1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3: %\usepackage{apjfonts}
4:
5: %\journalinfo{}
6: \slugcomment{To appear in The Astrophysical Journal}
7:
8: \shortauthors{HALPERN ET AL.}
9: \shorttitle{OUTBURST OF AXP 1E~1547.0$-$5408}
10:
11: \begin{document}
12:
13: % Definitions
14:
15: % Objects
16: \def\mag{1E~1547.0$-$5408}
17: \def\psr{PSR~J1550$-$5418}
18: \def\snr{G327.24$-$0.13}
19: \def\xte{XTE~J1810$-$197}
20: % Telescopes
21: \def\asca{{\em ASCA\/}}
22: \def\chandra{{\em Chandra\/}}
23: \def\einstein{{\em Einstein\/}}
24: \def\swift{{\em Swift\/}}
25: \def\xmm{{\em XMM-Newton\/}}
26:
27:
28: \title{Outburst of the 2 s Anomalous X-ray Pulsar 1E~1547.0--5408}
29:
30: \author{J.~P. Halpern,\altaffilmark{1}
31: E.~V. Gotthelf,\altaffilmark{1}
32: J.~Reynolds,\altaffilmark{2}
33: S.~M.~Ransom,\altaffilmark{3}
34: F.~Camilo\altaffilmark{1}
35: }
36:
37: \altaffiltext{1}{Columbia Astrophysics Laboratory, Columbia University,
38: New York, NY 10027.}
39: \altaffiltext{2}{Australia Telescope National Facility, CSIRO, Parkes
40: Observatory, Parkes, NSW 2870, Australia.}
41: \altaffiltext{3}{National Radio Astronomy Observatory, Charlottesville,
42: VA 22903.}
43:
44: \begin{abstract}
45: Following our discovery of radio pulsations from the newly
46: recognized Anomalous X-ray Pulsar (AXP) \mag, we initiated
47: X-ray monitoring with the \swift\ X-ray Telescope,
48: and obtained a single target-of-opportunity
49: observation with the {\em Newton X-ray Multi-Mirror Mission}
50: (\xmm). In comparison with its historic minimum flux of
51: $3 \times 10^{-13}$ ergs cm$^{-2}$ s$^{-1}$,
52: the source was found to be in a record high state,
53: $f_X(1$--$8\,\mbox{keV}) = 5 \times 10^{-12}$ ergs cm$^{-2}$ s$^{-1}$,
54: or $L_X = 1.7 \times 10^{35}(d/9\ {\rm kpc})^2$ ergs~s$^{-1}$,
55: and declining by 25\% in 1 month.
56: Extrapolating the decay, we bound the total energy in this outburst
57: to $10^{42} < E < 10^{43}$ ergs.
58: The spectra (fitted with a Comptonized blackbody)
59: show that an increase in the temperature and area
60: of a hot region, to 0.5~keV and $\sim 16\%$
61: of the surface area of the neutron star, respectively,
62: are primarily
63: responsible for its increase in luminosity.
64: The energy, spectrum, and timescale of decay are consistent with
65: a deep crustal heating event, similar to an interpretation of the
66: X-ray turn-on of the transient AXP \xte.
67: Simultaneous with the 4.6 hour \xmm\ observation,
68: we observed at 6.4~GHz with the Parkes telescope,
69: measuring the phase relationship of the radio and X-ray pulse.
70: The X-ray pulsed fraction of \mag\ is only $\sim 7\%$,
71: while its radio
72: pulse is relatively broad for such a slow pulsar,
73: which may indicate a nearly aligned rotator.
74: As also inferred from the transient behavior of \xte, the only other AXP
75: known to emit in the radio, the magnetic field rearrangement responsible
76: for this X-ray outburst of \mag\ is probably the cause of its radio
77: turn-on.
78:
79: \end{abstract}
80:
81: \keywords{ISM: individual (G327.24--0.13) --- pulsars: individual
82: (1E~1547.0--5408, PSR~J1550--5418, XTE~J1810--197) --- stars: neutron}
83:
84: \section{Introduction}\label{sec:intro}
85:
86: Anomalous X-ray pulsars (AXPs) and soft gamma-ray repeaters (SGRs)
87: are young neutron stars (NSs) with rotation period of 2--12\,s and inferred
88: surface magnetic field strength $B \approx 10^{14-15}$\,G. See
89: \citet{wt06} and \citet{kas07} for recent reviews. In the magnetar model
90: \citep[][1996]{dt92a,td95}\nocite{td96a}, the rearrangement and decay
91: of their extreme fields is responsible for their large and variable X-ray
92: luminosity, which exceeds that available from rotational braking.
93: Thirteen magnetars are confirmed\footnote{Nine
94: AXPs and four SGRs; there are two more candidates. See catalog at
95: http://www.physics.mcgill.ca/$\sim$pulsar/magnetar/main.html.}, of which
96: the AXP \mag\ is the most recent.
97: Discovered with the \einstein\ X-ray satellite in 1980 \citep{lam81},
98: \mag\ was
99: identified as a magnetar candidate in the center of the small candidate
100: SNR \snr\ by \citet{gg07}. It was subsequently observed to be emitting radio
101: pulsations at a period of $P=2$~s \citep{cam07b}, with spin-down properties
102: of a magnetar ($\dot P = 2.3 \times 10^{-11}, B = 2.2 \times 10^{14}$~G),
103: and a distance estimate of 9~kpc from its dispersion measure (DM).
104: \mag\ (\psr) and the 5.5\,s AXP \xte\ are the only magnetars known to emit in
105: the radio \citep{crh+06}. Both are demonstrably transient radio sources,
106: having not been detected in previous surveys of adequate sensitivity.
107:
108: Here, we report on new X-ray observations of \mag\ made shortly after
109: the radio discovery of its pulsations,
110: and compare them with archival X-ray data
111: in order to understand the nature of its apparent X-ray outburst
112: from its spectral and pulse properties.
113: Radio pulsar coverage was also obtained at the same time as one of the
114: X-ray observations, and the phase relationship of the radio and X-ray
115: pulse was determined.
116:
117: \section{X-ray Observations}\label{sec:obs}
118:
119: \subsection{Swift}
120:
121: Following the detection of radio pulsations from \mag\ on 2007 June 8,
122: seven observations were made with the \swift\
123: X-ray telescope \citep[XRT;][]{geh04,bur05} between 2007 June 22 and July 30.
124: The data were taken in photon counting mode, which provides imaging with
125: $18''$ half-power diameter resolution and 2.5\,s time sampling.
126: Neither the counts nor the time resolution
127: were sufficient to detect pulsations.
128: We used the archived event files
129: with standard screening criteria applied.
130: A total of 22.4~ks of exposure was obtained, although
131: 2.6~ks of this had incorrect attitude solution; the latter was not
132: used for spectral analysis.
133: The source was detected in the range 1--8\,keV with an initial count
134: rate of 0.095\ s$^{-1}$, and a systematic decrease of about 30\%
135: over the 40 day interval (see Fig.~\ref{fig:decay}).
136: The seven \swift\ observations were summed in order to accumulate
137: sufficient statistics for a useful spectral fit.
138: The spectral modeling is discussed in \S\ref{sec:spectra}.
139:
140: The maximum observed flux,
141: $\approx 5.2 \times 10^{-12}$\,ergs\,cm$^{-2}$\,s$^{-1}$ in the
142: 1--8~keV band on 2007 June 22, is 16 times greater
143: than the minimum seen in 2006 July--August, and 2.6 times greater than
144: the previous highest flux \citep{gg07}.
145: The small dynamic range of the data in Figure~\ref{fig:decay}
146: prevents a characterization of the decay curve, whether
147: linear, exponential, or power law.
148: Additional observations were made by \swift\
149: on 2007 September 28 -- October 1, after a 2 month gap caused by
150: a gyroscope anomaly, and finally on October 26.
151: These show the flux leveling off and
152: possibly rising again. The proximity
153: of the Sun prevents further observations until early 2008.
154:
155: \begin{figure}[t]
156: \begin{center}
157: \includegraphics[angle=0,scale=0.45]{f1.eps}
158: \caption{\label{fig:decay}
159: Absorbed, $1-8$~keV
160: X-ray flux of \mag\ beginning on 2007 June 22 from the \swift\ XRT
161: ({\it filled circles}) and including \xmm\ on 2007 August 9
162: ({\it open circle}). The solid line is fitted only to the early
163: \swift\ points. The dashed line indicates the minimum historical flux
164: level, which was seen by \chandra\ and \xmm\ in 2006 July and August
165: \citep[][and Table~\ref{tab:xrayspec}]{gg07}.
166: }
167: \end{center}
168: \end{figure}
169:
170: \subsection{XMM-Newton}
171:
172: The results of all archival X-ray observations of \mag\
173: were published by \citet{gg07}. Here, we reanalyze their
174: long (46~ks) \xmm\ observation of 2006 August 21, which found
175: the source in its lowest observed state,
176: in order to compare the components of its
177: X-ray spectrum with the high state and to make a
178: more sensitive search for pulsations at the now known period.
179: We reprocessed the observation data files using the {\tt emchain}
180: and {\tt epchain} scripts under Science Analysis System (SAS) version
181: xmmsas\_20060628\_1801-7.0.0. The 73.4~ms sampling of the
182: European Photon Imaging Camera pn CCD \citep[EPIC pn;][]{tur03}
183: in ``full frame'' mode was adequate to search for pulsations from \mag.
184: The 0.9~s sampling of the two EPIC MOS cameras,
185: operated in ``large window'' mode, is close to the
186: Nyquist frequency for the 2~s pulsar.
187:
188: On 2007 August 9 we obtained a new \xmm\ observation.
189: The exposure time was 4.2\,hr in the EPIC pn and 4.6\,hr
190: in the EPIC MOS. The pn CCD was
191: operated in ``large window'' mode, with 48~ms sampling.
192: The two MOS CCDs were operated in ``small window''
193: mode, with time sampling of 0.3~s, in order to be used to
194: study the pulsed light curve. We processed the data using
195: the SAS version mentioned above. This observation came 10 days
196: after a string of \swift\ observations, and its flux falls
197: on the linear decay fitted to the prior \swift\ points
198: (see Fig.~\ref{fig:decay}). Spectral fitting
199: of the \xmm\ observations is discussed in \S\ref{sec:spectra}
200: and timing analysis for pulsations in \S\ref{sec:pulse}.
201:
202: \section{X-ray Spectral Fitting}\label{sec:spectra}
203:
204: We first tried fitting a single blackbody to the spectra
205: from \xmm\ and \swift. A single blackbody is an inadequate fit
206: in either high or low states. It leaves a deficit
207: of flux at 1 keV, and an excess at the highest energies, $>5$~keV.
208: The low-energy end of the spectrum may be affected by an unmodeled
209: proton cyclotron absorption line, of resonance energy
210: $E_c = 0.63(1+z)^{-1}(B/10^{14}\,{\rm G})$~keV, where $z$ is the
211: gravitational redshift and $(1+z) = (1-2GM/c^2R)^{-1/2} \approx 1.3$.
212: For $B = 2.2 \times 10^{14}$~G, the line would appear at
213: $\approx 1$~keV. However, there is not enough coverage of the
214: continuum at low energies to model it.
215:
216: Similarly, the high-energy end of the spectrum is poorly characterized.
217: However, we can fit the high-energy excess with a
218: simple Comptonization model that introduces only
219: one additional parameter. In this model, surface thermal photons
220: are multiply scattered by relativistic electrons of mean energy
221: $\gamma mc^2$ and optical depth $\tau_{\rm es}$ \citep{poz76,ryb79}.
222: The mean energy boost factor per scattering is
223: $A=4\langle\gamma^2\rangle/3$.
224: In the limit $\tau_{\rm es}< 1$, an analytic approximation for the
225: spectrum can be derived that is a function only of the surface blackbody
226: temperature $T_{\rm BB}$ and the combined scattering parameter
227: $\alpha = -{\rm ln}\tau_{\rm es}/{\rm ln}A$.
228: The scattered spectrum resembles the original blackbody, plus a tail that
229: at high energies approaches a power law of photon index
230: $\Gamma = 2 + \alpha$.
231: This model was also used by \cite{tie05} to fit
232: spectra of another AXP, although they
233: derived (in our notation) $\Gamma = 1 + \alpha$.
234:
235: \begin{figure}[t]
236: \begin{center}
237: \includegraphics[angle=270,scale=0.35]{f2.eps}
238: \caption{\label{fig:xrayspec}
239: Spectra of \mag\ from the \xmm\ observation of 2007 August 9,
240: showing EPIC pn and MOS separately,
241: and the summed \swift\ XRT observations from 2007 June 22 -- July 30.
242: The fit is to the Comptonized blackbody model described in
243: the text, with parameters listed in Table~\ref{tab:xrayspec}.
244: }
245: \end{center}
246: \end{figure}
247:
248: \begin{deluxetable*}{lccc}
249: %\rotate
250: \tablecolumns{4}
251: \tablewidth{320pt}
252: \tablecaption{Comptonized Blackbody Spectral Fits}
253: \tablehead{
254: \colhead{} & \colhead{2006 Aug 21} & \colhead{2007 Jun 22--Jul 30} & \colhead{2007 Aug 9} \\
255: \colhead{Parameter} & \colhead{{\it XMM-Newton}} & \colhead{{\it Swift} average} & \colhead{{\it XMM-Newton}}
256: }
257: \startdata
258: %\multispan4{\hfill \hbox{Comptonized Blackbody Model}\hfill \vspace{5pt}}\\
259: %\tableline
260: $N_{\rm H}$ ($10^{22}$ cm$^{-2}$)\dotfill & $2.84 \pm 0.60$ & 3.12 (fixed) & $3.12 \pm 0.23$ \\
261: $kT_{\rm BB}$ (keV)\dotfill & $0.40 \pm 0.06$ & $0.50 \pm 0.03$ & $0.52 \pm 0.03$ \\
262: $\alpha$\dotfill & $2.10^{+0.76}_{-0.38}$ & 2.93 (fixed) & $2.93^{+0.97}_{-0.48}$ \\
263: $A_{\rm BB}$ (cm$^2$)\dotfill & $3.6 \times 10^{11}$ & $17.6 \times 10^{11}$ & $9.6 \times 10^{11}$ \\
264: $F$ (ergs cm$^{-2}$ s$^{-1}$)\tablenotemark{a}\dotfill & $3.3 \times 10^{-13}$ & $4.6 \times 10^{-12}$ & $3.2 \times 10^{-12}$ \\
265: $L_{\rm BB}$ ($10^{34}$ ergs s$^{-1}$)\tablenotemark{b}\dotfill & 0.95 & 11.3 & 7.2 \\
266: $L_{\rm tot}$ ($10^{34}$ ergs s$^{-1}$)\tablenotemark{c}\dotfill & 1.39 & 15.2 & 9.7 \\
267: $\chi^2_{\nu}({\nu})$\dotfill & 0.99(117) & 1.1(27) & 0.77(252) \\
268: %\cutinhead{Power-law~Model}
269: \enddata
270: \label{tab:xrayspec}
271: \tablecomments{\footnotesize Uncertainties are 90\% confidence intervals for three interesting parameters.
272: Luminosities and areas are computed for $d=9$~kpc, assuming isotropic flux.}
273: \tablenotetext{a}{\footnotesize Absorbed flux in the $1-10$ keV band.}
274: \tablenotetext{b}{\footnotesize Unabsorbed, bolometric blackbody luminosity.}
275: \tablenotetext{c}{\footnotesize Unabsorbed total luminosity including Comptonized component.}
276: \end{deluxetable*}
277:
278: While this approach neglects effects of the magnetic field,
279: which are treated by the more realistic
280: resonant cyclotron scattering model \citep{lyu06,fer07,rea07a,guv07},
281: the effects on the spectrum are similar, and our Comptonized
282: blackbody model is
283: in fact equivalent to equation (58) of \citet{fer07},
284: which they used to fit their own Monte Carlo spectral results.
285: Either scattering model is more
286: meaningful than the ``traditional'' pure power-law plus blackbody fit,
287: because there is no physical mechanism that extends such a power law
288: to photon energies less than the seed blackbody photons.
289:
290: The results of spectral fitting are summarized in
291: Table~\ref{tab:xrayspec} and Figure~\ref{fig:xrayspec}.
292: Acceptable values of $\chi^2$ and reasonably small errors on the
293: three fitting parameters $N_{\rm H}, kT_{\rm BB}$, and $\alpha$ were
294: obtained from both \xmm\ observations. In the case of \swift,
295: where fewer photons were collected, we could obtain a small
296: error on $kT_{\rm BB}$, but $\alpha$ was not constrained.
297: We therefore fixed $N_{\rm H}$ and $\alpha$ in the fitting
298: of the \swift\
299: spectrum to their nearly contemporaneous values measured by \xmm\
300: in 2007.
301:
302: From the average \swift\ spectrum of 2007 June--July, most of the
303: flux is fitted by a blackbody whose bolometric luminosity is
304: $L_{\rm BB} \approx 1.1 \times 10^{35}(d/9\ {\rm kpc})^2$ ergs~s$^{-1}$,
305: and the corresponding area is
306: $A_{\rm BB}\approx 1.8 \times 10^{12}(d/9\ {\rm kpc})^2$\,cm$^2$,
307: or $\sim 14\%$ of the NS surface. In addition, there is a Compton
308: scattered component of
309: $L_s \approx 3.9 \times 10^{34}(d/9\ {\rm kpc})^2$ ergs~s$^{-1}$.
310: Together, these components account for the total luminosity
311: $L_{\rm tot} \approx 1.5 \times 10^{35}(d/9\ {\rm kpc})^2$ ergs~s$^{-1}$
312: in Table~\ref{tab:xrayspec}. Relative to these averages of
313: 2007 June--July, the maximum luminosity and area recorded by
314: the first \swift\ observation of 2007 June 22 are $\approx 13\%$ higher, i.e.,
315: $L_{\rm BB, max} \approx 1.3 \times 10^{35}(d/9\ {\rm kpc})^2$ ergs~s$^{-1}$,
316: $A_{\rm BB, max}/A_{\rm NS} \approx 0.16$, and
317: $L_{s,{\rm max}} \approx 4.4 \times 10^{34}(d/9\ {\rm kpc})^2$ ergs~s$^{-1}$.
318: Smaller values are then obtained from the subsequent \xmm\ observation
319: on 2007 August 9, which, within the errors, are consistent with a
320: decrease of the blackbody area at a constant temperature.
321: We fitted individual \swift\ observations to look
322: for more subtle changes in spectral shape.
323: Within the larger uncertainties of the individual
324: spectra, there is no evidence for spectral changes
325: from 2007 June 22 through October 26.
326:
327: The \xmm\ and \swift\ spectra from 2006 August and 2007 June,
328: respectively, show that a combination
329: of the increase in the blackbody temperature from
330: 0.4~keV to 0.5~keV and an increase of the area of that
331: hot region by a factor of 5 is primarily
332: responsible for the dramatic increase in X-ray flux.
333: There is no evidence for a
334: major change in the efficiency of magnetospheric scattering
335: between the two states, as the scattering parameter $\alpha$
336: remained the same to within errors. Only $26-32\%$ of the
337: X-ray luminosity is in the Compton scattered component in
338: any state, as
339: can be seen by comparing $L_{\rm tot}$ and $L_{\rm BB}$
340: in Table~\ref{tab:xrayspec}.
341:
342: \section{X-ray Pulsations}\label{sec:pulse}
343:
344: Only the \xmm\ observations of \mag\ have sufficient photon statistics
345: and time resolution to search for pulsations.
346: The event times were converted to Barycentric
347: Dynamical Time (TDB) using the precise position of the pulsar,
348: (J2000) R.A. = $15^{\rm h}50^{\rm m}54.11^{\rm s}\pm0.01^{\rm s}$,
349: Decl. $-54\arcdeg18'23.7''\pm0.1''$
350: \citep{cam07b}. Events were extracted from the 2007 August
351: \xmm\ observation in a $30^{\prime\prime}$ radius around the
352: source in the EPIC pn CCD, and $25^{\prime\prime}$ in the MOS CCDs.
353: Photons in the range $0.5-8$~keV from the three detectors were
354: combined, and searched
355: for pulsations using the $Z_n^2$ test. A signal was found
356: consistent with the contemporaneous radio period
357: with $Z_1^2 = 33.9$, $Z_2^2 = 40.7$, and $Z_3^2 = 43.4$.
358: The $Z_1^2$ power spectrum is shown in Figure~\ref{fig:power}.
359: These values correspond to a chance probability of $\sim 10^{-7}$.
360: The power contributed by the harmonics is indicative of a pulse
361: that is narrower than a sinusoid.
362: From the $Z_3^2$ search, we derive $P = 2.069961(17)$~s ($1 \sigma$),
363: compared with $P = 2.0699348(5)$~s in the radio
364: (see \S \ref{sec:parkes}).
365:
366: Figure~\ref{fig:power} also
367: shows the $1-6$~keV X-ray pulse folded at the exact radio period
368: (determined in \S \ref{sec:parkes}).
369: We note that, while the 48~ms
370: time sampling of the pn is more than sufficient to resolve
371: the pulse, the 0.3~s sampling of the MOS also adequately
372: resolves the observed structure.
373: The X-rays are weakly modulated with pulsed amplitude
374: $\sim 7\%$, defined as the fraction of total counts above the
375: minimum of the light curve.
376: This is small compared to most AXPs, and comparable
377: only to the small amplitude of 4U~0142+614 \citep{rea07b,gon07}.
378: The latter has a complex pulse shape, which is energy dependent
379: as well as varying with luminosity.
380: Within the limited statistics available for \mag,
381: we do not have evidence for energy dependence of its pulse shape.
382:
383: \begin{figure}[t]
384: \begin{center}
385: \includegraphics[angle=270,scale=0.35]{f3a.eps}
386: \includegraphics[angle=270,scale=0.35]{f3b.eps}
387: \caption{\label{fig:power}
388: $Z_1^2$ power spectra of the combined \xmm\
389: EPIC pn and MOS observations of \mag.
390: The reference frequency, 0.4831049 Hz, is the radio measured value on
391: 2007 August 9. The range of frequencies searched corresponds to
392: $\dot P \leq 2.8 \times 10^{-10}$ between the two epochs,
393: or $\le 12$ times the discovery value.
394: The highest peak on 2006 August 21, if real,
395: requires an average
396: $\dot P = 2.8 \times 10^{-11}$ between the two epochs.
397: {\it Insets\/}: X-ray pulse profiles in the
398: optimal $1-6$~keV band, normalized to a mean
399: count rate of 1. Background has been
400: subtracted from nearby regions of the CCD.
401: The 2006 August 21 profile is from the EPIC pn only,
402: while the 2007 August 9 profile is from the EPIC pn and MOS.
403: Phase zero of the 2006 August 21 profile is arbitrary.
404: The 2007 August 9 profile is folded on the
405: contemporaneous radio ephemeris (see Fig.~\ref{fig:radioxraypulse}).
406: }
407: \end{center}
408: \end{figure}
409:
410: A search of the 2006 August 21 \xmm\ observation of \mag\ at the extrapolated
411: radio period was less sensitive because (1) the source was in a low
412: flux state at the time, (2) the 0.9~s sampling of the MOS detectors
413: in large window mode is close to the Nyquist frequency, and (3) the lack of
414: a contemporaneous radio ephemeris required a search over several
415: dozen independent trial periods. The latter is especially problematic,
416: as the period derivative of a magnetar can vary by
417: a factor of several on a timescale of 1 year \citep[e.g.,][]{gav04}.
418: Allowing for $0 < \dot P < 2.8 \times 10^{-10}$,
419: up to 12 times its discovery value of $2.3 \times 10^{-11}$
420: \citep{cam07b},
421: requires a search of $\approx 160$ independent periods in
422: the 2006 August observation of \mag.
423:
424: We extracted events from an optimized radius of
425: $20^{\prime\prime}$ around the source in the 2006 August EPIC pn
426: image, and applied the $Z_1^2$ test. The $1-6$~keV
427: band was chosen to optimize the signal-to-noise.
428: The strongest signal
429: in the $\dot P$ search range described above yielded
430: $Z_1^2=18.1$ and corresponds to $\dot P = 2.8 \times 10^{-11}$,
431: which is close to its 2007 June value of $\dot P = 2.3 \times 10^{-11}$,
432: and even closer to the latest value from
433: continued radio timing, $\dot P = 2.9 \times 10^{-11}$
434: (F. Camilo et al., in preparation).
435: The light curve folded at the corresponding period of
436: 2.069067(12)~s is quasi-sinusoidal with a pulsed fraction of
437: $\approx 15\%$ (see Fig.~\ref{fig:power}).
438: The overall statistical significance of this
439: signal is difficult to estimate because the range of
440: $\dot P$ considered is somewhat arbitrary, but it
441: is probably 98--99\%. When we add the source photons
442: from the MOS detectors to the search, $Z_1^2$ increases
443: to 24.0 (see Fig.~\ref{fig:power}),
444: which is encouraging of the reality of the signal,
445: as its significance rises to $\approx 99.9\%$ while the period
446: remains consistent, $P=2.069059(10)$~s.
447: Therefore, we consider this a tentative detection of pulsations
448: on 2006 August 21.
449:
450: \begin{figure}[t]
451: \begin{center}
452: \includegraphics[angle=0,scale=0.43]{f4.eps}
453: \caption{\label{fig:radioxraypulse}
454: Alignment of the simultaneously observed radio
455: and X-ray pulses on 2007 August 9, folded at
456: 0.4831049 Hz. Phase zero corresponds to MJD 54321.44 TDB.
457: The radio (DFB) total intensity profile is displayed in 512 bins, and
458: the small feature at phase $\approx 0.63$ is real, also being present
459: in analog filterbank data.
460: }
461: \end{center}
462: \end{figure}
463: \section{Contemporaneous Radio Pulsar Observation} \label{sec:parkes}
464:
465: We observed \mag\ with the Parkes telescope in Australia on 2007 August~9,
466: overlapping with the \xmm\ observation. We collected data for 4.3\,hr
467: beginning at 09:57 UT, 16\,min before the start of the X-ray observation.
468: We used the central beam of the methanol multibeam receiver and recorded
469: data using two different spectrometers. With an analog filterbank,
470: we sampled a bandwidth of 576\,MHz centered on 6.6\,GHz. Each of 192
471: polarization-summed frequency channels was sampled every 1\,ms with
472: one-bit precision. In parallel, we operated a digital filterbank (DFB)
473: with a bandwidth of 256\,MHz centered on 6.4\,GHz, which folded the data
474: into 2048-bin (1\,ms resolution) profiles in all Stokes parameters.
475:
476: We analyzed the data in standard fashion using the PRESTO, TEMPO, and
477: PSRCHIVE packages. The pulsar was detected at $P = 2.0699438(5)$\,s on
478: MJD~54321.44. In order to compare the radio and X-ray light curves in
479: absolute phase, radio pulse times-of-arrival were converted to TDB using
480: TEMPO, and corrected to infinite frequency using the measured $\mbox{DM}
481: = 830\pm50$\,cm$^{-3}$\,pc \citep{cam07b}. At the high radio frequency
482: used, an uncertainty in DM of 100\,cm$^{-3}$\,pc contributes a negligible
483: 10\,ms uncertainty to the infinite-frequency arrival time.
484:
485: The phase alignment is shown in Figure~\ref{fig:radioxraypulse}.
486: The centroid of the radio pulse lags the X-ray centroid by
487: 0.15 cycles, while the peak of the radio pulse
488: comes 0.19 cycles after the X-ray centroid.
489: There were three events in the radio observation in which
490: the pulse profile suddenly changed for times ranging from
491: 100 to 800 seconds, then returned to its otherwise steady
492: shape (see Fig.~\ref{fig:radiopulse}).
493: This behavior is peculiar to radio emitting AXPs,
494: and is also seen in \xte\ \citep{cam07a}. No changes
495: in X-rays are seen in connection with these events,
496: as the X-ray flux remained constant throughout the
497: observation.
498:
499: \begin{figure}[t]
500: \begin{center}
501: \hbox{
502: \hfil
503: \includegraphics[angle=0,scale=0.475,angle=270.]{f5a.eps}
504: \includegraphics[angle=0,scale=0.5,angle=270.]{f5b.eps}
505: \hfil
506: }
507: \caption{\label{fig:radiopulse}
508: {\it Left\/}: Radio pulse history obtained with the analog filterbank
509: during the 4.3 hr Parkes observation
510: at 6.6 GHz beginning on 2007 August 9 09:57 UT.
511: Note the transitory
512: enhancements on the trailing shoulder of the pulse beginning at
513: 3700 s and at 12,000 s. A fainter event can also be seen at
514: 800 s.
515: {\it Right\/}: Simultaneous count rate from the \xmm\ EPIC pn and MOS.
516: }
517: \end{center}
518: \end{figure}
519:
520: \section{Discussion}\label{sec:disc}
521:
522: \subsection{Distance}
523:
524: The X-ray-measured $N_{\rm H} \approx 3 \times 10^{22}$\,cm$^{-2}$
525: exceeds the total Galactic 21~cm \ion{H}{1} column density of
526: $1.8 \times 10^{22}$~cm$^{-2}$ in the direction of \mag\
527: \citep{dic90}, suggesting that most of the absorption is due
528: to molecular material. The $N_{\rm H}$ equivalent of
529: the CO column density in this direction is
530: $N_{\rm H} = 2N_{\rm H_2} \approx 4 \times 10^{22}$~cm$^{-2}$
531: according to the CO maps of \citet{bro89}. Half
532: of the total atomic plus molecular column is then sufficient to
533: account for the X-ray absorption.
534:
535: The $\mbox{DM} = 830$\,cm$^{-3}$\,pc
536: is consistent with the X-ray-measured $N_{\rm H}$
537: and implies a distance $d \approx 9$\,kpc according to the
538: free-electron model of \citet{cl02}.
539: A smaller $d \approx 4$\,kpc was suggested by \citet{gg07} assuming
540: an association with two nearby star forming regions with \ion{H}{1}
541: measured distances. This smaller distance estimate would
542: place \psr\ in or near the Crux--Scutum spiral arm, while
543: $d = 9$~kpc would be compatible with a location
544: near the Norma spiral arm. At $d \approx 4$\,kpc, the X-ray
545: luminosities and blackbody areas in Table~\ref{tab:xrayspec}
546: would be smaller by a factor of 5.
547:
548: The Galactic longitude of \psr,
549: $327.24^{\circ}$, falls on the tangent of our line of
550: sight to the Norma spiral arm as mapped in CO
551: \citep[see Fig.~9 of][]{bro89}, at a distance of 8.5~kpc.
552: In this direction the bulk of the CO lies at radial
553: velocities ranging from $-35$ to $-95$ km~s$^{-1}$.
554: If \psr\ ever becomes much brighter in the radio,
555: an \ion{H}{1} absorption study
556: may be able to pin down its distance more
557: precisely, since absorption in the Crux-Scutum arm
558: extends only to negative velocities of $-60$ km~s$^{-1}$,
559: while velocity as negative as $-90$ km~s$^{-1}$ would
560: indicate a distance closer to the Norma arm
561: \citep[see Fig.~4 of][]{mcc01}.
562:
563: \subsection{Emission Mechanisms and Geometry}
564:
565: The X-ray spectrum of \mag\
566: is similar to that of other AXPs, having a thermal
567: component that even in a low state is fitted by a blackbody of
568: $kT_{\rm BB} = 0.40$\,keV, which is hotter by at least
569: a factor of 3 than modeled cooling NSs of its same age
570: \citep[e.g.,][]{ykhg02}. This indicates that localized
571: heating of $\sim 3\%$ of the NS surface area
572: by magnetic field decay is the primary source
573: of luminosity in the faintest observed state
574: of \mag, with a smaller contribution to
575: the spectrum from magnetospheric scattering.
576: The remainder of the NS surface is presumably
577: too cool to be detected, given the large distance and
578: intervening absorption of soft X-rays.
579:
580: The 0.15 cycle
581: offset between the radio and X-ray pulses
582: recalls the question of whether radio emission from
583: magnetars arises on open or closed field lines, or
584: both. It is possible that magnetic field lines on which
585: large currents flow and
586: radio emission is generated are anchored in the area of
587: concentrated crustal heating from which the thermal
588: X-ray emission emerges. In this case, the phase lag
589: of the radio may by due to the azimuthal twist of
590: these closed field lines, with the radio emission
591: coming from higher altitude and different azimuth from the X-rays.
592: On the other hand, it is not necessary that the X-ray
593: and radio emission are associated with the same field
594: lines, especially if radio emission is restricted
595: to the open field line bundle while X-ray heating
596: occurs on closed field lines. The present data
597: leave this question open.
598:
599: If we accept the tentative detection of X-ray pulsations in
600: 2006 August, then the pulsed fraction decreased
601: by a factor of 2 while the pulsed flux
602: increased by a factor of 8 during the outburst.
603: Such a trend has been seen in other magnetars,
604: notably 1E 1048.1$-$5037 \citep{tie05,gav06}
605: and CXOU J164710.2$-$455216 \citep{mun07,isr07}.
606: One possible reason for this effect is the growth of the hot
607: region, which occupies a significant fraction of the NS
608: surface in outburst, but not in quiescence. Another explanation
609: involves the changing relative contributions of the
610: surface blackbody and scattered flux. A small surface hot spot
611: may be highly modulated if eclipsed by the rotation of the NS,
612: which may account for the 100\% pulsed
613: fraction of CXOU J164710.2$-$455216 in quiescence. In outburst,
614: enhanced magnetospheric scattering decreases that modulation.
615: In the case of \mag, the Comptonized component that we fitted
616: represents a small fraction ($26-32\%$) of the X-ray flux in
617: any of its states, so scattering probably does not account for a
618: change in its pulsed fraction. However, the increase in
619: heated area to 16\% of the NS surface may have that effect.
620:
621: On the other hand, if we discount the evidence of enhanced
622: pulsed amplitude in 2006, then the small X-ray pulsed
623: fraction of $\sim 7\%$ in the high state
624: (by inference in the low state as well),
625: and the broad radio pulse,
626: suggest that these emitting regions are both close to
627: the axis of rotation, i.e., that \psr\ is nearly
628: an aligned rotator. This is in contrast to the other
629: radio emitting magnetar, \xte. The pulsed fraction of \xte\
630: was larger ($\approx 50\%$) at the peak of the outburst
631: than in quiescence \citep{got04}.
632: This, plus the polarization properties of the radio pulse,
633: led \citet{cam07a} to favor large inclination angles for \xte.
634: The observed characteristics of the
635: quiescent state of \xte, for which only an upper limit of
636: $\approx 24\%$ on its pulsed fraction is known \citep{got04},
637: may also be controlled by its surface thermal emission,
638: since unlike \mag, the pre-outburst spectrum of \xte\ appears
639: to be a cool blackbody covering the entire surface,
640: with no evidence for a surface hot spot.
641:
642: \subsection{Energetics of the Outburst}
643:
644: \mag\ resembles the transient AXP \xte\ in that its maximum X-ray observed
645: luminosity, $\approx 1.7 \times 10^{35}(d/9\ {\rm kpc})^2$\,ergs\,s$^{-1}$,
646: is similar to the steady luminosity of most other AXPs, while \mag\ and
647: \xte\ are usually much fainter. Assuming that the high state
648: observed here belongs to a well-defined event, we can extrapolate
649: the decay to estimate the total integrated energy in the outburst.
650: Unfortunately, the functional form of the decay is not well established
651: due to the small dynamic range of the available data, and the
652: poorly constrained quiescent level. If we adopt the
653: linear decay fitted in Figure~\ref{fig:decay} as the
654: flux to be attributed to the outburst, then
655: the total energy beginning on 2007 June 22 will be
656: $\approx 1 \times 10^{42}(d/9\ {\rm kpc})^2$\,ergs.
657: In comparison, the outburst of \xte\ released
658: $\approx 6 \times 10^{42}$\,ergs \citep{gh07}.
659:
660: However, it is likely that the outburst of \mag\ began before our
661: earliest observation of it. A serendipitous detection of the
662: radio pulsar in a survey on 2007 April 23 is reported by S. Johnston
663: (private communication), from which we infer that the X-ray
664: outburst was already in progress at least two months prior
665: to our first \swift\ observation.
666: The above estimate is then a lower bound
667: on the outburst energy.
668: An upper bound on the total energy can be derived by
669: hypothesizing that the outburst began shortly after 2006 August 22,
670: when \mag\ was last observed in a low state.
671: In this case, the initial
672: luminosity was 3 times that seen on 2007 June 22,
673: and the integrated energy is
674: $E \approx 1 \times 10^{43}(d/9\ {\rm kpc})^2$\,ergs.
675:
676: An additional source of uncertainty in the total energy is the
677: possibly aligned geometry of the pulsar as discussed above.
678: This could cause us to overestimate the luminosity, which
679: was calculated assuming isotropic flux, by a factor of a few.
680: Nevertheless, the estimates are within an order of
681: magnitude of the energy released in the outburst of \xte.
682: The latter was interpreted as cooling following a deep crustal
683: heating event \citep{hal05,guv07}, which is a promising
684: hypothesis for \mag\ as well.
685:
686: CXOU J164710.2$-$455216 in Westerlund~1 \citep{isr07,mun07}
687: is also a transient AXP. Its quiescent flux was a factor
688: of $\sim 100$ fainter than its outburst. The peak luminosity
689: on the first day of the
690: outburst was $\approx 1 \times 10^{35}$ ergs~s$^{-1}$.
691: The flux decay followed a power law of index $-0.28 \pm 0.05$,
692: and the integrated energy in the outburst was
693: $\approx 5 \times 10^{41}$ ergs during the first 130 days \citep{isr07},
694: although it could end up being much higher as the flux
695: remained well above the quiescent level.
696: This behavior was similar to the $t^{-0.22}$
697: decay of the AXP 1E 2259+586 following a bursting episode \citep{woo04}.
698: The flux integral of such a shallow decay is unbounded, which
699: may imply that continuing energy input from the
700: magnetic field dominates the post-burst decay, rather
701: than the simple afterglow of an impulsive event.
702: In contrast to the slow decay of CXOU J164710.2$-$455216
703: and 1E 2259+586, models of cooling after a deep crustal
704: heating event follow approximate power-laws in the range $t^{-0.6}-t^{-1.0}$
705: \citep{lyu02}, while the observed decay of \xte\ was well fitted by an
706: exponential \citep{gh07}.
707:
708: Since there was no detection of X-ray bursting activity from
709: \mag\ prior to its discovery as a radio pulsar, we do not know
710: if a discrete event was responsible for its current outburst.
711: Accordingly, it is not yet clear if the decaying X-ray flux is
712: an afterglow of a sudden injection of energy,
713: or simply one in a series of fluctuations in its continuous
714: heating by magnetic field decay. We do know that earlier
715: X-ray observations of \mag, by \einstein\ and \asca,
716: found it at fluxes intermediate between the low and high levels
717: recorded here \citep{gg07}, suggesting that the recent
718: outburst may not be an isolated event. Indeed its X-ray flux
719: appears to have levelled off, and may even be rising again.
720:
721: \section{Conclusions}
722:
723: Prompted by its discovery as a transient radio pulsar,
724: we obtained new X-ray observations of \psr\
725: that reveal it in the highest state yet observed, with
726: $L_X \approx 1.7 \times 10^{35}$
727: ergs~s$^{-1}$ and declining.
728: The peak of the outburst was not
729: observed in X-rays, but
730: it could have been
731: several times higher. As this is the second
732: AXP to be detected as a transient radio source,
733: after \xte, we infer that magnetic field rearrangement
734: accompanying an X-ray outburst may be a necessary
735: (although not sufficient) condition for radio
736: turn-on of an AXP.
737:
738: We also detected X-ray pulsations from
739: \psr\ for the first time, although the pulsed fraction
740: $\sim 7\%$ is among the smallest observed from AXPs.
741: Together with the properties of the simultaneously
742: observed radio pulse, this may indicate a nearly
743: aligned geometry. In both quiescence and outburst,
744: the X-ray spectrum is dominated by a small blackbody
745: hot spot of $kT_{\rm BB} = 0.4-0.5$~keV,
746: with a lesser Comptonized
747: component, as is the case for most AXPs.
748: The changing luminosity is due mostly to the
749: increase in area of the blackbody, and partly
750: to its increase in temperature. This, plus
751: the energy and time scale of the decay, suggests an origin
752: of the brightening in a deep crustal heating
753: event rather than a change in magnetospheric
754: currents. The historical X-ray luminosities
755: of \mag\ span only a factor of 16, including
756: some intermediate states, which is not as
757: extreme a range as experienced by \xte\
758: and CXOU J164710.2$-$455216 (a factor of $\sim 100$).
759: It is not yet clear if we have sampled the total range of
760: luminosity of \mag, or what the duty cycle and
761: decay time of its largest outbursts are.
762:
763: \acknowledgements
764: This investigation is based on observations obtained with \xmm,
765: an ESA science mission with instruments and contributions directly
766: funded by ESA Member States and NASA.
767: We are grateful to the \swift\ and \xmm\
768: project scientists and staff for the timely scheduling
769: of observations.
770: The Parkes Observatory is part of the Australia Telescope,
771: which is funded by the Commonwealth of Australia for operation
772: as a National Facility managed by CSIRO. This work was
773: supported by NASA grant NNG05GC43G.
774:
775: \begin{thebibliography}{33}
776: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
777:
778: \bibitem[Bronfman et al.(1989)]{bro89}
779: Bronfman, L., Alvarez, H., Cohen, R. S., \& Thaddeus, P. 1989, ApJS, 71, 481
780:
781: \bibitem[Burke et al.(1997)]{bur97}
782: Burke, B. E., Gregory, J., Bautz, M. W., Prigozhin, G. Y., Kissel, S. E.,
783: Kosicki, B. N., Loomis, A. H., \& Young, D. J. 1997, IEEE Trans. Electron
784: Devices, 44, 1633
785:
786: \bibitem[Burrows et al.(2005)]{bur05}
787: Burrows, D. N., et al. 2005, \ssr, 120, 165
788:
789: \bibitem[{{Camilo} {et~al.}(2007{\natexlab{a}}){Camilo}, Cognard, {Ransom},
790: {Halpern}, {Reynolds}, {Zimmerman}, Gotthelf, Helfand, Demorest, Theureau, \&
791: Backer}]{cam07a}
792: {Camilo}, F., et al. 2007{\natexlab{a}}, ApJ, 663, 497
793:
794: \bibitem[Camilo et al.(2007b)]{cam07b}
795: Camilo, F., Ransom, S. M., Halpern, J. P., \& Reynolds, J. 2007b,
796: \apj, 666, L93
797:
798: \bibitem[{{Camilo} {et~al.}(2006){Camilo}, {Ransom}, {Halpern},
799: {Reynolds}, {Helfand}, {Zimmerman}, \& {Sarkissian}}]{crh+06}
800: {Camilo}, F., {Ransom}, S.~M., {Halpern}, J.~P., {Reynolds}, J., {Helfand},
801: D.~J., {Zimmerman}, N., \& {Sarkissian}, J. 2006, Nature, 442,
802: 892
803:
804: \bibitem[{{Cordes} \& {Lazio}(2002)}]{cl02}
805: {Cordes}, J.~M., \& {Lazio}, T.~J.~W. 2002, preprint (astro-ph/0207156)
806:
807: \bibitem[Dickey \& Lockman(1990)]{dic90}
808: Dickey, J. M., \& Lockman, F. J. 1990, \araa, 28, 215
809:
810: \bibitem[{Duncan \& Thompson(1992)}]{dt92a}
811: Duncan, R.~C., \& Thompson, C. 1992, ApJ, 392, L9
812:
813: \bibitem[Fern\'andez \& Thompson(2007)]{fer07}
814: Fern\'andez, R., \& Thompson, C. 2007, \apj, 660, 615
815:
816: \bibitem[Gavriil \& Kaspi(2004)]{gav04}
817: Gavriil, F. P., \& Kaspi, V. M. 2004, ApJ, 609, L67
818:
819: \bibitem[Gavriil et al.(2006)]{gav06}
820: Gavriil, F. P., Kaspi, V. M., \& Woods, P. M. 2006, ApJ, 641, 41
821:
822: \bibitem[Gehrels et al.(2004)]{geh04}
823: Gehrels, N., et al. 2004, \apj, 611, 1005
824:
825: \bibitem[{{Gelfand} \& {Gaensler}(2007)}]{gg07}
826: {Gelfand}, J.~D., \& {Gaensler}, B.~M. 2007, ApJ, 667, 1111
827:
828: \bibitem[Gonzalez et al.(2007)]{gon07}
829: Gonzalez, M. E., Dib, R., Kaspi, V. M., Woods, P. M., Tam, C. R.,
830: \& Gavril, F. P. 2007, \apj, submitted (arXiv:0708.2756)
831:
832: \bibitem[Gotthelf et al.(2004)]{got04}
833: Gotthelf, E. V., Halpern, J. P., Buxton, M., \& Bailyn, C. 2004, ApJ, 605, 368
834:
835: \bibitem[{{Gotthelf} \& {Halpern}(2007)}]{gh07}
836: {Gotthelf}, E.~V., \& {Halpern}, J.~P. 2007, Ap\&SS, 308, 79
837:
838: \bibitem[G\"uver et al.(2007)]{guv07}
839: G\"uver, T., \"Ozel, F., G\"o\u g\"u\c s, E., \& Kouveliotou, C. 2007,
840: ApJ, 667, L73
841:
842: \bibitem[Halpern \& Gotthelf(2005)]{hal05}
843: Halpern, J. P., \& Gotthelf, E. V. 2005, ApJ, 618, 874
844:
845: \bibitem[Israel et al.(2007)]{isr07}
846: Israel, G. L., Campana, S., Dall'Osso, S., Muno, M. P., Cummings, J.,
847: Perna, R., \& Stella, L. 2007, \apj, 664, 448
848:
849: \bibitem[Kaspi(2007)]{kas07}
850: Kaspi, V. M. 2007, Ap\&SS, 308, 1
851: \hfil\break
852:
853: \bibitem[Lamb \& Markert(1981)]{lam81}
854: Lamb, R. C., \& Markert, T. H. 1981, \apj, 244, 94
855:
856: \bibitem[Lyubarsky et al.(2002)]{lyu02}
857: Lyubarsky, Y., Eichler, D., \& Thompson, C. 2002, \apj, 580, L69
858:
859: \bibitem[Lyutikov \& Gavriil(2006)]{lyu06}
860: Lyutikov, M., \& Gavriil, F. 2006, \mnras, 368, 690
861:
862: \bibitem[McClure-Griffiths et al.(2001)]{mcc01}
863: McClure-Griffiths, N. M., Green, A. J., Dickey, J. M., Gaensler, B. M.,
864: Haynes, R. F., \& Wieringa, M H. 2001, \apj, 551, 394
865:
866: \bibitem[Muno et al.(2007)]{mun07}
867: Muno, M. P., Gaensler, B. M., Clark, J. S., de Grijs, R., Pooley, D.,
868: Stevens, I. R., \& Portegies Zwart, S. F. 2007, \mnras, 378, L44
869:
870: \bibitem[\"Ozel \& G\"uver(2006)]{oze06}
871: \"Ozel, F., \& G\"uver, T. 2006, \apj, 659, L141
872:
873: \bibitem[Pozdnyakov et al.(1976)]{poz76}
874: Pozdnyakov, L. A., Sobol, I. M., \& Sunyaev, R. A. 1976, Pis'ma Astron. Zh.,
875: 2, 140
876:
877: \bibitem[Rea et al.(2007a)]{rea07a}
878: Rea, N., Zane, S., Lyutikov, M., \& Turolla, R. 2007a, \apss, 308, 61
879:
880: \bibitem[Rea et al.(2007b)]{rea07b}
881: Rea, N., et al. 2007b, \mnras, 381, 293
882:
883: \bibitem[Rybicki \& Lightman(1979)]{ryb79}
884: Rybicki, G. B., \& Lightman, A. P. 1979, Radiative Processes in Astrophysics
885: (New York: Wiley)
886:
887: \bibitem[{{Thompson} \& Duncan(1995)}]{td95}
888: Thompson, C., \& {Duncan}, R.~C. 1995, MNRAS, 275, 255
889:
890: \bibitem[{Thompson \& Duncan(1996)}]{td96a}
891: ---------. 1996, ApJ, 473, 322
892:
893: %\bibitem[{Thompson {et~al.}(2002)Thompson, Lyutikov, \& Kulkarni}]{tlk02}
894: %Thompson, C., Lyutikov, M., \& Kulkarni, S.~R. 2002, ApJ, 574, 332
895:
896: \bibitem[Tiengo et al.(2005)]{tie05}
897: Tiengo, A., Mereghetti, S., Turolla, R., Zane, S., Rea, N., Stella, L.,
898: \& Israel, G. L. 2005, \aap, 437, 997
899:
900: \bibitem[Turner et al.(2003)]{tur03}
901: Turner, M. J. L., Briel, U. G., Fernando, P., Griffiths, R. G.,
902: \& Villa, G. E. 2003, Proc. SPIE, 4851, 169
903:
904: \bibitem[{Woods \& Thompson(2006)}]{wt06}
905: Woods, P.~M., \& Thompson, C. 2006, in Compact Stellar X-ray Sources, ed.
906: W.~H.~G. Lewin \& M.~van~der Klis (Cambridge: Cambridge Univ. Press),
907: 547
908: \bibitem[Woods et al.(2004)]{woo04}
909: Woods, P. M., et al. 2004, ApJ, 605, 378
910:
911: \bibitem[{{Woods} {et~al.}(2005){Woods}, {Kouveliotou}, {Gavriil}, {Kaspi},
912: {Roberts}, {Ibrahim}, {Markwardt}, {Swank}, \& {Finger}}]{wkg+05}
913: ---------. 2005, \apj, 629, 985
914:
915: \bibitem[{{Yakovlev} {et~al.}(2002){Yakovlev}, {Kaminker}, {Haensel}, \&
916: {Gnedin}}]{ykhg02}
917: {Yakovlev}, D.~G., {Kaminker}, A.~D., {Haensel}, P., \& {Gnedin}, O.~Y. 2002,
918: A\&A, 389, L24
919:
920: \end{thebibliography}
921:
922: \end{document}
923: