1: %\documentclass[12pt,preprint]{aastex}
2: %\documentclass{aastex}
3: \documentclass{emulateapj}
4: %\usepackage{natbib}
5: %\bibliographystyle{apj}
6:
7: %\usepackage{emulateapj}
8: %\usepackage{psfig}
9:
10: \newcommand{\be}{ \begin{eqnarray} }
11: \newcommand{\ee}{ \end{eqnarray} }
12: \newcommand{\bd}{ \begin{displaymath} }
13: \newcommand{\ed}{ \end{displaymath} }
14: \newcommand{\msun}{ M_{\odot}}
15: \newcommand{\mdot}{\dot{M}}
16:
17: \begin{document}
18: \title{Angular Momentum Transport in Accretion Disks and its Implications for
19: Spin Estimates in Black Hole Binaries}
20: \author{Chris Done\altaffilmark{1} and Shane W. Davis\altaffilmark{2,3}}
21: \altaffiltext{1}{Department of Physics, University of Durham, South Road,
22: Durham, DH1 3LE, UK}
23: \altaffiltext{2}{School of Natural Sciences, Institute for Advanced Study,
24: Einstein Drive, Princeton, NJ 08540}
25: \altaffiltext{3}{Chandra Fellow}
26:
27: \begin{abstract}
28:
29: The accretion flow in the disk dominated state of black hole binaries
30: has peak temperature and luminosity which vary together in such a way
31: as to indicate an approximately constant emitting area. The
32: association of this with the last stable orbit gives one of the few
33: ways to estimate spin when the mass of the black hole is
34: known. However, deriving this radius requires knowledge of how the
35: disk spectrum is modified by radiative transfer through the vertical
36: structure of the disk, as well as special and general relativistic
37: effects on the propagation of this radiation. Here we investigate the
38: extent to which differences in vertical structure change the derived
39: disk spectra by calculating these for a range of different stress
40: prescriptions. We find that at a given mass accretion rate the spectra
41: are almost identical for accretion rates of $L/L_{Edd} \lesssim
42: 0.1$. The spectra are remarkably similar even up to the highest
43: luminosities considered ($L/L_{Edd}\sim 0.6$) as long as the stresses
44: do not dissipate more than about 10 per cent of the gravitational
45: energy above the effective photosphere. This is exceeded only by
46: classic alpha disks with $\alpha\gtrsim 0.1$, but these models give
47: spectral variation which is incompatible with existing
48: data. Therefore, we conclude that disk spectral modelling can place
49: interesting constraints on angular momentum transport, but still
50: provide a robust estimate of the spin of the black hole.
51:
52: \end{abstract}
53:
54: \keywords{accretion, accretion disks --- black hole physics --- X-rays:binaries}
55:
56: \section{Introduction}
57: \label{intro}
58:
59: Black hole spin is difficult to measure as it gives a strong signature
60: only very close to the event horizon. Currently the only way to probe
61: this is via luminous accretion flows, which light up the regions of
62: dramatically curved spacetime close to the event horizon. However, to
63: use these as a diagnostic of spin and/or to test of Einsteins gravity
64: in the strong-field limit requires knowledge of the velocity and
65: emission structure of this material. Currently, this is best
66: understood for an geometrically thin, optically thick accretion disk,
67: where the material rotates in approximately Keplerian orbits (except
68: at the inner boundary) and the gravitational energy thermalizes to
69: produce a quasi--blackbody spectrum at each radius in the disk
70: \citep{sas73}.
71:
72: Black hole binaries (BHB) can indeed show spectra which look like this
73: simple `sum of blackbodies', though these are generally accompanied by
74: a nonthermal tail to higher energies whose origin is not well understood.
75: Nonetheless, where this tail carries only a minor fraction of the
76: bolometric power, then the disk model gives a fairly good description
77: of the spectral shape. More compellingly, a sequence of such disk
78: dominated spectra from the same object at different mass accretion
79: rates shows that the bolometric disk luminosity (spanning factors of
80: 10--50) relates to the maximum temperature as $L_{\rm bol} \propto
81: T_{\rm max}^4$ \citep{kme01}, or equivalently, that the disk has an
82: approximately constant inner radius \citep{emh91,ebi93}. This {\em
83: observation} is exactly the behavior predicted by General Relativity,
84: as the last stable orbit forms a fixed size scale for a geometrically
85: thin disk \citep[see e.g. the review by][ hereafter DGK07]{dgk07}.
86:
87: However, these disk dominated spectra {\em are} surprising in the
88: context of the Shakura--Sunyaev disk models as they are seen at high
89: luminosities, typically $0.1-0.5$ $L_{\rm Edd}$ (e.g. DGK07). The
90: classic prescription in which the viscous stress is directly
91: proportional to total (gas plus radiation, $P_{\rm tot}=P_{\rm
92: gas}+P_{\rm rad}$) pressure, is thermally and viscously unstable where
93: radiation pressure dominates, i.e. for luminosities above $\sim 0.05
94: L_{\rm Edd}$ for BHB \citep[e.g.][]{lig74,sas76}. Above this
95: luminosity, the inner regions of alpha disks exhibit limit cycles, in
96: direct conflict with the observed {\em stability} of the disk spectra
97: \citep[e.g.][]{now95}. Yet the alpha prescription does seem to be able
98: to reproduce the long timescale outburst/quiescence behavior of the
99: outer, gas pressure dominated region in the disk instability model
100: \citep[see e.g. the review by][]{las01}.
101:
102: Thus it seems likely that the stress is approximately proportional to
103: gas pressure where gas pressure dominates, but that it scales less
104: sensitively with temperature than predicted by radiation pressure when
105: radiation pressure dominates. Ultimately, the question of the stress
106: scaling will be answered by numerical simulations of the
107: Magneto-Rotational Instability \citep[MRI,][]{bah91} which forms the
108: physical basis for ``effective viscosity''. However, this is beyond
109: the scope of current codes as it requires coupled
110: radiative--magneto--hydrodynamic simulations in full general
111: relativity, covering a large radial extent but simultaneously
112: resolving the small scale height of the thin disk!. Until these become
113: available, {\em ad hoc} stress scalings are the only way to model the
114: structure of a radiative accretion disk.
115:
116: The choice of stress prescription can affect the spectrum as well as
117: the stability of the disk as it determines the surface density, and
118: therefore, the overall optical depth of the disk. This emission
119: thermalizes to a true blackbody only if the disk is effectively
120: optically thick to {\em absorption} at all frequencies. Free-free
121: (continuum) and bound--free (photo-electric edge) absorption opacity
122: both drop as a function of frequency while electron scattering is
123: constant, so the highest energy photons from each radii are less
124: likely to thermalize. The spectrum then becomes a modified blackbody
125: whose frequency dependence can, in general, differ significantly from
126: Planckian. However, the effects of Compton scattering tend to yield a
127: Wien-like tail and the hottest parts of the disk are generally well
128: modelled by a color-corrected (or diluted) blackbody, with effective
129: temperature which is a factor $f_{\rm col}$ (termed a
130: color-temperature correction) higher than for complete
131: thermalization. The full-disk spectrum is then a sum of modified black
132: bodies, but this can likewise be approximately described by a
133: color-temperature corrected blackbody disk spectrum
134: \citep{sat95}. This color-temperature correction depends on the
135: vertical temperature and density structure of the disk, both of which
136: can only be fully determined once the stress is specified.
137:
138: Since the observations show to zeroth order that $L\propto T^4$, the
139: color-temperature correction must stay approximately constant (the
140: alternative, that it changes in just such a way as to mask any change
141: in innermost extent of the disk, violates Occam's razor). This used to
142: be controversial, with different models of alpha disks (with only H
143: and He) giving different results. \cite{sat95} found that $f_{\rm
144: col}$ remains relatively constant over a wide range in luminosity,
145: while \cite{mfr00} had constant $f_{\rm col}$ only at $>0.1L_{\rm
146: Edd}$, {\em increasing} below \cite[ hereafter GD04]{gad04}. Since it
147: is physically rather unlikely that absorption becomes less effective
148: at lower temperatures, the \cite{mfr00} result seems rather to be an
149: artifact of the assumed constant vertical density structure. However,
150: both these calculations are now superseded by the models of
151: \cite{dav05}, which includes fully non-LTE calculations for all
152: abundant elements together with the self--consistent radiative
153: transfer and disk structure. These indeed show that $f_{\rm col}$ is
154: approximately constant for an alpha disk, but only below temperatures
155: of $\sim 1$~keV. Above this, the disk becomes so ionized that even
156: photo-electric absorption opacity becomes negligible, so $f_{\rm col}$
157: increases markedly \citep{dav05,ddb06}. A constant $f_{\rm col}$ can
158: be recovered even at these high temperatures by simply decreasing
159: $\alpha$ to $\sim 0.01$, but this is inconsistent with the $\alpha\sim
160: 0.1$ required in order to match the rapid rise to outburst seen in
161: transient systems \citep[e.g.][]{kpl07}, and still gives a disk which
162: should undergo limit cycle oscillations when radiation pressure
163: becomes dominant.
164:
165: Thus it seems that both spectra and stability of the observed disk
166: spectra in BHBs are inconsistent with an alpha stress prescription.
167: Here we use the methods of \cite{dav05} and \cite{dah06} to examine
168: two alternative models, where the stress is directly proportional to
169: gas pressure alone, hereafter called the beta disk, or to the
170: geometric mean of the gas and total pressure, hereafter called the
171: mean disk, \citep{sac81,sar84,tal84,lan89,mer03}. When we fix
172: $\alpha=0.1$ to agree with requirements of the disk instability model,
173: we find that either of these give a better match to the spectral
174: constraints than an alpha prescription as the resulting disk is denser
175: and more optically thick. Thus these alternative stress scalings give
176: disks which are able to maintain a relatively constant $f_{\rm col}$
177: even at high temperatures, providing a better match to the observed
178: spectra as well as the stability properties.
179:
180: While this gives important insight into the approximate form of the
181: MRI stresses, we show that {\em all} stress prescriptions considered
182: here (alpha, beta, and mean) give the {\em same} $f_{\rm col}$ below
183: $\sim 1$~keV, suggesting that disk spectra can be used to give a
184: robust estimate of the inner radius of the disk, hence observationally
185: constraining the black hole spin.
186:
187: \begin{figure*}
188: \begin{center}
189: \hbox{
190: \includegraphics[width=0.5\textwidth]{f1a.eps}
191: \includegraphics[width=0.5\textwidth]{f1b.eps}
192: %\psfig{figure=f1a.eps,width=0.5\textwidth,angle=0}
193: %\psfig{figure=f1b.eps,width=0.5\textwidth,angle=0}
194: }
195: \end{center}
196: \caption{Luminosity--Temperature relations derived from fits of
197: {\tt bhspec} with {\tt diskbb} as described in \S\ref{fcol_0.5}.
198: The left panel corresponds to fits
199: using a CCD ({\it Swift} XRT) response while the right panel utilizes a
200: proportional counter ({\it RXTE} PCA) response. We consider four different
201: stress prescriptions: alpha disks with $\alpha=0.01$ (black) and
202: $\alpha=0.1$ (red), beta disks with $\alpha=0.1$ (blue), and mean
203: disks with $\alpha=0.1$ (green). These all overlap apart from the
204: alpha disks with $\alpha=0.1$ at high luminosity, showing how robust
205: the spectra are to different stress prescriptions as long as the disk
206: remains optically thick. The solid black lines show
207: the $L-T$ relation expected from black hole with $a_*=0.5$
208: for a color-temperature correction of
209: 1.6, 1.8, 2.0 and 2.2 from top to bottom, respectively.
210: \label{f:lt} }
211: \end{figure*}
212:
213: \section{Alternative stress prescriptions}
214: \label{stress}
215:
216: The magnetic nature of the viscosity has long been recognized
217: \citep{sas73,eal75,ich77,grv79}, motivating alternative stress
218: prescriptions \citep[e.g.][]{sac81}. The two most popular are where
219: stress is directly proportional to gas pressure alone \citep[beta
220: disks,][]{sar84} or to the geometric mean of the gas and total
221: pressure, $\sqrt{P_{\rm gas}P_{\rm tot}}$ \citep[mean
222: disks,][]{tal84,mer03}. Beta disks are stable at all luminosities
223: below the Eddington limit (above which the equations break down as
224: winds and/or advection become important), while the mean disks show
225: limit cycles above $\sim 0.3 L_{\rm Edd}$ for stellar remnant black
226: holes \citep[e.g.][]{hkm91,man06}. The mean disk is thus inconsistent
227: with the observation that stable disk spectra exist up to at least
228: $\sim 0.7L_{\rm Edd}$ (DKG07), but the onset of the instability is
229: very sensitive to the exact scaling of the stress. For the more
230: general prescription which is proportional to $P_{\rm gas}^\mu P_{\rm
231: tot}^{1-\mu}$ then the observed stability limit requires $\mu\sim
232: 0.56$ \citep{kfm98}, which is negligibly different to the mean disk
233: ($\mu=0.5$) in terms of steady state disk structure.
234:
235: We use the {\tt bhspec} code of \cite{dah06} to calculate the
236: self-consistent vertical structure and resulting spectra for three
237: stress prescriptions: alpha, beta, and mean disks. We consider a
238: geometrically thin disk and break it up into a series of
239: logarithmically spaced annuli. We then calculate the one-dimensional
240: vertical structure and radiative transfer in each annulus using the
241: TLUSTY stellar atmosphere code \citep{hal95}. The total spectrum
242: is then integrated over the disk surface, accounting for the effects
243: of space-time curvature on the photon geodesics. Three parameters
244: determine the structure of each individual annulus: the radiative flux
245: at the disk surface; the surface density, $\Sigma$; and a parameter
246: which determines the strength of the tidal gravity.
247:
248: Since mechanical
249: energy dissipates directly within the disk ``atmosphere'', the
250: radiative flux $F$ varies with height $z$ in the disk, requiring an
251: additional assumption to uniquely determine the structure. Following
252: \cite{sas73}, we assume that the dissipation $d F/dz$ is locally
253: proportional to the density $\rho$. Using the definition of column
254: mass, $dm=-\rho dz$, this is equivalent to
255: \be
256: \frac{d F}{dm}=-\frac{2 \sigma T_{\rm eff}^4}{\Sigma},
257: \label{eq:dis}
258: \ee
259: where $T_{\rm eff}$ is the effective temperature of the annulus. Note
260: that the right hand side terms, and therefore $dF/dm$ are independent
261: of height with this assumption. We cannot use the stress prescription
262: to specify a local viscous-like dissipation as this is proportional to
263: some combination of radiation and gas pressure. These pressures are
264: largest at the disk midplane due to hydrostatic equilibrium, so the
265: dissipation is also largest there, and the resulting temperatures are
266: so high as to lead to the disk becoming Rayleigh-Taylor unstable. This
267: would lead to convection, presumably driving the disk structure
268: towards the marginally stable condition where the dissipation is
269: proportional to density. Time and horizontal averages of shearing box
270: simulations \citep{tur04,hks06} indicate that a somewhat greater
271: fraction of the dissipation occurs at low column near the surface.
272:
273: We use a vertically integrated disk model to determine the above
274: parameters for each annulus. The tidal gravity is specified by the
275: choice of spacetime and radius, while the radiative flux at the
276: surface is independent of stress prescription for a thin disk
277: \citep[see e.g.][]{sas73}. Therefore, we only require a stress
278: prescription to determine the surface density in this model. (Note,
279: however, that the theoretical arguments which motivate the stress
280: prescription may also require modifications to the dissipation
281: profile, a point we will discuss further in \S\ref{discus}.) The
282: surface density may be determined by enforcing angular momentum
283: conservation
284: \be
285: \int \tau_{r\phi} dz =
286: \frac{\mdot \Omega}{2 \pi}\frac{D(r)}{A(r)},
287: \label{eq:coa}
288: \ee
289: where $\tau_{r\phi}$ is the accretion stress, $\mdot$ is the accretion
290: rate, and $\Omega$ is the Keplerian frequency. The functions $A(r)$
291: and $D(r)$ (both $\to 1$ for radii much larger than the last stable
292: orbit) defined in \cite{rah95} incorporate the no--torque (stress
293: free) inner boundary condition and parameterize the deviation of
294: general relativistic disk structure from the Newtonian limit.
295:
296: \begin{figure*}
297: \begin{center}
298: \hbox{
299: \includegraphics[width=0.5\textwidth]{f2a.eps}
300: \includegraphics[width=0.5\textwidth]{f2b.eps}
301: %\psfig{figure=f2a.eps,width=0.5\textwidth,angle=0}
302: %\psfig{figure=f2b.eps,width=0.5\textwidth,angle=0}
303: }
304: \end{center}
305: \caption{The upper, middle and lower panels show the effective optical
306: depth to the disk midplane, total surface density and effective
307: temperature as a function of radius, respectively. We compare alpha
308: disks with $\alpha=0.01$ (black, dot-dashed) and $\alpha=0.1$ (red,
309: solid), beta disks
310: with $\alpha=0.1$ (blue, dotted), and mean disks with $\alpha=0.1$
311: (green, dashed). $T_{\rm eff}$ is independent of stress prescription.
312: (a) shows these disk parameters for $\log l=-1$ while (b) has $\log
313: l=-0.2$.\label{f:l-1}}
314: \end{figure*}
315:
316: The integral in eq. (\ref{eq:coa}) presents a computational difficulty
317: since $\tau_{r\phi}$ depends on local variable such as temperature and
318: density for the stress prescriptions considered here. Therefore, the
319: exact computation requires the full vertical structure which, in turn,
320: requires a choice of $\Sigma$. Therefore, an exact solution can only
321: be computed via iteration provided an initial guess for $\Sigma$
322: \citep[see e.g.][]{dav05}. Given the approximations inherent in our
323: stress prescriptions, we instead solve for $\Sigma$ using moderately
324: crude vertical averages to replace $\int\tau_{r\phi} dz$ with $2
325: H\langle\tau_{r\phi}\rangle$. Here $H$ is an approximate vertical
326: scale height for the disk determined via the equation of hydrostatic
327: equilibrium
328: \be
329: \frac{\partial P_{\rm tot}}{\partial z}=\rho \Omega^2 z \frac{C(r)}{B(r)},
330: \label{eq:he}
331: \ee
332: where $B(r)$ and $C(r)$ are relativistic factors (again both $\to 1$
333: for radii much larger than the last stable orbit) defined in
334: \cite{rah95}. This is solved by substituting $z \rightarrow H$ and
335: $\partial/\partial z \rightarrow H^{-1}$. We then specify $\langle
336: P_{\rm rad} \rangle$ and $\langle P_{\rm gas} \rangle$ in terms of
337: $\Sigma$, yielding two equations (\ref{eq:coa}, \ref{eq:he}) for two
338: unknowns ($\Sigma$, $H$). The resulting $\Sigma$ evolves smoothly
339: between the gas and radiation pressure dominated limits. Previous
340: work \citep{dav05} has shown that the resulting estimates for $\Sigma$
341: typically agree with a more precise iterative method to $\lesssim
342: 30\%$.
343:
344: Our results also agree with the gas and radiation pressure dominated limits
345: which have been worked out in detail by previous authors
346: \citep[e.g.][]{sas73,sar84,lan89,mer03}. We give the asymptotic
347: behavior at large radii
348: (away from the inner boundary condition) below as this is useful in
349: understanding the subsequent plots. For the gas pressure
350: dominated limit all three stress prescriptions give identical
351: results where
352: \be
353: \Sigma_{\rm gas} \propto m^{1/5} l^{3/5} r^{-3/5} \alpha^{-4/5}
354: \label{eq:siggas}
355: \ee
356: where $l\equiv L/L_{\rm Edd}$ is the Eddington ratio, $r \equiv R/R_g$
357: is radius in unit of the gravitational radius, and $m \equiv M/\msun$
358: is the black hole mass.
359: For the radiation pressure dominated limit the surface density is sensitive to the
360: stress prescription, and for our choices gives
361: \be
362: \Sigma_{\rm rad, \alpha} & \propto & l^{-1} r^{3/2} \alpha^{-1}, \nonumber \\
363: \Sigma_{\rm rad, \beta} & \propto & m^{1/5} l^{3/5} r^{-3/5} \alpha^{-4/5}, \nonumber \\
364: \Sigma_{\rm rad, mean} & \propto & m^{1/9} l^{-1/9} r^{1/3} \alpha^{-8/9}.
365: \label{eq:sigrad}
366: \ee
367:
368: Since the BHBs considered here cover only a small range in mass, the
369: weak dependence of $\Sigma$ on $m$ can be ignored. In what follows,
370: the Eddington ratio dependence will be most important. For the alpha
371: and mean disks $\Sigma$ decreases as $l$ increases, but with a much
372: weaker dependence for the mean disk. In contrast, $\Sigma$ increases
373: with $l$ in the beta disk model. The susceptibility of the stress
374: prescriptions to viscous instability follows directly from these
375: relations. The condition for stability is that $d \Sigma/d \mdot > 0$
376: \citep{lae74}. Since $\mdot \propto l$, we see that only the beta
377: disk is truly stable in the limit where radiation pressure dominates
378: entirely, though the mean disk is also very close to the stability
379: condition.
380:
381: \begin{figure*}
382: \begin{center}
383: \hbox{
384: \includegraphics[width=0.5\textwidth]{f3a.eps}
385: \includegraphics[width=0.5\textwidth]{f3b.eps}
386: %\psfig{figure=f3a.eps,width=0.5\textwidth,angle=0}
387: %\psfig{figure=f3b.eps,width=0.5\textwidth,angle=0}
388: }
389: \end{center}
390: \caption{Color-temperature corrections from fits of {\tt bhspec} with
391: {\tt kerrbb} as described in \S\ref{fcol_0.5}. The left panel
392: corresponds to fits using a CCD ({\it Swift} XRT) response while the
393: right panel is for a proportional counter ({\it RXTE} PCA) response.
394: We consider four different stress prescriptions: alpha disks with
395: $\alpha=0.01$ (black) and $\alpha=0.1$ (red), beta disks with
396: $\alpha=0.1$ (blue), and mean disks with $\alpha=0.1$
397: (green).\label{f:fcol1}. Comparison with Fig 1 shows that the results
398: for the color-temperature correction are not the same as for {\tt
399: diskbb}, where the CCD bandpass gives an almost constant
400: $f_{col}=1.65-1.75$ as opposed to the change from 1.6-1.9 seen here.
401: The proportional counter likewise has $f_{col}=1.7-2.0$ with
402: {\tt diskbb} compared to $1.4-1.8$ here.}
403: \end{figure*}
404:
405: \section{Luminosity-Temperature plots and the color-temperature
406: correction}
407: \label{fcol_0.5}
408:
409: We now use this {\tt bhspec} code \citep[ see also \S \ref{stress}]{dah06}
410: to investigate the
411: spectra produced by these different stress prescriptions for a black
412: hole of 10 $\msun$, inclined at 60$^\circ$, with spin of $a_*=0.5$. We
413: use the {\sc XSPEC} spectral fitting package to convolve the models with
414: the detector response from current X-ray instruments in order
415: to simulate what would be observed. We do this for
416: accretion rates equally spaced in $\log l$ from
417: -1.4 to -0.2 through both the {\it RXTE} PCA (3--20~keV) and {\it
418: Swift} XRT (0.1--10~keV) responses. These two instruments
419: are broadly characteristic
420: of any proportional counter (e.g. GINGA, ASTROSAT) and CCD
421: (e.g. Chandra, XMM-Newton, Suzaku XIS) responses, respectively.
422: We fix the source distance at 8.5~kpc (Galactic center), and
423: assume a Galactic absorbing column of $N_H=1.2\times 10^{21}$
424: cm$^{-2}$ (approximately the lowest possible at this distance).
425: We note that for $\log l =-0.6$, this gives a very good
426: estimate of the disk dominated spectrum of XTE J1817-330 at the peak
427: of its outburst \citep[spectrum 001:][ hereafter GDP08]{ryk07,gdp08}.
428:
429: We also include a small Comptonized tail, with bolometric luminosity
430: of $\sim 5$\% of that of the disk, with photon index of $\Gamma=2.2$
431: and electron temperature fixed at 100~keV (i.e. outside the bandpass
432: of either instrument). The seed photons are assumed to be a blackbody
433: at the maximum inner disk temperature, set by noting that disk fits to
434: the peak spectrum of XTE J1817-330 give $kT_{\rm disk}\sim 0.9$~keV
435: (GDP08). We use this as our fiducial seed temperature for $\log
436: l=-0.6$, and then change it by a factor $(10^{0.1})^{1/4}=1.059$ for
437: each 0.1 dex change in $\log l$. We also add 1\% systematic
438: uncertainties to our simulated PCA spectra as is usual with real data.
439:
440: These simulated spectra are then fit with a very simple multicolor
441: disk blackbody model, {\tt diskbb} \citep{mit84}. This assumes that
442: the disk temperature $T(r)\propto r^{-3/4}$, i.e. has continuous
443: stress at the inner boundary. The emitting area can still be derived
444: from the data by applying a correction factor for this \citep{kub98}
445: as the overall shape of this spectrum is very similar to that produced
446: by a stress-free inner boundary condition \citep{gme01}. We also
447: include a Comptonized tail produced from (blackbody) seed photons tied
448: to the inner disk temperature and absorption fixed at Galactic values.
449: The disk so dominates the CCD bandpass that the Compton tail is poorly
450: constrained, and allowing the spectral index to be free gives best fit
451: parameters, for which the Comptonized tail is very steep. This is due
452: to the {\tt bhspec} disk shape being subtly different to a simple sum
453: of blackbodies, firstly as relativistic effects smear out the Wien
454: tail and secondly as the color-temperature correction is not constant
455: as a function of radius. These combine to give a broader disk spectrum
456: than predicted by {\tt diskbb} so the low temperature Comptonization
457: is required to extend the emission to higher energies. Instead, we
458: make the assumption that simultaneous high energy data exist to
459: constrain the spectral index (such as from an {\it RXTE} monitoring
460: campaign: e.g. GDP08), hence we fix this at the input value of 2.2 in
461: the CCD fits, but allow it to be free in the proportional counter
462: models.
463:
464: Fig. \ref{f:lt} shows the resulting $L-T$ diagram from fitting the CCD
465: and proportional counter simulations for the two new stress
466: prescriptions considered here (green: beta disk with $\alpha=0.1$,
467: blue: mean disk with $\alpha=0.1$), together with the two previous
468: alpha disk models for comparison (black: $\alpha=0.01$ and red,
469: $\alpha=0.1$). All the stress prescriptions which yield large surface
470: density ($\alpha=0.01$, mean and beta disks) show very similar behavior.
471: Only the alpha disk with $\alpha=0.1$ gives significantly higher
472: temperatures at high luminosities. However, at low luminosities, all
473: the stress prescriptions give a fairly good $T^4$ relation, showing
474: clearly that the \cite{mfr00} result of a substantial increase in
475: color temperature at low luminosities is indeed an artifact, and is
476: {\em not} representative of a gas pressure dominated beta disk
477: \citep[c.f. the discussion in][ where the Merloni et al. result was
478: used to argue against a beta disk model]{gad04}.
479:
480: \begin{figure*}
481: \begin{center}
482: \hbox{
483: \includegraphics[width=0.5\textwidth]{f4a.eps}
484: \includegraphics[width=0.5\textwidth]{f4b.eps}
485: %\psfig{figure=f4a.eps,width=0.5\textwidth,angle=0}
486: %\psfig{figure=f4b.eps,width=0.5\textwidth,angle=0}
487: }
488: \end{center}
489: \caption{a) The best-fit {\tt diskbb} models to simulated {\tt bhspec}
490: disk spectra with $a_*=0.5$ and $\log l=-0.3$ for the proportional
491: counter (red, solid curve) and CCD (green, dashed curve)
492: responses. For comparison, we also plot actual {\tt bhspec} model
493: (blue, dotted curve) used to generate the simulated spectrum. The {\tt
494: diskbb} spectrum is narrower than {\tt bhspec} so the best fit models
495: adjust as the bandpass changes. The CCD gives a lower temperature by a
496: factor 1.15, and the discrepancy at 5-10~keV is taken up by the
497: freedom in fitting the Comptonized tail. b) A similar comparison of
498: {\tt kerrbb} models with the same {\tt bhspec} model plotted in panel
499: a. However, in this case the {\tt kerrbb} spectra are {\em not} best
500: fit models. Instead, they have spin and mass accretion rates set at the
501: intrinsic parameters for the {\tt bhspec} model, but with
502: color-temperature corrections chosen to have the same ratio (1.15, see text
503: for further details) as the best-fit {\tt diskbb} temperatures. It is
504: plain that the color-temperature corrections derived from the {\tt
505: diskbb} fits are not appropriate for {\tt kerrbb} because the models
506: have different spectral shapes. \label{f:spec} }
507: \end{figure*}
508:
509: The luminosity in these plots is derived as in the $L-T$ plots of GD04
510: and DGK07 by correcting the disk flux and temperature for inclination
511: angle and relativistic effects. The general relativistic corrections
512: are taken from \cite{zcc97} for a Schwarzschild black hole as this is
513: closer to an $a_*=0.5$ disk than the alternative tabulation for
514: $a_*=0.998$. However, these have very little effect at $60^\circ$, as
515: this is where the Doppler blueshift and gravitational redshifts
516: approximately cancel, thus the flux to luminosity conversion is very
517: close to a simple disk area correction of $L=2\pi D^2 F/\cos i=4\pi
518: D^2 F$ for $\cos i =0.5$, and the temperature correction is
519: negligible. However, the recovered flux is less than the input value
520: (e.g. for the highest luminosity points made from $\log l=-0.2$, while
521: the dense disks give $-0.26$) as limb darkening is present in
522: the {\tt bhspec} radiation transfer.
523:
524: The solid lines show the predicted $L-T$ relation for a constant color
525: temperature of 1.6, 1.8, 2.0 and 2.2 again as in GD04, but scaling
526: the disk area to that expected from an $a_*=0.5$ black hole. Clearly
527: the CCD data are very close to a constant value of $f_{\rm col}=1.7$,
528: while the proportional counter bandpass gives a small change in
529: $f_{\rm col}$ from 1.8-2.0 for the dense disks, and 1.8-2.2 for the
530: alpha disk with $\alpha=0.1$.
531:
532: The top panels of Fig. \ref{f:l-1} show the effective optical depth of
533: the disk ($\tau_{\rm eff} \equiv \sqrt{ \tau_{\rm abs}(\tau_{\rm
534: es}+\tau_{\rm abs})}$) as a function of radius for $\log l=-1$ and
535: $-0.2$ respectively. For comparison we also plot the surface density
536: (middle) and effective temperature (bottom) for the same models. For
537: alpha disks the minimum in $\tau_{\rm eff}$ closely corresponds to the
538: minimum in $\Sigma$. Furthermore, these minima are close to the radius
539: of maximum $T_{\rm eff}$, so the majority of the flux will be produced
540: in the regions of the disk which have the lowest opacity and hence the
541: highest color temperature. Comparison of the top and middle panels
542: show that $\tau_{\rm eff}$ is very similar in shape to $\Sigma$. For
543: a given alpha, the beta disk always yields a larger $\Sigma$ (and,
544: therefore, a larger $\tau_{\rm eff}$) than the alpha disk, with the
545: mean disk midway between them. Due to the different scalings with $l$
546: (see Eqs. \ref{eq:siggas} and \ref{eq:sigrad}), the models become more
547: discrepant as luminosity increases. The alpha disk with $\alpha=0.1$
548: (red, solid curve in Fig. \ref{f:l-1}) always has the lowest effective
549: optical depth, and so gives the largest color-temperature correction.
550: However, the effect of this is small at low luminosities
551: (Fig. \ref{f:lt}) as there is still substantial absorption opacity
552: (Fig. \ref{f:l-1}a) and the heating mainly occurs at high effective
553: optical depth. Thus, the photosphere acts simply as an atmosphere. By
554: contrast, at the highest luminosities, the alpha disk with
555: $\alpha=0.1$ approaches $\tau_{\rm eff}=1$. All of the flux is
556: dissipated above the effective photosphere itself, and the
557: color-temperature correction becomes much larger.
558:
559: We repeat the spectral fitting, replacing the simple {\tt diskbb}
560: model by the more physical {\tt kerrbb} spectrum \citep{li05}. Like
561: {\tt diskbb}, {\tt kerrbb} is a multicolor blackbody model, but it
562: includes all of the relativistic effects modeled in {\tt bhspec}
563: i.e. all the special and general relativistic smearing of the
564: intrinsic disk emission due to rapid rotation in strong gravity. These
565: broaden the spectrum so that it is not so sharply peaked as predicted
566: in the {\tt diskbb} models, but unlike {\tt bhspec} it assumes that
567: the intrinsic emission from each radius is just a color-temperature
568: corrected blackbody, and that this color correction factor is the same
569: at all radii. Therefore, the derived $f_{\rm col}$ will arise solely
570: from differences in the treatment of the surface emission \citep[see
571: e.g.][]{sha06}. We fix the mass, distance, inclination and spin, and
572: fit for the mass accretion rate and color-temperature correction
573: assuming no stress on the inner boundary and no returning radiation,
574: but including limb darkening. Fig. \ref{f:fcol1} shows these for the
575: CCD (left panel) and proportional counter (right panel) data
576: respectively. Again the three dense disk prescriptions all show
577: similar results, with the alpha disk with $\alpha=0.1$ giving higher
578: color-temperature correction at high luminosity, as before. However,
579: the values of the derived color-temperature corrections are {\em
580: different} to those derived from the {\tt diskbb} approach. The
581: color-temperature correction for {\tt kerrbb} does {\em not} stay
582: constant in the CCD bandpass, apparently in conflict with the observed
583: $L\propto T^4$ relation from {\tt diskbb}. It increases from 1.6--1.9,
584: which at least goes through the $f_{\rm col}=1.7$ constant value seen
585: with {\tt diskbb}. By contrast, the proportional counter bandpass
586: gives values of the {\tt kerrbb} color-temperature correction which
587: are always lower than those seen from {\tt diskbb}, as well as
588: spanning a wider range, from 1.35--1.8
589:
590: \begin{figure*}
591: %\begin{center}
592: \hbox{
593: \includegraphics[width=0.95\textwidth]{f5.eps}
594: %\psfig{figure=f5.eps,width=0.95\textwidth,angle=0}
595: }
596: %\end{center}
597:
598: \caption{Effective optical depth at flux maximum as a function of spin
599: and luminosity. The contour labels are values of $\log \tau_{\rm
600: eff}$. The panels correspond to alpha disks with $\alpha=0.01$ (a)
601: and $\alpha=0.1$ (b), mean disks with $\alpha=0.1$ (c), and beta disks
602: with $\alpha=0.1$ (d). Note the change in scale of the contours for
603: panel d. There is less than one order of magnitude change in effective
604: optical depth across the whole range, so we decrease the spacing to
605: 0.1 dex as opposed to 1 dex for the other panels.\label{f:taue}}
606: \end{figure*}
607:
608: The answer to these issues lies in the detailed shape of the
609: spectrum. Fig. \ref{f:spec}a shows the initial {\tt bhspec} disk
610: (blue, dotted line) and its best fit {\tt diskbb} spectra to the
611: proportional counter (red, solid line) and CCD (green, dashed line)
612: bandpasses for one of the dense disk prescriptions ($\alpha=0.01$) for
613: an input luminosity of $\log l=-0.3$. The higher energy bandpass of
614: the proportional counter weights the fit to higher temperatures/lower
615: normalization than those for the CCD, and the {\tt diskbb} model is a
616: good fit to the {\tt bhspec} above the 3~keV lower limit of the
617: proportional counter bandpass. However, the best fit to the CCD data
618: does not look so compelling. It is clear that the {\tt diskbb}
619: spectrum is narrower than the {\tt bhspec} disk emission, and the
620: 'best fit' matches this only in the 1-3~keV range where the
621: signal-to-noise is maximum. At higher energies, the Compton tail
622: provides additional freedom to match the data.
623:
624: \begin{figure*}
625: %\begin{center}
626: \hbox{
627: \includegraphics[width=0.95\textwidth]{f6.eps}
628: %\psfig{figure=f6.eps,width=0.95\textwidth,angle=0}
629: }
630: %\end{center}
631:
632: \caption{Color correction at flux maximum for {\tt kerrbb}
633: as a function of spin and
634: luminosity. The solid and dashed contours are computed using the {\it RXTE}
635: PCA and {\it Swift} XRT responses, respectively. See text for more details.
636: The panels correspond to different stress prescriptions as described
637: in Fig. \ref{f:taue}. \label{f:fcol}}
638: \end{figure*}
639:
640:
641: Fig. \ref{f:spec}b shows the {\tt kerrbb} models with mass accretion
642: rate chosen to give $\log l=-0.3$ for a black hole with $a_*=0.5$
643: (i.e. $8.537\times 10^{18}$~g/s), with color-temperature correction of
644: 1.7 for the proportional counter (red). The ratio of best fit
645: temperatures from the {\tt diskbb} models above implies that the
646: color-temperature correction should be a factor 1.15 smaller in the
647: CCD compared to the proportional counter i.e. giving $f_{col}=1.47$
648: for the CCD (blue). However, the {\tt kerrbb} spectrum is much
649: broader than {\tt diskbb} as it includes the relativistic smearing of
650: the continuum, and the proportional counter model is actually a fairly
651: good description of the {\tt bhspec} disk shape apart from around the
652: peak at $\sim 1$~keV. This is because {\tt bhspec} includes the
653: photoelectric edge features in the atmosphere, calculating the
654: radiative transfer through the ion populations rather than simply
655: assuming a diluted blackbody. These atomic features, from partially
656: ionized oxygen/silicon/iron L shell, depress the continuum around
657: 1~keV from that predicted by the {\tt kerrbb} models.
658:
659: Thus the {\tt kerrbb} model appropriate for the proportional counter
660: data gives a good match below $\sim 0.6$~keV but {\em overestimates}
661: the flux at the peak. Decreasing the color-temperature correction
662: simply increases the disparity between the {\tt kerrbb} and {\tt
663: bhspec} disk spectra in the CCD bandpass, especially in the crucial
664: 1--3~keV region where the signal-to-noise is high. Thus the
665: appropriate {\tt kerrbb} model would have slightly {\em higher}
666: color-temperature correction for the CCD than for the proportional
667: counter, as observed (see Figs. 3a and b). These differences in
668: spectral shape mean that color-temperature corrections derived from
669: {\tt diskbb} cannot be simply applied to {\tt kerrbb} (and vice
670: versa).
671:
672:
673: \section{Color-temperature correction and effective optical depth for all spin}
674: \label{fcol_all}
675:
676: We use the insights derived above from the particular case of
677: $a_*=0.5$ to extend our analysis of $\tau_{\rm eff}$ and $f_{\rm col}$
678: to a larger range of black hole spins and luminosities. We are
679: primarily interested the role $\tau_{\rm eff}$ plays in determining
680: $f_{\rm col}$. Since our fitting procedure is most sensitive to the
681: hottest annuli, we will focus only on $\tau_{\rm eff}$ at the radius
682: where $T_{\rm eff}$ is maximum. For each $a_*$ and $l$ we calculate
683: the radius of maximum $T_{\rm eff}$, and evaluate $\tau_{\rm eff}$ at
684: the midplane for the four stress prescriptions considered in
685: \S\ref{fcol_0.5}. Fig. \ref{f:taue} shows contour plots of the
686: effective optical depth at the radius of maximum flux
687: (c.f. Fig. \ref{f:l-1}) as a function of spin and luminosity for four
688: different stress prescriptions. The calculations span a range of $a_*$
689: from 0 to 0.8 and $\log l$ from -2 to 0. (We consider higher spins
690: below.)
691:
692: For comparison, we also fit the {\tt kerrbb} model to simulated {\tt
693: bhspec} spectra using the PCA and XRT responses. The procedure we
694: follow is identical to that described in \S\ref{fcol_0.5}, but we now
695: calculate $f_{\rm col}$ over the same range of $a_*$ and $l$ as in
696: Fig. \ref{f:taue}. Fig. \ref{f:fcol} shows the resulting
697: color-temperature corrections for {\tt kerrbb} fit over a proportional
698: counter (PCA, solid) and CCD (XRT, dashed) bandpass. (We stress again
699: that the different shape of {\tt kerrbb} to that of {\tt diskbb} means
700: that these are {\em not} applicable to {\tt diskbb} fits.) Except for
701: the highest luminosities of the alpha disk model with $\alpha=0.1$,
702: the CCD response always yields a higher $f_{\rm col}$ than the
703: proportional counter.
704:
705: A key issue which determines the spectrum is the fraction of energy
706: dissipated in the region where thermalization will be incomplete
707: i.e. above $\tau_{\rm eff}\sim 3$. This crucial translucent region of
708: the disk is simply an atmosphere (negligible dissipation) for very
709: optically thick disks, as the heating occurs at much greater
710: depths. However, when the whole disk has $\tau_{\rm eff}<30$ then a
711: substantial fraction of the energy can be emitted in the translucent
712: region, leading to a large increase in $f_{\rm col}$.
713: Fig. \ref{f:taue} shows that the alpha disk with $\alpha=0.1$ (panel
714: b) is the only prescription for which the majority of the flux can be
715: produced in the region with $\tau_{\rm eff}<3$ (i.e. $\log \tau_{\rm
716: eff}=0.5$). This happens only for luminosities near Eddington and is
717: only weakly dependent on spin. A comparison with Fig. \ref{f:fcol}
718: shows that these models are also the ones with highest $f_{\rm col}$.
719: The other three stress prescriptions (alpha disks with $\alpha=0.01$,
720: beta and mean disks with $\alpha=0.1$) tend to give more modest
721: $f_{\rm col}$ at these $l$. This comparison also identifies
722: $\tau_{\rm eff} \simeq 30$ as the point at which dissipation in the
723: atmosphere becomes important. At lower luminosities ($\log l \lesssim
724: -0.5$) $\tau_{\rm eff}$ is larger and $f_{\rm col}$ is rather
725: insensitive to the stress prescription.
726:
727: \begin{figure*}
728: %\begin{center}
729: \hbox{
730: \includegraphics[width=0.95\textwidth]{f7.eps}
731: %\psfig{figure=f7.eps,width=0.95\textwidth,angle=0}
732: }
733: %\end{center}
734:
735: \caption{As in Fig. \ref{f:taue} but extended to high spin
736: for an alpha disk with $\alpha=0.01$. The left panel gives effective
737: optical depth at flux maximum, while the right panel gives the
738: values of $f_{\rm col}$ for {\tt kerrbb} computed using the
739: {\it RXTE} PCA (solid) and {\it Swift} XRT (dashed) responses, respectively.
740: \label{f:a0.99}}
741: \end{figure*}
742:
743: For these low-to-moderate spins ($a_* \le 0.8$) the variations of
744: $f_{\rm col}$ and $\tau_{\rm eff}$ have a much stronger dependence on
745: $l$ than on $a_*$. These spins are probably the ones most appropriate
746: for the majority of BHB which have low mass companions, as there is
747: not enough mass in the companion for the accreted material to
748: significantly change the angular momentum of the black hole
749: \citep{kak99}. Thus they fairly accurately reflect the birth spin
750: distribution from massive star collapse, for which the best estimates
751: give spin $a_* \lesssim 0.75-0.9$ \citep{gsm04}. However, in the rarer
752: high mass X-ray binaries \citep[which probably also form the ULX
753: population e.g.][]{kin08} there is more mass in the companion star, so
754: accretion may change the mass and spin of the black hole if the
755: lifetime is long enough. Thus we also investigate higher spins, but
756: these are computationally problematic for alpha disks with
757: $\alpha=0.1$ as they become very effectively optically thin for $a_*
758: \sim 1$ and $l \sim 1$. Since this stress prescription is ruled out by
759: both spectral and variability behavior for the BHB, we take only alpha
760: disks with $\alpha=0.01$ as a guide to the behavior of the other dense
761: disks (mean and beta) as a function of spin for $0.8 < a_* \le
762: 0.99$. Contour plots for $f_{\rm col}$ and $\tau_{\rm eff}$ are shown
763: in Fig. \ref{f:a0.99}. The contours show a slightly stronger
764: dependence on $a_*$ at these higher values, with $\tau_{\rm eff}$
765: decreasing and $f_{\rm col}$ increasing as $a_*$ approaches
766: unity. Nonetheless, the overall effect is still relatively small
767: compared to the much stronger dependence on $l$, and the conclusions
768: drawn from Fig. \ref{f:lt} generalize quite broadly to other spins.
769: Thus this predicts that any of the dense disk stress prescriptions
770: will give disk spectra at high spin with $f_{\rm col}$ increasing
771: slightly with $l$ in a way which is not significantly different to
772: that for low spin.
773:
774: Thus extreme spins should simply manifest themselves as higher
775: temperature disks at a given luminosity, giving a clear signature of a
776: maximal Kerr black hole. However, none of the high mass X-ray
777: binaries (e.g. Cyg X-1) show any signs of this, so we
778: conclude that accretion has not yet had time to significantly spin up
779: the black holes in these systems. The ULXs are also potentially high
780: spin objects \citep{ebi03,hak08}, but these have
781: spectra which are often more complex than a simple disk model \citep[see,
782: however,][]{hak08}, in which case they do not give a
783: straightforward diagnostic of black hole mass and spin \citep{dak06}.
784:
785: \section{Discussion}
786: \label{discus}
787:
788: \subsection{Stress as a function of radius}
789:
790: The observation that stable disk spectra are seen spanning $l\sim
791: 0.05-0.5$ shows that disks are not subject to the limit cycles
792: predicted by the radiation pressure instability in this range. This
793: rules out an alpha type stress, and instead {\em requires} that the
794: surface density increases (or remains constant in the limit of
795: marginal stability) as a function of $l$ at each radius, and this in
796: itself implies that the accretion disk should remain effectively
797: optically thick at high luminosities. Therefore the color-temperature
798: corrections will be relatively constant, so {\em predicting} the
799: approximate $L\propto T^4$ relation for the disk dominated
800: spectra. The observation that most binaries show an approximate
801: $L\propto T^4$ is strong confirmation of this conclusion.
802:
803: Sadly, this also rules out the radiation pressure instability as being
804: the physical mechanism for any observed behavior for $l<0.5$. This is
805: otherwise an attractive possibility for the origin of the very high
806: state, the other spectral state seen in binaries at high mass
807: accretion rates. Although this does not show the predicted limit
808: cycles, there is the possibility that the non-linear outcome of the
809: instability could instead lead to strong Comptonization of the disk
810: \citep[e.g.][]{rac00b,kam04}. Nonetheless, the arguments above
811: preclude an alpha type stress, at least below $l\sim 0.5$, yet the
812: very high state is indeed seen in the same range in luminosity as the
813: disk dominated state \citep[e.g.][]{ram06}.
814:
815: At higher luminosities \citep[up to $l\sim 3$,][]{dwg04}, there is a
816: limit cycle observed in GRS1915+104 which could be the remnant of this
817: radiation pressure instability from a marginally stable stress such as
818: predicted with the mean disk \citep{hkm91,man06}. However, this limit
819: cycle behavior could alternatively be from other instabilities to do
820: with the super Eddington flows. This is an important distinction, as
821: the classic radiation pressure instability is not scale invariant with
822: mass, but is triggered at $l\propto m^{-1/8}$. Thus even if the disk
823: is stable in stellar mass black holes up to $l\sim 0.5$, it can be
824: unstable at $l\gtrsim 0.09$ for a $10^7 \msun$ AGN. In this case the
825: instability could play a role in producing the puzzling, 'soft excess'
826: seen in high mass accretion rate AGN, predominantly Narrow Line
827: Seyfert 1's \citep{rac00b,fab02}. However, there are multiple
828: similarities between the very high state and Narrow Line Seyfert 1
829: spectra to make it more likely that there is a similar explanation for
830: both types of object.
831:
832: Similarly, those properties of super Eddington flows which likewise
833: depend on the alpha stress are also probably not a realistic
834: description of these disks. Radiation trapping is a generic feature
835: of any of the stress prescriptions \citep[e.g.][]{hkm91}, resulting in
836: optically thick advection of energy especially in the disk midplane
837: \citep{wmm01}. However, the denser disks which result from the
838: alternative stress prescriptions are less likely to become effectively
839: optically thin to the escaping radiation, so are unlikely to show the
840: very large color-temperature correction which can arise from
841: overheating of alpha disks \citep[e.g.][]{bel98,kaw03}. We caution
842: that fitting such models to high mass accretion rate spectra from ULX
843: and NLS1's may not be appropriate.
844:
845: \subsection{Caveats}
846:
847: So what then are the caveats to using disk spectra as an estimator for
848: black hole spin? The first is that the spectra {\em must} be disk
849: dominated. The disk radius can be under -- or overestimated when a
850: substantial fraction of the dissipation goes instead into a corona,
851: either in the very high \citep{kad04,dak06} or low/hard state (GDP08).
852: This may be the origin of the discrepancy in spin determination by disk
853: spectral fitting in GRS 1915+104 \citep[compare][]{mid06,mcc06}.
854: The spectra of GRS 1915+104 may well be more
855: complex than the disk dominated spectra seen from BHB due to its
856: higher luminosity \citep{dwg04}, perhaps
857: powering strong winds (see below). It seems premature to apply the
858: uncertainties from this one pathological object to other sub-Eddington
859: BHB \citep[e.g.][]{raf08}.
860:
861: Secondly, the value of the color-temperature correction is robust for
862: all of the dense disk stress prescriptions, but varies with bandpass
863: and also varies with the detailed disk model used to fit the
864: spectra. We caution that differences in spectral shape between the
865: models means that the same color-temperature correction factors {\em
866: cannot} be simply applied to {\tt diskbb} and {\tt kerrbb}
867: fits. Ironically, for both the generic proportional counter and CCD
868: bandpasses, the {\tt diskbb} fits give a simpler (approximately
869: constant color temperature) representation of the data.
870:
871: Thirdly, the structure of the disk as $l\to 1$ is not well modeled by
872: this code. We neglect optically thick advection, and changes in the
873: scale height of the disk with radius which may lead to self shielding
874: for highly inclined objects \citep[e.g.][]{wtf05}, but these are
875: unlikely to have much effect for $l<0.5$, where the majority of BHB
876: data are taken.
877:
878: However, disks at high luminosities are likely to also power
879: substantial winds. There are clear signs of an equatorial disk wind
880: from observations of highly ionized, blueshifted, Fe K$\alpha$
881: absorption lines. These are ubiquitously seen in CCD data from
882: high/soft state BHB (i.e. $l>0.1$) with inclination angles $\ge
883: 60^\circ$ (e.g. DGK07). This material is probably a thermally driven
884: disk wind, where irradiation of the outer disk layers produces a
885: Compton heated skin whose high temperature allows gas particles to
886: have enough energy to escape at large radii \citep{bms83,woo96},
887: though this may be (substantially?) enhanced by magnetic fields
888: \citep{mil06}. Such winds almost certainly become more powerful with
889: $l$, especially as $l\to 1$ and radiation driving enhances them still
890: further \citep{pak02}. Counterintuitively, these stronger winds may
891: become less observable as the stronger irradiation means that they
892: become more ionized, losing even the He- and H-like Fe
893: \citep{woo96}. Thus there may be winds which have substantial column,
894: but are unobservable as they are completely ionized. Electron
895: scattering has the effect of suppressing the observed luminosity by a
896: factor $e^{-\tau}$ for lines of sight intersecting the wind, while the
897: scattered flux $(1-e^{-\tau})$ can enhance the luminosity along the
898: disk axis by a factor $1+(1-e^{- \tau})\Omega/4\pi$ where $\Omega$ is
899: the solid angle subtended by the wind. This may be the origin of the
900: occasionally observed `bends' in the $L-T$ diagram, where the data
901: deviate away from the generally observed slow increase in $f_{\rm
902: col}$ as a function of luminosity e.g. in GRO J1655-40 there is a much
903: stronger increase in $f_{\rm col}$ with $l$ than predicted by the
904: dense disk models (e.g. Fig 1b), while in LMC X--3 there is a slight
905: {\em decrease} \citep[see e.g.][]{ddb06}. While the inclination of
906: LMC X-3 is not well known, GRO J1655-40 is plainly at very high
907: inclination, and has clear wind features detected in multiple
908: observations \citep{mil06,sal07}.
909:
910: Thus as well as avoiding spectra with strong tails, and super
911: Eddington sources, we also caution against any serious attempt to
912: derive spin from `bent' $L-T$ diagrams, where winds may be important.
913:
914: Our results also hinge on the fact that the dissipation above the
915: effective photosphere is small when the surface density is
916: sufficiently large, so our assumptions about the dissipation profile
917: with height may be an important caveat. Previous work
918: \citep{dav05,bla06} has demonstrated that dissipation profiles
919: motivated by radiative, stratified shearing box simulations
920: \citep{tur04,hks06} do not put enough dissipation near the surface to
921: significantly alter the spectrum, although a more exhaustive range of
922: parameter space needs to be explored. However, the alternative stress
923: prescriptions discussed here are partially motivated by theoretical
924: arguments that limit the magnetic stresses due to transport of buoyant
925: magnetic field \citep[see e.g.][]{sar84,mer03}. If transport of
926: magnetic field is significantly larger than currently seen in
927: simulations, dissipation would be concentrated much nearer the surface
928: than our models assume \citep{mer03}. This would give significant
929: dissipation above the photosphere, and hence lead to a marked increase
930: in color-temperature correction compared to those calculated here for
931: the dense disks. As long as this is not a strong function of
932: luminosity then it could still fit the data, but would lead to the
933: spin being overestimated. Since the spins are generally
934: low-to-moderate, there is not much scope for the real stresses to
935: produce a significant fraction of the dissipation within the
936: photosphere.
937:
938: Similarly, the data also preclude a large amount of dissipation at the
939: last stable orbit. If this were to thermalize it would lead to higher
940: temperature emission, so current models would overestimate the spin,
941: yet these already favor low spins. Alternatively, if the dissipation
942: did not thermalize, it would instead produce a substantial
943: non--thermal component in the spectrum, yet the spectra are dominated
944: by the disk emission in the high/soft state. Thus there is only
945: limited scope for stresses at the last stable orbit in disk dominated
946: spectra, in contrast to the results from non-radiative, relativistic
947: magnetohydrodynamic simulations \citep[e.g.][]{bhk08}. It may be that
948: the character of the flow qualitatively changes when radiative cooling
949: is included in the simulations, as the small scale height fields
950: sustained by a thin (radiatively cooled) disk may lead to much less
951: stress at the last stable orbit than the large scale fields which can
952: be generated in the large scale height, non-radiative flows
953: \citep{aap03,snm08}.
954:
955: This potential difference in dissipation between large scale height
956: flows and thin disks may actually be observed in the behavior of the
957: transition between the low/hard state, where the spectrum is produced
958: by Compton scattering in a hot, optically flow, and the disk dominated
959: state where the emission is quasi--thermal. The low/hard state is
960: probably produced by a large scale height flow, which has an
961: efficiency which is at least a factor 3 lower than that of a
962: (stress--free inner boundary condition) thin disk down to the same
963: radius \citep[e.g.][]{nay95}. This predicts that the luminosity at
964: the transition should jump by at least a factor 3, yet the data show a
965: smooth transition of less than a factor $\sim 2$ \citep[e.g. the
966: compilation black hole binaries, especially Cyg X-1,
967: in][]{dag04}. This can be explained if the large scale height, less
968: efficient flow in the low/hard state has continuous dissipation across
969: the last stable orbit due to large scale height magnetic fields. In
970: effect this allows the flow to extend down to smaller radii, so it
971: taps a larger fraction of the gravitational potential energy, hence
972: somewhat compensating for its lower radiative efficiency compared to
973: that of the thin disk (Done et al. 2008, in preparation).
974:
975: \section{Conclusions}
976: \label{conc}
977:
978: The detailed nature of angular momentum transport remains a
979: significant uncertainty in our understanding of accretion flows and
980: their emission. It is not yet entirely certain that the MRI, which is
981: the best candidate for the source of the turbulent stress, truly does
982: saturate at a level high enough to satisfy observational constraints
983: \citep{pcp07,fap07,kpl07}. Even if this question is resolved in favor
984: of the MRI, a detailed understanding of the stress and resulting disk
985: structure will likely require fully-relativistic, global simulations
986: with realistic thermodynamics. Since such simulations require
987: significant advances in computing power, we are forced rely on much
988: simpler parameterizations of the stress until these become available.
989: Even though these prescriptions are ad hoc, they may still capture
990: important aspects of the physics in real astrophysical flows and
991: provide useful constraints by direct comparison with observations.
992:
993: With this motivation, we have computed the accretion disk spectra
994: predicted by several different stress prescriptions using the most
995: complete spectral code currently available. This includes both non-LTE
996: ion populations, with radiative transfer in the disk and full general
997: relativistic ray tracing to propagate this flux to the observer. These
998: models give results which are very close to the observed behavior of
999: the spectra of black hole binaries in the thermal dominant (high/soft)
1000: state from $L/L_{Edd}\sim 0.06\to 0.6$. The majority of this data
1001: comes from proportional counters, and generally shows that the
1002: temperature changes slightly more rapidly with luminosity than
1003: expected for a disk of constant inner radius, and constant
1004: color-temperature correction (e.g. the compilation of DG04 and DGK07).
1005: This is as predicted by all the high surface density disk models
1006: presented here, and should hold generally for all stress prescriptions
1007: where less than 10 per cent of the energy is dissipated above the
1008: effective photosphere. The surface layers then simply act like a
1009: passive atmosphere, with properties set by the effective
1010: temperature. The increase in temperature with luminosity gives rise to
1011: a small increase in color-temperature correction and the resulting
1012: spectra are remarkably similar irrespective of the detailed form of
1013: the stress.
1014:
1015: Such model fitting can also place interesting constraints on the
1016: angular momentum transport. The classic alpha disk with $\alpha=0.1$
1017: becomes effectively optically thin at the highest luminosities. The
1018: resulting increase in color temperature with luminosity is much more
1019: rapid than observed from existing data, ruling out such stress
1020: prescriptions. The standard alpha disk is also unstable to the
1021: thermal-viscous radiation pressure instability which predicts limit
1022: cycle behavior which is not observed, again showing that the disk is
1023: denser than such models predict. However, even these models give the
1024: same spectra at low luminosities ($L \lesssim 0.1 L_{\rm Edd}$), where
1025: the surface density is sufficiently large for the disk to remain very
1026: effectively optically thick. Therefore, where we have a clear view of
1027: the disk, unaffected by a moderately optically thick disk wind, the
1028: disk spectra should provide a relatively robust estimator of the disk
1029: inner radius, and plausibly, the spin of the black hole.
1030:
1031: \acknowledgements{We thank K. Beckwith, O. Blaes, M. Gierli{\'n}ski,
1032: I. Hubeny, J. Krolik, and A. Kubota for useful discussions. CD
1033: acknowledges support from a PPARC senior fellowship, and a Royal
1034: Society conference grant and Omer Blaes who together funded the visit
1035: to Santa Barbara at which this paper was begun. SD is supported by
1036: NASA grant number PF6-70045, awarded by the Chandra X-ray Center,
1037: which is operated by the Smithsonian Astrophysical Observatory for
1038: NASA under contract NAS8-03060.}
1039:
1040: \begin{thebibliography}{73}
1041: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
1042:
1043: \bibitem[{{Afshordi} \& {Paczy{\'n}ski}(2003)}]{aap03}
1044: {Afshordi}, N., \& {Paczy{\'n}ski}, B. 2003, \apj, 592, 354
1045:
1046: \bibitem[{{Balbus} \& {Hawley}(1991)}]{bah91}
1047: {Balbus}, S.~A., \& {Hawley}, J.~F. 1991, \apj, 376, 214
1048:
1049: \bibitem[{{Beckwith} {et~al.}(2008){Beckwith}, {Hawley}, \& {Krolik}}]{bhk08}
1050: {Beckwith}, K., {Hawley}, J., \& {Krolik}, J. 2008, ArXiv e-prints, 801
1051:
1052: \bibitem[{{Begelman} {et~al.}(1983){Begelman}, {McKee}, \& {Shields}}]{bms83}
1053: {Begelman}, M.~C., {McKee}, C.~F., \& {Shields}, G.~A. 1983, \apj, 271, 70
1054:
1055: \bibitem[{{Beloborodov}(1998)}]{bel98}
1056: {Beloborodov}, A.~M. 1998, \mnras, 297, 739
1057:
1058: \bibitem[{{Blaes} {et~al.}(2006){Blaes}, {Davis}, {Hirose}, {Krolik}, \&
1059: {Stone}}]{bla06}
1060: {Blaes}, O.~M., {Davis}, S.~W., {Hirose}, S., {Krolik}, J.~H., \& {Stone},
1061: J.~M. 2006, \apj, 645, 1402
1062:
1063: \bibitem[{{Davis} {et~al.}(2005){Davis}, {Blaes}, {Hubeny}, \&
1064: {Turner}}]{dav05}
1065: {Davis}, S.~W., {Blaes}, O.~M., {Hubeny}, I., \& {Turner}, N.~J. 2005, \apj,
1066: 621, 372
1067:
1068: \bibitem[{{Davis} {et~al.}(2006){Davis}, {Done}, \& {Blaes}}]{ddb06}
1069: {Davis}, S.~W., {Done}, C., \& {Blaes}, O.~M. 2006, \apj, 647, 525
1070:
1071: \bibitem[{{Davis} \& {Hubeny}(2006)}]{dah06}
1072: {Davis}, S.~W., \& {Hubeny}, I. 2006, \apjs, 164, 530
1073:
1074: \bibitem[{{Done} \& {Gierli{\'n}ski}(2004)}]{dag04}
1075: {Done}, C., \& {Gierli{\'n}ski}, M. 2004, Progress of Theoretical Physics
1076: Supplement, 155, 9
1077:
1078: \bibitem[{{Done} {et~al.}(2007){Done}, {Gierli{\'n}ski}, \& {Kubota}}]{dgk07}
1079: {Done}, C., {Gierli{\'n}ski}, M., \& {Kubota}, A. 2007, \aapr, 15, 1
1080:
1081: \bibitem[{{Done} \& {Kubota}(2006)}]{dak06}
1082: {Done}, C., \& {Kubota}, A. 2006, \mnras, 371, 1216
1083:
1084: \bibitem[{{Done} {et~al.}(2004){Done}, {Wardzi{\'n}ski}, \&
1085: {Gierli{\'n}ski}}]{dwg04}
1086: {Done}, C., {Wardzi{\'n}ski}, G., \& {Gierli{\'n}ski}, M. 2004, \mnras, 349,
1087: 393
1088:
1089: \bibitem[{{Eardley} \& {Lightman}(1975)}]{eal75}
1090: {Eardley}, D.~M., \& {Lightman}, A.~P. 1975, \apj, 200, 187
1091:
1092: \bibitem[{{Ebisawa} {et~al.}(1993){Ebisawa}, {Makino}, {Mitsuda}, {Belloni},
1093: {Cowley}, {Schmidtke}, \& {Treves}}]{ebi93}
1094: {Ebisawa}, K., {Makino}, F., {Mitsuda}, K., {Belloni}, T., {Cowley}, A.~P.,
1095: {Schmidtke}, P.~C., \& {Treves}, A. 1993, \apj, 403, 684
1096:
1097: \bibitem[{{Ebisawa} {et~al.}(1991){Ebisawa}, {Mitsuda}, \& {Hanawa}}]{emh91}
1098: {Ebisawa}, K., {Mitsuda}, K., \& {Hanawa}, T. 1991, \apj, 367, 213
1099:
1100: \bibitem[{{Ebisawa} {et~al.}(2003){Ebisawa}, {{\.Z}ycki}, {Kubota}, {Mizuno},
1101: \& {Watarai}}]{ebi03}
1102: {Ebisawa}, K., {{\.Z}ycki}, P., {Kubota}, A., {Mizuno}, T., \& {Watarai}, K.-y.
1103: 2003, \apj, 597, 780
1104:
1105: \bibitem[{{Fabian} {et~al.}(2002){Fabian}, {Ballantyne}, {Merloni}, {Vaughan},
1106: {Iwasawa}, \& {Boller}}]{fab02}
1107: {Fabian}, A.~C., {Ballantyne}, D.~R., {Merloni}, A., {Vaughan}, S., {Iwasawa},
1108: K., \& {Boller}, T. 2002, \mnras, 331, L35
1109:
1110: \bibitem[{{Fromang} \& {Papaloizou}(2007)}]{fap07}
1111: {Fromang}, S., \& {Papaloizou}, J. 2007, \aap, 476, 1113
1112:
1113: \bibitem[{{Galeev} {et~al.}(1979){Galeev}, {Rosner}, \& {Vaiana}}]{grv79}
1114: {Galeev}, A.~A., {Rosner}, R., \& {Vaiana}, G.~S. 1979, \apj, 229, 318
1115:
1116: \bibitem[{{Gammie} {et~al.}(2004){Gammie}, {Shapiro}, \& {McKinney}}]{gsm04}
1117: {Gammie}, C.~F., {Shapiro}, S.~L., \& {McKinney}, J.~C. 2004, \apj, 602, 312
1118:
1119: \bibitem[{{Gierli{\'n}ski} \& {Done}(2004)}]{gad04}
1120: {Gierli{\'n}ski}, M., \& {Done}, C. 2004, \mnras, 347, 885
1121:
1122: \bibitem[{{Gierlinski} {et~al.}(2008){Gierlinski}, {Done}, \& {Page}}]{gdp08}
1123: {Gierlinski}, M., {Done}, C., \& {Page}, K. 2008, ArXiv e-prints, 803
1124:
1125: \bibitem[{{Gierli{\'n}ski} {et~al.}(2001){Gierli{\'n}ski},
1126: {Macio{\l}ek-Nied{\'z}wiecki}, \& {Ebisawa}}]{gme01}
1127: {Gierli{\'n}ski}, M., {Macio{\l}ek-Nied{\'z}wiecki}, A., \& {Ebisawa}, K. 2001,
1128: \mnras, 325, 1253
1129:
1130: \bibitem[{{Hirose} {et~al.}(2006){Hirose}, {Krolik}, \& {Stone}}]{hks06}
1131: {Hirose}, S., {Krolik}, J.~H., \& {Stone}, J.~M. 2006, \apj, 640, 901
1132:
1133: \bibitem[{{Honma} {et~al.}(1991){Honma}, {Kato}, \& {Matsumoto}}]{hkm91}
1134: {Honma}, F., {Kato}, S., \& {Matsumoto}, R. 1991, \pasj, 43, 147
1135:
1136: \bibitem[{{Hubeny} \& {Lanz}(1995)}]{hal95}
1137: {Hubeny}, I., \& {Lanz}, T. 1995, \apj, 439, 875
1138:
1139: \bibitem[{{Hui} \& {Krolik}(2008)}]{hak08}
1140: {Hui}, Y., \& {Krolik}, J.~H. 2008, ArXiv e-prints, 803
1141:
1142: \bibitem[{{Ichimaru}(1977)}]{ich77}
1143: {Ichimaru}, S. 1977, \apj, 214, 840
1144:
1145: \bibitem[{{Kato} {et~al.}(1998){Kato}, {Fukue}, \& {Mineshige}}]{kfm98}
1146: {Kato}, S., {Fukue}, J., \& {Mineshige}, S., eds. 1998, {Black-hole accretion
1147: disks}
1148:
1149: \bibitem[{{Kawaguchi}(2003)}]{kaw03}
1150: {Kawaguchi}, T. 2003, \apj, 593, 69
1151:
1152: \bibitem[{{King}(2008)}]{kin08}
1153: {King}, A.~R. 2008, \mnras, 385, L113
1154:
1155: \bibitem[{{King} \& {Kolb}(1999)}]{kak99}
1156: {King}, A.~R., \& {Kolb}, U. 1999, \mnras, 305, 654
1157:
1158: \bibitem[{{King} {et~al.}(2007){King}, {Pringle}, \& {Livio}}]{kpl07}
1159: {King}, A.~R., {Pringle}, J.~E., \& {Livio}, M. 2007, \mnras, 376, 1740
1160:
1161: \bibitem[{{Kubota} \& {Done}(2004)}]{kad04}
1162: {Kubota}, A., \& {Done}, C. 2004, \mnras, 353, 980
1163:
1164: \bibitem[{{Kubota} \& {Makishima}(2004)}]{kam04}
1165: {Kubota}, A., \& {Makishima}, K. 2004, \apj, 601, 428
1166:
1167: \bibitem[{{Kubota} {et~al.}(2001){Kubota}, {Makishima}, \& {Ebisawa}}]{kme01}
1168: {Kubota}, A., {Makishima}, K., \& {Ebisawa}, K. 2001, \apjl, 560, L147
1169:
1170: \bibitem[{{Kubota} {et~al.}(1998){Kubota}, {Tanaka}, {Makishima}, {Ueda},
1171: {Dotani}, {Inoue}, \& {Yamaoka}}]{kub98}
1172: {Kubota}, A., {Tanaka}, Y., {Makishima}, K., {Ueda}, Y., {Dotani}, T., {Inoue},
1173: H., \& {Yamaoka}, K. 1998, \pasj, 50, 667
1174:
1175: \bibitem[{{Laor} \& {Netzer}(1989)}]{lan89}
1176: {Laor}, A., \& {Netzer}, H. 1989, \mnras, 238, 897
1177:
1178: \bibitem[{{Lasota}(2001)}]{las01}
1179: {Lasota}, J.-P. 2001, New Astronomy Review, 45, 449
1180:
1181: \bibitem[{{Li} {et~al.}(2005){Li}, {Zimmerman}, {Narayan}, \&
1182: {McClintock}}]{li05}
1183: {Li}, L.-X., {Zimmerman}, E.~R., {Narayan}, R., \& {McClintock}, J.~E. 2005,
1184: \apjs, 157, 335
1185:
1186: \bibitem[{{Lightman}(1974)}]{lig74}
1187: {Lightman}, A.~P. 1974, \apj, 194, 419
1188:
1189: \bibitem[{{Lightman} \& {Eardley}(1974)}]{lae74}
1190: {Lightman}, A.~P., \& {Eardley}, D.~M. 1974, \apjl, 187, L1+
1191:
1192: \bibitem[{{McClintock} {et~al.}(2006){McClintock}, {Shafee}, {Narayan},
1193: {Remillard}, {Davis}, \& {Li}}]{mcc06}
1194: {McClintock}, J.~E., {Shafee}, R., {Narayan}, R., {Remillard}, R.~A., {Davis},
1195: S.~W., \& {Li}, L.-X. 2006, \apj, 652, 518
1196:
1197: \bibitem[{{Merloni}(2003)}]{mer03}
1198: {Merloni}, A. 2003, \mnras, 341, 1051
1199:
1200: \bibitem[{{Merloni} {et~al.}(2000){Merloni}, {Fabian}, \& {Ross}}]{mfr00}
1201: {Merloni}, A., {Fabian}, A.~C., \& {Ross}, R.~R. 2000, \mnras, 313, 193
1202:
1203: \bibitem[{{Merloni} \& {Nayakshin}(2006)}]{man06}
1204: {Merloni}, A., \& {Nayakshin}, S. 2006, \mnras, 372, 728
1205:
1206: \bibitem[{{Middleton} {et~al.}(2006){Middleton}, {Done}, {Gierli{\'n}ski}, \&
1207: {Davis}}]{mid06}
1208: {Middleton}, M., {Done}, C., {Gierli{\'n}ski}, M., \& {Davis}, S.~W. 2006,
1209: \mnras, 373, 1004
1210:
1211: \bibitem[{{Miller} {et~al.}(2006){Miller}, {Raymond}, {Fabian}, {Steeghs},
1212: {Homan}, {Reynolds}, {van der Klis}, \& {Wijnands}}]{mil06}
1213: {Miller}, J.~M., {Raymond}, J., {Fabian}, A., {Steeghs}, D., {Homan}, J.,
1214: {Reynolds}, C., {van der Klis}, M., \& {Wijnands}, R. 2006, \nat, 441, 953
1215:
1216: \bibitem[{{Mitsuda} {et~al.}(1984)}]{mit84}
1217: {Mitsuda}, K., {et~al.} 1984, \pasj, 36, 741
1218:
1219: \bibitem[{{Narayan} \& {Yi}(1995)}]{nay95}
1220: {Narayan}, R., \& {Yi}, I. 1995, \apj, 452, 710
1221:
1222: \bibitem[{{Nowak}(1995)}]{now95}
1223: {Nowak}, M.~A. 1995, \pasp, 107, 1207
1224:
1225: \bibitem[{{Pessah} {et~al.}(2007){Pessah}, {Chan}, \& {Psaltis}}]{pcp07}
1226: {Pessah}, M.~E., {Chan}, C.-k., \& {Psaltis}, D. 2007, \apjl, 668, L51
1227:
1228: \bibitem[{{Proga} \& {Kallman}(2002)}]{pak02}
1229: {Proga}, D., \& {Kallman}, T.~R. 2002, \apj, 565, 455
1230:
1231: \bibitem[{{Remillard} \& {McClintock}(2006)}]{ram06}
1232: {Remillard}, R.~A., \& {McClintock}, J.~E. 2006, \araa, 44, 49
1233:
1234: \bibitem[{{Reynolds} \& {Fabian}(2008)}]{raf08}
1235: {Reynolds}, C.~S., \& {Fabian}, A.~C. 2008, \apj, 675, 1048
1236:
1237: \bibitem[{{Riffert} \& {Herold}(1995)}]{rah95}
1238: {Riffert}, H., \& {Herold}, H. 1995, \apj, 450, 508
1239:
1240: \bibitem[{{R{\'o}{\.z}a{\'n}ska} \& {Czerny}(2000)}]{rac00b}
1241: {R{\'o}{\.z}a{\'n}ska}, A., \& {Czerny}, B. 2000, \aap, 360, 1170
1242:
1243: \bibitem[{{Rykoff} {et~al.}(2007){Rykoff}, {Miller}, {Steeghs}, \&
1244: {Torres}}]{ryk07}
1245: {Rykoff}, E.~S., {Miller}, J.~M., {Steeghs}, D., \& {Torres}, M.~A.~P. 2007,
1246: \apj, 666, 1129
1247:
1248: \bibitem[{{Sakimoto} \& {Coroniti}(1981)}]{sac81}
1249: {Sakimoto}, P.~J., \& {Coroniti}, F.~V. 1981, \apj, 247, 19
1250:
1251: \bibitem[{{Sala} {et~al.}(2007){Sala}, {Greiner}, {Vink}, {Haberl},
1252: {Kendziorra}, \& {Zhang}}]{sal07}
1253: {Sala}, G., {Greiner}, J., {Vink}, J., {Haberl}, F., {Kendziorra}, E., \&
1254: {Zhang}, X.~L. 2007, \aap, 461, 1049
1255:
1256: \bibitem[{{Shafee} {et~al.}(2006){Shafee}, {McClintock}, {Narayan}, {Davis},
1257: {Li}, \& {Remillard}}]{sha06}
1258: {Shafee}, R., {McClintock}, J.~E., {Narayan}, R., {Davis}, S.~W., {Li}, L.-X.,
1259: \& {Remillard}, R.~A. 2006, \apjl, 636, L113
1260:
1261: \bibitem[{{Shafee} {et~al.}(2008){Shafee}, {Narayan}, \& {McClintock}}]{snm08}
1262: {Shafee}, R., {Narayan}, R., \& {McClintock}, J.~E. 2008, \apj, 676, 549
1263:
1264: \bibitem[{{Shakura} \& {Sunyaev}(1973)}]{sas73}
1265: {Shakura}, N.~I., \& {Sunyaev}, R.~A. 1973, \aap, 24, 337
1266:
1267: \bibitem[{{Shakura} \& {Sunyaev}(1976)}]{sas76}
1268: ---. 1976, \mnras, 175, 613
1269:
1270: \bibitem[{{Shimura} \& {Takahara}(1995)}]{sat95}
1271: {Shimura}, T., \& {Takahara}, F. 1995, \apj, 445, 780
1272:
1273: \bibitem[{{Stella} \& {Rosner}(1984)}]{sar84}
1274: {Stella}, L., \& {Rosner}, R. 1984, \apj, 277, 312
1275:
1276: \bibitem[{{Taam} \& {Lin}(1984)}]{tal84}
1277: {Taam}, R.~E., \& {Lin}, D.~N.~C. 1984, \apj, 287, 761
1278:
1279: \bibitem[{{Turner}(2004)}]{tur04}
1280: {Turner}, N.~J. 2004, \apjl, 605, L45
1281:
1282: \bibitem[{{Watarai} {et~al.}(2001){Watarai}, {Mizuno}, \& {Mineshige}}]{wmm01}
1283: {Watarai}, K.-Y., {Mizuno}, T., \& {Mineshige}, S. 2001, \apjl, 549, L77
1284:
1285: \bibitem[{{Watarai} {et~al.}(2005){Watarai}, {Takahashi}, \& {Fukue}}]{wtf05}
1286: {Watarai}, K.-Y., {Takahashi}, R., \& {Fukue}, J. 2005, \pasj, 57, 827
1287:
1288: \bibitem[{{Woods} {et~al.}(1996){Woods}, {Klein}, {Castor}, {McKee}, \&
1289: {Bell}}]{woo96}
1290: {Woods}, D.~T., {Klein}, R.~I., {Castor}, J.~I., {McKee}, C.~F., \& {Bell},
1291: J.~B. 1996, \apj, 461, 767
1292:
1293: \bibitem[{{Zhang} {et~al.}(1997){Zhang}, {Cui}, \& {Chen}}]{zcc97}
1294: {Zhang}, S.~N., {Cui}, W., \& {Chen}, W. 1997, \apjl, 482, L155+
1295:
1296: \end{thebibliography}
1297:
1298: \end{document}
1299:
1300:
1301:
1302: