0801.4613/ms.tex
1: \documentclass{pasj00}
2: \draft
3: 
4: \begin{document}
5: \SetRunningHead{Kawanaka et al.}{X-ray emissions from MHD accretion flows}
6: 
7: \title{X-ray Emissions from Three-dimensional Magnetohydrodynamic Coronal
8:  Accretion Flows}
9: \author{Norita \textsc{Kawanaka},\altaffilmark{1}
10:  Yoshiaki \textsc{Kato},\altaffilmark{2}
11:  and Shin \textsc{Mineshige}\altaffilmark{1}}
12: \altaffiltext{1}{Yukawa Institute for Theoretical Physics, Kyoto University,
13: Kyoto 606-8502, Japan}
14: \altaffiltext{2}{Center for Computational Sciences,
15:  University of Tsukuba 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8511, Japan}
16: \email{norita@yukawa.kyoto-u.ac.jp}
17: \KeyWords{accretion, accretion disks --- black hole physics ---
18:  radiative transfer -- X-rays: general}
19: 
20: \maketitle
21: \begin{abstract}
22: We calculate the radiation spectrum and its time variability of the black
23:  hole accretion disk-corona system
24:  based on the three-dimensional magnetohydrodynamic simulation. 
25:  In explaining the spectral properties of active galactic nuclei (AGNs),
26:  it is often assumed
27:  that they consist of a geometrically thin, optically thick disk and hot,
28:  optically thin corona surrounding the thin disk.
29:  As for a model of the corona, we adopt the simulation data of
30:  three-dimensional, non-radiative MHD accretion flows 
31:  calculated by Kato and coworkers, while for a thin disk we assume
32:  a standard type disk.  We perform
33:  Monte Carlo radiative transfer simulations in the corona, taking into
34:  account the Compton scattering of soft photons 
35:  from the thin disk by hot thermal electrons and coronal irradiation heating
36:  of the thin disk, which emits blackbody radiation.
37:   By adjusting the density parameter of the MHD coronal flow, 
38:  we can produce the emergent spectra which are consistent with those of
39:  typical Seyfert galaxies.  Moreover, we find 
40:  rapid time variability in X-ray emission spectra, originating from the
41:  density fluctuation produced by the magnetorotational instability
42:  in the MHD corona.  
43:  The features of reflection component including iron fluorescent line
44:  emission are also briefly discussed.
45: \end{abstract}
46: 
47: \section{Introduction}
48: Thanks to the rapid progress in the observational studies in recent years,
49:  our understanding of the radiation properties of black hole accretion flows
50:  have been
51:  really deepened.  Accreting black holes, such as active galactic nuclei
52:  (AGNs) and black hole binaries (BHBs) during their very-high
53:  spectral state [state with luminosities around a few
54:  tenths of the Eddington limit ($L_{\rm Edd}$)], show the radiation spectra
55:  dominated by two components; the thermal bump in UV/soft X-ray
56:  band and the power-law emission with a spectral index of $\alpha\sim 1$
57:  in the X-ray band (possibly with a high energy cutoff around MeV; for
58:  the multiwavelength spectrum of a typical AGN, see Reynolds et al. 1997).  
59:  These components are often explained by the disk-corona model
60:  (Liang \& Price 1975; Bisnovatyi-Kogan \& Blinnikov 1977;
61:  Haardt \& Maraschi 1991, 1993).  In this model, the accretion flow consists
62:  of geometrically thin, optically thick accretion disk whose structure is
63:  studied by Shakura \& Sunyaev (1973), and hot, optically thin corona
64:  surrounding the disk.  The thermal bump is believed as the thermal emission
65:  from the optically thick disk (see Kishimoto et al. 2005 for its
66:  observational implication), and the power-law component is interpreted
67:  to be formed by photons which are emitted from
68:  the disk and Compton up-scattered by hot electrons in the corona
69:  (see Thorne \& Price 1975; Shapiro et al. 1976 but in the context of the
70:  inner hot accretion flow model).  Such a structure that the hot gas
71:  (i.e. the corona) coexists with the cool gas
72:  (i.e. the disk) is justified by the existence of the reflection component
73:  observed in the X-ray spectra  (Pounds et al. 1990; see also
74:  Guilbert \& Rees 1988; Lightman and White 1988).  
75:  This reflection component is accompanied by the iron fluorescent line
76:  emission broadened relativistically (for reviews, see Mushotzky et al. 1993;
77:  Fabian et al. 2000; Reynolds \& Nowak 2003).
78: 
79: On the other hand, our theoretical understanding of black hole accretion has
80:  also made a rapid progress.  Now the dynamics of accretion flows is
81:  understood in terms of magnetohydrodynamics (MHD), since Balbus \& Hawley
82:  (1991) rediscovered the magnetorotational instability (MRI) as the
83:  fundamental mechanism of angular momentum transfer in accretion disks.
84:   Detailed dynamical features of MHD accretion flows have been investigated
85:  via global three-dimensional numerical simulations (Matsumoto 1999;
86:  Stone \& Pringle 2001; Machida et al. 2001; Hawley \& Krolik 2001, 2002;
87:  Machida \& Matsumoto 2003; Armitage \& Reynolds 2003;
88:  De Villiers et al. 2003; Gammie et al. 2003; Igumenshchev et al. 2003).
89:  Such magnetically dominated accretion flows as those simulated in these
90:  calculations are radiatively inefficient accretion flows (RIAFs) 
91:  whose mass accretion rates are much smaller than the critical value of
92:  $\sim L_{\rm Edd}$ (see, e.g., Mineshige et al. 1995).  
93:  In such flows, the dissipated energy would not be radiated away
94:  efficiently, because of their low density, and be advected inward to the
95:  central black hole (Ichimaru 1977; Narayan \& Yi 1994;
96:  Abramowicz et al. 1995; Kato et al. 1998).  
97: 
98: Although the detailed structures and behavior of magnetohydrodynamical
99:  accretion flows are shown by numerical methods, it is still unclear
100:  that if these simulational results can be applied in realistic
101:  situations in the universe.  In recent years, the researches concerning
102:  the observational properties like the spectra and their time
103:  variabilities expected from numerically simulated accretion flows have
104:  been extensively performed.  Since the RIAF model is believed to fit the
105:  emergent spectrum of Sagittarius ${\rm A}^*$, some authors have
106:  calculated the time dependent radiation spectra predicted from simulated
107:  MHD accretion flows with the aim to reproduce the observed behavior of
108:  Sgr ${\rm A}^*$ (Hawley \& Balbus 2002; Goldstone et al. 2005;
109:  Ohsuga et al. 2005; Moscibrodzka et al. 2007).  However, as for the
110:  accretion flows with moderately high mass accretion rate, which are
111:  considered to be a good model for AGNs and BHBs showing the thermal bump
112:  and the power-law emission in their spectra, no attempts have been made
113:  so far to calculate the spectra predicted from MHD simulations taking
114:  into account realistic radiation processes.  As noted in the above, in
115:  order to reproduce such spectra, the simulated accretion flows should
116:  have two components: geometrically thin cool disk and optically thin
117:  hot corona.  However, no simulation data which incorporate such
118:  two-component effects are available.
119: 
120: In this study we calculate for the first time the emergent spectra of
121:  two-component accretion flows based on the three-dimensional MHD
122:  simulation by Kato et al. (2004, hereafter KMS04).  Instead of fully
123:  solving the dynamics of two-component accretion flows, we adopt the
124:  simulation result of RIAF-like
125:  MHD accretion flow for the optically thin corona, and assume that
126:  the optically thick, geometrically thin disk is embedded in the corona
127:  with mutual interaction through radiation.  Other interactions,
128:  such as mass evaporation/condensation or heat conduction, are
129:  neglected, for simplicity (Meyer \& Meyer-Hofmeister 1994; Meyer
130:  et al. 2000; Liu et al. 2002, 2007).  The disk is emitting soft
131:  photons with thermal spectrum, and those photons are up-scattered
132:  by hot electrons in the corona.  After being scattered in the corona,
133:  a part of upscattered photons return to the disk and are
134:  thermalized there, thereby heating the accretion disk.  The disk
135:  emission is, hence, enhanced by this returning process.
136:  In this paper we perform three-dimensional Monte Carlo radiative
137:  transfer simulations to properly calculate  such radiation processes
138:  and the emergent spectra.
139: 
140: The plan of this paper is as follows.  We show the detail of the model
141:  and method used in our calculation in \S 2.  The results are presented in
142:  \S 3, and compare them with the spectra observed in typical AGNs
143:  and discuss the similar points and discrepancies between them in \S 4.
144:   In \S 5 we summarize our study.
145: 
146: \section{MODEL AND CALCULATION METHODS}
147: \subsection{Overview of Adopted MHD Simulations}
148: KMS04 investigated the evolution of a torus threaded by weak localized
149:  poloidal magnetic fields by performing the three-dimensional MHD
150:  simulation.  They solved the following basic equations of the resistive
151:  MHD in the cylindrical coordinates, $(r,~\phi ,~z)$:
152: \begin{eqnarray}
153: \frac{\partial \rho}{\partial t}+\mbox{\boldmath $\nabla$}\cdot
154:  (\rho \mbox{\boldmath $v$})=0, \\
155: \frac{\partial}{\partial t}(\rho \mbox{\boldmath $v$})+
156: \mbox{\boldmath $\nabla$}\cdot \left(\rho \mbox{\boldmath $v$}\otimes
157:  \mbox{\boldmath $v$}
158:  -\frac{\mbox{\boldmath $B$}\otimes \mbox{\boldmath $B$}}{4\pi} \right)
159:  = -\mbox{\boldmath $\nabla$}\left(p_{\rm gas}+\frac{B^2}{8\pi}\right)
160: -\rho\mbox{\boldmath $\nabla$}\psi ,\\
161: \frac{\partial}{\partial t}\left(\varepsilon +\frac{B^2}{8\pi}\right) 
162: +\mbox{\boldmath $\nabla$}\cdot \left[\left(\varepsilon 
163: +p_{\rm gas}\right)\mbox{\boldmath $v$}
164:  +\frac{c(\mbox{\boldmath $E$}\times \mbox{\boldmath $B$})}{4\pi}\right]
165: =-\rho\mbox{\boldmath $v$}\cdot \mbox{\boldmath $\nabla$}\psi, \\
166: \frac{\partial B}{\partial t}=-c\mbox{\boldmath $\nabla$}\times
167:  \mbox{\boldmath $E$},
168: \end{eqnarray}
169: where $\psi=-GM/(R-r_{\rm S})$ is the pseudo-Newtonian potential
170:  (Paczy\'{n}ski \& Wiita 1980), $\varepsilon =\rho v^2/2+p_{\rm gas}/
171: (\gamma-1)$ is the energy of the gas (here $\gamma$ is fixed to be 5/3)
172:  and $\mbox{\boldmath $E$}=
173: -(\mbox{\boldmath $v$}/c)\times \mbox{\boldmath $B$}+(4\pi\eta/c^2)
174: \mbox{\boldmath $J$}$ is Ohm's law.  Here $R~[\equiv (r^2+z^2)^{1/2}$
175:  is the distance
176:  from the origin,$r_{\rm S}~(\equiv 2GM/c^2)$ is the Schwarzschild
177:  radius (with $M$ and $c$ being the mass of a BH and the speed of
178:  light, respectively),
179:  and $\mbox{\boldmath $J$}=(c/4\pi)
180: \mbox{\boldmath $\nabla$}\times \mbox{\boldmath $B$}$ is the electric
181:  current.  As to $\eta$ we adopt the anomalous resistivity model which
182:  is used in many
183:  solar flare simulations (for the detail, see Yokoyama \& Shibata 1994).
184:   The calculation is started with a rotating torus in hydrostatic balance
185:  located around $r=r_0=40r_{\rm S}$.  The initial magnetic fields are
186:  confined within a torus and purely poloidal (see KMS04 for the detail).
187:   In the simulation
188:  they used $300\times 32\times 400$ nonuniform mesh points.  The grid
189:  spacing is uniform $(\Delta r=\Delta z=0.16r_{\rm S})$ within the inner
190:  calculation box of
191:  $0\leq z\leq 10r_{\rm S}$, and it increases by 1.5\% from one mesh to
192:  the adjacent outer mesh outside this box up to $r\leq 20r_{\rm S}$ and
193:  $z\leq 20r_{\rm S}$, and it
194:  increases by 3\% beyond that.  The entire computational box size is
195:  $0\leq r\leq 200r_{\rm S}$, $0\leq \phi \leq 2\pi$,
196:  and $-50r_{\rm S}\leq z\leq 50r_{\rm S}$, and they simulated a full
197:  $360^\circ $ domain (see Kato 2004 for more detail).
198: 
199:  The simulated MHD flow is slightly oscillating because of the turbulence
200:  driven by MRI, and geometrically thick density distribution is produced.
201:   In this quasi-steady accretion flow,
202:  the density profile is $\rho \propto r$ in the inner part
203:  ($r<20r_{\rm S}$), while $\rho \propto r^{-1}$ in the outer
204:  part ($r>20r_{\rm S}$) (see Fig. 4 in KMS04).  We use this quasi-steady
205:  density distribution as well as the ion temperature distribution in
206:  modeling the corona in which hot thermal electrons up-scatter the soft
207:  photons emerging from the cold disk virtually located in the equatorial
208:  plane.
209: 
210: \subsection{Physical Quantities of the Disk and Corona} 
211: As mentioned in \S 1 briefly, we adopt the following assumptions in
212:  constructing the disk-corona model.
213: 
214: 1. In the equatorial plane, we assume a standard accretion disk
215:  (Shakura \& Sunyaev 1973) with an infinitesimal height for a given
216:  mass accretion rate $\dot{M}_{\rm disk}$.
217: 
218: 2. The RIAF-like accretion flow, whose structure is provided by the
219:  three-dimensional MHD simulation (KMS04), is surrounding the standard
220:  thin disk
221:  as the hot corona.  That is, we assume that coronal flow is created at
222:  large radius by evaporation of the disk material and moves freely
223:  inward
224: \footnote{Note that evaporation dominates over condensation at large
225:  radii, whereas the opposite is the case at small radii
226:  (e.g. Liu et al. 2007)}.
227:   Mass evaporation and condensation occurring between them and heat
228:  conduction from the corona to the disk are neglected in the calculation.
229:   Radiative
230:  cooling of coronal plasma is also neglected.
231: 
232: 3. The photons radiated from the thin disk are partly up-scattered by
233:  hot electrons in the corona and the remaining portion penetrate
234:  through the
235:  corona, creating a power-law hard emission in the spectrum. 
236: 
237: 4. As the seed photon field from the underlying disk, we include the
238:  reprocessed radiation, which comes from the coronal irradiation
239:  on the disk, as
240:  well as the intrinsic disk radiation.
241: 
242: We take three-dimensional data of density and proton temperature
243:  distributions in the accretion flows calculated in Kato
244:  (2004). This has the same initial condition with Model B in KMS04.  The data
245:  of physical properties
246:  are given at each point in the simulation box associated with Cartesian
247:  coordinates $(x,~y,~z)$, in which the black hole is located at the
248:  origin of the
249:  coordinate axes, the $z$-axis is set to be the rotation axis of the
250:  accretion flow, and the $x$-$y$ plane corresponds to the equatorial
251:  plane.  We employ
252:  Cartesian grids with numbers $(N_x,~N_y,~N_z)=(101,~101,~101)$ of cells.
253:   The size of the calculating box is $2X\times 2Y\times 2Z$, where we set
254: $(X,~Y,~Z)=(99.9r_{\rm S},~99.9r_{\rm S},~99.9r_{\rm S})$.
255: 
256: In the MHD simulations with no radiative loss the density is given as
257:  non-dimensional number ${\tilde \rho}$ with the normalization factor
258:  $\rho_0$,
259:  which is treated a free parameter in our calculation.  Basically the
260:  coronal density is determined by evaporation of the disk gas, but here we
261:  determine $\rho_0$
262:  so that the power-law indices of the evaluated spectra agree with
263:  the observations.  The proton temperature does not depend on the
264:  density parameter and
265:  is given by the MHD simulation as
266:  $(\mu m_p c^2/k_B){\tilde c}_s^2$, where $\mu$ is the mean molecular
267:  weight $(=0.5)$, $m_p$ is the proton mass, $k_B$ is the Boltzmann constant,
268:  and ${\tilde c}_s$ is the normalized sound velocity obtained by the
269:  simulation.
270: 
271: Here we should note that ions (protons) and electrons in the plasma
272:  simulated in KMS04 have the same temperature, though electrons would be
273:  radiatively
274:  cooled and have lower temperature in the realistic situation.  Assuming
275:  that ion temperature coincides with the simulated one and that the
276:  electrons
277:  have a Maxwellian distribution, we evaluate the electron
278:  temperature, $T_e$, through the energy balance of the electrons between
279:  Coulomb collisions with ions and radiative cooling,
280: \begin{eqnarray}
281: \int_{-Z}^Z \int_{r_{\rm in}}^{r_{\rm out}} \lambda _{ie}2\pi r dr dz
282: =\int L_C(\nu;~r_{\rm in}\leq r\leq r_{\rm out})d\nu. \label{balance}
283: \end{eqnarray}
284: Here $\lambda_{ie}$ is the energy transfer rate from ions to electrons
285:  (Stepney \& Guilbert 1983) and $L_C(\nu)$ is the coronal luminosity at
286:  frequency $\nu$.  In the present study, we divide the corona into three
287:  regions (0<$r\leq 10r_{\rm S}$, $10r_{\rm S}<r\leq 30r_{\rm S}$,
288:  and $30r_{\rm S}<r$; ($r_{\rm in}$, $r_{\rm out})=(0,10r_{\rm S}),
289: ~(10r_{\rm S},30r_{\rm S})$ and $(30r_{\rm S},\infty )$) for simplicity,
290:  and suppose that the electrons in the accretion flow
291:  in each region have the temperature which is independent of the radius
292:  $r$ and the altitude $z$.  The coronal luminosity, $L_C$, is obtained
293:  by Monte Carlo simulations
294:  (see next subsection) for a guess value of $T_e$.  Since this $T_e$
295:  does not always satisfy Eq. (\ref{balance}), we should do some
296:  iterations to
297:  calculate the appropriate electron temperature and the emergent
298:  spectrum.
299: 
300: \subsection{Radiative Transfer Simulations}
301: Our model consists of a cold disk which produces blackbody radiation
302:  at each radius with temperature being determined from the standard
303:  model (Shakura \& Sunyaev 1973):
304: \begin{eqnarray}
305: T^0_{\rm disk}=\left[\frac{3GM\dot{M}_{\rm disk}}{8\pi r^3\sigma}
306: \left(1-\sqrt{\frac{r}{r_{\rm in}}}\right)\right]^{1/4}, \label{seed}
307: \end{eqnarray}
308:  (as long as the reprocessed radiation is unimportant) where $G$
309:  is the gravitational constant, $\dot{M}_{\rm disk}$ is the mass
310:  accretion rate in the disk,
311:  and $\sigma$ is the Stephan-Boltzmann constant.  Note that when
312:  irradiation flux $F_{\rm irr}$ by the corona is available, the
313:  disk emits blackbody
314:  with an enhanced temperature
315: \begin{eqnarray}
316: T_{\rm disk}=\left[\left(T_{\rm disk}^0\right)^4+F_{\rm irr}/
317: \sigma\right]^{1/4}. \label{irr}
318: \end{eqnarray}
319:   In the following calculation we set $r_{\rm in}=3r_{\rm S}$,
320:  $\dot{M}_{\rm disk}=10^{-3}\dot{M}_{\rm Edd}$ (with
321:  $\dot{M}_{\rm Edd}=10L_{\rm Edd}/c^2$)
322:  and the mass of a central black hole to be $M=10^8 M_{\odot}$,
323:  which is believed to be the typical value for AGNs.  Thus, the
324:  normalized time
325:  corresponds to ${\tilde t}\cong r_{\rm S}/c=10^3{\rm sec}$.
326: 
327: As the radiation process, we take into account both of the
328:  intrinsic disk radiation [Eq.(\ref{seed})] and the thermal
329:  reprocessing from irradiated
330:  disk as the seed photon field [see Eq. (\ref{irr})], and the
331:  Compton/inverse Compton scattering in the corona.  We neglect
332:  synchrotron emission/absorption and free-free emission/absorption.
333:   Some studies about the
334:  disk-corona model have shown that the power-law component of the
335:  emergent spectra is explained as the Comptonized emission
336:  (Haardt \& Maraschi 1991, 1993)
337:  , so for our
338:  present purpose this approximation is justified at least in the
339:  X-ray energy bands.
340:   According to the simulation by KMS04, the magnetic field in the
341:  coronal flow with $\rho_0=1.6\times 10^{-14}~{\rm g}~{\rm cm}^{-3}$
342:  is $B\sim 10^3{\rm G}$, for which
343:  the Compton scattering is the most efficient cooling mechanism.
344:   This fact justifies the method of deriving the coronal temperature
345:  described
346:  in the previous section.
347: 
348: The method of the Monte Carlo simulation is based on Pozdnyakov et
349:  al. (1977).  In order to efficiently calculate the emergent spectra,
350:  we introduce
351:  a photon weight $w$.  When emerged
352:  from the disk we set that each photon has the weight of $w_0=1$,
353:  and then we calculate
354:  the escape probability, $P_0$.  The escape probability of a photon
355:  after
356:  $i$-th scattering (for $i\geq 1$), $P_i$, is evaluated as
357: \begin{eqnarray}
358: P_i=\exp \left( -\int \left[ \frac{\rho(x_i,~y_i,~z_i)}{m_p}\right]
359: \sigma_{\rm KN}(x_i,~y_i,~z_i) dl\right),
360: \end{eqnarray}
361: where $(x,~y,~z)=(x_0,~y_0,~0)$ corresponds to the point on the
362:  equatorial plane
363:  (i.e. the disk plane), in which the thermal soft photons are
364:  generated,
365:  $(x_i~,y_i,~z_i)$ is the point where a photon is subject to the
366:  $i$-th
367:  scattering, $m_p$ is the proton mass, $\sigma_{\rm KN}$ is the
368:  Klein-Nishina cross section (Rybicki \& Lightman 1979), and the
369:  integral of $dl$ should be
370:  done along the photon direction there from the point
371:  $(x_i,~y_i,~z_i)$ to the boundary of the calculating
372:  box.  The quantity of $w_0P_0$ represents the transmitted portion
373:  of photons
374:  and is recorded to calculate the penetrated spectrum if the path
375:  of the photon
376:  does not cross with the equatorial plane, and will no longer
377:  continue to be
378:  counted and will be regarded to be absorbed by the disk.  As for
379:  the remaining
380:  portion of a photon, its weight becomes $w_1=w_0(1-P_0)$.
381:   The transmitted portion
382:  of photons after $i$-th scattering, $w_i P_i$ is recorded to
383:  calculate the emergent
384:  spectrum, and the remaining portion, $w_i(1-P_i)$, undergoes the
385:  $(i+1)$-th scattering.
386:   This calculation is continued until the weight $w_i$ becomes
387:  sufficiently small
388:  ($w_i\ll 1$) or the path of the remaining photon crosses the
389:  equatorial plane, regarding
390:  to be absorbed by the disk.  The whole process is simulated by the
391:  Monte Carlo method.
392:   Finally, we suppose that the inner boundary of the underlying
393:  standard disk is $r=3r_{\rm S}$,
394:  and general relativistic effects like the light bending and
395:  energy shift are neglected in
396:  our study.
397: 
398: \section{Results}
399: \subsection{Spectral Features}
400: First, we show the emergent spectra from the accretion disk with
401:  MHD coronal flow with various
402:  density parameters for the corona in Fig. 1.  In this calculation,
403:  the mass accretion rate of the underlying disk is set to be
404:  $\dot{M}_{\rm disk}=10^{-3}M_{\rm Edd}$.  This value is relatively
405:  lower than that assumed in the standard picture
406:  ($\dot{M}_{\rm disk}\sim 0.1-0.01M_{\rm Edd}$).  However,
407:  according to the idea of Haardt \& Maraschi (1991), the liberated
408:  gravitational energy accompanying the mass accretion $L_G$ should
409:  be mostly dissipated in
410:  the corona ($fL_G$, where $f\simeq 1$), while only a small fraction
411:  of the energy ($(1-f)L_G$) is dissipated in the underlying disk.
412:   Then the temperature of
413:  the intrinsic disk radiation would be reduced by a factor of
414:  $(1-f)^{1/4}$.  In our calculation the mass accretion rate of the
415:  disk only appears when
416:  giving the intrinsic disk temperature Eq. (\ref{seed}), and so the
417:  small mass accretion rate corresponds to the reduced disk temperature.
418:   Due to
419:  this fact, the relatively small mass accretion rate adopted in our
420:  calculation is justified.
421: 
422: \begin{figure*}[ht]
423: \begin{center}
424: \FigureFile(130mm,100mm){fig1.eps}
425: \end{center}	
426: \caption{Broadband spectra from a disk at the accretion
427:  rate of $10^{-3}{\dot M}_{\rm Edd}$ and a MHD coronal flow
428:  around a black hole
429:  of $10^8M_{\odot}$ with the density normalization parameters
430:  $5.1\times 10^{-15}$,
431:  $1.6\times 10^{-14}$ and
432:  $5.1\times 10^{-14}~{\rm g}~{\rm cm}^{-3}$ {\it solid lines}.  The spectra of
433:  the seed photons (including those from the cold disk and the
434:  reflection component; {\it dashed lines}) are also shown
435:  for the comparison.  The short solid lines near the spectra show the power-law fit of $\nu F_{\nu}$.}
436: \label{sedrho}
437: \end{figure*}
438: 
439:   The adopted density parameters $\rho_0$ for the corona
440:  are $\rho_0=5.1\times 10^{-15},~1.6\times 10^{-14}$, and $5.1\times
441:  10^{-14}{\rm g}~{\rm cm}^{-3}$.  These values correspond to
442:  the number density of $\sim 10^9{\rm cm}^{-3}$.
443:  Such density is consistent with some corona models (see Liu
444:  et al. 2002, 2003).  Due to the inverse Compton scattering
445:  of
446:  in which the soft thermal photons from the underlying disk in the
447:  corona, the power-law component with the spectral index of
448:  $\alpha \sim 1-2$ ($F_{\nu}\propto \nu^{-\alpha}$) appears in the
449:  higher energy band.  In the calculated spectra, we cannot find a
450:  bump-like structure 
451:  clearly in the UV/soft X-ray band which is usually seen in typical
452:  spectra of AGNs.  As $\rho_0$ increases, the total luminosities
453:  increases, while the
454:  power-law index decreases.  The corona with higher density (especially
455:  with higher scattering optical depth) would irradiate the underlying
456:  disk with
457:  higher energy flux because larger number of photons originated from the
458:  disk would gain the energy and be backscattered.  The the disk is
459:  heatened by
460:  the corona and so the energy flux of the seed photon field would
461:  increase.  This is why the total luminosity rises with the coronal
462:  density.
463: 
464: \begin{table}
465: \caption{Coronal properties}
466: \begin{center}
467: \begin{tabular}{|cccccc|}
468: \hline
469: $\rho_0$ (${\rm g}~{\rm cm}^{-3}$)&region&
470: $T_{\rm cor}~[{\rm K}]$&$\tau$&$y$&$\alpha$\\ \hline
471: $5.1\times 10^{-14}$&$0<r<10r_{\rm S}$&
472: $\sim 5.2\times 10^9$&$\sim 0.2$&$\sim 0.5$&$\sim 1.30$ \\
473: &$10r_{\rm S}<r<30r_{\rm S}$&$\sim 3.4\times 10^9$&
474: $\sim 0.9$&$\sim 3$& \\
475: &$30r_{\rm S}<r$&$\sim 7.6\times 10^8$&$\sim 0.7$&$\sim 0.5$&
476:  \\ \hline
477: $1.6\times 10^{-14}$&$0<r<10r_{\rm S}$&
478: $\sim 5.2\times 10^9$&$\sim 0.05$&$\sim 0.2$&$\sim 1.60$ \\
479: &$10r_{\rm S}<r<30r_{\rm S}$&$\sim 3.4\times 10^9$&
480: $\sim 0.25$&$\sim 0.55$& \\
481: &$30r_{\rm S}<r$&$\sim 7.5\times 10^8$&$\sim 0.2$&
482: $\sim 0.1$& \\ \hline
483: $5.1\times 10^{-15}$&$0<r<10r_{\rm S}$&
484: $\sim 5.2\times 10^9$&$\sim 0.02$&$\sim 0.05$&$\sim 2.00$ \\
485: &$10r_{\rm S}<r<30r_{\rm S}$&$\sim 3.4\times 10^9$&
486: $\sim 0.09$&$\sim 0.15$& \\
487: &$30r_{\rm S}<r$&$\sim 7.6\times 10^8$&$\sim 0.07$&
488: $\sim 0.03$& \\
489: \hline
490: \end{tabular}
491: \end{center}
492: \end{table}
493: 
494: In Table 1 we summarize the scattering optical depths $\tau$
495:  (evaluated by integrating the scattering
496:  opacity over the z-direction from the equatorial plane), and the
497:  Compton $y$ parameters of the corona (averaged in each region), and the spectral indices of the
498:  power law component estimated from the calculation results for 3
499:  density parameters.  The MHD simulation of coronal flow was
500:  started from the initial condition of a magnetized torus located
501:  around at
502:  $r\sim 40r_S$.  The radius of maximum gas density
503:  (and of maximum $\tau$) decreases inward with time and stays around
504:  $r\sim 20r_S$.
505:   According to the theory of unsaturated inverse Compton scattering,
506:  the power-law index of the Comptonized emission component depends on
507:  the plasma
508:  density and temperature through this equation:
509: \begin{eqnarray}
510: \alpha=-\frac{3}{2}+\sqrt{\frac{9}{4}+\frac{4}{y}}. \label{ypara}
511: \end{eqnarray}
512: Here $y\equiv (4k_B T/m_e c^2)\tau$ is the Compton $y$-parameter
513:  (Rybicki \& Lightman 1973), where $T$ and $\tau$ are the temperature
514:  and
515:  Thompson optical depth of the corona, respectively.  The spectral
516:  indices derived from this equation using $y$-parameters in Table 1 do
517:  not always agree with those estimated from the spectra.  This is
518:  because $\tau$ and $T_{\rm cor}$ have spatial distributions and we
519:  cannot evaluate
520:  $y$-parameter of the corona uniquely.  Even with such a situation,
521:  however, we can see the tendency that the spectra get flatter with
522:  higher
523:  coronal density (and then higher $y$-parameter), which is consistent
524:  with the theory above.
525: 
526:   In our calculation, $\rho_0$ is determined so as to reproduce the
527:  observation.
528:   By tuning the coronal density parameter, we can reproduce the
529:  luminosity
530:  of the power-law component which is as intense as that of the thermal
531:  component originated from the optically thick disk.  Such
532:  feature is typical in Seyfert galaxies.
533:   However, the coronal temperature cannot be chosen freely but should
534:  be determined by imposing the energy balance of the electrons
535:  between Coulomb collisions and the cooling via inverse Compton
536:  scattering, as we have done.  Nevertheless the resulting coronal
537:  temperature
538:  is substantially reduced from the plasma temperature derived by the
539:  simulation ($\sim 10^{13}~{\rm K}$) to $\sim 10^9{\rm K}$, which makes
540:  the high energy cutoff of computed spectra consistent with
541:  observations.
542: 
543: \subsection{Time variation}
544: The spectral variation caused by the time variation of MHD coronal flow
545:  structure is shown in Fig. 2.  In the highest energy range
546:  ($\gtrsim 10^{18}{\rm Hz}$) the spectra show fluctuations because of
547:  poor photon statistics.  As for the soft X-ray band
548:  (with ${\rm log}\nu\simeq 17-18$)
549:  where the spectra show a smooth power-law shape, the spectral index
550:  slightly changes with time, and then the X-ray flux fluctuates a little
551:  (see also Fig. 3).  According to the MHD simulation on which
552:  our radiative transfer calculations are based, the three-dimensional
553:  structure of the coronal accretion flow is fluctuating everywhere in
554:  each timestep.
555:   On the other hand, the spectral index depends on the distribution of
556:  $y$-parameter of the corona, as we note in the last subsection.  So we
557:  can conclude
558:  that the fluctuations of the spectral indices of the computed spectra
559:  in Fig. 2 reflect the fluctuation of $y$, which comes from the density
560:  fluctuations (which is supposed to be due to MRI) in the coronal flow.
561: 
562: \begin{figure*}[ht]
563: \begin{center}
564: \FigureFile(130mm,100mm){fig2.eps}
565: \end{center}
566: \caption{Spectral variation of the Comptonized emission
567:  predicted from the standard disk with a MHD coronal flow
568:  around a black hole of
569:  $10^8M_{\odot}$.  Here we set the density parameter as
570:  $1.6\times 10^{-14}{\rm g}~{\rm cm}^{-3}$.}
571: \label{sedtime}
572: \end{figure*}
573: 
574:   Fig. 3. shows the X-ray lightcurve derived from our simulations.
575:   From this plot we can see that the X-ray luminosity from our
576:  disk-corona
577:  system can change by factors of a few tens of percent on timescales of
578:  the orbital period at the last stable orbit ($r=3r_{\rm S}$),
579:  i.e. about $10^3(M/10^8) M_{\odot}$ sec.  In the whole calculation
580:  we do not vary the properties of the soft photon source (i.e. the
581:  underlying cold disk).  This variation which we obtained
582:  is due to the density fluctuation (and accompanying temperature
583:  fluctuation) of the coronal flow.
584: 
585: \begin{figure*}[ht]
586: \begin{center}
587: \FigureFile(130mm,100mm){fig3.eps}
588: \end{center}
589: \caption{Time variation of the X-ray luminosity (1-6keV).
590:   Here we set the density parameter as
591:  $1.6\times 10^{-14}{\rm g}~{\rm cm}^{-3}$.}
592: \label{xlc14}
593: \end{figure*}
594: 
595: \section{Discussion and Conclusion}
596: We have calculated the emergent spectra and their time variabilities
597:  predicted based on the disk-corona model, in which
598:  a cold standard disk at the equatorial plane is sandwiched by a hot
599:  coronal flow. As for the structure and
600:  dynamics of the corona we use the three-dimensional MHD simulation data
601:  by Kato (2004).  In this section we discuss our results
602:  especially in the context of AGNs.
603: 
604: We have shown that our black hole disk-corona system can reproduce the
605:  power-law X-ray emission with the photon index $\alpha\sim 1-2$ by
606:  adjusting the
607:  density parameter properly.  The power-law indices and the cutoff
608:  energy scales of the spectra are roughly
609:  in agreement with the observed spectra of Seyfert galaxies.  Moreover,
610:  we find significant variability of the power-law X-ray emission.
611:   The power-law X-ray emission flux predicted from our model changes
612:  by a few tens of percent on timescales of the orbital period
613:  near the last stable orbit, which is about
614:  $10^3(M/10^8M_{\odot})~{\rm sec}$,
615:  while the power-law index nor the cutoff energy do not change
616:  considerably. This variability comes purely from the fluctuation
617:  of the coronal flow around $r\sim 20r_S$,
618:  where the scattering optical depth of the coronal flow
619:  attains its largest value in its structure.  This fluctuation is
620:  driven by the turbulence as a result of MRI, and its amplitude is large
621:  enough to explain the observed X-ray variability in Seyfert galaxies
622:  (e.g. Miniutti et al. 2007).  In UV/soft X-ray band, however, we cannot
623:  see the bump-like structure which is often found in the spectra of AGNs.
624:   The same problem occurred in other disk-corona models
625:  (Shimura et al. 1995; Liu et al. 2003).
626:   This bump is supposed to be the thermal radiation coming directly from
627:  the optically thick disk,
628:  while in those disk-corona models the disk is wholly covered with corona.
629:   In order to reproduce the
630:  UV/soft X-ray bump with a model, the coronal structure may need to be
631:  patchy or locally concentrated.  In the present study we do
632:  not consider the viewing angle dependence of the spectra.  If we
633:  observe this accretion flow system face-on, then the scattering
634:  optical depth along the line of sight would be smaller than in the
635:  case observed with non-zero viewing angle and the optically thick
636:  component should be observed more clearly.
637:  
638:  
639: Iron K$\alpha$ fluorescent line emissions would occur when the disk is
640:  irradiated by X-ray from the corona.  The profile of this line emission
641:  would
642:  be broadened due to relativistic effects (see Reynolds \& Nowak 2003 for
643:  a review).  The detailed structure of this line profile is determined by
644:  the line emissivity distribution on the disk, which depends on the
645:  spectra of local X-ray irradiation from the corona.
646: 
647: Recently {\it Suzaku}
648:  has observed the broadened iron line profile from MCG-6-30-15 in detail
649:  and they implied that the emissivity profile should be as steep as
650:  $F_{\rm line}(r)\propto r^{-4.4}$ (Miniutti et al. 2007).  This is not
651:  easy to understand, since standard disks are known to produce
652:  continuum flux $F_{\rm cont}(r)\propto r^{-3}$.  Some authors proposed
653:  that this is a manifestation of Kerr black hole effects
654:  (e.g. Wilms et al. 2001).
655:   Such a steep emissivity profile can
656:  be reproduced by thermal Comptonization in the corona, if the
657:  irradiating corona has the temperature and density profiles which are
658:  arising inward, as was pointed out by Kawanaka et al. (2005). This is
659:  because the fraction
660:  of high energy photons increases inward so that the number of
661:  irradiating photons which are capable of producing iron fluorescent
662:  line emissions should increase inward more rapidly than $r^{-3}$.
663:   As for the MHD corona model which we adopted in this study, however,
664:  the temperature and density
665: have flatter profiles (see KMS04), so such a steep line emissivity
666:  profile is not expected as long as we consider only Compton
667:  upscattering to be the X-ray
668:  emission process in this corona.
669: 
670: One of the way of producing such a
671:  steep emissivity profile so as to produces the iron line profile as
672:  is observed is to introduce moderate radiative cooling in the
673:  simulation (Mineshige et al. 2002).
674:   Radiative cooling can be efficient in dense region which is
675:  generally located in the inner part of the disk. Then the density
676:  profile near the
677:  event horizon of the coronal flow can be steeper and Compton
678:  up-scattering would be more efficient. As a result, hard X-ray
679:  photons irradiating the disk would increase and iron fluorescence
680:  line photons would become
681:  more efficient. Another possibility is that the transient X-ray
682:  emission accompanying with magnetic reconnection flares in the
683:  localized region of the corona would contribute to the iron line
684:  emissions from the disk,
685:  which is not taken into account in our present calculation. Such
686:  small-scale X-ray sources are often assumed when explaining the
687:  constant reflection component including iron line emissions
688:  (Miniutti \& Fabian 2004; Malzac et al. 2006; Nayakshin 2007).
689:   If we analyze the local
690:  electron heating (or nonthermal acceleration) around magnetic
691:  reconnection flares and hard X-ray emission processes in the MHD
692:  coronal flow simulation,
693:  then we can know the local X-ray irradiation onto the disk, which
694:  may lead us to understanding the observed broad iron line profiles
695:  and their time variabilities from the microphysics of X-ray
696:  emission processes.
697: 
698: We can check the consistency of the assumption that the MHD coronal
699:  flow which we adopt in the calculation is cooled dominantly by
700:  advection, by comparing the heating rate ($\sim GMm_p/(2r)\Omega$)
701:  and the cooling rate ($\sim$Comptonization luminosity per electron).
702:   According to the result of our calculation, the innermost region
703:  ($0<r\lesssim 20r_{\rm S}$) is not cooled efficiently by inverse Compton
704:  scattering.  However, in the outer region ($r\gtrsim 20r_{\rm S}$)
705:  local heating rate and cooling rate are comparable, which means that
706:  the assumption of an advection-dominated flow is only marginally justified.
707:   Generally when Compton cooling becomes efficient the coronal
708:  temperature will become lower, and then Compton upscattering will be
709:  inefficient.  This means that the power-law X-ray emission would not
710:  be generated so much from that region.
711:   As far as we consider the spectrum and energy flux in X-ray and
712:  higher energy band, to which the photons from the innermost region
713:  mainly contribute, this inconsistency would hardly affect the results. 
714: 
715: Finally, we should mention the interaction between the corona and
716:  the underlying disk.  In the transition zone between the hot corona
717:  and the
718:  cold disk where density and temperature abruptly change, the heat
719:  conduction and
720:  the mass evaporation would be important as the mass and energy
721:  exchanging processes (Meyer \& Meyer-Hofmeister 1994;
722:  Liu et al. 2002).  Evaporation of photospheric material was
723:  actually shown to be essential in the context
724:  of solar flares (see e.g. Yokoyama \& Shibata 1994).  By including
725:  these effects in the simulation we will be able to obtain
726:  a more realistic model of the coronal structure and dynamics.  To
727:  what extent the disk-corona structure can extend
728:  to the inner region is important when we consider the relativistic
729:  skewing of the iron line emission profile because most of the
730:  observations of iron line profiles imply that the iron fluorescence
731:  line photons are
732:  emitted from around/inside the last stable orbit.  The detailed
733:  analysis
734:  of such disk-corona interactions and their observational
735:  implications are left as a future work. 
736: 
737: \bigskip
738: We would like to thank Masayuki Umemura, Toshihiro Kawaguchi
739:  and Ken Ohsuga for helpful comments and discussions.  We are also
740:  grateful to an anonymous referee for his/her valuable comments, which
741:  helped us to improve the manuscript in a great deal.
742:   The numerical calculations were carried out on Altix3700 BX2
743:  at YITP in Kyoto University.  This work is supported in part by
744:  Research Fellowship of the Japan Society for the Promotion of
745:  Science for Young Scientists (N. K.).
746: 
747: \begin{thebibliography}{}
748: \bibitem[Abramowicz et al. (1995)]{}
749: Abramowicz, M. A., Chen, X., Kato, S., Lasota, J.-P \&
750:  Regev, O. 1995, ApJ, 438, L37
751: \bibitem[Armitage \& Reynolds (2003)]{}
752: Armitage, P. J. \& Reynolds, C. S. 2003, MNRAS, 341, 1041
753: \bibitem[Balbus \& Hawley (1991)]{}
754: Balbus, S. A. \& Hawley, J. F. 1991, ApJ, 376, 214
755: \bibitem[Bisnovatyi-Kogan \& Blinnikov (1977)]{}
756: Bisnovatyi-Kogan, G. S. \& Blinnikov, S. I. 1977, A\&A, 59, 111
757: \bibitem[De Villers et al. (2003)]{}
758: De Villers, J. -P., Hawley, J. F. \& Krolik, J. H. 2003, ApJ,
759:  599, 1238
760: \bibitem[Fabian et al. (2000)]{}
761: Fabian, A. C., Iwasawa, K., Reynolds, C. S., Young, A. J., 2000,
762:  PASP, 112, 1145
763: \bibitem[Gammie et al. (2003)]{}
764: Gammie, C. F., McKinney, J. C. \& Toth, G. 2003, ApJ, 589, 444
765: \bibitem[Goldston et al. (2005)]{}
766: Goldston, J. E., Quataert, E. \& Igumenshchev, I. V. 2005, ApJ,
767:  621, 785
768: \bibitem[Guilbert \& Rees (1988)]{}
769: Guilbert, P. W. \& Rees, M. J. 1988, MNRAS, 233, 475
770: \bibitem[Haardt \& Maraschi (1991)]{}
771: Haardt, F. \& Maraschi, L. 1991, ApJ, 380, L51
772: \bibitem[Haardt \& Maraschi (1993)]{}
773: Haardt, F. \& Maraschi, L. 1993, ApJ, 413, 507
774: \bibitem[Hawley \& Krolik (2001)]{}
775: Hawley, J. F. \& Krolik, J. H. 2001, ApJ, 548, 348
776: \bibitem[Hawley \& Krolik (2002)]{}
777: Hawley, J. F. \& Krolik, J. H. 2002, ApJ, 566, 164
778: \bibitem[Ichimaru (1977)]{}
779: Ichimaru, S. 1977, ApJ, 214, 840
780: \bibitem[Igumenshchev et al. (2003)]{}
781: Igumenshchev, I. V., Narayan, R. \& Abramowicz, M. A. 2003,
782:  ApJ, 592, 1042
783: \bibitem[Kato et al. (1998)]{}
784: Kato, S., Fukue, J. \& Mineshige, S. 1998, Black-Hole
785:  Accretion Disks (Kyoto: Kyoto Univ. Press)
786: \bibitem[Kato (2004)]{}
787: Kato, Y. 2004, PASJ, 56, 931
788: \bibitem[Kato et al. (2004)]{}
789: Kato, Y., Mineshige, S. \& Shibata, K. 2004, ApJ, 605, 307
790:  (KMS04)
791: \bibitem[Kawanaka et al. (2005)]{}
792: Kawanaka, N., Mineshige, S. \& Iwasawa, K. 2005, ApJ, 635,
793:  167
794: \bibitem[Kishimoto et al. (2005)]{}
795: Kishimoto, M., Antonucci, R., \& Blaes, O. 2005, MNRAS,
796: 364, 640
797: \bibitem[Liang \& Price (1977)]{}
798: Liang, E. P. T \& Price, R. H. 1988, ApJ, 218, 247
799: \bibitem[Lightman \& White (1988)]{}
800: Lightman, A. P. \& White, T. R. 1988, ApJ, 335, 57
801: \bibitem[Liu et al. (2002)]{}
802: Liu, B. F., Mineshige, S., Meyer, F., Meyer-Hofmeister. E.
803:  \& Kawaguchi, T. 2002, ApJ, 575, 117
804: \bibitem[Liu et al. (2007)]{}
805: Liu, B. F., Taam, R. E., Meyer-Hofmeister, E. \& Meyer, F.
806:  2007, arXiv:0709.0143, ApJ accepted
807: \bibitem[Machida \& Matsumoto (2003)]{}
808: Machida, M. \& Matsumoto, R. 2003, ApJ, 585, 429
809: \bibitem[Machida et al. (2001)]{}
810: Machida, M., Matsumoto, R. \& Mineshige, S. 2001, PASJ,
811:  53, L1
812: \bibitem[Malzac et al. (2006)]{}
813: Malzac, J., Merloni, A. \& Suebsuwong, T. 2006, Astron.
814:  Nachr., 327, 985
815: \bibitem[Matsumoto (1999)]{}
816: Matsumoto, R. 1999, in Numerical Astrophysics, ed. S. M.
817:  Miyama, K. Tomisaka \& T. Hanawa (Boston: Kluwer), 195
818: \bibitem[Meyer et al. (2000)]{}
819: Meyer, F., Liu, B. F. \& Meyer-Hofmeister, E. 2000, A\&A,
820:  361, 175
821: \bibitem[Meyer \& Meyer-Hofmeister (1994)]{}
822: Meyer, F. \& Meyer-Hofmeister, E. 1994, A\&A, 288, 175
823: \bibitem[Mineshige et al. (2002)]{}
824: Mineshige, S., Negoro, H., Matsumoto, R., Machida, M. \&
825:  Manmoto, T. 2002, in Current High Energy Emission around
826:  Black Holes, ed. C. -H. Lee \& H. -Y. Chang (Singapore:
827:  World Scientific), 119
828: \bibitem[Miniutti \& Fabian (2004)]{}
829: Miniutti, G. \& Fabian, A. C. 2004, MNRAS, 349, 1435
830: \bibitem[Miniutti et al. (2007)]{}
831: Miniutti, G. et al. 2007, PASJ, 59, 315
832: \bibitem[Moscibrodzka et al. (2007)]{}
833: Moscibrodzka, M., Proga, D., Czerny, B. \&
834:  Siemiginowska, A. 2007, arXiv:0707.1403
835: \bibitem[Mushotzky et al. (1993)]{}
836: Mushotzky, R. F., Done, C. \& Pounds, K. A. ARA\&A, 31,
837:  717
838: \bibitem[Narayan \& Yi (1994)]{}
839: Narayan, R. \& Yi, I. 1994, ApJ, 428, L13
840: \bibitem[Nayakshin (2007)]{}
841: Nayakshin, S. 2007, MNRAS, 376, L25
842: \bibitem[Ohsuga et al. (2005)]{}
843: Ohsuga, K., Kato, Y. \& Mineshige, S. 2005, ApJ, 627,
844:  782
845: \bibitem[Pounds et al. (1990)]{}
846: Pounds, K. A., Nandra, K., Stewart, G. C.,
847:  George, I. M. \& Fabian, A. C. 1990, Nature, 344, 132
848: \bibitem[Pozdnyakov et al. (1977)]{}
849: Pozdnyakov, L. A., Sobol, I. M. \& Sunyaev, R.A. 1977,
850:  Soviet Astron., 21, 708
851: \bibitem[Reynolds et al. (1997)]{}
852: Reynolds, C. S., Ward, M. J., Fabian, A. C. \& Celotti, A. 1997,
853: MNRAS, 291, 403
854: \bibitem[Reynolds \& Nowak (2003)]{}
855: Reynolds, C. S. \& Nowak, M. A. 2003, Phys. Rep., 377,
856:  389
857: \bibitem[Rybicki \& Lightman (1979)]{}
858: Rybicki, G. B., \& Lightman, A. P. 1979,
859:  Radiative Processes in Astrophysics (New York: Wiley)
860: \bibitem[Shakura \& Sunyaev (1973)]{}
861: Shakura, N. I. \& Sunyaev, R. A. 1973, A\&A, 24, 337
862: \bibitem[Shapiro et al. (1976)]{}
863: Shapiro, S. L., Lightman, A. P. \& Eardley, D. M. 1976,
864:  ApJ, 204, 187
865: \bibitem[Shimura et al. (1995)]{}
866: Shimura, T., Mineshige, S. \& Takahara, F. 1995, ApJ,
867:  439, 74
868: \bibitem[Stone \& Pringle (2001)]{}
869: Stone, J. M. \& Pringle, J. E. 2001, MNRAS, 322, 461
870: \bibitem[Thorne \& Price (1975)]{}
871: Thorne, K. S. \& Price, R. H. 1975, ApJ, 195, L101
872: \bibitem[Wilms et al. (2001)]{}
873: Wilms, J., Reynolds, C. S., Begelman, M. C., Reeves, J.,
874:  Molendi, S., Staubert, R. \& Kendziorra, E. 2001,
875:  MNRAS, 328, L27
876: \bibitem[Yokoyama \& Shibata (1994)]{}
877: Yokoyama T. \& Shibata, K. 1994, ApJ, 436, L197
878: \end{thebibliography}
879: \end{document}
880: 
881: