1: \documentclass[12pt,preprint]{aastex}
2: % \usepackage{psfig,epsf}
3:
4: \shortauthors{J.J.M. in 't Zand et al.}
5: \shorttitle{Identification of counterparts to X-ray transient in NGC 6440}
6:
7: % LATEX start input:
8:
9: \begin{document}
10:
11: \slugcomment{Accepted for ApJL}
12: \title{Identification of the optical and quiescent counterparts to
13: the bright X-ray transient in NGC~6440}
14: \author{J.J.M.~in~'t~Zand\altaffilmark{1,2,3},
15: M.H.~van~Kerkwijk\altaffilmark{1},
16: D.~Pooley\altaffilmark{4},
17: F.~Verbunt\altaffilmark{1},
18: R.~Wijnands\altaffilmark{4,5},
19: W.H.G.~Lewin\altaffilmark{4}}
20: \affil{}
21: \altaffiltext{1}{Astronomical Institute, Utrecht University, P.O. Box 80 000, 3508 TA Utrecht, the Netherlands}
22: \altaffiltext{2}{SRON National Institute for Space Research,
23: Sorbonnelaan 2, 3584 CA Utrecht, the Netherlands}
24: \altaffiltext{3}{email jeanz@sron.nl}
25: \altaffiltext{4}{Center for Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139-4307}
26: \altaffiltext{5}{{\em Chandra\/} fellow}
27:
28: \begin{abstract}
29: After 3 years of quiescence, the globular cluster NGC~6440 exhibited a bright
30: transient X-ray source turning on in August 2001, as noted with the RXTE
31: All-Sky Monitor. We carried out a short target of opportunity observation
32: with the {\em Chandra\/} X-ray Observatory and are able to associate the
33: transient with the brightest of 24 X-ray sources detected during quiescence
34: in July 2000 with {\em Chandra}. Furthermore, we securely identify the optical
35: counterpart and determine that the 1998 X-ray outburst in NGC~6440 was from
36: the same object. This is the first time that an optical counterpart to a
37: transient in a globular cluster is securely identified. Since the transient is
38: a type I X-ray burster, it is established that the compact accretor is a
39: neutron star. Thus, this transient provides an ideal case to study the
40: quiescent emission in the optical and X-ray of a transiently accreting neutron
41: star while knowing the distance and reddening accurately. One model that fits
42: the quiescent spectrum is an absorbed power law plus neutron star hydrogen
43: atmosphere model. We find an intrinsic neutron star radius of
44: $17_{-12}^{+31}$~km and an unabsorbed bolometric luminosity for the neutron
45: star atmosphere of $(2.1\pm0.8)\times10^{33}$~erg~s$^{-1}$ which is consistent
46: with predictions for a cooling neutron star.
47: \end{abstract}
48:
49: \keywords{globular clusters: individual (NGC~6440) -- stars: neutron -- X-rays: binaries}
50:
51: %==============================================================================
52:
53: \section{Introduction}
54: \label{secintro}
55: Currently, thirteen bright X-ray sources are known to inhabit twelve Galactic
56: globular clusters (for a recent overview, see Sidoli et al. 2001; the 13th source
57: was recently discovered by White \& Angelini 2001). They are low-mass
58: X-ray binaries (LMXBs) of which at least 11
59: harbor a neutron star. Four are transient in nature: those in Liller 1,
60: Terzan 5 and 6, and NGC 6440 (one could argue that Terzan~1 contains a
61: transient source as well with an activity lasting decades, see Guainazzi,
62: Parmar \& Oosterbroek 1999). Thus far, none of these transients had a
63: securely identified optical counterpart. The prospects for finding one
64: were best for the transient in NGC~6440 because it has a relatively low
65: reddening of $E(B-V)=1.07$ and a non-extreme distance of $8.5\pm0.4$~kpc
66: (Ortolani, Barbuy \& Bica 1994). The transient in NGC~6440 was seen in
67: outburst in Dec. 1971 (Markert et al. 1975; Forman, Jones \& Tananbaum
68: 1976) and Aug. 1998 (In~'t~Zand et al. 1999; Verbunt et al. 2000). From ROSAT
69: HRI localizations of two sources in quiescence and optical observations during
70: the 1998 outburst, two candidate optical counterparts were identified,
71: V1 and V2 (Verbunt et al. 2000). V2 was evaluated to be the most likely one
72: with $B\approx22.7$ during outburst.
73:
74: Pooley et al. (2001) observed NGC~6440 for 23 ksec with the {\em Chandra\/}
75: X-ray Observatory on July 4, 2000, when the transient was quiescent. They
76: found 24 point sources with 0.5-2.5 keV luminosities between 1.8$\times10^{30}$
77: and $9.9\times10^{32}$~erg~s$^{-1}$, which they designate CX1 through CX24.
78: The brightest source, CX1, coincides with the optical counterpart candidate V2.
79:
80: The transient in NGC~6440 erupted again in August 2001, as detected with the
81: All-Sky Monitor (ASM) on the Rossi X-ray Timing Explorer RXTE\footnote{
82: as announced by the ASM team at {\tt http://xte.mit.edu}}.
83: We proceeded to secure the identification of the optical counterpart through
84: swift, high-resolution X-ray observations, and report the
85: results in the present paper. A preliminary report was already given by
86: In 't Zand et al. (2001).
87:
88: %==============================================================================
89: \section{Observations and source identification}
90: \label{section}
91:
92: The ASM (Levine et al. 1996; Bradt et al. 2000)
93: monitors NGC~6440 for 10 months of each year
94: at a detection threshold of about 15~mCrab in the 1.5 to 12 keV band per day of
95: observation. The transient exceeded this threshold during two weeks starting
96: on August 2, 2001. The peak flux was $60\pm10$~mCrab. Continued ASM monitoring
97: showed a reflaring episode starting on Sep. 10, with fluxes up to 0.2~Crab
98: and lasting for already almost 2 months at the time of this writing.
99: The Wide Field Cameras (WFCs) on BeppoSAX (Jager et al. 1997; In 't Zand 2001)
100: also observed NGC~6440 during this episode and detected at least 10 type-I
101: X-ray bursts. As soon as the community was alerted of the
102: outburst (on August 13), we proposed X-ray observations with {\em Chandra}
103: and optical observations with the European Southern Observatory's
104: New Technology Telescope in Director's Discretionary Time.
105:
106: The {\em Chandra\/} (Weiskopf, O'Dell \& van Speybroek 1996)
107: target-of-opportunity observation (TOO) was approved and carried out on August
108: 18.9 UT for an exposure time of 2.6~ksec with the Advanced CCD Imaging
109: Spectrometer (ACIS; Garmire, Nousek \& Bautz, in prep.) and no grating.
110: Despite the anticipated heavy pile up, we chose this configuration, and not
111: HRC, to optimize the sensitivity such that not only the transient but also a
112: few of the other sources in NGC~6440 would be detected (ACIS has 2 to 4
113: times more effective area for 0.3-8 keV photons than HRC-I). Thus, the
114: association of the transient with one
115: of the quiescent sources would become independent from the {\em Chandra\/}
116: attitude accuracy. Only a $2\arcmin\times8\arcmin$ subarray of the
117: back-illuminated S3 chip was read so that the CCD frame time was 0.8~s instead
118: of 3.2~s and pile up effects were reduced. Figure~\ref{figchandra} shows
119: the obtained image, zoomed in on the globular cluster and compared to the
120: observation during quiescence (Pooley et al. 2001). It is evident that the
121: transient was active. Furthermore, it is closest to CX1, the brightest of the
122: sources detected during quiescence. Based on the {\em Chandra\/}
123: aspect solution for the 2001 observation (accurate to about 0.5\arcsec) and
124: the astrometry carried out by Pooley et al. (2001) for the 2000 observation,
125: the angular distance between the transient and CX1 is 0.14\arcsec\ (the
126: position of the transient is determined through weighing pixel position with
127: pixel values). Of the remaining sources, CX2 and CX5 are clearly detected
128: during the TOO as well, with photon rates in the full bandpass of
129: $(8\pm2)\times10^{-3}$ and $(6\pm1)\times10^{-3}$~s$^{-1}$. These rates
130: compare well with those during the earlier observation
131: ($(7.6\pm0.6)\times10^{-3}$ and $(4.3\pm0.4)\times10^{-3}$~s$^{-1}$
132: respectively). The angular distances of the transient to CX2 and CX5 are
133: consistent with that to CX1 in the July 2000 image within 0.28$\pm$0.13\arcsec.
134: The distance is inconsistent with all of the other 23 sources from the July
135: 2000 observation by at least 15$\sigma$. This confirms the transient-CX1
136: association.
137:
138: A total of $2.6\times10^3$ events was detected within 8.0\arcsec\ from the
139: source centroid. In order to estimate the flux and avoid piled-up pixels,
140: we extracted a spectrum from the wings of the point spread
141: function (PSF) for pixels that are between 2.5 and 8.0\arcsec\ from the
142: centroid. We ignore negligible contributions from other sources in this
143: annulus. An absorbed power law fits the 0.7-8.0 keV spectral data well, with
144: a photon index of $1.2\pm0.1$, $N_{\rm H}=(8.2\pm1.6)\times10^{21}$~cm$^{-2}$
145: and $\chi^2_\nu=1.40$ ($\nu=75$ degrees of freedom).
146: After a correction for the fraction of the PSF covered (9\%), the flux is
147: $1.1\times10^{-10}$~erg~s$^{-1}$cm$^{-2}$ (0.5-10 keV) and the luminosity
148: $9\times10^{35}$~erg~s$^{-1}$ ($1.1\times10^{36}$~erg~s$^{-1}$ unabsorbed).
149: The statistical uncertainty is 7\%, that due to PSF uncertainties an estimated
150: 10\%. The flux is roughly 10
151: times fainter than at the peak and $10^3$ times brighter than during
152: quiescence (see below). The source shows no strong
153: variability, with an upper limit of 20\%
154: at 90\% confidence for 100~s time scales. No narrow features are apparent in
155: the power density spectrum.
156:
157: The position of CX1 (Pooley et al. 2001) is consistent with that of the
158: candidate optical counterpart V2 in Verbunt et al. (2000). They are
159: 0.46\arcsec\ apart while the uncertainty in the optical position alone is
160: 0.5\arcsec. We conclude that V2 is the secure optical counterpart to the
161: transient in NGC~6440, that CX1 is the quiescent X-ray counterpart, and that
162: the 1998 and 2001 outbursts were from the same object and not from any of the
163: other 3 or 4 transient low-mass X-ray binaries in quiescence that are likely
164: to be present in the cluster (Pooley et al. 2001). Most importantly, since
165: type-I X-ray bursts were detected from the transient during both 1998 and
166: 2001 outbursts, it is now firmly established that CX1 is a quiescent neutron
167: star.
168:
169: NGC~6440 was observed for us in Service Time with the New Technology Telescope
170: (NTT) at La Silla 3.5~d after the {\em Chandra\/} observation using the Superb
171: Seeing Imager SUSI2. The seeing was around 1\arcsec. Images were obtained in
172: the Bessell R and B filters. The analysis of these data is complicated
173: by proper motions of a number of stars in the field. However, the raw images
174: do show V2 to be active again in B at a level comparable to the 1998 outburst.
175:
176: %==============================================================================
177: \section{The quiescent X-ray spectrum}
178: \label{sectionspec}
179:
180: Quiescent neutron stars (qNSs) exhibit X-ray emission at typical luminosities
181: of $10^{32}-10^{33}$~erg~s$^{-1}$ (e.g., Asai et al. 1998, McClintock 2001).
182: A plausible explanation for this emission is thermal radiation from the
183: atmosphere of a neutron star that is cooling from the previous outburst
184: (Brown, Bildsten \&
185: Rutledge 1998). We extracted the spectrum of the transient during quiescence
186: by accumulating all 251 photons within 1.5\arcsec\ from the source centroid and
187: applying background corrections as found elsewhere in the image (amounting
188: merely to a 2.3 photons subtraction). To ensure applicability of the $\chi^2$
189: statistic, we
190: rebinned the data so that each channel contains at least 20 photons except
191: for the highest-energy bin which contains 15 photons. We used CIAO~2.1.3
192: Science Threads for the background subtraction and response matrix generation.
193: The spectrum was fitted between 0.3 and 6.7 keV (only 2 photons were detected
194: above this range) with a model combining a power law and a
195: neutron star hydrogen atmosphere model as calculated for the non-magnetic
196: case by Zavlin, Pavlov \& Shibanov (1996). Standard absorption by cold
197: interstellar matter of cosmological abundances was applied following Morrison
198: \& McCammon (1983). Due to the limited statistical quality, the fit errors
199: for the atmosphere temperature and radius are reasonably constrained only if
200: the mass of the neutron star, the distance, the interstellar hydrogen column
201: density $N_{\rm H}$ and the power law photon index are fixed. We did so at
202: values of 1.4~M$_\odot$ (for a 'canonical' neutron star), 8.5~kpc (this is well
203: determined), $N_{\rm H}=8.2\times10^{21}$~cm$^{-2}$ and 1.44 (these two
204: values were determined from the 1998 outburst; In 't Zand et al. 1999),
205: respectively. $N_{\rm H}$ has an uncertainty of roughly 40\%. The only three
206: parameters left free are the neutron star radius, the atmosphere temperature
207: and the power law normalization. The resulting fit is good, with $\chi^2_\nu=0.758$ (for $\nu=9$ degrees of freedom). The neutron star parameters are strongly
208: dependent. The fitted temperature is $92_{-33}^{+172}$~eV and the radius
209: $17_{-12}^{+31}$~km (90\% confidence). The radius to an observer at infinity is
210: $19_{-8}^{+29}$~km. A graph of the spectrum is presented in
211: figure~\ref{figspectrum} while in figure~\ref{figcontour} the 90\% confidence
212: region in temperature/radius space is presented. If $N_{\rm H}$ is left free,
213: it converges to $8.2^{+2.0}_{-1.3}\times10^{21}$~cm$^{-2}$ and the range of
214: allowed NS radii surpasses the 100 km mark. This upper radius margin is also
215: quite sensitive to the power law index which is poorly constrained by
216: the data.
217:
218: The inferred radius is consistent with that expected for a neutron star
219: (e.g., Lattimer \& Prakash 2001). We note that the model implies an intrinsic
220: radius that is 20\% larger if the NS mass is 1.8 instead of 1.4~M$_\odot$.
221: Fits with some other models are nearly as good (table~\ref{goodness}), so the
222: data are not discriminate in that sense. The absorbed 0.5-7 keV flux is
223: $(5.9\pm0.4)\times10^{-14}$~erg~s$^{-1}$cm$^{-2}$, the unabsorbed
224: flux $(2.0\pm0.6)\times10^{-13}$~erg~s$^{-1}$cm$^{-2}$ (the large error
225: reflects the uncertainty in $N_{\rm H}$). The power law contribution
226: in the latter flux is 20$\pm$6\% (90\% confidence); this is only 8\%
227: in 0.5 to 2 keV. The bolometric unabsorbed luminosity of the
228: NS atmosphere is $(2.1\pm0.8)\times10^{33}$~erg~s$^{-1}$ (the contribution
229: from the uncertainty in the distance is small).
230:
231: %==============================================================================
232: \section{Discussion}
233: \label{sectiondis}
234:
235: Pooley et al. (2001) find that there are 4 or 5 likely quiescent LMXB
236: transients in NGC~6440. Similar conclusions are currently being
237: reached from {\em Chandra\/} observations of other globular clusters such as
238: 47~Tuc (Grindlay et al. 2001a), Liller~1 (Homer et al. 2001), NGC~6652
239: (Heinke, Edmonds \& Grindlay 2001), $\omega$~Cen (Rutledge et al. 2001)
240: and NGC~6397 (Grindlay et al. 2001b).
241: Furthermore, for the first time two bright low-mass X-ray binaries have been
242: identified in a single cluster (M15; White \& Angelini 2001).
243: Our observations show unequivocally that the second and third transient
244: outburst in NGC 6440 (in 1998 and 2001) are from the same object.
245: Due to a limited positional accuracy of the source of the 1971 outburst,
246: it is not clear whether the same source was responsible for the 1971 and
247: the 1998/2001 outbursts. At first, indirect evidence suggested that the
248: source is different, because the duration of 1.5 to 3 months and the peak flux
249: of 250~mCrab deviate substantially from the values for the 1998 outburst
250: and for the initial phase of the 2001 outburst. However, at the time of this
251: writing the 2001 outburst shows activity for already 3 months and a peak flux
252: in excess of about 0.2 Crab units, according to ASM and WFC measurements.
253:
254: The recurrence time between the two most recent outbursts
255: is unexpectedly short. This may mean that the source has become
256: more active, or alternatively that earlier outbursts have been
257: missed. It is not uncommon for a previously quiet source, to
258: become active in a series of transient outbursts (e.g. GRO~J1744-28
259: and GRO~J1655-40). However, the 1998 outburst of the transient
260: in NGC6440 was so faint and short, that similar outbursts may
261: easily have passed undetected. We note that the RXTE ASM and
262: BeppoSAX WFCs became operational only in 1996. Similar coverage as the ASM
263: in the future should resolve this issue.
264:
265: This is the first time that an optical counterpart has firmly been established
266: for an X-ray transient in a globular cluster. Furthermore, it is the first
267: time that a quiescent neutron star has been identified without source
268: confusion in a globular cluster. These identifications are
269: important because they present the case of a quiescent neutron star
270: transient in the optical and X-rays for which an accurate distance is known.
271:
272: The spectral study of quiescent neutron stars is foremost interesting, as
273: Brown et al. (1998) pointed out, because it may result in an
274: accurate enough neutron star radius determination to be able to constrain the
275: equation of state of matter in the star's core. Measuring the radius to an
276: accuracy of about 0.5~km can exclude 50\% of the proposed equations of states
277: (Lattimer \& Prakash 2001). For neutron stars in globular clusters as compared
278: to those in the Galactic disk, one source of uncertainty is almost eliminated
279: --- the distance. The distance to NGC~6440 is accurate to 5\%. If other
280: uncertainties (in the spectral parameters) had been at least an order
281: of magnitude lower, the
282: prospects for constraining the equation of state for the transient in
283: NGC~6440 would have been good. However, this is not the case. This qNS would
284: have to be observed for a few million seconds with {\rm Chandra} before
285: it becomes worthwhile to consider this analysis, and that is without
286: considering uncertainties in the spectral models.
287:
288: Given a peak bolometric luminosity between $4\times10^{37}$ and
289: $1.6\times10^{38}$~erg~s$^{-1}$ (or 50 to 200~mCrab in 2-10 keV given the
290: broad-band spectrum measured by In~'t~Zand et al. 1999),
291: an e-folding decay time between 5~d (Verbunt et al. 2000) and 30~d, and
292: assuming that during accretion 100\% of the gravitational
293: potential energy is released through radiation, the recurrence time of
294: 2.95~yr implies a time-averaged accretion rate of
295: $<\dot{M}>\approx2\times10^{-11}$ to $4\times10^{-10}~M_\odot$~yr$^{-1}$.
296: Following Brown
297: et al.\ (1998), the quiescent time-averaged luminosity due to thermal
298: emission from the cooling NS is predicted to be between
299: $1\times10^{33}$ and $2\times10^{34}$~erg~s$^{-1}$. Our measurement of
300: $2.1\times10^{33}$~erg~s$^{-1}$ is consistent with this rough prediction,
301: also if less than 100\% of the liberated potential energy
302: is released as radiation.
303:
304: %------------------------------------------------------------------------------
305: \acknowledgements
306: We thank Harvey Tananbaum for the willingness to schedule the {\em Chandra\/}
307: observation at short notice, Yousaf Butt for help in planning the {\em
308: Chandra\/} observation, Olivier Hainaut
309: for carrying out the optical observations, Rob van der Meer for help in the
310: ACIS data analysis, Slava Zavlin and George Pavlov for providing their
311: neutron star atmosphere models in electronic form, and Jaap Schuurmans for
312: software support. JZ acknowledges financial support from the Netherlands
313: Organization for Scientific Research (NWO), MHvK through a fellowship of the
314: Royal Netherlands Academy of Sciences (KNAW), DP through a National Science
315: Foundation Graduate Fellowship, WL from NASA, and RW from NASA through a
316: Chandra fellowship.
317:
318: %==============================================================================
319: \begin{references}{}
320: \reference{Asai} Asai, K., Dotani, T., \& Hoshi, R. 1998, PASJ, 50, 611
321: \reference{Bradt} Bradt, H., Levine, A., Remillard, R., \& Smith, D.A. 2000,
322: in Proc. "X-ray Astronomy '99; Stellar Endpoints, AGN and the
323: Diffuse Background", eds. G.Malaguti, G.Palumbo, and N.White
324: (Gordon and Breach), in press (astro-ph/0003438)
325: \reference{Brown} Brown, E.F., Bildten, L., \& Rutledge, R.E. 1998, ApJ, 504, L95
326: \reference{Forman, Jones, Tananbaum} Forman, W., Jones, C., \& Tananbaum, H.
327: 1976, ApJ 207, L25
328: \reference{Grindlay} Grindlay, J., Heinke, C., Edmonds, P.D., \& Murray, S.S.
329: 2001a, Science, 292, 2290 (astro-ph/0105528)
330: \reference{Grindlay} Grindlay, J., Heinke, C., Edmonds, P.D., \& Murray, S.S.
331: 2001b, ApJL, submitted (astro-ph/0108265)
332: \reference{Guainazzi} Guainazzi, M., Parmar, A.N., \& Oosterbroek, T. 1999,
333: \aap, 349, 819
334: \reference{Homer} Homer, L., Deutsch, E.W., Anderson, S.F., \& Margon, B. 2001,
335: AJ, in press (astro-ph/0106140)
336: \reference{In 't Zand} In 't Zand, J.J.M., Verbunt, F., Strohmayer, T.E.,
337: Bazzano, A., Cocchi, M., Heise, J., van Kerkwijk, M.H., Muller,
338: J.M., Natalucci, L., Smith, M.J.S., \& Ubertini, P. 1999, A\&A, 345, 100
339: \reference{In 't Zand} In 't Zand, J.J.M. 2001, in Proc. 4th INTEGRAL
340: workshop, ``Exploring the Gamma-ray Universe'', Alicante (Spain),
341: Sep. 4-7, 2000, in press (astro-ph/0104299)
342: \reference{In 't Zand} In 't Zand, J.J.M., Pooley, D., Lewin, W.H.G.,
343: van Kerkwijk, M.H., \& Verbunt, F. 2001, ATEL 74
344: \reference{Jager} Jager, R., Mels, W.A., Brinkman, A.C., et al. 2001,
345: \aaps, 125, 557
346: \reference{Lattimer} Lattimer, J.M., Prakash, M. 2001, \apj, 550, 426
347: \reference{Levine et~al.} Levine, A., Bradt, H., Cui, W., et al. 1996,
348: ApJ 469, L33
349: \reference{Markert et al.} Markert, T.H., Backman, D.E., Canizares, C.R.,
350: Clark, G.W., \& Levine, A.M. 1975, Nat 257, 32
351: \reference{McClintock} McClintock, J.E. 2001, in Proc. Jan van Paradijs
352: Memorial Symposium, eds. E.P.J. van de Heuvel, L. Kaper, \&
353: E. Rol, PASP, in press (astro-ph//0109423)
354: \reference{Mitsuda}Mitsuda, K., Inoue, H., Koyama, K., et al. 1984, PASJ,
355: 36, 741
356: \reference{Morrison} Morrison, R., \& McCammon, D. 1983, \apj, 270, 119
357: \reference{Otolani, Barbuy \& Bica} Ortolani, S., Barbuy, B., \& Bica, E. 1994,
358: A\&AS 108, 653
359: \reference{Pooley} Pooley, D., Lewin, W.H.G., Verbunt, F., Margon, B., Miller,
360: J.M., Fox, D.W., Kaspi, V.M., \& van der Klis, M. 2001, submitted
361: \reference{Rutledge} Rutledge, R.E., Bildsten, L., Brown, E.F., Pavlov, G.,
362: \& Zavlin, V. 2001, ApJ, submitted (astro-ph/0105405)
363: \reference{Sidoli} Sidoli, L., Parmar, A.N., Oosterbroek, T., et al. 2001,
364: \aap, 368, 451
365: \reference{Verbunt} Verbunt, F., van Kerkwijk, M.H., In 't Zand, J.J.M.,
366: \& Heise, J. 2000, A\&A, 359, 960
367: \reference{Weiskopf} Weiskopf, M., O'Dell, S., \& van Speybroeck, L. 1996, in
368: Proc. SPIE , 2805, 142
369: \reference{White} White, N.E., \& Angelini, L. 2001, ApJ, 561, L101
370: \reference{Zavlin} Zavlin, V.E., Pavlov, G.G., \& Shibanov, Yu.A. 1996, \aap,
371: 315, 141
372: \end{references}
373:
374: %==============================================================================
375:
376: \newpage
377:
378: \begin{table}[t]
379: \caption[]{Goodness of fit for a variety of spectral models
380: to the 12-channel data between 0.3 and 6.7 keV.\label{goodness}}
381: \begin{tabular}{lllr}
382: \hline\hline
383: Model$^\dag$ & free & $\chi^2_\nu$ & $\nu$ \\
384: & params. & & \\
385: \hline
386: wa*nsa & $T,R$ & 2.74 &10 \\
387: wa*(nsa+po) & $T,R$,po-norm & 0.758 & 9 \\
388: wa*bb & $T$,bb-norm & 3.00 &10 \\
389: wa*(bb+po) & $T$,bb-norm,po-norm & 0.987 & 9 \\
390: wa*diskbb & $N_{\rm H}$,$T_{\rm in}$,norm & 1.94 & 9 \\
391: wa*diskbb & $T_{\rm in}$,norm & 2.44 &10 \\
392: \hline\hline
393: \end{tabular}
394:
395: \noindent
396: $^\dag$wa = interstellar absorption following Morrison
397: \& McCammon (1983), nsa = NS H-atmosphere emission according to Zavlin et
398: al. 1996, po = power law, bb = black body, diskbb = disk black body
399: according (Mitsuda et al. 1984)
400: \end{table}
401:
402: \begin{figure}
403: \plotone{f1.eps}
404: \caption{Chandra ACIS-S images of NGC~6440 during the transient active state
405: (left; 2.6 ks exposure) and quiescence (right; 23 ks exposure). The grids are
406: fixed to the equatorial coordinate system and defined in the same manner for
407: both cases. Pixels are 0\farcs5 on the side. The left panel shows the bright
408: transient with its large point spread function. The hole in the middle of the
409: PSF and the streak are artifacts introduced by the brightness of the transient.
410: The bright transient is coincident with quiescent source 'CX1' (Pooley et al.
411: 2001). CX1 is the
412: brightest X-ray source in the globular cluster during quiescence with a
413: 0.5-7 keV unabsorbed luminosity of $(1.5\pm0.5)\times10^{33}$~erg~s$^{-1}$.
414: \label{figchandra}
415: }
416: \end{figure}
417:
418: \begin{figure}
419: \plotone{f2.eps}
420: \caption{The spectrum of the quiescent emission of the
421: transient, as obtained in July 2000. The crosses refer to the measurements,
422: with the vertical lines indicating the errors, and the histogram refer to the
423: fluxes per channel as predicted by the best-fit NS atmosphere model plus
424: power law, after resetting the power law contribution to zero to illustrate
425: its impact at high energies. The resetting also illustrates that both
426: spectral components cover fairly distinct photon energies.
427: \label{figspectrum}
428: }
429: \end{figure}
430:
431: \begin{figure}
432: \plotone{f3.eps}
433: \caption{Contour plots of 90\%-confidence (R,T) region of NS atmosphere model
434: fit to CX1 spectrum during quiescence between 0.3 and 6.7 keV, for an observer
435: at infinity (left) or at the NS surface (right). It is assumed that the
436: distance is 8.5~kpc and
437: $N_{\rm H}=8.2\times10^{21}$~cm$^{-2}$. The location of the contour is
438: independent of NS mass for an observer at infinity.
439: \label{figcontour}
440: }
441: \end{figure}
442:
443: \end{document}
444: