1: %\documentclass[referee]{aa}
2: \documentclass{aa}
3: \usepackage{graphics,epsfig,lscape,times}
4:
5: \usepackage{natbib}
6: \bibpunct{(}{)}{;}{a}{}{,}
7:
8: \newcommand{\ergcm}[1]{$\times 10^{#1}$ erg cm$^{-2}$ s$^{-1}$}
9: \newcommand{\oergcm}[1]{$10^{#1}$ erg cm$^{-2}$ s$^{-1}$}
10: \newcommand{\ergs}[1]{$\times 10^{#1}$ erg s$^{-1}$}
11: \newcommand{\oergs}[1]{$10^{#1}$ erg s$^{-1}$}
12: \newcommand{\hcm}[1]{$\times 10^{#1}$ cm$^{-2}$}
13: \newcommand{\ohcm}[1]{$10^{#1}$ cm$^{-2}$}
14: \newcommand{\expo}[1]{$\times 10^{#1}$}
15: \newcommand{\oexpo}[1]{$10^{#1}$}
16: \newcommand{\nh}{$\mathrm{N_H}$}
17: \newcommand{\gr}{\hbox{g$_{\rm r}$}}
18: \newcommand{\ecp}{\hbox{E$_{\rm cp}$}}
19: \newcommand{\eline}{\hbox{E$_{\rm line}$}}
20: \newcommand{\wline}{\hbox{$\sigma_{\rm line}$}}
21: \newcommand{\eqw}{\hbox{EW}}
22: \newcommand{\lx}{\hbox{L$_{\rm x}$}}
23: \newcommand{\lmax}{\hbox{L$_{\rm x}^{\rm max}$}}
24: \newcommand{\ct}{cts s$^{-1}$}
25: \newcommand{\mv}{\hbox{$m_{\rm V}$}}
26: \newcommand{\mr}{\hbox{$m_{\rm R}$}}
27: \newcommand{\mb}{\hbox{$m_{\rm B}$}}
28: \newcommand{\HII}{\ion{H}{II}}
29: \newcommand{\Hone}{\ion{H}{I}}
30: \newcommand{\Hmol}{H$_2$}
31: \newcommand{\SII}{\ion{S}{II}}
32: \newcommand{\Halp}{H${\alpha}$}
33: \newcommand{\ltsima}{$\buildrel < \over \sim$}
34: \newcommand{\lsim}{\lower.5ex\hbox{\ltsima}}
35: \newcommand{\gtsima}{$\buildrel > \over \sim$}
36: \newcommand{\gsim}{\lower.5ex\hbox{\gtsima}}
37: %
38: \newcommand{\rxa}{\hbox{\object{RX\,J0420.0$-$5022}}}
39: \newcommand{\rxb}{\hbox{\object{RX\,J0720.4$-$3125}}}
40: \newcommand{\rxc}{\hbox{\object{RX\,J0806.4$-$4123}}}
41: \newcommand{\rxd}{\hbox{\object{1RXS\,J130848.6+212708}}}
42: \newcommand{\rxe}{\hbox{\object{RX\,J1605.3+3249}}}
43: \newcommand{\rxf}{\hbox{\object{RX\,J1856.4$-$3754}}}
44: \newcommand{\rbs}{\hbox{\object{RBS1223}}}
45: \newcommand{\rbt}{\hbox{\object{RBS1556}}}
46: \setlength{\topmargin}{-30mm}
47: %
48: \begin{document}
49:
50: \title{Evidence for precession of the isolated neutron star \object{RX\,J0720.4$-$3125}
51: \thanks{Based on observations with XMM-Newton,
52: an ESA Science Mission with instruments and contributions
53: directly funded by ESA Member states and the USA (NASA)}}
54:
55: \author{F.~Haberl\inst{1} \and R.~Turolla\inst{2} \and C.P.~De Vries\inst{3} \and
56: S.~Zane\inst{4} \and J.~Vink\inst{5} \and M.~M\'endez\inst{3} \and
57: F.~Verbunt\inst{5}}
58:
59: \titlerunning{Precession of the isolated neutron star RX\,J0720.4$-$3125}
60: \authorrunning{Haberl et al.}
61:
62: \offprints{F. Haberl, \email{fwh@mpe.mpg.de}}
63:
64: \institute{Max-Planck-Institut f\"ur extraterrestrische Physik,
65: Giessenbachstra{\ss}e, D-85748 Garching, Germany;
66: \and
67: Department of Physics, University of Padua, via Marzolo 8,
68: I-35131, Padova, Italy; %\email{turolla@pd.infn.it}
69: \and
70: SRON, Netherlands Institute for Space Research, Sorbonnelaan 2,
71: NL-3584 CA Utrecht, The Netherlands; %\email{C.P.de.Vries@sron.nl}
72: \and
73: Mullard Space Science Laboratory, University College London, Holmbury St. Mary,
74: Dorking, Surrey, RH5 6NT, UK; %\email{sz@mssl.ucl.ac.uk}
75: \and
76: University Utrecht, PO Box 80000,
77: NL-3508 TA Utrecht, The Netherlands; %\email{j.vink@astro.uu.nl}
78: }
79:
80: \date{Received 27 February 2006 / Accepted 23 March 2006}
81:
82: \abstract{
83: The XMM-Newton spectra of the isolated neutron star RX\,J0720.4$-$3125 obtained over
84: 4.5 years can be described by sinusoidal variations in the inferred blackbody
85: temperature, the size of the emitting area and the depth of the absorption line with a
86: period of 7.1$\pm$0.5 years, which we suggest to be the precession period of
87: the neutron star. Precession of a neutron star with two hot spots of
88: different temperature and size, probably not located exactly in antipodal
89: positions, may account for the variations in the X-ray spectra, changes in the
90: pulsed fraction, shape of the light curve and the phase-lag between soft
91: and hard energy bands observed from RX\,J0720.4$-$3125. An independent
92: sinusoidal fit to published and new pulse timing residuals from a coherent analysis
93: covering $\sim$12 years yields a consistent period of 7.7$\pm$0.6 years supporting
94: the precession model.
95:
96: \keywords{stars: individual: \rxb\ --
97: stars: neutron --
98: stars: magnetic fields --
99: X-rays: stars}}
100:
101: \maketitle
102:
103: \section{Introduction}
104:
105: The 8.39 s X-ray pulsar \rxb\ was discovered in the ROSAT
106: all-sky survey data by
107: \citet{1997A&A...326..662H} and identified with a faint blue optical star
108: \citep{1998A&A...333L..59M,1998ApJ...507L..49K,2003ApJ...590.1008K}
109: which shows a proper motion of about 100 mas/yr \citep{2003A&A...408..323M}.
110: The bright, soft X-ray source belongs to a small group of nearby radio-quiet isolated
111: neutron stars with blackbody-like thermal X-ray spectra
112: \citep[for recent reviews
113: see][]{2000PASP..112..297T,2001xase.conf..244M,2004AdSpR..33..638H,2005fysx.conf...39H}.
114:
115: Broad absorption lines in the X-ray spectra of most of these objects %, including \rxb,
116: were reported and are usually interpreted as due to resonant absorption at
117: the proton cyclotron energy and/or bound-bound transitions in H or H-like He
118: \citep[][ hereafter H04]{2003A&A...403L..19H,2004ApJ...608..432V,2005ApJ...627..397Z,2004A&A...419.1077H}.
119: This is suggestive of neutron star magnetic field strengths of
120: $\approx$10$^{13}$--$10^{14}$~G, values which are consistent with those derived
121: from pulse timing for the period derivatives of \rxb\
122: \citep[][ hereafter K05]{2002MNRAS.334..345Z,2004MNRAS.351.1099C,2005ApJ...628L..45K}
123: and \rbs\ \citep{2005ApJ...635L..65K} if one assumes magnetic dipole braking. Moreover,
124: \rxb\ shows variations of the absorption line depth with pulse phase (H04) with the
125: line being weakest near intensity maximum and deepest at the declining part of the pulse.
126:
127: Among the seven thermal isolated neutron stars discovered with ROSAT,
128: \rxb\ is unique by showing a gradual change of the X-ray spectrum on
129: a time scale of years accompanied by an energy-dependent change in
130: the pulse profile \citep[][ hereafter D04]{2004A&A...415L..31D}.
131: In this letter we report on further XMM-Newton observations of this enigmatic
132: source which show that the long-term trends have reversed. We present
133: pulse-phase averaged and pulse-phase resolved spectra from the EPIC-pn
134: instrument. The results of our spectral and temporal analysis show strong
135: evidence for a cyclic variation of the X-ray properties which strengthens the
136: case of a precessing neutron star.
137:
138: \section{XMM-Newton observations}
139:
140: \begin{table}
141: \caption[]{XMM-Newton EPIC-pn observations of \rxb.}
142: \begin{tabular}{lllll}
143: \hline\noalign{\smallskip}
144: \multicolumn{1}{l}{Orbit} &
145: \multicolumn{1}{l}{Observation} &
146: \multicolumn{1}{l}{Date} &
147: \multicolumn{1}{l}{Setup$^1$} &
148: \multicolumn{1}{l}{Exp. [s]} \\
149:
150: \noalign{\smallskip}\hline\noalign{\smallskip}
151: 0078 & 0124100101 & 2000 May 13 & FF thin & 44275 \\
152: 0175 & 0132520301 & 2000 Nov. 21-22 & FF medium & 22952 \\
153: 0533 & 0156960201 & 2002 Nov. 6-7 & FF thin & 25696 \\
154: 0534 & 0156960401 & 2002 Nov. 8-9 & FF thin & 27273 \\
155: 0622 & 0158360201 & 2003 May 2-3 & SW thick & 51022 \\
156: 0711 & 0161960201 & 2003 Oct. 27 & SW thin & 12711 \\
157: 0711 & 0161960201 & 2003 Oct. 27-28 & SW medium & 17429 \\
158: 0815 & 0164560501 & 2004 May 22-23 & FF thin & 26642 \\
159: 0986 & 0300520201 & 2005 April 28 & FF thin & 37101 \\
160: 1060 & 0300520301 & 2005 Sep. 23 & FF thin & 34989 \\
161: 1086 & 0311590101 & 2005 Nov. 12-13 & FF thin & 33777 \\
162: \noalign{\smallskip}\hline
163: \end{tabular}
164:
165: $^{(1)}$ Read-out mode and filter; FF: Full Frame;
166: SW: Small Window.
167: \label{xmm-obs}
168: \end{table}
169:
170: The soft X-ray source \rxb\ was observed with XMM-Newton \citep{2001A&A...365L...1J} as
171: calibration target until 2004 when it was recognised to exhibit a variable spectrum.
172: Since then we monitored the spectral evolution of the source.
173: Here we utilise the data collected with the EPIC-pn camera \citep[][]{2001A&A...365L..18S}.
174: The details of the XMM-Newton observations with the instrumental setup used for
175: EPIC-pn are summarised in Table~\ref{xmm-obs}. \rxb\ is slightly too bright
176: for the full-frame (FF) read-out mode causing photon pile-up effects which systematically
177: harden the spectrum (for details of this effect see H04).
178: For a relative comparison of the spectra we ignore this effect but emphasise
179: that small systematic differences are present between the FF mode spectra and those obtained
180: in the faster small window (SW) mode which are free of pile-up effects.
181: We do not use the observation with thick filter due to the much lower
182: efficiency and different detector response.
183: We screened out strong background flares and list the resulting net exposure times
184: (including dead time which is important for SW mode) in Table~\ref{xmm-obs}.
185: We processed the data using the XMM-Newton analysis system SAS6.5 and extracted spectra
186: and light curves from circular regions with radius of 30\arcsec.
187:
188: \subsection{Pulse-phase averaged spectra}
189:
190: We extracted and analyzed pulse-phase averaged spectra for the 10 observations
191: (at different epochs and different instrumental setup) following H04 who found that
192: an absorbed blackbody model with a broad Gaussian absorption line represents the
193: spectra of \rxb. We adopted this model and fitted the spectra simultaneously with
194: the absorption column density (\nh), the line energy and width common to all spectra.
195: We allowed temperature and the normalizations of blackbody and line to vary freely
196: between the spectra (except when observations were performed shortly after each other
197: as those from satellite revolutions 533/534 and the two parts of 711).
198: We present the EPIC-pn spectra obtained in FF mode with thin filter together with their
199: best-fit model in Fig.~\ref{fig-spectra} which shows that the long-term trend of spectral
200: hardening has reversed. We report the fit results in Table~\ref{tab-fits}. The common
201: parameters obtained from the best fit are \nh\ = 1.01$\pm$0.03 \hcm{20}, a line energy
202: of 280$\pm$6~eV with a width of $\sigma = 90\pm$5~eV.
203:
204: \begin{figure}
205: \begin{center}
206: \resizebox{\hsize}{!}{\includegraphics[clip=,angle=-90]{allffthin_phabs_absline_bbodyrad_pn_s.ps}}
207: \end{center}
208: \caption{Pulse-phase averaged EPIC-pn spectra of \rxb\ from the seven observations with
209: the same instrumental setup (FF read-out mode, thin filter). The same detector
210: efficiency for all the spectra allows a direct comparison and demonstrates the
211: long-term spectral changes. The softest spectrum (the uppermost at low
212: energies) was obtained in May 2000, while the hardest (the lowest at
213: low energies) is that from May 2004. After May 2004 the spectra
214: increased monotonically at low energies.}
215: \label{fig-spectra}
216: \end{figure}
217:
218: \begin{figure}
219: \begin{center}
220: \resizebox{7.19cm}{!}{\includegraphics[clip=,angle=-90]{allthinmed_kT.ps}}
221: \resizebox{7.19cm}{!}{\includegraphics[clip=,angle=-90]{allthinmed_eqw.ps}}
222: \resizebox{7.19cm}{!}{\includegraphics[clip=,angle=-90]{allthinmed_rns.ps}}
223: \end{center}
224: \caption{Variation of temperature, line equivalent width and radius of the emitting region
225: of \rxb\ derived from a simultaneous fit to the EPIC-pn spectra using an absorbed
226: blackbody model with a broad absorption line at 280 eV. The sine wave with a period
227: of 7.1 years indicates the best fit to the data derived from the FF mode (marked with
228: filled circle) observations. For the sine fit of EW and radius we fixed the period at
229: the value derived from kT. SW mode data from Oct. 2003 (open circle) are not used in
230: the fits.}
231: \label{fig-precession}
232: \end{figure}
233:
234: \begin{table}[t]
235: \caption[]{Spectral analysis of pulse-phase averaged X-ray spectra.}
236: \begin{tabular}{lccc}
237: \hline\noalign{\smallskip}
238: \multicolumn{1}{l}{Orbit} &
239: \multicolumn{1}{c}{kT [eV]} &
240: \multicolumn{1}{c}{\eqw\ [eV]} &
241: \multicolumn{1}{c}{Flux$^{(1)}$ [erg cm$^{-2}$]} \\
242:
243: \noalign{\smallskip}\hline\noalign{\smallskip}
244: 0078 & 86.6$\pm$0.4 & $-$5.02$\pm$4.5 & 0.98\expo{-11} \\
245: 0175 & 86.5$\pm$0.5 & $+$8.68$\pm$7.7 & 1.01\expo{-11} \\
246: 0533/534 & 88.3$\pm$0.3 & $-$21.5$\pm$2.6 & 0.98\expo{-11} \\
247: 0711/711 & 91.3$\pm$0.6 & $-$73.7$\pm$4.9 & 1.19\expo{-11} \\
248: 0815 & 93.8$\pm$0.4 & $-$72.4$\pm$4.7 & 1.02\expo{-11} \\
249: 0986 & 93.5$\pm$0.4 & $-$68.3$\pm$5.2 & 1.02\expo{-11} \\
250: 1060 & 93.2$\pm$0.4 & $-$67.4$\pm$4.3 & 1.02\expo{-11} \\
251: 1086 & 92.6$\pm$0.4 & $-$67.5$\pm$3.5 & 1.04\expo{-11} \\
252: \noalign{\smallskip}\hline\noalign{\smallskip}
253: \end{tabular}
254:
255: Throughout the paper errors are given for a 90\% confidence level. \\
256: $^{(1)}$Observed flux is given in the energy band 0.1$-$2.4 keV. The
257: systematically higher flux from satellite revolution 711 is caused
258: by the systematic difference of FF and SW mode (FF pile-up losses).
259: The statistical uncertainties on the flux are less then 2\ergcm{-14}, which are
260: negligible compared to the stability of the instrument of about 1\% between
261: observations. % (as derived from supernova remnant observations).
262: \label{tab-fits}
263: \end{table}
264:
265: The best fit values for the blackbody temperature (kT), the line equivalent width
266: (EW) and the inferred emitting area (radius for a circular region with an assumed source
267: distance of 300 pc) as function of time are shown in Fig.~\ref{fig-precession}.
268: All parameters can be described with a sinusoidal variation and we derive a period
269: of 2580$\pm$180 days for the variation in temperature. In the fit we do not
270: include the values obtained from the SW mode spectra because of the possible systematic
271: difference (pile-up, see above).
272: Among the three parameters the temperature can be probably best constrained from the
273: X-ray spectra. It is also not clear if there are additional variations in the
274: absorption line energy and width which we assume to be constant in our current analysis.
275: Therefore, we also fit a sine to the variations in EW and emission radius, but fix the
276: period at the value found from the temperature variation.
277:
278: \subsection{Pulse-phase variations}
279:
280: We assigned pulse phases to all detected events using the X-ray timing
281: ephemerides inferred by K05 (the ``All Data'' solution)
282: and folded the light curves in different energy bands to produce pulse profiles.
283: Examples from two XMM-Newton observations are presented in Fig.~\ref{fig-profile}.
284: As was found by D04 the pulse profile became deeper with time. In 2004
285: the shape of the profile changed in particular in the hard band and the hardness ratio was
286: higher on average and showed more modulation as compared to the observation in May 2000.
287: To investigate the spectral evolution as function of pulse phase
288: we divided the pulse into five phase intervals of equal length with the first interval from
289: phase 0.0 to 0.2.
290:
291: \begin{figure}
292: \begin{center}
293: \resizebox{6.5cm}{!}{\includegraphics[clip=,angle=-90]{P0124100101PNS003PIEVLI0000_sc_hr_phase.ps}}
294: \vspace{3mm}
295: \resizebox{6.5cm}{!}{\includegraphics[clip=,angle=-90]{P0164560501PNS001PIEVLI0000_sc_hr_phase.ps}}
296: \end{center}
297: \caption{Pulse profiles of \rxb\ in two different energy bands (soft S: 0.12-0.4 keV;
298: hard H: 0.4-1.0 keV) together with the hardness ratio HR=H/S. The top panels are
299: obtained from the observation in May 2000 and the bottom panels from that of May 2004.}
300: \label{fig-profile}
301: \end{figure}
302:
303:
304: To avoid any systematic shifts due to different instrumental setups, we use
305: only the seven observations in FF mode with thin filter. Similarly to the
306: analysis of phase-averaged spectra, we performed a simultaneous fit with the
307: same model to the 5$\times$7 spectra. \nh, the line
308: energy and width were treated common to all spectra and were found to be
309: consistent within the errors with the values derived from the phase-averaged
310: spectra. In Fig.~\ref{fig-pulse} the derived line EW is plotted versus
311: temperature kT. For each observation the evolution of the two parameters
312: during the X-ray pulse follows an ellipse-like track (sampled by five points
313: from our finite number of phase intervals) in the kT-EW plane. The evolution
314: proceeds counter-clockwise; the point marked with a circle indicates the
315: phase interval 0.0-0.2. Several remarkable features are seen in
316: Fig.~\ref{fig-pulse}: 1) The variation in kT was smaller during the
317: first observations, consistent with the $\sim$2.5~eV value reported by H04.
318: During the later observations the amplitude in the kT variation increased
319: to $\sim$6~eV, almost as large as the long-term change of $\sim$8~eV seen
320: in the phase-averaged spectra (Fig.~\ref{fig-precession}). 2) The
321: amplitude in the line EW variation is $\sim$40~eV and did not change
322: significantly between the observations.
323: % 3) The inclination of the
324: %ellipse (pentagon) changes during the long-term evolution which may
325: %however just be the consequence of points 1) and 2).
326: 3) The long-term
327: trend reversal of the evolution seen from the phase-averaged spectra
328: (Fig.~\ref{fig-precession}) after the May 2004 observation is similarly
329: seen at all pulse phases. We note that during the
330: pulse variation of the first observation the line is formally detected as
331: emission line in phase interval 0.8-1.0, i.e. before reaching the
332: intensity maximum. Also during the second observation there is an
333: indication that the line in the phase-averaged spectrum is in emission
334: rather than absorption. However, we can not exclude that this is caused
335: by uncertainties in the calibration which may result in a systematic shift of the
336: EW by $\sim$10~eV (for all spectra).
337:
338: \begin{figure}
339: \begin{center}
340: \resizebox{8.1cm}{!}{\includegraphics[clip=,angle=-90]{allffthin_phase5_kT_eqw.ps}}
341: \end{center}
342: \caption{Equivalent width of the absorption line vs. temperature kT derived from
343: the FF observations with thin filter. Lozenges denote the values derived
344: from the phase-averaged spectra. During the pulse the parameters evolve
345: counter-clockwise, the circle marks phase 0.0-0.2.}
346: \label{fig-pulse}
347: \end{figure}
348:
349:
350: \section{Discussion}
351:
352: The medium resolution X-ray spectra obtained by the
353: EPIC-pn instrument show that the long-term spectral evolution
354: discovered by D04 has reversed. The spectra
355: became softer following the observation of May 2004 when the
356: phase-averaged spectrum was hardest. All spectra can be well modeled
357: by an absorbed blackbody continuum with a broad absorption line of
358: Gaussian shape. The spectral variations both on time scales of years
359: and during the 8.39 s pulse can be explained by variations in the inferred
360: blackbody temperature, the depth of the absorption line and the size of the
361: emitting area.
362: In the simplest approach we did not allow changes in the energy and
363: width of the line. If the line is caused by cyclotron absorption
364: both parameters may be expected to vary. This requires a more
365: detailed investigation of the spectra also involving the higher resolution
366: RGS data which is work in progress.
367:
368: Blackbody temperature, size of emitting area and absorption line depth
369: (equivalent width) derived from the phase-averaged
370: spectra follow a sinusoidal variation with a period of ~7.1 years.
371: Although the XMM-Newton observations do not yet cover a full period,
372: it should be noted that the temperature of 79$\pm$4~eV derived from the
373: ROSAT spectrum (H04), obtained in Sept. 1993, $\sim$6.6 years before the first
374: XMM-Newton observation is also consistent with a periodic behavior of \rxb.
375: The involved time scale is strongly suggestive of a scenario with a
376: freely precessing neutron star, as first suggested by D04. The
377: observed period of $\sim$7.1 yr is then naturally interpreted as the precession
378: period of the neutron star.
379:
380: In that case the precession period should also be visible in the residual
381: phase shifts seen in the coherent timing analysis of the spin evolution of
382: \rxb\ (K05). Therefore, we fitted the phase residuals as inferred by K05
383: with a sinusoidal function, including the three EPIC-pn data points for
384: the new XMM-Newton observations of 2005.
385: The additional data points formally exclude the cubic model used by K05
386: and favour the sinusoidal model. We obtain a
387: preliminary period of 2830$\pm$220 days (7.7$\pm$0.6 years) from the phase
388: residual analysis. This period is somewhat longer, but consistent with that
389: inferred from the spectral analysis. The three ROSAT observations extend
390: the time coverage to more than 12 years, i.e. more than 1.5 precession
391: cycles. The results of a detailed timing analysis which directly includes
392: precession in the spin frequency model will be presented in a forthcoming
393: paper.
394:
395: A picture based on thermal emission from two hot spots on the
396: surface of a freely precessing neutron stars appears indeed
397: promising in explaining many of the peculiar characteristics of
398: \rxb\ which have been a challenge so far. In this scenario
399: the long-term change in temperature is produced by the different
400: (phase-averaged) fractions of the two spots which enter into view
401: as the star precesses. In order for such a model to work, the two
402: emitting regions need to have different temperatures and sizes, as
403: it has been recently proposed in the case of another member of the
404: same class \citep[\rbs, ][]{2005A&A...441..597S}.
405: The nearly, but not exactly sinusoidal pulse profile
406: (Fig.~\ref{fig-profile}, D04) and the - to first order - successful
407: fit of a phase-connected timing solution indicates that the poles can
408: not be exactly antipodal, but must be predominantly in the same
409: (East-West) hemisphere.
410:
411: This is again similar
412: to what was found for \rbs. During the first XMM-Newton observation (May
413: 2000), $\mathrm{kT}\sim 85$~eV and the cooler, larger spot was
414: predominantly visible, while four years later (May 2004),
415: precession brought into view mostly the second, hotter and smaller spot,
416: increasing the temperature to $\sim$95~eV. This may explain the
417: observed temperature/emitting area variations and their
418: anti-correlation (see Fig.~\ref{fig-precession}). It should also be noted
419: here that the total flux we see from the source stays quite constant, i.e.
420: the two poles contribute about the same amount. The presence of two different,
421: non-antipodal spots may also account for the observed changes in
422: the pulsed fraction, light curve shape and for the apparent (spin)
423: phase-lag seen between the soft and hard band.
424:
425: In Fig.~\ref{figmod} we show the results obtained for such a model with
426: a numerical code adapted from that discussed by \citet{2006MNRAS.366..727Z}.
427: First the phase-resolved spectrum is computed, having chosen values
428: of the temperature and size of the two spots ($\mathrm{T_{1,2}}$,
429: $\theta_{1,2}$), and of their relative angular displacement
430: $\theta_0$, to match the main observational features of
431: \rxb. We then compute the phase-averaged
432: spectrum and repeat the calculation at different precession phases for a chosen
433: value of the precession angle $\alpha$. The two angles $\chi$ and $\xi$ denote
434: the inclination of the line-of-sight
435: with respect to the precession axis and that between the centre of one spot and
436: the rotation axis, respectively. We assume isotropic blackbody emission from
437: the caps and no attempt has been made to model the absorption feature.
438: Relativistic ray-bending is included. Since the present calculation has mainly
439: illustrative purposes, we do not account for the detector response and
440: interstellar absorption nor do we look for parameter fine-tuning.
441:
442: \begin{figure*}
443: \begin{center}
444: \resizebox{7.8cm}{!}{\includegraphics[clip=]{precession_kT-R.ps}}% hoffset=20 voffset=-270 hscale=40 vscale=40 angle=0}
445: \resizebox{7.8cm}{!}{\includegraphics[clip=]{precession_lc-HR.ps}}% hoffset=260 voffset=-270 hscale=40 vscale=40 angle=0}
446: \end{center}
447: \caption[]{Left panel: the variation of the blackbody temperature and radius
448: (both normalised) as a function of the precession phase. Right panel: the
449: evolution of the light curve (0.12$-$1~keV band) and hardness ratio with
450: precession phase; full, dashed and dash-dotted lines refer to phases
451: 0.05, 0.4 and 0.75, respectively. Here $\alpha=20^\circ$,
452: $\mathrm{T_1}=80$~eV, $\mathrm{T_2}=100$~eV, $\sin\theta_1=0.8$,
453: $\sin\theta_2=0.6$, $\theta_0=160^\circ$, $\chi=75^\circ$ and $\xi=5^\circ$.}
454: \label{figmod}
455: \end{figure*}
456:
457: As can be seen from Fig.~\ref{figmod}, the model is capable of reproducing the
458: observed variations of the blackbody temperature and emitting area, and
459: also their phase anti-correlation. The anti-correlation of the hardness ratio
460: with the total intensity is also recovered, and the maximum of the hardness
461: ratio shifts ahead in phase. The light curve modulation evolves and the shape
462: changes, becoming more sharply peaked at certain epochs.
463: Although no proper fitting of the observed light curves and spectra has been
464: attempted at this stage (work in
465: progress), our preliminary analysis strongly supports the idea that \rxb\
466: is a precessing, nearly aligned rotator, seen almost equator-on.
467: First indications seem to favour precession angles $\ga 10^\circ$,
468: larger than those found for radio-pulsars.
469: We point out that the modulation is sensitive to the value of M/R
470: (2GM/Rc$^2$ = 0.42 in our example of Fig.~\ref{figmod}). Although present
471: uncertainties prevent us from reaching any firm conclusion, future improvements in
472: the modelling and new observations may allow us to derive
473: constraints on the equation of state.
474:
475: The variations we see in the absorption line could also be related to the different
476: properties of the two emitting spots. Two misaligned caps are suggestive of a magnetic
477: field structure more complex than a core-centred dipole. A different field
478: configuration could be then the origin of the different temperature and size
479: of the spots. In both \rbs\ and \rxb\ the inferred size of the
480: hotter spot is smaller than that of the cooler spot, probably because the former is
481: more confined due to a steeper surface gradient of the magnetic field. A further
482: possibility is that the line
483: changes (both with spin and precession phase) arise because of geometrical effects.
484: The rays which reach the observer are at different angles with the magnetic field at
485: different phases. This means that the contribution of ordinary and extraordinary
486: photons to the total spectrum is phase-dependent. Since extraordinary photons
487: contribute most to resonant absorption at the proton cyclotron energy, one expects
488: the line properties to change with phase.
489:
490: A detailed analysis of the implications of the precession of \rxb\
491: for the physics of neutron star interiors is outside
492: the scope of this letter. We note, however, that the star distortion
493: $\epsilon=(\mathrm{I}_3-\mathrm{I}_1)/\mathrm{I}_1=\mathrm{P_{spin}}/\mathrm{P_{prec}}\approx 4\times 10^{-8}$
494: is larger than that reported for radio-pulsars
495: \citep[e.g.][]{2001MNRAS.324..811J,2006MNRAS.365..653A}
496: but smaller than that for Her\,X-1 \citep{2000Ketsaris}.
497:
498: \begin{acknowledgements}
499: The XMM-Newton project is supported by the Bundesministerium f\"ur Bildung und
500: For\-schung / Deutsches Zentrum f\"ur Luft- und Raumfahrt (BMBF/DLR), the
501: Max-Planck-Gesellschaft and the Heidenhain-Stif\-tung. SZ acknowledges support
502: from a PPARC AF. We thank Joachim Tr\"umper, Mark Cropper and Gianluca Israel
503: for fruitful discussions.
504: \end{acknowledgements}
505:
506: \bibliographystyle{aa}
507: \bibliography{ins,general,myrefereed,myunrefereed}
508:
509:
510: \end{document}
511: