \citep[but see contribution by \"U. Ertan, these proceedings] [for the
549: fossil-disk viewpoint]{ea06}. A similar correlation was also reported
550: for AXP XTE~J1810$-$197 \citep{rti+04}.
551:
552: Meanwhile, significant near-IR variability has also been reported for
553: AXP 1E~1048.1$-$5937 \citep{ics+02,wc02,dvk05}. However, if anything, the
554: near-IR is {\it anti-correlated} with the X-ray flux. This is puzzling
555: given the \linebreak 1E~2259+586 and XTE~J1810$-$197 results. It is worth keeping
556: in mind, however, that there is now some evidence that 1E~1048.1$-$5937 has
557: been in an active phase from which it may have recently emerged (see \S\ref{sec:timing});
558: it will be interesting to see how its near-IR flux varies now that both
559: the X-ray flux and timing have stabilized (see Fig. \ref{fig:1048freq}).
560:
561: In addition, \citet{dv06c} report on near-IR observations of 4U~0142+61
562: and report no apparent correlation with the X-ray pulsed flux.
563: They argue that the situation for this source
564: is unclear and warrants additional, more frequent observations. Simultaneous
565: far-IR observations
566: would also be of interest to establish conclusively that they originate from
567: a separate mechanism, namely radiation from a passive, irradiated fall-back
568: disk \citep{wck06}.
569:
570: \section{AXP Spectra}
571: \label{sec:spectra}
572:
573: The description of spectra of AXPs has, until very recently, been
574: limited to the soft (0.5--10~keV) X-ray band, as that is where
575: AXPs were discovered and have been traditionally studied. However the
576: recent discoveries of optical and near-IR emission have cast them
577: firmly into the multi-wavelength regime, and the even more recent
578: detections in the hard X-ray band as well as in the far-IR and
579: radio bands broadens the situation even further.
580:
581: \subsection{X-ray Spectra}
582: \label{sec:xray_spectra}
583:
584: In the traditional 0.5--10~keV X-ray band, AXPs have long been known
585: to show two-component spectra, which are well described by
586: a blackbody plus a power law \cite[e.g.][]{wae+96,ios+01,mskk03}.
587: It is currently thought that the blackbody arises from internal heating
588: due to the decaying intense magnetic field, while the non-thermal
589: component is a result of resonant scattering of the thermal
590: seed photons off magnetospheric currents in the twisted magnetosphere
591: \citep{tlk02}. Detailed spectral modelling in this framework appears
592: to describe the spectrum of \linebreak 1E~1048.1$-$5937 \citep{lg06}, but
593: see the contribution by N. Rea in these proceedings.
594: Although some attempts were made to model the entire spectrum using
595: a single blackbody distorted by the effects of the intense magnetic
596: field on the atmosphere \citep[e.g.][]{oze01}, these could not
597: reproduce the non-thermal component adequately \citep[e.g.][]{phh+01,tlk02,lh03a}.
598:
599: There is evidence supporting a physical connection between the two components,
600: such as the very slow evolution of the pulse profile with
601: energy \citep[e.g.][]{ios+01,gk02}. For some AXPs (for example 1E~1048.1$-$5937
602: and 1E~1841$-$045), there is little if any variation in the pulse
603: profile, with no obvious difference between profiles in energy
604: bands that are thermally and non-thermally dominated. Even
605: in other AXPs for which the profile is energy-dependent, the
606: profiles in bands that are thermally and non-thermally dominated
607: are still very similar. This is in stark contrast to the situation
608: for rotation-powered pulsars, for which the thermal and non-thermal components generally
609: have radically different X-ray pulse profiles \citep[e.g.][]{hsg+02a}.
610:
611: Recently, \citet{hg05} have suggested on purely theoretical
612: grounds that the spectrum
613: of XTE J1810$-$197 is more appropriately described by a two-component
614: model consisting of two blackbodies (see also contribution by E. Gotthelf et
615: al. in this volume). Their main argument for this
616: interpretation is that an extrapolation of the power-law component to
617: low energies greatly exceeds that expected if the seeds are thermal
618: photons from the surface, as the blackbody eventually runs out of
619: photons to supply. Moreover, they argue, the expected blackbody cutoff
620: would then result in a substantial underestimate of the infrared flux,
621: assuming the latter is part of the non-thermal spectrum. Very recently,
622: \citet{dv06b} have shown using independently measured interstellar column
623: densities that the intrinsic spectra really are cut off, i.e. the power-law
624: component does not in fact extend far below the observable band. If correct,
625: the rationale for preferring the double blackbody over the blackbody plus
626: power law would not apply.
627:
628: This section would not be complete without some discussion of the
629: interesting recent results of \citet{dv06a,dv06b}, some of which are reported
630: on by Durant et al. in this volume. Specifically, using high-resolution
631: X-ray spectroscopy, they identified absorption edges whose amplitudes
632: determine $N_H$ under reasonable assumptions, independent of the overall
633: continuum spectral modelling. Using these newly measured values of $N_H$
634: and a novel distance estimation technique using reddening runs with distance
635: for red clump stars, they were able to first improve the spectral fits for
636: several AXPs significantly, and second determine
637: much improved distance estimates for them. Amazingly, among other things,
638: they find that the
639: soft X-ray luminosities of practically {\it all} AXPs are consistent with the
640: value $\sim 1.3 \times 10^{35}$~erg~s$^{-1}$. \citet{dv06a} argue that
641: this is consistent with the magnetar model's prediction
642: that there should be a saturation luminosity above which internal neutrino
643: cooling is at work.
644:
645: \subsection{Hard X-ray Spectra}
646:
647: A particularly interesting recent AXP discovery is that they are copious
648: hard X-ray emitters \citep{mcl+04,khm04}. Though their spectra in
649: $\nu F_{\nu}$ appear to fall off
650: in the softer X-ray band, they turn up again above $\sim$10~keV.
651: The luminosities above 10~keV independently greatly exceed the available spin-down power
652: by factors of over 100. This requires a new mechanism for accelerating particles
653: in the magnetosphere, in addition to an energy source, presumably the decaying
654: magnetic field. SGR 1806$-$20 has also been detected in this energy range,
655: though interestingly it is somewhat softer than the AXPs \citep{mgmh05}.
656: \citet{khdc06} have further shown that the hard X-ray emission is a generic
657: property of AXPs, and for at least three sources, extends without a break up to
658: 150~keV. They also show from {\it COMPTEL} upper limits that the break must
659: lie under $\sim$750~keV. Determining the location of the break could greatly
660: constrain emission models, possibly even providing independent evidence for
661: the magnetar-strength field.
662: See the contribution by P. den Hartog et al. in this volume for details regarding
663: hard-X-ray emission from 4U~0142+61.
664:
665: This hard X-ray emission, apart for being interesting for constraining the physics of
666: magnetars, offers a unique way of detecting AXPs throughout the Galaxy,
667: since the soft X-ray emission suffers from high absorption, especially in the inner
668: Galaxy. A focussing hard X-ray
669: instrument (like the NASA mission concept {\it NuSTAR})
670: would have the capability of detecting every magnetar in the Galaxy, provided their
671: hard X-ray emission is generic, even in quiescence.
672:
673: \subsection{Optical and Infrared Spectra}
674: \label{sec:spectra_optir}
675:
676: Currently, five of the known AXPs have been conclusively detected in the near-IR, with only
677: 4U 0142+61 (the closest, least absorbed AXP) having been detected optically.
678: None of the SGRs has been detected optically, and only one has been seen
679: in the near-IR \citep[SGR 1806$-$20;][]{kot05}, and only during a particulary active phase.
680: For a summary of these detections, see \\
681: www.physics.mcgill.ca/$\sim$pulsar/magnetar/main.html and references therein.
682:
683: The near-IR spectra of AXPs are an interesting puzzle. First, given the
684: variability in the near-IR (\S\ref{sec:ir_variability}), piecing together an accurate instantaneous
685: spectrum using photometry requires contemporaneous observations, not always
686: possible. Even more of a problem has been the generally unknown reddening
687: toward the sources, which have an enormous impact on the inferred intrinsic
688: spectrum. Nevertheless, some information regarding the optical and near-IR spectra
689: of AXPs is known.
690: Overall, the major mystery is how the optical and near-IR emission connects with the X-ray
691: spectrum. The blackbody emission seen in X-rays grossly underpredicts that in optical/near-IR,
692: while a simple extrapolation of the non-thermal component (when the spectrum is described
693: in this way -- see \S\ref{sec:xray_spectra}) generally overpredicts it.
694: Expecting at least the optical emission to connect spectrally
695: with the X-rays is reasonable given that the latter is pulsed in 4U~0142+61 \citep{km02,dmh+05}
696: hence seems likely to originate in the magnetosphere, as does, presumably, the non-thermal X-ray emission.
697:
698: Very recently, \citet{wck06} have shown using {\it Spitzer} data of 4U~0142+61 that in the far-IR,
699: there appears
700: to be a component that is spectrally distinct from the near-IR emission. They interpret this
701: component as resulting from a passive debris disk irradiated by the central X-ray source. They suggest
702: that such disks, which originate from matter that falls back following the supernova explosion,
703: may be ubiquitous around neutron stars. They further suggest that the correlated near-IR/X-ray flux decay
704: observed by \citet{tkvd04} following the 2002 outburst of 1E~2259+586 and for XTE J1810$-$197 (see \S 3.5)
705: could be a result of
706: a disk around that AXP as well, a possibility also discussed by \citet{ea06} and \"U. Ertan et al.
707: in
708: these proceedings. If the fallback-disk interpretation is correct for 4U 0142+61, the pulsed
709: X-ray flux change recently detected in this source (\S\ref{sec:flares}) may provide
710: an opportunity for a test of the disk hypothesis \citep[][and R. Dib et al. in this volume]{dkg06},
711: as there should be a correlated increase in the near-IR flux. This idea requires that the overall
712: X-ray flux, not just the pulsed component, also be increasing, which requires observations with an
713: imaging X-ray telescope to verify.
714:
715:
716: \subsection{Radio Spectrum}
717:
718: Very recently (in fact {\it after} this meeting took place!), \citet{crh+06} reported
719: the detection of radio pulsations from XTE~J1810$-$197. This was a magnetar first and
720: in many ways a welcome discovery, having provided a nice radiative link between magnetars
721: and radio pulsars, in addition to the similarity already established from timing
722: observations (see \S 2). It also demonstrated that pulsed radio emission can definitely be
723: produced in magnetar-strength fields in contrast to some predictions \citep[e.g.][]{bh98}.
724: Previous searches of other non-transient AXPs had come up empty \citep[e.g.][and see
725: contribution by M. Burgay et al., this volume]{mri+06,chk06}, suggesting the radio emission here
726: might somehow be related to this source's transient nature. Given the small radio beaming
727: fraction reasonably expected for such slow pulsars, the absence of radio pulsations from
728: other sources could also be due to small-number statistics.
729:
730: Also interesting is the unusual spectrum of the radio emission seen from XTE~J1810$-$197.
731: It has an unusually flat spectrum, with spectral index $>-0.5$, whereas radio pulsars have
732: very steep spectra, with most indices between $-$1 and $-3$. XTE~J1810$-$197 is the brightest radio
733: pulsar known at frequencies $>$20~GHz. Why this should be the case is an interesting new puzzle for
734: magnetar physics, one which has the potential to shed crucial new light on the long-standing
735: problem of the origin of radio emission in rotation-powered pulsars.
736:
737: \subsection{Spectral Features}
738:
739: Finally, there have been reports of features in AXP spectra. The first such report
740: was for 1E~1048.1$-$5937 during the first of its two observed 2001 bursts \citep{gkw02}. The
741: feature, which was most prominent in the first 1-s of the burst, was extremely
742: broad and seen apparently in emission at a central
743: energy of 14 keV (see Fig. \ref{fig:1048feature}). In terms of flux, it was comparable to the burst continuum
744: emission. It was very significant; Monte Carlo simulations showed that such
745: a feature at any energy would be seen only 0.01\% of the time. \citet{gkw06} saw a similar
746: feature in the third observed 1E~1048.1$-$5937 burst.
747: In that case, the central energy measured was also $\sim$13~keV.
748: In addition, \citet{wkg+05} observed a similar feature in one burst from XTE~J1810$-$197,
749: this time at energy 12.6~keV, with probability of chance occurence $4\times 10^{-6}$.
750: If interpreted as proton cyclotron lines, the energies of these features imply a magnetic
751: field above $\sim 10^{14}$~G, consistent with the magnetar hypothesis. However
752: if the features are interpreted as electron cyclotron lines, the implied field is correspondingly $\sim$2000 times lower.
753: The latter would not necessarily be inconsistent with the magnetar interpretation, as it
754: is unclear what altitude above the neutron-star surface these lines originate.
755:
756: \begin{figure}
757: \centering
758: \includegraphics[width=0.3\textwidth]{1048feature.ps}
759: \caption{Spectrum of the 1 s after the peak of the first burst observed from
760: 1E 1048.1$-$5937 \citep{gkw02}.}
761: \label{fig:1048feature}
762: \end{figure}
763:
764: Note that similar spectral features during SGR bursts have also been reported in the tails
765: of a few SGR bursts \citep[][see
766: contribution by A. Ibrahim et al., these proceedings]{iss+02,isp03}. However,
767: recently their statistical significance has been questioned (see contribution by
768: P. Woods et al.).
769:
770: \citet{roz+05} reported spectral features at certain rotational phases from
771: 1RXS 170849.0$-$400910 from {\it BeppoSax} data. These were claimed at the time
772: to be significant at the $\sim 4\sigma$ level.
773: Observations with XMM of the same source saw no such features, implying either that
774: the {\it BeppoSax} features were spurious or that the effect is time variable.
775: See the contribution by N. Rea et al. in these proceedings for more information.
776:
777:
778: \section{Conclusions}
779: \label{sec:conclusions}
780:
781: My hope in writing this review is to demonstrate that there remain many
782: important unsolved problems in the study of AXPs. Overall, the basic issue of
783: what differentiates AXPs from SGRs remains, as does the origin of the intense
784: magnetic fields inferred.
785: Other important issues for which there simply was not enough space for
786: discussion here include the possible association of AXPs (and SGRs) with massive
787: star progenitors \citep[e.g.][see contribution by B. Gaensler et al.]{fng+05,mcc+06},
788: the puzzling lack of ``anomalously" bright X-ray emission from high-magnetic field radio
789: pulsars (see contribution by M. Gonzalez et al.), and the proposed connection
790: between magnetars and the so-called ``RRATS'' \citep[][and see contribution
791: by A. Lyne in this volume]{mll+06}.
792:
793: As for how some of these problems will be solved, continued monitoring observations with
794: RXTE, as well as targetted studies with Chandra, XMM and INTEGRAL, will
795: obviously be of use. Greater concerted efforts in the optical and near- and far-IR
796: are warranted, as are careful and repeated radio searches for transient pulsations
797: or other phenomena. Finally, target-of-opportunity observations at the times of the
798: rare moments of AXP activity are definitely crucial for unravelling the overall physical
799: picture of these very interesting sources.
800:
801:
802: \begin{acknowledgements}
803: I thank C. Tam for maintaining the online McGill Magnetar Catalog
804: as well as R. Dib, F. Gavriil, and P. Woods for useful conversations.
805:
806: \end{acknowledgements}
807:
808: % BibTeX users please use
809: %\bibliographystyle{spmpsci}
810: \bibliographystyle{apj2}
811: %\bibliography{journals,psrrefs,crossrefs,vk_modrefs,temp_refs} % name your BibTeX data base
812: \bibliography{journals1,crossrefs,modrefs,psrrefs} % name your BibTeX data base
813:
814:
815: \end{document}
816: % end of file template.tex
817: