1: %\documentclass{article}
2: %\usepackage{natbib,emulateapj,psfig}
3: %\documentclass[manuscript]{aastex}
4: \documentclass[preprint,12pt]{aastex}
5: \usepackage{psfig}
6: \newcommand{\beq}{\begin{equation}}
7: \newcommand{\eeq}{\end{equation}}
8: \newcommand{\bea}{\begin{eqnarray}}
9: \newcommand{\eea}{\end{eqnarray}}
10: \newcommand{\bean}{\begin{eqnarray*}}
11: \newcommand{\eean}{\end{eqnarray*}}
12: \newcommand{\ba}{\begin{array}}
13: \newcommand{\ea}{\end{array}}
14: \newcommand{\bml}{\begin{mathletters}}
15: \newcommand{\eml}{\end{mathletters}}
16: \newcommand{\rem}[1]{{ }}
17: \bibpunct[,]{(}{)}{;}{a}{}{,}
18: \begin{document}
19: \title{On the Lack of Type I X-ray Bursts in Black Hole X-ray Binaries:
20: Evidence for the Event Horizon?}
21:
22: \author{Ramesh Narayan\altaffilmark{1,2} \& Jeremy S. Heyl\altaffilmark{2,3}}
23:
24: \altaffiltext{1}{Institute for Advanced Study, Princeton, NJ 08540}
25:
26: \altaffiltext{2}{Harvard-Smithsonian Center for Astrophysics, Cambridge,
27: MA 02138; rnarayan@cfa.harvard.edu; jheyl@cfa.harvard.edu}
28:
29: \altaffiltext{3}{Chandra Fellow}
30:
31: \begin{abstract}
32:
33: Type I X-ray bursts are very common in neutron star X-ray binaries,
34: but no Type I burst has been seen in the dozen or so binaries in which
35: the accreting compact star is too massive to be a neutron star and
36: therefore is identified as a black hole candidate. We have carried
37: out a global linear stability analysis of the accumulating fuel on the
38: surface of a compact star to identify the conditions under which
39: thermonuclear bursts are triggered. Our analysis, which improves on
40: previous calculations, reproduces the gross observational trends of
41: bursts in neutron star systems. It further shows that, if black hole
42: candidates have surfaces, they would very likely exhibit instabilities
43: similar to those that lead to Type~I bursts on neutron stars. The
44: lack of bursts in black hole candidates is thus significant, and
45: indicates that these objects have event horizons. We discuss possible
46: caveats to this conclusion.
47: \end{abstract} \keywords{accretion --- black hole physics --- X-rays:
48: binaries, bursts}
49:
50:
51: \section{Introduction}
52:
53:
54:
55: When gas accretes onto a neutron star (NS) in a low-mass X-ray binary
56: (LMXB), it is compressed and heated as it accumulates on the surface,
57: leading to thermonuclear reactions. In many NS LMXBs, the reactions
58: occur unsteadily and cause Type I X-ray bursts (Grindlay et al. 1976).
59: %(Type I bursts are so named to distinguish them from Type II bursts,
60: %which are not thermonuclear and are observationally distinct, e.g.,
61: %Hoffman, Marshall \& Lewin 1978.)
62: Type I bursts have been observed in a large number of NS LMXBs (see
63: Lewin, van Paradijs \& Taam 1993; Strohmayer, Swank \& Zhang 1998 for
64: reviews), and the theory of these bursts is relatively well understood
65: (Hansen \& van Horn 1975; Woosley \& Taam 1976; Joss 1977; Taam \&
66: Picklum 1978;
67: %Fujimoto, Hanawa \& Miyaji 1981,
68: Paczy\'nski 1982, hereafter P82; Fujimoto et al. 1984, 1987; Fushiki \&
69: Lamb 1987, hereafter FL87; Taam, Woosley \& Lamb 1996; Bildsten 1998).
70:
71: Among the over 100 LMXBs known in the Galaxy, roughly a dozen systems,
72: most of them transient sources (Tanaka \& Shibazaki 1996), have been
73: identified as black hole (BH) candidates. In these BH LMXBs,
74: dynamical measurements give mass estimates for the accreting stars
75: greater than the likely maximum mass $\sim3M_\odot$ of a NS (Shapiro
76: \& Teukolsky 1983, hereafter ST83; Narayan, Garcia \& McClintock 2001;
77: and references therein).
78:
79: No Type I burst has been seen in a BH LMXB, even though, as we show in
80: this {\it Letter}, BH LMXBs ought to produce bursts as efficiently as
81: NS LMXBs if the accreting BH candidates possess surfaces. The lack of
82: bursts thus represents possible evidence for the presence of event
83: horizons. In \S2 of this paper, we examine the stability of nuclear
84: burning on the surface of a compact star. In \S3, we discuss the
85: predictions of the model for accretion onto a $1.4M_\odot$ NS and a
86: $10 M_\odot$ BH candidate with a hypothetical surface. We conclude in
87: \S4 with a discussion.
88:
89:
90: \section{The Model}
91:
92: We consider a compact spherical star of mass $M$ and radius $R$,
93: accreting gas steadily at a rate $\dot\Sigma ~({\rm
94: g\,cm^{-2}\,s^{-1}})$. In the local frame, the gravitational
95: acceleration is $g=GM(1+z)/R^2$, where the redshift $z$ is given by
96: $1+z=(1-R_S/R)^{-1/2}$, and $R_S=2GM/c^2$ is the Schwarzschild radius.
97: We assume that the accreting material has mass fractions $X_0$, $Y_0$
98: and $Z_0=1-X_0-Y_0$, of hydrogen, helium and heavier elements (mostly
99: CNO).
100:
101: We consider a layer of accreted material of surface density
102: $\Sigma_{max}$ sitting on top of a substrate of fully burnt material
103: ($X=Y=0$, $Z=1$). (The particular composition of the substrate is
104: unimportant since we apply the inner boundary condition at its top.)
105: Since the physical thickness of the accreted layer is much less than
106: the radius, we work in plane parallel geometry and take $g$ to be
107: independent of depth. We solve for the density $\rho$, the
108: temperature $T$, the outgoing flux $F$, and the hydrogen, helium and
109: heavy element fractions, $X$, $Y$, $Z=1-X-Y$, as functions of the
110: column density $\Sigma$ ($0\le\Sigma\le\Sigma_{max}$).
111:
112: The evolution equations for the gas in the layer are the standard
113: equations of stellar physics. H- and He-burning give
114: \beq
115: {dX\over dt}=-{\epsilon_H\over E_H^*}, \quad {dY\over
116: dt}={\epsilon_H\over E_H^*} -{\epsilon_{He}\over E_{He}^*},
117: \eeq
118: where for our problem the total time derivative takes the form $d/dt
119: \equiv \partial/\partial t + \dot\Sigma\partial/\partial\Sigma$.
120: Here, $\epsilon_{H,He}$ are the respective nuclear energy generation
121: rates, and $E_{H,He}^*$ are the corresponding energy release per unit
122: mass of H and He burned (P82). For $\epsilon_H$, we include the pp
123: chain and the CNO cycle, including fast-CNO burning, saturated CNO
124: burning, and electron capture reactions, as described in Mathews \&
125: Dietrich (1984) and Bildsten \& Cumming (1998). Since we are not
126: concerned with modelling the bursts themselves, and since our
127: stability criterion does not depend on the detailed treatment of the
128: deep crust, we do not include proton captures onto heavier nuclei;
129: Schatz et al. (1999) illustrate some of the consequences of the
130: $rp$-process burning on accreting neutron stars. For He-burning, we
131: include the triple-$\alpha$ reaction, but not pycnonuclear reactions
132: (e.g., ST83). We do not correct the reaction rates to include
133: screening (e.g. FL87) since we are concerned only with determining
134: whether nuclear burning of H and He can proceed stably under given
135: conditions; stable burning of H and He utilizes almost exclusively the
136: reactions included.
137:
138: Hydrostatic equilibrium gives $\partial P/\partial\Sigma=g$. For the
139: pressure $P$ we use the expressions given in P82 for the gas,
140: radiation and degeneracy pressure, along with a correction when the
141: degenerate electrons become relativistic. Radiative transfer gives
142: $\partial T/\partial\Sigma=(3\kappa F)/(16 \sigma T^3)$, where
143: $\sigma$ is the Stefan-Boltzmann constant and we write the opacity
144: $\kappa$ in the form $1/\kappa = 1/\kappa_{rad} + 1/\kappa_{cond}$. We
145: employ Iben's (1975) fitting functions for the radiative opacity
146: $\kappa_{rad}$, and an analytical formula from Clayton (1968),
147: suitably modified for relativistic electrons, for the conductive
148: opacity $\kappa_{cond}$; the latter expression agrees well with more
149: modern treatments (e.g. Heyl \& Hernquist 2001). Finally, the energy
150: equation gives
151: \beq
152: \rho T{ds\over dt} = \rho(\epsilon_H+\epsilon_{He}) + \rho{\partial
153: F\over \partial\Sigma},
154: \eeq
155: where $s$ is the entropy per unit mass. The above five equations form
156: a closed set.
157:
158: We have four outer boundary conditions at the surface of the star
159: ($\Sigma=0$). Two of these are $(X,Y)=(X_0,Y_0)$. We obtain the
160: third boundary condition by equating the accretion luminosity of the
161: infalling gas, $L_{acc}=4\pi R^2\dot\Sigma c^2z/(1+z)$, to blackbody
162: emission from the surface: $4\pi R^2\sigma T_{out}^4=L_{acc}$. This
163: gives the surface temperature $T_{out}$. Then, using $T_{out}$ and an
164: assumed value of $F_{out}$, we solve for the surface density profile
165: $\rho(\Sigma)$ from the radiative transfer equation, thus obtaining
166: the fourth boundary condition.
167:
168: At the base of the accreted layer we have an inner boundary condition.
169: We assume that the temperature at the top of the substrate,
170: $T_{in}=T(\Sigma_{max})$, is fixed. We examine several values of
171: $T_{in}$ for layers with $\Sigma_{max}=10^9, 10^{10}$ and $10^{11}$~g
172: cm$^{-2}$. Applying the boundary condition at $\Sigma_{max}$ rather
173: than deeper down is an approximation, but the error due to this is not
174: serious. For the high surface densities we consider, the heat
175: transfer is dominated by conduction (Heyl \& Hernquist 2001), so the
176: temperature gradient for $\Sigma>\Sigma_{max}$ is small. Moreover, we
177: examine models for several values of $T_{in}$ which further mitigates
178: any error.
179:
180: The calculations proceed in two stages. First, we solve for the
181: steady state profile of the accretion layer by making the replacement
182: $d/dt \to \dot\Sigma d/d\Sigma$, $\partial/\partial\Sigma \to
183: d/d\Sigma$ in the governing equations. This gives five ordinary
184: differential equations with four outer boundary conditions and one
185: inner boundary condition. The solution gives the profiles of the
186: basic fluid quantities: $X(\Sigma), ~Y(\Sigma), ~P(\Sigma)=g\Sigma,
187: ~\rho(\Sigma), ~T(\Sigma)$.
188:
189: Having calculated the steady state structure of the accretion layer,
190: we next check its stability. Various local stability criteria have
191: been discussed in the literature (e.g., Bildsten 1998), in which one
192: considers the properties of the gas at a single depth. FL87 proposed
193: a global criterion involving an integral over the entire layer. While
194: an improvement, this approach is still unsatisfactory since the
195: authors assumed a constant temperature perturbation throughout the
196: layer.
197:
198: We have carried out a full linear stability analysis of the accretion
199: layer. We start with the steady state solution and assume that it is
200: slightly perturbed, $Q(\Sigma) \to Q(\Sigma) +Q'(\Sigma)\exp(\gamma
201: t)$, where $Q$ corresponds to each of our five variables, and the
202: perturbations $Q'(\Sigma)$ are taken to be small. We linearize the
203: five equations described earlier (three of which include time
204: derivatives), apply the boundary conditions, and solve for the
205: eigenvalue $\gamma$. We obtain a large number of solutions for
206: $\gamma$ (technically, there is an infinite number since we are
207: dealing with continuous functions), many of which have both a real and
208: imaginary part. We consider the accretion layer to be unstable if any
209: eigenvalue $\gamma$ has a real part (growth rate) greater than the
210: characteristic accretion rate $\gamma_{acc}=\dot\Sigma/\Sigma_{max}$.
211:
212: If the steady-state model is unstable according to the above
213: criterion, it cannot accrete matter stably with the particular
214: $\Sigma_{max}$ and $\dot\Sigma$. Whether this instability manifests
215: itself as a Type~I burst depends on how the burning flame, once
216: ignited, envelopes the surface of the star. This is as yet an
217: unsolved problem, although Spitkovsky, Levin \& Ushomirsky (2002)
218: present a possible solution. In the following we assume that the
219: instability will grow rapidly to the nonlinear regime and that the
220: system will indeed exhibit Type I bursts.
221:
222: \section{Results}
223:
224: Figure 1 shows results for solar composition material ($X_0=0.7$,
225: $Y_0=0.27$, $Z_0=0.03$) accreting on two kinds of compact stars.
226:
227: \clearpage
228: \begin{figure*}
229: \plotone{f1.eps}
230: \caption[f1.eps]{Regions of instability, shown by
231: dots, as a function of accretion luminosity and stellar radius. Top
232: left: $1.4M_\odot$ NS with a base temperature $T_{in}=10^{8.5}$~K.
233: Top center: $T_{in}=10^8$K. Top right: $T_{in}=10^{7.5}$ K. Bottom
234: left: $10M_\odot$ BH candidate with a surface, and a base temperature
235: $T_{in}=10^{7.5}$~K. Bottom center: $T_{in}=10^7$ K. Bottom right:
236: $T_{in}=10^{6.5}$ K.}
237: \label{fig2}
238: \end{figure*}
239: \clearpage
240:
241: The top three panels correspond to an accreting NS of mass
242: $1.4M_\odot$. We consider a range of accretion rates, parameterized
243: by the ratio $L_{acc}/L_{Edd}$, where we take the Eddington luminosity
244: to be $L_{Edd}=4\pi GMc/\kappa_{es}$ with $\kappa_{es}=0.4 ~{\rm
245: cm^2\,g^{-1}}$. We also consider a range of radii for the NS:
246: $\log(R/R_S)=0.2-0.6$ ($R=6.5-16$ km). For each choice of
247: $L_{acc}/L_{Edd}$ and $R/R_S$, we try three values of the surface mass
248: density of the accreted layer: $\Sigma_{max} = 10^9, ~10^{10},
249: ~10^{11} ~{\rm g\,cm^{-2}}$. If any of the three cases is unstable,
250: i.e., if it has any eigenvalue with ${\rm Re}(\gamma) >\gamma_{acc}$,
251: then we claim that the system will exhibit instabilities that may grow
252: into Type~I bursts.
253:
254: The results presented correspond to three choices of the temperature
255: at the base of the accreted layer: $T_{in}=10^{8.5}, ~10^8$ and
256: $10^{7.5}$~K from left to right. In this {\it Letter} we are
257: primarily interested in transient LMXBs. Since these sources have
258: very low luminosities in quiescence ($L_X<10^{33} ~{\rm
259: erg\,s^{-1}}$), the core temperatures of the NSs are likely to be
260: $T_{in}\la10^8$ K (Possenti et al. 2001;
261: %Brown, Bildsten \& Rutledge 1998;
262: Brown, Bildsten \& Chang 2002).
263:
264: The calculations shown in Fig. 1 indicate that NSs are unstable to
265: bursts for a wide range of $T_{in}$, but that the width of the
266: instability strip (as a function of $L_{acc}$) is less for higher
267: values of $T_{in}$. The reason for this is clear from the analysis of
268: P82 who showed that when the flux escaping from the stellar core into
269: the accretion layer increases (which happens when $T_{in}$ increases),
270: bursting behavior is restricted to a smaller range of $\dot \Sigma$.
271:
272: In the case of transient NS LMXBs, we expect $T_{in}$ to be $\la10^8$
273: K. Fig. 1 indicates that these sources should be unstable to bursts
274: for accretion luminosities up to $L_{acc} \approx 0.3 L_{Edd}$. The
275: predicted luminosity limit is generally consistent with observations;
276: the only NSs that are known not to burst are bright Z sources with
277: $L_{acc}\to L_{Edd}$ (e.g., Matsuba et al. 1995), and X-ray pulsars.
278: Although the latter are significantly less luminous than Eddington,
279: they accrete effectively at close to the Eddington rate since the
280: accreting matter is channeled onto a small area on the NS surface by
281: strong magnetic fields (see Lamb 2000 for a detailed discussion of
282: this argument).
283:
284: Below $L_{acc}\sim0.3L_{Edd}$, the instability is initially of a mixed
285: form in which a He-burning instability triggers a burst in which H and
286: He both burn explosively. At lower luminosities, nearly all the H is
287: burned steadily and the instability corresponds to a pure He burst.
288: These results are consistent with previous work (e.g., Bildsten 1998).
289: The calculations indicate that bursting behavior cuts off below an
290: accretion luminosity $L_{acc}\sim10^{-1.5}L_{Edd}$. The cutoff is the
291: result of the restriction $\Sigma_{max}\le10^{11} ~{\rm g\,cm^{-2}}$
292: in our models. Systems with luminosities below our cutoff are still
293: unstable to He bursts, but only at extremely high column density
294: $\Sigma_{max}$. We have ignored such bursts since the recurrence time
295: $t_{rec}$ is too long to be of interest for observations of transient
296: X-ray binaries.
297:
298: For $T_{in}=10^{7.5}$ K, there is a second instability strip at low
299: luminosities $L_{acc}\sim10^{-2}L_{Edd}$. This strip corresponds to
300: pure H bursts; the possibility of such bursts was first noted by
301: Fujimoto et al. (1987). An interesting difference between the two
302: instability strips is that the strip on the right generally has
303: complex eigenvalues $\gamma$ for the unstable modes while that on the
304: left has real eigenvalues. As explained above, the gap between the
305: two strips is real if we restrict ourselves to reasonable values of
306: $\Sigma_{max}$ and $t_{rec}$. We are not aware of clear observational
307: evidence for or against the gap, but such evidence could be searched
308: for in future observations.
309:
310: The bottom three panels in Fig. 1 show results for a $10M_\odot$ BH
311: candidate with a surface. The three panels correspond to different
312: choices of the base temperature: $T_{in}=10^{7.5}, ~10^7$ and
313: $10^{6.5}$~K from left to right. The particular choices of $T_{in}$
314: are motivated by the extraordinarily low quiescent luminosities of
315: many transient BH LMXBs ($L_X<10^{31} ~{\rm erg\,s^{-1}}$, Narayan et
316: al. 2001; a few BH LMXBs are brighter than this limit, but even these
317: are not likely to have $T_{in}\ga10^{7.5}$ K). We consider stellar
318: radii in the range $\log(R/R_S)=0.05-0.45$, corresponding to $R$
319: between $(9/8)R_S$ and about $3R_S$, a factor of nearly twenty in the
320: surface gravity. The choice $(9/8)R_S$ corresponds to the smallest
321: radius in general relativity for an object whose density either
322: decreases or remains level with increasing radius (ST83).
323:
324: The calculations indicate that BH candidates with surfaces are not
325: very different from NSs in their bursting behavior. Except for a
326: modest rightward shift of the positions of the instability zones, the
327: results in the lower three panels are quite similar to those for a NS
328: with a similarly low value of $T_{in}$ (upper right panel). As in the
329: case of NSs, bursts are expected for BH candidates even for larger
330: values of $T_{in}$ than we have considered, except that the
331: instability strip becomes narrower when $T_{in}$ exceeds $10^8$ K. We
332: conclude that BH candidates are as prone to the instabilities that
333: lead to Type I bursts as NSs are. The absence of bursts in BH LMXBs
334: is thus highly significant and argues for the lack of surfaces in
335: these systems.
336:
337: Although our analysis is for a spherical, non-rotating $10M_\odot$
338: star, the results are not expected to differ greatly for a rotating
339: object. The effective surface gravity and accretion rate vary by only
340: a factor of 2--3 from pole to equator even for a maximally rotating
341: oblate ellipsoid that is on the verge of the secular triaxial
342: instability; furthermore, the dependence on compactness and equation
343: of state is small (see Gondek-Rosinska \& Gourgoulhon 2002; ST83). In
344: comparison, Fig. 1 shows that bursts are present for a wide range of
345: surface gravity $g$ (factor of 20) and accretion rate $\dot\Sigma$
346: (factor of 100). The rather modest variation of $g$ and $\dot\Sigma$
347: with latitude in a rotating star is thus not likely to have an effect.
348:
349: Because our model focuses only on the most important physical effects,
350: and neglects many details, the exact positions of the instability
351: strips in Fig. 1 may be uncertain at the level of say a factor of two
352: in accretion luminosity. We believe, however, that the overall
353: pattern of instability we have computed is fairly robust. Also, the
354: calculations we present here do not directly predict what kind of
355: bursts are produced. We leave this for a more detailed paper, but
356: briefly, we find that for NSs with $L_{acc}\sim0.1L_{Edd}$, bursts
357: have durations of a few seconds and $t_{rec}\sim10$ hours. For lower
358: $L_{acc}$, as is well-understood, bursts have longer $t_{rec}$ and larger
359: fluences. These results agree qualitatively with observations.
360:
361:
362: \section{Discussion}
363:
364: It is clear from the theory of bursts (Bildsten 1998, and references
365: therein) that bursting behavior is largely determined by the surface
366: gravity $g$, the mass accretion rate, and the composition of the
367: accreting material. Since these parameters are similar in NS and BH
368: systems, the bursting behavior of the two should be similar. We have
369: quantified this argument with a global linear stability analysis which
370: represents a technical advance over previous calculations.
371:
372: The results presented in Fig. 1 show that if BH candidates had
373: surfaces they ought to experience thermonuclear instabilities as
374: commonly as NSs do; by inference, they should have frequent Type~I
375: bursts. However, no BH candidate has exhibited Type I bursts. The
376: most obvious explanation is that NSs have surfaces and BH candidates
377: have event horizons (Menou 2001). If there is no surface, material
378: cannot accumulate, and therefore cannot become hot or dense enough to
379: trigger a thermonuclear explosion.
380:
381: Before we can claim that this ``proves'' the reality of the event
382: horizon, more work is needed. First, we need to show that the model
383: is able to reproduce the more detailed features of Type I bursts as
384: observed in NS LMXBs. The statistics of burst durations and
385: recurrence times (e.g., van Paradijs, Penninx \& Lewin 1988) and the
386: occurrence of oscillatory behavior in some systems (Revnivtsev et
387: al. 2000) ought to appear naturally in a realistic model. Also, any
388: NS systems that burst when they should not by our calculations, or do
389: not burst when they should according to the model, need to be
390: explained. Second, the role of the inner boundary condition needs to
391: be explored in detail. In Fig. 1, we see that different choices of
392: the base temperature for a BH candidate give similar results. We have
393: tried other boundary conditions, and also tried changing the
394: composition of the accreting gas. In all cases we find that the
395: accumulating layer is unstable to bursts for a wide range of
396: luminosity, both in $1.4M_\odot$ NSs and $10M_\odot$ compact stars
397: with surfaces. Finally, the difficult issue of flame propagation over
398: the surface of the star once the instability has been triggered needs
399: to be addressed (e.g., Lamb 2000, Spitkovsky et al. 2002); the effect
400: of rotation, for instance, is presently unclear.
401:
402: We should caution that the analysis presented here assumes that the
403: accumulating gas on the surface of a BH candidate behaves like normal
404: matter, with nucleons and electrons; this is the case for standard
405: models of strange stars and Q-stars (Glendenning 1997). Obviously,
406: the argument becomes invalid if the properties of the gas change
407: drastically, e.g., if the nuclei disappeared and were replaced by
408: quarks. Whether such extreme changes are plausible remains to be
409: seen. The density and pressure at the base of the bursting layer do
410: not go above ${\rm few}\times10^8 {\rm g\,cm^{-3}}$ and $10^{26} ~{\rm
411: erg\,cm^{-3}}$ even in the most extreme cases we have considered. It
412: is hard to imagine exotic physics being important under these
413: conditions (Glendenning 1997).
414:
415: On the observational front, we should check whether some NSs that lie
416: within the unstable regions of Fig. 1 are stable to bursts. Any
417: obvious large-scale disagreement between the observed burst behavior
418: of NS LMXBs and the results presented here would indicate that the
419: model is missing important physics. In the case of BH binaries, we
420: should use observational data to derive quantitative limits on
421: bursting activity. The transient BH LMXBs are particularly important
422: for such work since they vary over a wide range of $L_{acc}$ during
423: their accretion outbursts (Tanaka \& Shibazaki 1996). If these
424: sources have surfaces it is virtually impossible to arrange matters
425: such that the objects have no bursts at all over the entire range of
426: $L_{acc}$. A firm demonstration that BH transients do not have Type I
427: bursts would thus be a strong argument for the presence of event
428: horizons in these systems.
429:
430:
431: \acknowledgements
432:
433: We thank Andrew Cumming, Alex Ene, Kristen Menou, Bohdan Paczy\'nski and
434: Greg Ushomirsky for useful discussions, and two referees for useful
435: comments. RN was supported in part by the W. M. Keck Foundation as a
436: Keck Visiting Professor. RN's research was supported by NSF grant
437: AST-9820686 and NASA grant NAG5-10780. JSH was supported by the
438: Chandra Postdoctoral Fellowship Award \# PF0-10015 issued by the
439: Chandra X-ray Observatory Center, which is operated by the Smithsonian
440: Astrophysical Observatory for and on behalf of NASA under contract
441: NAS8-39073.
442:
443: \clearpage
444:
445: \begin{thebibliography}{dummy}
446: %
447: \bibitem[Bildsten, L.(1998)]{B98} Bildsten, L. 1998, in The Many Faces
448: of Neutron Stars, eds R. Buccheri, J. van Paradijs, \& M. A. Alpar
449: (Dordrecht: Kluwer), 419
450: %
451: \bibitem[Bildsten \& Cumming (1998)]{BC98}
452: Bildsten, L. \& Cumming, A. 1998, \apj, 506, 842
453: %
454: \bibitem[Brown, Bildsten \& Chang(2002)]{BBC}
455: Brown, E. F., Bildsten, L., \& Chang, P. 2002, \apj, in press
456: %
457: %\bibitem[Brown, Bildsten \& Rutledge(1998)]{BBR} Brown, E. F.,
458: %Bildsten, L., \& Rutledge, R. E. 1998, \apj, 504, L95
459: %
460: \bibitem[Clayton (1968)]{Clay68} Clayton, D. D. 1968, Principles of
461: Stellar Evolution and Nucleosynthesis (New York: McGraw-Hill)
462: %
463: %\bibitem[Fujimoto, Hanawa \& Miyaji(1981)]{FHM} Fujimoto, M. Y.,
464: %Hanawa, T., \& Miyaji, S. 1981, \apj, 246, 267
465: %
466: \bibitem[Fujimoto et al.(1984)]{FHIR}
467: Fujimoto, M. Y., Hanawa, T., Iben, I., \& Richardson, M. B. 1984,
468: \apj, 278, 813
469: %
470: \bibitem[Fujimoto et al.(1987)]{FHIR87}
471: Fujimoto, M. Y., Hanawa, T., Iben, I., \& Richardson, M. B. 1987,
472: \apj, 315, 198
473: %
474: \bibitem[Fushiki \& Lamb(1987)]{FL87}
475: Fushiki, I., \& Lamb, D. Q. 1987, \apj, 323, L55 (FL87)
476: %
477: \bibitem[Glendenning(1997)]{1997csnp.conf.....G}
478: Glendenning, N.~K.\ 1997, Compact Stars, Nuclear Physics, Particle Physics,
479: and General Relativity (Berlin: Springer)
480: %
481: \bibitem[Gondek-Rosinska & Gourgoulhon]{Jacobi02}
482: Gondek-Rosinska, D. \& Gourgoulhon E. 2002, \prd, submitted (gr-qc/0205102)
483: %
484: \bibitem[Grindlay et al.(1976)]{G76}
485: Grindlay, J., et al. 1976, \apj, 205, L127
486: %
487: \bibitem[Hansen \& van Horn(1975)]{HvH}
488: Hansen, C. J., \& van Horn, H. M. 1975, \apj, 195, 735
489: %
490: \bibitem[Heyl \& Hernquist (2001)]{Heyl01}
491: Heyl, J. S., \& Hernquist, L. 2001, \mnras, 324, 292
492: %
493: %\bibitem[Hoffman et al. (1978)]{HML78}
494: %Hoffman, J. A., Marshall, H. L., \& Lewin, W. H. G. 1978, Nature, 271, 630
495: %
496: \bibitem[Iben(1975)]{1975ApJ...196..525I} Iben, I.\ 1975, \apj, 196, 525
497: %
498: \bibitem[Joss(1977)]{J77}
499: Joss, P. C. 1977, Nature, 270, 310
500: %
501: \bibitem[Lamb(2000)]{2000ApJS..127..395L} Lamb, D.~Q.\ 2000, \apjs, 127,
502: 395.
503: %
504: \bibitem[Lewin, van Paradijs \& Taam(1993)]{LvPT}
505: Lewin, W. H. G., van Paradijs, J., \& Taam, R.E. 1993, Sp. Sci. Rev., 62, 223
506: %
507: \bibitem[Mathews \& Dietrich (1984)]{MD84}
508: Mathews, G. J. \& Dietrich, F. S. 1984, \apj, 287, 969
509: %
510: \bibitem[Matsuba et al.(1995)]{M95}
511: Matsuba, E., et al. 1995, \pasj, 47, 575
512: %
513: \bibitem[Menou(2001)]{M01}
514: Menou, K. 2001, in Proc. 2nd KIA Astrophysics Workshop (astro-ph/0111469)
515: %
516: \bibitem[Narayan, Garcia \& McClintock 2001]{NGM}
517: Narayan, R., Garcia, M. R., \& McClintock, J. E. 2001, in Proc. IX Marcel
518: Grossmann Meeting, eds V. Gurzadyan, R. Jantzen, \& Ruffini (Singapore: World
519: Scientific) (astro-ph/0107387)
520: %
521: \bibitem[Paczy\'nski(1982)]{P82}
522: Paczy\'nski, B. 1982, \apj, 264, 282 (P82)
523: %
524: \bibitem[Possenti, Colpi, Page, \& Geppert(2001)]{2001ebms.conf..483P}
525: Possenti, A., Colpi, M., Page, D., \& Geppert, U.\ 2001, ASP Conf.~Ser.~229:
526: Evolution of Binary and Multiple Star Systems, 483
527: %
528: \bibitem[Revnivtsev et al.(2001)]{RCGS}
529: Revnivtsev, M., Churazov, E., Gilfanov, M., \& Sunyaev, R. 2001, A\&A, 372, 138
530: %
531: \bibitem[Schatz, Bildsten, Cumming, \& Wiescher(1999)]{1999ApJ...524.1014S}
532: Schatz, H., Bildsten, L., Cumming, A., \& Wiescher, M.\ 1999, \apj, 524, 1014
533: %
534: \bibitem[Shapiro \& Teukolsky (1983)]{Shap83}
535: Shapiro, S. L. \& Teukolsky, S. A. 1983, Black Holes, White Dwarfs, and Neutron
536: Stars (New York: Wiley-Interscience) (ST83)
537: %
538: \bibitem[Spitkovsky, Levin, \& Ushomirsky(2002)]{2002ApJ...566.1018S}
539: Spitkovsky, A., Levin, Y., \& Ushomirsky, G.\ 2002, \apj, 566, 1018
540: %
541: \bibitem[]{546}
542: Strohmayer, T. E., Swank, J. H. \& Zhang, W. 1998, Nucl. Phys. B
543: (Proc. Suppl.), 69, 129.
544: %
545: \bibitem[Taam \& Picklum(1978)]{TP78}
546: Taam, R. E., \& Picklum, R. E. 1978, \apj, 224, 210
547: %
548: \bibitem[Taam, Woosley \& Lamb(1996)]{TWL}
549: Taam, R. E., Woosley, S. E., \& Lamb, D. Q. 1996, \apj, 459, 271
550: %
551: \bibitem[Tanaka \& Shibazaki 1996]{TS96}
552: Tanaka, Y., \& Shibazaki, N. 1996, \araa, 34, 607
553: %
554: \bibitem[van Paradijs, Penninx \& Lewin(1988)]{vPPL}
555: van Paradijs, J., Penninx, W., \& Lewin, W. H. G. 1988, \mnras, 233, 437
556: %
557: \bibitem[Woosley \& Taam(1976)]{WT76}
558: Woosley, S. E., \& Taam, R. E. 1976, Nature, 263, 101
559: %
560: \end{thebibliography}
561:
562:
563: \end{document}
564: \end
565:
566: