1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{psfig}
3:
4:
5:
6:
7:
8: \newcommand\mdot {\hbox {${\dot M}$}}
9: \newcommand\sz {$S_{\rm z}$}
10: \newcommand\mzon {M$_{\odot}$}
11: \newcommand\pp {$\pm$}
12: \newcommand\pers {s$^{-1}$}
13: \newcommand\micros {$\mu$s}
14:
15: \def\degr{\hbox{$^\circ$}}
16: \newcommand\Lunit {ergs s$^{-1}$}
17: \newcommand\funit {ergs cm$^{-2}$ s$^{-1}$}
18: \newcommand\nh {$N_{\rm H}$}
19:
20:
21: \begin{document}
22:
23: \title{
24: {\itshape Chandra} observations of the accretion-driven millisecond
25: X-ray pulsars XTE J0929--314 and XTE J1751--305 in quiescence}
26:
27: \author{
28: Rudy Wijnands\altaffilmark{1}, Jeroen Homan\altaffilmark{2}, Craig
29: O. Heinke\altaffilmark{3,4,5} Jon M. Miller\altaffilmark{3,6}, Walter
30: H. G. Lewin\altaffilmark{2}}
31:
32: \altaffiltext{1}{Astronomical Institute ``Anton Pannekoek'',
33: University of Amsterdam, Kruislaan 403, 1098 SJ, Amsterdam, the
34: Netherlands; rudy@science.uva.nl}
35:
36: \altaffiltext{2}{Center for Space Research, Massachusetts Institute of
37: Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA;
38: jeroen@space.mit.edu, lewin@space.mit.edu}
39:
40:
41:
42: \altaffiltext{3}{Harvard-Smithsonian Center for Astrophysics,
43: 60 Garden Street, Cambridge, MA 02139, USA;
44: cheinke@head.cfa.harvard.edu; jmmiller@head.cfa.harvard.edu}
45:
46:
47: \altaffiltext{4}{Northwestern University, Dept. of Physics \&
48: Astronomy, 2145 Sheridan Rd., Evanston, IL 60208}
49:
50: \altaffiltext{5}{Lindheimer Postdoctoral Fellow}
51:
52:
53: \altaffiltext{6}{NSF Astronomy \& Astrophysics Fellow}
54:
55:
56:
57: \begin{abstract}
58:
59: We observed the accretion-driven millisecond X-ray pulsars XTE
60: J0929--314 and XTE J1751--305 in their quiescent states using {\it
61: Chandra}. From XTE J0929--314 we detected 22 source photons (in the
62: energy range 0.3--8 keV) in $\sim$24.4 ksec, resulting in a
63: background-corrected time-averaged count rate of $9\pm2 \times
64: 10^{-4}$ counts s$^{-1}$. The small number of photons detected did
65: not allow for a detailed spectral analysis of the quiescent spectrum,
66: but we can demonstrate that the spectrum is harder than simple thermal
67: emission which is what is usually presumed to arise from a cooling
68: neutron star that has been heated during the outbursts. Assuming a
69: power-law model for the time-averaged (averaged over the whole
70: observation) X-ray spectrum, we obtain a power-law index of
71: $1.8^{+0.6}_{-0.5}$ and an unabsorbed X-ray flux of $6^{+4}_{-2}
72: \times 10^{-15}$ ergs s$^{-1}$ cm$^{-2}$ (for the energy range 0.5--10
73: keV), resulting in a 0.5--10 keV X-ray luminosity of $7^{+5}_{-2}
74: \times 10^{31}$ ($d$/10 kpc)$^2$ ergs s$^{-1}$, with $d$ the distance
75: toward the source in kpc. No thermal component could be detected;
76: such a component contributed at most 30\% to the 0.5--10 keV flux.
77: Variability in the count rate of XTE J0929--314 was observed at the
78: 95\% confidence level. We did not conclusively detect XTE J1751--305
79: in our $\sim$43 ksec observation, with 0.5--10 keV flux upper limits
80: between 0.2 and 2.7 $\times 10^{-14}$ ergs s$^{-1}$ cm$^{-2}$
81: depending on assumed spectral shape, resulting in 0.5--10 keV
82: luminosity upper limits of 0.2 -- 2 $\times 10^{32}$ ($d$/8 kpc)$^2$
83: ergs s$^{-1}$. We compare our results with those obtained for other
84: neutron-star X-ray transients in their quiescent state, and in
85: particular with the quiescent properties of SAX J1808.4--3658. Using
86: simple accretion disk physics in combination with our measured
87: quiescent luminosity of XTE J0929--314 and the luminosity upper limits
88: of XTE J1751--305, and the known spin frequency of the neutron stars,
89: we could constrain the magnetic field of the neutron stars in XTE
90: J0929--314 and XTE J1751--305 to be less than $3\times 10^9 {d \over
91: {\rm 10~kpc}}$ and $3 - 7 \times 10^8 {d
92: \over {\rm 8~kpc}}$ Gauss (depending on assumed spectral shape of the
93: quiescent spectrum), respectively.
94:
95:
96: \end{abstract}
97:
98:
99: \keywords{
100: accretion, accretion disks --- stars: neutron stars: individual (XTE
101: J0929--314; XTE J1751--305)--- X-rays: stars}
102:
103: \section{Introduction}
104:
105:
106: Neutron stars in low-mass X-ray binaries accrete matter from solar
107: mass companion stars. The sub-group of neutron-star transients spend
108: most of their time in quiescence during which hardly any or no
109: accretion occurs onto their neutron stars. However, these transients
110: sporadically become very luminous in X-rays ($>$$10^{36-38}$~\Lunit)
111: owing to a huge increase in the accretion rate. During those
112: outbursts, these sources can be readily studied with the available
113: X-ray instruments, but so far only about a dozen sources have been
114: studied in their much dimmer quiescent states. In quiescence, they
115: typically exhibit 0.5--10 keV luminosities of $10^{32-34}$~\Lunit~and
116: their spectra are usually dominated by a soft component which can be
117: described by a thermal model (either a black-body or a neutron-star
118: atmosphere model). This emission is generally ascribed to the cooling
119: of the neutron star which has been heated during the outbursts (e.g.,
120: via deep crustal heating; Brown, Bildsten, \& Rutledge 1998). For
121: several quiescent systems an additional power-law shaped component is
122: present in their X-ray spectra and it dominates above a few keV (e.g.,
123: Asai et al. 1998; Rutledge et al. 2001). The origin of this power-law
124: component is not understood but it has been proposed to be due to
125: residual accretion on the neutron-star magnetic field, an active
126: pulsar mechanism, or shock emission due to the interaction of the
127: pulsar wind and matter which is still being transferred from the
128: companion star (e.g., Stella et al.~1994; Campana et al.~1998; Campana
129: \& Stella 2000).
130:
131: The fractional contribution of this power-law component to the flux in
132: the 0.5--10 keV energy range varies significantly between systems. In
133: some systems this component cannot be detected (with less than 10\% of
134: the 0.5--10 keV flux possibly due to such a component), but in other
135: systems it contributes up to half the emission in the 0.5--10 keV
136: energy range (e.g., Asai et al.~1998; Rutledge et
137: al.~2001). Currently, only two quiescent systems (SAX J1808.4--3658
138: and EXO 1745--248) have been found to exhibit quiescent spectra which
139: are fully dominated by the power-law component (with over 90\% of the
140: 0.5--10 keV flux due to the power-law component; Campana et al.~2002;
141: Wijnands et al.~2004). For several other systems (i.e., SAX
142: J1810.8--2609 and XTE J2123--058; Jonker, Wijnands, \& van der Klis
143: 2004a; Tomsick et al.~2004) it has also been found that the quiescent
144: spectra could be adequately fitted with only a power-law
145: model. However, the statistics of the data for those sources were
146: rather limited, still allowing for a thermal component which
147: contributed more than 50\%--70\% to the 0.5--10 keV flux.
148:
149: During outburst SAX J1808.4--3658 is an accretion-driven millisecond
150: X-ray pulsar and its anomalous X-ray properties in quiescence (its
151: hard quiescent spectrum and low luminosity of $\sim$$5\times 10^{31}$
152: ergs s$^{-1}$; Campana et al.~2002) might be related to the expected
153: higher magnetic field strength of the neutron star in this system,
154: compared to that of the neutron stars in the non-pulsating
155: systems. However, the recent study of the neutron-star X-ray transient
156: EXO 1745--248 located in the globular cluster Terzan 5 during its
157: quiescent state has cast doubt on this hypothesis. Wijnands et
158: al.~(2004) found that the Terzan 5 system, which is not observed to
159: pulsate during outburst, also had a quiescent spectrum which was fully
160: dominated by the power-law component (more than 90\% of the 0.5--10
161: keV flux was due to the power-law component) just like SAX
162: J1808.4--3658. However, EXO 1745--248 had a quiescent luminosity
163: ($\sim$$2\times10^{33}$ ergs s$^{-1}$) which was a factor of $\sim$40
164: higher than what has been observed for SAX J1808.4--3658.
165:
166: The physical process(es) behind the power-law component and the
167: differences among sources in terms of the properties of this spectral
168: component are unknown. However, some clues about the nature of this
169: spectral component were recently glimpsed in the work by Jonker et
170: al.~(2004a, b). They compared all quiescent neutron star X-ray
171: transients for which spectral information was available and found that
172: the fractional contribution of the power-law component to the 0.5--10
173: keV fluxes is lowest when these sources have quiescent luminosities of
174: $\sim 1 - 2 \times 10^{33}$ ergs s$^{-1}$. They found that both at
175: higher and lower quiescent luminosities the fractional contribution of
176: the power-law component to the 0.5--10 keV flux increases. So far,
177: only EXO 1745--248 did not follow this correlation. It remains to be
178: seen whether this transient is an unusual system or if the correlation
179: found by Jonker et al.~(2004a, b) is spurious and more sources similar
180: to EXO 1745--248 will be found.
181:
182: In recent years, four additional accreting millisecond X-ray pulsars
183: (after SAX J1808.4--3658) have been discovered (Markwardt et al.~2002;
184: Galloway et al. 2002; Markwardt, Smith, \& Swank 2003; Markwardt \&
185: Swank 2003). Those systems are prime targets to observe in quiescence
186: to test the hypothesis that the unusual quiescent properties of SAX
187: J1808.4--3658 are due to a relatively strong neutron-star magnetic
188: field strength compared to that of the non-pulsating systems: one
189: would expect that the magnetic fields of the neutron stars in those
190: additional systems are of similar strength as that of the neutron star
191: in SAX J1808.4--3658. To this end we had observations of XTE
192: J1751--305 and XTE J0929--314 scheduled during cycle 5 of {\it
193: Chandra}. Here we report on the results of those observations.
194:
195: The second accretion driven millisecond X-ray pulsar, XTE J1751--305,
196: was discovered on April 3, 2002 (Markwardt et al.~2002). It was found
197: to harbor a neutron star with a spin frequency of 435 Hz which is
198: located in a binary system with an orbital period of 42 minutes
199: (Markwardt et al.~2002). The third accreting millisecond X-ray
200: pulsar, XTE J0929--314, was discovered by Remillard (2002) on April
201: 13--18, 2002, using the all-sky monitor (ASM) aboard the {\it Rossi
202: X-ray Timing Explorer} ({\it RXTE}). Only later (May 2, 2002), when
203: the proportional counter array aboard {\it RXTE} observed the source,
204: was it discovered to be an accreting millisecond X-ray pulsar with a
205: pulse frequency of 185 Hz (Remillard, Swank, \& Strohmayer 2002). The
206: orbital period of the system was found to be 44 minutes (Galloway et
207: al.~2002).
208:
209: \section{Observations, analysis, and results}
210:
211:
212: We observed XTE J0929--314 with {\it Chandra} on March 18, 2004,
213: between 02:30 and 10:02 UT (resulting in an exposure time of
214: $\sim$24.4 ksec) and XTE J1751--305 on June 26, 2004 between 07:59 and
215: 21:12 UT (resulting in an exposure time of $\sim$43.0 ksec) We used
216: the ACIS-S3 CCD and we selected a 1/4 sub-array to limit the pile-up
217: in case the source fluxes exceeded $\sim$10$^{-12}$ ergs s$^{-1}$
218: cm$^{-2}$ (as we will show below, both sources had quiescent fluxes
219: significantly lower than this flux level so no pile-up occurred during
220: our observations). We checked for background flares during our
221: observations, but none were found allowing us to use all available
222: data. Our analyzes were performed using the CIAO software package
223: (version 3.1) and the standard {\it Chandra} analysis
224: threads\footnote{See http://cxc.harvard.edu/ciao/ for CIAO and the
225: analysis threads}.
226:
227: \subsection{XTE J0929--314}
228:
229: \subsubsection{Image analysis}
230:
231: We used the tool 'wavdetect' to search for point sources in our data
232: of XTE J0929--314 and to obtain the coordinates of each source which
233: was detected. For XTE J0929--314, we used the tool 'dmextract' to
234: extract the number of observed photons and the count rate (for the
235: energy range 0.3--8 keV). As source extraction region we used a circle
236: centered on the source position (as obtained with 'wavdetect') with a
237: radius of 1.5$''$. For background extraction region we used an annulus
238: centered on the source position with an inner radius of 5$''$ and an
239: outer radius of $20''$ (a larger outer radius could not be used since
240: the field source CXOU J092919.1--312244 was within $25''$ of XTE
241: J0929--314 as can be seen in Fig.~\ref{fig:image} left panel). We
242: detected four sources in our image, including XTE J0929--314; we will
243: not discuss the other three sources further in this article. Instead
244: we concentrate on the observed X-ray properties of XTE J0929--314 in
245: its quiescent state. The coordinates we obtained for XTE J0929--314
246: are right ascension $09^h ~29^m ~20^s.180$ and declination $-31\degr
247: ~23' ~03''.5$ (epoch J2000.0), which are consistent with the
248: coordinates of the source obtained during outbursts in the X-ray
249: (Juett, Galloway, \& Chakrabarty 2003), optical (Greenhill, Giles, \&
250: Hill 2002), and radio bands (Rupen, Dhawan, \& Mioduszewski 2002; see
251: Fig.~\ref{fig:image} right panel). The total number of source photons
252: detected in the energy range 0.3--8 keV from XTE J0929--314 is 22,
253: resulting in a net time-averaged count rate of $9\pm2 \times 10^{-4}$
254: counts s$^{-1}$ (after background subtraction; less than 0.4
255: background photons are expected in our source extraction region).
256:
257:
258: \subsubsection{Count rate analysis \label{subsubsection:cr_0929}}
259:
260: In Figure~\ref{fig:energy-curve} we plot the energies of the X-ray
261: photons (0.3--8 keV energy range) detected at the source position
262: against the arrival time of the photons (measured since the start of
263: the observation). From this figure it can be seen that 4 photons are
264: detected with energies above 3 keV (note that no photons were detected
265: with energies $>$5 keV) and the remainder of the photons have energies
266: below $\sim$2.5 keV. In the last $\sim$10 ksec of the observation only
267: two source photons are detected compared with twenty photons during
268: the first $\sim$15 ksec; this might suggest that the source exhibited
269: variability during our observation. To investigate if XTE J0929--314
270: was indeed variable during our observation, we applied
271: Kolmogorov-Smirnov and Cramer-Von Mises tests on the event list (as
272: shown in Figure~\ref{fig:energy-curve}) to attempt to disprove the
273: hypothesis that the source count rate is constant. Both tests showed
274: that XTE J0929--314 is variable for energies between 0.3 and 8 keV
275: (although effectively only up to 5 keV since no photons are detected
276: above that energy) at the 95\% confidence level. From
277: Figure~\ref{fig:energy-curve} it can be seen that the four photons
278: which have energies $>$3 keV all arrived within the first 5000 seconds
279: of the observation, which may be the main cause of the variability
280: detected in the source. Therefore, we also applied the above two tests
281: to the event list for photon energies between 0.3 and 2.5 keV. The
282: Kolmogorov-Smirnov test revealed that the 0.3--2.5 keV event list
283: might also be variable but only at a 90\% confidence level. The
284: Cramer-Von Mises test could not find evidence for variability in this
285: energy range. Therefore, we conclude that the variability observed in
286: the source is likely stronger at higher photon energies.
287:
288:
289: \subsubsection{Time-averaged X-ray spectrum \label{subsubsection:spectral_0929}}
290:
291:
292: Despite the low number of photons detected, we extracted the source
293: spectra (averaged over the whole duration of the observation) using
294: the CIAO tool 'psextract' for the energy range 0.3--8 keV, which also
295: created the response matrix and the ancillary response files (the
296: latter was also automatically corrected for the time-variable
297: low-energy quantum efficiency degradation of the CCD). We used a
298: circle with a radius of $1.5''$ as source extraction region. We fitted
299: the spectrum using Xspec version 11.3.0 (Arnaud 1996). The small
300: number of photons observed does not allow for $\chi^2$ statistics to
301: be used during the fits. Therefore, we fitted the data using Cash
302: statistics (Cash 1979). Since Cash statistics cannot be used on
303: background subtracted spectra, we did not subtract the background from
304: our data. The errors thus introduced are likely to be small since less
305: than 0.4 background photons are expected in the source extraction
306: region (see above). The quality of the fits was investigated by
307: generating 10,000 Monte Carlo simulations of the best-fit spectrum: if
308: the fit is good, roughly half the simulations should have values of
309: the Cash statistics which are lower than those of the data.
310:
311: In all our spectral fits we left the interstellar column density
312: $N_{\rm H}$ as a free parameter. To calculate the errors on the
313: obtained fluxes, we fixed each free fit parameter one at a time,
314: either to its minimum or maximum allowed value as obtained from the
315: fits. After that we refitted the data and recalculated the
316: fluxes. This process was then repeated for each free parameter and the
317: final flux range determined the flux errors. The spectra of
318: neutron-star X-ray transients in their quiescent states are usually
319: dominated by a soft thermal component which can adequately be fitted
320: with a neutron-star hydrogen atmosphere model for non-magnetized stars
321: (the NSA model; we used that of Zavlin, Pavlov, \& Shibanov 1996; we
322: also assumed a mass of $1.4~M_\odot$ and a radius of 10 km for the
323: neutron star in XTE J0929--314). In such a model the normalization is
324: given by $1\over d^2$, with $d$ the distance to the source in parsecs.
325: The distance toward XTE J0929--314 is not known but it is constrained
326: to be $>$5 kpc (Galloway et al.~2002). When leaving the normalization
327: free, 88\% of the simulated spectra have better Cash-statistics than
328: the data demonstrating that this model does not provide an accurate
329: description of the data (Fig.~\ref{fig:spectra} {\it bottom}). From
330: this model we obtained an effective temperature $kT_\infty$ (for an
331: observer at infinity) of $0.3\pm0.1$ keV and the column density is
332: $<$$3\times 10^{20}$ cm$^{-2}$. However, the normalization was
333: constrained to be $<$$ 1
334: \times 10^{-11}$, resulting in a distance constrain of $>$316
335: Mpc. This unrealistic distance limit further demonstrates that a
336: simple NSA model does not provide a good fit to the data.
337:
338: We also fitted the data using the NSA model but with the normalization
339: fixed and assuming three distances: 5, 10, or 15 kpc. The results are
340: listed in Table~\ref{tab:spectral_fits} and shown in
341: Figure~\ref{fig:spectra} (top). From this table it can be seen that
342: such a model does not provide an acceptable fit to the data since
343: 100\% of the simulated spectra have lower values for the
344: Cash-statistics than the data themselves. We also fitted a simple
345: blackbody model to the data but again such a model did not provide an
346: adequate fit to the data with a fit quality of 0.94 and a $kT$ of
347: $0.6^{+0.2}_{-0.1}$ keV (and $N_{\rm H}<3 \times 10^{20}$
348: cm$^{-2}$). The radius of the emitting area was $< 0.12 \times {d\over
349: {\rm 10~kpc}}$ km, with $d$ the distance in kpc, much smaller than the
350: expected radius of a neutron star. The inability of the thermal
351: models to provide an adequate fit to the data and the fact that four
352: photons are detected with energies above 3 keV point to a X-ray
353: spectrum which is significantly harder than a simple thermal
354: model. Therefore, we fitted the X-ray spectrum with a power-law model
355: (Fig.~\ref{fig:spectra}; Tab.~\ref{tab:spectral_fits}). The fit
356: quality of 0.68 using such a model is considerably better than for a
357: thermal model (0.68 versus $>$0.88). The fit resulted in a power-law
358: photon index $\Gamma$ of $1.8^{+0.6}_{-0.5}$, a column density of $<6
359: \times 10^{20}$ cm$^{-2}$, and a 0.5--10 keV flux of $6^{+4}_{-2}
360: \times 10^{-15}$ ergs s$^{-1}$ cm$^{-2}$ (corrected for
361: absorption). This gives an unabsorbed 0.5--10 keV X-ray luminosity of
362: $7^{+5}_{-2}
363: \times 10^{31} ({d \over {\rm 10~kpc}})^2$ ergs s$^{-1}$, with $d$ the
364: distance in kpc.
365:
366: To obtain limits on the temperature and luminosity of a possible
367: thermal component in the data, we fitted the spectrum with a power-law
368: component plus a neutron-star atmosphere component. We again fixed the
369: normalization of the neutron-star atmosphere component to correspond
370: to 5, 10, or 15 kpc but left the other parameters free (i.e., the
371: column density, the effective temperature, the photon index, and the
372: normalization of the power-law component). The maximum allowed
373: temperatures and bolometric luminosities are listed in
374: Table~\ref{tab:nsa_limits}, together with the minimum fraction of the
375: 0.5--10 keV flux which is due to the power-law component. The table
376: shows that the maximum allowed $kT_\infty$ is between 0.04 and 0.05
377: keV with a corresponding maximum allowed bolometric luminosity of
378: $0.4-1.7 \times 10^{32}$ ergs s$^{-1}$ (for both parameters the
379: highest values are obtained for the cases which assume the largest
380: distance). For all assumed distances, at least $\sim$70\% of the flux
381: in the 0.5--10 keV range is due to the power-law component. This
382: demonstrates that the power-law component dominates the X-ray emission
383: in the range 0.5--10 keV for XTE J0929--314.
384:
385: \subsubsection{Spectral variability}
386:
387: In \S~\ref{subsubsection:cr_0929} we demonstrated that XTE J0929--314
388: exhibited variability in its count rate at the 95\% confidence
389: level. It is possible that this count rate variability is accompanied
390: by or even due to changes in the X-ray spectrum of the source. If
391: true, the spectral results we obtained in
392: \S~\ref{subsubsection:spectral_0929} only represent a time-averaged
393: view of the spectral properties of the source. To investigate if
394: indeed XTE J0929--314 exhibited spectral variability we calculated
395: X-ray hardness ratios from the first part of the observation (between
396: 0 and 10 ksec after the start of the observation; 13 photons were
397: detected in this time interval) and the second part of the observation
398: (the remaining $\sim$14 ksec of the observation; 9 photons were
399: detected). As X-ray hardness ratio we used the logarithm of the ratio
400: between the number of photons detected with energies $>$1 keV and the
401: number of photons detected with energies $<$1 keV. In
402: Figure~\ref{fig:quantile} ({\it top panel}) we show the hardness ratio
403: versus time. During the second part of the observation the hardness
404: ratio is smaller than during the first part, but the error bars are
405: large and the data are consistent with a non-variable spectrum.
406:
407: Hong, Schlegel, \& Grindlay (2004) have argued that the X-ray hardness
408: ratios normally used in the literature (e.g., those we used in the
409: previous paragraph) are not appropriate for sources with low numbers
410: of counts. Since we have roughly 10 counts to construct our hardness
411: ratios, these ratios might not represent an accurate description of
412: the spectral shape of XTE J0929--314. Therefore, we also calculate
413: so-called ``quantiles'' using the quantile method outlined by Hong et
414: al.~(2004)\footnote{The errors on the quantiles were calculated using
415: the method of Maritz \& Jarret (1978; see Hong et al.~2004) but Hong
416: et al.~(2004) demonstrated, using simulations, that the errors on the
417: quantiles are overestimated when the source counts are very low, as in
418: our case. Therefore, we corrected the quantile errors using the
419: corrections provided by Hong et al.~(2004).}. Hong et al.~(2004) found
420: that the quantity $log_{10} {Q_{50}\over 1 - Q_{50}}$, with $Q_{50}$
421: the median quantile, provided a good indication for the spectral
422: hardness of a source. Therefore, in the bottom panel of
423: Figure~\ref{fig:quantile} we plot this quantity as a function of
424: time. Similar to the hardness ratios, the error bars on the quantiles
425: are such that a constant spectral shape throughout the observation
426: cannot be excluded.
427:
428: As a last way of investigating possible spectral variability, we
429: extracted the X-ray spectrum for each of the two time intervals used
430: above. The results of the spectral fitting are listed in
431: Table~\ref{tab:variability}. When fitting the two data sets
432: separately, a power-law model produces acceptable fits (with fit
433: quality of 0.6--0.7) with photon indices of 1.3$^{+0.9}_{-0.7}$ and
434: 2.6$^{+1.2}_{-0.9}$ for the first and second part of the observation,
435: respectively. The values again indicate that the source spectrum had
436: softened during the course of the observation, although the indices
437: are consistent with each other (within the errors). The 0.5--10 keV
438: X-ray flux decreased by a factor of 3--4 during the observation indeed
439: suggesting variability in the brightness of the source (note, again
440: the errors are large and only minimal variability might have been
441: present). To further investigate how significant the softening of the
442: X-ray spectrum is, we fitted the two data sets simultaneously. During
443: this fit, we assumed that the column density did not vary during the
444: observation and we tied it between the two data sets. The fit results
445: were consistent with those obtained when fitting the data sets
446: separately (Tab.~\ref{tab:variability}). In Figure~\ref{fig:contour}
447: we plot the confidence regions of the photon indices, showing that the
448: two indices are only different from each other at the $<$90\%
449: confidence level. This is illustrated in the figure by the line of
450: equal index between two data sets which intersects the 90\% confidence
451: contour. We conclude that the limited statistics does not allow for
452: any strong conclusions on the possible softening of the X-ray spectrum
453: suggested by Figure~\ref{fig:energy-curve}.
454:
455: We investigated if a thermal model could fit the two data sets, but
456: when using a NSA model with fixed normalization (to a distance of 5,
457: 10, or 15 kpc), the fit qualities were always 1.00 for the first part
458: of the observation and between 0.89 and 0.94 (larger for smaller
459: assumed distance) for the second part of the observation. Also a
460: simple black-body model does not produce a good fit with fit qualities
461: of 0.85 and 0.87 for the first and second part of the observation,
462: respectively. This clearly demonstrates that during the whole
463: observation, the source had an X-ray spectral shape which was
464: significantly different (i.e., harder) from a simple thermal
465: model. This confirms our conclusions in
466: \S~\ref{subsubsection:spectral_0929}. Investigations into whether or
467: not the fractional contribution of the power-law component to the
468: 0.5--10 keV flux changed during the observation did not result in
469: useful insights due to the limitation on the statistics of the
470: individual data sets.
471:
472:
473: \subsection{XTE J1751--305 \label{subsection:1751}}
474:
475: We begin a similar analysis for XTE J1751--305 by searching for point
476: sources in our data (using 'wavdetect'). We detected 34 sources in our
477: image but XTE J1751--305 itself was not detected (a discussion of the
478: properties of these 34 sources, unrelated to XTE J1751--305, is beyond
479: the scope of this article). In Figure~\ref{fig:image_1751} we show the
480: 0.5--7 keV X-ray image near the position of XTE J1751--305. Several
481: point sources are detected close to the position of XTE J1751--305
482: (obtained during outburst using {\it Chandra}; Markwardt et al.~2002),
483: but only 1 photon was detected (in the energy range 0.5--7 keV) in a
484: circle with a radius of 1$''$ around the position of XTE J1751--305
485: (for {\it Chandra}'s point-spread-function, 1$''$ corresponds to the
486: encirclement of 90\% of the energy), which does not constitute a
487: significant detection. When using also the data at lower energies
488: (i.e., using the energy range 0.3--7.0 keV), we still could only
489: detect 2 photons within an 1$''$ radius. When using a larger radius of
490: 1.5$''$, 3 photons are detected for the energy range 0.5--7 keV, and 5
491: photons for the energy range 0.3--7 keV. However, in these cases more
492: than half the possible source photons would fall outside the circle
493: which should encircle $>$90\% of the energy, which is
494: unlikely. Moreover, the positions of the two softest photons (those
495: with energies $<$0.45 keV) were consistent with the position of a
496: bright $K$ star close to the position of XTE J1751--305 (1.2$''$ away;
497: the two photons have a distance of 0.5$''$ and 0.8$''$ from the
498: position of this star). Thus these two photons could have possibly
499: originated from this star and not from XTE J1751--305, especially
500: because the high column density toward the pulsar ($\sim$$10^{22}$
501: cm$^{-2}$; Miller et al.~2003) would make unlikely any detection of
502: its photons which have energies below 0.5 keV.
503:
504: The closest, significantly detected X-ray source (the one to the
505: south-east of XTE J1751--305) is located 3.8$''$ away\footnote{The
506: coordinates of this source are: right ascension $17^h ~51^m
507: ~13^s.673$, declination $-30\degr ~37' ~26''.40$. The coordinates are
508: for epoch J2000.0 and have errors of $0.6''$ (90\% confidence
509: levels).} from XTE J1751--305. The absolute positional accuracy of
510: {\it Chandra} is 0.6$''$ (90\% confidence level; $0.8''$ for 99\%
511: confidence level\footnote{See
512: http://cxc.harvard.edu/cal/ASPECT/celmon/}) strongly suggesting that
513: this south-east source is not the quiescent counterpart of XTE
514: J1751--305. It might be possible that during our observation the
515: positional accuracy of {\it Chandra} was (for some reason) extremely
516: inaccurate. To test this hypothesis, we compared our X-ray image with
517: the $K$-band image of the region around XTE J1751--305 published by
518: Jonker et al. (2003). Of the 13 X-ray sources located in the region
519: covered by this $K$-band image, 8 have $K$ objects located within a
520: radius of 0.6$''$ (including the X-ray source located 3.8$''$ away
521: from XTE J1751--305 as well as the four other reasonably bright X-ray
522: point sources visible in Fig.~\ref{fig:image_1751}). We simulated
523: 10,000 X-ray point sources with random positions in the field of the
524: $K$ image. Only $<$7\% of the simulated sources have a $K$ star (of
525: similar or higher brightness than the potential $K$ counterparts of
526: the detected X-ray point sources) falling within a 0.6$''$ radius of
527: their positions. The coincidence rate between the detected X-ray
528: sources and the $K$ stars is significantly higher ($\sim$60\%) than
529: the simulated one demonstrating that most of the $K$ stars located in
530: the X-ray error circles are indeed the infrared counterparts of those
531: X-ray sources. Our analysis strongly suggests that the positional
532: accuracy of our X-ray image is within the standard accuracy quoted for
533: {\it Chandra}. Therefore, we conclude that XTE J1751--305 was not
534: conclusively detected during our {\it Chandra} observation.
535:
536: To calculate a flux upper limit for XTE J1751--305, we assume that
537: $<$5 photons\footnote{Using 5 photons as an upper limit can be
538: interpreted (Gehrels 1986) as a 97\% or 90\% confidence upper limit
539: for 1 or 2 detected photons from the source, respectively, which is
540: the number of photons we possible detected from XTE J1751--305.} (for
541: the energy range 0.5--7 keV) were detected from the source, resulting
542: in a count rate upper limit of 1.16 $\times 10^{-4}$ counts s$^{-1}$
543: (0.5--7 keV). We used PIMMS\footnote{Available
544: http://heasarc.gsfc.nasa.gov/Tools/w3pimms.html} to calculate the flux
545: upper limits assuming different spectral model. As column density we
546: used the value as observed during outburst ($N_{\rm H} = 9.8 \times
547: 10^{21}$ cm$^{-2}$; Miller et al.~2003) and we assume a black-body
548: spectral shape\footnote{Although a neutron-star atmosphere model is
549: usually fitted to the X-ray spectra of quiescent neutron star X-ray
550: transients, such a model cannot be used in PIMMS and we had to resort
551: to the black-body model as an approximation.} (with temperatures $kT$
552: of 0.05--0.7 keV) or a power-law spectral shape (with a photon index
553: $\Gamma$ of 0--4). Using a black-body model, we find that the 0.5--10
554: keV flux upper limit $F_{\rm upper}$ (in units of ergs s$^{-1}$
555: cm$^{-2}$) can be accurately approximated by
556:
557: \begin{equation}
558: log_{10} F_{\rm upper} =
559: -14.787 + 4.09 e^{-{kT\over 0.0827~{\rm keV}}}
560: \label{eq:bblimit}
561: \end{equation}
562:
563: \noindent
564: Using a power-law model, we find
565:
566: \begin{equation}
567: F_{\rm upper} = 5.76\times 10^{-15} - 3.05 \times
568: 10^{-15} \Gamma + 7.41\times 10^{-16} \Gamma^2
569: \label{eq:pllimit}
570: \end{equation}
571:
572:
573: \noindent
574: For a reasonable range in $kT$ of 0.1--0.3 keV (as observed in other
575: quiescent neutron star X-ray transients), we find 0.5--10 keV upper
576: limits of $0.2 - 2.7 \times 10^{-14}$ ergs s$^{-1}$ cm$^{-2}$ (the
577: highest upper limits are obtained when the assumed $kT$ is
578: smallest). For a reasonable $\Gamma$ of 1--3, the 0.5--10 keV flux
579: upper limit is $2.6-3.5\times 10^{-15}$ ergs s$^{-1}$ cm$^{-2}$ (the
580: minimum upper limit is obtained when $\Gamma \sim$2).
581:
582: \section{Discussion}
583:
584: We have observed the accretion-driven millisecond X-ray pulsars XTE
585: J0929--314 and XTE J1751--305 in their quiescent states. This triples
586: the number of such sources to be observed in their quiescent states
587: (the first one being SAX J1808.4--3658; Stella et al.~2000; Dotani,
588: Asai, \& Wijnands 2000; Campana et al.~2002). We could not detect XTE
589: J1751--305 during our observations, which resulted in upper limits on
590: the quiescent X-ray flux (0.5--10 keV; corrected for absorption) of
591: this system of $2.6-3.5\times 10^{-15}$ ergs s$^{-1}$ cm$^{-2}$ when
592: we assume that the source exhibited a power-law spectral shape with
593: photon index between 1 and 3, or $0.2 - 2.7 \times 10^{-14}$ ergs
594: s$^{-1}$ cm$^{-2}$ when assuming a black-body spectral shape with $kT$
595: of 0.1--0.3 keV. We conclusively detected XTE J0929--314, but we
596: detected only 22 photons (energy range 0.3--8 keV) during our
597: observation which limits any detailed analysis. The X-ray spectrum
598: could not be fitted with a simple thermal model alone (such as a
599: black-body or a neutron-star atmosphere model) but a single power-law
600: model provided an adequate fit to the data resulting in a 0.5--10 keV
601: X-ray luminosity of $7^{+5}_{-2} \times 10^{31} ({d \over {\rm
602: 10~kpc}})^2$ ergs s$^{-1}$ (with $d$ in kpc). The upper limits on the
603: effective temperature of a thermal component ranged from 0.04 to 0.05
604: keV with corresponding upper limits on the bolometric luminosity of
605: 0.4--1.7 $\times 10^{32}$ ergs s$^{-1}$ (both quantities are for an
606: observer at infinity). We determined that at most $\sim$30\% of the
607: 0.5--10 keV flux could have come from such a thermal component. We
608: could demonstrate that the source was variable at the 95\% confidence
609: level but, within the limitation of the statistics, we could not find
610: evidence for spectral variability during our observation. Next, we
611: will first discuss what can be inferred from the results of XTE
612: J0929--314 before we discuss the non-detection of XTE J1751--305.
613:
614:
615: \subsection{The power-law component of XTE J0929--314
616: \label{subsection:0929_pl}}
617:
618: Our results suggest that the X-ray emission of XTE J0929--314 in its
619: quiescent state is dominated by a power-law component and not by a
620: thermal component as is usually observed during the quiescent states
621: of neutron-star X-ray transients. This makes XTE J0929--314 similar to
622: SAX J1808.4--3658, which was until now the only accreting millisecond
623: pulsar to have been observed in its quiescent state. In quiescence,
624: SAX J1808.4--3658 was found to have a 0.5--10 keV X-ray luminosity of
625: $\sim$$5\times10^{31}$ ergs s$^{-1}$ when fitted with a power-law
626: model with index of $1.5\pm0.3$ (Campana et al.~2002). This is rather
627: similar to the 0.5--10 keV luminosity ($\sim 7 \times 10^{31}$ ergs
628: s$^{-1}$ for a distance of 10 kpc) and power-law index
629: ($1.8^{+0.6}_{-0.5}$) of XTE J0929--314. Furthermore, the quiescent
630: X-ray emission between 0.5 and 10 keV was in both sources dominated by
631: the power-law component ($>$90\% in SAX J1808.4--3658 and $>$70\% in
632: XTE J0929--314). Although our statistics on the XTE J0929--314 data
633: are poor, the only difference between the two sources seems to be that
634: SAX J1808.4--3658 has a somewhat lower luminosity in quiescence and
635: perhaps a bit harder spectrum. We note that the errors on the
636: parameters also suggest that both sources could have equal
637: luminosities and spectral hardness. Furthermore, XTE J0929--314
638: exhibited luminosity variability during our observation possibly
639: accompanied by (or even due to) spectral variability. This means that
640: XTE J0929--314 could be at times intrinsically fainter than SAX
641: J1808.4--3658 or have a harder spectrum (note that it is less likely
642: that XTE J0929--314 was fainter {\em and} harder than SAX
643: J1808.4--3658, since the indication is that the hardest spectrum was
644: observed when XTE J0929--314 was brightest).
645:
646: The results on XTE J0929--314 and SAX J1808.4--3658 might suggest that
647: a power-law dominated X-ray spectrum in quiescence and a low quiescent
648: 0.5--10 keV luminosity are common properties of accretion-driven
649: millisecond X-ray pulsars. However, recent work (as discussed below)
650: on two other weak quiescent neutron-star X-ray transients, which do
651: not exhibit pulsations, suggests that this might instead be a property
652: of many weak quiescent neutron-star X-ray transients in general and
653: not only of the accreting millisecond X-ray pulsars. Jonker et
654: al.~(2004a) and Tomsick et al.~(2004) reported on quiescent X-ray
655: observations of SAX J1810.8--2609 and XTE J2123--058,
656: respectively. For both sources it was found that their quiescent X-ray
657: spectra could be fitted with a power-law model with indices of $\sim$3
658: and an 0.5--10 keV X-ray luminosity around $\sim$$10^{32}$ ergs
659: s$^{-1}$. No thermal component could be detected with a maximum
660: contribution to the 0.5--10 keV flux of $\sim$50\% in SAX
661: J1810.8--2609 (Jonker et al.~2004a) and $\sim$60\%--70\% in XTE
662: J2123--058 (Tomsick et al.~2004). These luminosities are somewhat
663: higher than those of SAX J1808.4--3658 and XTE J0929--314 and the
664: X-ray spectra somewhat softer, but they are among the lowest
665: luminosity quiescent neutron-star X-ray transients for which no X-ray
666: pulsations were seen during their outbursts.
667:
668: Jonker et al.~(2004a, b) used the results obtained for SAX
669: J1808.4--3658, SAX J1810.8--2609, and XTE J2123--058 together with
670: those obtained for the other quiescent neutron star X-ray transients,
671: to show evidence that, in their quiescent spectra, the contributions
672: of the power-law components to the 0.5--10 keV flux seem to increase
673: (from at most several tens of percents to over 90\%) when the
674: quiescent source luminosities decrease from $\sim$$10^{33}$ ergs
675: s$^{-1}$ to approximately $5\times 10^{31}$ ergs s$^{-1}$. They also
676: found a similar trend for higher quiescent luminosities of the
677: transients: the fractional power-law contribution also increases when
678: the quiescent source luminosities increase from the comparison
679: luminosity of $1-2 \times 10^{33}$ ergs s$^{-1}$. Our results show
680: that XTE J0929--314 fits between SAX J1808.4--3658 on the
681: low-luminosity side, and SAX J1810.8--2609 and XTE J2123--058 on the
682: higher-luminosity side, which provides additional support to the
683: possible correlations found by Jonker et al.~(2004a, b). This result
684: would suggest that hard quiescent emission is not a feature unique to
685: accretion-driven millisecond X-ray pulsars but rather a general
686: feature of quiescent neutron-star X-ray transients when they are at
687: low quiescent luminosities (with increasingly harder spectra when the
688: luminosities get lower). The current data still suggest that the
689: accretion-driven millisecond X-ray pulsars might have slightly lower
690: luminosities and harder spectra than the other sources, but it remains
691: to be seen if this holds when additional neutron-star X-ray transients
692: are observed in their quiescent states. Additional neutron star X-ray
693: transients need to be studied in their quiescent states to better
694: understand the power-law spectral component in the quiescent
695: spectra. Such studies might also clarify if the quiescent transient in
696: Terzan 5 (EXO 1745--248; Wijnands et al.~2004), which does not follow
697: this correlation (Jonker et al.~2004b), is unique among the quiescent
698: neutron star systems, or if it is the first example of many similar
699: systems which would suggest that this correlation is spurious.
700:
701: \subsection{The non-detection of a thermal component in XTE J0929--314}
702:
703: We can investigate whether our upper limits on the contribution of the
704: thermal component to the quiescent spectrum of XTE J0929--314 are
705: consistent with what would be expected from the cooling neutron star
706: model proposed by Brown et al.~(1998). The quiescent thermal
707: luminosity predicted by this model depends on the time-averaged (over
708: $>$10,000 years) accretion rate of the source. For XTE J0929--314 we
709: can estimate its time-averaged accretion rate in two ways; either by
710: using the outburst fluence of the source and the limited information
711: about the outburst recurrence time, or by assuming that the mass
712: transfer is solely due to gravitational radiation and all matter is
713: eventually accreted onto the neutron star.
714:
715: \subsubsection{Constraints obtained when using the 2002 outburst fluence
716: \label{subsubsection:0929_outburstfluence}}
717:
718: According to Galloway et al.~(2002), the 2--60 keV fluence of the 2002
719: outburst of XTE J0929--314 was $4.2\times 10^{-3}$ ergs cm$^{-2}$,
720: with an estimated bolometric correction of $\sim$2.34. This results in
721: a bolometric fluence of $\sim$$9.8\times 10^{-3}$ ergs
722: cm$^{-2}$. During the lifetime of {\it RXTE} (which is at present 8.5
723: years), only 1 outburst has been observed, resulting in a time
724: averaged accretion flux $\langle F_{\rm acc} \rangle$ of $<3.7\times
725: 10^{-11}$ ergs s$^{-1}$ cm$^{-2}$. Using the Brown et al.~(1998) model
726: and assuming standard core cooling, the expected quiescent flux
727: $F_{\rm q}$ is approximately given by $F_{\rm q} = {\langle F_{\rm
728: acc} \rangle\over 135}$ (Brown et al.~1998; Wijnands et al.~2001;
729: Rutledge et al.~2002). For XTE J0929--314, this results in $F_{\rm q}
730: < 2.7 \times 10^{-13}$ ergs s$^{-1}$ cm$^{-2}$, which is consistent
731: with the upper limits on the bolometric fluxes of a possible
732: neutron-star atmosphere component in the quiescent spectra (which are
733: $<$1.5 $\times 10^{-14}$ ergs s$^{-1}$ cm$^{-2}$). However, in this
734: scenario the source would have to remain quiescent for a significant
735: period (i.e., much longer than the 8.5 years we assumed here) to allow
736: the predicted limits to come down to the measured flux limits.
737:
738: Using our measured flux limits, we can actually estimate how long XTE
739: J0929--314 has to be in quiescence for the neutron star surface to be
740: as cool as we measure if the neutron star cools down using standard
741: core cooling processes. From $F_{\rm q} = {\langle F_{\rm acc}
742: \rangle\over 135}$ it can be derived (Wijnands et al.~2001) that $F_{\rm
743: q} \approx {t_{\rm o} \over t_{\rm o} + t_{\rm q}} \times {\langle
744: F_{\rm o} \rangle \over 135} $, with $\langle F_{\rm o} \rangle$ the
745: averaged flux during outburst, $t_{\rm o}$ the averaged time the
746: source is in outburst, and $t_{\rm q}$ the averaged time the source is
747: in quiescence. From Galloway et al.~(2002), it follows that $t_{\rm o}
748: \approx 73$ days (if the 2002 outburst had a duration which is typical
749: for the source) and $\langle F_{\rm o} \rangle =1.6
750: \times 10^{-9}$ ergs s$^{-1}$ cm$^{-2}$. Using our limits on $F_{\rm
751: q}<1.5 \times 10^{-14}$ ergs s$^{-1}$ cm$^{-2}$, then $t_{\rm q}>
752: 5.8\times 10^4$ days, resulting in $t_{\rm q}> 160 $ years. Although
753: this predicted quiescent period is quite long compared to the known
754: recurrence time of several other neutron-star X-ray transients
755: (ranging from less than 1 year up to several decades; e.g., Chen,
756: Shrader, \& Livio 1997), the disk instability model proposed for the
757: outburst behavior of X-ray transients (Lasota 2001) might be able to
758: accommodate such long quiescent episodes. We note that due to the very
759: low peak fluxes during the 2002 outburst, we cannot rule out that,
760: prior to the launch of {\it RXTE}, one or more outbursts occurred
761: during the last four decades (i.e., since the birth of X-ray
762: astronomy). Weak outbursts, such as the 2002 outburst of XTE
763: J0929--314, could have easily been missed by X-ray satellites other
764: than {\it RXTE}.
765:
766:
767: \subsubsection{Constraints obtained when assuming mass
768: transfer solely via gravitational radiation}
769:
770: Galloway et al.~(2002) stated that if the mass transfer in XTE
771: J0929--314 is driven by gravitational radiation then the time-averaged
772: mass transfer rate $\dot{M}_{\rm GR}$ in this system should be
773:
774:
775:
776: \begin{equation}
777: \dot{M}_{\rm GR} = 5.5 \times 10^{-12} ({ M_{\rm ns} \over
778: 1.4~M_\odot})^{2\over 3} ({M_{\rm c} \over 0.01~M_\odot})^2
779: ~M_\odot~{\rm yr}^{-1} \label{equation:gw}
780: \end{equation}
781:
782: \noindent
783: with $M_{\rm ns}$ the mass of the neutron star and $M_{\rm c}$ the
784: mass of the companion star (both in solar masses). From the model by
785: Brown et al.~(1998; see also Rutledge et al.~2002) the quiescent
786: luminosity $L_{\rm q}$ can be estimated via
787:
788: \begin{equation}
789: L_{\rm q} = 8.7 \times 10^{33} ( {\langle \dot{M} \rangle \over
790: 10^{-10} M_\odot~{\rm yr}^{-1}} ) {Q \over 1.45~{\rm MeV}} {\rm
791: ~ergs~s}^{-1} \label{equation:lq}
792: \end{equation}
793:
794: \noindent
795: with $\langle \dot{M} \rangle$ the time-averaged mass accretion rate
796: onto the neutron star, and $Q$ the amount of heat deposited in the
797: crust per accreted nucleon (e.g., Haensel \& Zdunik 1990). If we
798: assume that the matter transfer from the companion star is solely
799: driven by gravitational radiation and that all the matter transferred
800: from the companion star is eventually accreted onto the neutron star
801: then $\langle \dot{M} \rangle = \dot{M}_{\rm GR}$. Substituting
802: equation~\ref{equation:gw} in equation~\ref{equation:lq} gives
803:
804: \begin{equation}
805: L_{\rm q} = 4.8 \times 10^{32} ({ M_{\rm ns} \over 1.4
806: M_\odot})^{2\over 3} ({M_{\rm c} \over 0.01 M_\odot})^2 {Q \over
807: 1.45~{\rm MeV}} {\rm ~ergs~s}^{-1}
808: \end{equation}
809:
810: \noindent
811: If $ M_{\rm ns} =1.4~M_\odot$, then Galloway et al.~(2002) obtained a
812: minimum mass for the companion star of $0.008~M_\odot$ (thus $M_{\rm
813: c}>0.008~M_\odot$). If we further assume $Q= 1.45$ MeV, then $L_{\rm
814: q} > 3 \times 10^{32}$ ergs s$^{-1}$. Only for large distances ($>$15
815: kpc) can the upper-limits on any contribution of a thermal component
816: be consistent with this lower-limit on the quiescent luminosity (if
817: $M_{\rm c}$ is not much greater than the minimum allowed value of
818: 0.008 $M_\odot$). For smaller distances the predicted quiescent
819: luminosity for a thermal component is a factor 2--8 times too high
820: (this discrepancy becomes worse if $M_{\rm c}$ becomes larger). Firm
821: conclusions can only be obtained when the exact luminosity of the
822: thermal component can be measured with better data. However, our
823: results already suggest that the quiescent luminosity predicted is
824: higher than that measured if we assume that the matter transfer from
825: the companion star is solely driven by gravitational radiation and
826: that all the matter transferred is accreted onto the neutron
827: star. Possible reasons for this discrepancy are that not all the
828: matter transferred is accreted onto the neutron star or that enhanced
829: cooling processes occur in the neutron-star core.
830:
831: \subsection{The non-detection of XTE J1751--305}
832:
833: Despite the fact that we could not conclusively detect XTE J1751--305
834: during our observation, the limits on the X-ray flux of this source
835: can still be used to provide further insight into the quiescent
836: emission of neutron star X-ray transients. We do not know the spectral
837: shape of the quiescent emission of this system and therefore in
838: \S~\ref{subsection:1751} we calculated the flux upper limits assuming
839: two different spectral models (either a black-body or a power-law
840: model; equations~\ref{eq:bblimit} and~\ref{eq:pllimit}). The limits
841: obtained using the black-body model are most useful for constraining
842: the cooling neutron star model which assumes that the emission is due
843: to thermal radiation from the neutron-star surface. We will attempt to
844: constrain this model once again taking into account the limited
845: observational knowledge about the time-averaged accretion rate of the
846: source (\S~\ref{subsubsection:1751_fluence}) and assuming that the
847: mass transfer is solely due to gravitational radiation and that all
848: matter will eventually be accreted by the neutron star
849: (\S~\ref{subsubsection:1751_gr}) . Here we first discuss the
850: possibility that the source emitted a power-law shaped quiescent X-ray
851: spectrum.
852:
853: The two other accretion-driven millisecond X-ray pulsar which have
854: been observed in quiescence have shown a power-law dominated quiescent
855: X-ray spectrum with photon indices of 1.5$\pm0.3$ (SAX J1808.4--3658;
856: Campana et al.~2002) and 1.8$\pm0.6$ (XTE J0929--314;
857: Tab.~\ref{tab:spectral_fits}). If we assume that XTE J1751--305 has
858: also a power-law shaped quiescent spectrum, then, using
859: equation~\ref{eq:pllimit}, we get 0.5--10 keV flux upper limits on the
860: quiescent emission of $2.5- 3.5 \times 10^{-15}$ ergs s$^{-1}$
861: cm$^{-2}$ (assuming a range of 1--3 for the photon index, which covers
862: the allowed values of the photon indices for SAX J1808.4--3658 and XTE
863: J0929--314), giving 0.5--10 keV luminosity upper limits of $1.9 - 2.7
864: \times 10^{31} ({d\over 8 {\rm ~kpc}})^2$ ergs s$^{-1}$ ($d$ in
865: kpc). If the distance toward XTE J1751--305 is $<$11 kpc (thus the
866: source is located in the Galactic center region) and the quiescent
867: spectrum follows a power-law shape, then XTE J1751--305 would be the
868: intrinsically faintest quiescent neutron-star X-ray transients so far
869: known. However, if the source is located at a larger distance, SAX
870: J1808.4--3658 would still remain the faintest quiescent system (with a
871: luminosity of $\sim 5 \times 10^{31}$ ergs s$^{-1}$; Campana et
872: al.~2002), unless it can be confirmed that XTE J0929--314 can become
873: at times fainter than SAX J1808.4--3658 (see
874: \S~\ref{subsection:0929_pl}).
875:
876:
877:
878:
879: \subsubsection{Constraints obtained when using the 2002 outburst fluence
880: \label{subsubsection:1751_fluence}}
881:
882: According to Markwardt et al.~(2002), the 2--200 keV fluence of the
883: 2002 outburst of XTE J1751--305 was $2.5 \times 10^{-3}$ ergs
884: cm$^{-2}$. Since no bolometric correction factor is currently
885: available, we will use this fluence, but we note that the bolometric
886: fluence could be significantly higher (a factor of a few) and thus
887: also the predicted quiescent thermal flux. Markwardt et al.~(2002)
888: found a previous outburst of the source in June 1998, resulting in a
889: recurrence time of $\sim$3.8 years. If this is the typical recurrence
890: time of the source, then the time-averaged accretion rate will be
891: $\langle F_{\rm acc} \rangle \sim 2.1 \times 10^{-11}$ ergs s$^{-1}$
892: cm$^{-2}$. Since the predicted quiescent flux is given by $F_{\rm q} =
893: {\langle F_{\rm acc} \rangle\over 135}$
894: (\S~\ref{subsubsection:0929_outburstfluence}), this results in $F_{\rm
895: q} = 1.6 \times 10^{-13}$ erg s$^{-1}$ cm$^{-2}$, which is
896: significantly larger than the maximum bolometric flux ($2.7
897: \times 10^{-14}$ ergs s$^{-1}$ cm$^{-2}$) we obtained for a possible
898: thermally shaped quiescent spectrum. It might be possible that the
899: average recurrence time of the source is larger than the 3.8 years
900: found for the two outbursts so far observed. Using the method outlined
901: in \S~\ref{subsubsection:0929_outburstfluence}, we find that the
902: source must have typical quiescent episodes of $>$22 years for the
903: predicted quiescent flux to come down to the calculated flux
904: limits. Such a recurrence time scale is not uncommon among the
905: neutron-star X-ray transients. Another possibility for the discrepancy
906: between the predicted quiescent flux and the limits on the actual flux
907: is that the neutron star in XTE J1751--305 is colder than expected due
908: to enhanced cooling in its core.
909:
910: \subsubsection{Constrains obtained when assuming mass transfer solely
911: via gravitational radiation \label{subsubsection:1751_gr}}
912:
913: Similar to XTE J0929--314, it also has been suggested for XTE
914: J1751--305 (Markwardt et al.~2002) that the mass transfer might be due
915: to gravitational radiation. If true, then $\dot{M}_{\rm GR}$ in this
916: system should be (Markwardt et al.~2002)
917:
918: \begin{equation}
919: \dot{M}_{\rm GR} = 1.2 \times 10^{-11} ({ M_{\rm ns} \over
920: 1.4~M_\odot})^{2\over 3} ({M_{\rm c} \over 0.0137~M_\odot})^2
921: ~M_\odot~{\rm yr}^{-1}. \label{equation:gw2}
922: \end{equation}
923:
924: \noindent
925: If we assume that the matter transfer is solely driven by
926: gravitational radiation and that all the matter is eventually accreted
927: onto the neutron star then we can again set $\langle \dot{M} \rangle =
928: \dot{M}_{\rm GR}$. Substituting equation~\ref{equation:gw2} in
929: equation~\ref{equation:lq} gives for the predicted quiescent
930: luminosity (again assuming $Q = 1.45$ MeV)
931:
932: \begin{equation}
933: L_{\rm q} = 1.0 \times 10^{33} ({ M_{\rm ns} \over 1.4
934: M_\odot})^{2\over 3} ({M_{\rm c} \over 0.0137 M_\odot})^2 {\rm
935: ~ergs~s}^{-1}.
936: \end{equation}
937:
938: \noindent
939: Markwardt et al~(2002) obtained a minimum mass for the companion star
940: of $M_{\rm c} > 0.0137~M_\odot$, when assuming a neutron star mass of
941: $1.4~M_\odot$. Using these values for the masses of the companion and
942: the neutron star, we find that the predicted quiescent luminosity is
943: given by $L_{\rm q} > 1\times 10^{33}$ ergs s$^{-1}$. The 0.5--10 keV
944: flux upper limit assuming a black body spectral model is given by
945: equation~\ref{eq:bblimit}. If $kT> 0.1$ keV (which is typically
946: observed for other neutron star X-ray transients in their quiescent
947: states), this gives a 0.5--10 keV luminosity upper limit of $>2\times
948: 10^{32} {d \over ~8~{\rm~kpc}}$ ergs s$^{-1}$ ($d$ in kpc). The
949: bolometric luminosity limit could be a factor of a few larger. Only
950: for relatively large distances ($>10-15$ kpc; depending on the
951: bolometric correction factor) can the measured upper limit on the
952: luminosity become consistent with the predicted luminosity. For
953: effective temperatures below 0.1 keV, the bolometric luminosity as
954: calculated via $L_{\rm bol} = 4\pi
955: \sigma R^2 T_{\rm eff}^4$ (with $R$ and $T$ the neutron star radius
956: and surface temperature for an observer at infinity and $\sigma$ the
957: Stefan-Boltzmann constant) will become quite low and will not be
958: consistent with the predicted luminosity. This indicates that the
959: effective temperature cannot be too low in order to be consistent with
960: the predicted quiescent luminosity.
961:
962: \subsection{Constraining the magnetic field of the neutron stars in
963: XTE J0929--314 and XTE J1751--305}
964:
965: Recently, Burderi et al.~(2002) and Di Salvo \& Burderi (2003)
966: constrained the magnetic field strengths of the neutron stars in
967: several neutron-star X-ray transients (i.e., KS 1731--260, SAX
968: J1808.4--3658 and Aql X-1) based on their measured quiescent
969: luminosities and on the knowledge of the spin rate of their neutron
970: stars\footnote{SAX J1808.4--3658 exhibits coherent oscillations in its
971: persistent X-ray emission during outbursts which reflect the spin rate
972: of the neutron star (Wijnands \& van der Klis 1998). KS 1731--260 and
973: Aql X-1 exhibit nearly-coherent oscillations during thermonuclear
974: flashes on the neutron star surfaces (Smith, Morgan, \& Bradt 1997;
975: Zhang et al.~1998). Those oscillations are likely to be directly
976: related to the neutron star spin rate as well (see, e.g., Strohmayer
977: \& Bildsten 2004 for a recent review).}. To constrain the magnetic
978: field strength, they used two proposed mechanisms which might produce
979: X-rays in quiescence: {\it a}) residual accretion onto the neutron
980: star at a very low accretion rate; {\it b}) neutron star rotational
981: energy which is converted into radiation (a fraction of which might be
982: released in X-rays) because of the presence of a rotating magnetic
983: dipole. They note that thermal emission from a cooling neutron star
984: which is heated during outburst might also contribute to the quiescent
985: X-ray emission and that the measured quiescent luminosities are
986: therefore an upper limit on the X-ray contribution due to the above
987: discussed scenarios.
988:
989: They divided scenario {\it a} into two possibilities. The first ({\it
990: a1}) assumes that the radius of the magnetosphere is inside the
991: co-rotation radius and accretion onto the neutron star surface is
992: possible; the second ({\it a2}) assumes that the magneto-spheric
993: radius is outside the co-rotation radius but inside the radius of the
994: light cylinder and accretion onto the neutron star is not possible but
995: an X-ray emitting accretion disk might still exist with an inner
996: radius larger than the co-rotation radius. Scenario {\it b} can also
997: be divided into two possibilities. The first ({\it b1}) assumes that
998: the X-ray emission in quiescence is due to reprocessing of part of the
999: bolometric luminosity from the rotating neutron star into X-rays in
1000: the shock front between the pulsar wind and the circumstellar matter;
1001: the second ({\it b2}) assumes that the X-ray emission is intrinsic
1002: emission from the radio pulsar.
1003:
1004: Following the method outlined by Burderi et al.~(2002) and Di Salvo \&
1005: Burderi (2003) and using the maximum quiescent luminosity from our
1006: fits ($1.2\times 10^{32} ({d \over {\rm 10~kpc}})^2$ ergs s$^{-1}$,
1007: with $d$ in kpc) and the spin rate (185 Hz; Galloway et al.~2002) of
1008: XTE J0929--314, we found that the magnetic field strength of the
1009: neutron star in this system should be $<2\times 10^7 {d \over {\rm
1010: 10~kpc}}$ Gauss (scenario {\it a1}), $<3\times 10^{9} {d \over {\rm
1011: 10~kpc}}$ Gauss ({\it a2}), $<5 \times 10^8 {d \over {\rm 10~kpc}}$
1012: Gauss ({\it b1}), and $<3\times 10^9 ({d \over {\rm 10~kpc}})^{0.76}$
1013: Gauss ({\it b2}). Therefore, from these four scenarios the
1014: magnetic field strength of the neutron star in XTE J0929--314 can be
1015: constrained to be $<3\times 10^9 {d \over {\rm 10~kpc}}$ Gauss.
1016:
1017: We also calculated the limits on the magnetic field strength of the
1018: neutron star in XTE J1751--305. The maximum flux limit of $2.7\times
1019: 10^{-14}$ ergs s$^{-1}$ cm$^{-2}$ we obtained was for a spectral model
1020: which we assumed to be a simple black body shape, resulting in a
1021: luminosity limit of $2\times 10^{32} {d \over 8 {\rm ~kpc}}$ ergs
1022: s$^{-1}$ (with $d$ in kpc). Using the spin frequency of 435 Hz
1023: (Markwardt et al.~2002), this gives limits on the magnetic field
1024: strength of $<1\times 10^7 {d \over 8 {\rm ~kpc}}$ Gauss (scenario
1025: $a1$), $<5 \times 10^8 {d \over 8 {\rm ~kpc}}$ Gauss ($a2$), $<1
1026: \times 10^8 {d \over 8 {\rm ~kpc}}$ Gauss ($b1$), and $7 \times 10^8
1027: {d \over 8 {\rm ~kpc}}$ Gauss ($b2$). When assuming a power-law
1028: spectral model for the quiescent spectrum of XTE J1751--305, the upper
1029: limits on the fluxes are significantly smaller, with a maximum
1030: luminosity limit of $3\times 10^{31} {d \over 8 {\rm ~kpc}}$ (with $d$
1031: in kpc), which would make the magnetic field limits a factor of $>$2
1032: less (the maximum upper limit would be again for scenario $b2$ and is
1033: $<3 \times 10^8 {d \over 8 {\rm ~kpc}}$ Gauss). Therefore, from these
1034: four scenarios the magnetic field strength of the neutron star in XTE
1035: J1751--305 can be constrained to be $< 3- 7 \times 10^8 {d \over {\rm
1036: 8~kpc}}$ Gauss (depending on assumed spectral model for the quiescent
1037: spectrum).
1038:
1039:
1040:
1041: Although this method still has significant uncertainties (i.e., due to
1042: the lack of understanding of the structure of the accretion flow
1043: geometry in quiescent X-ray transients; see Burderi et al.~2002 and Di
1044: Salvo \& Burderi 2003 for a more detailed discussion), our limits on
1045: the magnetic field of the neutron stars in XTE J0929--314 and XTE
1046: J1751--305 are consistent with the expectation that these neutron
1047: stars have a low but non-negligible (since they exhibit pulsations
1048: during accretion outbursts) magnetic field strength.
1049:
1050:
1051:
1052: \acknowledgements
1053:
1054: We thank Peter Jonker for allowing us to use his $K$-band image of the
1055: region around XTE J1751--305 and for comments on an early version of
1056: this paper. We thank Jae Sub Hong for useful discussion about the
1057: quantile analysis method.
1058:
1059: \begin{references}
1060:
1061: \reference{}Arnaud, K. 1996, in G. Jacoby \& J. Barnes (eds.), {\it
1062: Astronomical Data Analysis Software and Systems V.}, Vol. 101, p. 17,
1063: ASP Conf. Series.
1064:
1065: \reference{}
1066: Asai, K., Dotani, T., Hoshi, R., Tanaka, Y., Robinson, C. R., \&
1067: Terada, K. 1998, \pasj, 50, 611
1068:
1069: \reference{}Brown, E. F., Bildsten, L., \& Rutledge, R. E. 1998, \apj,
1070: 504, L95
1071:
1072: \reference{}
1073: Burderi, L. et al.~2002, \apj, 574, 930
1074:
1075: \reference{}
1076: Cash, W. 1979, \apj, 228, 939
1077:
1078:
1079: \reference{}
1080: Campana, S. \& Stella, L. 2000, \apj, 541, 849
1081:
1082: \reference{}Campana, S., Colpi, M., Mereghetti, S., Stella, L.,
1083: Tavani, M. 1998, \aapr, 8, 279
1084:
1085:
1086: \reference{}
1087: Campana, S., et al. 2002, \apj, 575, L15
1088:
1089: \reference{}
1090: Chen, W., Shrader, C. R., \& Livio, M. 1997, \apj, 491, 312
1091:
1092:
1093: \reference{}
1094: Di Salvo, T. \& Burderi, L. 2003, \aap, 397, 723
1095:
1096: \reference{}
1097: Dotani, T., Asai, K., \& Wijnands, R. 2000, \apj, 543, L145
1098:
1099: \reference{}
1100: Galloway, D. K., Chakrabarty, D., Morgan, E. H., \& Remillard,
1101: R. A. 2002, \apj, 576, L137
1102:
1103: \reference{}
1104: Gehrels, N. 1986, \apj, 303, 336
1105:
1106: \reference{}
1107: Greenhill, J. G., Giles, A. B., \& Hill, K. M. 2002, \iaucirc, 7889
1108:
1109: \reference{}
1110: Haensel, P. \& Zdunik J. L. 1990, \aap, 227, 431
1111:
1112: \reference{}Hong, J., Schlegel, E. M., \& Grindlay, J. E. 2004, ApJ,
1113: in press (astro-ph/040646)
1114:
1115: \reference{}Jonker, P. G., et al.~2003, \mnras, 344, 201
1116:
1117: \reference{}Jonker, P. G., Wijnands, R., \& van der Klis,
1118: M. 2004a, \mnras, 349, 94
1119:
1120: \reference{}
1121: Jonker, P. G., Galloway, D. K., McClintock, J. E., Buxton, M., Garcia,
1122: M., \& Murray, S. 2004b, \mnras, in press (astro-ph/0406208)
1123:
1124:
1125: \reference{}
1126: Juett, A. M., Galloway, D. K., \& Chakrabarty, D. 2003, \apj, 587, 754
1127:
1128: \reference{}
1129: Lasota, J.-P. 2001, NewA Rev., 45, 449
1130:
1131: \reference{}
1132: Maritz, J. S. \& Jarrett, R. G. 1978, Journal of American Statistical
1133: Association, 73, 194
1134:
1135: \reference{}
1136: Markwardt, C.B \& Swank, J. H. 2003, \iaucirc, 8144
1137:
1138: \reference{}
1139: Markwardt, C. B., Swank, J. H., Strohmayer, T. E., in 't Zand,
1140: J. J. M., \& Marshall, F. E. 2002, \apj, 575, L21
1141:
1142: \reference{}
1143: Markwardt, C. B., Smith, E., \& Swank, J. H. 2003, \iaucirc 8080
1144:
1145:
1146: \reference{}
1147: Miller, J. M., et al.~2003, \apj, 583, L99
1148:
1149:
1150: \reference{}
1151: Remillard, R. A. 2002, \iaucirc, 7888
1152:
1153: \reference{}
1154: Remillard, R. A., Swank, J., \& Strohmayer, T. 2002, \iaucirc, 7893
1155:
1156:
1157: \reference{}
1158: Rutledge, R. E., Bildsten, L., Brown, E. F., Pavlov, G. G., \& Zavlin,
1159: V. E., 2001, \apj, 551, 921
1160:
1161: \reference{}Rutledge, R. E., Bildsten, L., Brown, E. F., Pavlov,
1162: G. G., Zavlin, V. E., Ushomirsky, G., 2002, \apj, 580, 413
1163:
1164: \reference{}
1165: Rupen, M. P., Dhawan, V., \& Mioduszewski, A. J. 2002, \iaucirc, 7893
1166:
1167: \reference{}
1168: Smith, D. A., Morgan, E. H., \& Bradt, H. 1997, \apj, 479, L137
1169:
1170: \reference{}
1171: Stella, L., Campana, S., Colpi, M., Mereghetti, S., \& Tavani,
1172: M. 1994, \apj, 423, L47
1173:
1174: \reference{}
1175: Stella, L., Campana, S., Mereghetti, S., Ricci, D., \& Israel,
1176: G. L. 2000, \apj, 537, L115
1177:
1178: \reference{}
1179: Strohmayer, T. \& Bildsten, L. 2004, To appear in 'Compact Stellar
1180: X-ray Sources', eds. W. H. G. Lewin \& M. van der Klis, Cambridge
1181: University Press (astro-ph/0301544)
1182:
1183: \reference{}
1184: Tomsick, J. A., Gelino, D. M., Halpern, J. P., \& Kaaret, P. 2004,
1185: \apj, 610, 933
1186:
1187: \reference{}
1188: Wijnands, R. \& van der Klis, M. 1998, \nat, 394, 344
1189:
1190: \reference{}Wijnands, R., Miller, J. M., Markwardt, C., Lewin,
1191: W. H. G., van der Klis, M. 2001, \apj, 560, L159
1192:
1193: \reference{}
1194: Wijnands, R. et al. 2004, \apj, submitted (astro-ph/0310144)
1195:
1196: \reference{}
1197: Zhang, W., Jahoda, K., Kelley, R. L., Strohmayer, T. E., Swank, J. H.,
1198: \& Zhang, S. N. 1998, \apj, 495, L9
1199:
1200: \reference{}Zavlin, V. E., Pavlov, G. G., \& Shibanov, Yu. A., 1996,
1201: \aap, 315, 141
1202:
1203: \end{references}
1204:
1205:
1206:
1207: \clearpage
1208: \begin{figure}
1209: \begin{center}
1210: \begin{tabular}{c}
1211: \psfig{figure=f1.ps,width=16cm}
1212: \end{tabular}
1213: \figcaption{
1214: The {\it Chandra} image of the field around XTE J0929--314 (left) and
1215: a close-up of XTE J0929--314 (right) with the radio error circle for
1216: the source obtained during outburst (Rupen et al.~2002). In both
1217: figures, north is up and east is to the left. In the left panel the
1218: arrow indicates which source is XTE J0929--314. The source located to
1219: the north-west of XTE J0929--314 is CXOU J092919.1--312244 and the
1220: source to the south-east (in the bottom left corner) is CXOU
1221: J092924.4--312420.
1222: \label{fig:image} }
1223: \end{center}
1224: \end{figure}
1225:
1226: \clearpage
1227: \begin{figure}
1228: \begin{center}
1229: \begin{tabular}{c}
1230: \psfig{figure=f2.ps,width=16cm,angle=-90}
1231: \end{tabular}
1232: \figcaption{
1233: The energy of the detected photons versus the time of detection (since
1234: the start of the observation).
1235: \label{fig:energy-curve} }
1236: \end{center}
1237: \end{figure}
1238:
1239:
1240:
1241:
1242: \clearpage
1243: \begin{figure}
1244: \begin{center}
1245: \begin{tabular}{c}
1246: \psfig{figure=f3.ps,width=14cm,angle=-90}
1247: \end{tabular}
1248: \figcaption{
1249: The time-averaged spectrum of XTE J0929--314. In the top panel, we
1250: show the best power-law fit to the data (solid line) and the best NSA
1251: fit through the data with a fixed normalization (to a distance of 10
1252: kpc; dashed line). In the bottom panel, we again show the best
1253: power-law fit (solid line) but now the best NSA fit with the
1254: normalization left free (dashed line). For display purposes, the data
1255: were rebinned so that each bin has 3 counts but the spectra were
1256: fitted without any rebinning.
1257: \label{fig:spectra} }
1258: \end{center}
1259: \end{figure}
1260:
1261:
1262: \clearpage
1263: \begin{figure}
1264: \begin{center}
1265: \begin{tabular}{c}
1266: \psfig{figure=f4.ps,width=14cm}
1267: \end{tabular}
1268: \figcaption{
1269: The hardness ratio ({\it top}) and the $\log{}_{10} {Q_{50} \over 1 -
1270: Q_{50}}$ ({\it bottom}) versus the time during the observation. The
1271: hardness ratio is defined as the logarithm of the ratio between the
1272: number of photons detected with energies $>$1 keV and the number
1273: detected with energies $<$1 keV, and $Q_{50}$ is the median quantile
1274: as defined by Hong et al.~(2004).
1275: \label{fig:quantile} }
1276: \end{center}
1277: \end{figure}
1278:
1279:
1280: \clearpage
1281: \begin{figure}
1282: \begin{center}
1283: \begin{tabular}{c}
1284: \psfig{figure=f5.ps,width=14cm,angle=-90}
1285: \end{tabular}
1286: \figcaption{
1287: A comparison of the photon index of the first part of the data (0--10
1288: ksec; $\Gamma_1$) with that of the second part of the data (the
1289: remaining $\sim$14 ksec of the observation; $\Gamma_2$). The contours
1290: are the 68\%, 90\% and 99\% confidence levels. The cross marks the
1291: position of the best-fit values of the indices. The dashed line
1292: indicates where both indices are equal.\label{fig:contour} }
1293: \end{center}
1294: \end{figure}
1295:
1296:
1297:
1298: \clearpage
1299: \begin{figure}
1300: \begin{center}
1301: \begin{tabular}{c}
1302: \psfig{figure=f6.ps,width=14cm}
1303: \end{tabular}
1304: \figcaption{
1305: The 0.5--7 keV image of the region around XTE J1751--305. East is to
1306: the left and north is up. Several point sources are detected close to
1307: the position of XTE J1751--305 but none of these sources is consistent
1308: with the position of XTE J1751--305 (as indicated by the circle which
1309: has a radius of 1$''$, which means that 90\% of the energy of the
1310: photons is encircled in this region).
1311: \label{fig:image_1751} }
1312: \end{center}
1313: \end{figure}
1314:
1315:
1316: \clearpage
1317:
1318: \begin{deluxetable}{lccccc}
1319: \tablecolumns{6}
1320: \tablewidth{0pt}
1321: \tablecaption{Spectral fits of XTE J0929--314 \label{tab:spectral_fits}}
1322: \tablehead{
1323: Model$^a$ & $N_{\rm H}$ & $kT_\infty$ & $\Gamma$ & $F^b$ & Fit quality$^c$ \\
1324: &($10^{22}$ cm$^{-2}$) & (keV) & & ($10^{-15}$ ergs s$^{-1}$ cm$^{-2}$) &
1325: }
1326: \startdata
1327: NSA ($d$ = 5 kpc) & $0.3^{+0.2}_{-0.1}$ & 0.045$\pm0.005$ & --- & 9$^{+7}_{-4}$ & 1.00 \\
1328: NSA ($d$ = 10 kpc) & $0.2\pm0.2$ & 0.058$^{+0.008}_{-0.009}$ & --- & 9$^{+9}_{-5}$ & 1.00 \\
1329: NSA ($d$ = 15 kpc) & $0.2^{+0.2}_{-0.1}$ & 0.067$\pm0.08$ & --- & 9$^{+7}_{-4}$ & 1.00 \\
1330: Power-law & $<$0.06 & --- & $1.8^{+0.6}_{-0.5}$ & 6$^{+4}_{-2}$ & 0.68 \\
1331: \enddata
1332:
1333: \tablenotetext{\,}{$^a$ For the neutron-star hydrogen atmosphere (NSA) model we used that of
1334: Zavlin et al.~(1996) with a mass of 1.4 $M_\odot$, a radius of 10 km,
1335: and a zero magnetic field strength for the neutron star in XTE
1336: J0929--314. The errors on the fit parameters are for 90\% confidence
1337: levels.}
1338:
1339: \tablenotetext{\,}{$^b$ The 0.5--10 keV unabsorbed flux.}
1340:
1341: \tablenotetext{\,}{$^c$ The fit quality represents the fraction of 10,000 simulated spectra
1342: which have Cash-statistics better than that obtained for the actual
1343: data.}
1344:
1345: \end{deluxetable}
1346:
1347: \clearpage
1348:
1349: \begin{deluxetable}{lccccc}
1350: \tablecolumns{6}
1351: \tablewidth{0pt}
1352: \tablecaption{Limits on any NSA contribution in XTE J0929--314$^a$ \label{tab:nsa_limits}}
1353: \tablehead{
1354: Distance $d$ & $N_{\rm H}$ & $\Gamma$ & $kT_\infty$ & Bolometric luminosity$^b$ & Fraction in the \\
1355: (kpc) & ($10^{22}$ cm$^{-2}$) & & (keV) & ($10^{32}$ ergs s$^{-1}$) & power-law component$^c$ }
1356: \startdata
1357: 5 & $<$0.14 & 1.2$^{+0.9}_{-1.1}$ & $<$0.04 & $<$0.4 & $>$0.77 \\
1358: 10 & $<$0.10 & 1.1$^{+1.0}_{-1.2}$ & $<$0.05 & $<$1.2 & $>$0.70 \\
1359: 15 & $<$0.08 & 1.1$^{+1.0}_{-1.3}$ & $<$0.05 & $<$1.7 & $>$0.78 \\
1360: \enddata
1361:
1362: \tablenotetext{\,}{$^a$
1363: Results of spectral fits assuming a NSA plus a power-law model. For
1364: the NSA model we used that of Zavlin et al.~(1996) with a mass of 1.4
1365: $M_\odot$, a radius of 10 km, and a zero magnetic field strength for
1366: the neutron star in XTE J0929--314. The upper limits are for 90\%
1367: confidence levels. }
1368:
1369: \tablenotetext{\,}{$^b$ Bolometric luminosity of the neutron-star
1370: atmosphere component as seen by an observer at infinity}
1371:
1372: \tablenotetext{\,}{$^c$ The minimum fraction of the 0.5--10 keV flux in the power-law
1373: component}
1374:
1375: \end{deluxetable}
1376:
1377:
1378: \clearpage
1379:
1380: \begin{deluxetable}{lcccc}
1381: \tablecolumns{5}
1382: \tablewidth{0pt}
1383: \tablecaption{Possible spectral variability in XTE J0929--314 \label{tab:variability}}
1384: \tablehead{
1385: Time selection & $N_{\rm H}$ & $\Gamma$ & Flux$^a$ & Fit quality \\
1386: (ksec since start) & ($10^{22}$ cm$^{-2}$) & & ($10^{-15}$ ergs s$^{-1}$ cm$^{-2}$) & }
1387: \startdata
1388: \multicolumn{5}{c}{Separate fits}\\
1389: \hline
1390: 0 -- 10 & $<$0.2 & $1.3^{+0.9}_{-0.7}$ & 14$^{+12}_{-4}$ & 0.70 \\
1391: 10 -- end & $<$0.1 & $2.6^{+1.2}_{-0.9}$ & 3$^{+2}_{-1}$ & 0.62 \\
1392: \hline
1393: \multicolumn{5}{c}{Simultaneous fits with tied $N_{\rm H}$}\\
1394: \hline
1395: 0 -- 10 & $<$0.07 & $1.3\pm0.7$ & 14$^{+12}_{-6}$ & 0.66 \\
1396: 10 -- end & `` & $2.5^{+1.0}_{-0.9}$ & 3$^{+2}_{-1}$ & `` \\
1397: \enddata
1398:
1399: \tablenotetext{\,}{$^a$ The 0.5--10 keV unabsorbed flux}
1400:
1401:
1402: \end{deluxetable}
1403:
1404:
1405:
1406: \end{document}
1407:
1408: