1: \documentclass[12pt,preprint]{aastex}

2: \usepackage{graphicx,lscape}

3: \def\ea{\it et al.}

4: \def\exo{{\sl EXOSAT}}

5: \def\ein{{\sl Einstein}}

6: \def\uhu{{\sl Uhuru}}

7: \def\ros{{\sl ROSAT}}

8: \def\cha{{\sl Chandra}}

9: \def\xmm{XMM-{\sl{Newton}}}

10: \def\psr{{PSR~B0628-28}}

11: \def\ergsec{\hbox{erg s$^{-1}$ }}

12: \def\ergcm{\hbox{erg cm$^{-2}$ s$^{-1}$ }}

13: \def\kevcs{\hbox{keV cm$^{-2}$ s$^{-1}$ }}

14: \def\lx{$L_{\mbox{x}}$}

15: 

16: 

17: \shorttitle{X-ray emission from \psr}

18: \shortauthors{Tepedelenlio\v{g}lu \& \"{O}gelman}

19: 

20: \begin{document}

21: \title{\cha\ and \xmm\ observations of the exceptional pulsar B0628-28}

22: 

23: \author{E. Tepedelenlio\v{g}lu\altaffilmark{1}

24: \email{emre@cow.physics.wisc.edu}

25: \and

26: H. \"{O}gelman\altaffilmark{1,2}}

27: \email{ogelman@cow.physics.wisc.edu}

28: \altaffiltext{1}{Department of Physics, University of

29: Wisconsin-Madison, 1150 University Ave., Madison, WI 53703, USA}

30: \altaffiltext{2}{Faculty of Engineering and Natural Sciences, Sabanci

31: University, Orhanli Tuzla, Istanbul 34956, Turkey}

32: 

33: 

34: \begin{abstract} 

35: \psr\ is a radio pulsar which was first detected in the X-ray band by

36: \ros\ and then later observed with \cha\ and \xmm. The \cha\

37: observation yielded an X-ray luminosity two orders of magnitude higher

38: than what is expected for spin-powered pulsars, also there were no

39: pulsations detected. The \xmm\ observation, however, reveals

40: pulsations at the expected radio period, $P=1.244$ s. The

41: simultaneously analyzed spectra also give a luminosity (in cgs)

42: $\log{L_{\mbox{x}}}=30.34$, which is $\sim 350$ times greater than

43: what would be expected from the correlation between

44: $L_{\mbox{x}}$-$\dot{E}$. 

45: \end{abstract}

46: 

47: \keywords{pulsars:individual (\objectname{PSR

48: B0628-28}) --- stars:neutron --- X-rays:stars}

49: 

50: \section{Introduction}\label{intro}

51: \psr\ is an old pulsar that has a characteristic age

52: ($\tau=P/2\dot{P}$) of 2.8 Myr. Since older pulsars have radiated away

53: their initial heat content and have relatively low rotational energy

54: loss ($\dot{E}=I\Omega\dot{\Omega}$), one would expect to detect X-ray

55: radiation coming either from the reheating of the surface

56: \citep[e.g. see][]{tsuruta,shibazaki} or by heated polar caps

57: \citep{halpern}. In young pulsars, however, thermal radiation from the

58: neutron star surface is dominated by the nonthermal component from the

59: neutron star magnetosphere, whose spectrum can be described primarily

60: by a power-law model. We, also, know from X-ray observations that the

61: pulsar's nonthermal luminosity (\lx) shows a correlation with the

62: spin-down power $\dot{E}$. For instance, \citet{possenti} found a best

63: fit based on 37 pulsars. Although the variance is large, pulsars

64: follow a general trend that can be formulated as

65: $\log{L_{\mbox{x}}^{2-10}}=1.34\log{\dot{E}}-15.34$. This set included

66: only three old ($\tau > 10^{6}$~years) pulsars. The reason for having

67: only three is two fold. The first one is that older pulsars are less

68: active and relatively cooler, hence most of them were not detected

69: until \cha\ and \xmm. The detected ones have low countrates,

70: consequently statistically poor spectra. This makes it challenging to

71: distinguish between spectral models. The other reason is that it is

72: expected that surface emission from old neutron stars will be

73: dominant, resulting in spectra to be more like a blackbody. Hence, the

74: X-ray radiation from these pulsars are not expected to be

75: characterized by a power-law model. However, recent observation of

76: PSR~B0943+10 \citep{zhang}, PSR~B0823+26, PSR~B0950+08, PSR~J2043+2740

77: with \xmm\ \citep{becker} and of PSR~B2224+65 with \cha\

78: \citep{zavlin} has unfolded a new perspective in understanding old

79: neutron stars. Not only many previously undetected sources were

80: detected but also it has been possible to distinguish between

81: different spectral models for some of these old pulsars. These old

82: pulsars have spectra harder than expected and is best described by a

83: power-law and/or a polar cap (black-body) model, as opposed to a

84: black-body model representing the thermal emission from the whole

85: neutron star surface. The integrated model flux of these sources

86: convert to unusually high X-ray efficiencies ($L_{\mbox{x}}/\dot{E}$).

87: 

88: \psr\ is the longest period radio pulsar detected in X-rays. However,

89: physical parameters inferred from radio observations are not in any

90: means extreme. The \cha\ data yielded a luminosity two orders of

91: magnitude greater than what is expected for a spin-powered pulsar

92: \citep{ogelman}. The varying efficiency of converting spin-power in to

93: X-ray luminosity in pulsars can be explained by geometrical

94: effects. But, non of these can amount to such a great excess flux in

95: the X-ray band. In this paper we try to address the reasons for this

96: extreme luminosity. We describe the \xmm\ and \cha\ data in

97: $\S$~\ref{obs} and present the result of our timing and spectral

98: analysis in $\S$~\ref{res}. We discuss the implications of the

99: analysis results in $\S$~\ref{dis}.

100: 

101: \section{\cha\ and \xmm\ observations of \psr}\label{obs}

102: \subsection{\cha} 

103: \psr\ was observed twice, first on 2001 November 04 and again on 2002

104: March 25 for 2000 s and 17000 s, respectively. For both observations

105: photons were collected using the Advanced CCD Imaging Spectrometer

106: (ACIS). Data were collected in the nominal timing mode, with 1.141 s

107: exposures between CCD readouts. We reprocessed the Level 1

108: event data to correct the detrimental effects of charge transfer

109: efficiency.  The imaging, timing and spectral analysis presented here

110: were done only on the 17 ks observation, the 2.0 ks observation was

111: disregarded due to an interruption by a large solar storm. The

112: background countrate during this solar storm increased by a factor of

113: $\sim$20. The net source countrate during the first observation period

114: was $0.011\pm 0.001$ counts per second (cps) where the error is in the

115: 68\% confidence range. Since by taking into account the 2.0 ks

116: observation we gain only $\sim$27 counts, which is not enough to

117: improve our statistics significantly, we preferred to disregard these

118: counts, which potentially can be misleading.

119: 

120: The measured PSF of \psr\ is consistent with the ACIS point-source

121: response, hence the ACIS image reveals a point-like X-ray source at the

122: pulsar position. The Chandra position of \psr\ is

123: $\alpha=06^{\mbox{h}}30^{\mbox{m}}49\farcs43$,

124: $\delta=-28^{\circ}34^{\prime}43\farcs60$ (J2000.0), which considering

125: 0\farcs5 rms error and the $\sim$0\farcs6 ~absolute astrometric

126: accuracy of \cha, is in good agreement with the radio position

127: $\alpha$=06$^{\mbox{h}}$30$^{\mbox{m}}$49\farcs53, $\delta=-28^{\circ}34

128: \arcmin43\farcs60$ (J2000.0), which was taken from the ATNF pulsar

129: catalogue{\footnote{http://www.atnf.csiro.au/research/pulsar/psrcat}}.

130: 

131: \subsection{\xmm}

132: 

133: \psr\ was observed with \xmm\ on 2004 February 28 for a total ontime

134: of 48 ks. MOS1/2 were both operated in imaging (PrimeFullWindow) mode

135: and the medium filter was used. During the EPIC-pn exposure the thin

136: filter was used and the detector was operated in imaging

137: (PrimeLargeWindow) mode for 47 ks. The temporal resolution achieved

138: with this choice of science modes were 2.6 s and 43 ms for MOS1/2 and

139: pn, respectively.

140: 

141: The background of EPIC camera is known to be effected by soft proton

142: flares. In order to screen for times of high background we inspected

143: the light curves of MOS1/2 and pn data separately at energies above 10

144: keV. Both sets of data were effected by high background. The light

145: curves were formed with 100 s bins and after inspection we rejected

146: bins with countrate greater than 0.4 cps. The removal of high

147: background time intervals from the data leaves us with effective

148: exposure times of 42.5 ks and 33.3 ks for MOS1/2 and pn,

149: respectively. The pulsar is clearly detected in the EPIC image at

150: $\alpha=06^{\mbox{h}}30^{\mbox{m}}49\farcs48$, $\delta=-28^{\circ}

151: 34^{\prime}43\farcs10$ (J2000.0) which differs from the radio position

152: by only 0\farcs8 well within the 2\arcsec-3\arcsec\ uncertainty of the

153: EPIC absolute astrometry. The shape of the radial profile of the

154: source is also consistent with that expected for a point-like source.

155: 

156: \section{Results}\label{res}

157: \subsection{Timing}

158: For searching pulsations from \psr\ \cha\ ACIS and \xmm\ EPIC-MOS data

159: were not suitable due to their limited temporal resolution. The

160: sampling frequency in both cases set by the detector read out rate give

161: a Nyquist frequency greater than the pulsar frequency (see

162: Table~\ref{tab1}). Thus we used EPIC-pn data that has 43 ms timing

163: resolution.

164: 

165: We extracted source plus background photons from a 30\arcsec\ radius

166: circle centered at the pulsar position, which encircles about 85\% of

167: all detected source counts. The extraction region contained 1047

168: counts of which 16\% is background. The photon arrival times were

169: solar system barycenter corrected. The pulsar's spin parameters

170: (Table~\ref{tab1}) are well known from radio observations and can be

171: extrapolated to the mean epoch of the \xmm\ observation:

172: MJD=53063.339. Around this predicted pulsar frequency, we then

173: generated a periodogram using the $Z_{1}^{2}$-statistic. The X-ray

174: periodogram is shown in Figure~\ref{fig1}. We found a peak at

175: $f=0.80358444\pm 0.00000112$ Hz which is consistent with the

176: extrapolated pulsar frequency $f=0.80358551$ Hz. The $Z_{1}^{2}$ for

177: this peak is 40.4, which has a probability of chance occurrence of

178: 1.69$\times 10^{-9}$. The pulse profile of \psr\ over the whole energy

179: band (bottom panel in Figure~\ref{fig2}) is broad and single-peaked

180: with a pulse fraction of $f_{p}=35\pm 12$\%. Here we defined the pulse

181: fraction as

182: $(C_{\mbox{max}}-C_{\mbox{min}})/(C_{\mbox{max}}+C_{\mbox{min}})$,

183: where $C_{\mbox{max}}$ and $C_{\mbox{min}}$ are the maximum and

184: minimum counts per bin, respectively.

185: 

186: We also looked at the pulse profile of the pulsar at different energy

187: bands (upper panels of Figure~\ref{fig2}). Most of the counts are in

188: the soft band, 67\% and 87\% of all counts are in the 0.2-1 and 0.2-2

189: keV energy bands, respectively. Pulsed fraction in each selected

190: energy band seems to be consistent with each other within the

191: 1$\sigma$ errors associated. The pulse shape does not seem to be

192: energy dependent which would suggest that all pulsed X-rays are coming

193: from the same region.

194: 

195: 

196: \subsection{Spectral}\label{spec}

197: 

198: The pulsar's energy spectrum was extracted from the MOS1/2 data by

199: selecting all events detected in a circle of radius 50\arcsec\

200: centered on the pulsar position. This region includes 90\% of all

201: event from the pulsar. Due to a source located close to the pulsar we

202: extracted the background spectrum from a nearby circular region with

203: radius 87\arcsec. For the EPIC-pn data we extracted the spectrum from

204: a circle centered on the pulsar with radius 30\arcsec\ (includes 80\%

205: of source counts). \psr\ was being located very close to the chip

206: boundary precluded the extraction of the background spectrum from an

207: annular region around the pulsar. Hence, we used an off-source

208: circular region with radius 66\arcsec.

209: 

210: To extract the spectrum from the \cha\ data we used a circular region

211: centered on the pulsar position with 2\arcsec\ radius. This region

212: contains 95\% of all source counts. The background spectrum was

213: extracted from an annulus of radii 3\arcsec$<r<$50\arcsec. 

214: 

215: In total, the extracted spectra include 780 EPIC-pn source counts and

216: 754 EPIC-MOS1/2 source counts. Both spectra were binned so that each

217: bin contained minimum 25 counts per bin. \cha\ data had a total of 184

218: source counts. The extracted photons were binned and regrouped such

219: that each fitted spectral bin contained minimum of 20 counts. All

220: three extracted spectra were then simultaneously fitted with model

221: spectra.

222: 

223: Among the single-component spectral models, an absorbed power-law

224: model gave the statistically best representation ($\chi^{2}=56.2$ for

225: 62 degrees of freedom (dof)) of the observed spectrum. A single

226: black-body ($\chi^{2}=93$ for 62 dof) did not give a statistically

227: acceptable fit. This fit when absorbing column is left to vary also

228: yields a very low $N_{\mbox{H}}$. The best fit power-law spectrum and

229: residuals are shown in Figure~\ref{fig3}. 

230: 

231: The power-law model yields a column density of

232: $N_{\mbox{H}}=1.38^{+0.37}_{-0.23}\times10^{21}$ cm$^{-2}$, a photon

233: index $\Gamma=3.20^{+0.26}_{-0.23}$, and a normalization of

234: $1.73^{+0.26}_{-0.22}\times10^{-5}$ photons cm$^{-2}$ s$^{-1}$

235: keV$^{-1}$ at $E=1$ keV. The errors are the upper and lower bounds of

236: the 1$\sigma$ confidence range. The normalization converts to an

237: unabsorbed energy flux of $f^{2-10}_{\mbox{x}}

238: =8.61^{+2.15}_{-0.33}\times10^{-15}$ \ergcm in the 2-10 keV

239: band. Given the distance of $d=1.45$ kpc this yields an X-ray

240: luminosity of $L^{2-10}_{\mbox{x}}

241: =2.17^{+0.56}_{-0.07}\times10^{30}$ \ergsec. This luminosity implies a

242: rotational energy to X-ray energy conversion factor

243: $L_{\mbox{x}}/\dot{E}= 0.015$ within 2-10 keV band.

244: 

245: It is natural to assume that, in addition to the magnetospheric

246: emission, thermal emission from polar caps or from the surface due to

247: reheating of the superfluid interior contributes to the observed X-ray

248: flux. In order to explore this possibility and how it represents the

249: data we used a two component model, thermal and magnetospheric, to fit

250: the spectra. As a first approach we let every parameter vary. The fit

251: gives a blackbody temperature of $T=3.28_{-0.62}^{+1.31}\times 10^{6}$

252: K and effective radius $R=59_{-46}^{+65}$ m. The power-law photon

253: index does not change significantly and has a value of

254: $\Gamma=2.98_{-0.65}^{+0.91}$. The bolometric luminosity of the

255: thermal component is $L_{\mbox{bol}}=2.87\times 10^{30}$ \ergsec

256: whereas the total luminosity in the 2-10 keV is $L^{2-10}_{\mbox{x}}

257: =1.67^{+0.91}_{-0.62}\times10^{30}$ \ergsec. The nonthermal X-ray

258: luminosity in the 2-10 keV band converts to an X-ray efficiency of

259: $L_{\mbox{x}}/\dot{E}= 0.01$. The hydrogen column density is not well

260: bound $N_{\mbox{H}}=0.62^{+0.98}_{-0.62}\times10^{21}$ cm$^{-2}$ but

261: still is consistent with the estimate obtained from the single

262: power-law fit. The quality of the composite model ($\chi^{2}=54.5$ for

263: 62 dof) is slightly better than that obtained from single power-law

264: model. However, both fits are statistically acceptable and addition of

265: a new model will always tend to decrease the $\chi^{2}$.

266: 

267: 

268: 

269: \section{Discussion}\label{dis}

270: 

271: With the observations of new sensitive X-ray telescopes like \xmm\ and

272: \cha\ we are uncovering a class of new sources; X-ray luminous

273: old-radio pulsars. Up to now there are only seven such sources that

274: have been detected. Namely, these are PSRs B2224+65, J2043+2740,

275: B1929+10, B0823+26, B0950+08, B0943+10 and B0628-28

276: \citep[see][]{zavlin,becker,zhang,ogelman}. Of these six sources only

277: two of them (PSRs B0950+08, B0628-28) have high quality spectra so

278: that one can distinguish between spectral models (e.g. thermal

279: vs. nonthermal). \citet{becker} suggested that the three pulsars

280: observed with \xmm\ (PSRs B0950+08, 0823+26, J2043+2740) all have

281: spectra described better by a single power-law model, indicating

282: nonthermal emission is dominant. For these pulsars, however, there are

283: alternate representations that do give statistically and physically

284: acceptable results \citep{zavlin}. For example, \citet{zavlin} argues

285: that the soft part of the spectrum of PSR B0950+08 can be interpreted

286: as radiation from heated polar caps on the neutron star surface

287: covered with a hydrogen atmosphere. For PSR B2224+65 \citet{zavlin}

288: arrives at the conclusion, from the analysis of the \cha\ ACIS

289: observation, that the pulsar definitely shows non-thermal

290: emission. All these pulsars have X-ray luminosities greater than what

291: is predicted from the correlation between

292: $L_{\mbox{x}}^{2-10}$-$\dot{E}$ \citep{possenti}. For example, in the

293: case of PSR B0628-28 and PSR B2224+65 the luminosities inferred from

294: spectral fits are greater then the so-called ``maximum efficiency

295: line'' derived by \citet{possenti} such that all pulsar lie below this

296: line in the $\log{L_{\mbox{x}}}$-$\log{\dot{E}}$ plane. \psr\ has been

297: known to be an exceptional emitter and with the detection of

298: pulsations from this source at the radio frequency has left no doubt

299: that it is the X-ray counterpart of the pulsar. These luminous old

300: pulsars (in particular \psr) seem not to follow the trend their

301: possible progenitors do. Which suggests that pulsars become more X-ray

302: efficient as they grow older, given that $\tau$ is inversely

303: proportional to $\dot{E}$. From multi-wavelength observations

304: \citet{zharikov} have arrived at the same conclusion. However,

305: \citet{zavlin} suggest that the most plausible reason for the

306: observed high nonthermal X-ray efficiencies associated with the old

307: pulsars is the geometrical effects. We discuss these below.

308: 

309: As mentioned in $\S$~\ref{intro}, \citet{possenti} used data from 39

310: X-ray emitting pulsars to find a best fit that describes the

311: correlation between $\log{L_{\mbox{x}}}$-$\log{\dot{E}}$. These 39 sources

312: were divided in to subcategories as; {\it{millisecond pulsars}}

313: ($P\la$10 ms), {\it{Crab-like}} ($\tau\sim 10^{4}$ years),

314: {\it{Vela-like}} ($\tau\sim 10^{4}$-$10^{5}$ years),

315: {\it{Geminga-like}} ($\tau\ga 10^{5}$ years, where substantial amount

316: of the X-ray flux comes from the internal cooling) and finally

317: {\it{Old-pulsars}} ($\tau\ga 10^{6}$ years). In Figure~\ref{fig4} we

318: show a similar plot. In this plot we only include Crab, Vela and

319: Geminga-like pulsars. We did not include millisecond pulsars because

320: they are re-cycled and do not represent the naturally aging pulsars

321: with the same $\dot{E}$. Also when converting countrates in to

322: luminosities, \citet{possenti} adopted a power-law spectrum with

323: $\Gamma=2$ for all millisecond pulsars. Hence, the luminosities do not

324: come from spectral analysis and rather from assumptions based on other

325: pulsars. Also we excluded the three Old-pulsars because their fluxes

326: were obtained by scaling their \ros\ countrates to that of PSR

327: B1929+10.

328: 

329: This newly formed set of pulsars represent the evolution of pulsars on

330: the $\log{L_{\mbox{x}}}$-$\log{\dot{E}}$ plane. Where older pulsars

331: are on the bottom left (low $\dot{E}$) and young Crab-like pulsars are

332: on the top right (high $\dot{E}$). Using only these three subclasses

333: (26 sources) we performed a linear fit of the form

334: $\log{L_{\mbox{x}}}=a\log{\dot{E}}+b$ (see Figure~\ref{fig4}). Due to

335: exclusion of the stated pulsars our fit yields a line with a steeper

336: slope ($a\sim 1.5$). We also identified the line of maximum efficiency

337: as the line for which every pulsar lies underneath. We then overlayed

338: only two of the seven old pulsars, with well known spectra. We

339: should note that, however, with their luminosities derived from the

340: power-law fits, PSR B0943+10 \citep{zhang} and PSR B2224+65

341: \citep{zavlin}, have efficiencies exceeding the maximum efficiency.

342: 

343: From Figure~\ref{fig4} it is apparent that old pulsars have very high

344: nonthermal X-ray efficiencies. This is contrary to what

345: \citet{possenti} has found and to what has been found when millisecond

346: pulsars were excluded. The observed luminosities for younger pulsars

347: also show large deviations, from this dependence. But in their case

348: the deviation is symmetric around the trend line. The scatter that

349: young pulsars exhibit could be due to uncertainties in pulsar

350: distances, spread in the orientations of magnetic and rotational axes

351: versus the line of sight. For example, seeing a certain fraction of

352: the beam would result in lower inferred X-ray efficiency and vice

353: versa.

354: 

355: Although there should be a thermal contribution to the overall

356: luminosity, this effect will not change the nonthermal luminosity

357: significantly. For instance, when we subtract the thermal luminosity,

358: obtained for \psr\ (see $\S$~\ref{spec}), from the total, the

359: nonthermal luminosity only changes by a factor of 0.5. The resulting

360: luminosity is still too big and gives a nonthermal X-ray efficiency

361: 2-3 orders of magnitudes greater than the maximum efficiency.

362: 

363: The quality of the observed spectrum for these old pulsars are in

364: general not high enough to distinguish between a thermal and a

365: non-thermal model, or a combination of both. The current set of old

366: pulsars all either show a non-thermal emission or the statistics is

367: not high enough to rule it out. For example, \citet{becker} suggests

368: that the X-ray emission from old, nonrecycled rotation powered pulsars

369: is dominated by non-thermal radiation. When the data is fit by a

370: power-law model, which in all cases is statistically acceptable, the

371: obtained power-law indices are on average larger than the younger

372: rotation powered pulsars. The mean of the photon indices of the 26

373: pulsars (Figure~\ref{fig4}) is $\sim$1.9 as opposed to $\sim$2.5 for

374: the old pulsars.  The power-law fits suggests that the spectra of

375: these old pulsars being steeper should result in lower luminosities in

376: the 2-10 keV band. However, we observe an inverse effect where not

377: only the spectra are steeper but also they have higher

378: luminosities. Further X-ray observations of old radio pulsars should

379: help us understand these puzzling features.

380: 

381: 

382: 

383: \begin{thebibliography}{}

384: \bibitem[Becker \ea(2004)]{becker} Becker, W., Weisskopf, M.C.,

385: Tennant, A.F., Jessner, A., Dyks, J., Harding, A.K, \& Zhang, S.N.,

386: 2004, \apj, 615, 908

387: \bibitem[Cordes \& Lazio(2002)]{cordes} Cordes, J.M., \& Lazio, T.J.W.,

388: 2002, preprint (astro-ph/0207156)

389: \bibitem[Halpern \& Ruderman(1993)]{halpern} Halpern, J.P., \& Ruderman,

390: M., 1993, \apj, 415, 286

391: \bibitem[\"Ogelman \& Tepedelenlio\v{g}lu(2004)]{ogelman} \"Ogelman,

392: H.B., \& Tepedelenlio\v{g}lu, 2004, Adv. Space Res., 33(4), 597

393: \bibitem[Possenti \ea(2002)]{possenti} Possenti, A., Cerutti, R.,

394: Colpi, M., Mereghetti, S., 2002, \aap, 387, 993

395: \bibitem[Shibazaki \& Lamb(1989)]{shibazaki} Shibazaki, N., \& Lamb,

396: F.K., 1989, \apj, 346, 808

397: \bibitem[Tsuruta(1998)]{tsuruta} Tsuruta, S., 1998, Phys. Rep., 292, 1

398: \bibitem[Zavlin \& Pavlov(2004)]{zavlin} Zavlin, V.E., \& Pavlov, G.G.,

399: 2004, \apj, 616, 452

400: \bibitem[Zhang \ea(2005)]{zhang} Zhang, B., Sanwal, D., \& Pavlov,

401: G.G., 2005, preprint (astro-ph/0503423)

402: \bibitem[Zharikov \ea(2004)]{zharikov} Zharikov, S., Shibanov, Yu., \&

403: Komarova, V., 2004, preprint (astro-ph/0410152 v3)

404: \end{thebibliography}

405: 

406: \newpage

407: \input{tab1.tex}

408: \newpage

409: \begin{figure}

410: \begin{center}

411: \includegraphics[angle=270,width=16cm]{f1.eps}

412: \end{center}

413: \caption{\small{The power spectrum of the timing data around the

414: expected frequency. The inset is the expanded view $\pm 500$ $\mu$Hz

415: around the peak, which is located at $f=0.80358551$ Hz.}}

416: \label{fig1}

417: \end{figure}

418: \newpage

419: \begin{figure}

420: \begin{center}

421: \includegraphics[angle=270,width=16cm]{f2.eps}

422: \end{center}

423: \caption{\small{X-ray light curves of \psr\ extracted from the EPIC-pn

424: data in four energy bands, with the values of the intrinsic pulsed

425: fraction $f_{p}$ and its 1$\sigma$ errors. Two phase cycles are shown

426: for clarity. For the expression used for the pulse fraction see

427: text.}}

428: \label{fig2}

429: \end{figure}  

430: \newpage

431: \begin{figure}

432: \begin{center}

433: \includegraphics[angle=270,width=16cm]{f3.eps}

434: \end{center}

435: \caption{\small{Energy spectrum of \psr\ as observed with pn (upper

436: spectrum), MOS1/2 (lower spectrum) and ACIS-S (middle spectrum). The

437: lines are the best fit power-law model. The lower panel is fit

438: residuals.}}

439: \label{fig3}

440: \end{figure}

441: \newpage

442: \begin{figure}

443: \begin{center}

444: \includegraphics[angle=0,width=16cm]{f4.eps}

445: \end{center}

446: \caption{\small{The X-ray luminosity in the band 2-10 keV versus the

447: spin-down power for the 26 sources together with two old pulsars. The

448: value for the X-ray luminosity of PSR B0950+08 is taken from

449: \citet{becker} and the rest are taken from \citet{possenti}. The solid

450: line is the best fit to the set of 26 pulsars. Other two lines are

451: labeled in the legend.}}

452: \label{fig4}

453: \end{figure}  

454: 

455: \end{document}

456: 

457: 

458: