1: \documentclass{emulateapj}
2:
3: %\documentclass[12pt,preprint]{aastex}
4:
5: \shorttitle{Pulsation coherence in XTE J1814-338}
6: \shortauthors{Watts, Patruno \& van der Klis}
7:
8: \begin{document}
9:
10: \title{Coherence of burst oscillations and accretion-powered pulsations in the accreting millisecond pulsar XTE J1814-338}
11:
12:
13: \author{Anna L. Watts\altaffilmark{1}, Alessandro Patruno \& Michiel van der Klis}
14: \affil{Astronomical Institute ``Anton Pannekoek'', University of
15: Amsterdam, Kruislaan 403, 1098 SJ Amsterdam, the Netherlands}
16: \altaffiltext{1}{A.L.Watts@uva.nl}
17:
18: \begin{abstract}
19: X-ray timing of the accretion-powered pulsations during the 2003
20: outburst of the accreting millisecond pulsar XTE J1814-338 has revealed
21: variation in the pulse time of arrival residuals. These can be interpreted in several ways, including spin-down and wandering of the fuel impact point around the magnetic pole. In this Letter we show that the
22: burst oscillations of this source are coherent with the
23: persistent pulsations, to the level where they track all of the
24: observed fluctuations. Only one burst, which occurs at
25: the lowest accretion rates, shows a significant phase offset. We
26: discuss what might lead to such rigid phase-locking between the modulations in the accretion and thermonuclear burst emission, and consider the implications for spin variation and
27: the burst oscillation mechanism. Wandering of the fuel impact hot spot
28: around a fixed magnetic pole seems the most likely cause for the accretion-powered
29: pulse phase variations. This means that the burst asymmetry is coupled
30: to the hot spot, not the magnetic pole. If premature ignition at this point (due to higher local
31: temperatures) triggers a burning front that stalls before
32: spreading over the entire surface, the resulting localized nuclear hot
33: spot may explain the unusual burst and burst oscillation properties of this source.
34: \end{abstract}
35:
36: \keywords{binaries: general, stars: individual (XTE J1814-338), stars:
37: neutron, stars: rotation, X-rays: bursts, X-rays: stars }
38:
39: \section{Introduction}
40:
41: The accreting millisecond X-ray pulsars (AMXPs) are a small class of neutron stars in
42: Low Mass X-ray Binaries that show pulsations in outburst,
43: thought to be caused by magnetic channeling of accreting plasma. Detailed timing studies of these stars reveal
44: diverse behavior that can be interpreted in terms of
45: spin variation or shifts in emission pattern. Both processes may play a role, and the
46: degree to which we can be confident in inferred values of
47: spin-up or spin-down remains a hot topic \citep{gal02, bur06, bur07,
48: pap07, pap08, har08, rig08}.
49:
50: Some additional way of verifying the
51: timing analysis obtained from the accretion-powered pulsations would be useful, and in
52: this respect two of the AMXPs are particularly valuable. SAX
53: J1808.4-3658 (J1808) and XTE J1814-338 (J1814)
54: also show thermonuclear-powered
55: pulsations, or burst oscillations. These are high frequency
56: variations seen during Type I X-ray bursts, powered by unstable
57: burning of accreted fuel. In these systems the burst oscillation
58: frequency is at (J1814), or very close to (J1808), the spin frequency
59: (\citealt{cha03, str03}, hereafter S03). The frequency is stable in the decaying tails of the bursts, with
60: no sign of the large frequency drifts seen in other burst oscillation
61: sources \citep{mun02}. In J1814 the frequency is also stable during
62: the rising phase of the bursts, making it the most straightforward
63: candidate for burst oscillation timing (\citealt{wat05}, hereafter
64: W05).
65:
66: J1814's accretion-powered pulsations show significant pulse time of
67: arrival (TOA) residuals even after correction for orbital Doppler
68: shifts (\citealt{pap07}, hereafter P07). The cause is still a matter
69: of debate: P07 interpreted the observations in terms of a steady
70: spin-down coupled with some jitter due to wandering of the fuel impact
71: hotspot around the magnetic pole. However there are other
72: possibilities, such as changes in beaming due to the accretion shock,
73: that may also lead to the observed variation. In this respect
74: analysis of the burst oscillations may be simpler: although the
75: process responsible is not yet understood, the thermal spectrum
76: suggests a purely surface mechanism, with the accretion shock
77: contributing little to the observed asymmetry.
78:
79: Some initial investigation of this issue was carried out by S03, who
80: found that burst oscillations in the first 12 bursts were phase-locked
81: to within 2.5$^\circ$ of the persistent pulsations. Their analysis, however, covered only the
82: first 10 days of the $\approx$ 50 day outburst; before any of the variation reported by
83: P07 is apparent. The level of coherence between the two types of
84: pulsations is also important in our efforts to understand the burst
85: oscillation mechanism, which remains mysterious (see the
86: reviews by \citealt{str06}, \citealt{gal06}). The AMXPs are
87: the only sources in which we can quantify the role of the magnetic
88: field. We want to know, for example, whether the burst oscillations couple to
89: the magnetic field or to the fuel stream impact point.
90:
91:
92:
93: \section{Timing analysis}
94: \label{data}
95:
96:
97: J1814 was discovered in 2003 in the {\it Rossi
98: X-ray Timing Explorer (RXTE)} Galactic bulge monitoring campaign \citep{mar03a}, and
99: remained in outburst for nearly 2 months. The pulsar has a spin
100: frequency of 314.4 Hz and resides in a binary with an orbit of 4.3 hours
101: \citep{mar03}. During the outburst over 425 ks of high time
102: resolution data were taken with {\it RXTE}'s Proportional Counter Array
103: (PCA, \citealt{jah06}). A total of 28 X-ray bursts were detected, all with burst
104: oscillations at the spin frequency. Both accretion-powered pulsations
105: and burst oscillations have a strong overtone at twice the spin
106: frequency (S03).
107:
108: For our timing analysis we use all available pointed observations from
109: the {\it RXTE} PCA with Event mode data (time resolution 122
110: $\mu$s, 64 binned energy channels) or Good Xenon mode data (time resolution 1 $\mu$s, 256 unbinned energy channels). The latter
111: were rebinned in time to 122 $\mu$s time resolution. Data were
112: barycentered using the JPL DE405
113: ephemeris and a spacecraft ephemeris including fine clock
114: corrections which together provide an absolute timing accuracy of 3-4
115: $\mu$s \citep{rot04}, using the source position of
116: \citet{kra05}.
117:
118: In analysing the accretion-powered pulsations we discard the X-ray
119: bursts, removing all data from 50 s before to 200 s after the burst
120: rise, and select only photons in the 2.5-17 keV range to maximize the
121: signal to noise ratio. Pulse profiles
122: are built using the fixed frequency solution of P07
123: to fold segments of approximately 500 s of data
124: after subtracting the background contamination (using the FTOOL
125: {\it{pcabackest}}). The TOAs were then obtained by cross-correlating the folded profiles
126: with a pure sinusoid whose frequency represents the spin frequency of
127: the neutron star. The same procedure is repeated for the first
128: overtone. The fiducial point used in measuring the TOAs was
129: the peak of the sine wave being cross-correlated. The determination of
130: pulse TOAs and their statistical uncertainties follows the
131: standard radio pulsar technique \citep{tay93}. Fitting a
132: Keplerian orbit plus a constant spin frequency $\nu$, or spin
133: derivative $\dot{\nu}$, we obtain solutions consistent with those of
134: P07.
135:
136: Like P07, we find a
137: strong anti-correlation between flux and the accretion-powered
138: pulse TOA residuals. It is interesting, however, that this
139: anti-correlation becomes weaker when one considers the residuals from
140: an ephemeris that includes a constant $\dot{\nu}$ term. A rank
141: correlation test between flux and residuals
142: for a constant frequency model gives a Spearman
143: coefficient of $\rho=-0.71$ (fundamental) and $\rho=-0.90$ (first
144: overtone), with a probability of $< 0.01$\% that
145: the two variables are not anti-correlated. Including spin-down, the
146: magnitudes of the
147: Spearman coefficients fall to $\rho=-0.56$ (probability still $<
148: 0.01$\% that the two variables are not anti-correlated, but larger
149: than the probabilities for the zero spin derivative case) and
150: $\rho=-0.12$ (a probability of 27\% that the two variables are
151: not anti-correlated) respectively. As we will argue in Section \ref{disc}, an
152: accretion-rate dependent hot spot location may be able to explain the entire
153: residual record, with no need for spin variation.
154:
155: We then apply the same timing procedure to the X-ray bursts. As in W05 and
156: \citet{wat06} we use data where the count rate is at least twice
157: the pre-burst level. For the first 27 bursts there is no evidence of frequency
158: variability during the bursts (W05), so we use the entire burst to
159: generate a folded profile. The final burst requires
160: more care, as there is a statistically significant frequency drop in
161: the late stages of the burst rise (Fig.20 of W05). For this burst we use
162: only the first 2 s of the burst rise, before the frequency starts to
163: shift. For each folded burst profile we then compute residuals using
164: the same ephemeris that we used for the accretion-powered pulsations.
165: Again we cross-correlate the folded profile using both a fundamental
166: and an overtone. Figure \ref{mult} shows the TOA residuals for the accretion-powered
167: pulsations and the burst oscillations.
168:
169: \begin{figure}
170: \centering
171: \includegraphics[width=8.5cm, clip]{f1.eps}
172: \caption{Phase residuals for the accretion-powered pulsations (bars; red) and the
173: burst oscillations (crosses; blue), compared to a model that has a
174: constant frequency and orbital Doppler shifts. The final burst,
175: which has no detectable 1st overtone component, requires special
176: treatment; see text. The units on the right axis are {\it rotational}
177: cycles. The top panel shows the outburst light curve. }
178: \label{mult}
179: \end{figure}
180:
181:
182: To check whether the TOAs of the burst oscillations are consistent
183: with having the same temporal dependence as the accretion-powered
184: pulsations we fit a constant frequency model to the two TOA sets
185: separately (excluding the final burst, see below). The fitted frequency is the
186: same within the statistical uncertainties both for the fundamental and
187: the first overtone. We also tried fitting the two TOA sets
188: with a spin frequency plus a frequency derivative. Again the two
189: solutions are consistent within the statistical uncertainties.
190: However P07
191: already noted that both ephemerides are a poor fit to the
192: accretion-powered pulse TOA residuals, and the same is true for the
193: burst oscillation residuals. They
194: are formally inconsistent with a constant frequency model: for the fundamental
195: this assumption gives a $\chi^{2}$ of 459 for 23 degrees of freedom
196: (dof). They are also inconsistent with a constant $\dot{\nu}$, as
197: $\chi^{2}$ in this case is still large, 80 for 22 dof. Similar
198: results are obtained for the first overtone.
199:
200: It is clear from the Figure that, for all but the final burst, the burst
201: oscillation TOA residuals track the accretion pulsation
202: residuals. To test the phase-lock we computed the phase difference $\Delta
203: \phi_m = \nu[{\rm TOA}_{\rm bur} - {\rm TOA}_{\rm acc}]$, using the accretion
204: pulse TOA residuals from the observation containing each burst. We
205: then fitted a constant. For
206: the fundamental we found a $\chi^2$ of 15.3 (26
207: dof), with best fit $\Delta \varphi_m = (0.004 \pm 0.002)$ rotational cycles.
208: For the first overtone we found a $\chi^2$ of 11.3 (26 dof), with
209: best fit $\Delta \varphi_m = -(0.001 \pm 0.003)$ rotational cycles. The fact that a constant
210: $\Delta \varphi_m$ is such a
211: good fit confirms that the two sets of
212: pulsations are phase-locked. The small ($2\sigma$) non-zero offset in the fundamental
213: bears comment. Unlike the burst oscillations, the
214: accretion-powered pulsations have soft lags across the 2.5-20 keV
215: band, with higher energy photons arriving earlier in phase
216: \citep{wat06}. Taking this into account, we find that the burst
217: oscillation TOAs are completely coincident with the softer (2.5 - 5
218: keV) component of the
219: accretion pulsations, thought to originate from
220: stellar surface (Section \ref{disc}). The only exception to the
221: phase-locking is the final burst, with $\Delta \varphi_m = (0.2 \pm
222: 0.04)$ rotational cycles.
223:
224:
225: When calculating burst oscillation residuals some care is required,
226: since if accretion continues during the burst, there might still be a contribution from the accretion-powered
227: pulsations. The resulting bias in the burst
228: oscillation phase can be
229: calculated easily by considering the profile that results from the
230: addition of two offset sinusoidal profiles. Standard trigonometric
231: identities yield a relation between $\Delta \varphi_m$ (the measured
232: offset between burst oscillation phase and accretion-powered
233: pulse phase) and $\Delta \varphi_b$ (the bias, i.e. the offset between
234: measured and true burst
235: oscillation phase caused by residual accretion):
236:
237: \begin{equation}
238: \tan \Delta \varphi_m = \frac{\sin(\Delta\varphi_m -
239: \Delta\varphi_b)}{\left[N_\mathrm{acc} r_\mathrm{acc}/N_\mathrm{bur}
240: r_\mathrm{bur}\right] + \cos(\Delta\varphi_m -
241: \Delta\varphi_b)}.
242: \label{offset}
243: \end{equation}
244: where $N_\mathrm{acc}$ and $N_\mathrm{bur}$ are the number of
245: accretion and burst photons in the folded profile,
246: $r_\mathrm{acc}$ and $r_\mathrm{bur}$ being the associated fractional
247: amplitudes. The quantity $\Delta\varphi_m$ was measured earlier: $\Delta\varphi_m \lesssim 0.01$ cycles
248: at the 99\% confidence level. Using the values of $N_\mathrm{acc}$,
249: $N_\mathrm{bur}$, $r_\mathrm{acc}$ and $r_\mathrm{bur}$ from Table 1
250: of W05, Equation (\ref{offset}) gives a 99\% confidence upper limit on the
251: bias introduced by any residual accretion pulsation of $10^{-3}$
252: cycles. This is sufficiently small that it does not
253: affect our analysis. Similar conclusions can be reached for the
254: overtone. Note that in computing these limits we assume
255: that the
256: accretion flow parameters ($N_\mathrm{acc}, r_\mathrm{acc},
257: \Delta\varphi_m$) are unchanged during a burst. If accretion is
258: inhibited during a burst, due to radiation pressure, the bias will be lower.
259:
260: \section{Discussion}
261: \label{disc}
262:
263: J1814 has unusual burst oscillation properties
264: compared to other sources: they occur in hydrogen-rich bursts, have
265: negligible frequency and amplitude variation, and have a soft spectrum
266: (S03, W05, \citealt{wat06}). Our analysis has shown that the burst
267: oscillations are also phase-locked to the accretion-powered pulsations (to within
268: $3^\circ$ at 99\% confidence) despite the substantial phase-wander
269: exhibited by the latter over the course of the outburst.
270:
271: In fact the burst oscillations are not only phase-locked but also
272: coincident, having the same phase as the soft (2.5-5 keV), lagging, part of the
273: accretion-powered pulsations (although they are also at the 2$\sigma$ level
274: coincident with the entire 2.5-20 keV accretion-powered pulse). Detailed modeling of the accretion
275: pulsations has yet to be done for
276: J1814, but modeling for other AMXPs with similar pulse properties
277: suggests that the soft pulsed component comes from a hot spot on the stellar surface, with the hard component
278: originating in the accretion funnel \citep{pou03, gie05, fal07}.
279:
280: We
281: first consider what the exceptional degree of phase-locking
282: implies about the cause
283: of the variation in the TOA residuals. There
284: are several parts of the system whose variation might affect both types of
285: pulsation: surface rotation, the accretion funnel/disk, the magnetic field, and the fuel deposition
286: footprint.
287:
288: {\it Case 1: Genuine changes in the spin rate of the stellar
289: surface.} Our result would be consistent with a model where all of
290: the variation is due to spin changes, both sets
291: of pulsations being locked to the surface. However this
292: requires alternating spin-up and spin-down with $|\dot{\nu}| \sim
293: 10^{-12}$ Hz/s. Even if the crust were decoupled from the
294: fluid core, this is high compared to what is
295: achievable from the
296: expected accretion or
297: gravitational wave torques \citep{and05, bil98b}. Fitting a constant spin-down term
298: $\dot{\nu} \approx 6\times 10^{-14}$ Hz/s, as argued by P07, does
299: improve the quality of the fits somewhat. However spin derivatives $|\dot{\nu}| \sim
300: 10^{-12}$ Hz/s would still be required on shorter timescales to explain the
301: remaining excursions. Our results would also be consistent with
302: precession, but modeling by \citet{chu08} suggests that
303: precession is unlikely in this source.
304:
305: {\it Case 2: Changes in beaming/scattering by the
306: accretion funnel or disk.}
307: The accretion shock in the funnel is thought to contribute to the pulsed emission of the
308: accretion-powered pulsations, leaving a signature of hard emission in the
309: spectrum. If the funnel were to have a similar effect on the much
310: stronger burst emission it would have to do this without leaving any
311: trace in the spectrum \citep{str03, kra05}. This does not seem
312: feasible. Our result also casts doubt on the accretion disk being
313: the source of the soft lagging component of the accretion-powered
314: pulsations (one of the possibilities considered by \citealt{fal07}), since it is hard to understand why the
315: burst oscillations (a surface process) would track a component
316: generated in the surrounding environs.
317:
318: {\it Case 3: Wander of the magnetic pole, or changes in field
319: geometry.} Motion of the magnetic pole would affect
320: location of the accretion hot spot. If the magnetic field also
321: determines the location of the nuclear burning hot spot through
322: modulation of ignition or emission, this could also explain our
323: result. However, the observed variability would require
324: localized burial or amplification of the poloidal field component on timescales
325: of order a day. The accretion rate in the peak of the outburst is at
326: most a few percent of the Eddington rate \citep{gal04}. Current
327: modeling suggests that this is insufficient to cause burial of the
328: polar
329: field on the required timescales \citep{bro98, cum01, pay04}.
330: There is also no obvious mechanism for field amplification:
331: material arriving via a funnel flow will have almost no angular
332: momentum differential compared to the stellar surface. The heating
333: associated with accretion could bring a buried field to the
334: surface \citep{cum01}, but as previously stated burial is unlikely at
335: such low accretion rates.
336:
337: {\it Case 4: Wander of the fuel deposition point around the magnetic
338: pole.} Simulations of funneled accretion have shown that
339: the fuel deposition point can exhibit phase excursions around
340: a fixed magnetic pole as accretion rate fluctuates, particularly for small
341: misalignment angles between the magnetic pole and the spin axis
342: \citep{rom03, rom04, lam08}\footnote{The pulse profile modeling for this
343: source which has been attempted has suggested large
344: misalignment angles \citep{bha05,
345: lea08}. If this were the case, the observed phase variability would require
346: the fuel impact point to migrate back and forth by several km over the course
347: of the outburst - an uncomfortably large amount. It seems more likely
348: that some of the assumptions in these models, particularly the use
349: of single temperature circular hotspots (known to be problematic,
350: see \citealt{wat06}), need to be revisited.}. Such a model might
351: neatly account for the
352: correlation between residuals and flux without requiring any non-zero $\dot{\nu}$, since the stable position for
353: the fuel impact point will vary in azimuth depending on accretion
354: rate. The fact that the anti-correlation between
355: flux and residuals is stronger when we set $\dot{\nu}=0$ supports this idea.
356:
357: If the most plausible explanation for the variability in the two sets of
358: pulsations is the last, what physical
359: mechanisms might lead to phase-locking between the fuel impact point -
360: which moves
361: with accretion rate relative to field
362: geometry on timescales of a few days - and the nuclear burning hot
363: spots?
364:
365: One possibility is some degree of magnetic
366: confinement, leading to accumulation of fuel at the accretion hot
367: spot. Material deposited near the polar cap will be prevented from spreading until the
368: over-pressure is sufficient to distort the field lines \citep{bro98},
369: even if the impact point is not precisely on the polar cap. To ensure that the magnetic propeller effect does not
370: disrupt accretion, the magnetic field of J1814 must be $\lesssim 10^9$ G
371: \citep{psa99, rap04}. This is consistent with the upper
372: limit on spin-down inferred from pulse timing (P07). At this
373: upper limit helium could be confined until a column depth $\sim 10^6$ g
374: cm$^{-2}$, but this is still well before the material reaches the
375: bursting layer at column depth $\sim 10^8$ g cm$^{-2}$
376: \citep{bil98}. Hydrogen will spread even more easily. Based on these
377: estimates we conclude that fuel confinement is not effective, although
378: it has been advanced as a possible explanation for the short burst
379: recurrence times \citep{gal04}.
380:
381: The other factor distinguishing the fuel impact point is its
382: temperature, which is higher than the rest of the star. The magnitude
383: of the temperature differential has not been
384: determined observationally, but it could certainly affect burst
385: emission. One possibility is that the higher temperatures modify the
386: local composition (via steady burning between bursts, for example).
387: The higher starting temperature and/or composition could in principle
388: modify the flux from the burst once it starts; perhaps driving more
389: energetic reactions or enhancing convection. Whether this effect
390: would be large enough to explain the high fractional amplitudes is,
391: however, not clear.
392:
393: An alternative is the effect that a higher local temperature would
394: have on ignition conditions. Previous studies that concluded that
395: ignition would occur predominantly near the equator \citep{spi02,
396: coo07} did not consider the effect of non-uniformities in
397: temperature. A small increase in temperature can have
398: a large effect on the column depth required for ignition, with
399: the hotter area requiring a lower column depth \citep{bil98}.
400: Material at the fuel impact point could therefore reach
401: ignition conditions well ahead of the rest of the star. In this scenario
402: the burning front might stall before spreading across the whole star,
403: depending on the rate of heat transfer across the burning front. This
404: would result in a brightness asymmetry centered on the fuel impact point.
405: Premature ignition at the fuel deposition point, followed by stalling, could explain
406: several observational features: the rather small black body radius of
407: the bursts
408: \citep{gal06}, the shorter than expected burst recurrence times
409: \citep{gal04}, and the burst shapes, which suggest off-equatorial
410: ignition \citep{mau08}. Note that off-equatorial ignition alone is not
411: sufficient to explain the presence of an asymmetry in the burst tail:
412: something else, such as stalling, is required.
413:
414: Whatever the mechanism, it must fail once the accretion rate
415: drops (or perhaps when the burst is more energetic),
416: since the oscillations in the final burst are offset. The fuel impact
417: footprint is by this time thought to be smaller, since the fractional
418: amplitude of the accretion-powered pulsations is
419: rising (W05). In addition the accreted fuel has more time
420: to spread and equilibrate between bursts. Both factors will act to
421: reduce the temperature differential between the fuel stream impact point and the rest of
422: the star, which would make ignition at the fuel impact point less likely. The oscillation
423: properties of this final burst (substantially lower fractional
424: amplitude, frequency drift) are very different to the rest of
425: the sample, and it is quite plausible that a different burst
426: oscillation mechanism operates in this final burst.
427:
428: \acknowledgments
429: We thank Mariano Mendez and the participants of the workshop ``A
430: Decade of Accreting
431: Millisecond X-ray Pulsars'' for the lively discussions that prompted
432: and informed
433: this work.
434:
435:
436: \begin{thebibliography}{99}
437:
438: \bibitem[\protect\citeauthoryear{Andersson et al.}{2005}]{and05}
439: Andersson N. et al., 2005, MNRAS, 361, 1153
440:
441: %Constraints on Neutron Star Parameters from Burst Oscillation Light Curves of the Accreting Millisecond Pulsar XTE J1814-338
442: \bibitem[\protect\citeauthoryear{Bhattacharyya et al.}{2005}]{bha05}
443: Bhattacharyya S. et al., 2005, ApJ, 619, 483
444:
445: \bibitem[\protect\citeauthoryear{Bildsten}{1998a}]{bil98b}
446: Bildsten L., 1998, ApJ, 501, L89
447:
448:
449: \bibitem[\protect\citeauthoryear{Bildsten}{1998b}]{bil98}
450: Bildsten L., 1998, in The Many Faces of Neutron Stars, eds Buccheri et
451: al., Kluwer Academic Publishers, p.419
452:
453: %The Ocean and Crust of a Rapidly Accreting Neutron Star: Implications for Magnetic Field Evolution and Thermonuclear Flashes
454: \bibitem[\protect\citeauthoryear{Brown \& Bildsten}{1998}]{bro98}
455: Brown E.F., Bildsten L., 1998, ApJ, 496, 915
456:
457: %Order in the Chaos: Spin-up and Spin-down during the 2002 Outburst of SAX J1808.4-3658
458: \bibitem[\protect\citeauthoryear{Burderi et al.}{2006}]{bur06}
459: Burderi L. et al., 2006, ApJ, 653, L133
460:
461:
462: %Timing an Accreting Millisecond Pulsar: Measuring the Accretion Torque in IGR J00291+5934
463: \bibitem[\protect\citeauthoryear{Burderi et al.}{2007}]{bur07}
464: Burderi L. et al., 2007, ApJ, 657, 961
465:
466:
467: %Nuclear-powered millisecond pulsars and the maximum spin frequency of neutron stars
468: \bibitem[\protect\citeauthoryear{Chakrabarty et al.}{2003}]{cha03}
469: Chakrabarty D. et al., 2003, Nature, 424, 42
470:
471: \bibitem[\protect\citeauthoryear{Chung et al.}{2008}]{chu08}
472: Chung C. et al., MNRAS in press, eprint arXiv:0808.3820
473:
474: \bibitem[\protect\citeauthoryear{Cooper \& Narayan}{2007}]{coo07}
475: Cooper R.L., Narayan R., 2007, ApJ, 657, L29
476:
477:
478: %Magnetic Screening in Accreting Neutron Stars
479: \bibitem[\protect\citeauthoryear{Cumming et al.}{2001}]{cum01}
480: Cumming A. et al., 2001, ApJ, 557, 958
481:
482:
483: %Energy-dependent 100 microsecond time lags as observational evidence of Comptonization effects in the neutron star plasma environment
484: \bibitem[\protect\citeauthoryear{Falanga \& Titarchuk}{2007}]{fal07}
485: Falanga M., Titarchuk L., 2007, ApJ, 661, 1084
486:
487: \bibitem[\protect\citeauthoryear{Galloway et al.}{2002}]{gal02}
488: Galloway D.K. et al., 2002, ApJ, 576, L137
489:
490: \bibitem[\protect\citeauthoryear{Galloway et al.}{2004}]{gal04}
491: Galloway D.K. et al., 2004, BAAS, 36, 954
492:
493: %Thermonuclear (type-I) X-ray bursts observed by the Rossi X-ray Timing Explorer
494: \bibitem[\protect\citeauthoryear{Galloway et al.}{2006}]{gal06}
495: Galloway D.K. et al., ApJS in press, astro-ph/0608259
496:
497: %Physics of accretion in the millisecond pulsar XTE J1751-305
498: \bibitem[\protect\citeauthoryear{Gierli\'nski \& Poutanen}{2005}]{gie05}
499: Gierli\'nski M., Poutanen J., 2005, MNRAS, 359, 1261
500:
501:
502: %The Long-Term Evolution of the Spin, Pulse Shape, and Orbit of the Accretion-powered Millisecond Pulsar SAX J1808.4-3658
503: \bibitem[\protect\citeauthoryear{Hartman et al.}{2008}]{har08}
504: Hartman J.M. et al., 2008, ApJ, 675, 1468
505:
506: %Calibration of the Rossi X-Ray Timing Explorer Proportional Counter Array
507: \bibitem[\protect\citeauthoryear{Jahoda et al.}{2006}]{jah06}
508: Jahoda K. et al., 2006, ApJSS, 163, 401
509:
510:
511: %
512: \bibitem[\protect\citeauthoryear{Krauss et al.}{2005}]{kra05}
513: Krauss M.I. et al., 2005, ApJ, 627, 910
514:
515: \bibitem[\protect\citeauthoryear{Lamb et al.}{2008}]{lam08}
516: Lamb F.K. et al., eprint arXiv:0808.4159
517:
518: \bibitem[\protect\citeauthoryear{Leahy et al.}{2008}]{lea08}
519: Leahy D.A. et al., eprint arXiv:0806.0824
520:
521: \bibitem[\protect\citeauthoryear{Markwardt \& Swank}{2003}]{mar03a}
522: Markwardt C.B., Swank J.H., 2003, IAU Circ. 8144, 1
523:
524: %
525: \bibitem[\protect\citeauthoryear{Markwardt et al.}{2003}]{mar03}
526: Markwardt C.B. et al., 2003, Atel 164.
527:
528:
529: %Ignition latitude and the shape of Type I X-ray bursts
530: \bibitem[\protect\citeauthoryear{Maurer \& Watts}{2008}]{mau08}
531: Maurer I., Watts A.L., 2008, MNRAS, 383, 387
532:
533: \bibitem[\protect\citeauthoryear{Muno et al.}{2002}]{mun02}
534: Muno M.P. et al., 2002, ApJ, 580, 1048
535:
536:
537: \bibitem[\protect\citeauthoryear{Payne \& Melatos}{2004}]{pay04}
538: Payne D.J.B., Melatos A., 2004, MNRAS, 351, 569
539:
540:
541: %Timing of the accreting millisecond pulsar XTE J1814-338
542: \bibitem[\protect\citeauthoryear{Papitto et al.}{2007}]{pap07}
543: Papitto A. et al., 2007, MNRAS, 375, 971
544:
545:
546: %Measuring the spin-up of the accreting millisecond pulsar XTE J1751-305
547: \bibitem[\protect\citeauthoryear{Papitto et al.}{2008}]{pap08}
548: Papitto A. et al., 2008, MNRAS, 383, 411
549:
550: %On the nature of the X-ray emission from the accreting millisecond pulsar SAX J1808.4-3658
551: \bibitem[\protect\citeauthoryear{Poutanen \& Gierli\'nski}{2003}]{pou03}
552: Poutanen J., Gierli\'nski M., 2003, MNRAS, 343, 1301
553:
554:
555: \bibitem[\protect\citeauthoryear{Psaltis \& Chakrabarty}{1999}]{psa99}
556: Psaltis D., Chakrabarty D., 1999, ApJ, 521, 332
557:
558:
559: %Accretion onto Fast X-Ray Pulsars
560: \bibitem[\protect\citeauthoryear{Rappaport et al.}{2004}]{rap04}
561: Rappaport S.A. et al., 2004, ApJ, 606, 436
562:
563:
564: %Spin up and phase fluctuations in the timing of the accreting millisecond pulsar XTE J1807-294
565: \bibitem[\protect\citeauthoryear{Riggio et al.}{2008}]{rig08}
566: Riggio A. et al., 2008, ApJ, 678, 1273
567:
568:
569: %3D Simulations of Disk Accretion to an Inclined Dipole. I. Magnetospheric Flows at Different Î
570: \bibitem[\protect\citeauthoryear{Romanova et al.}{2003}]{rom03}
571: Romanova M.M. et al., 2003, ApJ, 595, 1009
572:
573:
574: %3D sims to inclined dipole: hotspots and variability
575: \bibitem[\protect\citeauthoryear{Romanova et al.}{2004}]{rom04}
576: Romanova M.M. et al., 2004, ApJ, 610, 920
577:
578:
579: %Absolute Timing of the Crab Pulsar with the Rossi X-Ray Timing Explorer
580: \bibitem[\protect\citeauthoryear{Rots et al.}{2004}]{rot04}
581: Rots A.H. et al., 2004, ApJ, 605, L129
582:
583:
584: \bibitem[\protect\citeauthoryear{Spitkovsky et al.}{2002}]{spi02}
585: Spitkovsky A. et al., 2002, ApJ, 566, 1018
586:
587:
588: %X-ray bursts from the accreting millisecond pulsar XTE J1814-338
589: \bibitem[\protect\citeauthoryear{Strohmayer et al.}{2003}]{str03}
590: Strohmayer T.E. et al., 2003, ApJ, 596, L67
591:
592:
593: \bibitem[\protect\citeauthoryear{Strohmayer \& Bildsten}{2006}]{str06}
594: Strohmayer T., Bildsten L., 2006, in Compact stellar X-ray sources,
595: eds Lewin W., van der Klis M., Cambridge Astrophysics Series 39,
596: Cambridge, UK, p.113
597:
598: \bibitem[\protect\citeauthoryear{Taylor}{1993}]{tay93}
599: Taylor J.H., 1993, in Pulsars as Physics Laboratories, ed. R.D.Blandford (New York: Oxford Univ. Press), 117
600:
601:
602: %Analysis of variability in the burst oscillations of the accreting millisecond pulsar XTE J1814-338
603: \bibitem[\protect\citeauthoryear{Watts et al.}{2005}]{wat05}
604: Watts A.L. et al., 2005, ApJ, 634, 547
605:
606:
607: %The energy dependence of burst oscillations from the accreting millisecond pulsar XTE J1814-338
608: \bibitem[\protect\citeauthoryear{Watts \& Strohmayer}{2006}]{wat06}
609: Watts A.L., Strohmayer T.E., 2006, MNRAS, 373, 769
610:
611:
612: \end{thebibliography}
613:
614: \end{document}
615: