1: %% --------------------------------------------------------
2: %\documentclass[preprint2]{aastex}
3: \documentclass[12pt,preprint]{aastex}
4: %% --------------------------------------------------------
5: \newcommand{\Msun}{\ifmmode\mbox{M}_{\odot}\else$\mbox{M}_{\odot}$\fi}
6: \newcommand{\Rsun}{\ifmmode\mbox{R}_{\odot}\else$\mbox{R}_{\odot}$\fi}
7: \newcommand{\Mearth}{\ifmmode\mbox{M}_{\oplus}\else$\mbox{M}_{\oplus}$\fi}
8: \newcommand{\Rearth}{\ifmmode\mbox{R}_{\oplus}\else$\mbox{R}_{\oplus}$\fi}
9: \newcommand{\chandra}{\textit{Chandra}\,}
10: \newcommand{\spitzer}{\textit{Spitzer}\,}
11: %\newcommand{\xmm}{\textit{XMM-Newton}\,}
12: \newcommand{\xmm}{\textit{XMM}\,}
13: \newcommand{\xte}{\textit{RXTE}\,}
14: \newcommand{\rosat}{\textit{ROSAT}\,}
15: \newcommand{\ooft}{4U~0142$+$61\,}
16: \newcommand{\seven}{RXS~J170849.0$-$400910\,}
17: \newcommand{\eight}{1E~1841$-$045\,}
18: \newcommand{\etff}{AX~1845$-$0258}
19: \newcommand{\ttfn}{1E~2259$+$586}
20: \newcommand{\tfe}{1E~1048.1$-$5937\,}
21: \newcommand{\etten}{XTE~J1810$-$197\,}
22: \newcommand{\ninet}{SGR~1900$+$14}
23: \newcommand{\etos}{SGR~1806$-$20\,}
24:
25: \shorttitle{1E 2259+586 radiative evolution}
26: \shortauthors{Zhu, Kaspi, Dib, Woods , Gavriil \& Archibald}
27:
28: %% --------------------------------------------------------
29: %% \slugcomment{Submitted to ApJ}
30: %% --------------------------------------------------------
31: \begin{document}
32: %% --------------------------------------------------------
33: \title{The Long-term Radiative Evolution of Anomalous X-ray Pulsar 1E~2259$+$586 after its 2002 Outburst}
34: %% --------------------------------------------------------
35: \author{Weiwei Zhu\altaffilmark{1},
36: Victoria M. Kaspi\altaffilmark{1,2},
37: Rim Dib\altaffilmark{1},
38: Peter M. Woods \altaffilmark{3,4},
39: Fotis P. Gavriil \altaffilmark{5,6} and
40: Anne M. Archibald \altaffilmark{1}}
41: %% --------------------------------------------------------
42: %% \email{}
43: %% --------------------------------------------------------
44: \altaffiltext{1}{\footnotesize Department of Physics,
45: McGill University, Montreal, QC, H3A 2T8, Canada;
46: zhuww@physics.mcgill.ca, vkaspi@physics.mcgill.ca, rim@physics.mcgill.ca,
47: aarchiba@physics.mcgill.ca}
48: \altaffiltext{2}{Canada Research Chair; Lorne Trottier Chair;
49: R. Howard Webster Fellow of CIFAR}
50: \altaffiltext{3}{Dynetics, Inc., 1000 Explorer Boulevard, Huntsville,
51: AL, 35806}
52: \altaffiltext{4}{NSSTC, 320 Sparkman Drive, Huntsville, AL, 35805}
53: \altaffiltext{5}{NASA Goddard Space Flight Center, Astrophysics
54: Science Division, Code 662, Greenbelt, MD, 20771}
55: \altaffiltext{6}{Center for Research and Exploration in Space Science and
56: Technology, University of Maryland Baltimore County, 1000 Hilltop Circle,
57: Baltimore, MD 21250}
58: %% --------------------------------------------------------
59:
60: \begin{abstract}
61: We present an analysis of five X-ray Multi-Mirror Mission (\xmm)
62: observations of the anomalous X-ray pulsar (AXP) 1E 2259+586 taken in 2004
63: and 2005 during its relaxation following its 2002 outburst.
64: We compare these data with those of five previous \xmm observations taken in 2002 and 2003, and find the observed flux decay is well described by a power law of index $-0.69\pm0.03$.
65: As of mid-2005, the source may still have been brighter than preoutburst, and was certainly hotter.
66: We find a strong correlation between hardness and flux, as seen in other AXPs.
67: We discuss the implications of these results for the magnetar model.
68:
69: \end{abstract}
70:
71: %% --------------------------------------------------------
72: \keywords{pulsars: individual (1E 2259$+$586) --- X-rays: stars --- stars: neutron}
73: %% --------------------------------------------------------
74: %% A small citation guide:
75: %% \citep{jon90} (Jones et al. 1990) [no comma]
76: %% \cite(t){jon90} Jones et al. (1990) [no comma]
77: %%
78: %% \citep*{jon90} (Jones, Baker, and Williams 1990) [one arg, no comma]
79: %% \citet*{jon90} Jones, Baker, and Williams (1990) [one arg]
80: %%
81: %% \citealp(jon90) Jones et al. 1990 [no comma]
82: %% \citealt{jon90} Jones et al. 1990 [no comma, been using this]
83: %%
84: %% \citealp*{jon90} Jones, Baker, and Williams 1990 [all and no comma,this]
85: %% \citealt*{jon90} Jones, Baker, and Williams 1990 [all and no comma]
86: %% --------------------------------------------------------
87:
88: \section{Introduction}
89: \label{sec:intr}
90: It is now commonly believed that soft gamma-ray repeaters (SGRs) and anomalous
91: X-ray pulsars (AXPs) are neutron stars with ultra-strong magnetic fields, i.e.
92: magnetars \citep{dt92a}. Their common nature was conclusively demonstrated
93: when AXP \tfe was observed to emit SGR-like bursts in 2001 \citep{gkw02} and \ttfn, in the supernova remnant (SNR) CTB 109, was seen to undergo a major SGR-like outburst in 2002 \citep{kgw+03,wkt+04}.
94: Subsequently, a variety of different types of activity in AXPs have been seen,
95: including short- and long-term flux variations \citep{gk04,dkg07} and slow and rapid pulse profile changes \citep{ikh92,kgw+03,wkt+04,icd+07,dkg07,dkg07b}, in addition to bursts and outbursts (\citealp{gkw06,wkg+05,dkg07}; see \citealt{kas07} for a recent review).
96:
97: During \ttfn's 2002 outburst, the pulsed and persistent fluxes rose suddenly by a factor of $\ge$20 and decayed on a timescale of months.
98: Coincident with the X-ray brightening, the pulsar suffered a large glitch of fractional frequency change $4\times10^{-6}$ \citep{kgw+03,wkt+04}.
99: In the first few hours of the outburst, the pulsar's pulse profile changed significantly, its pulsed fraction decreased, and its spectrum hardened dramatically.
100: Over 80 short SGR-like bursts from the pulsar were observed at the same time \citep{gkw04}.
101: A near-infrared (K$_s$) enhancement was also observed during the epoch of the outburst \citep{kgw+03}.
102:
103: Combining {\it Rossi X-Ray Timing Explorer } (\xte) observations and \xmm
104: observations of \ttfn\, taken during and after the outburst, \cite{wkt+04}
105: found that the decay of \ttfn's unabsorbed flux (mostly inferred from \xte
106: pulsed fluxes) after the outburst was well characterized by two power law
107: components: a rapid steep decay visible only during the first several hours
108: ($<1$ day) of the outburst, and a slower decay of index $-0.22$ for the next several months.
109: \cite{tkvd04} found that the near-infrared enhancement at late times decayed at the same rate as the slow X-ray decay,
110: although there were no IR observations during the first few hours of the outburst.
111:
112: Other AXPs have also exhibited transient behavior that could be explained by SGR-like outbursts. AXP \etten is called transient because it was only discovered in 2003 when it suddenly became brighter by a factor of 100 \citep{ims+04,ghbb04}.
113: \cite{gh07} found that the flux of \etten after 2003 followed an exponential decay of timescale 233.5 days.
114: Similarly, the AXP CXOU~J164710.2$-$455216 was found to have brightened by a factor of $\sim$300 between two \xmm observations taken 5 days apart in 2006 September \citep{icd+07}.
115: Candidate AXP \etff, was discovered in an observation made in 1993 by {\it ASCA} \citep{gv98,tkk+98}. Follow-up observations in 1999 showed that the source's flux was smaller by a factor of $\sim$10 \citep{vgtg00}.
116: \cite{tkgg06} found that \etff\, remains undetected in \chandra observations
117: taken in 2003, with its flux $\sim $260-430 times fainter than observed in 1993.
118:
119: The transient AXP phenomena summarized above are qualitatively similar to the 1998 August 27 flare of \ninet, in which the X-ray flux decayed with a power law of index $\sim -0.9$ \citep{fmw+03}, and the flux decay of SGR 1627$-$41 since 1998, which followed a power law of index $\sim -0.47 $ and lasted for $\sim$800 days \citep{kew+03}.
120: However, thus far, the AXP outbursts have been much less energetic than most SGR outbursts.
121: Also, most of the burst energy was released during the afterglows of the AXP outbursts, while for SGR outbursts, the X-ray afterglows have less integrated energy than the burst itself.
122:
123: With now a handful of AXP and SGR outbursts and subsequent relaxations observed, we can begin to look for correlations between different outburst and relaxation properties in the hope of constraining magnetar physics.
124: For example, SGR outburst recoveries have been modeled as crustal cooling following impulsive heat injection, and in principle can yield constraints on the nature of the crustal matter \citep{let02}.
125: Alternatively, the AXP events have been interpreted in terms of magnetospheric twisting \citep{tlk02,bt07}, whose recovery depends on electrodynamics in the region of the
126: magnetosphere immediately outside the stellar surface.
127: On the other hand, \citet{gogk07} suggest that AXP recoveries can be
128: modeled with a stationary magnetosphere, with only the surface temperature
129: changing. They argue that their model, which includes the stellar
130: atmosphere, can be used to quantitatively determine the source's magnetic
131: field.
132:
133: In this paper we present a spectral and pulsed flux analysis of 10 \xmm\, observations of AXP 1E~2259+586\ taken between 2002 and 2005, as the source relaxed back toward quiescence following its 2002 outburst.
134: We compare the X-ray flux and spectral evolution of \ttfn\, with those of other magnetars, and interpret these results in terms of the magnetar model.
135:
136:
137: \section{Observations}
138: \subsection{{\textit XMM-Newton} Observations}
139: \label{sec:xmmobs}
140: Ten \xmm \citep{jla+01} observations were analyzed for this paper.
141: The first five observations of \ttfn\, were taken between 2002 and 2003, just prior to and after the 2002 June outburst.
142: Data from these five observations have already been presented in \cite{wkt+04}.
143: We re-analyzed these observations using the \xmm calibrations published on
144: 2007 September 4 (XMM-CCF-REL-239\footnote{See http://xmm.esac.esa.int/external/xmm\_sw\_cal/calib/rel\_notes/index.shtml}).
145: The later five observations were taken between 2004 and 2005.
146: Most of these observations pointed at \ttfn, with the European Photon Imaging Camera (EPIC) pn camera \citep{sbd+01} in Small Window Mode.
147: However, three observations were obtained with \xmm pointing at a portion of the SNR CTB 109's shell and with the pn camera in extended Full Frame Mode.
148: Details about the observational modes, pointing offsets, and exposure times are presented in Table \ref{tab:obs}.
149: The EPIC mos cameras \citep{taa+01} were operating in Full window mode with
150: the medium filter in four of the first five observations, the exception being
151: the third, and therefore the observed spectra are highly piled-up. The mos
152: cameras were operated in Small Window Mode with the thick filter in the
153: remaining observations, and hence with lower efficiency than for the pn camera.
154: Nevertheless, we analyzed the mos data and found the resulting fluxes and
155: parameters were quantitatively in agreement with those from pn data, given the
156: current knowledge of cross-calibration uncertainties between the two
157: instruments.\footnote{See
158: http://xmm.esac.esa.int/docs/documents/CAL-TN-0018-2-6.pdf, on the current
159: calibration status of the EPIC cameras.}
160: In this paper we report only the higher quality EPIC pn data.
161:
162: The data were analyzed with the \xmm Science Analysis System (SAS) version
163: 7.1.0\footnote{See http://xmm.esac.esa.int/sas/7.1.0/} and the latest calibrations.
164: Strong background flares can sometimes contaminate source events.
165: To exclude possible flares, we extracted light curves from the entire field of
166: view for events having energy $>$ 10 keV.
167: We then examined these light curves for flares. We defined bad time intervals to be when flares occurred, and excluded these intervals for all subsequent analyses.
168: For all the \xmm observations, we filtered a total of 20 ks of bad time intervals.
169: Then we corrected the event times to the barycenter using the SAS {\tt barycen } tool.
170:
171: \subsection{\textit{RXTE} observations}
172: \label{sec:xteobs}
173: We have observed AXP 1E~2259+586 regularly since 1997 with \xte
174: (see, e.g., \citealt{gk02}). Our data were obtained using the Proportional
175: Counter Array (PCA) on board \xte \citep{jmr+06}. The PCA consists of five identical
176: and independent xenon/methane Proportional Counter Units (PCUs). We use our
177: \xte observations of 1E~2259+586 to monitor its pulsed flux,
178: and its frequency evolution using phase-coherent timing, and to look for
179: bursts and pulse profile changes.
180:
181:
182: For the purposes of this paper we analysed 193 observations that took place
183: between 2001 April~1 (MJD 52,000) and 2006 September~22 (MJD 54,000): 15
184: preoutburst observations, 1 observation during the outburst, and 177
185: postoutburst observations. All 193 observations with the exception of the
186: two observations immediately following the outburst were taken in {\tt
187: GoodXenonwithPropane} or {\tt GoodXenon} data modes. Both data modes record
188: photon arrival times with 1 $\mu$s resolution and bin photon energies into
189: one of 256 channels. To maximize the signal-to-noise ratio, we analysed only
190: those events from the top xenon layer of each PCU. The remaining two
191: observations were in event modes with a time resolution of
192: $\sim$125~$\mu$s, a smaller number of energy channels, and no possibility of
193: layer selection.
194: For each of the observations we created barycentered light curves in the
195: 2--10~keV band with 31.25~ms time resolution.
196:
197: We then folded each of the
198: light curves using an ephemeris determined iteratively by maintaining phase
199: coherence (see, e.g., \citealt{gk02}). We then used the folded profiles to
200: calculate the pulsed flux for each observation using both an rms estimator
201: (see, e.g., \citealt{wkt+04}) and an area estimator after baseline subtraction (see
202: \citealt{adk08} in preparation, for details). The results obtained using the two
203: methods were consistent with each other. Here we only report the area
204: pulsed flux because, while more sensitive to noise,
205: it is the quantity of primary interest.
206: %, and the rms pulsed flux's relation to it depends on profile shape.
207: %by definition not sensitive to pulse profile changes.
208:
209:
210: To calculate the area pulsed flux for a given folded time series, we used
211: the following:
212: \begin{equation}
213: \label{eq:pf}
214: PF_{area} = \sum\limits_{i=1}^{N} (p_i - p_{min})/N,
215: \end{equation}
216: where $p_i$ refers to the count rate in the $i$th bin, $N$ is the number of
217: phase bins, and $p_{min}$ is the average count rate in the off-pulse phase of the profile, determined by
218: cross-correlating with a high signal-to-noise ratio template, and calculated in
219: the Fourier domain after truncating the Fourier series to six harmonics.
220: Finally, we combined the pulsed flux numbers from each of two consecutive weeks
221: into a single number, with the exception of the burst observation and the
222: two observations that followed it, which remained unbinned. The results are
223: presented in Figure \ref{fig:all}{\em a}.
224:
225:
226:
227:
228: \section{Analysis and Results}
229: \label{sec:res}
230:
231: \subsection{Spectrum evolution}
232: Source spectra were extracted from circular regions of 32$''$.5 radius around
233: the source center, using the barycentered, filtered event file described in
234: \S \ref{sec:xmmobs}.
235: Background spectra were extracted from circular regions of 50$''$ radius centered $\sim 3'$ away from the source center.
236: For the observations taken in Small Window Mode, we extracted single- and double-photon events and excluded events on or close to a bad pixel using the filter expression ``FLAG $=0$ \&\& PATTERN $<=4$''.
237: In the Full Frame Mode observations, the source is highly off-center in the CCD image (Table \ref{tab:obs}), and bad pixels were found close to the source center region.
238: For these observations, the event list was filtered using the selection expression \#XMMEA\_EP to exclude only photons which fall directly on the bad pixels.
239: However, we did not exclude photon events located adjacent to the bad pixel (which normally would be excluded by the expression FLAG $=0$),
240: because when there is a bad pixel close to the center of the source region,
241: the effective area is evaluated more accurately with pixels around the bad
242: ones taken into account by the {\rm SAS} command {\tt arfgen} (\xmm help desk
243: 2008, private communication).
244: In order to avoid events that affected multiple pixels, we used only single events (PATTERN$=0$) in the Full Frame Mode data.
245: Event lists thus extracted were input to {\tt ftool} {\tt grppha},
246: which grouped the events by at least 25 photons per bin.
247: A systematic uncertainty of 2\% was also appended to the output spectra using
248: {\tt grppha} in order to characterize the current level of calibration
249: accuracy.\footnote{See http://xmm.esac.esa.int/docs/documents/CAL-TN-0018.pdf,
250: (\it EPIC Status of Calibration and Data Analysis)}
251:
252: The resultant spectra were fitted in {\rm XSPEC 12.3.0}\footnote{See
253: http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/} with the commonly used
254: photoelectrically absorbed blackbody plus power law model in the energy range
255: $0.6$-$12$ keV.
256: Because the hydrogen column density $N_H$ is not expected to be variable, we fixed this parameter for all the data sets and performed a joint fit.
257: The goodness of fit is reasonable (see $\chi^2_{\nu}$ in Table \ref{tab:par}).
258: The best-fit $N_H$ is $(1.012\pm0.007) \times 10^{22}\, \rm cm^{-2}$.
259:
260: This value is consistent with that estimated from fitting individual
261: absorption edges of elements O, Fe, Ne, Mg, and Si in the \xmm RGS spectra \citep{dv06b}.
262: The other parameters were set free to vary and their best-fit values are presented in Table \ref{tab:par}.
263: The best-fit blackbody temperature, blackbody radius, and power law index are plotted versus time in Figure \ref{fig:all}.
264:
265: %The pulsed flux of \ttfn from \xte monthly monitoring is also plotted in Figure \ref{fig:all}{\rm a} to show the AXP's flux history.
266:
267:
268:
269:
270: In order to look for correlations between spectral hardness and flux as
271: observed in other AXPs \citep{roz+05,cri+07,tgd+07,gdk+08}, we have looked for a correlation between hardness ratio and observed flux.
272: We define the hardness ratio to be the ratio of 2--10 keV absorbed flux to 0.1--2 keV absorbed flux.
273: We find the hardness ratio to be strongly correlated with the 2--10 keV
274: absorbed flux (as shown in Fig. \ref{fig:flxhr}{\em a}) in our observations.
275: An anti-correlation between photon index and 2--10 keV unabsorbed flux is also
276: seen, but has more scatter (as shown in Fig. \ref{fig:flxhr}{\em b}).
277: This is likely because the photon index is not a perfect measure of spectral hardness, as it can be strongly influenced by the spectral fit at the low end of the band.
278:
279:
280: \subsection{Pulsed fractions}
281: \label{sec:pf}
282: We folded the 0.1--2 and 2--10 keV light curves of each \xmm observation at the pulsar's period, determined using an ephemeris derived by phase coherent timing, from \xte monitoring (Table \ref{tab:obs}; see \citealt{dkg07} for details).
283: Each pulse profile was constructed by folding the photons into 32 phase bins.
284: We measured area pulsed flux of the \xmm the same way we did for \xte (see
285: eq.\ref{eq:pf}), except that we used eight harmonics instead of six when smoothing
286: the light curves (for details see \citealt{adk08} in preparation).
287:
288: The measured area pulsed fractions are plotted in Figure \ref{fig:all}{\em f}.
289: A possible correlation between the 0.1--2 keV area pulsed fraction and the
290: 2--10 keV unabsorbed flux is seen (Fig. \ref{fig:flxpf}, {\em filled circles}).
291: A similar correlation was also found between the 0.1--2 keV area pulsed fraction and the 0.1--2 keV absorbed flux.
292: However, the correlation between 2--10 keV pulsed fraction and flux is not
293: significant (Fig. \ref{fig:flxpf}, {\em open boxes}).
294:
295:
296:
297: We also measured the rms pulsed fraction from the profiles to compare with the area pulsed fraction results.
298: The 2--10 keV rms pulsed fractions are consistent with being constant, while the 0.1--2 keV rms pulsed fractions have some variance, but no significant trend or correlation with other parameters.
299: The area and rms pulsed fractions are different by a factor that depends on the shape of the profile; as the pulse profile of \ttfn\, did change temporarily after the outburst (from a simple double peaked profile to triple peaked; \citealp{kgw+03,wkt+04}), the different result is not surprising.
300:
301:
302: %To verify that our measured pulsed fractions were not influenced by the changing pulse profile of \ttfn, we also found the pulsed fraction using an estimator based on the area under the profile (after baseline subtraction, \citealt{adk08} in progress). With this method, we obtained qualitatively very similar results for the measured pulsed fractions.
303:
304:
305:
306: \subsection{Flux evolution}
307: We fit the unabsorbed fluxes measured in our \xmm observations after the
308: outburst with a power law plus constant decay model, $F(t)=F_b[(t-t_g)/({\rm
309: 1~ day})]^{\alpha}+F_q$, where $F(t)$ denotes the unabsorbed flux, $F_b$ is
310: the unabsorbed source flux one day after the onset of the outburst, $F_q$ is
311: the quiescent flux and $t_g$ marks the glitch epoch MJD 52,443.13 \citep{wkt+04}.
312: A good fit of $\chi^2_{\nu}(\nu) =0.66(5)$ (Fig. \ref{fig:flux}, {\em dashed
313: line}) was found.
314: The best-fit power law index $\alpha=-0.69\pm0.03$ (Table \ref{tab:fit}).
315: The quiescent flux level we found from this power law fit is $(1.75 \pm
316: 0.02)\times 10^{-11}\,\rm ergs\,s^{-1}cm^{-2}$, considerably higher than that
317: measured one week before the outburst [$(1.59\pm0.01)\times 10^{-11}\, \rm
318: ergs\,s^{-1}cm^{-2}$; Table \ref{tab:par}].
319: We also fit the \xmm unabsorbed fluxes with an exponential decay plus
320: quiescent level model, $F(t)=F_pe^{-(t-t_g)/\tau}+F_q$, where $F(t)$ is unabsorbed flux, $F_p$ is the peak flux, $F_q$ is the quiescent flux, $\tau$ is the decay timescale and $t_g$ marks the glitch epoch.
321: The fit is worse than that of the power law decay model but still acceptable, with $\chi^2_{\nu}(\nu)$ of $1.08(5)$.
322: The best-fit decay timescale $\tau$ is $13.3\pm0.7$ days.
323: Best-fit flux decay parameters are presented in Table \ref{tab:fit}.
324:
325: We also fit power law and exponential models to the area pulsed flux of
326: \ttfn\, measured by \xte from 12 to 1649 days after the glitch.
327: A power law model fits the data
328: much better than the exponential model [$\chi^2_{\nu}(\nu)=1.18(69)$ for the
329: power law model, $\chi^2_{\nu}(\nu)=1.65(69)$ for the exponential decay model; Table \ref{tab:fit}],
330: which is evidence against the latter.
331: The best-fit exponential decay timescale for \xte data is $134\pm15$ days,
332: an order of magnitude different from the $\sim13$ day timescale found for the \xmm data.
333:
334: The best-fit power law plus constant model for the evolution of the \xte pulsed fluxes
335: is different from that of the \xmm total fluxes.
336: This suggests that the 2--10~keV pulsed fractions were varying.
337: In principle, we can check this with the pulsed fraction measurements we made
338: with \xmm (see \S~\ref{sec:pf}). Given the uncertainties on the \xmm 2--10~keV pulsed
339: fractions (Table~\ref{tab:par}),
340: as well as those of the best-fit evolution models (Table~\ref{tab:fit}),
341: we find that the two are in agreement.
342:
343: %\cite{wkt+04} measured several unabsorbed phase-averaged fluxes with \xte when the flux of \ttfn\, was high.
344: %We plot these \xte unabsorbed fluxes and our \xmm fluxes against time since the glitch in Figure \ref{fig:flux}.
345: %Because \xte does not have the focusing capability of {\it XMM}, the unabsorbed fluxes measured from the \xte observations could be systematically offset.
346: %We introduced an artificial 10\% uncertainty on the \xmm fluxes (to avoid them completely dominating the fit) to account for the uncertainty in comparing \xte and \xmm fluxes, and fit \xmm and \xte fluxes together to the power law plus constant model.
347: %The fit is significantly worse this way ($\chi^2_{\nu}(\nu)=1.69(11)$) and the best-fit decay index is $-0.49\pm0.11$.
348: %The best-fit model is shown in Figure \ref{fig:flux} (solid line) and best-fit parameters are listed in Table \ref{tab:fit}.
349:
350:
351:
352:
353: %The 2002 outburst of \ttfn\, may be similar to the unseen event that caused the transient AXP \etten to brighten in late 2002/early 2003 \citep{ims+04}.
354: \cite{gh07} fit the spectrum of \etten using a double-blackbody model when studying that source's relaxation following its outburst.
355: In order to compare the spectrum and evolution of \ttfn\, to that of XTE~J1810$-$197, we also fit a photoelectrically absorbed double-blackbody model to \ttfn's spectra jointly.
356: A double-blackbody model does not fit the spectra as well as the blackbody plus power law model (see Table \ref{tab:par} for details).
357: The best-fit $N_H$ for the double-blackbody model [$(0.568\pm0.03)\times
358: 10^{22}\, \rm cm^{-2}$] is smaller than that from our blackbody plus power law
359: fit and is not consistent with the value measured model independently from
360: RGS spectra [$(1.12\pm0.33)\times10^{22}\,\rm cm^{-2}$; \citealp{dv06b}],
361: but is consistent with the best-fit $N_H$ [$(0.5$-$0.7)\times10^{22}\,\rm cm^{-2} $] of CTB 109 measured by \cite{spg+04}.
362:
363: Unabsorbed fluxes obtained using the double-blackbody spectral model can also be fitted to a power law decay or an exponential decay model.
364: The best-fit power law index is $-0.73\pm0.04$, and the best-fit exponential timescale is $12.7\pm0.7$ days (Table \ref{tab:fit}), consistent with what we obtained using the blackbody plus power law spectral model.
365: This indicates that our results for the decay parameters are independent of the choice of spectral model.
366: In the analysis of XTE~J1810$-$197 by \cite{gh07}, they found that both of the two-blackbody components' flux followed an exponential decay after \etten's 2003 outburst.
367: However, we find that the flux of \ttfn's soft blackbody component measured from our fourth and fifth observations (only $\sim$21 days after the outburst and glitch) were lower than that measured for the last five observations (see Table \ref{tab:par} for details).
368: This flux variation of the soft blackbody component therefore cannot be well fitted with an exponential or power law decay model.
369: The temperatures of both the hotter and cooler components were also lower in the fourth and fifth observations than in the last five observations.
370: %It seems that the higher flux after outburst was due only to larger blackbody radii in this model.
371: The non-monotonic variation of the soft blackbody flux and the two components' temperature are different from what was observed by \cite{gh07} and suggest that the double-blackbody model is not a reasonable representation of the spectrum of \ttfn.
372: On the other hand, the spectral evolution from the blackbody plus power law spectral fit looks more reasonable.
373: Using this spectral model, the blackbody radius in the first postoutburst
374: observation was small compared to that of the preoutburst observations and was
375: even smaller in the second and third postoutburst observations (Fig.
376: \ref{fig:all}{\em d}), suggesting that one or more hot spots formed after the outburst and were fading away in the next few months.
377: In the last five observations, the blackbody radius was as large as the preoutburst value, suggesting that the putative hot spots had completely faded away and the thermal radiation then mostly came from the bulk surface of the neutron star as it did before outburst.
378: Perhaps a more realistic spectral model such as that of \cite{gogk07,gog07} could describe the spectral evolution of \ttfn\, better, but such an analysis is outside the scope of this paper.
379:
380: Based on \xte observations, \citet{wkt+04} found that the decay of \ttfn's
381: 2002 outburst consisted of two parts: a steeper power law decay in the
382: first few hours, and a slower power law decay afterwards. They also
383: found that the total energy released (2--10 keV) in the slower decay was
384: $2.1\times10^{41}$ ergs, which is much larger than the total energy (2--60
385: keV) released in the bursts ($6\times10^{37}$ergs; \citealp{gkw04}).
386: We also studied the slower decay, by fitting a power law plus constant
387: model, instead of the simple power law model used by \citet{wkt+04}.
388: The total released energy, according to our best-fit model, is
389: roughly consistent with that calculated by \citet{wkt+04}: we find
390: $\simeq3\times10^{41}$ ergs (2--10~keV), assuming that the outburst
391: will be over in 10000 days. However, based on our best-fit exponential model,
392: the total energy released was somewhat smaller, $\simeq(3-4)\times10^{40}$ ergs (2--10 keV).
393:
394: \section{Discussion}
395: \label{sec:disc}
396:
397: In this paper, we have presented a comprehensive study of the X-ray recovery
398: of AXP 1E~2259+586 following its 2002 outburst. Here we discuss the properties
399: of this recovery, compare them with those of other magnetar outbursts, and consider
400: how they constrain the magnetar model.
401:
402: \subsection{Return to ``Quiescence''}
403:
404: In our 2004 and 2005 {\it XMM} observations, the source's temperature
405: and unabsorbed fluxes were still higher than the preoutburst value
406: (Fig. \ref{fig:all}). This suggests that the source was not fully back to
407: the preoutburst flux level. Our power law fit to the flux decay shows
408: that the after-outburst quiescent flux level is $(1.75 \pm 0.02)\times
409: 10^{-11}\, \rm ergs,s^{-1} cm^{-2}$, which is significantly higher than
410: the preoutburst value [$(1.59\pm0.01) \times 10^{-11}\,\rm ergs\,s^{-1}
411: cm^{-2} $; Table \ref{tab:fit}]. Either the 2005 flux had still not
412: returned to its quiescent level, or perhaps it had returned to quiescence
413: but the flux just before the event was unusually low. Also possible is
414: that this (and other) AXPs do not have well-defined constant quiescent
415: fluxes, but have long-term flux variations. Indeed, there is evidence for some
416: X-ray flux variability in \ttfn\, over the years since its discovery in 1981
417: \citep{bs96}.
418: Other AXPs also show variability on a variety of timescales
419: (see \citealp{kas07} for a review).
420: % although its pulsed flux as observed by {\it RXTE} was admittedly
421: %stable from 1996 until 2002 \citep{gk02}.
422:
423: \subsection{Comparison with other Magnetar Recoveries}
424:
425: It is useful to compare the behavior observed from \ttfn\, with
426: that of other magnetars. \ninet's flux was found to follow a
427: power law of index $-0.713\pm0.025 $ after its 1998 August 27 flare
428: \citep{wkg+01}.\footnote{Later the afterglow of the \ninet\, August 27 flare
429: was fitted with a power law plus constant model instead of the single
430: power law model used by \cite{wkg+01}, and a decay index of $\sim0.9$
431: was obtained \citep{fmw+03}.} This has been interpreted as the cooling
432: of the magnetar outer crust following a sudden release of magnetic
433: energy \citep{let02}. This model predicts a power law decay of index
434: $\sim -2/3$. The flux of SGR 1627$-$41 was found to decay following a power law
435: of index $\sim-0.47$ since its 1998 source activation. Approximately
436: 800 days after the source activation, SGR 1627$-$41's flux suddenly
437: declined by a factor of 10. This behavior is also well fitted by
438: the crust cooling model, although with some fine tuning \citep{kew+03}.
439: We fit the \xmm 2--10 keV unabsorbed fluxes of \ttfn\, with a power law
440: plus constant model, and found the best-fit power law index to be $-0.69
441: \pm 0.03 $, close to that of \ninet, and that predicted by the model.
442: This suggests that the \ttfn\, outburst afterglow may also be explained
443: by the diffusion of heat in the outer crust.
444:
445: The transient AXP \etten exhibited an outburst in 2003. \cite{ims+04}
446: found that the afterglow of the \etten\, outburst as observed by \xte
447: could be described by a power law decay model ($F\propto t^{-\beta} $)
448: with $\beta=0.45-0.73 $. This is similar to the behavior of \ttfn\,
449: and other SGRs. However, \cite{gh07} found that, with more observations
450: taken by \chandra\, from 2003 to 2006, the afterglow of the \etten\,
451: outburst actually followed an exponential decay of timescale 233.5 days.
452: As we have shown in this paper, the pulsed and unabsorbed X-ray flux decay
453: of \ttfn\, favors the power law decay model over the exponential decay.
454: Perhaps the physical processes involved in the 2003 outburst of \etten
455: were different from those in 2002 outburst of \ttfn.
456:
457: \subsection{Twisted Magnetosphere Model}
458:
459: \cite{tlk02} reported that, if there exists a global twist of the
460: magnetosphere, the decay timescale $\tau$ of this twist would be
461: \begin{equation}
462: \label{eq:twist}
463: \tau = 40\Delta \phi^2 (\frac{L_X}{10^{35}{\rm
464: ergs\,s^{-1}}})^{-1}(\frac{B_{\rm pole}}{10^{14} \rm G})^2(\frac{R_{\rm
465: NS}}{10 \rm km})^3 \rm yr.
466: \end{equation}
467: \cite{wkt+04} argued that,
468: for \ttfn, the twist angle $\Delta \phi$ should be $\sim 10^{-2}$ rad.
469: Thus, the predicted twist relaxation timescale of \ttfn\, is several
470: hours, which is coincidently the timescale of the steeper flux decay
471: observed at the beginning of the afterglow.
472:
473: However, \citet{bt07} have
474: shown more recently that this decay timescale is actually
475: expected to be much larger than equation~(\ref{eq:twist}) suggests. This is because,
476: in their model, the self-induction of the twisted portion of the magnetosphere
477: accelerates particles from the stellar surface and initiates avalanches of pair
478: creation which forms the corona. This corona persists in dynamic equilibrium,
479: maintaining the electric current, as long as dissipation permits. The relevant
480: timescale in this picture for the decay of a sudden twist is given by
481: \begin{equation}
482: \tau \simeq 0.3 (\frac{L_X}{10^{35}{\rm ergs \, s^{-1}}}) (\frac{e\Phi_e}{\rm
483: GeV})^{-2}(\frac{R_{\rm NS}}{10 \rm km}) \,\, {\rm yr},
484: \end{equation}
485: where $L_X$ is the peak X-ray luminosity and $e\Phi_e$ is the voltage along the
486: twisted magnetic field lines and should nearly universally be $\sim$1~GeV \citep[see][]{bt07}.
487: For 1E~2259+586, we find $\tau
488: \simeq 1.2$~yr, given the peak luminosity $L_X \sim 4 \times 10^{35}(d/3
489: \; {\rm kpc})$~ergs~s$^{-1}$. Thus, the longer observed decay after the
490: initial steep decline may indeed correspond to the untwisting of a coronal
491: flux tube in the \citet{bt07} picture, although the predicted timescale is somewhat
492: smaller than the observed time to return to quiescence. We note that the
493: \citet{bt07} model predicts a linear flux decline, in contrast
494: to what we have observed for 1E~2259+586 and what has been
495: observed for XTE~J1810$-$197 \citep{gh07}. Moreover, in the $\sim$5 yr of {\it
496: RXTE} monitoring of 1E~2259+586 prior to its 2002 outburst \citep{gk02},
497: its pulsed X-ray luminosity in the 2--10~keV band was roughly constant at $\sim 2 \times
498: 10^{34}$~ergs~s$^{-1}$. This also is puzzling given the \citet{bt07}
499: prediction that if the time between large-scale events is longer than
500: the decay time from the previous event, the magnetar should enter a
501: quiescent state in which the observed luminosity is dominated by the
502: surface blackbody emission. Why should the ``quiescent'' blackbody
503: emission from 1E~2259+586 be a full order of magnitude larger than that
504: from XTE~J1810$-$197, especially given the latter's much larger inferred
505: magnetic field ($1.7 \times 10^{14}$ versus $6 \times 10^{13}$~G)?
506: This disparity in ``quiescent,'' steady luminosities is even larger
507: when considering AXP 1E~1841$-$045, which has an apparently steady 2--10~keV
508: luminosity of $1.4 \times 10^{35}\rm ~ergs~s^{-1}$, and comparing with probable AXP
509: AX~1845$-$0258, which has quiescent luminosity approximately 2 orders of
510: magnitude smaller \citep{tkgg06}. Distance uncertainties may contribute
511: but not on a scale that can significantly alleviate this problem.
512: This remains an interesting puzzle in magnetar physics.
513:
514: The twisted magnetosphere or flux tube models generically predict that
515: the flux and spectral hardness of magnetars in outburst should be roughly
516: correlated due to increased scattering optical depth when the twist is
517: larger. However, a similar prediction for a flux/hardness correlation was
518: made by \cite{og07} in their thermally emitting magnetar model, using a
519: simple prescription for the magnetosphere and scattering geometry, with
520: the latter stationary, i.e. invoking no variable magnetospheric twists.
521: \citet{gogk07} found that their model could reproduce the existing data
522: for XTE~J1810$-$197. We note that hardness-intensity correlations have
523: now been observed for \seven\, \citep{cri+07}, \tfe\, \citep{tgd+07},
524: and as we report, in our \ttfn\, \xmm observations. It would be interesting
525: to apply analysis of \cite{og07} to these data, but it is outside the scope of this paper.
526:
527: \subsection{Other Observed Recovery Properties}
528:
529: The fact that the rms and area pulsed fractions remained largely constant while the
530: blackbody radius (in the blackbody plus power law model) changed by
531: a factor of $\sim$2 (Fig. \ref{fig:all}) is worth considering, if the
532: empirical blackbody plus power law spectrum model somehow resembles the
533: real radiation mechanism. %If we only consider simple viewing geometry,
534: Pulsed fraction should generally decrease when the thermally radiating
535: region on the star grows, provided that this region is not very small
536: compared to the entire surface. Any realistic spectral model which
537: takes radiative transfer in the atmosphere and scattering through the
538: magnetosphere into account should be able to reproduce the observation
539: in this regard as well.
540:
541: A clear anti-correlation between \tfe's pulsed fraction and unabsorbed
542: flux has been observed \citep{tmt+05,gkw06,tgd+07}.
543: However, we found no such correlation in the 2--10 keV band for \ttfn.
544: On the contrary, its 0.1--2 keV area pulsed fractions seem to be correlated with
545: both 0.1--2 and 2--10 keV unabsorbed fluxes (see Fig. \ref{fig:flxpf}).
546: \cite{gh07} found that \etten's pulsed
547: fraction measured between 2003 and 2006 after its outburst decreased
548: with the decay of its flux, i.e. \etten's pulsed fraction is also
549: correlated with flux. Thus, the striking anti-correlation between pulsed
550: fraction and flux observed from \tfe\, is clearly not universal.
551:
552: Finally, we note that the near-infrared flux decay of \ttfn\, was found
553: to follow a power law of index $-0.75^{+0.22}_{-0.33}$ when fitted to
554: a power law plus constant model \citep{tkvd04}. This decay index is
555: close to what we found for the X-ray flux decay, thus confirming the
556: reported correlation between near-IR and X-ray fluxes postoutburst.\footnote{The $-$0.22 X-ray decay index reported by \cite{wkt+04} and
557: the $-$0.21 near-infrared flux decay index reported by \cite{tkvd04} were
558: obtained from a simple power law fitting, i.e. with no quiescent level
559: included in the fit.} \citet{tgd+08} and \citet{wbk+08} showed that
560: the near-IR flux of 1E~1048.1$-$5937 do show correlation with X-rays at
561: times of outbursts. However, \citet{crp+07} show that the near-IR flux
562: variation of XTE~J1810$-$197 is not simply correlated with X-ray flux
563: nor even monotonic postoutburst. Thus, the AXP picture with regard to
564: near-IR variability is not yet fully clear.
565:
566:
567:
568: \acknowledgements
569: We thank A. Beloborodov, A. Cumming, and G. Vasisht for useful discussions.
570: Support for this work was provided by an NSERC Discovery grant Rgpin
571: 228738-03, an R. Howard Webster Fellowship of the Canadian Institute for
572: Advanced Research, Les Fonds de la Recherche sur la Nature et les
573: Technologies, a Canada Research Chair, and the Lorne Trottier Chair in Astrophysics and Cosmology to VMK.
574: %FPG is supported by the NASA Postdoctoral Program administered by Oak Ridge Associated Universities at NASA Goddard Space Flight Center.
575: RD was supported by an NSERC PGSD scholarship.
576:
577:
578:
579:
580: \bibliographystyle{apj}
581: \bibliography{myrefs,journals1,modrefs,psrrefs,crossrefs}
582:
583: \clearpage
584:
585:
586: \begin{deluxetable}{ccccccc}
587: \tabletypesize{\footnotesize}
588: \tablecaption{\label{tab:obs} \xmm observation log for 1E~2259$+$586.}
589: \tablehead{
590: Name & \xmm & Date & Date & pn on-times \tablenotemark{a} & Off-axis angle & Frequencies \tablenotemark{b} \\
591: & Obsid & (MJD TDB) & (YY-MM-DD) & (ksec) & (arcmin) & (s$^{-1}$)
592: }
593: \startdata
594: Obs1\tablenotemark{c \rm d} & 0057540101 & 52,296.791 & 02-01-22 & 8.5 & 8.7 & 0.1432871204(7) \\
595: Obs2\tablenotemark{c} & 0038140101 & 52,436.413 & 02-06-11 & 26.3 & 2.0 & 0.143287000(9) \\
596: Obs3 & 0155350301 & 52,446.446 & 02-06-21 & 16.4 & 2.0 & 0.14328759(1) \\
597: Obs4\tablenotemark{d} & 0057540201 & 52,464.368 & 02-07-09 & 5.2 & 10.7 & 0.14328754(1) \\
598: Obs5\tablenotemark{d} & 0057540301 & 52,464.602 & 02-07-09 & 10.2 & 10.3 & 0.14328754(1) \\
599: Obs6 & 0203550301 & 53,055.596 & 04-02-20 & 3.6 & 1.9 & 0.143286974(7) \\
600: Obs7 & 0203550601 & 53,162.655 & 04-06-06 & 4.8 & 2.0 & 0.143286882(4) \\
601: Obs8 & 0203550401 & 53,178.634 & 04-06-22 & 3.4 & 2.0 & 0.143286868(2) \\
602: Obs9 & 0203550501 & 53,358.014 & 04-12-19 & 3.5 & 2.0 & 0.143286714(1) \\
603: Obs10 & 0203550701 & 53,579.970 & 05-07-28 & 3.3 & 1.9 & 0.143286523(7) \\
604:
605: \enddata
606: \tablenotetext{a}{On-times quoted reflect on-source times after filtering of background flares.}
607: \tablenotetext{b}{Frequencies are from contemporaneous \xte observations. }
608: \tablenotetext{c}{Observations taken before the outburst, MJD 52,443.13 \citep{wkt+04}. }
609: \tablenotetext{d}{These three observations were taken in extended Full Frame Mode; all the others were taken in Small Window Mode. }
610: \end{deluxetable}
611:
612:
613: \begin{deluxetable}{lllllllllll}
614: %\begin{table}
615: %\scriptsize
616: \tablewidth{495pt}
617: \tabletypesize{\scriptsize}
618: %\tabletypesize{\footnotesize}
619: %\rotate
620: %\caption{\label{tab:par} 1E~2259$+$586's best-fit spectral parameters and pulsed fractions. }
621: %\rotate
622: %\begin{tabular}{lllllllllll}
623: %\hline
624: %\hline
625: %\tabletypesize{\footnotesize}
626: \tablecaption{\label{tab:par} 1E~2259$+$586's best-fit spectral parameters and pulsed fractions. }
627: \tablehead{
628: Parameter\tablenotemark{ a} \tablenotemark{b} & Obs1 & Obs2 & Obs3 & Obs4 & Obs5 & Obs6 & Obs7 & Obs8 & Obs9 & Obs10 %\hline
629: }
630: \startdata
631: \multicolumn{11}{c}{Blackbody plus power law model} \\ \hline
632:
633: $N_{H}$ (10$^{22}$ cm$^{-2}$) & 1.012(7) & 1.012(7) & 1.012(7) & 1.012(7) & 1.012(7) & 1.012(7) & 1.012(7) & 1.012(7) & 1.012(7)& 1.012(7)\\
634: $kT$ (keV) & 0.37(1) & 0.406(2) & 0.510(4) & 0.48(2) & 0.49(1) & 0.400(7) & 0.400(5) & 0.400(7) & 0.405(7) & 0.400(7)\\
635: $\Gamma$ & 3.75(4) & 3.89(2) & 3.49(2) & 3.71(5) & 3.72(4) & 3.77(3) & 3.78(3) & 3.79(3) & 3.79(3) & 3.75(3) \\
636: Flux\tablenotemark{c} & 1.15(2) & 1.29(1) & 3.45(3) & 1.95(5) & 2.03(4) & 1.51(2) & 1.48(2) & 1.48(2) & 1.47(2) & 1.45(2) \\
637: Unabs Flux\tablenotemark{d} & 1.41(3) & 1.59(1) & 4.12(3) & 2.34(7) & 2.44(5) & 1.84(3) & 1.82(2) & 1.81(3) & 1.80(3) & 1.77(3) \\
638: PL/BB ratio\tablenotemark{e} & 1.8(3) & 1.2(2) & 1.6(2) & 2.6(4) & 2.1(3) & 1.6(2) & 1.5(2) & 1.7(2) & 1.5(2) & 1.7(2) \\
639: Hardness \tablenotemark{f} & 0.93(3) & 0.94(1) & 1.43(1) & 1.13(4) & 1.15(4) & 0.98(2) & 0.98(2) & 0.97(2) & 0.98(2) & 0.99(2) \\
640: %\hline
641: $\chi^2_{\nu}(\nu)$ & \multicolumn{2}{l}{ 1.02(5800)} &\multicolumn{2}{c}{ ($P =0.12$)\tablenotemark{g} } & & & & & & \\
642: \hline
643: \multicolumn{11}{c}{Double blackbody model} \\ \hline
644: $N_{H}$ (10$^{22}$ cm$^{-2}$) & 0.568(3) & 0.568(3) & 0.568(3) & 0.568(3) & 0.568(3) & 0.568(3) & 0.568(3) & 0.568(3) & 0.568(3)& 0.568(3)\\
645: Cooler $kT$ (keV) & 0.362(5) & 0.372(2) & 0.390(3) & 0.330(7) & 0.335(5) & 0.371(3) & 0.380(3) & 0.371(3) & 0.370(4) & 0.371(3)\\
646: Hotter $kT$ (keV) & 0.77(4) & 0.82(1) & 0.86(1) & 0.74(3) & 0.73(2) & 0.85(3) & 0.94(2) & 0.89(3) & 0.85(3) & 0.86(3) \\
647: Flux\tablenotemark{c} & 1.12(6) & 1.27(2) & 3.34(4) & 1.85(9) & 1.93(7) & 1.48(4) & 1.46(3) & 1.44(4) & 1.44(4) & 1.41(4) \\
648: Unabs Flux\tablenotemark{d} & 1.26(6) & 1.42(2) & 3.69(4) & 2.05(11) & 2.15(8) & 1.65(5) & 1.63(4) & 1.62(5) & 1.61(5) & 1.58(5) \\
649: HB/CB ratio\tablenotemark{h} & 0.8(1) & 0.6(1) & 1.2(2) & 2.1(3) & 2.0(3) & 0.7(1) & 0.53(7) & 0.64(9) & 0.7(1) & 0.7(1) \\
650: Hardness \tablenotemark{f} & 0.91(5) & 0.92(1) & 1.37(2) & 1.07(6) & 1.09(4) & 0.96(3) & 0.95(2) & 0.94(3) & 0.95(3) & 0.96(3) \\
651: %\hline
652: $\chi^2_{\nu}(\nu)$ & \multicolumn{2}{l}{ 1.11(5800)} &\multicolumn{2}{c}{($P=4.6\times10^{-9}$)\tablenotemark{g} } & & & & & & \\
653: \hline
654: \multicolumn{11}{c}{Pulsed fractions} \\ \hline
655: PF(0.1--2 keV)\tablenotemark{i} & 0.18(3) & 0.234(6) & 0.322(6) & 0.30(2) & 0.28(2) & 0.29(2) & 0.26(1) & 0.24(2) & 0.27(2) & 0.26(2)\\
656: PF(2--10 keV)\tablenotemark{i} & 0.23(5) & 0.30(1) & 0.339(9) & 0.33(4) & 0.36(3) & 0.30(3) & 0.33(2) & 0.34(3) & 0.31(2) & 0.29(3)\\
657: %\hline
658:
659: %\end{tabular}
660: \enddata
661: \tablenotetext{\rm a}{ Numbers in parentheses indicate the 1$\sigma$ uncertainty in the least significant digit. Note that these
662: uncertainties reflect the 1$\sigma$ error for a reduced $\chi^2$ of unity. }
663: \tablenotetext{\rm b}{Best-fit parameters from a joint fit to all data sets. $N_H$ in all data sets was set to be the same; other parameters were allowed to vary from observation to observation.}
664: \tablenotetext{\rm c} {(10$^{-11}$ ergs s$^{-1}$cm$^{-2}$). Observed flux from
665: both spectral components in the range 2--10 keV. }
666: \tablenotetext{\rm d} {(10$^{-11}$ ergs s$^{-1}$cm$^{-2}$). Unabsorbed flux
667: from both spectral components in the range 2--10 keV. }
668: \tablenotetext{\rm e} {The ratio of power law flux to blackbody flux in the 2--10 keV band (corrected for absorption).}
669: \tablenotetext{\rm f} {Spectral hardness defined as the ratio of 2--10 keV
670: absorbed flux to 0.1--2 keV absorbed flux.}
671: \tablenotetext{\rm g} {The probability for the $\chi^2_{\nu}$ to be higher
672: than that was observed, assuming the model is correct.}
673: \tablenotetext{\rm h} {The ratio of hot blackbody flux to cool blackbody flux in the 2--10 keV band (corrected for absorption).}
674: \tablenotetext{\rm i} {The area pulsed fractions. }
675: \end{deluxetable}
676:
677: %\end{table}
678:
679: %\begin{table}
680: \begin{deluxetable}{lcccll}
681: %\footnotesize
682: %\tabletypesize{\footnotesize}
683: \tabletypesize{\scriptsize}
684: \tablecaption{Best-fit parameters for the \ttfn\, flux decay
685: %\caption{Best-fit parameters for the \ttfn\, flux decay
686: \label{tab:fit} }
687: %\begin{tabular}{lcccll}
688: %\hline
689: %\hline
690: \tablehead{
691: \colhead{Power law decay\tablenotemark{a}} & \colhead{$F_q (10^{-11}\rm
692: ergs\,s^{-1}cm^{-2})$} & \colhead{$F_b(10^{-11}\rm ergs\,s^{-1}cm^{-2})$} & \colhead{$\alpha$} &
693: \colhead{$\chi^2$} & \colhead{$\chi^2/\nu$} %}
694: }
695: %\hline
696: \startdata
697: %\hline
698: \xmm UF (BB+PL) \tablenotemark{b}& $(1.75 \pm 0.02)$ & $(5.40 \pm 0.21)$ & $-0.69 \pm 0.03$ & 3.31 & 0.66\\
699: \xmm UF (BB+BB) \tablenotemark{c}& $(1.58 \pm 0.01)$ & $(5.07 \pm 0.22)$ & $-0.73 \pm 0.04$ & 1.22 & 0.24\\
700: \xte PF \tablenotemark{d}&$(0.14 \pm 0.06)\rm cts~s^{-1} PCU^{-1}$ & $(1.4 \pm 0.1)\rm cts~s^{-1} PCU^{-1}$ & $-0.27 \pm 0.05$ & 81.3 & 1.18\\
701: %\hline
702: \hline
703: \colhead{Exponential decay\tablenotemark{e}} & \colhead{$F_q(10^{-11}\rm
704: ergs\,s^{-1}cm^{-2})$ } & \colhead{$F_p(10^{-11}\rm ergs\,s^{-1}cm^{-2})$
705: } & \colhead{$\tau$ (days) }&
706: \colhead{$\chi^2$} & \colhead{$\chi^2/\nu$} \\ %}
707: \hline
708: \xmm UF (BB+PL) \tablenotemark{b}& $(1.81 \pm 0.01)$ & $(2.97 \pm 0.06)$ & $13.3 \pm 0.7$ & 5.39 & 1.08\\
709: \xmm UF (BB+BB) \tablenotemark{c}& $(1.62 \pm 0.01)$ & $(2.69 \pm 0.05)$ & $12.7 \pm 0.7$ & 1.737 & 0.35\\
710: \xte PF \tablenotemark{d}&$(0.362 \pm 0.005 )\rm cts~s^{-1} PCU^{-1}$ & $(0.46 \pm 0.03)\rm cts~s^{-1} PCU^{-1}$ & $134 \pm 15$ & 113.5 & 1.65\\
711: %\hline
712: \enddata
713: %\end{tabular}
714: \tablenotetext{\rm a}{Power law decay model defined as $F(t)=F_b((t-t_g)/({\rm
715: 1~day}))^{\alpha}+F_q$, where $F(t)$ is unabsorbed flux, $F_q$ is the
716: quiescent flux, $\alpha$ is the power law index and $t_g$ is the glitch epoch
717: MJD 52,443.13.}
718: \tablenotetext{\rm b}{\xmm unabsorbed flux decay measured using a blackbody plus power law spectral model.}
719: \tablenotetext{\rm c}{\xmm unabsorbed flux decay measured using a double-blackbody spectral model.}
720: \tablenotetext{\rm d}{\xte area pulsed flux in units of {$\rm cts~s^{-1} PCU^{-1}$}.}
721: \tablenotetext{\rm e}{Exponential decay model, defined as
722: $F(t)=F_pe^{-(t-t_g)/\tau}+F_q$, where $F(t)$ is unabsorbed flux, $F_p$ is the
723: peak flux, $F_q$ is the quiescent flux, $\tau$ is the decay timescale and
724: $t_g$ is the glitch epoch MJD 52,443.13.}
725: \end{deluxetable}
726: %\end{table}
727:
728: \clearpage
729:
730: \begin{figure}
731: \includegraphics[scale=.8]{f1.ps}
732: %\includegraphics{f1.ps}
733: \caption{\label{fig:all} Spectral and pulsed fraction evolution of \ttfn\,
734: during and following its 2002 outburst. ({\em a}) 2--10 keV
735: area pulsed flux measured in \xte monitoring observations; ({\em b}) 2--10 keV
736: unabsorbed phase-averaged flux from \xmm observations (all lower panels are
737: also from \xmm observations); ({\em c}) blackbody temperature ($kT$), ({\em
738: d}) blackbody radius; ({\em e}) photon index; ({\em f}) 0.1--2 keV area
739: pulsed fraction ({\em filled circles}) and 2--10 keV area pulsed fraction
740: ({\em open boxes}).
741: A distance of 3 kpc \citep{kuy02} is assumed to calculate the blackbody radius.
742: The vertical line denotes the 2002 glitch epoch, MJD 52,443.13 \citep{wkt+04}.
743: }
744: \end{figure}
745:
746: \clearpage
747:
748:
749: \begin{figure}
750: \includegraphics[scale=.9]{f2.ps}
751: %\includegraphics{f2.ps}
752: \caption{\label{fig:flxhr}$(a)$ Hardness vs. absorbed flux. Hardness ratio
753: is defined as the ratio of 2--10 to 0.1--2 keV absorbed flux. $(b)$ Photon
754: index vs. 2--10 keV unabsorbed flux. }
755: \end{figure}
756:
757: \clearpage
758:
759: \begin{figure}
760: \includegraphics[scale=0.6]{f3.ps} \\
761: \caption {\label{fig:flxpf} The 2--10 keV area pulsed fraction ({\em open
762: boxes}) vs. 2--10 keV unabsorbed flux ; 0.1--2 keV area pulsed fraction ({\em
763: filled circles}) vs. 2--10 keV unabsorbed flux.}
764: \end{figure}
765:
766: \clearpage
767:
768: \begin{figure}
769: %\includegraphics[scale=.45]{figs/1708-observations.ps}
770: \includegraphics[scale=.8]{f4.ps}
771: \caption{\label{fig:flux} Evolution of \ttfn's 2--10 keV unabsorbed
772: phase-averaged flux ({\em squares}) measured from \xmm observations and 2--10 keV area
773: pulsed flux ({\em cross}) from \xte observations following its 2002 outburst.
774: \xmm unabsorbed fluxes are in units of $\rm 10^{-11} ergs~s^{-1}cm^{-2}$.
775: \xte area pulsed fluxes are in units of $\rm count~s^{-1}PCU^{-1}$.
776: The time axis is relative to the estimated glitch epoch (MJD 52,443.13).
777: The solid line is the power law plus constant model fit to the \xte area pulsed fluxes.
778: The dashed line is a fit of the same model, although having different best-fit
779: parameters, to the \xmm fluxes.
780: See Table \ref{tab:fit} for the best-fit parameters.
781: The dotted line is the flux level in $\rm 10^{-11} ergs~s^{-1}cm^{-2}$ observed with \xmm one week before the outburst.
782: The uncertainty on this preoutburst flux is approximately the width of the line.
783: }
784: \end{figure}
785:
786:
787:
788:
789:
790: \end{document}
791:
792:
793: