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 {Flame Spreading on Neutron Stars}
29: \shortauthors {Bhattacharyya and Strohmayer}
30:
31: \begin{document}
32:
33: \title {Thermonuclear Flame Spreading on Rapidly Spinning Neutron Stars:
34: Indications of the Coriolis Force?}
35:
36: \author {Sudip Bhattacharyya\altaffilmark{1,2}, and Tod
37: E. Strohmayer\altaffilmark{3}}
38:
39: \altaffiltext{1}{CRESST and X-ray Astrophysics Lab, Astrophysics
40: Science Division, NASA's Goddard Space Flight Center, Greenbelt, MD
41: 20771; sudip@milkyway.gsfc.nasa.gov} \altaffiltext{2}{Department of
42: Astronomy, University of Maryland, College Park, MD 20742}
43: \altaffiltext{3}{X-ray Astrophysics Lab, Astrophysics Science
44: Division, NASA's Goddard Space Flight Center, Greenbelt, MD 20771;
45: stroh@clarence.gsfc.nasa.gov}
46:
47: \begin{abstract}
48:
49: Millisecond period brightness oscillations during the intensity rise
50: of thermonuclear X-ray bursts are likely caused by an azimuthally
51: asymmetric, expanding burning region on the stellar surface. The time
52: evolution of the oscillation amplitude during the intensity rise
53: encodes information on how the thermonuclear flames spread across the
54: stellar surface. This process depends on properties of the accreted
55: burning layer, surface fluid motions, and the surface magnetic field
56: structure, and thus can provide insight into these stellar
57: properties. We present two examples of bursts from different sources
58: that show a decrease in oscillation amplitude during the intensity
59: rise. Using theoretical modeling, we demonstrate that the observed
60: amplitude evolution of these bursts is not well described by a
61: uniformly expanding circular burning region. We further show that by
62: including in our model the salient aspects of the Coriolis force (as
63: described by Spitkovsky, Levin, and Ushomirsky) we can qualitatively
64: reproduce the observed evolution curves. Our modeling shows that the
65: evolutionary structure of burst oscillation amplitude is sensitive to
66: the nature of flame spreading, while the actual amplitude values can
67: be very useful to constrain some source parameters.
68:
69: \end{abstract}
70:
71: \keywords{relativity --- stars: neutron --- stars: rotation ---
72: X-rays: binaries --- X-rays: bursts --- X-rays: individual (4U
73: 1636--536, SAX J1808.4--3658)}
74:
75: \section {Introduction} \label{sec: 1}
76:
77: Millisecond period brightness oscillations, ``burst oscillations'',
78: during thermonuclear X-ray bursts (Strohmayer \& Bildsten 2006;
79: Woosley \& Taam 1976; Lamb \& Lamb 1978) from accreting neutron stars
80: were first discovered with the Rossi X-ray Timing Explorer ({\it
81: RXTE}) in 1996 (Strohmayer et al.). Since then, a total of 19 low mass
82: X-ray binaries (LMXBs; including three tentative detections; Thompson
83: et al. 2005; Kaaret et al. 2007; Bhattacharyya 2007) have exhibited
84: this timing feature (see Lamb \& Boutloukos 2007; Bhattacharyya et
85: al. 2006; Markwardt et al. 2007). The discovery of burst oscillations
86: at the spin frequencies of the accreting millisecond pulsars SAX
87: J1808.4--3658 and XTE J1814--338 (see Chakrabarty et al. 2003;
88: Strohmayer et al. 2003) has conclusively linked these oscillations to
89: neutron star spin, and showed that the burst oscillation frequency
90: gives a direct measure of the stellar spin frequency. This discovery
91: has also established that burst oscillations originate at the stellar
92: surface, and hence can be very useful to constrain neutron star mass
93: and radius (Miller, \& Lamb 1998; Muno et al. 2002). Measurements of
94: these stellar parameters are critical to understand the nature of
95: high-density matter inside neutron stars, which is a fundamental
96: problem of physics (Bhattacharyya et al. 2005).
97:
98: Thermonuclear X-ray bursts also provides a powerful tool to probe the
99: physics of flame spreading under the extreme conditions that exist on
100: neutron stars. This is because bursts ignite at a particular point on
101: the stellar surface, and then spread to burn all the surface fuel
102: (Fryxell \& Woosley 1982; Spitkovsky et al. 2002; Bhattacharyya \&
103: Strohmayer 2006a; 2006b). As flame spreading depends on shearing flows
104: in the surface layers (Spitkovsky et al. 2002), the study of this
105: spreading can be very useful to understand surface fluid motions.
106: Moreover, flame spreading can be helpful for mapping the magnetic
107: field structure on the stellar surface. This is because the shearing
108: flows present during flame spreading can make this field locally
109: strong, which may in turn act back on the flow (Spitkovsky et
110: al. 2002; Bhattacharyya \& Strohmayer 2006c). However, a detailed
111: theoretical study of flame spreading including all relevant physical
112: effects has not yet been done, perhaps partly due to a lack of
113: observational motivation. Such motivation includes the significant
114: detection and measurement of the diagnostic features, such as the
115: evolution of spectral and burst oscillation properties during the
116: spreading. But, flame spreading is expected to occur during burst
117: rise, and in this short period (sub-second to a few seconds) the
118: significant detection of the diagnostic aspects is difficult due to
119: the statistical quality of the present data.
120:
121: Recently Spitkovsky et al. (2002) have proposed a theoretical model of
122: flame spreading on rapidly spinning neutron stars including the
123: effects of the Coriolis force. According to these authors, after
124: ignition the Coriolis force is initially not important due to the
125: small size of the burning region (which implies a large Rossby
126: number). During this time the flame spreads quickly with the so-called
127: geostrophic speed. So, the burning region (hot spot) attains a
128: considerable size almost at the beginning of the burst rise. After
129: this, for a rapidly spinning neutron star the Coriolis force becomes
130: important, and the flame spreads as a shearing flow with the much
131: slower ageostrophic speed. This speed decreses as the burning front
132: latitude increases, which causes a portion of the burning region to
133: encircle the stellar equator quickly, maintaining an asymmetric
134: portion at higher latitudes for some more time (for mid-latitude
135: ignition; Fig. 8c of Spitkovsky et al. 2002; see also Bhattacharyya \&
136: Strohmayer 2006c).
137:
138: Strohmayer et al. (1997) found that during the rise of some bursts
139: from the LMXB 4U 1728--34, the oscillation amplitude diminished, which
140: is expected for an expanding burning region during burst rise. These
141: initial results suggested that the tracking of amplitude evolution can
142: be a useful probe of flame spreading. However, more detailed
143: observational measurements of amplitude evolution have been found and
144: reported only recently (for 4U 1636--536 and SAX J1808.8--3658;
145: Bhattacharyya \& Strohmayer 2005; 2006c). In this Letter (in \S~2), we
146: first show that these more detailed evolutionary curves exhibit
147: amplitude changes (decreasing with burst rise). Then, using
148: theoretical models we demonstrate that these observed curves cannot be
149: easily explained with a uniformly expanding hot spot. Finally, we show
150: that models that include the salient features of the Coriolis force
151: can qualitatively reproduce the observed curves. In \S~3, we discuss
152: the implications of our analysis.
153:
154: \section {Analysis} \label{sec: 2}
155:
156: Tracking the burst oscillation amplitude evolution during burst rise
157: requires a strong oscillation throughout the rise. Therefore, given
158: the (1) small duty cycle of bursts, (2) non-detection of burst rise
159: oscillations for some bursts, (3) short burst rise period, (4) low
160: count rate during the early burst rise, and the fact that only {\it
161: RXTE} can currently detect burst oscillations, so far clear amplitude
162: evolution curves (for burst rise) have been reported only for two
163: bursts (See Fig. 1) from two rapidly spinning neutron stars (4U
164: 1636--536: spin frequency $\nu_{\rm *} = 582$ Hz; and SAX
165: J1808.8--3658: $\nu_{\rm *} = 401$ Hz). Hence, if flame spreading is
166: influenced by the Coriolis force, then the bursts from these two
167: sources should show this influence. The evolutionary curve for each of
168: these bursts is consistent with a rapid amplitude decay, followed by a
169: nearly constant level of significantly non-zero oscillation amplitude
170: for some time. Before using these observed properties to probe flame
171: spreading, first we show that the data are not consistent with a
172: constant amplitude level. In order to demonstrate this, we have fitted
173: each data set with a constant amplitude, and the best-fit levels are
174: shown in Fig. 1. The resulting $\chi^2/\nu$ for these levels (20.7/7
175: for 4U 1636--536, and 18.1/9 for SAX J1808.8--3658) clearly shows the
176: significant decrease of the amplitude during burst rise. Motivated by
177: this, we have theoretically computed oscillation amplitude evolution
178: for an expanding hot spot on a neutron star's surface assuming various
179: source parameter values. Although some previous studies explored such
180: model computations (Strohmayer et al. 1997; Nath et al. 2002;
181: Strohmayer 2004), here we calculate and compare various models in
182: detail for the first time.
183:
184: We have started with the simplest spreading geometry, viz, a uniformly
185: expanding circular hot spot. If the effect of stellar spin (and hence
186: the Coriolis force) on spreading is negligible, then this may be a
187: reasonable approximation to how flames actually spread. This is
188: because, with a relatively low stellar magnetic field ($10^7-10^9$ G)
189: for LMXBs, the accreted matter (fuel) should not be confined, and
190: should be present all over the surface almost uniformly. In our
191: calculations, we have considered eight source parameters (see also
192: Bhattacharyya et al. 2005): (1) the dimensionless neutron star
193: radius-to-mass ratio $R/M$, (2) the stellar mass $M$, (3) $\nu_{\rm
194: *}$, (4) the observer's inclination angle $i$, (5) the polar angle
195: $\theta_{\rm c}$ of the hot spot center, (6) the angular radius
196: $\Delta \theta$ of the spot, (7) the blackbody temperature of the spot
197: $T_{\rm BB}$ (burst spectra can normally be well fitted with a
198: blackbody model; Strohmayer \& Bildsten 2006), and (8) a parameter $n$
199: that gives a measure of the beaming in the emitter's frame, where the
200: specific intensity as a function of the angle $\psi$ (in the emitter's
201: frame) from the surface normal is $I(\psi) \propto \cos^n\psi$. This
202: beaming may be due to the scattering of burst photons in an optically
203: thick layer of thermal electrons. Note that for semi-infinite
204: plane-parallel layers with a constant net flux and Thomson scattering,
205: $n \sim 0.5$ (Chandrasekhar 1960). In our calculation of energy
206: dependent flux from a hot spot, we have taken into account (1) Doppler
207: and special relativistic effects, (2) gravitational redshift, and (3)
208: light-bending (in a Schwarzschild spacetime). In order to include
209: light-bending, we have backtracked the paths of the photons from the
210: observer to the source (see Bhattacharyya et al. 2001; 2005). The
211: model light curve is calculated by repeating the same procedure for
212: many identical spots at different $\phi$-positions (but the same
213: $\theta$-position) on the surface of the star. The actual phase
214: points of the light curve are calculated from these $\phi$-positions,
215: the stellar spin frequency, and the time delay considerations. The
216: time delays result from the fact that photons emitted at different
217: points on the stellar surface take different times to reach the
218: observer. We have then folded this model light curve with a suitable
219: {\it RXTE} PCA response matrix, and determined the oscillation
220: amplitude in the same way it was determined from the data. We have
221: repeated this procedure for many hot spots with increasing $\Delta
222: \theta$ (for a chosen combination of other parameter values) to
223: determine the oscillation amplitude evolution.
224:
225: In Fig. 2, we have plotted these evolutionary curves for various
226: parameter values. Note that, here, we only show the points with
227: fractional peak amplitude less than 1. Fig. 2 clearly shows that the
228: model curves for widely different parameter values are qualitatively
229: very similar, that is, the amplitude first decreases slowly, and then
230: decreases rapidly. On the other hand, the observed curves of Fig. 1
231: first decrease rapidly, and then maintain a significantly non-zero
232: level. A qualitative comparison of Fig. 1 \& 2 suggests that the
233: uniform expansion of a circular hot spot cannot easily explain the
234: observed amplitude evolution during rise for these bursts. Fig. 2
235: also shows that the oscillation amplitude increases with the increase
236: of $R/M$, $M$, $\nu_{\rm *}$ $i$, $\theta_{\rm c}$ (except when
237: $\theta_{\rm c}$ is close to $180^{\rm o}$), and $n$, and with the
238: decrease of $\Delta \theta$ and $T_{\rm BB}$. Among these parameters,
239: $i$, $\theta_{\rm c}$, $R/M$ and $n$ affect the amplitude the most
240: (apart from $\Delta \theta$, which is not a free parameter for an
241: evolutionary curve), with the first two having the maximum effects.
242: Consequently, these parameter values can be measured more confidently
243: from oscillation amplitudes (see the last sentence of \S~2). We note
244: that a large variation of $T_{\rm BB}$ does not change the amplitude
245: very much, which justifies (given the quality of data) our use of the
246: same blackbody temperature throughout the flame spreading.
247:
248: Motivated by the suggestion that the flames probably do not spread
249: isotropically and uniformly on a rapidly spinning neutron star's
250: surface, we included some qualitative features of the Coriolis force
251: (mentioned in \S~1) in our modeling. For a chosen combination of
252: source parameter values, we started with an already large hot spot
253: (because of initially fast geostrophic flow; \S~1) in a chosen
254: $\phi$-range and $\theta$-range in the low- to mid-latitudes. Most of
255: the hot spot was in the northern hemisphere (through which the
256: observer's line of sight passes), and crossed the equator only
257: slightly. Then we allowed a very narrow $\theta$-portion (of the spot)
258: on the equator to spread in the $\phi$-directions with a speed
259: proportional to $1/\sqrt{\cos\theta}$ (Spitkovsky et al. 2002). Here,
260: we assumed an average $\theta$-value, and kept the $\phi$-direction
261: speed unchanged until a $\phi$-symmetric belt around the equator was
262: formed. This narrow $\phi$-expanding portion was also allowed to
263: simultaneously expand in $\theta$-directions with a speed proportinal
264: to $1/\sqrt{\cos\theta}$, where $\theta$ is the instantaneous polar
265: angle of the northern and southern burning fronts. In addition, the
266: north edge of the original hot spot was allowed to expand northwards
267: with the same speed formula. As a result of such spreading, the hot
268: spot had two portions (see Fig. 3): (1) an equatorial portion that
269: encircled the equator quickly, and expanded north and southwards
270: simultaneously (but progressively at a slower rate); and (2) a
271: $\phi$-asymmetric portion, that moved northwards, and became gradually
272: narrower (in $\theta$-direction) because of the assumed spreading
273: speed formula. The rapid formation of the equatorial belt is expected
274: to cause a fast decay of the amplitude, and subsequently the mid- to
275: high-latitude $\phi$-asymmetric portion should give rise to a lower
276: amplitude that decreases (more slowly) primarily because of the
277: increasing persistent contribution from the expanding $\phi$-symmetric
278: belt. Our model calculation shows this expected evolution (see
279: Fig. 4), which qualitatively remains the same for widely different
280: values of $R/M$, $M$, $\nu_{\rm *}$ $i$, $T_{\rm BB}$, and $n$, and is
281: qualitatively similar to the observed evolution shown in Fig. 1. Here
282: we note that, while the models show the amplitude evolution from the
283: beginning of flame spreading (except the initial very rapid spreading)
284: up to the later part, the data of Fig. 1 may be for an intermediate
285: time phase of spreading. This is because, the observed oscillation
286: power may not be significant (1) in the initial phase due to the low
287: count rate, and (2) in the final phase due to the low
288: amplitude. Finally, although the evolutionary structures of the models
289: of Fig. 2 and the models of Fig. 4 are different, the dependence of
290: amplitude on each source parameter is the same for both sets of
291: models.
292:
293: \section {Discussion and Conclusions} \label{sec: 3}
294:
295: %It is expected that during the intensity rise of an X-ray burst,
296: %thermonuclear flames spread, and the resulting $\phi$-asymmetric,
297: %expanding burning region should cause brightness oscillations with a
298: %frequency close to the neutron star's spin frequency. Therefore, the
299: %evolution of such oscillation properties should be useful to probe the
300: %flame spreading phenomenon.
301: In this Letter, we have, for the first time, reported a detailed
302: comparison among theoretical models of burst oscillation amplitude
303: evolution during burst rise in order to probe the flame spreading
304: phenomenon. Our fitting of burst rise data from two different bursts
305: from two rapidly spinning neutron stars indicate that the oscillation
306: amplitude first decays quickly, and then maintains a non-zero
307: near-constant level. We have shown that such behavior cannot be
308: qualitatively explained with a simple uniform expansion of a circular
309: hot spot. However, we note that if the circular hot spot first
310: expands rapidly, and then, after attaining a large size (covering most
311: of the stellar surface), expands with more than an order of magnitude
312: slower speed, then the observed amplitude curves may be
313: reproduced. But, such a dramatic time dependence of expansion may not
314: be realistic. Inclusion of Coriolis force effects, on the other hand,
315: can provide a natural explanation of the observed amplitude curves.
316: We have included some salient features of these effects (based on the
317: work of Spitkovsky et al. 2002) into our spreading model, and found
318: that these new model amplitude evolutions are qualitatively similar to
319: those observed. This suggests that we may be seeing, for the first
320: time, effects of the Coriolis force on the surface layers of fast
321: spinning neutron stars. This also provides additional motivation to
322: theoretically study thermonuclear flame spreading phenomena
323: considering all the main physical effects. Our modeling also shows
324: that the time evolution of burst oscillation amplitude is sensitive to
325: the nature of flame spreading, while the actual amplitude values
326: strongly depend on some source parameters, such as $i$, $R/M$,
327: etc. Therefore, understanding flame spreading, and constraining the
328: source parameters should be achieveable with detailed modeling of
329: higher signal to noise ratio observations of the burst oscillation
330: sources. Such data may be obtainable with future X-ray missions such
331: as {\it ASTROSAT}, {\it Constellation-X}, {\it XEUS}, or a large-area
332: ``Super-{\it RXTE}'' timing mission.
333:
334:
335: \acknowledgments
336:
337: \begin{thebibliography}{}
338:
339: %\bibitem[]{361} Belian, R. D., Conner, J. P., \& Evans, W. D. 1976, \apj,
340: %206, L135
341:
342: \bibitem[]{363} Bhattacharyya, S. 2007, \mnras, 377, 198
343:
344: \bibitem[]{365} Bhattacharyya, S., Bhattacharya, D., \& Thampan, A. V. 2001,
345: \mnras, 325, 989
346:
347: \bibitem[]{367} Bhattacharyya, S., \& Strohmayer, T. E. 2005, \apj, 634, L157
348:
349: \bibitem[]{369} Bhattacharyya, S., \& Strohmayer, T. E. 2006a, \apj, 636, L121
350:
351: \bibitem[]{371} Bhattacharyya, S., \& Strohmayer, T. E. 2006b, \apj, 641, L53
352:
353: \bibitem[]{373} Bhattacharyya, S., \& Strohmayer, T. E. 2006c, \apj, 642, L161
354:
355: \bibitem[]{375} Bhattacharyya, S., Strohmayer, T. E., Markwardt, C. B., \&
356: Swank, J. H. 2006, \apj, 639, L31
357:
358: \bibitem[]{378} Bhattacharyya, S., Strohmayer, T. E., Miller, M. C., \&
359: Markwardt, C. B. 2005, \apj, 619, 483
360:
361: \bibitem[]{} Chakrabarty, D., Morgan, E. H., Muno, M. P.,
362: Galloway, D. K., Wijnands, R., van der Klis, M., \& Markwardt,
363: Craig B. 2003, Nature, 424, 42
364:
365: \bibitem[]{381} Chandrasekhar, S. 1960, Radiative Transfer, Dover Publications.
366:
367: \bibitem[]{383} Fryxell, B. A., \& Woosley, S. E. 1982, \apj, 261, 332
368:
369: %\bibitem[]{385} Galloway, D. K. et al. 2006 (astro-ph/0608259)
370:
371: %\bibitem[]{387} Grindlay, J. E. et al. 1976, \apj, 205, L127
372:
373: %\bibitem[]{389} Joss, P. C. 1977, \nat, 270, 310
374:
375: \bibitem[]{391} Kaaret, P. et al. 2007, \apj, 657, L97
376:
377: \bibitem[]{393} Lamb, F. K., \& Boutloukos, S. 2007 (arXiv: 0705.0155)
378:
379: \bibitem[]{395} Lamb, D. Q., \& Lamb, F. K. 1978, \apj, 220, 291
380:
381: \bibitem[]{397} Markwardt, C. B., Klein-Wolt, M., Swank, J. H., \& Wijnands, R.
382: 2007, ATel, 1068
383:
384: \bibitem[]{400} Miller, M. C., \& Lamb, F. K. 1998, \apj, 499, L37
385:
386: \bibitem[]{402} Muno, M. P., \"Ozel, F., \& Chakrabarty, D. 2002, \apj, 581,
387: 550
388:
389: \bibitem[]{404} Nath, N. R., Strohmayer, T. E., \& Swank, J. H. 2002, \apj,
390: 564, 353
391:
392: \bibitem[]{406} Spitkovsky, A., Levin, Y. \& Ushomirsky, G. 2002, \apj, 566,
393: 1018
394:
395: \bibitem[]{408} Strohmayer, T. E. 2004, in {\it X-ray Timing 2003: Rossi and
396: Beyond}, AIP Conference Proceedings, Eds. P. Kaaret, F. K. Lamb and
397: J. H. Swank, (Melville, NY: American Institute of Physics), 714, 245
398:
399: \bibitem[]{410} Strohmayer, T. E., \& Bildsten, L. 2006, in {\it Compact
400: Stellar X-ray Sources}, Eds. W.H.G. Lewin and M. van der Klis,
401: (Cambridge University Press: Cambridge), p. 113
402:
403: \bibitem[]{} Strohmayer, T. E., Markwardt, C. B., Swank, J. H., \&
404: in't Zand, J. 2003, \apj, 596, L67
405:
406: \bibitem[]{414} Strohmayer, T. E. et al. 1996, \apj, 469, L9
407:
408: \bibitem[]{416} Strohmayer, T. E., Zhang, W., \& Swank, J. H. 1997, \apj,
409: 487, L77
410:
411: \bibitem[]{} Thompson, T. W. J., Rothschild, R. E., Tomsick, J. A., \&
412: Marshall, H. L. 2005, \apj, 634, 1261
413:
414: \bibitem[]{418} Woosley, S. E., \& Taam, R. E. 1976, \nat, 263, 101
415:
416: \end{thebibliography}{}
417:
418: \clearpage
419: \begin{figure}
420: \vspace{-7.0 cm}
421: \hspace{-2.5 cm}
422: \epsscale{0.8}
423: \plotone{f1.ps}
424: \vspace{-3.0 cm}
425: \caption {Observed evolution of burst oscillation amplitude during the
426: rising portions of two thermonuclear X-ray bursts (solid lines with
427: error bars). Panel {\it a}: for a burst from the LMXB 4U 1636--536
428: (Fig. 1 of Bhattacharyya \& Strohmayer 2005); and panel {\it b}: for a
429: burst from the LMXB SAX J1808.4--3658 (Fig. 1 of Bhattacharyya \&
430: Strohmayer 2006c). The dotted lines give the best-fit constant
431: amplitude levels, and show that the amplitudes evolve significantly.
432: }
433: \end{figure}
434:
435: \clearpage
436: \begin{figure}
437: \vspace{-3.0 cm}
438: \hspace{-1.9 cm}
439: \epsscale{0.5}
440: \plotone{f2.ps}
441: \vspace{0.5 cm}
442: \caption {Theoretical models of burst oscillation amplitude evolution
443: for a uniformly expanding circular hot spot. For each panel, the solid
444: curve is for the parameter values (see \S~2): $R/M = 4.5$, $M = 1.6
445: M_{\odot}$, $\nu_{\rm *} = 582$ Hz, $i = 60^{\rm o}$, $\theta_{\rm c}
446: = 70^{\rm o}$, $T_{\rm BB} = 1.7$ K, and $n = 0$. In each panel, we
447: change one parameter value, while keeping the other parameter values
448: fixed at those for the solid curve. Panel {\it a}: dotted curve: $R/M
449: = 3.8$, dashed curve: $R/M = 5.5$; panel {\it b}: dotted: $M = 1.4
450: M_{\odot}$, dashed: $M = 2.0 M_{\odot}$; panel {\it c}: dotted:
451: $\nu_{\rm *} = 200$ Hz, dashed: $\nu_{\rm *} = 401$ Hz; panel {\it d}:
452: dotted: $i = 30^{\rm o}$, dashed: $i = 80^{\rm o}$; panel {\it e}:
453: dotted: $\theta_{\rm c} = 40^{\rm o}$, dashed: $\theta_{\rm c} =
454: 130^{\rm o}$; panel {\it f}: dotted: $T_{\rm BB} = 1.0$ K, dashed:
455: $T_{\rm BB} = 2.5$ K; and panel {\it g}: dotted: $n = 0.5$, dashed: $n
456: = 0.8$. This figure shows that the qualitative nature of burst
457: oscillation amplitude evolution for a uniformly expanding circular hot
458: spot for any source parameter values is qualitatively different from
459: those observed (see Fig. 1). }
460: \end{figure}
461:
462: \clearpage
463: \begin{figure}
464: \hspace{-1.9 cm}
465: \epsscale{0.5}
466: \plotone{f3.eps}
467: \vspace{-0 cm}
468: \caption {Schematic diagram of flame spreading on a spinning neutron
469: star including salient features of the Coriolis force. We have used
470: this spreading scheme to compute our model evolutionary curves in
471: Fig. 4. The spin axis and the direction of spin are shown. The meshed
472: region represents the hot spot, and the direction and speed of
473: expansion are indicated by the arrows. The flame speed is higher at
474: lower latitudes, as approximately shown by the the length of the
475: arrows. Here, the equatorial belt has already ignited, and the
476: residual $\phi$-asymmetry in the northern hemisphere gives rise to the
477: low amplitude oscillations. }
478: \end{figure}
479:
480: \clearpage
481: \begin{figure}
482: \hspace{-1.9 cm}
483: \epsscale{0.5}
484: \plotone{f4.ps}
485: \vspace{1.0 cm}
486: \caption {Theoretical models of burst oscillation amplitude evolution
487: for a hot spot expanding in a way that captures some features of the
488: Coriolis force (see \S~2; also see Fig. 3). The solid curve is for the
489: parameter values (see \S~2): $R/M = 4.5$, $M = 1.6 M_{\odot}$,
490: $\nu_{\rm *} = 582$ Hz, $i = 60^{\rm o}$, $T_{\rm BB} = 1.7$ K, and $n
491: = 0$. For each of the other curves, we change one parameter value,
492: while keeping the other parameter values fixed at those for the solid
493: curve. Upper dotted curve: $R/M = 3.8$; short-dashed curve: $M = 2.0
494: M_{\odot}$; dash-dot curve: $\nu_{\rm *} = 401$ Hz; lower dotted
495: curve: $i = 30^{\rm o}$; long-dashed curve: $T_{\rm BB} = 2.5$ K; and
496: dash-triple-dot curve: $n = 0.8$. This figure shows that the
497: qualitative nature of amplitude evolution for this model is similar to
498: those observed (see Fig. 1). }
499: \end{figure}
500:
501: \end{document}
502: