1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{psfig}
3: %\usepackage{emulateapj5}
4:
5:
6:
7: \newcommand\mdot {\hbox {${\dot M}$}}
8: \newcommand\sz {$S_{\rm z}$}
9: \newcommand\mzon {M$_{\odot}$}
10: \newcommand\pp {$\pm$}
11: \newcommand\pers {s$^{-1}$}
12: \newcommand\micros {$\mu$s}
13:
14: \def\degr{\hbox{$^\circ$}}
15: \newcommand\Lunit {erg s$^{-1}$}
16: \newcommand\funit {erg cm$^{-2}$ s$^{-1}$}
17: \newcommand\nh {$N_{\rm H}$}
18:
19: \slugcomment{Accepted for publication in ApJ Letters: March 19, 2004}
20:
21: \begin{document}
22:
23: \title{Monitoring {\itshape Chandra} observations of the
24: quasi-persistent neutron-star X-ray transient MXB 1659--29 in
25: quiescence: the cooling curve of the heated neutron-star crust}
26:
27: \author{Rudy Wijnands\altaffilmark{1,2}, Jeroen Homan\altaffilmark{3},
28: Jon M. Miller\altaffilmark{4,5}, Walter H. G. Lewin\altaffilmark{3}}
29:
30: \altaffiltext{1}{School of Physics and Astronomy,
31: University of St Andrews, North Haugh, St Andrews, Fife KY16 9SS,
32: Scotland, UK; radw@st-andrews.ac.uk}
33:
34: \altaffiltext{2}{Present address: Astronomical Institute ``Anton Pannekoek'',
35: University of Amsterdam, Kruislaan 403, 1098 SJ, Amsterdam, the
36: Netherlands; rudy@science.uva.nl}
37:
38: \altaffiltext{3}{Center for Space Research, Massachusetts Institute of
39: Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA;
40: jeroen@space.mit.edu, lewin@space.mit.edu}
41:
42: \altaffiltext{4}{Harvard-Smithsonian Center for Astrophysics,
43: 60 Garden Street, Cambridge, MA 02139, USA; jmmiller@head.cfa.harvard.edu}
44:
45: \altaffiltext{5}{NSF Astronomy \& Astrophysics Fellow}
46:
47:
48:
49: \begin{abstract}
50:
51: We have observed the quasi-persistent neutron-star X-ray transient and
52: eclipsing binary MXB 1659--29 in quiescence on three occasions with
53: {\it Chandra}. The purpose of our observations was to monitor the
54: quiescent behavior of the source after its last prolonged ($\sim$2.5
55: years) outburst which ended in September 2001. The X-ray spectra of
56: the source are consistent with thermal radiation from the neutron-star
57: surface. We found that the bolometric flux of the source decreased by
58: a factor of 7--9 over the time-span of 1.5 years between our first and
59: last {\it Chandra} observations. The effective temperature also
60: decreased, by a factor of 1.6--1.7. The decrease in time of the
61: bolometric flux and effective temperature can be described using
62: exponential decay functions, with $e$-folding times of $\sim$0.7 and
63: $\sim$3 years, respectively. Our results are consistent with the
64: hypothesis that we observed a cooling neutron-star crust which was
65: heated considerably during the prolonged accretion event and which is
66: still out of thermal equilibrium with the neutron-star core. We could
67: only determine upper-limits for any luminosity contribution due to the
68: thermal state of the neutron-star core. The rapid cooling of the
69: neutron-star crust implies that it has a large thermal
70: conductivity. Our results also suggest that enhanced cooling processes
71: are present in the neutron-star core.
72:
73: \end{abstract}
74:
75:
76: \keywords{accretion, accretion disks --- stars: neutron stars: individual (MXB
77: 1659--29)--- X-rays: stars}
78:
79: \section{Introduction}
80:
81:
82: Neutron stars in low-mass X-ray binaries accrete matter from solar
83: mass companions. Among those systems, the sub-group of neutron-star
84: transients spend most of their time in quiescence during which hardly
85: any or no accretion occurs. However, these transients sporadically
86: become very bright ($>$$10^{36-38}$~\Lunit) owing to a huge increase
87: in the accretion rate onto their neutron stars. During those
88: outbursts, these sources can be readily studied with the available
89: X-ray instruments, but obtaining high quality quiescent data remains a
90: challenge. In spite of this, several systems have now been studied in
91: detail: they typically exhibit 0.5--10 keV luminosities of
92: $10^{32-33}$~\Lunit~and their spectra are usually dominated by a soft
93: component which can be described by a thermal model. This emission is
94: thought to be due to the cooling of the neutron star which has been
95: heated during the outbursts (Brown, Bildsten, \& Rutledge 1998;
96: Campana et al.~1998a).
97:
98: Most neutron-star transients are active for only weeks to months, but
99: several systems have remained active for years and even decades (the
100: 'quasi-persistent' neutron-star transients; Wijnands et
101: al. 2003). Wijnands et al.~(2001) realized that those systems are
102: excellent targets to study the effects of accretion on the behavior of
103: neutron stars by observing them in quiescence. The accreting material
104: is expected to have a larger effect on the neutron stars in such
105: systems than on the neutron stars in short-duration transients
106: (Wijnands et al.~2001; Rutledge et al.~2002). In the latter systems,
107: the crust is only marginally heated during the outbursts and will
108: quickly return to thermal equilibrium with the core after the end of
109: the outbursts. In the quasi-persistent transients, however, the crust
110: is heated to high temperatures and becomes significantly out of
111: thermal equilibrium with the core (Rutledge et al.~2002). After the
112: end of the prolonged outbursts, it will cool until it returns to
113: equilibrium with the core. The exact cooling time depends on the
114: thermal conductivity of the crust, the core cooling processes, and the
115: accretion history of the source.
116:
117: KS 1731--260 was the first quasi-persistent transient to be studied in
118: detail in quiescence. It was observed using {\it Chandra} shortly
119: after the end of its $\sim$12.5 year outburst (Wijnands et al.~2001)
120: and it was found to have a luminosity of $\sim10^{33}$~\Lunit~(for a
121: distance $d=7$ kpc; 0.5--10 keV). Half a year later it was observed
122: with {\it XMM-Newton} and it was found that its luminosity had
123: decreased by a factor of 2--3 (Wijnands et al.~2002b). Using the
124: cooling curves calculated by Rutledge et al.~(2002), this drop in
125: brightness can be explained if the neutron star has a large crustal
126: conductivity and enhanced core cooling processes. In September 2001, a
127: second quasi-persistent neutron-star transient (MXB 1659--29) turned
128: off after having accreted for $\sim$2.5 years. Wijnands et al.~(2003)
129: obtained a {\it Chandra} observation of this source within a month
130: after the end of its outburst and detected it at a luminosity of
131: $\sim3-4 \times 10^{33}$~\Lunit~(0.5--10 keV; $d=10$ kpc). Several
132: years before this outburst, the source was observed with {\it ROSAT},
133: but could not be detected (Verbunt 2001). The flux upper limit was
134: $\sim$10 times lower than the {\it Chandra} flux (Oosterbroek et
135: al.~2001; Wijnands 2002). Wijnands et al.~(2003) concluded that during
136: the {\it Chandra} observation the observed radiation was due to a hot
137: crust and not associated with the core.
138:
139: \section{Observations, analysis, and results}
140:
141: {\it Chandra} observed MXB 1659--29 twice for $\sim$27 ksec: on
142: October 15, 2002 (the 2002 observation), and on May 9, 2003 (the 2003
143: observation). We also used the $\sim$19 ksec observation performed on
144: October 15--16, 2001 (the 2001 observation; Wijnands et
145: al.~2003). During all observations the ACIS-S3 chip was used. The data
146: were reduced and analyzed using CIAO 3.0. To make use of the latest
147: calibration products, we reprocessed the 2001 observation. A minor
148: background flare occurred during the 2003 observation (factor of
149: $\sim$2; lasting $\sim$2 ksec). Its effect on the quality of the
150: source data was negligible and we did not to remove this flare from
151: the data. No flares occurred during the other observations.
152:
153: For each observation, we extracted the number of source photons, the
154: light curve, and the spectrum, using a circle with a radius of 3$''$
155: as source extraction region and an annulus with an inner radius of
156: $7''$ and an outer radius of 22$''$ as background region. We detected
157: 948\pp31, 263\pp16, and 107\pp10 counts (0.3--7 keV; background
158: corrected) for the 2001, 2002, and 2003 observations, respectively,
159: resulting in corresponding count rates of 0.050\pp0.002,
160: 0.0097\pp0.0006, and 0.0039\pp0.0004 counts~s$^{-1}$. Wijnands et
161: al. (2003) observed an eclipse and dipping behavior during the 2001
162: observation (similar to the outburst behavior of the source; Lewin
163: 1979; Cominsky et al. 1983; Cominsky \& Wood 1984, 1989). To search
164: for eclipsing behavior during the 2002 and 2003 observations, we
165: determined the orbital phase range covered by those observations using
166: the time of the eclipse in the 2001 observation as the reference
167: time. Given the orbital phase range traced during each observation, we
168: expect to see a single eclipse per observation and, as anticipated, we
169: did not detect any photons during the expected eclipse intervals.
170: However, we also found that no photons were detected during several
171: time intervals (of equal duration as the lengths of the eclipses) at
172: different phases of the orbital period. Therefore, without prior
173: knowledge of the eclipsing nature of MXB 1659--29, we could not have
174: concluded that we saw eclipses during the 2002 and 2003
175: observations. Owing to the limited statistics of the 2002 and 2003
176: observations, no conclusions can be drawn about possible dipping
177: behavior during these observations.
178:
179: When extracting the spectra, we used all data, including those taken
180: during the intervals of eclipses and possible dipping behavior. The
181: eclipses could not be removed from the data before extracting the
182: spectra because the uncertainties in the ephemeris presented by
183: Oosterbroek et al. (2001) are sufficiently large so that the exact
184: start and end times of the expected eclipses could not be
185: determined. Instead we decreased the exposure time in the resulting
186: spectral files by 900 s since the eclipse duration during outburst was
187: found to be $\sim$900 s (Wachter et al. 2000) and Wijnands et
188: al. (2003) reported an eclipse duration of 842\pp90 seconds for the
189: 2001 observation. Small differences in the eclipse duration might be
190: present between the observations but the expected effects on the
191: resulting fluxes will be marginal. We also did not remove the data
192: obtained during the dipping interval observed in the 2001
193: observation. Such dipping intervals are likely present during the
194: other two observations but they cannot be identified in the light
195: curves due to limited statistics. For those two observations all data
196: had to be used and to obtain a homogeneous data selection across
197: observations, we included the dipping interval observed during the
198: 2001 observation. Wijnands et al. (2003) found evidence that this
199: dipping behavior is likely due to a change in internal absorption in
200: the system and not due to actual changes in the neutron-star
201: properties. Therefore, the inclusion of the (possible) dipping
202: intervals will likely result in a somewhat higher column density
203: ($N_{\rm H}$) in the spectral fits than the true interstellar $N_{\rm
204: H}$ toward the source, but should not significantly impact other
205: source properties.
206:
207: We grouped the spectra in bins of 15 counts to validate the use of the
208: $\chi^2$ fitting method and simultaneously fitted the three spectra
209: using Xspec (Arnaud 1996). A variety of one-component
210: models\footnote{E.g., a power-law model could fit the spectra but with
211: an index of 4.7--5.8 suggesting soft thermal spectra. We also fitted a
212: NSA plus power-law model to determine the upper-limits on the
213: contribution of such a power-law tail to the 0.5--10 keV flux. Those
214: limits are $<$20\%--25\%, $<$35\%--45\%, and $<$50\%--100\%, for the
215: 2001, 2002, and 2003 observation, respectively. The range of upper
216: limits is due to the range assumed in photon indices (between 1 and
217: 2).} could fit the individual spectra satisfactorily, but since we
218: expect that the X-rays from MXB 1659--29 are due to the cooling of the
219: neutron-star surface, for this paper we only fit the data using a
220: neutron-star hydrogen atmosphere model (NSA; for weakly magnetized
221: neutron stars; Zavlin et al. 1996). In such models the normalization
222: is given by $1/d^2$, with $d$ in pc. The distance should be constant
223: between observations and therefore we left the normalization tied
224: among the different spectra (when leaving the normalizations free
225: between observations, we find that they are consistent with each
226: other). We expect the $N_{\rm H}$ toward the source to be very similar
227: between observations (only minor variations are expected due to
228: variable internal absorption) and this parameter was also tied. We
229: assume a 'canonical' neutron star with a radius of 10 km and a mass of
230: 1.4 \mzon.
231:
232: From the fits, we found that the normalization was
233: $1.4^{+2.2}_{-0.8}\times 10^{-8}$ which yields a source distance of
234: 5--13 kpc. This is consistent with the distance range given in the
235: literature (10--13 kpc; Oosterbroek et al. 2001; Muno et
236: al. 2001). However, we found that the errors on the fit parameters
237: were dominated by the large uncertainties in the normalization and did
238: not allow us to realize the full potential of the data. If the source
239: distance were established through an independent method, we could fix
240: the normalization in the NSA models, resulting in considerably smaller
241: errors on the remaining fit parameters. Therefore, instead of leaving
242: the normalization as a free parameter, we fixed it so that it
243: corresponded to a distance of 5, 10, and 13 kpc, covering the full
244: range of allowed distances obtained when the normalization was a free
245: parameter. To estimate the bolometric fluxes ($F_{\rm bol}$) we
246: extrapolated the model to the energy range 0.01--100 keV which gives
247: approximate bolometric fluxes\footnote{We verified that the 0.01--100
248: keV fluxes approximate $F_{\rm bol}$ by calculating the bolometric
249: luminosity $L_{\rm bol} = 4\pi\sigma R^{2}_{\infty} T^{\infty 4}_{\rm
250: eff}$, with $\sigma$ Stefan-Boltzmann constant, $T^{\infty}_{\rm eff}$
251: the effective temperature (at infinity), and $R_{\infty}$ the
252: neutron-star radius (at infinity). The 0.01--100 keV fluxes were
253: indeed consistent with the calculated $F_{\rm bol}$. We use the
254: measured fluxes because their errors takes into account the
255: uncertainties in $N_{\rm H}$ and the $T^{\infty}_{\rm eff}$ obtained
256: for all observations. The $L_{\rm bol}$ errors are only calculated
257: using the $T^{\infty}_{\rm eff}$ errors during one specific
258: observation. \label{footnote}}. To calculate the flux errors, we
259: fixed each free fit parameter (only one at a time) either to its
260: minimum or maximum allowed value. After that we refitted the data and
261: recalculated the fluxes. This process was repeated for each free
262: parameter and the final flux range determined the flux errors. The
263: fit parameters obtained are listed in Table~\ref{tab:spectra}.
264:
265: This table shows that $T^{\infty}_{\rm eff}$ and $F_{\rm bol}$
266: decreased in time (Fig.~\ref{fig:decay}). We fitted the
267: $T^{\infty}_{\rm eff}$ and $F_{\rm bol}$ curves with an exponential
268: decay function $y(t) = c_0 e^{- {t - t_0\over
269: \tau}}$, with $c_0$ a normalization constant, $t_0$ the start time,
270: and $\tau$ the $e$-folding time. We found that the other fit
271: parameters were not very sensitive to the value of $t_0$, but when
272: $t_0$ was left free it had adverse effects on the errors on those
273: parameters. Therefore, we fixed $t_0$ to MJD 52159.5 which
274: corresponds to midday September 7, 2001 (the last day MXB 1659--29 was
275: found to be active; Wijnands et al. 2002a) and which can be regarded
276: as an approximation of the time when $T^{\infty}_{\rm eff}$ and
277: $F_{\rm bol}$ began to decrease. The assumed exponential functions
278: could adequately describe the decrease in $T^{\infty}_{\rm eff}$ and
279: $F_{\rm bol}$ (Fig.~\ref{fig:decay}; alternative functions did not
280: provide adequate fits). We found that $\tau$ and $c_0$ for the
281: $F_{\rm bol}$ curve were 289\pp37, 262\pp33, and 254\pp29 days, and
282: 70\pp9, 48\pp6, and 43\pp6 $\times 10^{-14}$ \funit, when assuming a
283: distance of 5, 10, or 13 kpc, respectively, in the spectral fits. The
284: corresponding $\tau$ and $c_0$ for the $T^{\infty}_{\rm eff}$ curve
285: were 1153\pp160, 1060\pp126, and 1055\pp112 days, and 0.099\pp0.004,
286: 0.126\pp0.004, and 0.139\pp0.004 keV. We saw no evidence that the
287: curves approached a rock-bottom value: we found an upper limit on such
288: a value of 3.5--7.5 $\times 10^{-14}$ \funit~for the $F_{\rm bol}$
289: curve (resulting in bolometric luminosity limits of 2.2--7.0
290: $\times 10^{32}$ \Lunit), and 0.06--0.07 keV for the $T^{\infty}_{\rm
291: eff}$ curve.
292:
293: \section{Discussion}
294:
295: We have presented monitoring {\it Chandra} observations of MXB
296: 1659--29 in quiescence. The first observation was taken only a month
297: after the end of its last outburst which lasted 2.5 years; the second
298: and third observations were taken $\sim$1 and $\sim$1.5 years after
299: this initial one. Because it is expected that the emission should be
300: dominated by thermal emission from the hot neutron-star crust (see
301: Wijnands et al. 2003), we fitted the data with a NSA model for weakly
302: ($B<10^{8-9}$ G) magnetized neutron stars. We found that $F_{\rm bol}$
303: decreased by a factor of $\sim$8 in $\sim$1.5 years and the rate of
304: decrease followed an exponential decay function. Furthermore, $T_{\rm
305: eff}^{\infty}$ also decreased and the rate of decrease again followed
306: an exponential decay function. We found that the $e$-folding time of
307: the $T_{\rm eff}^{\infty}$ curve was consistent with four times that
308: of the $F_{\rm bol}$ curve, as expected if the emission is caused by a
309: cooling black body for which the bolometric luminosity is given by
310: $L_{\rm bol}=4\pi\sigma R^{2}_{\infty} T^{\infty 4}_{\rm eff}$ (see
311: footnote~\ref{footnote}): if $T^{\infty}_{\rm eff}$ decays
312: exponentially, $L_{\rm bol}$ (and thus $F_{\rm bol}$) will also decay
313: exponentially but with an $e$-folding time four times smaller than
314: that of $T^{\infty}_{\rm eff}$, exactly what we observe.
315:
316: Our results support the suggestion that the crust was heated to high
317: temperatures during the prolonged accretion event, which ended a month
318: before our first observation, and that it is now cooling until it
319: reaches thermal equilibrium with the core. Rutledge et al. (2002)
320: calculated cooling curves for the neutron star in KS 1731--260,
321: assuming different behaviors of the crustal micro-physics and the core
322: cooling processes. Those curves can be used as a starting point to
323: investigate how our results of MXB 1659--29 could be explained. Of
324: those curves, only the one which assumes a large crustal conductivity
325: and the presence of enhanced core cooling processes exhibits a large
326: luminosity decrease in the first two years after the end of the last
327: outburst, suggesting that the neutron star in MXB 1659--29 has similar
328: properties. This conclusion was already tentatively reached by
329: Wijnands et al. (2003) based on a comparison of the luminosity seen
330: during the October 2001 {\it Chandra} observation with the
331: significantly lower luminosity upper-limit found with {\it ROSAT}. But
332: detailed cooling curves for the neutron star in MXB 1659--29 need to
333: be calculated to fully explore (and exploit) the impact of our
334: observations on our understanding of the structure of neutron
335: stars. The cooling curves calculated by Rutledge et al. (2002) for KS
336: 1731--260 only give us a hint of the behavior of MXB 1659--29 because
337: they depend on the long-term ($>10^4$ years) accretion history of the
338: source. For KS 1731--260, this long-term accretion behavior was quite
339: unconstrained due to large uncertainties in the averaged duration of
340: the outbursts, the time-averaged accretion rate during the outbursts,
341: and the time the source spent in quiescence. However, the accretion
342: history of MXB 1659--29 over the last three decades is much better
343: constrained (Wijnands et al. 2003), which will help to reduce the
344: uncertainties in its long-term averaged accretion history allowing for
345: more detailed cooling curves to be calculated for MXB 1659--29. This
346: might help to constrain the physics of the crust better for MXB
347: 1659--29 than for KS 1731--260. The only significant uncertainty left
348: is that of the source distance; however, we found that this only
349: affects the exact values of the bolometric fluxes and the effective
350: temperatures, but not their rate of decay.
351:
352: Our 0.5--10 keV flux during the May 2003 {\it Chandra} observation is
353: still higher than the upper limit found with {\it ROSAT}, suggesting
354: that the crust will cool even further in quiescence and that we have
355: not yet reached thermal equilibrium between the crust and
356: core. Further monitoring observations are needed to follow the cooling
357: curve of the crust to determine the moment when the crust is thermally
358: relaxed again. When this occurs, no significant further decrease of
359: the quiescent luminosity is expected and from this bottom level the
360: state of the core can be inferred. As of yet, we have found no
361: evidence that the flux and temperature are reaching a leveling-off
362: value, associated with the temperature of the core, although the
363: limits we obtained are not very stringent.
364:
365: Jonker, Wijnands, \& van der Klis (2004) suggested that the difference
366: in luminosity of MXB 1659--29 between the {\it ROSAT} non-detection
367: and the 2001 {\it Chandra} observation might be due to differences in
368: residual accretion rate onto the surface. Residual accretion could
369: indeed produce soft spectra (e.g., Zampieri et al. 1995), but to
370: explain the exponential decay we observe for $F_{\rm bol}$ and $T_{\rm
371: eff}^{\infty}$, the residual accretion rate must also decrease
372: exponentially with a timescale of a year. Although this cannot be
373: completely ruled out, we believe this is unlikely since other
374: neutron-star transients have been observed to reach their quiescent
375: states on timescales of only tens to several tens of days at the end
376: of their outbursts (e.g., Campana et al. 1998b; Jonker et al. 2003)
377: and the variations in accretion rate tend to be more
378: stochastic. Moreover, if the neutron star has a significant magnetic
379: field strength, this might inhibit material from reaching the surface
380: when accreting at the inferred low rates.
381:
382:
383:
384: \begin{references}
385:
386: \reference{}Arnaud, K. 1996, in G. Jacoby \& J. Barnes (eds.), {\it
387: Astronomical Data Analysis Software and Systems V.}, Vol. 101, p. 17,
388: ASP Conf. Series.
389:
390: \reference{}Brown, E. F., Bildsten, L., \& Rutledge, R. E. 1998, \apj,
391: 504, L95
392:
393:
394: \reference{}Campana, S., Colpi, M., Mereghetti, S., Stella, L.,
395: Tavani, M. 1998a, \aapr, 8, 279
396:
397: \reference{}Campana, S., Stella, L., Mereghetti, S., Colpi, M., Tavani, M.,
398: Ricci, D., Dal Fiume, D., Belloni, T. 1998b, \apj, 499, L65
399:
400:
401: \reference{}Cominsky, L. R. \& Wood, K. S., 1984, \apj, 283, 765
402:
403: \reference{}Cominsky, L. R. \& Wood, K. S., 1989, \apj, 337, 485
404:
405:
406: \reference{}Cominsky, L. R., Ossmann, W., \& Lewin, W. H. G., 1983,
407: \apj, 270, 226
408:
409:
410: \reference{}G\"ansicke, B. T., Braje, T. M., \& Romani, R. W. 2002, \aap,
411: 386, 1001
412:
413: \reference{}Jonker, P. G., M\'endez, M., Nelemans, G., Wijnands, R.,
414: van der Klis, M., 2003, \mnras, 341, 823
415:
416: \reference{}Jonker, P. G., Wijnands, R., \& van der Klis, M.
417: 2004, \mnras, in press (astro-ph/0311560)
418:
419: \reference{}Lewin, W. H. G., 1979, Proc. IAU/COSPAR Symposium on
420: X-ray Astronomy, Innsbruck, Austria, May 1978, in X-ray Astronomy,
421: ed. W. A. Baity and L. E. Peterson (Oxford: Pergamon Press), Vol. 3.,
422: p. 133
423:
424:
425: \reference{}Muno, M. P., Chakrabarty, D., Galloway, D. K., Savov, P.,
426: 2001, \apj, 553, L157
427:
428:
429: \reference{}Oosterbroek, T., Parmar, A. N., Orlandini, M., Segreto,
430: A., Santangelo, A., Del Sordo, S. 2001, \aap, 375, 922
431:
432:
433: \reference{}Rutledge, R. E., Bildsten, L., Brown, E. F., Pavlov,
434: G. G., Zavlin, V. E., Ushomirsky, G., 2002, \apj, 580, 413
435:
436: \reference{}Verbunt, F., 2001, \aap, 368, 137
437:
438: \reference{}Wachter, S., Smale, A. P., \& Bailyn, C., 2000, \apj, 534,
439: 367
440:
441: \reference{}Wijnands, R. 2002 in "The High Energy Universe
442: at Sharp Focus: Chandra Science", proceedings of the 113th Meeting of
443: the Astronomical Society of the Pacific. 16-18 July 2001, St. Paul,
444: MN, ed. E. M. Schlegel \& S. D. Vrtilek, p.235
445:
446: \reference{}Wijnands, R., Miller, J. M., Markwardt, C., Lewin,
447: W. H. G., van der Klis, M. 2001, \apj, 560, L159
448:
449: \reference{}Wijnands, R., Muno, M. P., Miller., J. M., Franco, L. M.,
450: Strohmayer, T., Galloway, D., Chakrabarty, D. 2002a, \apj, 566, 1060
451:
452: \reference{}Wijnands, R., Guainazzi, M., van der Klis, M., M\'endez,
453: M. 2002b, \apj, 573, L45
454:
455: \reference{}Wijnands, R., Nowak, M., Miller, J. M., Homan, J., Wachter, S.,
456: Lewin, W. H. G., 2003, \apj, 594, 952
457:
458:
459: \reference{}Zampieri, L., Turolla, R., Zane, S., Treves, A.,
460: 1995, \apj, 439, 849
461:
462: \reference{}Zavlin, V. E., Pavlov, G. G., \& Shibanov, Yu. A., 1996,
463: \aap, 315, 141
464:
465: \end{references}
466:
467:
468:
469: \clearpage
470: \begin{figure}
471: \begin{center}
472: \begin{tabular}{c}
473: \psfig{figure=f1.eps,width=16cm,angle=-90}
474: \end{tabular}
475: \figcaption{
476: The {\it Chandra} X-ray spectra obtained during the quiescent state of
477: MXB 1659--29. The top spectrum was obtained on October 15--16, 2001,
478: the middle spectrum (indicated by the crosses) was obtained on October
479: 15, 2002, and the bottom spectrum (indicated by the open circles) was
480: obtained on May 9, 2003. The solid lines through the spectra indicate
481: the best fit neutron-star hydrogen atmosphere model (that of Zavlin et
482: al. 1996; for weakly magnetized neutron stars).
483: \label{fig:spectra} }
484: \end{center}
485: \end{figure}
486:
487: \clearpage
488: \begin{figure}
489: \begin{center}
490: \begin{tabular}{c}
491: \psfig{figure=f2.eps,width=10cm,angle=-90}
492: \end{tabular}
493: \figcaption{
494: The bolometric flux (top panel) and effective temperature (bottom
495: panel; for an observer at infinity) of the neutron-star crust as a
496: function of time (as obtained with the neutron-star hydrogen
497: atmosphere model for weakly magnetized neutron stars of Zavlin et
498: al. 1996). The solid curves are the best fit exponential function
499: through the data points. The bolometric fluxes are plotted on a
500: logarithmic scale, but for clarity, the effective temperatures are
501: plotted on a linear scale. The data used in this figure are those
502: obtained when assuming a distance of 10 kpc in the spectral fits; the
503: figures for a distance of 5 or 13 kpc are very similar and therefore
504: we omit them. Only the absolute values of the bolometric flux and
505: effective temperature are different but the overall decay trend is
506: nearly identical.
507: \label{fig:decay} }
508: \end{center}
509: \end{figure}
510:
511:
512: \clearpage
513:
514: \begin{deluxetable}{lccc}
515: \tablecolumns{4}
516: \tablewidth{0pt}
517: \tablecaption{Spectral results for MXB 1659--29\label{tab:spectra}}
518: \tablehead{
519: Parameter &\multicolumn{3}{c}{Distance assumed} \\
520: & 5 kpc & 10 kpc & 13 kpc }
521: \startdata
522: $N_{\rm H}$ ($10^{21}$ cm$^{-2}$) & 2.8\pp0.3 & 1.8\pp0.2 & 1.5\pp0.2 \\
523: \hline
524: \multicolumn{4}{l}{$kT_{\rm eff}^{\infty}$ (keV)} \\
525: ~~~2001 & 0.096\pp0.002 & 0.122\pp0.002 & 0.134\pp0.002 \\
526: ~~~2002 & 0.069\pp0.002 & 0.086\pp0.002 & 0.094$^{+0.002}_{-0.003}$ \\
527: ~~~2003 & 0.059\pp0.002 & 0.072\pp0.003 & 0.079\pp0.003 \\
528: \hline
529: \multicolumn{4}{l}{Flux ($10^{-14}$ \funit; 0.5--10 keV; unabsorbed)}\\
530: ~~~2001 & 41.8\pp3.2 & 31.3\pp2.3 & 28.4\pp2.1 \\
531: ~~~2002 & 9.1\pp1.0 & 6.4$^{+0.8}_{-0.6}$ & 5.7\pp0.6 \\
532: ~~~2003 & 4.0$^{+0.8}_{-0.2}$ & 2.8$^{+0.7}_{-0.5}$ & 2.5\pp0.4 \\
533: \hline
534: \multicolumn{4}{l}{Bolometric flux ($10^{-14}$ \funit; unabsorbed)}\\
535: ~~~2001 & 61.6\pp4.2 & 42.1\pp2.9 & 37.7\pp2.8 \\
536: ~~~2002 & 16.9\pp1.5 & 10.1\pp1.0 & 8.5\pp0.8 \\
537: ~~~2003 & 8.9$^{+1.4}_{-0.4}$ & 5.1\pp0.9 & 4.2\pp0.6 \\
538: \hline
539: $\chi^2$/d.o.f. & 59.2/65 & 56.1/65 & 58.6/65 \\
540: \enddata
541:
542:
543: \tablenotetext{\,}{Note: The error bars represent 90\% confidence
544: levels. We used a neutron-star mass of 1.4 \mzon~and radius of 10 km
545: and the neutron-star hydrogen atmosphere model for weakly magnetized
546: neutron stars of Zavlin et al. 1996. }
547:
548:
549: \end{deluxetable}
550:
551:
552:
553: \end{document}
554:
555: