1: \documentstyle[11pt,aaspp4]{article}

2: \singlespace

3: 

4: \begin{document}

5: 

6: \title{Discovery of 16.6 and 25.5 s Pulsations from the Small Magellanic Cloud}

7: \author{R.C.Lamb$^1$, D.J. Macomb$^2$, T.A. Prince$^1$, and W. A. Majid$^3$}

8: 

9: $^1$ Space Radiation Laboratory, California Institute of Technology, Pasadena, CA

10: 91125

11: 

12: $^2$ Physics Department, Boise State University, Boise, ID 83725-1570

13: 

14: $^3$ NASA Jet Propulsion Laboratory, Pasadena, CA 91109

15: 

16: \keywords{accretion, accretion disks -- binaries: general -- pulsars,

17: X-rays}

18: 

19: Submitted to ApJ Letters, 2001 December 13

20: 

21: \begin{abstract}

22: 

23: We report the serendipitous detection of two previously unreported

24: pulsars from the direction of the Small Magellanic Cloud, with periods

25: of 16.6 and 25.5 seconds.  The detections are based on archival PCA

26: data from the {\it Rossi X-Ray Timing Explorer} (RXTE).  The

27: observation leading to these detections occurred in September 2000

28: extending over 2.1 days with an exposure of 121ks.  A possible

29: identification of the 16.6 pulsar with an X-ray source (RX J

30: 0051.8-7310) seen by both ROSAT and ASCA imaging X-ray satellites

31: is presented.

32: 

33: \end{abstract}

34: 

35: \section{INTRODUCTION AND RXTE RESULTS}

36: 

37: In the course of a survey of archival data from the Proportional

38: Counter Array (PCA) of the {\it Rossi X-Ray Timing Explorer}(RXTE:

39: Bradt, Rothschild, \& Swank 1993), we have found evidence for two

40: previously unreported pulsars with periods of 16.6 and 25.5 seconds..

41: The evidence is based on observations taken 2000 September 13-15, with

42: the PCA field of view, $\sim$1.0$^\circ$ FWHM, (Jahoda et al. 1996)

43: viewing an area near the southwestern edge of the Small Magellanic

44: Cloud, centered on 00$^h$ 50$^m$ 44.64$^s$, -73$^\circ$ 16$^{'}$

45: 04.8$^{``}$.

46: 

47: The observation in question extended for 2.1 days, with minimal

48: interruptions, leading to 68\% of the time devoted to source region

49: coverage.  Such extended ``dense'' observations are well-suited for

50: sensitive searches for periodic phenomena from relatively faint

51: sources.

52: 

53: The PCA observations used the so-called Good Xenon data mode, in which

54: the arrival of each photon at the detector is time tagged to an

55: accuracy of better than 1$\mu$s.  For the timing analysis the

56: electrical pulses were required to originate in either one of the top

57: two layers (of three) of the PCA, with pulse heights in the X-ray

58: energy range 2 to 15 keV.  Such an energy and layer selection is

59: effective in improving signal-to-noise for pulsar detection.

60: 

61: The times of arrival were corrected to the barycenter of the solar

62: system and a discrete Fourier transform (fft) of these times was

63: performed.  The portion of this transform from 0.02 to 0.2 Hz is shown

64: in Figure 1.

65: 

66: 

67: \begin{figure}

68: \figurenum{1}

69: \plotone{f1.eps}

70: \caption{FFT of the RXTE Observation}

71: \end{figure}

72: 

73: 

74: There are four distinct frequencies and one second harmonic which are

75: evident in the spectrum of Figure 1.  The frequencies, corresponding

76: periods, and Fourier power (normalized to 1) are listed in Table 1.

77: Two of the periods are consistent with those that have been previously

78: reported from this region of the SMC.  The 8.9 s period is identified

79: with the same period pulsation from the Be transient RX J0051.8-7231

80: (Israel et al. 1997).  The 15.7 period may be identified with the

81: reported 15.3 s period from the SMC transient RX J0052.1-7319 (Finger

82: et al. 2001).  The observations reported by Finger et al. for this

83: object occurred 3.8 years earlier than the RXTE observation of Figure

84: 1.  The difference in period, 2.5\%, can be accounted for by a

85: spin-down episode(s) consistent with spin histories of other Be

86: transients (Bildsten et al. 1997), therefore we regard this

87: identification as reasonable.  For the remaining periods, 16.6 and

88: 25.5 s, we have found no previous reports covering this portion of the

89: sky.

90: 

91: \begin{deluxetable}{rrr}

92: 

93: \tablecolumns{3}

94: 

95: \tablecaption{List of Frequencies with Power $>$20}

96: \tablewidth{0pt}

97: \tablehead{\colhead{Frequency(Hz)} & \colhead{Nomalized Power} & \colhead{Period (s)}}

98: \startdata

99: 

100: 0.039230 & 48. & 25.5 \\

101: 0.060339 & 196. & 16.6 \\

102: 0.063861 & 54. & 15.7 \\

103: 0.112401 & 58. & 8.90 \\

104: 0.120685 & 22. & 16.6 (2Harm) 

105: 

106: \enddata

107: \end{deluxetable}

108: 

109: 

110: In what follows we present the lightcurves for both the 16.6 and the

111: 25.5 s pulsars and discuss evidence for a possible identification of

112: the 16.6 s pulsar with a ROSAT source, RX J0051.8-7310.  In the

113: concluding section we discuss the luminosities of sources.  If these

114: pulsars are associated with the SMC, as is likely, their discovery

115: further accentuates the remarkable overdensity of binary pulsars in

116: the SMC relative to our own galaxy.

117: 

118: \section{LIGHTCURVES}

119: 

120: There is evidence for a significant spin-down of the frequency of the

121: 16.6 s pulsar during the course of the RXTE observation.  We find that

122: the time derivative of the frequency which maximizes the sum of the

123: power at the fundamental and second harmonic ($\sim$0.12 Hz) is

124: -4.6$\pm$1.0$\times$10$^{-11}$ Hz/s, at a frequency of 0.0603435 Hz.

125: With these values the sum of the power at the fundamental and second

126: harmonic is 265; with no frequency derivative the power sum was

127: 218. The lightcurve resulting from these values of frequency and its

128: time derivative is shown in Figure 2 a).  Using the lowest bin of the

129: lightcurve to establish an unpulsed level we find a pulsed counting

130: rate of 0.60$\pm$0.08 photons/s.

131: 

132: The upper limit for a non-zero frequency derivative for the 25.5 s

133: pulsar is 3$\times$10$^{-11}$ Hz/s.  Its lightcurve is shown in Figure

134: 2 b).  Its pulsed counting rate is 0.49$\pm$0.08 photons/s.

135: 

136: \begin{figure}

137: \figurenum{2} \epsscale{0.60} \plotone{f2.eps}

138: \caption{Lightcurves for both the 16.6 and 25.5 s pulsars}

139: \end{figure}

140: 

141: 

142: 

143: 

144: \section{POSSIBLE IDENTIFICATION OF THE 16.6 S PULSAR}

145: 

146: We have examined the public archive of the ROSAT satellite for data

147: relevant for the possible identification of either of these newly

148: discovered pulsars.  There are 4 PSPC observations which are within

149: 30$^{'}$ of the center of the RXTE PCA's field-of-view.  Two of the

150: four observations have insufficient exposure ($<\sim$2 ks) for a

151: sensitive search for pulsations from any of the relatively faint

152: sources in the field.  Both of the remaining observations have similar

153: deep exposures, $\sim$ 20ks.  However one of these (rp500249n00) is

154: significantly more useful than the other (rp600149n01) since the

155: latter observation has a duration more than 3 times as long.  For a

156: fixed number of pulsed photons, the detection sensitivity for a pulsar

157: will decline as the length of the observation is increased.  This is

158: because the number of independent frequencies required to cover a

159: given frequency range increases as T where T is the total observation

160: duration.  Also the frequency variations which are to be expected (on

161: the basis of the RXTE observations for the 16.6 s pulsar) will further

162: vitiate the sensitivity of long observations.  Therefore, we have

163: examined only rp500249n00 for possible counterparts to the RXTE

164: sources.

165: 

166: In the field of rp500249n00 there are 5 sources with a counting rate

167: greater than 0.01 cts/s.  None of these 5 are identified with

168: previously reported pulsars.

169: 

170: We have Fourier transformed the arrival times of the photons from each

171: of the 5 sources with the view of establishing the identity and

172: precise location of either of the 16.6 or the 25.5 s pulsars.  The

173: ROSAT observation occurred 6.8 years earlier than the RXTE

174: observation.  In order to account for possible episodes of spin-up and

175: spin-down over that interval we have searched a range of $\pm5$\%

176: around the signal frequencies of 0.06034 and 0.03923 Hz.

177: 

178: In order to assess the significance of any possible detection the

179: number of trials must be included in the calculation of probability

180: that a given outcome could be due to chance.  The number of trials is

181: given by the number of independent frequencies which span a given

182: search range times a possible oversampling factor if the frequency

183: digitization is finer than 1/T.  We have oversampled the time series

184: and use an oversampling factor of 3.

185: 

186: For the 0.039 Hz search region, there are 1060 independent

187: frequencies.  None of the 5 sources showed any power $>8$ over that

188: region.  The probability of a power 8 or more occurring by chance is

189: therefore (1060 ind. trials)(3 oversampling factor)(5 sources)(exp(-8)) $>$

190: 1.  Thus there is no evidence for the 25.5 s pulsar in this dataset.

191: 

192: For the 16.6 s search range, it is a somewhat different story.  The

193: largest power occurring in the search range for any of the 5 sources

194: was 14.0.  The probability that any one of the sources would give a

195: power $>$14 is (1630)(3)(5)(exp(-14)) = 2.0\%.  Therefore this

196: source, RX J0051.8-7310, is a candidate for identification with the

197: 16.6 s pulsar.  The portion of the fft from 0.02 to 0.2 Hz for this

198: source is shown in Figures 3 a) and b).

199: 

200: The 4 peaks evident in the figure near 0.0615 Hz are due to the sparse

201: sampling of the ROSAT data, which extended for 270 ks, with only 19 ks

202: of exposure.  The spacing of the 4 peaks (0.174 mHz between peaks) is

203: due to the orbital period of the ROSAT spacecraft ($\sim$95 minutes).

204: We have verified this behavior, using a simulation of the observation

205: with an artifical 0.06 Hz signal arranged to mimic the exposure time

206: of the ROSAT observation.  In the simulation, a ``picket fence'' of

207: power peaks around 0.06 Hz similar to that of Figure 3 b) was seen,

208: with the separation between adjacent power peaks given by 0.174 mHz.

209: 

210: We have searched for evidence for a possible frequency derivative of the

211: putative signal and find none.

212: 

213: Further support for the identification comes from an analysis of an

214: ASCA observation (48003010).  In this observaion RX J0051.8-7310 was

215: in ASCA's Gas Imagining Spectrometer's field of view.  This

216: observation extended for 2.8 days beginning 2000 April 11.  Again we

217: use a search range of $\pm$5\% around the RXTE frequency of 0.06034

218: Hz.  Although the ASCA observation occurred less than six months prior

219: to the RXTE observations, episodes of appreciable spin-up or spin-down

220: may occur on rather short time scales (Bildsten et al. 1997).  We

221: selected photons from the region of RX J0051.8-7310 and performed an

222: fft.  In the search range, the highest power was 10.1.  There were 145

223: independent frequencies in the search range and again, we use an

224: oversampling factor of 3.  The probability of a power exceeding this

225: value in the search range is therefore: (1450)(3)exp(-10.1) = 18\%.

226: 

227: We then added a search on a possible frequency derivative, varying it

228: from -1.0$\times10^{-9}$ to +1.0$\times$10$^{-9}$ Hz/s.  The spacing

229: between independent values of the frequency derivative is given by

230: $1/2T^2$.  Thus to cover this search range 230 trials with an

231: additional factor of 3 for oversampling are required.  From this

232: search, a power 19.5 occurred at a frequency of 0.06028 Hz and a

233: frequency derivative of 9.0$\pm$0.5$\times$10$^{-11}$ Hz/s.  The

234: probability of this occurring by chance is given by:

235: (230)(3)(1450)(3)exp(-19.5) = 1.0$\times10^{-2}$.  Figures 3 c) and d)

236: show a portion of the fft from the ASCA observation.

237: 

238: 

239: We may combine the probabilities of the ROSAT and ASCA observations to

240: arrive at an overall probability that these two observations are due

241: to chance.  That probability is 2$\times$10$^{-4}$.  This number is

242: sufficiently small to suggest that the identification is correct.

243: However we propose that the identification remain tentative until it

244: is supported by further imaging X-ray satellite observations.

245: 

246: If the identification is correct, then we may use ROSAT HRI data to

247: establish an accurate position.  Using an 27ks exposure from 1996

248: November (rh600811n00) which produced $\sim$ 80 signal photons on a

249: background of 10, we find a position of RA = 00$^h$ 51$^m$ 51.2$^s$,

250: Dec. = 73$^\circ$ 10$^{'}$ 32$^{''}$, with an error of 7$^{''}$.  We

251: note that this position agrees, within errors, with its position

252: derived from PSPC data alone by Kahabka et al. (1999).

253: 

254: \begin{figure}

255: \figurenum{3} \epsscale{1.00}

256: 

257: \plotone{f3.eps}

258: 

259: \caption{ROSAT and ASCA observations of a possible counterpart to

260: the 16.6s pulsar.  For the ROSAT fft, Figure 3 a) and b), no frequency

261: derivative is used.  For the ASCA fft, Figure 3 c) and d), a frequency 

262: derivative of 9.0$\pm$0.5$\times$10$^{-11}$ Hz/s is used.}

263:  

264: \end{figure}

265: 

266: 

267: \section{DISCUSSION}

268: 

269: For the 25.5s pulsar we derive a pulsed flux value of

270: 6.9$\pm$0.8$\times$10$^{-12}$ (2-15 keV) under model assumptions of a

271: power-law spectrum with an energy index of -2.0 and a hydrogen column

272: density of 1.0$\times$10$^{21}$atoms/cm$^2$.  Under the same assumptions

273: the 16.6 pulsar flux is 8.5$\pm$1.1$\times$10$^{-12}$ (2-15 keV).

274: At the distance of the SMC (nominally 60 kpc) these pulsed fluxes correspond

275: to isotropic luminosities of 3.0 and 3.7$\times$10$^{36}$ ergs/s.

276: These fluxes are a few percent of the Eddington luminosity, 2$\times$10$^{38}$

277: ergs,s, for a 1.4 solar mass object.

278: 

279: Two arguments can be made to support the idea that at least one if not

280: both of these pulsars are binary.  Isolated pulsars with periods

281: greater than 10s have yet to be seen.  Also there is a significant

282: non-zero period derivative for the 16.6 s pulsar with a time-scale of

283: $\sim$ 50 years.  Further analysis of existing and future RXTE observations 

284: may be able to constrain possible orbital parameters for this object.

285: 

286: If these pulsars are binaries in the SMC, their discovery further

287: accentuates the dramatic difference between the SMC and our Galaxy

288: with regard to the population of such systems.  This fact has been

289: noted by several authors (Schmidtke et al. 1999, Yokogawa et

290: al. 2000).

291: 

292: In a recent compilation of known X-ray pulsars $^4$ there are 18 X-ray

293: pulsar binaries listed for the SMC all of which are either high-mass

294: or transient and therefore likely to be high mass systems.  We cannot

295: say if these newly discovered pulsars are high or low-mass.  However,

296: if they are high mass, then this increases the number of such systems

297: in the SMC to 20.  For the Galaxy the corresponding number is 40.

298: Therefore, using a mass ratio of the SMC to the Galaxy of 1/100, this

299: suggests that such systems are over-abundant by a factor of $\sim$50

300: relative to the Galaxy.  This simple analysis ignores important issues

301: regarding the uncertain coverage of the Galaxy for transient X-ray

302: binaries versus the rerlatively complete coverage of the SMC.  It also

303: ignores differences in X-ray absorption effects.  Nevertheless,

304: pending careful analysis of such issues, there appears to be a

305: significant overabundance of high mass binaries in the SMC relative to

306: the Galaxy.  Since high mass X-ray binaries have lifetimes which are

307: $\sim10^{-3}$ the age of the Galaxy and possibly the SMC, this

308: difference between the SMC and the Galaxy may point to a rather

309: recent outburst of star-formation in the SMC within the past

310: $\sim10^{7}$ years.  Yokogawa et al. (2000) reach a similar

311: conclusion.

312: 

313: $^4$ http://gammaray.msfc.nasa.gov/batse/pulsar/asm.pulsars.html

314: 

315: \begin{thebibliography}{}

316:  \bibitem[Bildsten]{bil} Bildsten, L., et al. 1997,ApJS, 113, 367 

317: 

318: \bibitem[Bradt(1993)]{bra} Bradt, H.V., Rothschild, R.E., \&Swank, J.H. 1993, A\&AS, 97, 355

319: 

320:  \bibitem[Finger(2001)]{fin} Finger, M.H., et al. 2001, ApJ, 560, 378

321: 

322: \bibitem[Jahoda(1996)]{jah} Jahoda, K., et al. 1996, Proc. SPIE, 2808, 59

323: 

324: 

325: \bibitem[Israel(1997)]{isr} Israel, G.L., et al. 1997, ApJ, 484, L141

326: 

327: \bibitem[Kahabka(1999)]{kah} Kahabka, P., Pietsch, W., Filipovic, M.D., \&

328: Haber, F. 1999, A\&AS, 136, 81

329: 

330: \bibitem[Schmidtke(1999)]{sch} Schmidtke, P.C., et al. 1999. AJ, 117, 927

331: 

332: \bibitem[Yokogawa(2000)]{yok} Yokogawa, J., et al. 2000, ApJS, 128, 491

333: 

334: \end{thebibliography}

335: 

336: 

337: \end{document}

338: