1: \documentclass[12pt,preprint]{aastex}
2: % \documentclass[manuscript]{aastex}
3: % \documentclass[preprint2]{aastex}
4:
5: \newcommand{\etal}{et al.}
6: \newcommand{\fe}{Fe~K$\alpha$}
7: \newcommand{\kev}{keV}
8: \newcommand{\fouru}{4U~1820-30}
9:
10: %% You can insert a short comment on the title page using the command below.
11: %\slugcomment{}
12:
13: %\shorttitle{}
14: %\shortauthors{}
15:
16: \usepackage{emulateapj5}
17: %\usepackage{onecolfloat5}
18: \usepackage{times}
19: \usepackage{graphicx}
20:
21: %\received{2003 December 9}
22: \begin{document}
23:
24: \title{The evolution of the accretion disk around \fouru\ during a
25: superburst}
26: \author{D.R. Ballantyne\altaffilmark{1} and
27: T.E. Strohmayer\altaffilmark{2}}
28: \altaffiltext{1}{Canadian Institute for Theoretical
29: Astrophysics, 60 St. George Street, Toronto, Ontario, Canada
30: M5S~3H8; ballantyne@cita.utoronto.ca}
31: \altaffiltext{2}{NASA Goddard Space Flight Center, Laboratory
32: for High Energy Astrophysics, Code 680, Greenbelt, MD, USA 20771;
33: stroh@clarence.gsfc.nasa.gov}
34:
35: \begin{abstract}
36: Accretion from a disk onto a collapsed, relativistic star -- a neutron
37: star or black hole -- is the mechanism widely believed to be
38: responsible for the emission from compact X-ray
39: binaries. Because of the extreme spatial resolution
40: required, it is not yet possible to directly observe the evolution or
41: dynamics of the inner parts of the accretion disk where general
42: relativistic effects are dominant. Here, we use the bright X-ray
43: emission from a superburst on the surface of the neutron star \fouru\
44: as a spotlight to illuminate the disk surface. The X-rays cause iron
45: atoms in the disk to fluoresce, allowing a determination of the
46: ionization state, covering factor and inner radius of the disk over
47: the course of the burst. The time-resolved spectral fitting shows that
48: the inner region of the disk is disrupted by the burst, possibly being
49: heated into a thicker, more tenuous flow, before recovering its
50: previous form in $\sim 1000$~s. This marks the first instance that the
51: evolution of the inner regions of an accretion disk has been observed
52: in real-time.
53: \end{abstract}
54:
55: \keywords{accretion, accretion disks --- line: formation --- stars:
56: individual (\fouru) --- stars: neutron --- X-rays: binaries ---
57: X-rays: bursts}
58:
59: \section{Introduction}
60: \label{sect:intro}
61: Superbursts \citep[e.g.][]{cor00,sb01,kuul02} are the more powerful and rarer
62: relations of the Type~I X-ray bursts, and are thought to be due to the
63: nuclear burning of the ashes (principally carbon and rp-process
64: nuclei) that remain after thermonuclear processing of the accreted
65: light elements \citep{sb01,cumm03}. One of the defining characteristics
66: of a superburst is its long decay time of a number of hours,
67: three orders of magnitude greater than a typical Type~I
68: burst \citep{kuul03}. This allows high signal-to-noise spectra to be
69: accumulated at regular intervals throughout the course of the burst.
70:
71: \fouru\ is a compact low-mass X-ray binary with a period of only
72: 11.4~minutes \citep{spw87}, and lies within the globular cluster
73: NGC~6624 at a distance of $\sim 7$~kpc
74: \citep{vlv86,hs85,rml93}. Type~I X-ray bursts have been observed from
75: this system for nearly 30 years \citep{gg76}. The accretion rate
76: inferred from the persistent X-ray luminosity is $\sim 1.5\times
77: 10^{17}$~g~s$^{-1}$ \citep{cumm03}. The first, and so far only,
78: superburst seen from \fouru\ was observed by the \textit{Rossi X-ray Timing
79: Explorer} (\textit{RXTE}) on 1999 September 9 (UT) during a scheduled
80: monitoring observation. The initial analysis and interpretation of the
81: data was presented by \citet{sb01}. Time-resolved spectroscopy
82: presented by these authors showed that an iron K$\alpha$ line and absorption
83: edge were present in the spectra, and that their properties changed
84: over time. A likely origin for these features is reprocessing of the
85: X-rays by the accretion disk \citep{dd91}. This process, called X-ray
86: reflection, is commonly used to explain similar features in the
87: spectra of Seyfert galaxies \citep*[e.g.][]{pou90,gf91,bal03}.
88:
89: In this paper we present the results of fitting the time-resolved
90: spectra of the superburst with detailed models of X-ray
91: reflection. The next section describes the model calculations and
92: details of the spectral fitting. The results are presented in
93: Section~\ref{sect:res} and discussed in Section~\ref{sect:discuss}.
94:
95: \section{Model Calculations}
96: \label{sect:models}
97: The X-ray reflection models were calculated with the code of
98: \citet{ros93} (see also \citealt*{ros99,bif01,brf01}). A
99: one-dimensional slab of uniform density gas is illuminated by a
100: blackbody continuum (defined from 1~eV to 100~\kev) of temperature
101: $kT$, and allowed to come into thermal and ionization balance. The
102: emergent spectrum is calculated with a Fokker-Planck/diffusion method
103: that naturally accounts for the energy redistribution of photons by
104: Compton scattering \citep*{rwm78}. The following ions are included in the
105: calculations: \ion{C}{5} \ion{--}{7}, \ion{N}{6} \ion{--}{8},
106: \ion{O}{5} \ion{--}{9}, \ion{Mg}{9} \ion{--}{13}, \ion{Si}{11}
107: \ion{--}{15} and \ion{Fe}{16} \ion{--}{27}. Fluorescence and
108: recombination lines from these ions are computed and transferred
109: simultaneously with the continuum, resulting in a self-consistent
110: prediction of the emitted spectrum.
111:
112: The compact orbit of this binary suggests that the donor is an evolved
113: low-mass helium star \citep{rap87}, and thus the accretion disk
114: material will be He-rich and H-poor \citep{cumm03}. Therefore, we
115: assumed the reflecting medium has the He, C, N, and O abundances of
116: cool extreme helium stars from the tabulation of \citet{pan01}. The
117: Mg, Si and Fe abundances were left at their solar values
118: \citep{gns96}, but the effective metallicity is increased (by $\sim
119: 3$) due to the negligible levels of hydrogen. The helium is assumed to
120: be completely ionized everywhere in the reflecting region.
121:
122: The computed reflection spectra are parameterized by the ionization
123: parameter $\xi=4 \pi F_{\mathrm{X}} / n_{\mathrm{He}}$ of the
124: irradiated gas, where $F_{\mathrm{X}}$ is the flux of the incident
125: X-rays and $n_{\mathrm{He}}=10^{15}$~cm$^{-3}$ is the assumed number
126: density of helium. Since the blackbody emission is also seen directly
127: from the neutron star, the fitted spectrum is the sum of the blackbody
128: and the reflected emission, weighted by a factor $R$, the reflection
129: fraction. This parameter measures the relative importance of
130: reflection in the spectrum, and can be related to the covering factor
131: of the accretion disk.
132:
133: \citet{sb01} describes the \textit{RXTE} data and the process of
134: accumulating the $\sim 80$ 64~s spectra during the burst. The preburst
135: spectrum was used as the background, and was thus subtracted from the
136: data during spectral fitting \citep{sb01}. This persistent emission
137: (amounting to a luminosity of $1.8\times 10^{37}$~erg~s$^{-1}$) is
138: driven by the accretion onto the neutron star, so using this spectrum
139: as a background implicitly assumes that the burst will not effect this
140: underlying emission. As we see below, we cannot be sure that this does
141: not happen; however, \citet{sb01} found little difference in their
142: spectral fits between the preburst background and the detector
143: background. Since the flux of the superburst varies from 20$\times$ to
144: 3$\times$ greater than the preburst emission, then the shape of the
145: background will not greatly effect the fit results.
146:
147: The spectral fitting was performed with \textsc{xspec}v.11.2.0bp
148: \citep{arn96}. The model consisted of a grid of reflection spectra
149: (read in using the \texttt{atable} command) and attenuation from
150: neutral gas along the line of sight (using the \texttt{wabs} model),
151: parameterized by the column density $N_{\mathrm{H}}$. The Galactic
152: absorption toward \fouru\ is $1.5\times 10^{21}$~cm$^{-2}$
153: \citep{dl90}. The reflection spectra were blurred by the effects of
154: relativistic motion within the Schwarzschild metric \citep{fab89}. The
155: inner radius of emission ($r_{\mathrm{in}}$) was allowed to vary, but
156: the emissivity was fixed as $r^{-3}$, where $r$ is distance along the
157: accretion disk, and the outer radius was frozen at $200$~$GM/c^{2}$,
158: where $M$ is the mass of the neutron star. The inclination angle of
159: the disk to the line of sight was fixed at 30$^{\circ}$. The fitted
160: energy range was 3--40~\kev, but was changed to 3--15~\kev\ 7000~s
161: into the burst because of background dominating at higher
162: energies. The errorbars on the best fit parameters were computed using
163: the 2$\sigma$ uncertainties for one parameter of interest.
164:
165: \section{Spectral Fitting Results}
166: \label{sect:res}
167: The results of fitting the reflection models to the superburst data
168: are shown in Figure~\ref{fig:fitres}. We find that an ionized
169: reflector provides a good description to the spectra for most of the
170: burst. During the first 1000~s the spectra are well fit with a large
171: $\xi$, indicating significant ionization of the accretion disk. Over
172: the next $\sim 500$~s $\xi$ decreases by more than an order of
173: magnitude. For times between $\sim 2500$ and $3000$~s the spectra
174: cannot be fit by the reflection model, except for the last three
175: points, where only very loose constraints could be placed on
176: $\xi$. \citet{sb01} showed the spectrum was hardening very rapidly
177: during that period, which indicates that some rapid evolution of the
178: system was ongoing, and is not well modeled by our equilibrium
179: calculations. However, the recombination timescale for H-like Fe is
180: $t_{\mathrm{recomb}} \sim 1/n_{e}\alpha_{\mathrm{Fe}}(T)$. The slowest
181: timescale occurs at the highest temperature, which is $10^7$~K at the
182: surface of the reflector, so that $t_{\mathrm{recomb}} \sim
183: 10^{-4}$~s, assuming the recombination coefficient of
184: \citet{vf96}. This is the same order of magnitude as the disk
185: dynamical time at 10~$GM/c^2$. The recombination timescale of
186: \ion{He}{2} is $\sim 0.1$~s, much longer than the dynamical time, but
187: faster than the 64~s spectra examined here. Therefore, the assumptions
188: of ionization equilibrium or fully stripped helium cannot be the cause of the poor fits in this
189: region of the lightcurve. A more fundamental change in the X-ray
190: source or reflecting geometry may be required. Finally, after the last
191: Earth occultation passage, the system was revealed to have settled
192: into a final decaying state, with $\xi \sim 100$~erg~cm~s$^{-1}$,
193: although one other non-equilibrium event took place at $\sim 6000$~s.
194:
195: While the disk inclination angle was fixed at 30$^{\circ}$ for the
196: spectral fits, there is a previous constraint of
197: 35$^{\circ}$--50$^{\circ}$ found by \citet{and97} from modulations in
198: the ultraviolet flux (due to X-ray heating of the secondary). Relativistic disk
199: features are also dependent on the accretion disk inclination angle
200: \citep{fab89}, so can provide a consistency check on the reflection
201: hypothesis. In Figure~\ref{fig:incl} we plot the 68\%, 90\% and 99\%
202: confidence contours for $r_{\mathrm{in}}$ and the disk inclination
203: angle 5500~s into the superburst. This segment was chosen as it has a
204: low reduced $\chi^2$ and strong reflection features. While the disk
205: fit seems to prefer a lower inclination angle, it is consistent with
206: the results of \citet{and97} at the 2$\sigma$ level.
207:
208: The observed evolution of $\xi$ from highly to weakly ionized is to be
209: expected since the observed flux decays over the course of the burst
210: \citep{sb01}. More fundamental insights from this analysis arise from
211: the variations of $N_{\mathrm{H}}$, $R$ and $r_{\mathrm{in}}$ over the
212: first 2000~s of the superburst. The changes of both
213: $r_{\mathrm{in}}$, which measures the inner radius of the disk (as
214: judged from the width of the iron line and edge), and $R$, the
215: reflection fraction, strongly indicates that the accretion disk is
216: significantly affected by the superburst out to a radius of $\sim
217: 100$~$GM/c^2$. In addition, the line-of-sight absorbing column also
218: went through a similar variation, either as a result of the burst
219: itself or the changes to the inner accretion disk. All three of these
220: parameters follow a similar pattern where they increase
221: ($r_{\mathrm{in}}$ and $N_{\mathrm{H}}$) or decrease ($R$) before
222: returning to close to their initial value. This behavior seems to be
223: closely linked to the blackbody temperature $kT$, rather than the
224: flux, which decreases monotonically during the burst.
225:
226: \section{Discussion}
227: \label{sect:discuss}
228: A likely scenario to explain the observed behavior is that the
229: accretion flow in the inner disk is altered by the superburst so that
230: it loses the ability to produce reflection features. This could be
231: accomplished if the scale-height of the disk, $H$, increased due to
232: heating of the gas by the radiation field. This would decrease the
233: surface density of the disk, allowing it to be more easily and
234: completely ionized by the burst radiation, removing any spectral
235: features in the observable band, and effectively becoming a mirror for
236: the blackbody spectrum. As there would be no evidence for reprocessing
237: from this innermost region, the estimate of the inner radius would be
238: from further out in the disk where iron can still produce an emission
239: line. The additional blackbody emission reflected from the ionized
240: mirror, would dilute the reflection features, resulting in the low
241: value of $R$. Alternatively, the lower surface density could reduce
242: the optical depth of the gas to the point where the inner disk becomes
243: a very inefficient reflector, while only the outer disk would be
244: optically thick enough for efficient reprocessing.
245:
246: The main evidence for this thermal effect is seen in the correlations
247: between $r_{\mathrm{in}}$, $R$ and $kT$. The Compton temperature of
248: the gas is $kT_{\mathrm{C}} \approx kT$, and reached $\sim 3$~\kev,
249: about twice the maximum disk temperature \citep{ss73}. Since $H
250: \propto c_{\mathrm{s}}r^{3/2} \propto T^{1/2} r^{3/2}$, where
251: $c_{\mathrm{s}}$ is the local sound speed, then $H$ will increase by
252: $\sim 40$\% from its non-illuminated value. Changes to the disk
253: surface density can take place over a viscous time which is $\sim
254: 1000$~s at 100~$GM/c^{2}$ if $\alpha \sim 0.2$, where $\alpha$ is the
255: Shakura-Sunyaev viscosity parameter \citep{ss73}. This timescale
256: compares favorably with the observed changes in $R$ and
257: $r_{\mathrm{in}}$. An increase in $H$ may also explain the evolution
258: of $N_{\mathrm{H}}$ as the thicker disk may bring clumps of disk
259: material into the line-of-sight. Data which extend below the 3~\kev\
260: limit of \textit{RXTE} would be needed to test if the absorption was
261: caused by ionized or neutral material. Under this hot disk scenario,
262: once the peak in $kT$ is passed the disk material cooled to lower and
263: lower temperatures, and thus contracted to its original
264: size. Simultaneously, the amount of material available to reflect
265: increased, so $R$ grew larger, but $\xi$ dropped due to a combination
266: of the increasing surface density and the decaying burst flux.
267:
268: The connection between the ionization parameter and the disk surface
269: density $\Sigma$ can be formulated as $\xi \approx L_{\mathrm{X}}
270: m_{\mathrm{He}} H / r^2 \Sigma$, where $L_{\mathrm{X}}$ is the
271: luminosity of the burst, and $m_{\mathrm{He}}$ is the mass of a helium
272: nucleus. Using the equations for a gas pressure dominated
273: Shakura-Sunyaev disk with electron scattering opacity \citep{ss73}, a
274: peak burst luminosity of $3.4 \times 10^{38}$~erg~s$^{-1}$ \citep{sb01}, and
275: $\alpha=0.2$, we find $\xi \approx 100$~erg~cm~s$^{-1}$ for
276: $r=15$~$GM/c^2$, which is consistent with the values of $\xi$ found
277: near the end of the burst (Fig.~1). From the observed decay in
278: flux \citep{sb01}, the surface density of the disk must be $\sim 100$
279: times less than the Shakura-Sunyaev value at the start of the
280: superburst, but only $\sim 6$ times less near the end in order to
281: obtain the necessary level of ionization. This indicates the surface
282: density of the inner accretion disk was significantly lower at the
283: peak of the burst but increased as the burst decayed, supporting the
284: interpretation that the disk material was severely affected by the explosion.
285:
286: Another process that could lower the surface density at the inner edge
287: of an accretion disk is the interaction between the radiation field
288: and material flowing in the disk. Radiation can remove angular
289: momentum from the disk material, causing the disk to empty inward,
290: thus lowering the surface density. There exists only a few
291: calculations of the interaction between an accretion disk and the
292: radiation field from a luminous central source
293: \citep{wm89,walk92,ml93,ml96}. These models predict that the radiation
294: will exert a significant torque on the accretion flow, which increases
295: the accretion rate close to the central object. This effect may be
296: enhanced in the case of \fouru\ due to the power of the superburst (as
297: compared to the typical Type~I burst considered in these models), as
298: well as an interaction between the disk and the expanding shell that
299: was ejected from the neutron star just prior to the superburst
300: \citep{sb01} -- the radius-expansion phase \citep{lvb84}. It is
301: difficult to rigorously compare this idea to the observations because
302: detailed modeling of the disk-radiation interaction for a superburst
303: or a radius-expansion burst has not been performed.
304:
305: By the end of the observation, the accretion disk has recovered from
306: its disruption by the burst, and is reflective down to a radius of
307: 10--20~$GM/c^2$, which is curiously larger than the innermost stable
308: circular orbit of 6~$GM/c^2$. Since this is not an X-ray pulsar, the
309: disk would not be truncated by the star's magnetosphere. Rather, this
310: may be evidence for the disk-radiation interaction described above,
311: and the material within 10--20~$GM/c^2$ is no longer dense enough to
312: provide a reflection signature. At these late times, the fits also
313: require strong reflection, with $R>1$, indicating that the disk covers
314: a large solid angle from the X-ray source, and thus is thin relative
315: to the radius of the neutron star or has a non-negligible flare at
316: large radii. The latter possibility is supported by the reflection
317: from $r > 100$~$GM/c^2$ observed early in the burst. If the inner disk
318: expanded from the X-ray heating, then a significant disk flare might
319: be necessary in order to allow the distant material to be irradiated
320: by the blackbody emission.
321:
322: Dipping behavior is seen in the X-ray lightcurve of \fouru\ between
323: 6000--7500~s into the burst (see Fig.~7 in \citealt{sb01}). This kind
324: of variability in low-mass X-ray binaries is often explained as
325: variable absorption from a two-component ionized medium close to the
326: neutron star \citep{fkl87}. The dipping observed from \fouru\ could
327: then be due to the reformation of the accretion disk, however, the
328: timescale is incorrect, with the dipping seen $\sim 2000$~s after the
329: probable recollapse of the disk. However, there is an interesting
330: correspondence between the times when dipping is observed and an
331: increase in $N_{\mathrm{H}}$ derived from spectral fitting
332: (Fig.~\ref{fig:fitres}). This may indicate that the dipping is caused
333: by partially ionized material that is slowly recombining, but farther away
334: from the accretion disk. An obvious source for such material is the
335: gas that was cast off from the neutron star during the radius
336: expansion phase. This material could be at a large distance from the
337: system ($10^{14}$~cm if ejected at the escape velocity).
338:
339: \acknowledgments
340:
341: We thank E. Agol, L. Bildsten, E. Quataert, N. Murray \& C. Matzner
342: for helpful discussions, and A. Fabian for providing the relativistic
343: blurring functions. DRB is supported by the Natural Sciences and
344: Engineering Research Council of Canada.
345:
346: \begin{thebibliography}{}
347: \bibitem[\protect\citeauthoryear{Anderson
348: \etal}{1997}]{and97}Anderson, S.F., Margon, B., Deutsch, E.W.,
349: Downes, R.A. \& Allen, R.G., 1997, \apj, 482, L69
350: \bibitem[\protect\citeauthoryear{Arnaud}{1996}]{arn96} Arnaud K.A.,
351: 1996, in Jacoby G., Barnes J., eds, Astronomical Data Analysis
352: Software and Systems V, ASP Conf.\ Ser.\ Vol.\ 101, 17
353: \bibitem[\protect\citeauthoryear{Ballantyne, Iwasawa \&
354: Fabian}{Ballantyne \etal}{2001}]{bif01} Ballantyne, D.R.,
355: Iwasawa, K. \& Fabian, A.C., 2001, \mnras, 323, 506
356: \bibitem[\protect\citeauthoryear{Ballantyne, Ross \&
357: Fabian}{Ballantyne \etal}{2001}]{brf01} Ballantyne, D.R., Ross, R.R. \&
358: Fabian A.C., 2001, \mnras, 327, 10
359: \bibitem[\protect\citeauthoryear{Ballantyne, Vaughan \&
360: Fabian}{Ballantyne \etal}{2003}]{bal03} Ballantyne, D.R., Vaughan,
361: S. \& Fabian, A.C., 2003, \mnras, 342, 239
362: \bibitem[\protect\citeauthoryear{Cornelisse
363: \etal}{2000}]{cor00}Cornelisse, R., Heise, J., Kuulkers, E.,
364: Verbunt, F. \& in't Zand, J.J.M., 2000, \aap, 357, L21
365: \bibitem[\protect\citeauthoryear{Cumming}{2003}]{cumm03}Cumming, A.,
366: 2003, \apj, 595, 1077
367: \bibitem[\protect\citeauthoryear{Day \& Done}{1991}]{dd91}Day,
368: C.S.R. \& Done, C, 1991, \mnras, 253, 35\textsc{p}
369: \bibitem[\protect\citeauthoryear{Dickey \&
370: Lockman}{1990}]{dl90}Dickey, J.M. \& Lockman, F.J., 1990, \araa, 28,
371: 215
372: \bibitem[\protect\citeauthoryear{Fabian \etal}{1989}]{fab89}Fabian,
373: A.C., Rees, M.J., Stella, L. \& White, N.E., 1989, \mnras, 238, 729
374: \bibitem[\protect\citeauthoryear{Frank, King \& Lasota}{1987}]{fkl87}
375: Frank, J., King, A.R. \& Lasota, J.-P., 1987, \aap, 178, 137
376: \bibitem[\protect\citeauthoryear{George \& Fabian}{1991}]{gf91} George,
377: I.M. \& Fabian, A.C., 1991, \mnras, 249, 352
378: \bibitem[\protect\citeauthoryear{Grevesse, Noels \&
379: Sauval}{1996}]{gns96}Grevesse, N., Noels, A. \& Sauval, A.J. 1996,
380: in Holt, S.S. \& Sonneborn, G., eds, Cosmic Abundances, ASP Conf.\
381: Ser.\ Vol.\ 99, 117
382: \bibitem[\protect\citeauthoryear{Grindlay \&
383: Gursky}{1976}]{gg76}Grindlay, J. \& Gursky, H., 1996, \apj, 205,
384: L131
385: \bibitem[\protect\citeauthoryear{Hesser \& Shawl}{1985}]{hs85}Hesser,
386: J.E. \& Shawl, S.J., 1985, \pasp, 97, 465
387: \bibitem[\protect\citeauthoryear{Kuulkers
388: \etal}{2002}]{kuul02}Kuulkers, E. \etal, 2002, \aap, 382, 503
389: \bibitem[\protect\citeauthoryear{Kuulkers}{2003}]{kuul03}Kuulkers,
390: E. 2003, in van den Heuvel, E.P.J., in 't Zand, J.J.M. \&
391: Wijers, R.A.M.J. eds, The Restless High-Energy Universe, in press
392: (astro-ph/0310402)
393: \bibitem[\protect\citeauthoryear{Lewin, Vacca \&
394: Basinska}{1984}]{lvb84}Lewin, W.H.G., Vacca, W.D. \& Basinska, E.M., 1984,
395: \apj, 277, L57
396: \bibitem[\protect\citeauthoryear{Miller \& Lamb}{1993}]{ml93}Miller,
397: M.C. \& Lamb, F.K., 1993, \apj, 413, L43
398: \bibitem[\protect\citeauthoryear{Miller \& Lamb}{1996}]{ml96}Miller,
399: M.C. \& Lamb, F.K., 1996, \apj, 470, 1033
400: \bibitem[\protect\citeauthoryear{Pandey \etal}{2001}]{pan01}Pandey,
401: G., Rao, N.K., Lambert, D.L., Jeffery, C.S. \& Asplund, M., 2001,
402: \mnras, 324, 937
403: \bibitem[\protect\citeauthoryear{Pounds \etal}{1990}]{pou90} Pounds,
404: K.A., Nandra, K., Stewart, G.C., George, I.M. \& Fabian A.C., 1990, \nat,
405: 344, 132
406: \bibitem[\protect\citeauthoryear{Rappaport
407: \etal}{1987}]{rap87}Rappaport, S., Nelson, L.A., Ma, C.P. \& Joss,
408: P.C., 1987, \apj, 322, 842
409: \bibitem[\protect\citeauthoryear{Rich, Minniti \&
410: Liebert}{1993}]{rml93}Rich, R.M., Minniti, S. \& Liebert, J., 1993,
411: \apj, 406, 489
412: \bibitem[\protect\citeauthoryear{Ross \& Fabian}{1993}]{ros93} Ross,
413: R.R. \& Fabian, A.C., 1993, \mnras, 261, 74
414: \bibitem[\protect\citeauthoryear{Ross, Weaver \& McCray}{Ross
415: \etal}{1978}]{rwm78} Ross, R.R., Weaver, R. \& McCray, R., 1978, \apj, 219, 292
416: \bibitem[\protect\citeauthoryear{Ross, Fabian \& Young}{Ross
417: \etal}{1999}]{ros99} Ross, R.R., Fabian, A.C. \& Young, A.J., 1999, \mnras,
418: 306, 461
419: \bibitem[\protect\citeauthoryear{Shakura \& Sunyaev}{1973}]{ss73}
420: Shakura, N.I. \& Sunyaev, R.A., 1973, \aap, 24, 337
421: \bibitem[\protect\citeauthoryear{Stella, Priedhorsky \&
422: White}{1987}]{spw87}Stella, L., Priedhorsky, W. \& White, N.E.,
423: 1987, \apj, 312, L17
424: \bibitem[\protect\citeauthoryear{Strohmayer \&
425: Brown}{2002}]{sb01}Strohmayer, T.E. \& Brown, E.F., 2002, \apj, 566,
426: 1045
427: \bibitem[\protect\citeauthoryear{Vacca, Lewin \& van
428: Paradijs}{1986}]{vlv86}Vacca, W.D., Lewin, W.H.G. \& van Paradijs,
429: J., 1986, \mnras, 220, 339
430: \bibitem[\protect\citeauthoryear{Verner \& Ferland}{1996}]{vf96}
431: Verner, D.A. \& Ferland, G.F., 1996, \apjs, 103, 467
432: \bibitem[\protect\citeauthoryear{Walker \&
433: M\'{e}sz\'{a}ros}{1989}]{wm89}Walker, M.A. \& M\'{e}sz\'{a}ros, P.,
434: 1989, 346, 844
435: \bibitem[\protect\citeauthoryear{Walker}{1992}]{walk92}Walker, M.A.,
436: 1992, \apj, 385, 642
437: \end{thebibliography}
438:
439:
440:
441: \clearpage
442:
443: \begin{figure}
444: \centerline{
445: \includegraphics[angle=-90,width=0.7\textwidth]{f1a.eps}
446: }
447: \centerline{
448: \includegraphics[angle=-90,width=0.7\textwidth]{f1b.eps}
449: }
450: \caption{\footnotesize Results of fitting the time-resolved superburst spectra of
451: \fouru\ with ionized reflection models. The top panel shows the
452: evolution of the reflection parameters: the ionization parameter
453: $\xi$, the reflection fraction, and inner radius of the disk,
454: $r_{\mathrm{in}}$. The lower panel describes the evolution of the
455: absorption column, $kT$ of the incident blackbody, and the reduced
456: $\chi^2$ of the spectral fits. The reflection fraction,
457: $r_{\mathrm{in}}$, and $N_{\mathrm{H}}$ seem to be closely related to
458: the evolution of $kT$. Data were fit from 3--40~\kev\ for the points
459: to left of the dashed line, and from 3--15~\kev\ for the points to the
460: right. Time equals zero corresponds to about 100~s after the start of
461: the burst. The hatched region denotes the interval where the spectra
462: were hardening rapidly, and were not well described by the equilibrium
463: reflection model.}
464: \label{fig:fitres}
465: \end{figure}
466:
467: \clearpage
468:
469: \begin{figure}
470: \centerline{
471: \includegraphics[angle=-90,width=0.8\textwidth]{f2.eps}
472: }
473: \caption{Joint 68\%, 90\% and 99\% confidence contours on
474: $r_{\mathrm{in}}$ and the disk inclination angle for a spectral fit
475: 5500~s into the superburst. Modulations in the ultraviolet flux from
476: \fouru\ previously constrained the inclination angle to between
477: $35^{\circ}$ and $50^{\circ}$ \citep{and97}, which is consistent with
478: the results from disk reflection at the 2$\sigma$ level.}
479: \label{fig:incl}
480: \end{figure}
481:
482: \end{document}
483:
484:
485:
486:
487: