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

2: %\documentclass[manuscript]{aastex}

3: %\documentclass[preprint2]{aastex}

4: %\documentclass{aastex}

5: %\usepackage{emulateapj5}

6: 

7: \def\xr {X--ray}

8: \newcommand{\gcc}{g~cm$^{-3}\ $}

9: \newcommand{\sfun}[2]{$#1(#2)\ $}

10: \newcommand{\rhonot}{$\rho_{\circ}\ $}

11: \newcommand{\msun}{$M_{\odot}\ $}

12: \newcommand{\greq}{$\stackrel{>}{ _{\sim}}$}

13: \newcommand{\lteq}{$\stackrel{<}{ _{\sim}}$}

14: \newcommand{\etal}{{\it et al.}}

15: \newcommand{\lsim}{\raisebox{-0.3ex}{\mbox{$\stackrel{<}{_\sim} \,$}}}

16: \newcommand{\gsim}{\raisebox{-0.3ex}{\mbox{$\stackrel{>}{_\sim} \,$}}}

17: \def\gta{\ifmmode {\mathbin{\lower 3pt\hbox   %> or of order

18:     {$\,\rlap{\raise 5pt\hbox{$\char'076$}}\mathchar"7218\,$}}}

19:     \else {${\mathbin{\lower 3pt\hbox

20:     {$\rlap{\raise 5pt\hbox{$\char'076$}}\mathchar"7218\,$}}}

21:     $}\fi}

22: \def\lta{\ifmmode {\,\mathbin{\lower 3pt\hbox   %< or of order

23:     {$\,\rlap{\raise 5pt\hbox{$\char'074$}}\mathchar"7218\,$}}}

24:     \else {${\mathbin{\lower 3pt\hbox

25:     {$\rlap{\raise 5pt\hbox{$\char'074$}}\mathchar"7218\,$}}}

26:     $}\fi}

27: 

28: \shorttitle {Double-peaked burst from 4U 1636--536}

29: \shortauthors {Bhattacharyya and Strohmayer}

30: 

31: \begin{document}

32: 

33: \title {A non-PRE double-peaked burst from 4U 1636--536: evidence for

34: burning front propagation}

35: 

36: \author {Sudip Bhattacharyya\altaffilmark{1,2}, and Tod

37: E. Strohmayer\altaffilmark{2}}

38: 

39: \altaffiltext{1}{Department of Astronomy, University of Maryland at

40: College Park, College Park, MD 20742-2421}

41: 

42: \altaffiltext{2}{X-ray Astrophysics Lab,

43: Exploration of the Universe Division,

44: NASA's Goddard Space Flight Center,

45: Greenbelt, MD 20771; sudip@milkyway.gsfc.nasa.gov,

46: stroh@clarence.gsfc.nasa.gov}

47: 

48: \begin{abstract}

49: We analyse Rossi X-ray Timing Explorer (RXTE) Proportional Counter

50: Array (PCA) data of a double-peaked burst from the low mass X-ray

51: binary (LMXB) 4U 1636--536 that shows no evidence for photospheric

52: radius expansion (PRE). We find that the X-ray emitting area on the

53: star increases with time as the burst progresses, even though the

54: photosphere does not expand. We argue that this is a strong indication

55: of thermonuclear flame spreading on the stellar surface during such

56: bursts. We propose a model for such double-peaked bursts, based on

57: thermonuclear flame spreading, that can qualitatively explain their

58: essential features, as well as the rarity of these bursts.

59: \end{abstract}

60: 

61: \keywords{accretion, accretion disks --- relativity --- stars: neutron --- 

62: X-rays: binaries --- X-rays: bursts ---  X-rays: individual (4U 1636--536)}

63: 

64: \section {Introduction} \label{sec: 1}

65: 

66: X-ray bursts are produced by thermonuclear burning of matter

67: accumulated on the surfaces of accreting neutron stars (Grindlay et

68: al. 1976; Belian, Conner, \& Evans 1976; Woosley \& Taam 1976, Joss

69: 1977; Lamb \& Lamb 1978). For most bursts, profiles are single

70: peaked, with rise times of the order of a fraction of a second to a

71: few seconds, and decay times of the order of ten or a few tens of

72: seconds.  However, for some bursts, double-peaked structure is

73: observed.  These peaks (with time separation of a few seconds) in a

74: single luminous burst can normally be explained in terms of

75: photospheric radius expansion (PRE; due to radiation pressure) and

76: contraction (Paczynski 1983; Ebisuzaki, Hanawa, \& Sugimoto 1984). As

77: the photosphere expands, the effective temperature decreases, and the

78: emitted photons shift towards lower energies. A subsequent contraction

79: of the photosphere has the opposite effect. This can cause a dip (and

80: hence the double-peaked structure) in the high-energy burst profile

81: (Lewin et al. 1976; Hoffman, Cominsky, \& Lewin 1980), although such a

82: structure is not frequently seen in bolometric or low-energy profiles

83: (see Smale 2001).

84: 

85: Double-peaked structure in weak X-ray bursts was discovered by Sztajno

86: et al. (1985) using EXOSAT observations of the low mass X-ray binary

87: (LMXB) system 4U 1636--536. For these bursts, two peaks are seen in

88: the bolometric profile, and even in low-energy profiles.  For this

89: reason, and as these bursts are not strong enough to cause

90: photospheric expansion, some other physical effects are needed to

91: explain them. Several models have been put forward to explain these

92: non-PRE double-peaked bursts: (1) two-step energy generation due to

93: shear instabilities in the fuel on the stellar surface (Fujimoto et

94: al. 1988), (2) a nuclear waiting point impedance in the thermonuclear

95: reaction flow (Fisker, Thielemann, \& Wiescher 2004), (3) heat

96: transport impedance in a two-zone model (Regev \& Livio 1984), and

97: (4) interactions with the accretion disk (Melia \& Zylstra 1992). As

98: we will elaborate in \S~3, none of these models can explain various

99: aspects of these bursts satisfactorily.

100: In this Letter, we propose a model for the double-peaked bursts based on thermonuclear flame

101: spreading on neutron stars, and 

102: comparing it qualitatively with the RXTE data of a double-peaked burst from

103: 4U 1636--536, we show that our model can explain the essential features of these bursts.

104: 

105: \section {Data Analysis and Model Calculations} \label{sec: 2}

106: 

107: We analyse the RXTE PCA archival data of a double-peaked burst (Date

108: of observation: Jan 8, 2002; ObsId: 60032-01-19-000) from 4U

109: 1636--536.  The heights of the two peaks are almost identical $(\sim

110: 2200$ counts/s/PCU), with a dip depth more than half

111: the peak height (Fig. 1).  This is a weak burst

112: compared to PRE bursts from this source, which can have $\sim 7000$

113: counts/s/PCU (see Strohmayer et al. 1998).  The burst

114: profiles at different energy bands are very similar

115: (Fig. 1), showing that this is not a PRE burst. However, the hardness

116: in panel {\it b} of Fig. 1 shows two striking features: (1) the first

117: peak of the hardness occurs $2-3$ seconds before that of the burst

118: profile; (2) the second hardness peak is much lower than the first

119: one, while the burst profile peaks are of similar height. As the

120: emitted flux primarily depends on source hardness (which is a measure

121: of temperature) and source emission area, feature (1) indicates that

122: the emission area increases with time. Feature (2) is possible if the

123: emission area at the time of the second peak is much higher than that

124: at the time of the first peak. As for a non-PRE burst, the emission area

125: can increase only if the burning region spreads on the stellar surface

126: from an initially small size, these two features

127: are consistent with thermonuclear flame spreading (Strohmayer, Zhang, \& Swank 

128: 1997; Kong et al. 2000).

129: 

130: As a next step, we break the burst profile into smaller time bins, and

131: for each bin perform spectral fitting. This gives the time evolution

132: of the spectral parameters. We fit the data with a single temperature

133: blackbody model (bbodyrad in XSPEC), as generally burst spectra are

134: well fit by a blackbody (Strohmayer \& Bildsten 2003). In doing this,

135: we fix the hydrogen column density $N_{\rm H}$ at a value

136: $0.56\times10^{22}$~cm$^{-2}$ (van Paradijs et al. 1986).

137: The results of these fits are shown in panels {\it c} \& {\it d} of

138: Fig. 1.  The radius is calculated from the ``normalization'' and

139: provides a measure of the source emitting area. The panels show that

140: the evolution of the temperature is similar to that of the hardness

141: (as expected), and the size of the emission area increases with time

142: (indicating flame spreading), first quickly, and then more slowly. The

143: temporal behavior of the radius also shows that this is not a PRE

144: burst, otherwise the radius would decrease from the time when the

145: burst profile attains its minimum between the two peaks.  However, the

146: reduced $\chi^2$ values are high for these fits $(> 1.5$ 

147: for 13 out of 29 time bins).  Considering

148: the arguments of the previous paragraph, this may be because of the

149: following reason: the emission is locally blackbody, but temperatures

150: at different locations on the stellar surface are significantly

151: different (as a result of slow flame spreading in comparison to the

152: timescale of temperature decay at a given location), and hence a

153: single temperature blackbody model can not fit the observed spectra

154: well.  However, the similar evolution of temperature to that of the

155: hardness indicates that these fits give

156: average blackbody temperatures on the stellar surface. This

157: explains the smaller height of the second temperature

158: peak (panel {\it c}, Fig. 1), as with the slow flame

159: spreading, temperature decays on most part of the star

160: before the flame engulfs the whole star, making the average

161: temperature smaller during the second peak.

162: The error bars in panels {\it c} \& {\it d} of Fig. 1 give $1\sigma$

163: errors.  As the reduced $\chi^2$ values for some of the time bins are high, 

164: increasing $\chi^2$ by 1 from the best fit value would

165: underestimate these errors. Therefore, we increase $\chi^2$ by

166: the amount of the reduced $\chi^2$ of the fit to calculate the

167: $1\sigma$ errors.

168: 

169: From the above analysis we infer that double-peaked bursts may be

170: caused by thermonuclear flame spreading on the stellar surface.

171: In order to show this, in our simple model, we consider that the fuel

172: (accreted matter) is distributed over the entire stellar surface,

173: the burst is ignited at a certain point, 

174: and then propagates on the surface to ignite all the fuel

175: gradually. For the particular double-peaked burst

176: analysed here, we assume that 

177: the burning region forms a $\phi$-symmetric belt very

178: quickly after ignition at or near the north pole

179: (while the observer's inclination angle, measured from this pole,

180: is $\le 90^{\rm o})$, in order to explain the non-observation

181: of millisecond period brightness oscillations, and the initially fast moving

182: front ``stalls'' for a time as it approaches the equator, before

183: speeding up again into the opposite hemisphere. 

184: This causes the burning front to take more time to reach the equator from

185: the mid-latitudes, and during that time hot portions of the star can

186: cool, causing a decrease in the emitted flux. Approaching the

187: equator, the front propagation speed increases again,

188: causing an increase of the emitted flux and the

189: observed double-peaked structure.

190: 

191: To qualitatively test this hypothesis we calculate the corresponding

192: model, assuming that the emitting region is a $\phi$-symmetric

193: belt extending from the north pole to a polar angle $\theta_{\rm edge}$.

194: To compare this model with the data, we need

195: to calculate the flux and spectrum at a certain time elapsed since burst onset

196: $(\Delta t)$, and hence it is essential to know $\theta_{\rm edge}$ and the temperature

197: at a given $\theta$-position in the belt as functions of $\Delta t$.

198: The first one can be determined from 

199: $\Delta t = \int_0^{\theta_{\rm edge}} d\theta/\dot\theta(\theta)$, 

200: if the burning front speed $\dot\theta(\theta)$ is known. To calculate

201: the temperature, we assume that after ignition the temperature

202: increases from $T_{\rm low}$ to 

203: $(T_{\rm low} + (0.99\times (T_{\rm high} - T_{\rm low})))$

204: following the equation $T(t) = T_{\rm low} + (T_{\rm high} - T_{\rm

205: low})\times(1-\exp(-t/t_{\rm rise}))$, and then decays exponentially with an

206: e-folding time $t_{\rm decay}$. In our model, we assume that

207: $\dot\theta(\theta) = F(\theta) = 1/(t_{\rm total}\times\cos\theta)$

208: for $\theta \le 90^{\rm o}$, and $\dot\theta(\theta) = F(180^{\rm o}-\theta)$

209: for $\theta \ge 90^{\rm o}$, where 

210: $t_{\rm total}$ is the time needed by the front to propagate from a pole to

211: the equator in the absence of any stalling. 

212: This expression of $\dot\theta(\theta)$ 

213: follows from Spitkovsky et al. (2002), as the neutron star in 4U 1636--536 is

214: rapidly rotating (spin frequency $\nu_* \approx 582$~Hz; Giles et al. 2002;

215: Strohmayer \& Markwardt 2002), and hence the effect of

216: the Coriolis force on the flame speed should be important. We assume that the

217: stalling of the front happens between the polar angles $\theta_1$ and $\theta_2$

218: in the northern hemisphere: $\dot\theta(\theta)$ decreases linearly from $\theta = \theta_1$ to

219: $\theta = \theta_m$, reaching a value $s/t_{\rm total}$, and then

220: increases linearly up to $\theta = \theta_2$ reaching a value $F(\theta_2)$.

221: In our calculations, we consider the Doppler, special relativistic, and

222: general relativistic (gravitational redshift and light-bending in

223: Schwarzschild spacetime) effects.

224: We compute model lightcurves and spectra for a range of parameter

225: values, and show an example in Fig. 2, which qualitatively reproduces

226: the observed features of the double-peaked burst. 

227: In panel {\it a} of Fig. 2, the burst profiles

228: qualitatively match (including the

229: depth of the dip) the data (see Fig. 1), except the

230: initial rise. For the model, the initial rise time is longer than that

231: for the data. An effect which may account for this discrepancy is the

232: radiative diffusion time, ie. the delay between ignition at depth and

233: emergence of the radiation.  Note also that we calculate the model

234: flux only up to the time when the burning front reaches the south

235: pole, while in Fig. 1, the real data probably extend beyond that time.

236: In panels {\it b} and {\it c} of

237: Fig. 2, we plot the model hardness and average temperature on the

238: stellar surface, respectively. We also fit our normalised model

239: spectra with the XSPEC model bbodyrad, in the same manner as for the

240: data. The resulting blackbody temperature and radius are shown in

241: panels {\it d} and {\it e} of Fig. 2.  Panels {\it b}, {\it c} and

242: {\it d} show a similar temporal behavior: both hardness and

243: temperature increase at the beginning rapidly, then decrease up to the

244: point when the burst profile reaches a minimum, increase slightly up

245: to the point when the burst profile reaches the second peak, and then

246: decrease again. This behavior is strikingly similar to that seen in

247: the burst data (Fig. 1). We note that

248: the temporal behavior of the model average temperature 

249: (Fig. 2) suggests that spectral fitting with a single temperature

250: blackbody model actually does give the average temperature on the

251: stellar surface. In panel {\it e} of Fig. 2, the evolution of the

252: radius shows an initial rapid increase, and then a slower increase,

253: which is also quite similar to the data (Fig. 1).

254: Therefore, simple modeling of pole to pole flame spreading (with a temporary stalling)

255: can reproduce the essential features of this double-peaked burst.

256: 

257: We note that the burst lightcurves are sensitive to the values of the parameters,

258: such as $\theta_1$, $\theta_2$, $\theta_{\rm m}$, etc. quantitatively, 

259: but not qualitatively. For example, the main effects of the increase of 

260: $\theta_1$, $\theta_2$, $\theta_{\rm m}$, $s$, $t_{\rm total}$, $t_{\rm rise}$ 

261: and $t_{\rm decay}$ are 

262: to decrease $d$, slightly decrease $d$, decrease $p$, decrease $l$,

263: increase the timescale of the whole burst, increase the rise time of the second peak,

264: and decrease $l$ respectively. Here,

265: $d$ is the time separation between the two peaks relative to the burst duration, 

266: $p$ is the ratio of the flux of second peak to that of first peak, and 

267: $l$ is the ratio of the depth of the dip (from the second peak) to the second peak

268: height.

269: 

270: \section {Discussion and Conclusions}

271: 

272: In this Letter we have presented a new model for double-peaked bursts,

273: that naturally explains the observed increase in emission area, 

274: which other models do not.  Moreover, it appears unlikely that model 1 (see

275: \S~1) can reproduce both the burst profile and the evolution

276: of hardness (or, temperature) simultaneously, as it does not consider

277: the emission area increase. There is also no real calculation of

278: double-peaked profiles from this model. In addition, if thermonuclear

279: flames spread in the way Spitkovsky et al. (2002) argue, it is very

280: difficult to see how sufficient unburnt fuel (as required by

281: model 1) can be maintained on top of the burnt fuel, as the full scale

282: height of the hot fuel is likely overturned and mixed with the cold

283: fuel.  We suggest that models 2 \& 3 (see \S~1) are probably unable to

284: reproduce the large dip (judging from the figures of Regev \& Livio

285: 1984; Fisker, Thielemann, \& Wiescher 2004), seen in the observed

286: burst. It is also unclear whether these models, as well as model 4

287: (see \S~1), can explain the observed hardness and/or temperature

288: evolution, and the rarity of the double-peaked

289: bursts. Note that the naive interpretation of a double-peaked burst

290: as two subsequent bursts (possibly in two hemispheres) can be ruled

291: out, because these two bursts would have to be localized (probably

292: by two magnetic poles; otherwise flame spreading without stalling

293: would give rise to a single peak), and in that case we would expect to observe

294: millisecond period brightness oscillations.

295: Our model can qualitatively reproduce the essential features of the

296: double-peaked bursts (see \S~2), including the burst profile (with a

297: large dip) and the hardness evolution. 

298: However it requires the burning front to stall for a few seconds, which

299: clearly warrants some justification and further study.

300: We suggest that accretion may provide a

301: mechanism to slow the front, although a detailed theoretical

302: calculation and modeling of the data are required to establish this.

303: The magnetic field of the neutron star in 4U 1636-53 is

304: probably comparatively low (as the source is not a millisecond X-ray pulsar), 

305: and accretion likely proceeds via a disk

306: around the equatorial plane. 

307: Therefore, a weak burst ignited near the north pole and proceeding towards

308: the equator may be impeded and stalled by the pole-ward flow of accreted matter in

309: the mid-latitudes, as this matter spreads from the equator towards

310: the poles, first rapidly, then more slowly (Inogamov \& Sunyaev 1999).

311: After reaching the vicinity of the equator, the burning region may be able to

312: inhibit accretion sufficiently to allow the front to speed up

313: again. This is because  the gravitational force on

314: particles falling onto the star via a disk is closely balanced by the centrifugal

315: force, and hence even weak bursts can probably inhibit accretion (see Inogamov \& Sunyaev

316: 1999), if thermonuclear flux is radiated near the equator. The cessation of accretion

317: may also allow the observer to get X-ray flux from the burning region

318: in some parts of the southern hemisphere. We emphasize that the credibility of these

319: arguments depends on the justification of accretion induced impedance of

320: front propagation, as the accretion flows exist much above the burning layer.

321: Here we give the following qualitative arguments. The $\phi$ component of linear speed 

322: $(v_{\phi})$ of accreted matter in the stellar atmosphere (in mid-latitudes and near

323: the equator) is $\sim 10^{5}$ km s$^{-1}$

324: (Fujimoto et al. 1988), and the corresponding latitudinal $(\theta)$ component 

325: $(v_{\theta})$ may be $\sim 1000$ km s$^{-1}$ (Inogamov \& Sunyaev 1999). Such 

326: accreted matter is likely to produce differential rotation in the inner layers

327: by the inflow of the angular momentum, which may extend down through the

328: burning shell (at the column depth of $\sim 10^8$ gm cm$^{-2}$; Fujimoto et al. 1988).

329: Therefore, in the burning layer, $v_{\theta}$ may be comparable to the

330: flame speed, which may be $\le 10$ km s$^{-1}$ for a rapidly spinning star

331: $(g \sim 10^{14}$ cm s$^{-2}$, $f = 3657$ rad s$^{-1}$ at $\theta = 60^{\rm o}$;

332: Spitkovsky et al. 2002). As a result, the burning front could plausibly be influenced

333: by the accretion-induced pole-ward motion of burning shell matter.

334: 

335: Double-peaked structure appears

336: only to be associated with weak bursts, perhaps because strong bursts

337: would tend to disrupt accretion sufficiently to preclude the kind of

338: front stalling that is required for the occurence of two peaks

339: according to our model.  The double-peaked feature is somewhat rare

340: even among the weak bursts.  This may be because in order to have the

341: double-peaked structure, the burst needs to be ignited at or near a

342: pole (so that the accretion can continue for a few seconds), which is

343: less probable than equatorial ignition (Spitkovsky et al. 2002). The

344: fact that double-peaked bursts are seen from only a few sources

345: (mostly from 4U 1636--536) can be understood in our model as

346: follows. These bursts require a low stellar magnetic field (for a

347: given accretion rate), so that accretion happens mostly through a disk

348: in the equatorial plane, and the disk must closely approach the star

349: (so that the gravitational force is closely balanced by the

350: centrifugal force near the surface). This is possible, if the

351: stellar equatorial dimensionless radius to mass ratio 

352: $R/M$ is large, and $\nu_*$ is high (making the radius of

353: the innermost stable circular orbit small; Bhattacharyya

354: et al. 2000). This relatively fine tuning among magnetic field,

355: accretion rate, equatorial $R/M$ and $\nu_*$ may exist for a

356: relatively small fraction of LMXBs. Therefore, our model

357: qualitatively explains the enigmatic rarity of the non-PRE

358: double-peaked bursts, and may also, in principle, enable constraints

359: on stellar magnetic fields and equatorial $R/M$ to be obtained.

360: 

361: Our work suggests that non-PRE double-peaked bursts can be important

362: in understanding thermonuclear flame spreading on neutron stars, which

363: may provide important insights about the millisecond period

364: brightness oscillations during X-ray bursts, and hence can be useful 

365: for constraining equation of state models of the dense matter in

366: the cores of neutron stars (Bhattacharyya \& Strohmayer 2005; 

367: Bhattacharyya et al. 2005).  However, the

368: rarity of such bursts has been an obstacle to understanding them, and

369: thus new attempts to expand the sample of these bursts seems well

370: warranted.

371: 

372: \acknowledgments

373: 

374: This work was supported in part by NASA Guest Investigator grants.

375: 

376: \clearpage

377: 

378: \begin{thebibliography}{}

379: 

380: \bibitem[]{392} Belian, R. D., Conner, J. P., \& Evans, W. D. 1976, \apj,

381: 206, L135.

382: 

383: \bibitem[]{} Bhattacharyya, S., \& Strohmayer, T. E. 2005, ApJ,

384: 634, L157 (astro-ph/0509370).

385: 

386: \bibitem[]{395} Bhattacharyya, S., Strohmayer, T. E., Miller, M. C., \&

387: Markwardt, C. B. 2005, \apj, 619, 483.

388: 

389: \bibitem[]{398} Bhattacharyya, S., Thampan, A. V., Misra, R., \& Datta,

390: B. 2000, \apj, 542, 473.

391: 

392: \bibitem[]{401} Ebisuzaki, T., Hanawa, T., \& Sugimoto, D. 1984, \pasj,

393: 36, 551.

394: 

395: \bibitem[]{404} Fisker, J. L., Thielemann, F., \& Wiescher, M. 2004,

396: \apj, 608, L61.

397: 

398: \bibitem[]{407} Fujimoto, M. Y., Sztajno, M., Lewin, W. H. G., \& van

399: Paradijs, J.  1988, A\&A, 199, L9.

400: 

401: \bibitem[]{410} Giles, A. B., Hill, K. M., Strohmayer, T. E., \&

402: Cummings, N. 2002, \apj, 568, 279.

403: 

404: \bibitem[]{415} Grindlay, J. E. et al. 1976, \apj, 205, L127.

405: 

406: \bibitem[]{417} Hoffman, J. A., Cominski, L. R., \& Lewin, W. H. G. 1980,

407: \apj, 240, L27.

408: 

409: \bibitem[]{420} Inogamov, N. A., \& Sunyaev, R. A. 1999, Astronomy

410: Letters, 25, 269.

411: 

412: \bibitem[]{423} Joss, P. C. 1977, \nat, 270, 310.

413: 

414: \bibitem[]{431} Kong, A. K. H., Homer, L., Kuulkers, E., Charles, P. A., \& Smale, A. P.

415: 2000, \mnras, 311, 405.

416: 

417: \bibitem[]{425} Lamb, D. Q., \& Lamb, F. K. 1978, \apj, 220, 291.

418: 

419: \bibitem[]{429} Lewin et al. 1976, \mnras, 177, 83P.

420: 

421: \bibitem[]{431} Melia, F., \& Zylstra, G. J. 1992, \apj, 398, L53.

422: 

423: \bibitem[]{433} Paczynski, B. 1983, \apj, 276, 315.

424: 

425: \bibitem[]{435} Regev, O., \& Livio, M. 1984, A\&A, 134, 123.

426: 

427: \bibitem[]{437} Smale, A. P. 2001, \apj, 562, 957.

428: 

429: \bibitem[]{439} Spitkovsky, A., Levin, Y., \& Ushomirsky, G. 2002, \apj,

430: 566, 1018.

431: 

432: \bibitem[]{442} Sztajno, M. et al. 1985, \apj, 299, 487.

433: 

434: \bibitem[]{444} Strohmayer, T. E., \& Bildsten, L. 2003, in {\it Compact

435: Stellar X-ray Sources}, Eds. W.H.G. Lewin and M. van der Klis,

436: (Cambridge University Press: Cambridge), (astro-ph/0301544).

437: 

438: \bibitem[]{448} Strohmayer, T. E., \& Markwardt, C. B. 2002, \apj, 577, 337.

439: 

440: \bibitem[]{457} Strohmayer, T. E., Zhang, W., \& Swank, J. H. 1997, \apj, 487, L77.

441: 

442: \bibitem[]{450} Strohmayer, T. E., Zhang, W., Swank, J. H., White, N. E., \&

443: Lapidus, I. 1998, \apj, 498, L135.

444: 

445: \bibitem[]{462} van Paradijs, J. et al. 1986, \mnras, 221, 617.

446: 

447: \bibitem[]{453} Woosley, S. E., \& Taam, R. E. 1976, \nat, 263, 101.

448: 

449: \end{thebibliography}{}

450: 

451: \clearpage

452: 

453: \begin{figure}

454: \epsscale{.80}

455: \hspace{-6.0cm}

456: \plotone{f1.eps}

457: \vspace{-4.0cm}

458: \caption{Double-peaked burst from 4U 1636--536: panel {\it a} gives

459: the burst profiles (for 3 PCUs on): curve 1 is for the channel range $0-63$ (nearly

460: bolometric), curve 2 is for the channel range $0-10$ (energy $<

461: 6.52$~keV), and curve 3 is for channel range $11-63$ (energy $>

462: 6.52$~keV). Panel {\it b} shows the time evolution of hardness (ratio

463: of counts in $11-63$ channel range to that in $0-10$ channel

464: range). For both these panels, the size of the time bin is 0.125 s.

465: Panels {\it c} \& {\it d} show the time evolution of the blackbody

466: temperature and the apparent radius (assuming 10 kpc source distance)

467: of the emission area respectively, obtained by fitting the burst

468: spectrum (persistent emission subtracted) with a single temperature

469: blackbody model.} 

470: \end{figure}

471: 

472: \clearpage

473: 

474: \begin{figure}

475: \epsscale{.90}

476: \hspace{-8.0cm}

477: \plotone{f2.eps}

478: \vspace{-5.0cm}

479: \caption{Model (convolved with a PCA response matrix) of

480: double-peaked bursts: for all the panels, the burst is normalised so

481: that its first intensity peak has the same count rate as that of the

482: first peak of the observed burst. Panels {\it a} \& {\it b} are

483: similar to those of Fig. 1. Panel {\it c} gives the time evolution of

484: average blackbody temperature on the stellar surface.  Panels {\it d}

485: \& {\it e} are similar to panels {\it c} \& {\it d} of Fig. 1

486: respectively. For these two panels, spectra are calculated for 0.5 s

487: time bins for each point. Model parameter values are the following:

488: stellar mass $M = 1.5 M_{\odot}$, dimensionless stellar radius to mass

489: ratio $R/M = 5.5$, stellar spin frequency $\nu_* = 582$~Hz, observer's

490: inclination angle (measured from north pole) $i = 50^{\rm o}$,

491: $\theta_1 = 67^{\rm o}$, $\theta_m = 83^{\rm o}$, $\theta_2 = 87^{\rm

492: o}$, $s = 0.04$, $t_{\rm total} = 11$~s, $t_{\rm rise} = 0.05$~s,

493: $t_{\rm decay} = 6$~s, $T_{\rm low} = 1$~keV, and $T_{\rm high} =

494: 2.8$~keV (see text for the definitions of the parameters).}

495: \end{figure}

496: 

497: \end{document}

498: