1: \documentclass[12pt,preprint]{aastex} % for FASTTRACK submission!
2: % \documentclass{aastex}
3:
4: % Convert to emulate:
5: % \usepackage{emulateapj5}
6: % Move figures to main body
7: % Remove \begin{figure} and \end{figure}
8: % Change \caption to \figcaption
9: % Enclose each figure in braces
10:
11: \slugcomment{Submitted to ApJ Letters}
12:
13: % New commands
14: \newcommand{\xtej}{XTE J1751--305}
15: \newcommand{\saxj}{SAX J1808.4--3658}
16: \newcommand{\rxte}{{\it RXTE}}
17: \newcommand{\chandra}{{\it Chandra}}
18: \newcommand{\sax}{{\it BeppoSAX}}
19: \newcommand{\xmm}{{\it XMM-Newton}}
20:
21: \begin{document}
22:
23: \title{Discovery of a Second Millisecond Accreting Pulsar: XTE J1751--305}
24: \author{C. B. Markwardt,\altaffilmark{1,2}
25: J. H. Swank,\altaffilmark{2}
26: T. E. Strohmayer,\altaffilmark{2}
27: J. J. M. in 't Zand,\altaffilmark{3,4}
28: F. E. Marshall\altaffilmark{2}}
29:
30: \altaffiltext{1}{Department of Astronomy, University of Maryland,
31: College Park, MD 20742; craigm@lheamail.gsfc.nasa.gov}
32: \altaffiltext{2}{Laboratory for High Energy Astrophysics,
33: Mail Code 662, NASA Goddard Space Flight Center, Greenbelt, MD 20771}
34: \altaffiltext{3}{Astronomical Institute, Utrecht University, P.O. Box 80000,
35: NL - 3508 TA Utrecht, the Netherlands}
36: \altaffiltext{4}{SRON National Institute for Space Research, Sorbonnelaan 2,
37: NL - 3585 CA Utrecht, the Netherlands}
38:
39: \begin{abstract}
40: We report the discovery by the \rxte\ PCA of a second transient
41: accreting millisecond pulsar, \xtej, during regular monitoring
42: observations of the galactic bulge region. The pulsar has a spin
43: frequency of 435 Hz, making it one of the fastest pulsars. The
44: pulsations contain the signature of orbital Doppler modulation, which
45: implies an orbital period of 42 minutes, the shortest orbital period
46: of any known radio or X-ray millisecond pulsar. The mass function,
47: $f_x = (1.278 \pm 0.003)\times 10^{-6} M_\sun$, yields a minimum mass
48: for the companion of between 0.013 and 0.017 $M_\sun$, depending on
49: the mass of the neutron star. No eclipses were detected. A previous
50: X-ray outburst in June, 1998, was discovered in archival All-Sky
51: Monitor data. Assuming mass transfer in this binary system is driven
52: by gravitational radiation, we constrain the orbital inclination to be
53: in the range 30\arcdeg--85\arcdeg, and the companion mass to be
54: 0.013--0.035 $M_\sun$. The companion is most likely a heated helium
55: dwarf. We also present results from the \chandra\ HRC-S observations
56: which provide the best known position of \xtej.
57: \end{abstract}
58:
59: \keywords{binaries: close --- pulsars: general --- pulsars:
60: individual: XTE J1751$-$305 --- stars: neutron --- x-rays: binaries
61: --- white dwarfs}
62:
63: \section{Introduction}
64:
65: Accreting neutron stars in low mass X-ray binaries (LMXBs) are
66: conventionally thought to be the progenitors of millisecond or
67: ``recycled'' radio pulsars \citep{alpar82}. Firm evidence supporting
68: this theory remained elusive until the launch of NASA's {\it Rossi
69: X-ray Timing Explorer\/} (\rxte) in December, 1995. The discovery of
70: 300--600 Hz nearly coherent oscillations during thermonuclear X-ray
71: bursts \citep[e.g.,][]{stroh97} was a first solid indicator that
72: neutron stars in LMXBs rotate rapidly. This was followed by the
73: discovery in April, 1998, of the first accreting millisecond pulsar,
74: \saxj, with a spin period of 2.5 ms and orbital period of 2.1 hr
75: \citep{wijnands-vdk98,chakmorgan98}. This discovery convincingly
76: established a link between accreting neutron stars and recycled
77: pulsars. The presence of quasi-periodic oscillations in the range
78: 300--1300 Hz in many LMXBs has also been used to infer rapid neutron
79: star spin \citep[e.g.,][]{vdk00-khz}.
80:
81: Binary evolution models are becoming more sophisticated
82: \citep{podsiad02}, but still involve significant assumptions about
83: mass transfer and the effects of magnetic fields. \citet{kulkarni88}
84: have questioned whether the birthrate of LMXBs can account for the
85: number of millisecond radio pulsars. On the other hand, there have
86: been speculations that there should be a significant number of
87: low-luminosity transient LMXBs in the galaxy \citep{heise99,king00},
88: whose mass transfer and binary separation are driven primarily by the
89: emission of gravitational radiation. Although the discovery of \saxj\
90: provided convincing evidence that recycled pulsars can form in LMXBs,
91: it is difficult to draw inferences on binary and stellar evolution
92: based on a single case.
93: % Clearly, finding more
94: % examples of actively accreting millisecond pulsars would assist in
95: % population modeling.
96:
97: In this paper we report the discovery of an accreting millisecond
98: pulsar, \xtej, which was discovered by \rxte\ in regular monitoring of
99: the galactic center region. This is the fastest known accreting
100: pulsar and the second of its kind to be found. Recently, a third
101: pulsar XTE J0929$-$314, was discovered
102: \citep{remillard02,galloway02-psr}. Interestingly, all three systems
103: have very low mass companions. In \S\ref{sec:xte} and
104: \S\ref{sec:chandra}, we present the discovery by \rxte, and results of
105: a short \chandra\ observation to determine the source position. In
106: \S\ref{sec:timing} we develop a pulsar timing solution, and in
107: \S\ref{sec:spectral} we present basic spectral results.
108: \S\ref{sec:disc} contains a discussion of the binary system
109: properties. In this paper we focus on the pulsar timing properties,
110: and defer more detailed analyses of other issues to future work.
111:
112: % These new findings are demonstrating that a little recognized
113: % population of low luminosity transient LMXBs does exist and many of
114: % them apparently harbor rapidly rotating neutron stars.
115:
116:
117: \section{\rxte\ Observations\label{sec:xte}}
118:
119: \xtej\ was discovered using the \rxte\ Proportional Counter Array
120: (PCA) in a monitoring program of the galactic bulge region
121: \citep{swankmark01}. The PCA instrument has an effective area of
122: $\sim$6500 cm$^2$, and is sensitive to 2--60 keV X-rays within a
123: collimated field of view, which has a triangular profile and a full
124: width at half-maximum (FWHM) of 1\arcdeg. A region of approximately
125: 250 square degrees around the galactic center region has been scanned
126: by the PCA twice weekly since February 1999, except for several months
127: when sun constraints interfere. The nominal $1\sigma$ sensitivity of
128: the scans to variations is approximately 0.5--1 mCrab, but the
129: sensitivity is degraded somewhat within a few degrees of the galactic
130: center, where source confusion becomes important.
131:
132: % Individual source count rates are modulated by the PCA collimator as
133: % they pass into and out of the field of view. The resulting light
134: % curves are fitted to a model of known sources, convolved with the
135: % collimator response function.
136:
137: In a bulge scan on 2002 April 3.62, we detected a source whose
138: identification was not previously known \citep{mark02-disc}.
139: Subsequently, detections were made and positions were determined by
140: the \sax\ WFC and \xmm\ \citep{intzand02-pos,ehle02}. All of the
141: positions are essentially consistent with a previously unknown source
142: situated 2.1 degrees from the galactic center.
143:
144: % \footnote{The uncertainty of the PCA determination
145: % is itself somewhat uncertain, due to the crowded X-ray field.}
146:
147: % \begin{deluxetable}{llllc}
148: % \tablecaption{Position Determinations of \xtej\label{tab:position}}
149: % \tablehead{\colhead{Description}&\colhead{R.A. (J2000)\tablenotemark{a}}&
150: % \colhead{Dec. (J2000)}&\colhead{$\sigma$}&\colhead{Ref.}}
151: % \startdata
152: % PCA Scan & 17 51 30 & $-$30 30 & 5\arcmin & 1 \\
153: % BeppoSax WFC & 17 51 16 & $-$30 37 30 & 1\farcm2 & 2 \\
154: % XMM EPIC MOS2 & 17 51 13.5 & $-$30 37 22 & 10\arcsec & 3 \\
155: % %\chandra\ HRC-S & 17 51 13.52 & $-$30 37 22.9 & 0\farcs6 & 4 \\
156: % % Following data have been corrected for aspect offset
157: % \chandra\ HRC-S & 17 51 13.49 & $-$30 37 23.4 & 0\farcs6 & 4 \\
158: % \enddata
159: % \tablenotetext{a}{Values for R.A. are Hours, Minutes, Seconds;
160: % for Declination they are Degrees, Arcminutes, Arcseconds}
161: % \tablerefs{1---\citet{mark02-disc}, 2---\citet{intzand02-pos},
162: % 3---\citet{ehle02}, 4---\citet{mark02-pos}}
163: % \end{deluxetable}
164:
165: Follow-up \rxte\ pointed observations continued from April 4.6--30.9
166: on an essentially daily basis, for a total good exposure time of 398
167: ks. X-ray pulsations at a frequency of $\approx$435 Hz were discovered
168: in the first 200 seconds of the pointed observation on April 4.6, with
169: a semi-amplitude of $\simeq 5\%$ \citep{mark02-disc}. This makes
170: \xtej\ only the second accreting millisecond pulsar to be discovered.
171:
172: The PCA can be configured in a variety of data modes to optimize the
173: usage of telemetry and science return. The primary PCA data modes for
174: timing were {\tt GoodXenon} and {\tt E\_125us\_64M\_0\_1s}, which
175: provide 1 and 125 $\mu$s temporal resolutions, respectively.
176: Spectroscopy was performed using the {\tt Standard2} mode. For light
177: curves and spectroscopy, the PCU0 detector was excluded, because of a
178: missing propane layer which increases the background level. For high
179: resolution timing, all PCU detectors were used.
180:
181: % The primary PCA data mode for timing on the first two days was ``{\tt
182: % GoodXenon},'' which provides full spectral and temporal ($1 \mu$s)
183: % information. The remainder of the data were taken primarily in ``{\tt
184: % E\_125us\_64M\_0\_1s}'' mode, which has a coarser energy resolution
185: % and 125 $\mu$s time resolution, but has a lower telemetry usage.
186:
187:
188: \section{\chandra\ Observations \label{sec:chandra}}
189:
190: We obtained \chandra\ HRC-S target of opportunity observations of
191: \xtej\ on April 10.7 \citep{mark02-pos}. The HRC-S is a micro-channel
192: plate instrument sensitive to 0.1--10 keV X-rays, which can be placed
193: at the \chandra\ focal plane. Due to a problem in the assignment of
194: event times in the on-board electronics, the HRC must be operated in a
195: special ``Imaging'' mode with background reducing filters disabled.
196: Event timestamps are then corrected in the ground software. The total
197: exposure was 2969 s. A correction for spacecraft bore sight offset
198: was also applied. A source of $\sim$16000 counts was clearly detected
199: at a position consistent with those of \sax\ and \xmm\ (Table
200: \ref{tab:timing}). The nominal position error for \chandra\ is
201: dominated by uncertainties in the spacecraft aspect
202: solution\footnote{\tt http://cxc.harvard.edu/cal/ASPECT/celmon/}
203: \citep{aldcroft00}, whose 90\% confidence uncertainty is quoted in
204: Table \ref{tab:timing}. No other sources were detected in the HRC-S
205: field of view, which might have helped to refine the position further.
206: To date, this is the best known position of \xtej.
207:
208: % \footnote{See ``Check for aspect offset and fix''
209: % \chandra\ analysis thread, May 2, 2002 ({\tt
210: % http://cxc.harvard.edu/\-cal/\-ASPECT/\-fix\_offset/\-fix\_offset.cgi}).}
211:
212: % The instrument was configured in a special ``Imaging'' mode in order
213: % to mitigate the excess background counting rate from exceeding the
214: % telemetry bandwidth limit.
215:
216:
217: % It is in principle possible to refine the position even further if one
218: % or more known sources are detected in the field, and a frame tie can
219: % be formed to a high precision catalog. We searched for other sources
220: % using the XIMAGE version 4.1 software, but none were found above a
221: % $4\sigma$ threshold. Therefore we assign the nominal \chandra\ aspect
222:
223: \section{X-ray Timing \label{sec:timing}}
224:
225: An X-ray light curve of both scanning and pointed data is shown in
226: Figure~\ref{fig:ltlc}. Throughout this paper, particle background
227: subtraction was performed using the ``CMVLE'' model available from the
228: \rxte\ Guest Observer Facility. In addition to particle backgrounds,
229: there is also a significant astrophysical background component due to
230: nearby sources and galactic diffuse emission, which affects the
231: baseline determination. The mean quiescent flux levels in the scan
232: and pointing data (2.4 ct s$^{-1}$ and 5.63 ct s$^{-1}$ per PCU
233: respectively) were subtracted.
234:
235: Figure~\ref{fig:ltlc} shows that the X-ray peak flux probably occurred
236: around April 4.5, and thereafter the light curve exhibits a nearly
237: exponential decay, $e^{-t/\tau}$, where $\tau = 7.1\pm 0.1$ d (compare
238: to the same fit for \saxj, which yields $\tau = 14\pm 1$ d). A sudden
239: turn-off occurs around April 13, which is very similar to the turn-off
240: of \saxj. The rise time is less than four days. The figure also
241: shows that there was a detection of a small outburst of $\sim 0.3$
242: mCrab around April 27--30, and an X-ray burst on April 30.9. Given
243: the large PCA field of view, the source identification must be done
244: with care. A preliminary analysis of the burst indicates that it is
245: not from \xtej, but that the low-level flux may be.
246:
247: % which is accurate to within $\pm 3 \mu s$ of UTC
248: % (Markwardt, in preparation)
249:
250: When a more complete set of data were available, a sinusoidal
251: variation in the centroid frequency became apparent. To
252: systematically investigate this effect, we first corrected the X-ray
253: event arrival times at the satellite to the solar system barycenter
254: using the JPL planetary ephemeris DE405, the definitive \rxte\ orbit
255: ephemeris, and the best known source position. We also applied a fine
256: clock correction, which has a magnitude of 53--78 $\mu$s. Next, we
257: divided the data into 100 s segments and computed the Rayleigh or
258: $Z_1^2$ statistic \citep{buccheri83}, which is defined as
259: \begin{equation}
260: Z_1^2 = \left(\sum_j \cos \phi(t_j)\right)^2 +
261: \left(\sum_j \sin \phi(t_j)\right)^2
262: \end{equation}
263: where $\phi(t_j)$ is the pulse phase of a photon which arrives at time
264: $t_j$, according to a model of pulse phase evolution. For the initial
265: investigation, $\phi(t) = 2\pi ft$ for a grid of constant frequencies
266: $f$ around 435 Hz, and thus the $Z_1^2$ statistic is essentially the
267: Fourier power spectrum in a narrow bandpass. Each $Z_1^2$ transform
268: peak was fitted by a Gaussian function. The pulsation frequencies for
269: data from April 4.6--8.7, folded on a trial orbital period, are shown
270: in Figure~\ref{fig:orbfit}, and clearly establish the orbital
271: frequency modulation.
272:
273: % \begin{equation}
274: % Z_n^2 = \sum_{i=1}^{n}\left[\left(\sum_j \cos n\phi(t_j)\right)^2 +
275: % \left(\sum_j \sin n\phi(t_j)\right)^2\right]
276: % \end{equation}
277:
278: The timing model was refined by constructing a single pulse phase
279: connected solution over the data span from April 4.6--14.0. Events
280: from PCA channels 5--40 from all enabled PCUs (energies of 2.5--16.9
281: keV) were coherently summed to form a single $Z_1^2$ statistic. For
282: the phase model, we translated to IDL a version of the BNRYBT binary
283: pulsar model \citep{bt76} from the program TEMPO \citep{taylor89}.
284: The parameters of the binary were varied iteratively in order to
285: achieve the global maximum value of the $Z_1^2$ statistic. Table
286: \ref{tab:timing} contains the optimal binary parameters. After this,
287: we separated the data set into smaller (10 ks) segments and examined
288: the timing residuals by computing the cosine and sine terms of the
289: $Z_1^2$ statistic separately and converting to a time delay. The
290: r.m.s. residual was $\sim 30$ $\mu$s.
291:
292: % Maximization of
293: % the $Z^2$ is equivalent to least squares fitting of times of arrival,
294: % but without the need to form a template or perform cross correlation.
295:
296: % In this case the X-ray pulse profile is highly sinusoidal (see below),
297: % so the use of the $Z_1^2$ statistic is justified, as opposed to higher
298: % order statistics.
299:
300: We estimated the parameter confidence intervals by employing a Monte
301: Carlo estimation procedure. An ideal pulsar light curve model was
302: constructed with similar orbital and spin properties to \xtej. An
303: ensemble of 100 light curve realizations was drawn from this model
304: using Poisson statistics. Each realization was then converted to
305: events, and a $Z_1^2$ optimization was performed. For all parameters
306: of the model, the sample variances corresponded very closely to a
307: region within $\Delta Z_1^2 = 1$ of the peak value. Therefore, we
308: associate this region with the $1\sigma$ single-parameter confidence
309: region. For the actual X-ray data we stepped each parameter of
310: interest through a fixed grid while allowing the other parameters to
311: be optimized. Confidence intervals were set conservatively using
312: $\Delta Z_1^2 = 9$, corresponding to $\sim 3\sigma$ uncertainties, but
313: we quote enough precision to recover the $1\sigma$ errors.
314:
315: % While this may appear to be a novel result, similar
316: % precisions have been obtained using an odds ratio test
317: % \citep{zavlin00,gl96}.
318:
319: Using our best timing solution, we constructed an overall folded X-ray
320: pulse profile for the time range April 4.0--14.0. The pulsed fraction
321: is defined as $p_f = \sqrt{2(Z^2-2)/N_{ev}}$, which is then rescaled
322: to account for background. The mean pulsed fraction in the
323: fundamental was 4.41\%$\pm$0.02\%, but slowly decreases throughout the
324: outburst from 5.3\% to 3.9\%. After the light curve turn-off of April
325: 13, the pulsations are still detected at 3--4\% on April 14. The mean
326: amplitude of the first harmonic was 0.12\%$\pm$0.02\%, increasing from
327: $<0.12\%$ on April 4.6--6.5 (95\% confidence), to 0.27\%$\pm$0.05\% on
328: April 11.5--15.0. The amplitude of the second harmonic was $<
329: 0.04\%$, with no evidence of time evolution.
330:
331: % The dominance of the fundamental also justifies
332: % the choice of $Z_1^2$ for the timing analysis.
333:
334: We also examined the \chandra\ HRC-S data for pulsations. We
335: barycentered the data and selected 16414 events within $8.5\arcsec$ of
336: the source position, and computed the $Z_1^2$ statistic using the PCA
337: timing solution. We found $Z_1^2 = 27.65$ (a $\sim 5\sigma$
338: detection), which implies a pulsed fraction of about 5.6\%. Cross
339: calibration of the HRC and PCA data is a subject of further research.
340:
341: % Upon expanding to a grid search, the peak $Z_1^2\ (=27.85)$ occurred
342: % within 20 $\mu$Hz of the expected peak based on the PCA solution, and
343: % is also within the expected $1\sigma$ confidence limits.
344:
345: Finally, we applied the PCA solution to the small outburst of April
346: 28--30.9 and tested for pulsations. While the data for April 29.9
347: alone indicates a detection of $Z_1^2 = 6.8$ (a significance of
348: $2.1\sigma$), there are no pulsations detected in the other days from
349: April 27.8--30.9. Thus the overall 95\% upper limit for pulsations in
350: the small outburst is 5.5\%, and an unambiguous identification with
351: \xtej\ on the basis of pulsations is not possible.
352:
353: % the overall (incoherent) signal level is only
354: % significant at a $1\sigma$ level, which we do not deem to be strong
355: % enough to warrant an identification with \xtej.
356:
357: \section{Spectral Analysis \label{sec:spectral}}
358:
359: We sought to determine the total broadband X-ray flux by modeling the
360: X-ray spectrum. For the purposes of this paper, we chose
361: representative observations on April 5.6, 9.2 and 12.7, which
362: correspond to an approximately even sampling of the light curve decay.
363: PCA and HEXTE spectra were extracted using the analysis techniques
364: recommended by the \rxte\ Guest Observer Facility. The PCA response
365: matrix was computed using PCARSP version 8.0. The HEXTE instrument
366: has two clusters of detectors which are sensitive to 15--200 keV
367: X-rays and has field of view of $\sim 1^\circ$ (FWHM). The collimated
368: detectors are physically rocked back and forth by $\pm1.5^\circ$ in
369: order to sample time-varying particle backgrounds. In this analysis,
370: spectra from HEXTE cluster ``A'' were background subtracted using the
371: $-1.5^\circ$ rocking position.
372:
373: % ; the spectra from the $+1.5^\circ$
374: % position were clearly contaminated by another source.
375:
376: % We verified that once the source became undetectable in the PCA, the
377: % HEXTE ``source'' spectrum also converged to the background level.
378:
379: The joint spectrum is well modeled by an absorbed power law with
380: photon index 1.7--1.9, and an exponential cut-off energy of between
381: 100 and 200 keV. The spectrum remains more or less constant during
382: the decline of the outburst, and so we estimate that approximately
383: $2.6\pm0.5$ times as much flux appears in the 2--200 keV band as in
384: the 2--10 keV band depicted in Figure~\ref{fig:ltlc}. Using this
385: conversion we determine the integrated outburst fluence to be
386: $(2.5\pm0.5)\times 10^{-3}$~erg cm$^{-2}$ (2--200 keV).
387:
388: % The neutral hydrogen absorption value ranged from
389: % $(1.5-2.5)\times 10^{22}$ cm$^{-2}$.
390:
391: % However, \citet{miller02-epic} has found a low
392: % temperature black body component ($kT_{BB} = 1.06$ keV) in XMM-Newton
393: % EPIC spectra of the source --- which is largely below the PCA band ---
394: % so we do not ascribe much physical meaning to the absorption value.
395:
396:
397: We searched the \rxte\ All Sky Monitor (ASM) light curve for \xtej,
398: available from the MIT ASM web site. After filtering the data to
399: remove points with large uncertainty ($> 2.5$ ct/s), and rebinning to
400: form one day averages, we detected another probable outburst in June
401: of 1998 (Figure \ref{fig:asm}). The peak flux in the 2--12 keV band
402: was approximately $80\pm 10$ mCrab, and the duration was 3--4 days.
403: We also examined the PCA bulge monitoring data since February 1999,
404: but no outbursts exceeding 0.5 mCrab were detected. While there are
405: gaps in the ASM and PCA monitoring programs where a several day
406: outburst would be missed, we tentatively assign $T_{\rm rec} = 3.8$ yr
407: as the mean recurrence time.
408:
409: % The \sax\ WFC has
410: % accumulated over $\sim 6$ Ms of galactic center observations since
411: % 1996 (irregularly spaced) with no X-ray bursts or persistent emission
412: % detected from the source position \citep[in 't Zand 2002, private
413: % communication;][]{intzand01-bursters}.
414:
415: \section{Discussion \label{sec:disc}}
416:
417: % M_jup = 0.00095 M_sun
418:
419: The binary orbital parameters allow us to estimate the properties of
420: the companion. The 42 minute orbital period immediately reveals that
421: \xtej\ is a highly compact binary system. \xtej\ has the shortest
422: orbital period of all known millisecond pulsars (both X-ray and
423: radio). The amplitude of the modulations also determines the mass
424: function of the pulsar shown in Table~\ref{tab:timing}, defined as
425: $f_x = (M_c \sin i)^3 / (M_x + M_c)^2$, where $i$ is the binary
426: inclination to our line of sight, and $M_x$ and $M_c$ are the masses
427: of the pulsar and companion respectively. Given $M_x$, the mass
428: function provides the minimum possible value of $M_c$, which would
429: occur when viewing the binary edge-on ($i = 90^\circ$). The minimum
430: mass range shown in Table~\ref{tab:timing} reflects a reasonable range
431: of neutron star masses between 1.4--2.0 $M_\sun$. It is clear that
432: the companion of \xtej\ is in the regime of very low mass dwarfs, of
433: order 15 Jupiter masses. This is a factor of $\sim$3 smaller than the
434: companion of \saxj, which \citet{bildchak00} speculate is a $0.05
435: M_\sun$ brown dwarf.
436:
437: It is reasonable to assume that the companion must fill its Roche lobe
438: in order to transfer mass to the neutron star \citep{eggleton83}.
439: Combining the mass function and Roche lobe constraints results in a
440: curve in the $M_c$ vs. $R_c$ plane, shown in Figure~\ref{fig:mvr}.
441: This line can be compared to the equations of state of other types of
442: bodies, including hydrogen main sequence stars, brown dwarfs
443: \citep{chabrier00}, \saxj\ \citep{bildchak00}, and a cold helium dwarf
444: \citep{zapolsky69,rappaport84}. While none of the models intersect
445: the trace of \xtej, the hydrogen models are clearly unlikely. The
446: oldest brown dwarf model is unlikely, given the probability that
447: irradiation by the compact object will cause bloating
448: \citep{bildchak00}. A warm helium dwarf model, which would lie above
449: the ``cold helium dwarf'' curve in Figure~\ref{fig:mvr}, appears to be
450: the most likely scenario, but there appear to be no calculations of
451: such a configuration in the literature.
452:
453: There were no X-ray eclipses or dips detected in the PCA light curves.
454: We therefore put an upper limit of $i<85^\circ$. There is also no
455: evidence of X-ray modulations at the binary period (at a limit of
456: $~\sim 0.5\%$), which might have implied propagation through a
457: scattering atmosphere in a near edge-on geometry \citep{bildchak00}.
458: On the other hand, a very low inclination is also unlikely, since that
459: would imply a high companion mass, and thus a large mass transfer
460: rate. Following the reasoning of \citet{bildchak00}, we find that the
461: time-averaged mass accretion rate, based on the measured X-ray fluence
462: and a recurrence time of 3.8 yr, is $\dot{M}_x = 2.1\times 10^{-11}
463: M_\sun$ yr$^{-1} d_{10}^2 m_{1.4}^{-1} T_{3.8}^{-1}$, where $d_{10}$
464: is the distance in units of 10 kpc, $m_{1.4} = M_x / (1.4 M_\sun)$,
465: and assuming a neutron star radius of 10 km. On the other hand, the
466: mass transfer in these systems is thought to be driven by
467: gravitational radiation \citep{king00,rappaport84}, in which case the
468: mass transfer rate should be $\dot{M}_{GR} = 1.2\times 10^{-11}
469: M_\sun$ yr$^{-1} m_c^2 m_{1.4}^{2/3}$, where $m_c = M_c / (0.0137
470: M_\sun$), and we have assumed the companion is nearly degenerate. If
471: the two mass transfer rates are equal, we arrive at the constraint,
472: $\sin i = 0.74 m_{1.4}^{3/2} d_{10}^{-1}$. For distances within 15
473: kpc, this constraint implies inclinations of 30\arcdeg--85\arcdeg and
474: companion masses in the range 0.013--0.035 $M_\sun$, for neutron star
475: masses between 1.4 $M_\sun$ and 2.0 $M_\sun$.
476:
477: An interesting result of this constraint is that the distance is at
478: least 7 kpc, significantly farther than \saxj\ \citep[$d = 2.5$
479: kpc;][]{intzand01-1808}. The peak luminosity of the persistent
480: emission would then be $\gtrsim 2\times 10^{37}$ erg s$^{-1}$, an
481: order of magnitude higher than that of \saxj\ \citep{cui98}. It is
482: probable that \xtej\ is near the galactic center, in which case more
483: pulsars like it should be detectable by the PCA bulge monitoring
484: program over the next few years.
485:
486: \acknowledgments This work was partially supported by a NASA
487: Astrophysics Data Program grant.
488:
489: % \xtej\ is only the second accreting millisecond pulsar to be
490: % discovered, and it shares many similarities to the first, \saxj. Both
491: % have similar fast-rise, exponential-decay type light curves, with a
492: % sudden turn-off after decaying to about 30\% of peak flux.
493:
494: % \begin{equation}
495: % \left<\dot{M}_x\right> = 2.1\times 10^{-11} M_\sun\ {\rm yr}^{-1}
496: % \left({r_{ns}\over 10\ {\rm km}}\right)
497: % \left({M_x\over 1.4 M_\sun}\right)^{-1}
498: % \left({d\over 10\ {\rm kpc}}\right)^2
499: % \left({T_{\rm rec}\over 3.8\ {\rm yr}}\right)^{-1}
500: % \end{equation}
501:
502: % \begin{equation}
503: % \dot{M}_{GR} = 5.6\times 10^{-11} M_\sun\ {\rm yr}^{-1}
504: % \left({M_c\over 0.03 M_\sun}\right)^2
505: % \left({M_x\over 1.4 M_\sun}\right)^{2/3}
506: % \left({P_b\over 42\ {\rm min}}\right)^{-8/3}
507: % \end{equation}
508:
509:
510: % \begin{equation}
511: % f_x = {(M_x \sin i)^3\over (M_x + M_c)^2} \\
512: % \end{equation}
513: % where $i$ is the binary inclination, and $M_x$ and $M_c$ are the
514: % masses of the neutron star and companion.
515:
516:
517: % % pb = 2545.341358158969d ; Binary period (s)
518: % % epb = 3.8d-3 ; Uncertainty in pb (s)
519: % % asini = 0.010113371688d * 2.99792458d10 ; a sin(i) (cm)
520: % % easini = 8.3d-6 * 2.99792458d10 ; Uncertainty in a sin(i) (s)
521: % % msol = 1.98892d33 ; Solar mass (gm)
522: % % f1 = 4d*!dpi^2*asini^3/6.67259d-8/pb^2/msol ; Mass function (gm)
523: % % ef1 = f1*sqrt((3*easini/asini)^2 + (2*epb/pb)^2)
524:
525:
526:
527: % \citet{chakmorgan98} and \citet{bildchak00} have suggested that
528: % modulations detected in the X-ray flux in \saxj, at the orbital
529: % period, are evidence of scattering by the companion's wind or other
530: % circumbinary material, and hence constrain the binary to be viewed
531: % nearly edge-on. We searched for X-ray modulations in \xtej\ at its
532: % orbital period, but found none, at a limit of a few tenths of a
533: % percent. However this search is confounded somewhat by unmodeled
534: % variations in the \rxte\ particle background rate, which have a
535: % typical period of $0.5\times 5709$ s = 2854 s, very close to the
536: % binary orbital period of \xtej.
537:
538: % The small outburst and X-ray burst between April 28--30 is formally
539: % from an unidentified source. In principle it could be due to another
540: % source in the $1^\circ$-radius field of view. If another source were
541: % off axis, it would be attenuated by the PCA collimator, and hence be
542: % brighter when viewed on-axis. No such brighter sources were detected
543: % in subsequent PCA bulge monitoring observations, which scan over the
544: % region with a rather dense scan pattern. Also, the marginal detection
545: % of pulsations on April 30 suggests \xtej\ is probably the source.
546: % \saxj\ also showed post-outburst fluctuations in 1998 and 2000
547: % \citep{wijnands01}, so this would not be completely unexpected.
548:
549:
550: \begin{thebibliography}{dummy}
551:
552: \bibitem[Aldcroft et al.(2000)]{aldcroft00} Aldcroft, T. L., Karovska,
553: M., Cresitello-Dittmar, M. L., Cameron, R. A. \& Markevitch,
554: M. L. 2000, in Proc. SPIE, 4012, X-Ray Optics, Instruments and
555: Missions III, ed. J. Tr\"umper \& B. Aschenbach, 650
556:
557: \bibitem[Alpar et al.(1982)]{alpar82} Alpar, M.~A., Cheng, A.~F.,
558: Ruderman, M.~A., \& Shaham, J.\ 1982, \nat, 300, 728
559:
560: % \bibitem[Backer(1998)]{backer98} Backer, D.~C.\ 1998, \apj, 493, 873
561:
562: \bibitem[Bildsten \& Chakrabarty(2001)]{bildchak00} Bildsten,
563: L.~\& Chakrabarty, D.\ 2001, \apj, 557, 292
564:
565: \bibitem[Blandford \& Teukolsky(1976)]{bt76} Blandford,
566: R.~\& Teukolsky, S.~A.\ 1976, \apj, 205, 580
567:
568: % \bibitem[Brazier(1994)]{brazier94} Brazier, K.~T.~S.\ 1994,
569: % \mnras, 268, 709 % Confidence intervals from the Rayleigh test
570:
571: \bibitem[Buccheri et al.(1983)]{buccheri83} Buccheri, R.~et al.\
572: 1983, \aap, 128, 245
573:
574: \bibitem[Chabrier et al. (2000)]{chabrier00} Chabrier, G., Baraffe,
575: I., Allard, F., \& Hauschildt, P.\ 2000, \apj, 542, 464
576: % 2000ApJ...542..464C
577:
578: \bibitem[Chakrabarty \& Morgan(1998)]{chakmorgan98} Chakrabarty,
579: D.~\& Morgan, E.~H.\ 1998, \nat, 394, 346
580:
581: % \bibitem[Chen, Shrader, \& Livio(1997)]{chen97} Chen, W.,
582: % Shrader, C.~R., \& Livio, M.\ 1997, \apj, 491, 312
583:
584: \bibitem[Cui, Morgan, \& Titarchuk(1998)]{cui98} Cui, W., Morgan,
585: E.~H., \& Titarchuk, L.~G.\ 1998, \apjl, 504, L27
586:
587: \bibitem[Eggleton(1983)]{eggleton83} Eggleton, P.~1983, \apj, 268, 368
588:
589: \bibitem[Ehle et al.(2002)]{ehle02} Ehle, M.~et al.\ 2002,
590: \iaucirc, 7872
591:
592: \bibitem[Galloway et al.(2002)]{galloway02-psr} Galloway, D.~K.,
593: Morgan, E.~H., Remillard, R.~A., \& Chakrabarty, D.\ 2002, \iaucirc,
594: 7900
595: % 2002IAUC.7900....2G
596:
597: % \bibitem[Giles et al. (2002)]{giles02}
598: % Giles, A.~B., Hill, K.~M., Strohmayer, T.~E., \& Cummings, N.\ 2002, \apj,
599: % 568, 279
600:
601: % \bibitem[Gregory \& Loredo(1996)]{gl96} Gregory, P.~C.~\&
602: % Loredo, T.~J.\ 1996, \apj, 473, 1059
603:
604: \bibitem[Heise, J. et al. (1999)]{heise99} Heise, J. et al. 1999, in
605: Astrophys. Lett. Comm., 38, Proc. Third INTEGRAL Workshop,
606: ed. A. Bazzano, G.G.C. Palumbo \& C. Winkler, 297
607:
608: % \bibitem[in't Zand(2001)]{intzand01-bursters} in't Zand, J.\ 2001, in
609: % ESA SP-459, Proc. Fourth INTEGRAL Workshop, ed. A. Gimenez, V. Reglero
610: % \& C. Winkler (Nordwijk: ESA Publications), 463
611:
612: \bibitem[in't Zand et al.(2001)]{intzand01-1808} in't Zand,
613: J.~J.~M.~et al.\ 2001, \aap, 372, 916
614: % 2001A&A...372..916I
615:
616: % \bibitem[in 't Zand et al.(1998)]{intzand96} in 't Zand,
617: % J.~J.~M., Heise, J., Muller, J.~M., Bazzano, A., Cocchi, M., Natalucci, L.,
618: % \& Ubertini, P.\ 1998, \aap, 331, L25
619:
620: \bibitem[in 't Zand et al.(2002)]{intzand02-pos} in 't Zand,
621: J.~J.~M., Lowes, P., Heise, J., Bazzano, A., \& Ubertini, P.\ 2002,
622: \iaucirc, 7867
623:
624: \bibitem[King(2000)]{king00} King, A.~R.\ 2000, \mnras, 315, L33
625: % 2000MNRAS.315L..33K
626:
627: \bibitem[Kulkarni \& Narayan(1988)]{kulkarni88} Kulkarni, S.~R.~\&
628: Narayan, R.\ 1988, \apj, 335, 755
629:
630: \bibitem[Markwardt \& Dobrzycki(2002)]{mark02-pos} Markwardt, C.~B.~\&
631: Dobrzycki, A.\ 2002, \iaucirc, 7876
632:
633: \bibitem[Markwardt \& Swank(2002)]{mark02-disc} Markwardt, C.~B.~\&
634: Swank, J.~H.\ 2002, \iaucirc, 7867
635:
636: % \bibitem[Markwardt, et al. (2000)]{mark00-bulge} Markwardt, C.~B.,
637: % Swank, J.~H., Marshall, F.~E., \& in't Zand, J.~J.~M.\ 2000,
638: % Rossi2000: Astrophysics with the Rossi X-ray Timing Explorer.~March
639: % 22-24, 2000 at NASA's Goddard Space Flight Center, Greenbelt, MD USA,
640: % p.E7
641:
642: % \bibitem[Miller et al.(2002)]{miller02-epic} Miller, J.~M.~et al.\
643: % 2002, Astronomer's Telegram, 90
644:
645: % \bibitem[Miller, Lamb, \& Psaltis(1998)]{miller98} Miller,
646: % M.~C., Lamb, F.~K., \& Psaltis, D.\ 1998, \apj, 508, 791
647:
648: \bibitem[Podsiadlowski, Rappaport, \& Pfahl(2002)]{podsiad02}
649: Podsiadlowski, P., Rappaport, S., \& Pfahl, E.~D.\ 2002, \apj, 565,
650: 1107
651:
652: % \bibitem[Protheroe(1987)]{1987PASAu...7..167P} Protheroe, R.~J.\ 1987,
653: % Proceedings of the Astronomical Society of Australia, 7, 167
654: % Periodic analysis of gamma-ray data
655:
656: \bibitem[Rappaport \& Joss(1984)]{rappaport84} Rappaport, S.~\&
657: Joss, P.~C.\ 1984, \apj, 283, 232
658: % 1984ApJ...283..232R
659:
660: \bibitem[Remillard(2002)]{remillard02} Remillard, R.~A.\ 2002,
661: \iaucirc, 7888
662: % 2002IAUC.7888....2R
663:
664: % \bibitem[Savonije, de Kool, \& van den Heuvel(1986)]{savonije86}
665: % Savonije, G.~J., de Kool, M., \& van den Heuvel, E.~P.~J.\ 1986, \aap,
666: % 155, 51
667:
668: % \bibitem[Strohmayer \& Markwardt(1999)]{strohmark99} Strohmayer,
669: % T.~E.~\& Markwardt, C.~B.\ 1999, \apjl, 516, L81
670:
671: % \bibitem[Strohmayer \& Markwardt (2002)]{stroh02-superburst}
672: % Strohmayer, T.~E., \& Markwardt, C. B.\ 2002, \apj, in press
673: % (astro-ph/0205435)
674:
675: \bibitem[Strohmayer, Zhang, \& Swank(1997)]{stroh97}
676: Strohmayer, T.~E., Zhang, W., \& Swank, J.~H.\ 1997, \apjl, 487, L77
677: % 1997ApJ...487L..77S
678:
679: \bibitem[Swank \& Markwardt (2001)]{swankmark01} Swank, J. \&
680: Markwardt, C. 2001, in ASP Conf. Ser. 251, New Century of X-ray
681: Astronomy, ed. H. Inoue \& H. Kunieda (San Francisco: ASP), 94
682:
683: \bibitem[Taylor \& Weisberg(1989)]{taylor89} Taylor, J.~H.~\&
684: Weisberg, J.~M.\ 1989, \apj, 345, 434
685:
686: \bibitem[van der Klis(2000)]{vdk00-khz} van der Klis, M.\ 2000,
687: \araa, 38, 717 % 2000ARA&A..38..717V
688:
689: % \bibitem[van der Klis(1999)]{vdk99-khz} van der Klis, M.\ 1999,
690: % Pulsar Timing, General Relativity and the Internal Structure of
691: % Neutron Stars, 259
692:
693: % \bibitem[Webbink, Rappaport, \& Savonije(1983)]{webbink83}
694: % Webbink, R.~F., Rappaport, S., \& Savonije, G.~J.\ 1983, \apj, 270, 678
695:
696: % \bibitem[Wijnands et al.(2001)]{wijnands01} Wijnands, R.,
697: % M{\'e}ndez, M., Markwardt, C., van der Klis, M., Chakrabarty, D., \& Morgan,
698: % E.\ 2001, \apj, 560, 892
699:
700: \bibitem[Wijnands \& van der Klis(1998)]{wijnands-vdk98} Wijnands,
701: R.~\& van der Klis, M.\ 1998, \nat, 394, 344
702:
703: \bibitem[Zapolsky \& Salpeter(1969)]{zapolsky69} Zapolsky,
704: H.~S.~\& Salpeter, E.~E.\ 1969, \apj, 158, 809 % 1969ApJ...158..809Z
705:
706: % \bibitem[Zavlin et al. (2000)]{zavlin00}
707: % Zavlin, V.~E., Pavlov, G.~G., Sanwal, D., \& Tr{\" u}mper, J.\ 2000,
708: % \apjl, 540, L25
709:
710:
711: \end{thebibliography}
712:
713: \begin{deluxetable}{lc}
714: \tablecaption{Timing Parameters of \xtej\label{tab:timing}}
715: \tablehead{\colhead{Parameter} & \colhead{Value}}
716: \startdata
717: Right ascension, $\alpha$ (J2000) &
718: $17^{\rm h}51^{\rm m}13\fs49(5)$\tablenotemark{a}\\
719: Declination, $\delta$ (J2000) &
720: $-30\arcdeg37\arcmin23\farcs4(6)$\tablenotemark{a}\\
721: Barycentric pulse frequency, $f_o$ (Hz) & 435.317993681(12)\tablenotemark{b} \\
722: Pulsar frequency derivative, $|\dot f|$ (Hz s$^{-1}$) & $<3\times 10^{-13}$\\
723: Projected semimajor axis, $a_x \sin i$ (lt-ms) & 10.1134(83) \\
724: Binary orbital period, $P_b$ (s) & 2545.3414(38) \\
725: Epoch of mean longitude $90^\circ$, $T_{90}$ & 54118.7563591(87)\tablenotemark{c} \\
726: Orbital eccentricity, $e$ & $<1.7\times 10^{-3}$\\
727: Pulsar mass function, $f_x$ ($10^{-6} M_\sun$) & 1.2797(31) \\
728: Minimum companion mass, $M_c$ ($M_\sun$) & 0.0137--0.0174 \\
729: Maximum Power, $Z_{\rm max}^2$ & 36237 \\
730: \enddata
731: \tablenotetext{a}{Parameter was fixed; 90\% confidence limits from \chandra\
732: aspect uncertainty.}
733: \tablenotetext{b}{Uncertainties and upper limits are $3\sigma$ in last quoted digits.}
734: \tablenotetext{c}{Modified Julian days, referred to TDB timescale.}
735: \end{deluxetable}
736:
737: \begin{figure}
738: \plotone{f1.eps}
739: % \centerline{\psfig{file=fig-ltlc.ps,angle=+90,width=\textwidth}}
740: \caption{\rxte\ PCA light curve of \xtej\ in the 2--10 keV band from
741: galactic bulge monitoring (crosses) and pointings (vertical bars; 1600
742: s bins). The inset shows an expanded vertical scale. The arrow
743: indicates the time of an X-ray burst. Note that 1 mCrab (2--10 keV) =
744: $2.42\times 10^{-11}$ erg s$^{-1}$ cm$^{-2}$ = 2.27 ct s$^{-1}$
745: PCU$^{-1}$.
746: \label{fig:ltlc}}
747: \end{figure}
748:
749: \begin{figure}
750: \plotone{f2.eps}
751: % \centerline{\psfig{file=fig-orbfit.ps,angle=+90,width=\textwidth}}
752: \caption{Measured pulse frequency as a function of orbital phase,
753: folded on a trial period of $P_b = 2545.35$ s, for data from April
754: 4.6--8.7. The best fit sinusoid model is shown (thick line).
755: \label{fig:orbfit}}
756: \end{figure}
757:
758: \begin{figure}
759: \plotone{f3.eps}
760: % \centerline{\psfig{file=fig-asm.ps,angle=+90,width=\textwidth}}
761: \caption{\xtej\ light curves. (top) ASM (2--12 keV); (bottom) PCA
762: bulge monitoring (2--10 keV). (bottom inset) ASM Light curve from
763: June 6--26, 1998. Triangles highlight the two outburst dates.
764: \label{fig:asm}}
765: \end{figure}
766: % 1 Crab = 75 ASM ct s$^{-1}$.
767:
768: \begin{figure}
769: \plotone{f4.eps}
770: % \centerline{\psfig{file=fig-mvr.ps,angle=+90,width=\textwidth}}
771: \caption{Companion mass ($M_c$) vs. radius ($R_c$) plane, showing the
772: Roche lobe constraints for \xtej\ (highlighted solid) and \saxj (thin
773: solid). The equations of state are shown for hydrogen main sequence
774: (dash-dot), brown dwarfs (dashed) and a cold helium dwarf (thick
775: solid). Shaded contours of represent lines of constant $\dot{M}_{GR}$
776: for a $1.4 M_\sun$ neutron star \citep{rappaport84}. Brown dwarf
777: models are for ages of (from top to bottom) 0.1, 0.5, 1, 5 and 10
778: billion years.
779: \label{fig:mvr}}
780: \end{figure}
781:
782:
783: \end{document}
784: