1: % paper on RXTE obs of XTE J1739-285
2:
3: %\documentclass[12pt,preprint]{aastex} % for ApJ submission
4: %\documentclass[preprint2]{aastex} % for preprints
5: \documentclass{emulateapj} % for ApJ look
6: \usepackage{times}
7:
8: \shorttitle{Millisecond Variability from XTE J1739-285}
9: \shortauthors{Kaaret et al.}
10:
11: \begin{document}
12:
13: \title{Evidence for 1122~Hz X-Ray Burst Oscillations from the
14: Neutron-Star X-Ray Transient XTE J1739-285}
15:
16: \author{P.\ Kaaret\altaffilmark{1}, Z.\ Prieskorn\altaffilmark{1},
17: J.J.M.\ in 't Zand\altaffilmark{2}, S.\ Brandt\altaffilmark{3}, N.\
18: Lund\altaffilmark{3}, S.\ Mereghetti\altaffilmark{4}, D.\
19: G\"{o}tz\altaffilmark{5}, E.\ Kuulkers\altaffilmark{6}, J.A.\ Tomsick
20: \altaffilmark{7,8}}
21:
22: \altaffiltext{1}{Department of Physics and Astronomy, University of
23: Iowa, Iowa City, IA, 52242; philip-kaaret@uiowa.edu,
24: zachary-prieskorn@uiowa.edu}
25:
26: \altaffiltext{2}{SRON Netherlands Institute for Space Research,
27: Sorbonnelaan 2, 3584 CA Utrecht, the Netherlands;
28: J.J.M.in.t.Zand@sron.nl}
29:
30: \altaffiltext{3}{Danish National Space Center, Juliane Maries Vej 30,
31: DK-2100 Copenhagen, Denmark; sb@dsri.dk, nl@dsri.dk}
32:
33: \altaffiltext{4}{INAF - Istituto di Astrofisica Spaziale e Fisica
34: Cosmica, Milano, via Bassini 15, 20133 Milano, Italy;
35: sandro@iasf-milano.inaf.it}
36:
37: \altaffiltext{5}{CEA - Saclay, DSM/DAPNIA/Service d'Astrophysique, Orme
38: des Merisiers, Bat.\ 709, F-91191 Gif-sur-Yvette, France;
39: diego@mi.iasf.cnr.it}
40:
41: \altaffiltext{6}{SOC, ESA/ESAC, Urb.\ Villafranca del Castillo, PO Box
42: 50727, 28080 Madrid, Spain; ekuulker@rssd.esa.int}
43:
44: \altaffiltext{7}{Space Sciences Laboratory, 7 Gauss Way, University of
45: California, Berkeley, CA 94720; jtomsick@ssl.berkeley.edu}
46:
47: \altaffiltext{8}{Center for Astrophysics and Space Sciences, Code 0424,
48: 9500 Gilman Drive, University of California at San Diego, La Jolla, CA
49: 92093}
50:
51:
52: \begin{abstract}
53:
54: We report on millisecond variability from the X-ray transient XTE
55: J1739-285. We detected six X-ray type I bursts and found evidence for
56: oscillations at $1122\pm 0.3$~Hz in the brightest X-ray burst. Taking
57: into consideration the power in the oscillations and the number of
58: trials in the search, the detection is significant at the 99.96\%
59: confidence level. If the oscillations are confirmed, the oscillation
60: frequency would suggest that XTE J1739-285 contains the fastest
61: rotating neutron star yet found. We also found millisecond
62: quasiperiodic oscillations in the persistent emission with frequencies
63: ranging from 757~Hz to 862~Hz. Using the brightest burst, we derive an
64: upper limit on the source distance of about 10.6~kpc.
65:
66: \end{abstract}
67:
68: \keywords{accretion, accretion disks --- gravitation --- relativity ---
69: stars: individual (XTE J1739-285) --- stars: neutron --- X-rays:
70: stars}
71:
72:
73: \section{Introduction}
74:
75: Weakly magnetized neutron stars can be spun up to rates of several
76: 100~Hz by accretion in low-mass X-ray binaries \citep{Alpar82}. The
77: first direct measurements of millisecond spin rates in actively
78: accreting neutron stars in low-mass X-ray binaries (LMXBs) came from
79: the discovery of oscillations in thermonuclear X-ray bursts occurring
80: on the neutron star surface \citep{Strohmayer96}. The burst
81: oscillation frequencies were later found to be nearly equal to those of
82: coherent pulsations
83: \citep{intZand01,Strohmayer02,Chakrabarty03,Strohmayer03} implying that
84: the burst oscillation frequency indicates the neutron star spin rate.
85: This X-ray technique has no known biases against the detection of very
86: high spin rates, unlike radio pulsation searches, and the sample of
87: X-ray burst oscillation frequencies has been exploited to constrain the
88: neutron star spin rate distribution. Analysis of a sample of 11 X-ray
89: measured spin frequencies in the range 270-619~Hz suggests a limiting
90: spin rate near 760~Hz if the distribution is uniform and bounded
91: \citep{Chakrabarty03}. This is below the expected maximum spin
92: frequency possible without centrifugal breakup and has been interpreted
93: as evidence that some physical process, possibly gravitational
94: radiation, limits the maximum possible spin rate. However, the
95: discovery of a radio pulsar spinning at 716~Hz, a frequency above any
96: previously measured in X-rays, suggests that the true maximum spin rate
97: is higher \citep{Hessels06}.
98:
99: Here, we describe observations made with the Rossi X-Ray Timing
100: Explorer (RXTE; Bradt, Rothschild, \& Swank 1993) following the
101: detection of X-ray bursts from the transient source XTE J1739-285 as
102: part of a program to search for millisecond oscillations in both X-ray
103: bursts and persistent emission from neutron star X-ray binaries which
104: are newly discovered or found to be active
105: \citep{Kaaret02,Kaaret03,Kaaret06}. We detected six X-ray bursts and
106: found oscillations at a frequency of 1122~Hz in the brightest burst.
107: This suggests that XTE J1739-285 contains the most rapidly rotating
108: neutron star yet discovered. We describe our observations in \S 2, the
109: X-ray bursts in \S 3, and the persistent emission and the discovery of
110: kHz QPOs from the source in \S 4. We discuss the results in \S 5.
111:
112:
113: \section{Observations of XTE J1739-285}
114:
115: XTE J1739-285 is a transient neutron-star low-mass X-ray binary
116: (NS-LMXB) that was discovered during RXTE PCA scans of the Galactic
117: bulge on 1999 October 19 \citep{Markwardt99} and underwent short
118: outbursts in May 2001 and October 2003. The source became active again
119: in August 2005 \citep{Bodaghee05} and two X-ray bursts were detected
120: with the JEM-X instrument on INTEGRAL on 2005 September 30 and October
121: 4 \citep{Brandt05}.
122:
123: Triggered by the detection of the X-ray bursts, we obtained 19
124: observations using RXTE in the period beginning 2005 October 12 and
125: ending 2005 November 16. Data were obtained with the Proportional
126: Counter Array (PCA) in a spectral mode (Standard 2) with 256 energy
127: channels and 16~s time resolution, a low-resolution timing mode
128: (Standard 1) with no energy information and 0.125~s time resolution,
129: and a high-resolution timing mode (Event mode) with 122~$\mu$s time
130: resolution and 64 energy channels \citep{Jahoda06}.
131:
132:
133: \section{X-Ray Bursts}
134:
135: We searched the Standard 1 data for X-ray bursts and found six, see
136: Table~\ref{bursttable}. We examined their evolution by extracting
137: spectra for 0.25~s intervals of event mode data using all Proportional
138: Counter Units (PCUs) that were on during each burst and all layers.
139: The fluxes were corrected for dead time effects with a maximum
140: correction of 5.5\%. To remove the contribution of the persistent
141: emission, we subtracted off a spectrum from 10~s of data preceding each
142: burst. Spectra were fitted in the 3-18~keV band (channels 3-24 in the
143: event mode data) with an absorbed blackbody model with the column
144: density fixed to $N_H = 7.5 \times 10^{21} \rm \, cm^{-2}$ which is the
145: hydrogen column density along the line of sight in the Milky Way
146: \citep{Dickey90}. We found a burst-like event at Nov 11 10:27:22 UTC
147: which appears in only one of four PCUs on at the time and is likely a
148: detector breakdown event and not an X-ray burst.
149:
150: \begin{deluxetable}{cccc}
151: \tablecolumns{3}
152: \tabletypesize{\scriptsize}
153: \tablecaption{Properties of X-Ray Bursts
154: \label{bursttable}}
155: \tablewidth{0pt}
156: \tablehead{\colhead{\#} & \colhead{Time} &
157: \colhead{Peak flux} & \colhead{Decay}}
158: \startdata
159: 1 & Oct 31 07:59:09 & $1.6 \pm 0.3$ & 5.0 \\
160: 2 & Nov 4 11:34:23 & $2.8 \pm 0.5$ & 5.4 \\
161: 3 & Nov 7 05:34:04 & $2.2 \pm 0.4$ & 5.0 \\
162: 4 & Nov 7 07:30:53 & $1.1 \pm 0.3$ & 3.0 \\
163: 5 & Nov 8 08:19:08 & $1.0 \pm 0.2$ & 3.1 \\
164: 6 & Nov 11 09:47:10 & $1.2 \pm 0.3$ & 5.7 \\
165: \enddata
166:
167: \tablecomments{The table lists for each burst: Time -- the time (UTC)
168: at the start of each burst (all bursts occurred in 2005); Peak flux --
169: the bolometric peak flux corrected for absorption in units of $10^{-8}
170: \rm \, erg \, cm^{-2} \, s^{-1}$; and Decay -- the decay ($e$-folding)
171: time in seconds.} \vspace{0.2in}
172: \end{deluxetable}
173:
174: The brightest burst is \# 2, which had a peak bolometric flux, as
175: measured in 0.25~s intervals, of $(2.8 \pm 0.5) \times 10^{-8} \rm \,
176: erg \, cm^{-2} \, s^{-1}$. The blackbody temperature rose rapidly
177: during the flux rise, reached a maximum near 2.5~keV, and then decayed
178: as the flux decayed. This behavior is indicative of heating during the
179: rise and cooling during the decay and is characteristic of type I X-ray
180: bursts. \citet{Kuulkers03} found that the empirical maximum bolometric
181: peak luminosities for photospheric radius expansion bursts from a
182: sample of 12 bursters located in globular clusters with known distances
183: was $3.8 \times 10^{38} \rm \, erg \, s^{-1}$, with an accuracy of
184: $\sim$15\%. None of our bursts showed evidence of photospheric radius
185: expansion and, thus, their luminosities should be below this value.
186: The peak flux of the brightest burst implies an upper limit on the
187: distance to XTE J1739-285 of about 10.6~kpc. This is consistent with,
188: but somewhat closer than, the upper limit of 12~kpc from
189: \citet{Torres06}.
190:
191: Bursts 1 and 4-6 have similar peak fluxes, while bursts 2 and 3 were
192: brighter, see Table~\ref{bursttable}. Burst 3 had a rise time near
193: 3~s, while the other rise times were near 1~s. Burst 6 had a
194: relatively broad maximum extending over 5~s. The decay ($e$-folding)
195: times are in the range 3--6~s. The fact that the bursts have fast
196: rises and durations of tens of seconds suggests that they are typical
197: helium bursts.
198:
199: \begin{figure}
200: \centerline{\includegraphics[width=3.2in]{f1.eps}}
201: \caption{Dynamical power spectrum of burst 2 with the burst light curve
202: (count rate) and the power at 1122.0~Hz superposed. The contours are at
203: Leahy powers of 12, 20, 28, and 36. The contours are generated from
204: power spectra for overlapping 4~s intervals of data with the power
205: plotted at the midpoint of the interval. The curve which peaks near
206: 2~s is the rate in the 3-18~keV band from all the PCUs on during the
207: observation (PCU0 and PCU2). The curve which peaks near 15~s is the
208: Leahy power at 1122.0~Hz. The scales for both curves are on the right
209: side. The oscillation is significant at the 99.96\% confidence level.}
210: \label{burst2dyn} \end{figure}
211:
212: Using high time resolution data, we computed power spectra for
213: overlapping 4~s intervals with 0.125~s between the starts of successive
214: intervals using events in PHA channels 4-24 for PCU0 and channels 3-24
215: all other PCUs. The background, particularly at low energies, is
216: higher in PCU0 due to the loss of its propane layer. We searched over
217: the duration of each burst defined as the interval where the total
218: counting rate exceeds the background plus persistent counting rate,
219: measured a few seconds before the burst, by at least a factor of 2. We
220: searched for excess power in the range 20-2000~Hz, corresponding to
221: 7920 frequency bins in the 4~s FFTs. We found oscillations in the
222: second burst with a Leahy normalized power \citep{Leahy83} of 42.82 at
223: a frequency of $1122 \pm 0.3$~Hz, occurring in the burst decay 15~s
224: after the burst rise, see Fig.~\ref{burst2dyn}. The four consecutive
225: FFTs around the one with the peak power all have powers above 40.31 at
226: 1122.0~Hz.
227:
228: To evaluate the significance of the signal, we used a Monte Carlo
229: simulation which generates Poisson distributed events following the
230: light curve from burst 2 in 0.125~s bins after smoothing with a moving
231: average over 9 bins. The deadtime of the PCA is modeled by removing
232: any event which occurs within 10~$\mu$s after a previous event
233: \citep{Jahoda06}. The number of events generated in each time bin is
234: larger than the observed counts so that after the deadtime correction
235: the number matches that in the actual light curve within Poisson
236: fluctuations. We generated 400,000 trial bursts. Each simulated burst
237: was analyzed using exactly the same analysis done on the real data,
238: specifically by calculating 4~s FFTs at overlapping intervals with
239: starts each 0.125 seconds and searching for power in the 20-2000~Hz
240: interval. The chance probability of occurrence of the observed signal
241: is calculated by counting the fraction of trial bursts with powers
242: equal to or exceeding those observed. This procedure eliminates the
243: ambiguity in estimating the equivalent number of independent trials for
244: the overlapping FFTs. We found 58 bursts with at least one power above
245: 42.82 of which 14 had at least 4 consecutive powers above 40.31 at the
246: same frequency. We estimate the chance probability of occurrence of
247: the observed signal to be $3.5\times 10^{-5}$. Allowing two trials for
248: two energy bands, the probability is $7.0\times 10^{-5}$ equivalent to
249: a $3.97\sigma$ confidence level.
250:
251: The signal at 1122~Hz was detected in the brightest burst. Since the
252: bursts are not a uniform sample and the presence of oscillations may
253: depend on burst properties and/or accretion history (e.g.\ Watts,
254: Strohmayer, Markwardt 2005), one can reasonably argue that the
255: significance should be evaluated for each burst individually,
256: particularly in this case since the brightest burst is the most likely
257: to give a detectable signal. However, a more conservative approach is
258: to consider the full set of six bursts. The durations of the bursts,
259: as defined above, are 21, 21, 22, 10, 11, and 18~s for bursts 1--6,
260: respectively. To account for all the trials in all the bursts, we
261: multiply the number of trials by an additional factor of 103/21 = 4.9
262: for a chance probability of $3.4 \times 10^{-4}$. The oscillation is
263: detected at a confidence level of 99.966\%, equivalent to a $3.6\sigma$
264: significance.
265:
266: \begin{figure}
267: \centerline{\includegraphics[width=3.0in,angle=0]{f2.eps}}
268: \caption{Power at 1122~Hz in independent 1~s intervals (crosses). The
269: solid line is the count rate in the same 1~s intervals. The powers at
270: 1122~Hz used in the significance calculation are, in time order
271: starting at 12.5~s, 3.90, 9.51, 9.34, 15.37, and 9.67. We note that
272: the power at 17.5~s at 1123~Hz is not shown on the plot and is 7.94.}
273: \label{burst2i} \end{figure}
274:
275: To obtain an estimate of the significance of the oscillations which
276: does not depend on a simulation, we calculated FFTs in independent 4~s,
277: 2~s, and 1~s intervals of data from burst 2. There are five successive
278: 1~s FFTs with power at 1122 Hz, see Fig.~\ref{burst2i}. The single
279: trial probability for having five powers at or above the observed
280: levels in the 1~s FFTs is $4.2 \times 10^{-11}$. Accounting for all
281: sequential combinations of FFTs of fixed length of 1, 2, or 4~s with a
282: total duration of 5~s or less covering the 20-2000~Hz range in two
283: energy bands, we estimate a total of 605880 trials:
284: (17+18+19+20+21)$\times$1980$\times$2 + (9+10)$\times$3960$\times$2 +
285: 5$\times$7920$\times$2. The chance probability taking into account the
286: number of trials in the one burst is then $2.5 \times 10^{-5}$
287: ($4.2\sigma$). Accounting for the other bursts, the chance probability
288: is $1.2\times 10^{-4}$ ($3.8\sigma$). This probability agrees within a
289: factor of 3 with that in the previous paragraph.
290:
291: In the 4~s interval with the maximum power, the fractional rms
292: amplitude of the 1122.0~Hz oscillation is $0.13 \pm 0.02$. The quality
293: factor, $Q = \nu/\Delta\nu$, of the oscillation in burst 2 is $Q >
294: 1000$. This is consistent with the values found in burst oscillations
295: and inconsistent with the values found in high frequency QPOs in the
296: persistent emission. We searched for oscillations near half the
297: frequency. The strongest signal has a Leahy power of 13.6 at 561.25~Hz
298: and occurs 7.75~s after the burst rise. This is not a significant
299: signal; the chance probability of occurrence within the narrow window
300: searched, 562.5-558.5~Hz, is 0.09. At the time of the maximum strength
301: of the 1122~Hz signal, the Leahy power near 561~Hz is less than 6.
302:
303:
304: \section{Persistent emission}
305:
306: To study the persistent emission spectra, we used Standard-2 data from
307: PCU 2, removing data around the X-ray bursts, and estimating the
308: background using bright source background files. The source was in a
309: relatively hard state, with a hard X-ray color, $HC$, defined as the
310: count rate in the 9.7--16~keV band divided by the 6.0--9.7~keV rate, of
311: $HC > 0.35$ for all observations before October 25. The flux was
312: roughly constant and near $4 \times 10^{-10} \rm \, erg \, cm^{-2} \,
313: s^{-1}$ in the 2-20~keV band. The spectra were adequately described by
314: the absorbed ($N_H = 7.5 \times 10^{21} \rm \, cm^{-2}$) sum of a
315: power-law with a photon index near 2 and a gaussian emission line with
316: a centroid of 6.7~keV and a flux consistent with that from Galactic
317: ridge emission at the source position \citep{Revnivtsev06}. After
318: October 25, the source stayed in a softer state, $HC < 0.35$. The flux
319: varied by a factor of two, with the maximum near $1.4 \times 10^{-9}
320: \rm \, erg \, cm^{-2} \, s^{-1}$ in the 2-20~keV band. The spectra
321: showed distinct curvature at high energies and were adequately fitted
322: with the absorbed sum of a Comptonization model (compst in xspec) with
323: a temperature in the range 4--12~keV and an optical depth in the range
324: 3--10, a blackbody with a temperature near 1.6~keV, and a gaussian
325: emission line with a centroid of 6.7~keV, a width less than 0.9~keV,
326: and an equivalent width of 100-300~eV, in some cases larger than the
327: expected Galactic ridge emission. The blackbody component was not
328: required for some lower flux observations. The spectra are similar to
329: those of other moderate luminosity NS-LMXBs \citep{Kaaret02,Kaaret03}.
330:
331: \begin{deluxetable}{lcccc}
332: \tablecolumns{5}
333: \tabletypesize{\scriptsize}
334: \tablecaption{High frequency peaks in the persistent emission
335: \label{qpotable}}
336: \tablewidth{0pt}
337: \tablehead{
338: \colhead{Time} & \colhead{Centroid} &
339: \colhead{Width} & \colhead{Amplitude} \\
340: \colhead{(UTC)} & \colhead{(Hz)} &
341: \colhead{(Hz)} & \colhead{(\%)} \\
342: }
343: \startdata
344: Oct 31 06:22:11 & 837.9$\pm$0.3 & 5.3$\pm$0.9 & 1.7$\pm$0.2 \\
345: Nov 01 02:48:14 & 784.6$\pm$1.1 & 14$\pm$3 & 2.4$\pm$0.4 \\
346: Nov 08 04:37:47 & 814.4$\pm$3.2 & 36$\pm$9 & 3.1$\pm$0.6 \\
347: Nov 08 09:46:31 & 756.7$\pm$0.7 & 9$\pm$3 & 2.1$\pm$0.4 \\
348: Nov 10 08:30:35 & 861.8$\pm$0.7 & 13$\pm$2 & 2.7$\pm$0.3 \\
349: Nov 10 10:04:43 & 846.0$\pm$2.9 & 33$\pm$8 & 2.7$\pm$0.5 \\
350: \enddata
351: \tablecomments{The table includes: Time -- the UTC time at the
352: beginning of the observation, all observations were in 2005; Centroid
353: and Width - of the fitted Lorentzian; Amplitude - RMS fractional
354: amplitude of the fitted Lorentzian.}
355: \end{deluxetable}
356:
357: To further investigate the source state, we produced low frequency
358: power spectra using event mode data from all PCUs on during each
359: observation. The sum power spectrum for observations with $HC > 0.35$
360: shows very-low-frequency noise (VLFN) and so-called high-frequency
361: noise (HFN). The total noise power in the 0.01-100~Hz band is 1.0\%
362: fractional rms. We fit the power spectrum with a model consisting of a
363: power-law to describe the VLFN and an exponentially cutoff power-law
364: for the HFN \citep{vdk95}. The best fit parameters are a power law
365: index for the VLFN of $1.81 \pm 0.13$ and a power law index of $0.8 \pm
366: 0.1$ and a cutoff frequency in the range of 20-60~Hz for the HFN. The
367: weakness of the signal at high frequencies prevents an accurate
368: determination of the HFN parameters, but a HFN noise component is
369: required, $\Delta \chi^2 = 49$. This power spectrum is consistent with
370: those in the ``lower banana'' state. There is no detectable HFN in the
371: power spectrum from the observations with the softer hard color, $HC <
372: 0.35$. The total noise power in the 0.01-100~Hz band is 0.2\%
373: fractional rms and the spectrum is adequately described by a single
374: power law with an index of $1.08 \pm 0.02$. The shape of the power
375: spectrum is consistent with those seen in the ``upper banana'' state,
376: but the total noise power is lower than usual and spectrally the state
377: would be classified as the ``lower banana'' \citep{vdk95}.
378:
379: Based on the timing and color information, we identify XTE J1739-285 as
380: an atoll source. The source appears to have been in the ``banana''
381: state during all of the observations analyzed here, although its
382: evolution along the ``banana'' appears somewhat unusual.
383:
384: \begin{figure}
385: \centerline{\includegraphics[width=2.5in,angle=0]{f3.eps}}
386: \caption{Power spectrum showing a kHz QPO from October 31. The power
387: is Leahy normalized.} \label{khzqpo} \end{figure}
388:
389: We searched for high frequency quasiperiodic oscillations (QPOs) in
390: each uninterrupted RXTE observation window. We calculated averages of
391: 2~s power spectra for PCA events in the 4.2-20.4~keV energy band
392: (channels 6-26 in the event mode data). We searched for peaks in the
393: range from 100~Hz to 3000~Hz and fit any peak found with a Lorentzian
394: plus a constant equal to the calculated the Poisson noise level. There
395: were several strong QPO detections, see Table~\ref{qpotable}. The
396: second most significant detection occurred on October 31 and is shown
397: in Fig.~\ref{khzqpo}. Allowing for 1102 trials, which is calculated by
398: dividing the search interval of 2900~Hz by the QPO trial widths of 5,
399: 10, 20, 50, and 100~Hz, we estimate a chance probability of occurrence
400: of $1.2 \times 10^{-13}$ for this QPO. The chance probabilities of
401: occurrence, taking into account the number of trials, of all the other
402: QPOs listed in the Table are less than $1 \times 10^{-4}$. These QPO
403: detections establish XTE J1739-285 as a new member of the class of
404: neutron-star low-mass X-ray binaries producing kHz QPOs. We did not
405: detect two simultaneous kHz QPOs in any observation.
406:
407: %We report a marginal detection of a single QPO at a frequency of $716
408: %\pm 2$~Hz in a 3168~s long observation beginning on November 8 at
409: %06:11:39 UTC. However, the chance probability of occurrence of this
410: %signal, taking into account the number of trials, is 0.002 and the
411: %detection must be considered tentative. There are also several other
412: %possible, low significance QPOs in other observations overlapping the
413: %frequency range of the QPO detections in Table~\ref{qpotable}.
414:
415:
416: \section{Discussion}
417:
418: These observations establish that XTE J1739-285 exhibits millisecond
419: oscillations in its persistent X-ray emission. The properties of the
420: source are generally similar to those of other atoll neutron-star X-ray
421: binaries, although the evolution of the timing noise in the banana
422: state is unusual.
423:
424: The signal at 1122~Hz present in burst 2 is significant at the 99.96\%
425: confidence level. If this signal represents a true burst oscillation
426: then it would be of substantial interest. The near equality of the
427: burst oscillation frequency with the frequency of coherent pulsations
428: in the millisecond pulsars SAX J1808.4-3658
429: \citep{intZand01,Chakrabarty03} and XTE J1814-338 \citep{Strohmayer03}
430: and the frequency of coherent pulsations in a superburst from 4U
431: 1636-536 \citep{Strohmayer02} strongly suggests that the burst
432: oscillation frequency indicates the neutron star spin frequency. The
433: lack of any significant signal near 561~Hz in the burst from XTE
434: J1739-285 supports this interpretation and suggests that the possible
435: 1122~Hz oscillation would be most naturally interpreted as the spin
436: rate of the neutron star.
437:
438: If the burst oscillation frequency of 1122~Hz is the spin rate of the
439: neutron star, then XTE J1739-285 contains the most rapidly rotating
440: neutron star yet discovered. This spin rate is close to the
441: centrifugal breakup limit for some equations of state of nuclear matter
442: \citep{Burgio03} and, therefore, may remove the motivation for a
443: physical limit on neutron star spin other than the centrifugal breakup
444: limit. Furthermore, such a high spin rate would place constraints on
445: the nuclear equation of state, particularly if combined with a
446: measurement of the mass and/or radius of the neutron star.
447:
448:
449: \acknowledgments
450:
451: We thank Anna Watts and an anonymous referee for useful discussions and
452: the RXTE team, particularly Jean Swank and Evan Smith, for scheduling
453: these observations. PK and ZP acknowledge support from NASA grant
454: NNGO5GM77G.
455:
456:
457: %--------------
458:
459: %\clearpage
460:
461: \begin{thebibliography}{}
462:
463: \bibitem[Alpar et al.(1982)]{Alpar82} Alpar, M.A., Cheng, A.F.,
464: Ruderman, M. A., \& Shaham, J.\ 1982, Nature, 300, 728
465:
466: \bibitem[Bodaghee et al.(2005)]{Bodaghee05} Bodaghee, A., Mowlavi, N.,
467: Kuulkers, E., Wijnands, R., et al. 2005, ATEL 592
468:
469: \bibitem[Bradt, Rothschild, \& Swank(1993)]{Bradt93} Bradt, H.V.,
470: Rothschild, R.E., \& Swank, J.H. 1993, A\&AS, 97, 355
471:
472: \bibitem[Brandt et al.(2005)]{Brandt05} Brandt, S., Kuulkers, E.,
473: Bazzano, A., Courvoisier, T. J.-L., Domingo, A., et al. 2005, ATEL 622
474:
475: \bibitem[Burgio, Schulze, \& Weber(2003)]{Burgio03} Burgio, G.F,
476: Schulze, H.-J., Weber, F.\ 2003, A\&A, 408, 675
477:
478: \bibitem[Chakrabarty et al.(2003)]{Chakrabarty03} Chakrabarty, D.,
479: Morgan, E.H., Muno, M.P., Galloway, D.K., Wijnands, R., van der Klis,
480: M., Markwardt, C.B.\ 2003, Nature, 424, 42
481:
482: \bibitem[Dickey \& Lockman(1990)]{Dickey90} Dickey, J.M.\ \& Lockman,
483: F.J.\ 1990, ARA\&A, 28, 215
484:
485: \bibitem[Hessels et al.(2006)]{Hessels06} Hessels, J.W.T, Ranson, S.M.,
486: Stairs, I.H., Freire, C.C., Kaspi, V.M., Camilo, F.\ 2006, Science,
487: 311, 1901
488:
489: \bibitem[in 't Zand et al.(2001)]{intZand01} in 't Zand, J.J.M.\ et
490: al.\ 2001, A\&A, 372, 916
491:
492: \bibitem[Jahoda et al.(2006)]{Jahoda06} Jahoda, K., Markwardt, C.B.,
493: Radeva, Y., Rots, A.H., Stark, M.J., Swank, J.H., Strohmayer, T.E.,
494: Zhang, W.\ 2006, ApJS, 163, 401
495:
496: \bibitem[Kaaret et al.(2002)]{Kaaret02} Kaaret, P., in 't Zand,
497: J.J.M., Heise, J., Tomsick, J.A.\ 2002, ApJ, 575, 1018
498:
499: \bibitem[Kaaret et al.(2003)]{Kaaret03} Kaaret, P., in 't Zand,
500: J.J.M., Heise, J., Tomsick, J.A.\ 2003, ApJ, 598, 481
501:
502: \bibitem[Kaaret et al.(2006)]{Kaaret06} Kaaret, P., Morgan, E.H.,
503: Vanderspek, R., Tomsick, J.A.\ 2006, ApJ, 638, 963
504:
505: \bibitem[Kuulkers et al.(2003)]{Kuulkers03} Kuulkers, E., den Hartog,
506: P.R., in 't Zand, J.J.M., Verbunt, F.W.M., Harris, W.E., Cocchi, M.\
507: 2003, A\&A, 399, 663
508:
509: \bibitem[Leahy et al.(1983)]{Leahy83} Leahy, D.A., Darbro, W., Elsner,
510: R.F., Weisskopf, M.C., Sutherland, P.G., Kahn, S., Grindlay, J.E.\
511: 1983, ApJ, 266, 160
512:
513: \bibitem[Markwardt et al.(1999)]{Markwardt99} Markwardt, C. B.,
514: Marshall, F. E., Swank, J. H., \& Wei, C. 1999, IAU Circ. 7300
515:
516: \bibitem[Revnivtsev, Molkov, Sazonov(2006)]{Revnivtsev06} Revnivtsev,
517: M., Molkov, S., Sazonov, S.\ 2006, MNRAS, 373, L11
518:
519: \bibitem[Strohmayer et al.(1996)]{Strohmayer96} Strohmayer, T.E.,
520: Zhang, W., Swank, J.H., Smale, A., Titarchuk, L., Day, C., Lee, U.
521: 1996, ApJL, 469, L9
522:
523: \bibitem[Strohmayer \& Markwardt(2002)]{Strohmayer02} Strohmayer, T.E.\
524: \& Markwardt, C.B.\ 2002, ApJ, 577, 337
525:
526: \bibitem[Strohmayer et al.(2003)]{Strohmayer03} Strohmayer, T.E.\ \&
527: Markwardt, C.B., Swank, J.H., in 't Zand, J.\ 2003, ApJ, 596, L67
528:
529: \bibitem[Torres et al.(2006)]{Torres06} Torres, M. A. P.\ et al.\ 2006,
530: ATEL 784
531:
532: \bibitem[van der Klis(1995)]{vdk95} van der Klis, M.\ 1995,
533: in X-Ray Binaries, ed. W.H.G.\ Lewin, J.\ van Paradijs, \&
534: E.P.J.\ van den Heuvel (Cambridge: Cambridge Univ. Press),
535: 252
536:
537: \bibitem[Watts, Strohmayer, Markwardt(2005)]{Watts05} Watts, A.L.,
538: Strohmayer, T.E., Markwardt, C.B.\ 2005, ApJ, 634, 547
539:
540:
541: \end{thebibliography}
542:
543: \end{document}
544:
545: