astro-ph0404582/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: 
3: %% manuscript produces a one-column, double-spaced document:
4: %% \documentclass[manuscript]{aastex}
5: 
6: %% preprint2 produces a double-column, single-spaced document:
7: %%\documentclass[preprint2]{aastex}
8: 
9: 
10: %\usepackage{emulateapj5}
11: %\usepackage{psfig}
12: 
13: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
14: 
15: 
16: \newcommand{\axp}{1E\ 1841-045}
17: \newcommand{\gtap}{\mathrel{\hbox{\rlap{\lower.55ex \hbox {$\sim$}}
18:                    \kern-.3em \raise.4ex \hbox{$>$}}}}
19: \newcommand{\ltap}{\mathrel{\hbox{\rlap{\lower.55ex \hbox {$\sim$}}
20:                    \kern-.3em \raise.4ex \hbox{$<$}}}}
21: 
22: %% You can insert a short comment on the title page using the command below.
23: \slugcomment{}
24: 
25: \shorttitle{Detection of hard pulsed X-rays from \axp}
26: \shortauthors{Kuiper et al.}
27: 
28: 
29: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
30: 
31: \begin{document}
32: 
33: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
34: 
35: \title{Discovery of hard non-thermal pulsed X-ray emission\\ from the anomalous X-ray pulsar \axp}
36: 
37: \author{L. Kuiper\altaffilmark{1}, W. Hermsen\altaffilmark{1,2} and M. Mendez\altaffilmark{1,2}}
38: %\affil{SRON-National Institute for Space Research}
39: %\affil{Sorbonnelaan 2, 3584 CA, Utrecht, The Netherlands}
40: \email{L.M.Kuiper@sron.nl}
41: \altaffiltext{1}{SRON-National Institute for Space Research, Sorbonnelaan 2, 3584 CA, Utrecht, The Netherlands }
42: \altaffiltext{2}{Astronomical Institute ``Anton Pannekoek", University of Amsterdam, Kruislaan 403, 1098 SJ Amsterdam, The Netherlands}
43: 
44: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45: 
46: \begin{abstract}
47: We report the discovery of non-thermal pulsed X-ray/soft $\gamma$-ray emission up to $\sim 150$ keV from the anomalous X-ray pulsar 
48: AXP \axp\ located near the centre of supernova remnant Kes 73 using RXTE PCA and HEXTE data. The morphology of the double-peaked 
49: pulse profile changes rapidly with energy from 2 keV up to $\sim 8$ keV, above which the pulse shape remains more or less stable. The 
50: pulsed spectrum is very hard, its shape above 10 keV can be described well by a power law with a photon index of $0.94\pm 0.16$. 
51: \axp\ is the first AXP for which such very-hard pulsed emission has been detected, which points to an origin in the magnetosphere of a 
52: magnetar. We have also derived the total emission spectrum from the Kes73/\axp\ complex for the $\sim$ 2-270 keV energy range using 
53: RXTE HEXTE and XMM-Newton PN data. A comparison of the total emission from the complex with the pulsed+DC emission from \axp\ as derived from 
54: Chandra ACIS CC-mode data \citep{morii03} leaves little room for emission from Kes 73 at energies near 7 keV or above. This suggests 
55: that the HEXTE spectrum above $\sim$ 10 keV, satisfactorily described by a power law with index $1.47 \pm 0.05$, is dominated by emission 
56: from \axp. In that case the pulsed fraction for energies above 10 keV would increase from about 25\%  near  10 keV to 100\%  near
57: 100 keV. The origin of this DC-component extending up to $\sim$ 100 keV is probably magnetospheric and could be a manifestation of
58: pulsed emission which is ``on" for all phases.
59: \end{abstract}
60: 
61: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
62: 
63: %\psfigurepath{/home/kuiper/....}
64: 
65: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
66: 
67: \keywords{pulsars: individual (\axp), X-rays: stars}
68: 
69: %% The \notetoeditor{TEXT} command allows the author to communicate
70: %% information to the copy editor.  This information will appear as a
71: %% footnote on the printed copy for the manuscript style file.  Nothing will
72: %% appear on the printed copy if the preprint or preprint2 style files are used.
73: 
74: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
75: 
76: \section{Introduction}
77: 
78: The quest whether anomalous X-ray pulsars (AXPs) and soft gamma-ray repeaters (SGRs)
79: are both manisfestations of isolated neutron stars with ultra-strong magnetic fields 
80: ($\sim$ $10^{14}-10^{15}$ G), so-called ``magnetars" \citep{thompson96}, has recently been decided.
81: Notably, the first detection of SGR-like bursts from the direction of AXP 1E 1048.1-5937 \citep{gavriil02},
82: followed by the detection of SGR-like bursts from AXP 1E 2259+586 \citep{kaspi03} conclusively 
83: unified AXPs and SGRs \citep[see also][]{gavriil04}. Similarly, \citet{kulkarni03} showed that the originally 
84: classical SGR 0526-66 was recently found to behave like an AXP. This was predicted uniquely by the 
85: magnetar model \citep{thompson96}. For recent reviews on AXPs and SGRs see \citet{mereghetti02} 
86: and \citet{woods04}.
87: 
88: In this paper we present detailed timing and spectral characteristics of AXP \axp, located
89: in the centre of the supernova remnant (SNR) Kes 73 (G27.4+0.0), which is at a kinematic distance
90: between 6 and 7.5 kpc \citep{sanbonmatsu92}. The X-ray pulsations of the source were discovered with 
91: ASCA \citep{vasisht97} and a phase-connected timing solution was published by \citet{gotthelf02}, who 
92: analyzed observations with  the Rossi X-ray Timing Explorer (RXTE) spanning 2 years. A linear ephemeris appeared to
93: be consistent with the pulse periods measured over 15 years with Ginga, ASCA, RXTE and BeppoSAX.
94: The measured constant, long-term spin-down of \axp\ as well as the phase-connected timing with
95: ``timing-noise-like" residuals supported a magnetar identification.
96: 
97: \citet{morii03} reported detailed results for  \axp\ from Chandra ACIS CC-mode observations.
98: They could for the first time discriminate the compact object from the surrounding SNR Kes 73.  Like other AXPs,
99: the phase integrated spectrum (pulsed and DC emission) was well fitted with a power law (photon index 2.0 $\pm$ 0.3)
100: plus blackbody model (kT$_{\hbox{\tiny \rm BB}}$= 0.44 $\pm$ 0.02 keV). The photon index is the flattest among AXPs. 
101: It should be noted, however, that a two-blackbody fit rendered a similarly good fit. They also reported
102: that the pulse profile is double-peaked, with the second pulse exhibiting a harder spectrum between 3 and 7 keV.
103: 
104: \citet{molkov04} published recently a source catalog of 28 sources detected by INTEGRAL/IBIS between 18 and 60 keV
105: in a survey of the Sagittarius Arm tangent region. The source in this catalog with the hardest spectrum was identified with  \axp\ 
106: and SNR Kes 73. Since AXPs are ``known" to be soft-spectrum sources, the SNR seemed at first glance the most likely
107: counterpart. This identification stimulated us to analyse archival RXTE PCA and HEXTE data to search for
108: a pulsed signature from \axp, particularly in the hard X-ray range covered by HEXTE. We also analyzed
109: archival XMM-Newton data in an attempt to unravel the contributions from Kes 73 and the pulsed and unpulsed emissions
110: from \axp, combining the informations from our RXTE PCA/HEXTE and XMM-Newton analyses, and the Chandra results 
111: on \axp\ published by \citet{morii03} .
112: 
113: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
114: 
115: \section{Instruments and observations}
116: In this study the results come mainly from the analysis of data from two of the three X-ray instruments aboard RXTE , the Proportional Counter Array 
117: (PCA; 2-60 keV) and the High Energy X-ray Timing Experiment (HEXTE; 15-250 keV). Both are non-imaging instruments.
118: The PCA \citep{jahoda96} consists of five collimated xenon proportional counter units (PCUs) with a total effective area of $\sim 6500$ cm$^2$
119: over a $\sim 1\degr$ (FWHM) field of view. Each PCU has a front Propane anti-coincidence layer and three Xenon layers which provide the
120: basic scientific data, and is sensitive to photons with energies in the range 2-60 keV. The energy resolution is about 18\% at 6 keV.
121: 
122: The HEXTE instrument \citep{rothschild98} consists of two independent detectors clusters, each containing four Na(Tl)/CsI(Na) scintillation
123: detectors. The HEXTE detectors are mechanically collimated to a $\sim 1\degr$ (FWHM) field of view and cover the 15-250 keV energy range
124: with an energy resolution of $\sim$ 15\% at 60 keV. The collecting area is 1400 cm$^2$ taking into account the loss of the spectral capabilities of
125: one of the detectors. The maximum time resolution of the tagged events is $7.6\mu$s. In its default operation mode the field of view of each
126: cluster is switched on and off source to provide instantaneous background measurements.
127: Due to the co-alignment of HEXTE and the PCA, they simultaneously observe the sources in their field of view. Table \ref{obs_table} lists
128: the publicly available RXTE observations used in this study\footnote{\axp\ is regularly monitored
129: by RXTE since February 1999}. All observations listed in Table \ref{obs_table} have been carried out with the instrumental pointing axis within 
130: $\sim 0\farcm 5$ from \axp. In the fourth column the PCU unit 2 screened exposure is given (see Sect. \ref{pca_timing}). A typical observation
131: run consists of several sub-observations spaced more or less uniformly between the start and end date of the observation.
132: In this study the total number of sub-observations amounts 45.
133: 
134: \begin{table}[t]
135: \caption{List of  RXTE observations used in this study of \axp.\label{obs_table}}
136: \begin{center}
137: \begin{tabular}{cccr}
138: \hline\hline
139: \textbf{Obs.} & \multicolumn{2}{c}{\textbf{Begin/End Date}}    & \textbf{Exp.$^{\dagger}$}\\
140: \textbf{id.}  & \multicolumn{2}{c}{\textbf{(dd/mm/yyyy)}}      & \textbf{(ks)}\\
141: \hline
142: 40083            & 15-02-1999          & 23-02-2000           & 105.73 \\
143: 50082            & 10-04-2000          & 08-03-2001           &  59.91 \\
144: 60069            & 02-04-2001          & 26-01-2002           &  52.24 \\
145: 70094            & 16-03-2002          & 27-01-2003           &  53.26 \\
146: \hline
147: \multicolumn{4}{l}{$^{\dagger}$PCU-2 exposure after screening} \\
148: \end{tabular}
149: \end{center}
150: \end{table}
151: 
152: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
153: 
154: \section{Timing analysis}
155: 
156: \subsection{Timing analysis of PCA data}
157: \label{pca_timing}
158: The PCA data from the observations listed in Table \ref{obs_table} have all been collected in {\em Goodxenon} or {\em GoodxenonwithPropane}
159:  mode allowing high-time-resolution ($0.9\mu$s) studies in 256 spectral channels. Because we are mainly interested in the medium/hard X-ray timing
160:  properties of \axp\ we ignored the events triggered in the Propane layers of each PCU. Furthermore, we used only data from the top xenon layers of each
161:  PCU in order to improve the signal-to-noise ratio. Because the number of active PCUs at any time was changing we treated the five PCUs
162:  constituting the PCA separately. For each PCU good time intervals have been determined by including only time periods when the PCU in question is on 
163:  and during which the pointing direction is within $0\fdg 05$ from the target, the elevation angle above Earth's horizon is greater than $5\degr$, a time delay 
164:  of 30 minutes since the peak of a South-Atlantic-Anomaly passage holds, and a low background level from contaminating electrons is observed. 
165:  These good time intervals have subsequently been applied in the screening process to the data streams from each of the PCUs (e.g. see Table 
166:  \ref{obs_table} for the resulting screened exposure of PCU-2 per observation run).
167: 
168:  \begin{figure}[t]
169:   \epsscale{0.5}
170:   \plotone{f1.ps}
171:   \caption{\label{pca_pulse_template}RXTE PCA pulse profile of \axp\ for energies in the range 3.9-7.8 keV combining data collected between
172:   February 1999 and February 2003 (see Table \ref{obs_table}). Two cycles are shown for clarity. The best fitting truncated Fourier series
173:   (3 harmonics) is superposed as a dashed line.}
174: \end{figure}
175: 
176: Next, for each sub-observation the arrival times of the selected events (for each PCU unit) have been converted to arrival times at the solar
177:  system barycenter (in TDB time scale) using the instantaneous spacecraft position and celestial position of \axp\  \citep[see][]{gotthelf02}.
178:  These barycentered arrival times have been folded with the phase connected timing solution given in \citet{gotthelf02} using only the first
179:  three frequency coefficients to obtain pulse phase distributions for selected energy windows. Combining now the phase distributions from the various PCUs
180:  the well-known 2-10 keV profile of \axp\  \citep[see][]{gotthelf02} could be recognized in each sub-observation, however, phase shifts between 
181:  the sub-observations made a direct combination impossible.
182:  Therefore we correlated the pulse phase distribution of each sub-observation with a chosen initial template and applied the measured phase shifts
183:  to obtain an aligned combination with much higher statistics. The correlation is then once repeated with, instead of the initially chosen template, the aligned
184:  combination from the first correlation to obtain the final summed profile \citep[see e.g.][for a similar iterative method applied for PSR B0540-69]{deplaa03}.
185:  The summed profile for energies between 3.9-7.8 keV deviates from uniformity at a $\sim 32\sigma$ level and is shown in Fig. \ref{pca_pulse_template}. 
186:  Its morphology mimics the profile shown by \citet{gotthelf02}. For the first time, however, pulsed emission has been detected up to $\sim 25$ keV -
187:  the deviation from uniformity is $9.5\sigma$ and $3.2\sigma$ in the 11.7-16.1 keV and 16.1-23.8 keV energy ranges, respectively - providing
188:  strong evidence for a non-thermal origin of the pulsed high-energy emission.
189:  This result motivated us to search also the higher-energy HEXTE data for a pulsar signal.
190: 
191: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
192: \begin{figure}[t]
193:   \epsscale{0.5}
194:   \plotone{f2.ps}
195:   \caption{\label{hexte_pulse_prof_gt50} RXTE HEXTE pulse profile of \axp\ for energies greater than 50.3 keV demonstrating the existence
196:   of pulsed emission from this source at hard X-rays/soft $\gamma$-rays. The deviation from uniformity is about $4.1\sigma$. The dashed line is the
197:   best fitting truncated Fourier series (3 harmonics) like shown in Fig. 1. The structured profile and the minimum level 
198:   approximately coincide in phase with those in the 3.9-7.8 keV profile of Fig. \ref{pca_pulse_template}}.
199: \end{figure}
200: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
201: \begin{figure*}[t]
202:   \centerline{\includegraphics[height=16cm,width=10cm,angle=90]{f3.ps}}
203:   \caption{\label{profile_collage}A RXTE PCA/HEXTE collage of pulse profiles of \axp\ for energies greater than 2.1 keV. Two cycles are shown
204:   for each panel. Panels {\em a-e}: RXTE PCA pulse profiles for energies between 2.1 and 28.2 keV. Panels {\em f-h}: RXTE HEXTE pulse
205:   profiles for the energy ranges 10.3-26 keV, 35.2-102 keV and 102-267.5 keV. The three HEXTE profiles deviate from uniformity at a 4.2, 3.1
206:   and 2.7$\sigma$ level, respectively. Notice the morphology change of the pulse profiles, particularly in the PCA profiles.}
207: \end{figure*}
208: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
209: 
210: \subsection{Timing analysis of HEXTE data}
211: 
212: HEXTE operated in its default rocking mode during the observations listed in Table \ref{obs_table} allowing the 
213: collection of real-time background data
214: from two independent positions $\pm 1\fdg 5$ to either side of the on-source position. For the timing analysis 
215: we selected only the on-source data.
216: Good time intervals have been determined using similar screening filters as used in the case of the PCA. The 
217: selected on-source HEXTE event times
218: have subsequently been barycentered and folded according to the ephemeris given in \citet{gotthelf02} using 
219: again only the first three frequency coefficients.
220: Applying the phase shifts as derived from the contemporaneous PCA measurements to the HEXTE phase distributions of each sub-observation 
221: we could obtain the HEXTE pulse phase distributions in 256 spectral channels for the combination of observations listed in Table \ref{obs_table}. 
222: The HEXTE integral ($>10.3$ keV) pulse profile deviates from uniformity at a $\sim 4.7\sigma$ level 
223: and shows strong similarities with the PCA profile for energies above $\sim 10$ keV. Most remarkably,
224: the pulse profile for energies greater than 50.3 keV reaches a significance of $\sim 4.1\sigma$, 
225: proving the existence of pulsed emission from \axp\ at 
226: hard X-rays/ soft $\gamma$-rays (see Fig. \ref{hexte_pulse_prof_gt50}).
227: 
228: \subsection{Energy dependence of the pulse profile morphology}
229: 
230: We have investigated the morphology of the pulse phase distributions as a function of
231: energy using both the RXTE PCA and HEXTE data. This kind of information could provide clues to the
232: origin and physics of the processes responsible for the pulsed high-energy radiation. 
233: In Fig. \ref{profile_collage} a collage 
234: of pulse-phase distributions is shown for different energy bands between $\sim 2$ and $\sim 270$ keV
235: \footnote{For HEXTE the 26-35.2 energy window is ignored in this collage because of a huge background line
236: near 30 keV due to the activation of iodine.}. 
237: A striking feature is that pulse minimum occurs in the phase range $[0.7,0.8]$ irrespective of the energy band.
238: Furthermore, the structured broad profile, measured with high statistics in the PCA data, seems to behave
239: as two components (``pulses") separated $\sim 0.4$ in phase, with the relative contributions varying with energy. 
240: The first ``pulse" near phase 0.1 is dominant over the second ``pulse" near phase 0.5 for energies
241: below $\sim 9$ keV. Above this energy the second pulse, however, has a similar intensity level as the first pulse. 
242: Quantitatively this behaviour is shown in Fig. \ref{p1_p2_ratio}, where the ratio of the number of excess counts in 
243: phase range 0.8-1.25 (P1) and 0.25-0.7 (P2) for RXTE PCA data is plotted as a function of energy (phase range 0.7-0.8 is chosen 
244: as a DC-reference window).
245: \begin{figure}[h]
246:   \epsscale{0.45}
247:   \plotone{f4.ps}
248:   \caption{\label{p1_p2_ratio} The ratio of excess counts in phase range 0.8-1.25 (P1) and 0.25-0.7 (P2) for RXTE PCA
249:   data as a function of energy. The dotted line represents the best fit to a function of form $a+b/E$ with $(a,b)$ parameters to be 
250:   optimized. The improvement relative to a constant function $a$ is about $3\sigma$.}
251: \end{figure}
252: The $\chi^2$ of a fit to a function parameterized as $a+b/E$ improves by about $3\sigma$ (1 degree of freedom) relative
253: to the $\chi^2$ of a fit assuming a constant, i.e. no energy dependence of the ratio. Thus, we conclude that the morphology of the 
254: high-energy ($> 2$ keV) pulse profiles of \axp\ changes with energy.
255: 
256: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
257: \begin{figure}[t]
258:   \centerline{\includegraphics[height=9cm,width=11cm,angle=0]{f5.ps}}
259:   \caption{\label{high_energy_spc}A $\nu F_{\nu}$ spectral representation of the total pulsed high-energy emission
260:   from \axp\ is shown in red (RXTE PCA; open circles, RXTE HEXTE; open squares). The total spectrum from 
261:   the Kes73/\axp\ complex is represented in blue (triangles; XMM EPIC PN, filled squares; RXTE HEXTE). The total 
262:   (pulsed plus DC) 1-7 keV X-ray spectrum from \axp\ \citep{morii03} is plotted as a solid dark orange line. The two magenta 
263:   flux points are INTEGRAL IBIS ISGRI measurements given in \citet{molkov04}. Fits ($>10$ keV) to the total complex (blue) and 
264:   total pulsed (red) spectra of \axp\ are drawn as dashed lines.}
265: \end{figure}
266: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
267: 
268: \section{Spectral analysis}
269: 
270: In this section high-energy ($\sim 2-300$ keV) spectra are presented 
271: for: 1) the total pulsed emission from \axp\ based on RXTE PCA and HEXTE data, and 2) 
272: the total high-energy emission from the Kes 73 supernova remnant plus 
273: the embedded AXP \axp\ using RXTE HEXTE on/off data and XMM Newton PN data. 
274: 
275: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
276: 
277: \subsection{Pulsed X-ray spectrum of \axp}
278: The number of pulsed counts in the differential PCA/HEXTE energy bands has been determined by fitting a truncated Fourier series, 
279: $a_0+\sum_{k=1}^{N} a_k \cos(2\pi k \phi) + b_k \sin(2\pi k \phi) $, with $\phi$ the pulse phase, to the measured pulse phase distributions. It turned out that 3 
280: harmonics ($N=3$) were sufficient to describe the measured distributions accurately for all energy intervals. For example, in Figs. \ref{pca_pulse_template} and 
281: \ref{hexte_pulse_prof_gt50} the best fit truncated Fourier series are shown as dashed lines for the PCA 3.9-7.8 keV and HEXTE $> 50.3$ keV pulse phase 
282: distributions, respectively.
283: 
284: In case of the PCA we derived for each PCU the energy response matrix (energy redistribution including the  sensitive area) for the combination of observations
285: listed in Table  \ref{obs_table} and subsequently took the different PCU (screened) exposure times into account in the construction of the weighted PCU-combined 
286: energy response. The pulsed excess counts per energy band are fitted in a procedure assuming a power-law  photon spectrum folded through the PCU-combined 
287: energy response. The best fit power-law model to the PCA pulsed data  has a photon index of $1.93\pm 0.01$. The derived 2-10 keV total pulsed 
288: flux\footnote{Spectral cross-calibration between the BeppoSAX MECS and RXTE PCA instruments indicate that the normalization of the PCA spectra is about 
289: 20\% higher than that of the MECS. The PCA fluxes are therefore multiplied by 0.81 \citep[see e.g.] [about this subject]{kuiper04}} is 
290: ($5.20\pm 0.14)\cdot 10^{-12}$ erg/cm$^2$s, which is consistent with the 2-10 keV pulsed 
291: flux of ($5.70\pm 0.70)\cdot 10^{-12}$ erg/cm$^2$s given in \citet{gotthelf02} who used a smaller dataset and a different extraction method (on/off).
292: 
293: In the spectral ``deconvolution" process of the HEXTE total pulsed counts the on-axis\footnote{It was not necessary to derive off-source HEXTE response matrices 
294: taking into account the reduction in the sensitive area due to the collimator response because all observations were performed nearly on-axis.} cluster A and B 
295: energy response matrices have been employed taking into account the (slightly) different screened on-source exposure times of each cluster. The exposure 
296: times have been corrected for the considerable deadtime effects. 
297: 
298: The total pulsed PCA and HEXTE flux measurements are shown  in Fig. \ref{high_energy_spc} as red open circles and squares, respectively. In this $\nu F_{\nu}$
299: spectral representation the hardening of the pulsed flux near 10 keV is striking. A power-law model fitted  to the PCA/HEXTE total pulsed flux measurements for energies above 
300: 10 keV yielded a photon index of $0.94\pm 0.16$. Such a spectral hardening has not been observed yet for any of the AXPs currently known. 
301: 
302: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
303: 
304: \subsection{Total X-ray spectrum of Kes 73 plus \axp}
305: For HEXTE a model independent way for deriving the total flux from the Kes 73/\axp\ complex exists based on the source on/off observations.
306: The off-source observations make it possible to obtain instantaneous background estimates. Subtracting the properly scaled background 
307: contribution (every 32 s on-source observation is accompanied by a 28 s off-source observation; 4 s is consumed in the slew process) from 
308: the on-source measurements yielded the total excess counts from the Kes 73/\axp\ complex for energies above $\gtap 10$ keV.
309: These excess counts have been converted to flux values using similar procedures as discussed before and the resulting flux measurements
310: are shown as blue filled squares in Fig. \ref{high_energy_spc}. A power-law fit through these points yielded a photon index of $1.47\pm 0.05$,
311: slightly softer than the pulsed emission for energies above 10 keV.
312: 
313: In order to extend the total emission spectrum towards lower energies we analysed archival XMM Newton data from an observation 
314: targetted at \axp. We used data taken on October 5, 2002, starting at 03:16:47 UTC. We processed the data using the latest version 
315: of the XMM-Newton analysis software, SAS 6.0.0. For the analysis, we first produced light curves in different energy bands
316: of a region in the sky far from the supernova remnant to check for solar flares that might have affected the observation. Since no flares
317: were detected in the light curves, we used the total length of the observation, $\sim 4000$ s of on source time, for the rest of the
318: analysis. We produced images and spectra both for the EPIC PN and the two EPIC MOS instruments, and checked that all instruments 
319: yielded consistent results. Since the EPIC PN has the largest overall effective area and is sensitive to higher energies than MOS, we 
320: only concentrated on this instrument for the rest of our analysis.
321: 
322: We extracted a spectrum of a circular region of 2 arcmin radius centered on \axp\ containing both the AXP and the supernova
323: remnant. We checked that the spectrum is not affected by pile up\footnote{Pile up occurs when two or more photons hit the same CCD
324: pixel within a read-out cycle. In that case the instruments count those photons as single event with an energy approximately equal to the sum
325: of the energies of the individual photons. Pile up reduces the apparent flux of the source and at the same time makes the spectrum appear
326: harder than it really is.}. To produce a background spectrum we selected an area near the supernova remnant, but sufficiently far from it to
327: avoid contamination from the supernova itself.
328: Next we produced an energy  response matrix and an ancilliary response file, the latter properly weighted using the surface brightness
329: distribution of the extended emission of the source over the detector. We rebinned the spectrum such that we had about 3 bins per 
330: resolution element. Since the source is relatively bright, with this rebinning we obtained at least 25 counts per bin in the range 0.3 -- 7 keV, 
331: and at least 16 counts per bin up to 11 keV.
332: The ``deconvolved'' spectrum ($\nu F_{\nu}$) is shown in Fig. \ref{high_energy_spc}:  it displays a relatively bright continuum, on top of which 
333: several emission lines of Mg, Ar, S, and Si coming from the supernova remnant can be seen.
334: 
335: 
336: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
337: 
338: \section{Discussion}
339: 
340: We have presented the surprising discovery of very hard, photon index 0.94 $\pm$ 0.16, 
341: pulsed hard X-ray emission extending up to energies of $\sim$ 150 keV from the AXP \axp,
342: obtained using archival monitoring observations by the PCA and HEXTE aboard RXTE. Such hard, non-thermal
343: emission can only originate within the magnetosphere of a neutron star/magnetar. In fact, the total pulsed X-ray spectrum
344: of \axp\ is very reminiscent to that of young spin-down powered pulsars. Particularly the spectrum of the Vela pulsar
345: with a blackbody spectrum at lower X-ray energies and a hard power-law extension to higher energies looks very similar.
346: These hard power-law  spectral extensions with photon indices  of $\sim 1$ are also found for the young (radio weak/quiet) 
347: spin-down powered  pulsars PSR B1509-58, PSR B1846-0258, PSR J1811-1925, PSR J1930+1852, all associated with SNRs
348: , like  \axp, and the old millisecond pulsars PSR J0218+4232, PSR B1821-24 and PSR B1937+21.
349: 
350: Such hard X-ray emission from AXPs has hardly been considered in the modelling of AXPs. However, \citet{cheng01}
351: studied the production of high-energy gamma-ray radiation in the outer magnetospheres of AXPs. They argue that due to the
352: strong field of a magnetar, the gamma-ray emission rooted at the polar caps will be quenched. However, far away from the 
353: pulsar surface, i.e. in the outer gaps, the gamma-rays could be emitted because the local magnetic field will drop 
354: below the critical quantum limit. The production process would be curvature radiation by the acceleration of electrons/positrons, 
355: resulting from the collisions between high-energy photons from the outer gap and the soft X-rays originating from the stellar surface. 
356: They applied their model calculations also to the case of \axp, and predict an integral gamma-ray photon flux for energies 
357: above 100 MeV of $\sim$ 8.7 $\times$ 10$^{-9}$ (d/6 kpc)$^{-2}$ cm$^{-2}$ s$^{-1}$, which is below the EGRET sensitivity 
358: but above the sensitivity of GLAST. The very-hard pulsed spectrum we measured with HEXTE might be consistent with this interpretation.
359: 
360: We studied the energy dependence of the pulse profile morphology and confirm the findings by \citet{morii03}, namely
361: the pulse profile is double-peaked, and the second pulse has a harder spectrum for energies below $\sim$ 7 keV. However,
362: this trend does not continue above $\sim 10$ keV. Within the statistics, both pulses remain visible at about the same
363: strengths up to $\sim$ 150 keV. Significant variations with energy of pulse profile morphologies below $\sim 8$ keV 
364: have been reported earlier for three other AXPs (4U 0142+61, 1RXS 1708-4009 and 1E 2259.1+586) by \citet{gavriil02a}.
365: 
366: The set of spectra shown in Fig. \ref{high_energy_spc} summarizes our present knowledge on the contributions from
367: the different components constituting the total emission from the direction of Kes 73 and AXP \axp, i.e.
368: the pulsed and DC components of \axp, and the emission from SNR Kes 73.
369: In addition to the total pulsed spectrum (PCA/HEXTE) and the pulsed+DC+Kes-73 spectrum (XMM EPIC PN/HEXTE) derived in this work,
370: we show the INTEGRAL IBIS ISGRI flux values for energies 18--120 keV \citep{molkov04}, and the Chandra ACIS spectrum \citep{morii03}
371: for the pulsed+DC spectrum of \axp. We note that the INTEGRAL flux points are consistent with our total HEXTE spectrum
372: (one should realize that these preliminary INTEGRAL IBIS ISGRI flux values have still systematic uncertainties of $\sim$ 10\%).
373: 
374: For energies up to about 7 keV we have information on the relative contributions:  
375: 1) the pulsed fraction of \axp\ is $\sim$ 25\%, 2) there is very little 
376:  margin for emission from SNR Kes 73 around 7 keV. This suggests that the HEXTE spectrum above 10 keV is dominated
377:  by emission from \axp\ (pulsed+DC), and above $\sim 100$  keV the spectrum is totally due to pulsed emission.
378:  This would mean that also the DC emission from \axp\ extends up to $\sim 100$  keV!
379:  An alternative interpretation is that the relative contribution from Kes 73 increases again above 10 keV due to a new hard component.
380:  Although the latter seems unlikely, higher spatial-resolution measurements at energies above 10 keV would be required
381:  to exclude this possibility.
382: 
383:  If we assume that the contribution from Kes 73 is small or negligible above 10 keV, the challenge is now to explain
384:  the hard DC emission from \axp. Most plausible might be an origin in the magnetosphere, namely for
385:  the situation that the pulsed emission is ``on" for all phases. In the competing polar cap \citep[e.g. as discussed for AXPs by][] 
386:  {zhang00} and outer gap \citep[see e.g.][]{cheng01} scenarios explaining the high-energy emission from spin-down powered pulsars 
387:  there are geometries  for which this is indeed possible. Allowing for a small contribution from Kes 73, e.g. up to about 30 keV, and 
388:  knowing  that pulsar spectra depend on pulse phase, detailed model calculations are required to investigate this interpretation further.
389: 
390: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
391: 
392: \acknowledgments
393: This research has made use of data obtained from the High Energy Astrophysics Science Archive Research Center (HEASARC), 
394: provided by NASA's Goddard Space Flight Center.
395: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
396: \begin{thebibliography}{}
397: 
398: \bibitem[Cheng \& Zhang(2001)]{cheng01} Cheng, K.S. and Zhang, L. 2001, \apj, 562, 918
399: \bibitem[Gavriil et al.(2002a)]{gavriil02a} Gavriil, F.P. and Kaspi, V.M. 2002, \apj, 567, 1067
400: \bibitem[Gavriil et al.(2002b)]{gavriil02} Gavriil, F.P., Kaspi, V.M., Woods, P.M. 2002, \nat, 419, 142
401: \bibitem[Gavriil et al.(2004)]{gavriil04} Gavriil, F.P., Kaspi, V.M., Woods, P.M. 2004, AdSpR., 33, 654
402: \bibitem[Gotthelf et al.(2002)]{gotthelf02} Gotthelf, E.V., Gavriil, F.P., Kaspi, V.M., et al. 2002, \apjl, 564, L31
403: \bibitem[Jahoda et al.(1996)]{jahoda96} Jahoda, K., Swank, J.H., Giles, A.B., et al., 1996, Proc. SPIE, 2808, 59
404: \bibitem[Kaspi et al.(2003)]{kaspi03} Kaspi V.M., Gavriil, F.P., Woods, P.M., et al. 2003, \apjl, 588, 93
405: \bibitem[Kuiper et al.(2004)]{kuiper04} Kuiper, L., Hermsen, W., Stappers, B.  2004, AdSpR., 33, 507
406: \bibitem[Kulkarni et al.(2003)]{kulkarni03} Kulkarni, S.R., Kaplan, D.L., Marshall, H.L., et al. 2003, \apj, 585, 948
407: \bibitem[Mereghetti et al.(2002)]{mereghetti02} Mereghetti, S., Chialone, L., Israel, G.L., et al. 2002, Proc. 270 WE-Heraeus Seminar, 29
408: \bibitem[Morii et al.(2003)]{morii03} Morii, M., Sato, R., Kataoka, J., et al. 2003, \pasp, 55, L45
409: \bibitem[Molkov et al.(2004)]{molkov04} Molkov, S.V., Cherepashchuk, A.M., Lutovinov, A.A., et al. 2004, preprint (astro-ph/0402416v1)
410: \bibitem[de Plaa et al.(2003)]{deplaa03} de Plaa, J., Kuiper, L., Hermsen, W. 2003, \aap, 400, 1013
411: \bibitem[Rothschild et al.(1998)]{rothschild98} Rothschild, R.E., Blanco, P.R., Gruber, D.E., et al. 1998, \apj, 496, 538
412: \bibitem[Sanbonmatsu \& Helfand(1992)]{sanbonmatsu92} Sanbonmatsu, K.Y. and Helfand, D.J. 1992, \aj, 104, 2189
413: \bibitem[Thompson \& Duncan(1996)]{thompson96} Thompson, C. and Duncan R.C. 1996, \apj, 473, 322
414: \bibitem[Vasisht \& Gotthelf(1997)]{vasisht97} Vasisht, G. and Gotthelf, E.V. 1997, \apjl, 486, 129
415: \bibitem[Woods(2004)]{woods04} Woods, P.M. 2004, AdSpR., 33, 630
416: \bibitem[Zhang \& Harding(2000)]{zhang00} Zhang, B. and Harding A.K. 2000, \apj, 535, L51
417: \end{thebibliography}
418: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
419: 
420: 
421: 
422: \clearpage 
423: \end{document}
424: