1:
2: %\documentclass[12pt,preprint]{aastex}
3: \documentclass[apjl]{emulateapj}
4: %
5: \usepackage{apjfonts,mathptmx}
6:
7:
8: \def\xmm {\emph{XMM-Newton}}
9: \def\cha {\emph{Chandra}}
10: \def\src {CXOU\,J010043.1--721134}
11: \def\flux {\mbox{erg cm$^{-2}$ s$^{-1}$}}
12: \def\lum {\mbox{erg s$^{-1}$}}
13: \def\nh {$N_{\rm H}$ }
14:
15: %
16: \begin{document}
17:
18: %\special{!userdict begin /bop-hook{gsave 150 90 translate 55
19: %rotate /Times-Roman findfont 60 scalefont setfont 0 0 moveto 0.7
20: %setgray (V1.0- 5/9/04 - confidential) show grestore}def end}
21:
22: \title{\emph{XMM-Newton} observations of \src: the first deep look at the soft X--ray emission of a magnetar\altaffilmark{1}}
23:
24: \author{A.~Tiengo, P.~Esposito\altaffilmark{2}, and S.~Mereghetti}
25: \email{tiengo@iasf-milano.inaf.it}
26: \affil{INAF - Istituto di Astrofisica Spaziale e Fisica Cosmica Milano,\\
27: Via Edoardo Bassini 15, 20133 Milano, Italy}
28:
29: \altaffiltext{1}{Based on observations obtained with
30: \emph{XMM-Newton}, an ESA science mission with instruments and
31: contributions directly funded by ESA Member States and NASA.}
32: \altaffiltext{2}{Universit\`{a} di Pavia, Dipartimento di Fisica
33: Nucleare e Teorica and INFN-Pavia,
34: via Agostino Bassi 6, 27100 Pavia, Italy}
35: %\altaffiltext{}{tiengo@iasf-milano.inaf.it}
36:
37:
38: \shortauthors{A. Tiengo et al.}
39: \shorttitle{\xmm\ observations of \src}
40:
41: \journalinfo{The Astrophysical Journal Letters, in press}
42: \submitted{Received 2008 February 11; accepted 2008 May 14}
43:
44: \begin{abstract}
45: We present the analysis of six \xmm\ observations of the Anomalous
46: X--ray Pulsar \src, the magnetar candidate characterized by the
47: lowest interstellar absorption.
48: In contrast with all the other magnetar candidates, its X--ray
49: spectrum cannot be fit by an absorbed power-law plus blackbody
50: model. The sum of two (absorbed) blackbody components with
51: $kT_1=0.30\pm0.02$ keV and $kT_2=0.7\pm0.1$ keV gives an
52: acceptable fit, and the radii of the corresponding blackbody
53: emission regions are $R^\infty_{\rm BB1}=12.1^{+2.1}_{-1.4}$ km
54: and $R^\infty_{\rm BB2}=1.7^{+0.6}_{-0.5}$ km. The former value is
55: consistent with emission from a large fraction of a neutron star
56: surface and, given the well known distance of \src, that is located
57: in the Small Magellanic Cloud, it provides the most constraining
58: lower limit to a magnetar radius ever obtained. A more physical
59: model, where resonant cyclotron scattering in the magnetar
60: magnetosphere is taken into account, has also
61: been successfully applied to this source.
62:
63: \end{abstract}
64:
65: \keywords{pulsars: individual (\src) -- stars: neutron}
66:
67: %\maketitle
68:
69: %
70: %________________________________________________________________
71:
72: \section{Introduction}
73:
74: The Anomalous X--ray Pulsars \citep[AXPs; see][for a recent
75: review]{magnetarsSandro} were initially identified as a subclass of
76: accreting X--ray pulsars.
77: %HELLIER \citep{mereghetti95,vanparadijs95}.
78: Their much softer X--ray
79: spectrum and the lack of a bright optical counterpart distinguished
80: them from the vast majority of X--ray pulsars, which are neutron
81: stars accreting in high mass X--ray binary systems. AXPs have
82: rotation periods of several seconds and show a secular spin-down on
83: timescales of 10$^3$--10$^5$ years, but their rotational energy loss
84: is smaller than their X--ray luminosity, excluding the possibility
85: that they are rotation-powered, like radio pulsars. It is generally
86: believed that the AXPs, as well as another small class of
87: high-energy sources with similar properties, the Soft Gamma-ray
88: Repeaters (SGRs), are magnetars, i.e. neutron stars powered by their
89: extremely high magnetic field
90: \citep[$\sim$10$^{15}$ G;][]{duncan92,thompson96}.
91:
92: The soft X-ray (1--10 keV) spectra of magnetars cannot be
93: adequately fit with single component models whenever data with good
94: count statistics are available. Successful fits are instead obtained
95: by a two component model consisting
96: %, e.g.,
97: of a steep power-law (photon index $\sim$3--4) and a blackbody
98: ($kT\sim0.5$ keV).
99: Some attempts have been done, also based on
100: phase-resolved spectroscopy, to attribute the two components to
101: physically distinct processes \citep[e.g.][]{tmt05}, but no
102: particularly compelling interpretations could be obtained.
103: %\citet{halpern05} discussed several reasons to prefer the two
104: %blackbody model.
105: One problem of
106: %the blackbody plus power-law
107: this model is that it tends to give best-fit values of the
108: interstellar absorption higher than those independently estimated in
109: other ways \citep[e.g.][]{durant06}. Another problem is that the
110: power-law component cannot be extrapolated at lower energies without
111: exceeding the flux of the near infrared (NIR) and optical
112: counterparts \citep[e.g.][]{hulleman04}. Drastic, and possibly
113: unphysical, cut-offs in the power-law component are required to
114: match the low optical/NIR fluxes.
115:
116: In some AXPs
117: %equally
118: good spectral fits are obtained with the sum of
119: two blackbody components with different temperatures. Since this
120: model does not suffer of the problems described above,
121: %in these cases
122: it is usually preferred to the power-law plus blackbody model
123: \citep[e.g.][]{halpern05}. However, also this model is only
124: phenomenological and it is inadequate to represent the non-thermal
125: phenomena that are expected to occur in the highly magnetized
126: magnetosphere of magnetars \citep[e.g.][]{lyutikov06}. More physical
127: models of the X--ray spectra, including the effects of the strong
128: magnetic field and charged currents,
129: % expected in a magnetar magnetosphere,
130: have recently been developed and successfully applied
131: to a sample of magnetar candidates
132: \citep[][]{fernandez07,guver07,RCSnanda}.
133:
134: %However, f
135: From a purely observational point of view, it has not been possible
136: to discriminate between the different models reproducing the
137: magnetar X--ray spectra.
138: %%%se two possibilities.
139: %In fact equally good fits are obtained with
140: %the power-law plus blackbody and with the double blackbody models in
141: %most magnetars.
142: This is mainly due to the low sensitivity of hard X--ray detectors
143: above $\sim$10 keV and to the large uncertainties in the fits
144: introduced by the high interstellar absorption, that severely
145: suppresses the flux below $\sim$1 keV. Being young neutron stars
146: born from massive progenitors, all the Galactic magnetars are
147: located in highly absorbed regions of the Galactic plane. All of
148: them have column densities N$_{\rm{H}}$ ranging from
149: $\sim$$5\times10^{21}$ to $\sim$$10^{23}$ cm$^{-2}$.
150: % while the soft X--ray spectrum of SGR~0526--66, located in the Large
151: % Magellanic Cloud, is heavily contaminated by the surrounding
152: % supernova remnant N49, which is particularly bright in soft
153: % X--rays.
154: The two known magnetars in the Magellanic Clouds, being considerably
155: less absorbed, offer the possibility to better constrain the spectra
156: in the low energy range. The study of SGR~0526--66, located in the
157: Large Magellanic Cloud, is complicated by the presence of the
158: surrounding supernova remnant N49, which is particularly bright in
159: soft X--rays \citep{kulkarni03}. Here we concentrate therefore on
160: the spectral properties of \src\
161: \citep{lamb02,lamb03errata,mcgarry05}, the only known AXP in the
162: Small Magellanic Cloud (SMC).
163:
164: \newpage
165: \section{Observations and data analysis}
166:
167: The field containing \src\ was observed six times\footnote{Only
168: the data of the two first observations have already been
169: published \citep{lamb02,naze04,majid04,mcgarry05,nakagawa08}; we reanalyzed
170: them using more recent calibration files, in order to consistently
171: compare the results with those of the new observations.} with the
172: \xmm\ satellite (see Table \ref{log}). Here we report the analysis
173: of the data collected by the EPIC instrument, which is composed by
174: one PN \citep{struder01} and two MOS X--ray cameras \citep{turner01}.
175: \begin{deluxetable}{ccccc}
176: % \tabletypesize{\scriptsize}
177: \tablecolumns{1}
178: \tablewidth{0pt}
179: \tablecaption{\label{log} Log of the \emph{XMM-Newton} Observations of \src. The Pulse Periods
180: and Corresponding 1$\sigma$ Errors Are Also Reported.}
181: \tablehead{
182: \colhead{Obs.} & \colhead{Date} & \colhead{PN exposure} & \colhead{MOS exposure} & \colhead{Period}\\
183: \colhead{} & \colhead{} & \colhead{(ks)} & \colhead{(ks)} & \colhead{(s)}
184: }
185: \startdata
186: A & 17 Oct 2000 & 14 & 20 & 8.019(1)\phantom{1}\\
187: B & 20 Nov 2001 & 22 & 27&8.0193(9)\\
188: \,\,C\tablenotemark{a} & 27 Mar 2005 & \nodata & 24 & 8.0215(9)\\
189: D & 27 Nov 2005 & 14 & 17 & 8.022(1)\phantom{1}\\
190: E & 29 Nov 2005 & 13 & 16 & 8.022(1)\phantom{1}\\
191: F & 11 Dic 2005 & 9 & 16 & 8.022(2)\smallskip
192: \enddata
193: \tablenotetext{a}{The PN data were not considered, since the PN
194: instrument was operated with the filter wheel in closed position.}
195: \end{deluxetable}
196:
197: \src\ was not the main target of the observations,
198: but, being at an off-axis angle of $\sim$$6\arcmin$, it was always well
199: inside the field of view of the EPIC instrument ($\sim$$15\arcmin$
200: radius).
201: All the observations were performed with the medium optical blocking
202: filter and in Full Frame mode (time resolution of 73 ms and 2.6 s
203: for the PN and MOS, respectively), except for the first PN
204: observation, done in Extended Full Frame mode (time resolution of
205: 200 ms). All the data were processed using the \xmm\ Science
206: Analysis Software (SAS version 7.1.0) and the calibration files
207: released in August 2007. The standard pattern selection criteria
208: (patterns 0--4 for PN and 0--12 for MOS) were adopted.
209:
210: Source spectra were extracted for each observation
211: %according to the same event selection criteria and extraction
212: from circular regions with 25$\arcsec$ radius.
213: The background spectra were extracted from a
214: 94$\arcsec$$\times$72$\arcsec$ box centered at
215: RA $=01^{\rm{h}}\,00^{\rm{m}}\,56\fs8$,
216: Dec. $=-72\degr\,11\arcmin\,33\arcsec$ and rotated such that it
217: intercepts no CCD gaps in any PN image. Response matrices and
218: ancillary files for each spectrum were produced using the SAS
219: software.
220:
221: The spectra
222: %of each observation
223: were fitted to a set of models
224: (power-law, blackbody, blackbody plus power-law, and two
225: blackbodies, all modified by photoelectric absorption)
226: % and with the overall normalization factor described above)
227: using XSPEC version 11.3.1. The single component models gave only
228: marginally acceptable fits in most observations, while better
229: results were obtained with the two component models. The best-fit
230: parameters of the latter models are reported in Table \ref{fits}.
231:
232: \begin{deluxetable*}{cccccccccc}
233: %\rotate
234: % \tabletypesize{\scriptsize}
235: \tablecolumns{10}
236: \tablewidth{0pc}
237: \tablecaption{Summary of the EPIC Spectral Results in the 0.1--10 keV Energy Range. Errors Are Given at the 90\% Confidence Level.\label{fits}} \tablehead{ \colhead{Observation} &
238: \colhead{Model\tablenotemark{a}} & \colhead{$N_{\rm
239: H}$\tablenotemark{b}} & \colhead{$\Gamma$}
240: & \colhead{PL norm.\tablenotemark{c}} & \colhead{$k_B T_{BB1}$} & \colhead{$R_{BB1}$\tablenotemark{d}} & \colhead{$k_B T_{BB2}$} & \colhead{$R_{BB2}$\tablenotemark{d}} & \colhead{$\chi^{2}_{r}$ (d.o.f.)} \\
241: \colhead{} & \colhead{} & \colhead{($10^{20}\rm cm^{-2}$)} & \colhead{} & \colhead{} & \colhead{(keV)} & \colhead{(km)} & \colhead{(keV)} & \colhead{(km)} & \colhead{}
242: }
243: \startdata
244: A
245: %& PL & $46^{+3}_{-5}$ & $3.8^{+0.1}_{-0.2}$ & $50^{+4}_{-7}$ & \nodata & \nodata & \nodata & \nodata & 1.45 (85) \\
246: % & BB & $2.8^{+1.8}_{-1.6}$ & \nodata & \nodata & 0.37$\pm$0.01 & $8.3^{+0.8}_{-0.7}$ & \nodata & \nodata & 1.27 (85) \\
247: & PL+BB & $4.8^{+8.4}_{-2.1}$ & $1.3^{+1.1}_{-1.5}$ & $0.8^{+3.9}_{-0.7}$ & $0.35^{+0.02}_{-0.03}$ & $9.1^{+1.4}_{-0.9}$ & \nodata & \nodata & 1.18 (83)\phantom{1} \\
248: & BB1+BB2 & $8.1^{+7.9}_{-4.0}$ & \nodata & \nodata & 0.27$^{+0.07}_{-0.09}$ & $12.9^{+15.7}_{-3.4}$ & 0.5$^{+0.3}_{-0.1}$& $3.6^{+3.5}_{-2.9}$ & 1.11 (83)\phantom{1} \\
249: %\hline
250: %----------------------------------------------------------------------------
251: B
252: %& PL & $38^{+3}_{-2}$ & $3.3\pm0.1$ & $41^{+4}_{-3}$ & \nodata & \nodata & \nodata & \nodata & 1.09 (117) \\
253: % & BB & $<$1.2 & \nodata & \nodata & 0.39$\pm$0.01 & $7.5^{+0.5}_{-0.3}$ & \nodata & \nodata & 1.17 (117) \\
254: & PL+BB & $4.3^{+3.0}_{-2.8}$ & $1.7^{+0.7}_{-0.5}$ & $2.0^{+5.7}_{-1.3}$ & $ 0.36^{+0.02}_{-0.03}$ & $8.5^{+1.3}_{-0.6}$ & \nodata & \nodata & 0.93 (115) \\
255: & BB1+BB2 & $4.2^{+2.7}_{-2.1}$ & \nodata & \nodata & 0.31$\pm$0.03 & $11.0^{+2.6}_{-1.6}$ & 0.7$\pm$0.1& $1.6^{+0.9}_{-0.6}$ & 0.72 (115) \\
256: %\hline
257: %----------------------------------------------------------------------------
258: %$^{\phantom{1}}$
259: \,\,C\tablenotemark{e}
260: %& PL & 45$^{+9}_{-7}$ & $3.5^{+0.4}_{-0.3}$ & $47^{+14}_{-10}$ & \nodata & \nodata & \nodata & \nodata & 1.22 (109) \\
261: % & BB & $4.3^{+4.6}_{-3.8}$ & \nodata & \nodata & 0.36$\pm$0.03 & $9.0^{+2.0}_{-1.5}$ & \nodata & \nodata & 1.14 (109) \\
262: & PL+BB & $7.8^{+22}_{-7.0}$ & $\sim$2 & $<$16 & 0.34$^{+0.04}_{-0.05}$ & $9.6^{+2.8}_{-2.0}$ & \nodata & \nodata & 1.16 (107) \\
263: & BB1+BB2 & $5.3^{+21.3}_{-4.4}$ & \nodata & \nodata & $<$0.39 & $<$11.6 & $\sim$1& $<$3.7 & 1.16 (107) \\
264: %\hline
265: %----------------------------------------------------------------------------
266: D
267: %& PL & 36$\pm$3 & $3.2\pm0.1$ & 37$\pm$4 & \nodata & \nodata & \nodata & \nodata & 1.40 (85) \\
268: % & BB & $<$0.9 & \nodata & \nodata & 0.39$\pm$0.01 & $7.4^{+0.4}_{-0.3}$ & \nodata & \nodata & 1.38 (85) \\
269: & PL+BB & $3.6^{+6.6}_{-3.5}$ & $1.6^{+0.7}_{-0.9}$ & $2.0^{+5.0}_{-1.6}$ & $ 0.36^{+0.02}_{-0.03}$ & 8.2$^{+1.0}_{-0.7}$ & \nodata & \nodata & 1.07 (83)\phantom{1} \\
270: & BB1+BB2 & $<3.6$ & \nodata & \nodata & 0.35$^{+0.02}_{-0.03}$ & $8.9^{+0.9}_{-1.0}$ & 0.9$^{+0.4}_{-0.2}$& $0.7^{+0.7}_{-0.4}$ & 1.02 (83)\phantom{1} \\
271: %\hline
272: %----------------------------------------------------------------------------
273: E
274: %& PL & 39$\pm$3 & $3.3\pm0.1$ & $40^{+5}_{-4}$ & \nodata & \nodata & \nodata & \nodata & 0.99 (86) \\
275: % & BB & $<$2.5 & \nodata & \nodata & 0.40$\pm$0.01 & $7.4^{+0.7}_{-0.6}$ & \nodata & \nodata & 1.08 (86) \\
276: & PL+BB & $4.2^{+7.2}_{-3.2}$ & $1.4^{+0.5}_{-1.2}$ & $1.5^{+5.0}_{-1.3}$ & $ 0.37^{+0.02}_{-0.03}$ & $8.3^{+1.1}_{-1.0}$ & \nodata & \nodata & 0.82 (84)\phantom{1} \\
277: & BB1+BB2 & $4.4^{+3.7}_{-2.5}$ & \nodata & \nodata & 0.33$^{+0.04}_{-0.06}$ & $10.1^{+3.5}_{-1.7}$ & 0.8$^{+0.2}_{-0.3}$& $1.2^{+1.6}_{-0.7}$ & 0.75 (84)\phantom{1} \\
278: %\hline
279: %----------------------------------------------------------------------------
280: F
281: %& PL & 41$\pm$4 & $3.4\pm0.2$ & $44^{+7}_{-5}$ & \nodata & \nodata & \nodata & \nodata & 1.41 (67) \\
282: % & BB & $<$3.2 & \nodata & \nodata & 0.39$\pm$0.02 & $7.7^{+0.9}_{-0.7}$ & \nodata & \nodata & 1.78 (67) \\
283: & PL+BB & $7.6^{+32.6}_{-5.6}$ & $1.6^{+2.0}_{-1.0}$ & 2.6$\pm$2.1 & $ 0.33^{+0.03}_{-0.06}$ & $9.9^{+2.6}_{-1.5}$ & \nodata & \nodata & 1.39 (65)\phantom{1} \\
284: & BB1+BB2 & $6.6^{+6.4}_{-3.4}$ & \nodata & \nodata & 0.31$^{+0.04}_{-0.08}$ & $11.7^{+5.4}_{-2.3}$ & 0.8$^{+0.4}_{-0.3}$& $1.0^{+2.3}_{-0.6}$ & 1.30 (65)\phantom{1} \\
285: %\hline
286: %----------------------------------------------------------------------------
287: B+D+E+F\tablenotemark{f}
288: %& PL & $39\pm2$ & $3.35^{+0.09}_{-0.08}$ & 42$\pm3$ & \nodata & \nodata & \nodata & \nodata & 1.96 (102) \\
289: % & BB & $1.6\pm0.8$ & \nodata & \nodata & $0.391\pm0.009$ & $7.7\pm0.4$ & \nodata & \nodata & 2.43 (102) \\
290: & PL+BB & $9.1^{+7.9}_{-3.8}$ & $2.0^{+0.5}_{-0.4}$ & $3.7^{+5.5}_{-1.9}$ & $0.34\pm0.02$ & $9.3^{+0.9}_{-0.7}$ & \nodata & \nodata & 1.75 (100) \\
291: & BB1+BB2 & $6.3^{+2.0}_{-1.6}$ & \nodata & \nodata & $0.30\pm0.02$ & $12.1^{+2.1}_{-1.4}$ & $0.68^{+0.09}_{-0.07}$ & $1.7^{+0.6}_{-0.5}$ & \phantom{1}\phantom{1}1.14 (100)\smallskip
292: \enddata
293: \tablenotetext{a}{A free normalization factor has been introduced to
294: account for inaccurate flux reconstruction in the MOS spectra,
295: where, in most cases, the source is located on a CCD gap.}
296: \tablenotetext{b}{Assuming photoelectric absorption cross section
297: from \citet{balucinska92} and abundances from \citet{anders89}.}
298: \tablenotetext{c}{In units of 10$^{-5}$ ph. cm$^{-2}$ s$^{-1}$
299: keV$^{-1}$, at 1 keV.} \tablenotetext{d}{Assuming a distance of 60
300: kpc.} \tablenotetext{e}{Only MOS data.} \tablenotetext{f}{Only PN
301: data.}
302: \end{deluxetable*}
303:
304: In order to check for flux variability, we have also
305: simultaneously fit the five available PN spectra and the two MOS
306: spectra for observation C with a double blackbody model with all
307: parameters linked to the same value and a variable normalization
308: factor.
309: From this analysis we can exclude significant ($>$3 $\sigma$) flux
310: variations larger than $\sim$20\% among the different \xmm\
311: observations.
312:
313: Since no significant differences in the spectral parameters are
314: detected and the calibration of the PN instrument has proven to be
315: very stable throughout the \xmm\ mission \citep[see,
316: e.g.,][]{kirsch05stab},
317: %(see, e.g., \textbf{Kirsch...}),
318: a cumulative spectrum of the PN data of observations B, D, E, and
319: F\footnote{Observation A has been excluded because it was taken in a
320: different operating mode, while
321: %, for which a different response matrix must be used.
322: no PN data were available for observation C.} has also been
323: extracted.
324: The resulting net exposure time is 58 ks.
325: %As can be seen in Table \ref{fits}, o
326: Only the double blackbody model gives an acceptable fit to the
327: cumulative spectrum with $kT_1\sim0.3$ keV and $kT_2\sim0.7$ keV
328: (see Fig.~\ref{spec} and Table \ref{fits}). The hydrogen column
329: density is in good agreement with the average value of N$_{\rm
330: H}=5.9\times10^{20}$ cm$^{-2}$ expected towards this region of
331: the SMC \citep{dickey90}. The observed flux in the
332: %0.1
333: 2--10 keV energy range
334: %, obtained from the double blackbody model of the cumulative spectrum,
335: is $1.4\times10^{-13}$ erg cm$^{-2}$ s$^{-1}$,
336: %4.3$\times$10$^{-13}$ erg cm$^{-2}$ s$^{-1}$ and the
337: corresponding to an unabsorbed
338: %bolometric
339: luminosity of $6.1\times10^{34}$ erg s$^{-1}$ for a distance of 60
340: kpc.
341: %is
342: %1.46e35 + 7.7e34
343: %2.2$\times$10$^{35}$ erg s$^{-1}$.
344: The double blackbody model gave
345: also the lowest $\chi^2$ values for the spectra of the single
346: observations, but in those cases it was not the only one compatible
347: with the data. In particular, the power-law plus blackbody model,
348: that is rejected with high confidence by the fit to the cumulative
349: spectrum, gave acceptable fits to all the single spectra.
350: \begin{figure}
351: \resizebox{\hsize}{!}{\includegraphics[angle=-90]{f1.eps}}
352: \caption{\label{spec} Cumulative PN spectrum of \src\ during
353: observations B, D, E, and F. The best-fit double blackbody model is
354: also shown together with the residuals and the separate contribution
355: of the two blackbody components (dotted and dash-dotted lines).}
356: % The residuals refer to the best-fit blackbody plus power-law (upper
357: %panel) and two-blackbodies (lower panel) model.}
358: \end{figure}
359:
360: In addition to the phenomenological models described above, we have
361: also fitted the cumulative spectrum with the magnetar spectral model
362: described in \citet{RCSnanda}. This model, originally proposed by
363: \citet{lyutikov06}, is based on cyclotron resonant scattering of
364: blackbody radiation in a twisted magnetosphere \citep{tlk02}.
365: Although the resulting $\chi^{2}_{r}$ (1.20 for 100 degrees of
366: freedom) is slightly worse than for the double blackbody model, the
367: fit is acceptable. The photoelectric absorption (N$_{\rm
368: H}=5\pm1\times10^{20}$ cm$^{-2}$) and blackbody temperature
369: ($kT=0.32\pm0.08$ keV) are consistent with the values derived from
370: the double blackbody fit (N$_{\rm
371: H}=6.3^{+2.0}_{-1.6}\times10^{20}$ cm$^{-2}$ and
372: $kT=0.30\pm0.02$ keV for the cooler blackbody). The best-fit values
373: of the other spectral parameters
374: %\citep[see][for a detailed description of the parameters]{RCSnanda}
375: are a resonant scattering optical depth of
376: $\tau_{res}=1.2\pm0.2$
377: %$\times$10$^-2$
378: and a particle velocity of $\beta_T=0.48\pm0.12$. These parameters
379: are in the same range as the ones observed in the other magnetar
380: candidates \citep{RCSnanda}. Although
381: %does not provide, in its current version,
382: a direct information on the size of the emitting region cannot be
383: derived in the current version of this model, an approximate
384: estimate gives a radius similar to the one of the cooler component
385: in the double blackbody model.
386:
387:
388: By inspecting the residuals from the best-fit models, we found no
389: significant absorption or emission narrow-line features. We computed
390: upper limits on narrow lines' equivalent widths as a function of the
391: assumed line energy and width $\sigma_E$. This was done by adding
392: Gaussian components to the double blackbody model and computing the
393: allowed range in their normalization. The results for the
394: high-statistics cumulative PN spectrum are summarized in
395: Fig.\,\ref{lines}, where the plotted curves represent the
396: 3\,$\sigma$ upper limits for $\sigma_E=0$ eV.
397: \begin{figure}
398: \centering
399: \resizebox{\hsize}{!}{\includegraphics{f3.eps}}
400: \caption{\label{lines} Upper limits (at 3 $\sigma$) to the
401: equivalent width of narrow lines (either in emission or absorption)
402: in the PN cumulative spectrum of \src\ during observations B, D, E,
403: and F.}
404: \end{figure}
405:
406: We performed a timing analysis to measure the source pulse period in
407: each data set. After correcting the photon arrival times to the
408: Solar system barycenter,
409: we derived the best period values based on a $Z_2^2$ periodogram
410: analysis \citep{buccheri83}. The resulting values are indicated in
411: Table \ref{log}.
412: Considering also the periods measured by \emph{Chandra}
413: \citep{mcgarry05}, a linear fit to the ten
414: %\emph{XMM-Newton} and \emph{Chandra}
415: %available period measurements %(Fig.~\ref{periods})
416: values yields a period derivative
417: $\dot{P}=(1.9\pm0.1)\times10^{-11}$ s s$^{-1}$ ($\chi^2_{r}$ of 1.32
418: for 8 degrees of freedom).
419:
420: Since observation D and E were performed only two days apart, we
421: tried to better constrain the spin-down rate through a
422: phase-coherent timing analysis of the two datasets. However, the
423: periods uncertainties during each observation
424: %have not an uncertainty small enough
425: are too large to allow the prediction of the phase of the next
426: observation to better than a pulse cycle.
427:
428: Many AXPs and SGRs are known to exhibit significant changes in their
429: pulse profiles \citep[e.g.][]{kaspi07,gogus02}. To search for
430: possible pulse shape variations in \src\ as a function of time, we
431: compared the folded light curves using a Kolmogorov-Smirnov test.
432: Taking into account the unknown relative phase alignment, all the
433: light curves are compatible with the same profile. We therefore
434: summed them after appropriate phase shifts\footnote{We selected the
435: shifts that maximized the Kolmogorov-Smirnov statistics comparing
436: subsequent observations.}.
437: The resulting pulse profiles in the soft (0.2--1 keV, $S$) and hard
438: (1--6 keV, $H$) energy ranges, together with their hardness ratio
439: (computed as $(H-S)/(H+S)$), are shown in Fig. \ref{profile}. This
440: analysis does not show any significant profile changes with energy.
441: The pulsed fraction\footnote{The pulsed fraction is defined as
442: $(C_{\rm{max}}-C_{\rm{min}})/(C_{\rm{max}}+C_{\rm{min}})$, where
443: $C_{\rm{max}}$ and $C_{\rm{min}}$ are the background-subtracted
444: count rates at the peak and at the minimum.} in the 0.2--6 keV energy
445: range is $(32\pm3)$\%.
446: \begin{figure}
447: \centering
448: \resizebox{\hsize}{!}{\includegraphics{f2.eps}}
449: \caption{\label{profile} Background subtracted pulse profile of
450: \src\ in different energy ranges (as indicated in the panels) and
451: corresponding hardness ratio (PN data of observations B, D, E, F).}
452: \end{figure}
453:
454: \section{Discussion}
455: The new \xmm\ observations reported here indicate that \src\
456: continued to spin-dow at a rate of $\sim$$1.9\times10^{-11}$ s
457: s$^{-1}$, consistent with previous results \citep{mcgarry05}.
458: Although the errors on P are quite large, there is no evidence for
459: major torque changes. Also the pulse profile, spectral shape and
460: flux are consistent with no major changes, confirming that this AXP
461: is one of the most stable members of its class. This characteristic
462: allows us to sum up all the \xmm\ observations taken with the same
463: instrumental settings.
464:
465: The cumulative PN spectrum of \src\ cannot be adequately fit by a
466: power-law plus blackbody model. This is the first magnetar for which
467: such a conclusion can be derived based only on the X--ray data,
468: thanks to the very low interstellar absorption.
469:
470: A good fit is instead obtained with a double blackbody model. The
471: known distance of the SMC \citep[60 kpc,][]{keller06} allows a
472: precise measure
473: %of the luminosity and
474: of the size of the emitting
475: region of the two blackbodies.
476: The $\sim$2 km radius of the region associated to the hotter
477: component is compatible with a hot spot on the neutron star surface.
478: The radius of the emitting region of the cooler blackbody,
479: 12.1$^{+2.1}_{-1.4}$ km,
480: %(assuming a distance of 60 kpc)
481: %is instead derived, indicating that such emission might come from
482: is consistent with a large fraction of the magnetar surface.
483: However, the strong pulsation below 1 keV (see Fig.~\ref{profile}),
484: where this component dominates
485: %Since this spectral component dominates the emission of
486: %\src\ in the 0.1--1 keV energy range
487: (see Fig.~\ref{spec}), indicates
488: % and strong pulsations are detected below 1
489: %keV (see Fig.~\ref{profile}),
490: that it cannot come from the whole neutron star surface. A similarly
491: large blackbody radius ($\sim$10 km) was also derived from the
492: spectrum of the AXP XTE~J1810--197 observed by \emph{ROSAT} before
493: the onset of its outburst in 2003 \citep{halpern05} and from
494: phase-resolved spectroscopy of \xmm\ observations of the same object
495: \citep{israel08}
496: and of the AXP 1E~1048.1--5937 \citep{tmt05}. However,
497: in these cases, the less accurately known distance and the high
498: interstellar absorption produce large uncertainties on the emitting
499: region size.
500:
501: Assuming that the thermal photons are produced on the neutron star
502: surface (and not, for instance, in the magnetosphere) and
503: considering that the blackbody is the most efficient thermal
504: emission process at a given temperature, the radius of the region
505: emitting the colder blackbody in \src\ is a firm lower limit to the
506: radius of the compact object. This limit is not large enough to
507: exclude any of the most popular equations of state for neutron
508: stars, but it is the most constraining lower limit ever obtained for
509: a magnetar.
510:
511: Magnetar spectra are expected to be more complex than a double
512: blackbody. In fact most magnetar spectra cannot be fitted by such a
513: simple model, that underestimates their emission in the 5--10 keV
514: energy range. In all these cases a power-law tail in hard X--rays
515: ($>$20 keV) has been detected \citep[see,
516: e.g.,][]{kuiper06,gotz06,leyder08} and it is likely responsible also
517: for the hard excess below 10 keV \citep{nakagawa08}. From a
518: theoretical point of view, the emission expected from a magnetar has
519: two components, with a thermal part directly from the surface, and a
520: non-thermal one due to emission reprocessed in the magnetosphere
521: \citep{lyutikov06,fernandez07}. We found that also \src\ can be fit
522: by a model of this kind \citep{RCSnanda}.
523:
524: No compelling detections of spectral features in the persistent
525: X--ray emission of magnetars have been reported so far. \xmm\ and
526: \emph{Chandra} observations yielded strong upper limits (equivalent
527: width $\lesssim$10 eV) for 4U\,0142+61 \citep{juett02},
528: 1E\,1048.1$-$5937 \citep{tmt05}, and 1E\,2259+586 \citep{woods04} in
529: the 0.7--5 keV range. However, the high interstellar absorption
530: towards these objects causes a series of absorption edges in the
531: observed spectrum at low energies, introducing large systematic
532: uncertainties in the search for features in the intrinsic AXP
533: spectrum.
534: %due to the unknown abundances of the interstellar medium.
535: A hint of a spectral feature at $\sim$0.9 keV was noted by
536: \cite{durant06} in the spectrum of the AXP 4U\,0142+61, after its
537: deconvolution from interstellar absorption using the edges directly
538: observed in the high resolution X--ray spectra, but also this
539: measure is affected by the poorly constrained abundances of most
540: interstellar elements.
541: %prevented a sensitive analysis at lower energies.
542: As shown in Figure \ref{lines}, for \src\ we did not find evidence
543: for lines, but, although this source is one of the dimmest AXPs, we
544: could put stringent limits on narrow features
545: %down to 0.4 keV.
546: in the soft X--ray band, which are virtually independent of the
547: photoelectric absorption model. The dipolar magnetic field derived
548: from the spin-down rate of \src\ is \mbox{$4\times10^{14}$ G},
549: corresponding to a proton cyclotron energy of $\sim$2.5 keV;
550: however, a cyclotron line at lower energies is expected if the
551: cyclotron emission or absorption process occurs far from the neutron
552: star surface, while a feature at higher energies is produced if
553: strong multipolar magnetic field components are present. These
554: effects, in addition to other processes that can suppress the
555: spectral features \citep[see, e.g.,][]{ho03}, make the lack of
556: proton cyclotron lines in the X--ray spectra of magnetars compatible
557: with their magnetic fields of (6--$250)\times10^{13}$ G
558: (corresponding to proton cyclotron energies of 0.4--15 keV) derived
559: from their timing properties.
560: \acknowledgements
561:
562: We thank Nanda Rea for fitting the spectrum with the RCS model and
563: useful discussion. This research has been partially supported by the
564: Italian Space Agency.
565: \newpage
566: \bibliographystyle{apj}
567: \bibliography{biblio}
568:
569: \end{document}
570: