The Spectral States of Black Hole X-ray Binary Sources

Xingming Chen1 and Ronald E. Taam2

1 Department of Astronomy & Astrophysics, Göteborg University and Chalmers University of
Technology, 412 96 Göteborg, Sweden
2 Department of Physics & Astronomy, Northwestern University, Evanston, IL 60208
arXiv:astro-ph/9606010v1 4 Jun 1996

ABSTRACT

A framework for the interpretation of the spectral states of black hole X-ray
transients based on the diversity of accretion disk models is introduced. Depending on
the mass accretion rate, it is proposed that the accretion disk is described by one or
a combination of the following structures: optically thick disk, advection-dominated
disk, corona-disk, and non-steady state disk. In particular, it is suggested that the
very high, high, low, and off states are characterized by mass accretion rates of
decreasing magnitude. The very high state corresponds to mass accretion rates near
the Eddington limit in which an optically thin non steady inner region is surrounded
by an optically thick structure. In the high state, the inner region is optically thin and
advection-dominated or optically thick. The low hard state is interpreted in terms of a
disk-corona system and the off state in terms of an optically thin disk dominated by
advective energy transport into the black hole. The possible observational consequences
of such a paradigm are discussed.

Subject headings: accretion, accretion disks — binaries: close — black hole physics —
stars: individual (Cyg X-1, A 0620–00, GS 1124-68, GS 2000+25, GS 2023+33, GRO
J0422+32) — X-rays: stars

1. INTRODUCTION

One of the most intriguing properties of black hole candidate X-ray sources (BHCs) is their
distinct spectral states. The first known BHC, Cyg X-1, was discovered by Bowyer et al. (1965)
and was recognized as probably containing a black hole by Webster & Murdin (1972) and Bolton
(1972). Cyg X-1 has an estimated mass in excess of ∼ 7 M⊙ and exhibits high and low states as
defined by the X-ray flux in 1–10 keV band. The luminosity difference in this energy band between
these states is approximately an order of magnitude. In the high state, the spectrum consists of
two components, namely, a relatively stable soft blackbody component and a weak, highly variable
hard power-law component. On the other hand, in the low state, the intensity shows rapid chaotic
temporal fluctuations with fractional rms amplitudes of several times 10% and which do not
 –2–

depend strongly on photon energy. The spectrum is exceptionally hard and can be described as a
power-law with a photon index αN less than 2 in the energy band between ∼ 10 keV and a few
100 keV (Liang & Nolan 1984). Other persistent BHCs include LMC X-3 and LMC X-1. Another
type of BHC is the soft X-ray transient among which includes A 0620–00 (V616 Mon), Nova
Muscae (GS/GRS 1124–68), and GS 2000+25 (Nova Vul). A recent review of the observational
data obtained from these sources has been given, for example, by Tanaka & Lewin (1995).

The spectral and time-dependent behavior of BHCs contain valuable information about the
underlying physics of the accretion process. It is generally believed that the soft blackbody spectral
component from these sources emanates from an optically thick cool accretion disk (Shakura &
Sunyaev 1973). The properties of the low hard state and the power-law component in the high
state imply the existence of optically thin, hot matter. However, the structure of the hot optically
thin flow is unclear. Specifically, several types of models have been suggested including the
optically thin two-temperature disk (Shapiro, Lightman, & Eardley 1976), and various corona-disk
models (Liang & Price 1977; Ionson & Kuperus 1984; Haardt & Maraschi 1993; Svensson &
Zdziarski 1994). Although the dynamics and stability properties of the corona-disk models have
not been well established, it is well-known that the optically thin disk models are catastrophically
thermally unstable (Piran 1978).

BHC X-ray transients may provide important clues for a model description of the different
spectral states since they exhibit a wide intensity range in which they can be studied. In the case
of Nova Muscae 91, it exhibited all the spectral states during its decay after the burst in January
1991 ( Kitamoto et al. 1992, Ebisawa et al. 1994). Since the X-ray nova outburst is very probably
due to mass flow modulations in the disk due to a thermal limit-cycle instability in the accretion
disk and/or a mass transfer induced instability in the companion star (see a recent review by
Lasota 1996), it is conceivable that the decay of the light curve reflects the decrease of the mass
flow rate in the disk. Near the peak of the burst, the soft X-ray blackbody component and the
hard X-ray, power law component, are comparable with the latter component much more variable
than the former. This state is called the very-high-state to distinguish it from the high-state which
occurs later. In the high state, the soft component remains, but the hard component is very
weak or totally disappears. During this state, a “reflare” in the light curve (by a factor about 2)
occurred about 70 to 80 days after peak (see Ebisawa et al. 1994). It is important to note that
this same kind of reflare is present in the light curve of BHC X-ray transients A 0620–00 and
GS 2000+25, which therefore suggests a common origin (Chen, Livio, & Gehrels 1993). At lower
luminosity levels (about a factor of 100 from the peak value) Nova Muscae exhibited a low-state
in which the spectrum is very similar to that of Cyg X-1.

A unified view for the spectral states of black hole candidate sources has yet to emerge.
Recently it has been suggested that the spectral changes are related to the mass flow rate in
the accretion disk (van der Klis 1995, Nowak 1995). However, this was adopted as a working
hypothesis and no physical explanation was provided. We note that an interpretation of the
transition from a low to a high state as due to an increase of mass accretion rate is opposite to
 –3–

the standard accretion disk theory which predicts thermal instabilities (and a possible transition
to an optically thin state) only for high mass accretion rates (e.g., Piran 1978).

Accordingly, in this paper we present a new model for the interpretation of spectral states in
black hole candidate systems in terms of accretion disk models characterized by an optically thick
component, advection-dominated component, and an optically thin corona. We suggest that the
very high, high, low, and off states are a sequence characterized by decreasing mass accretion rates
(see also van der Klis 1995). A similar picture, but differing in detail, has recently been suggested
by Narayan & Yi (1995) and Narayan (1996). In the next section, the model is presented in terms
of current ideas in accretion disk theory. The implications of such a model and its observational
consequences are discussed in the last section.

2. MODEL

Recently, Chen et al. (1995) have presented a unified description of accretion disks around
black holes. They showed that, at a given radius r, exactly four physically distinct types of
accretion disks may exist. Two of them correspond to a low viscosity in the disk, α < αcrit (r), and
the other two correspond to a high viscosity, α > αcrit (r). These disks are further differentiated by
the optical depth in the disk in the former case (i.e., optically thick or thin). For the high viscosity
case, the type of disk solution depends upon whether advective energy transport is negligible or
dominant, and advection-dominated solutions exist for all values of the mass accretion rate. On
the other hand, for a small viscosity, advection-dominated solutions exist except for a gap near
the Eddington rate, where no stable solution is possible and instability may occur. More recent
developments in disk models, which have included the detailed microphysics of the hot plasma,
have shown that αcrit may, in fact, exceed unity (Björnsson et al. 1996, Kusunose & Mineshige
1996). Therefore, we consider only the case of α < 1 < ∼ αcrit .
In Figure 1a the thermal equilibria of an accretion disk is illustrated at a fixed radius for
the case in which the viscous stress is proportional to the total pressure. The solution at high
column densities is optically thick and exhibits an S-shaped curve. The solid curve denotes
thermal-viscous stability whereas the dotted curve denotes instability. The disk is stable at low
mass accretion rates where it is gas pressure dominated and at high mass accretion rates where
it is advection-dominated. Instability occurs when the ratio of gas pressure to total pressure is
less than 0.4. In contrast, the solution at low column densities is optically thin. Here, the lower
branch is local cooling dominated (only bremsstrahlung cooling is included) and is thermally
unstable, whereas the upper branch is advection-dominated and is thermally stable (Narayan &
Yi 1994, Abramowicz et al. 1995a). We identify the maximum mass accretion rate (due to the
radial advection) at the tip of the optically thin branch as Ṁ1 and the mass accretion rate at the
lower turning point (due to the radiation pressure effects, prad /pgas ≥ 3/2) of the optically thick
branch as Ṁ2 . Note that at the upper turning point (due to the radial advection again), Ṁ is
usually larger than 10 − 50ṀE (see the S-curves of Abramowicz et al. 1988 and Chen & Taam
 –4–

1993). Hence, we consider only Ṁ less than that value. We estimate,

Ṁ1 /ṀE = 2.0 × 103 f (r/rg )−1/2 α2 , (1)

Ṁ2 /ṀE = 4.0 × 10−3 f −1 (r/rg )21/16 (αM/M⊙ )−1/8 . (2)

Here ṀE = 4πGM/(cκes ) is defined as the Eddington rate, where κes is the electron scattering
opacity taken to be equal to 0.34 and f = 1 − ℓ∗ /ℓ is considered for the inner boundary effect,
where ℓ and ℓ∗ are the specific angular momenta at radius r and the inner edge r∗ respectively.
The specific angular momentum is Keplerian and is calculated under the pseudo-Newtonian
potential of Paczyński & Wiita (1980). Formula (1) is calculated as that in Abramowicz et al.
(1995a) where only the bremsstrahlung is included for the local cooling.

The variations of Ṁ1 and Ṁ2 with respect to the radius of the disk are shown in Figure 1b.
It can be seen that Ṁ1 decreases slowly with increasing radius in the outer part of the disk. This
trend reflects the fact that local cooling processes become more efficient relative to the heating
rate at larger radii. A similar formula for Ṁ1 (r) has been calculated by Narayan & Yi (1995)
under the self-similar assumption. It has the same scaling with α but the coefficient is smaller by
a factor of approximately 10. In their calculation, the detailed radiative cooling processes such
as synchrotron and Componization are included. Since these processes become more important
for smaller radii, Ṁ1 decreases as r decreases. In their case, Ṁ1 /ṀE ≈ 10 − 20α2 for r <
∼ 1000rg ,
while our formula (1) gives Ṁ1 /ṀE ≈ 100α2 for r ≈ 1000rg . More recent calculations have shown
that, with similar detailed microphysics, the self-similar solution gives a smaller Ṁ1 than that
calculated under the Keplerian disk assumption. For example, for M = 10M⊙ , α = 0.1, and
r = 30rg , Björnsson et al. (1996) obtained Ṁ1 /ṀE ≈ 0.4 (see also Kusunose & Mineshige 1996)
while Narayan & Yi (1995) gave Ṁ1 /ṀE ≈ 0.035 (see their Figure 1 and scale Ṁ1 to the same α
and ṀE ). There are probably two factors contributing to this difference. One is the self-similar
assumption which underestimates the value of the angular velocity (see global solutions of Chen,
Abramowicz, & Lasota 1996 and Narayan, Kato, & Honma 1996a) and thus results in a smaller
heating rate. A smaller heating rate shifts the advection dominated thermal equilibrium curve
lower and shifts the optically thin local cooling dominated thermal equilibrium curve higher,
therefore Ṁ1 becomes smaller. Note that the Keplerian approximation overestimates the angular
velocity and, hence, the heating rate. Therefore, Ṁ1 lies within the range determined by these
approximations. The second factor is related to the inner boundary condition which cannot be
included in the self-similar solution. An accurate Ṁ1 can only be obtained by constructing global
solutions of the disk including detailed radiative microphysics. For the purpose here, considering
both effects of the radiative microphysics and the angular velocity, we may assume Ṁ1 as a very
weak function of r, defined as Ṁc1 for r <
∼ 1000rg :

Ṁc1 /ṀE ≈ 50α2 . (3)

Note that even a smaller Ṁc1 as that of Narayan & Yi (1995) will not effect our conclusion later,
since a slight increase of α will easily compensate for it.
 –5–

The increase of Ṁ2 with radius in the outer part of the disk is attributable to the tendency
that radiation pressure becomes important at larger radii only at higher mass accretion rates.
Figure 1b clearly shows that Ṁ2 has a minimum of Ṁc2 at radius of about rc ≈ 8 − 9 rg , where
rg = 2GM/c2 :
Ṁc2 /ṀE = 0.2(αM/M⊙ )−1/8 . (4)
It is clear from Figure 1b that no steady solutions exist for Ṁ > ∼ Ṁc1 . That is, in this mass
accretion rate range the region characterized by r < r2 is non steady (see Figure 1b). We note that
this region could be extensive, r2 ∼ 100rg , if Ṁ is large. It can be calculated that for Ṁ ≈ ṀE and
α>∼ 0.1, the effective optical depth of the disk in regions r < ∼ 100rg is less than unity. We suggest
that this region is hot and optically thin and is the seat of the chaotic hard X-ray variability.
Outflow is highly likely especially if Ṁ > ṀE . The region exterior to r2 is optically thick which
produces the soft X-rays. The hard and soft X-ray luminosities may be comparable depending on
the location of r2 . This state is identified with the very high-state.

At lower accretion rates, Ṁc1 > Ṁ > Ṁc2 , region r < r2 is described by an advection-
dominated flow which is hot, optically thin and stable (see also Narayan & Yi 1995). It is seen
that, as Ṁ decreases to near the bottom of the line r2 , the optically thick gas pressure dominated
disk solution starts to be available in the region to the left of the line r2 inside a few rg . The global
solution in this small region will however remain optically thin because its ”outer” boundary
condition is optically thin and advection dominated at this stage. In addition, since the flow
is transonic and the radial drifting time is almost a free-fall time scale, the once hot advection
dominated flow does not have time to settle to a cool state there. Since most of the gravitational
binding energy is advected into the black hole, the disk produces very little hard X-rays. On the
other hand, the region exterior to r2 is optically thick and the soft X-ray emission dominates the
total luminosity. This disk configuration describes the system in its high-state. Note that, as the
mass accretion rate decreases, the ratio of the hard X-rays to the soft X-rays decreases since the
spatial extent of the advection-dominated region decreases while that of the optically thick region
increases.

As the mass accretion rate declines further, Ṁc2 > Ṁ > Ṁc3 , the entire disk becomes
optically thick and the spectrum is similar to that described above for the high state. This
structural change, however, may result in an increase of luminosity, corresponding to a reflare, due
to the sudden decrease of the inner radius of the disk, rin , from r2 = rc ≈ 8 − 9rg to r2 ∼ 3rg
(since now, the whole disk can be optically thick). Since the disk luminosity is approximately
inversely proportional to the inner disk radius, its shift results in a luminosity increase by a factor
of ∼ 2 − 3.

A transition to a new stage occurs for Ṁc3 > Ṁ > Ṁc4 , where the system is in a corona-disk
configuration. The formation of corona above the disk may be due to an evaporation mechanism
facilitated by the electron conduction process as envisaged by Meyer & Meyer-Hofmeister (1994).
The corona is hot and optically thin and is responsible for the production of the hard X-rays. The
underlying disk is optically thick. This state corresponds to the low-state. In this corona-disk
 –6–

model, the soft X-rays (produced by the viscous dissipation in the underlying cold disk as well
as by the reprocessing of the hard X-rays from the above corona) are also Comptonized in the
corona. Therefore the overall energy spectrum can be described as a power-law (Haardt et al.
1993). The soft blackbody component, however, has a temperature of only Tbb < ∼ 0.1 keV because
of the low mass accretion rate. This component, however, is difficult to detect since the spectrum
peaks in the wavelength region where interstellar absorption is high.

At the lowest rates of mass accretion, Ṁ < Ṁc4 , the electron conduction process may lead
to the formation of a totally optically thin disk. In this case, the flow is advection-dominated,
thereby producing very little luminosity corresponding to the off-state. In particular, Narayan,
McClintock, & Yi (1996b) have applied the advection-dominated accretion flows to the off-state
of A 0620–00 (see also Lasota, Narayan, & Yi 1996). We note that Narayan (1996) has applied
the advection-dominated accretion flows to the low-state of BHCs, which however require a high
viscosity parameter (α ∼ 1) to produce the observed X-ray luminosity. In addition, the low-state
has also been modeled by a small inner advection-dominated flow plus an outer disk-corona
structure by Abramowicz, Chen, & Taam (1995b). The model, presented here, may reconcile these
differences depending on the values of Ṁc3 and Ṁc4 .

The critical values of Ṁc3 and Ṁc4 are difficult to estimate, but further work along the
directions advocated by Meyer & Meyer-Hofmeister (1994) on CV’s may be fruitful. For example,
Ṁc4 may be determined by the balance between the evaporation rate into the corona and the mass
accretion rate of the disk.

A schematic diagram illustrating the above description for the systematic variation of the
disk configuration with declining mass accretion rates is shown in Figure 2.

3. DISCUSSION

An interpretive framework for the spectral states of black hole candidate sources has been
presented in terms of the mass accretion rate only. The accretion disk models upon which the
framework is based require a moderately large α. Since Ṁc1 ∝ α2 , both very small and very
large α will give an extremely small or large Ṁc1 respectively, which is largely excluded by the
interpretation of observational data (see below). Considering the uncertainties involved with Ṁc1 ,
we estimate α ∼ 0.1 − 0.3, which gives an Ṁc1 corresponding to a luminosity near the Eddington
limit. This is probably the case for Nova Muscae. This value of α is consistent with the disk
instability model which produces the outburst and fits the exponential decay of the light curve
with an e-folding time of about 30-40 days (Cannizzo, Chen, & Livio 1995, and references within).

It is seen that the critical mass accretion rate, Ṁc2 , below which the disk can be stable and
optically thick depends on the disk parameter very weakly. Since it corresponds to the luminosity
level where the “reflare” occurs, it provides an estimate of the absolute value for the luminosity for
a given accretion efficiency. If the distance to the source and the inclination angle of the binary
 –7–

system are known, the observed luminosity can be used to provide an estimate of the mass of the
black hole. Independent of the distance to the source, Ṁc1 /Ṁc2 can be determined from the ratio
of the peak flux during the high state to the flux just before the reflare. The observations of Nova
Muscae (Ebisawa et al. 1994) and GS 2000+25 (Tsunemi et al. 1989) indicate that this ratio is
about ∼ 7 − 10, which is consistent with α ≈ 0.1 − 0.3.

In the present model, the “reflare” occurs as a result of the inward shift of the inner edge
of the optically thick disk. Thus, the ”reflare” is unique to a black hole source, with significant
emission in the soft X-ray band expected. In fact, this model predicts a decline in hard X-rays
during the reflare (see, for example, Ebisawa et al. 1994). This phenomenon cannot be explained
by other models (see Wheeler et al. 1996 for a review). It also predicts that, if there is also
an optical/UV “reflare”, it would result from the reprocessing of X-rays and, thus, it should
be delayed in this model. This is in contrast to the irradiation (of the secondary star) model
suggested by Chen et al. (1993) and the irradiation of the outer disk as envisioned by Kim et al.
(1995) in which, the optical/UV “reflare” is expected to arise prior to the X-ray reflare (see also
Mineshige 1994).

Our model does not predict a very high state if the accretion rate is less than Ṁc1 . In this
case, the “reflare” should occur earlier provided that the mass accretion rate is greater than Ṁc2 .
This may be the case in A 0620–00 in which the ”reflare” occured after 50–60 days after peak
luminosity (see the light curve in Kaluzienski et al. 1977 and Tanaka & Lewin 1995).

Our model also implies that, if a source has no high-state (i.e., stays hard), then there should
be no reflare. This appears to be the case in GS2023+33 (V404 Cyg: Kitamoto et al. 1989,
Miyamoto et al. 1995) and in GRO J0422+32 (Nova Persei: Sunyaev et al. 1993, Callanan et al.
1995, Vikhlinin et al. 1995). Note that the big-bump of hard X-rays in J0422+32 which occurred
more than 100 days after peak (Callanan et al. 1995) may have the same origin as the second
larger reflare observed in A 0620-00 and Nova Muscae, which we believe is different from the
“reflare” we have studied here (see Chen at al. 1993).

We note here that although the efficiency of producing luminosity is very different, the
energy spectra from the optically thin hot medium, of the inner region nonsteady flow in the
very high-state, the corona in the low-state, and the advection dominated flow in the off-state,
are rather similar since the radiative cooling mechanism is similar (for a spectral calculation, see
Narayan et al. 1996b). The contribution of soft X-rays from the outer optically thick disk in
the very high-state is easily observed since the mass accretion rate there is high and the typical
blackbody temperature is around 1keV. On the other hand, in the low state, the underlying
cold disk inside the corona has a very low blackbody temperature, < ∼ 0.1 keV, and therefore,
its detection is difficult. Interesting, Balucińska-Church et al. (1995) recently observed a soft
X-ray (Tbb = 0.13 ± 0.02 keV) excess in Cyg X-1 and they identified it as the disk emission. The
luminosity of this soft excess is about 4.7 × 1036 erg s−1 assuming a distance of the source as
2.5 kps.
 –8–

Narayan (1996) modeled the low-state as an advection-dominated accretion flow. In the

black hole X-ray transient case, then, the luminosity should drop significantly from the high state
(local cooling dominated) to the low state (advection dominated), independent of the viscosity
parameter. This has however not been observed and is unlike the off-state where the X-ray
luminosity and the mass accretion rate inferred from other observations indicate a very low
luminosity efficiency implying an advection dominated disk (Narayan et al. 1996b). There are
other observational tests to distinguish the advection dominated model from the ’sandwiched’
corona-disk model for the low state. For example, in the latter case, the spectrum should consist
of a reprocessed (in the corona above) hard X-ray reflection (off the disk) component. Haardt et
al. (1996) have constructed a corona-disk model to explain this type of spectral component from
Cyg X-1 (Gierliński et al. 1995). Theoretically, time-dependent simulation of both types of models
could determine whether or not the chaotic variability of the low-state can be reproduced within
each model.

In conclusion, the unified description of accretion disks around black holes have appealing
characteristics which can be applied to the spectral states of black hole candidate systems. In the
proposed model the spectral states are governed by the mass accretion rate alone. This picture,
however, is incomplete since the evidence from a combination of observational results from a
number of BHC sources reveal that the luminosity level at which the transition from a low state
to a high state on the ascending portion of the outburst differs from the transition from the high
state to the low state during the descending portion by a factor of about 100 (see Miyamoto et al.
1995). A possible explanation for this hysteretic behavior is that the transition from the low to
high state is physically distinct from the transition from the high to low state. For example, the
spectral state during the rising portion of the outburst may correspond to an optically thin disk
in its advection-dominated state (i.e. the solution corresponding to low column densities in Fig.
1a). The disk may remain advection-dominated until a transition occurs to the high soft state
at accretion rates near Ṁ1 (r). On the other hand, the transition from the high state to the low
state on the decay phase of the outburst reflects the operation of the evaporation mechanism in
the corona. Thus, the transition from optically thin to optically thick and vice versa are not
necessarily symmetric and differences in the luminosity level at which such transitions occur may
be expected to be different. To confirm the framework outlined in this paper and to quantify the
interpretation for such transitions, models for accretion disks based on a global solutions rather
than the local solutions must be constructed. The implications of such solutions on the hysteretic
behavior inferred in observed sources remain for future investigations.

X.C. thanks Jean Swank for hospitality and discussions while he was a visitor at GSFC/NASA
during March 1995. We thank Jean-Pierre Lasota and Ramesh Narayan for their comments on
an earlier version of the manuscript. This research was supported in part by NASA under grant
NAGW-2526.
 –9–

REFERENCES

Abramowicz, M. A., Chen, X., Kato, S., Lasota, J.-P., & Regev, O. 1995a, ApJ, 438, L37
Abramowicz, M. A., Chen, X., & Taam, R. E. 1995b, ApJ, 452, 379
Abramowicz, M. A., Czerny, B., Lasota, J.-P., & Szuszkiewicz, E. 1988, ApJ, 332, 646
Balucińska-Church, M., Belloni, T., Church, M. J., & Hasinger, G. 1995, A&A, 302, L5
Björnsson, G., Abramowicz, M. A., Chen, X., & Lasota, J.-P. 1996, ApJ, submitted
Bolton, C. T. 1972, Nature, 235, 271
Bowyer, S., Byram, E. T., Chubb, T. A., & Friedman, H. 1965, Science, 147, 394
Callanan, P. J., et al. 1995, ApJ, 441, 786
Cannizzo, J. K., Chen, W., & Livio, M. 1995, ApJ, 454, 880
Chen, W., Livio, M., & Gehrels, N. 1993, ApJ, 408, L5
Chen, X., Abramowicz, M. A.,& Lasota, J.-P. 1996, ApJ, submitted
Chen, X., Abramowicz, M. A., Lasota, J.-P., Narayan, R., & Yi, I. 1995, ApJ, 443, L61
Chen, X., & Taam, R. E. 1993, ApJ, 412, 254
Ebisawa, K. et al. 1994, PASP, 46, 375
Gierliński, M., Zdziarski, A. A., Johnson, W. N., Phlips, B. F., Ebisawa, K., & Done, C. 1995, in
Proc. Roentgenstrahlung from the Universe, Wuerzburg, Germany, in press
Haardt, F., Done, C., Matt, G. & Fabian, A. C. 1993, ApJ, 411, L95
Haardt, F., & Maraschi, L. 1993, ApJ, 413, 507
Haardt, F., et al. 1996, in preparation
Ionson, J. A., & Kuperus, M. 1984, ApJ, 284, 389
Kaluzienski, L. J., Holt, S. S., Boldt, E. A., & Serlemitsos, P. J. 1977, ApJ, 212, 303
Kim, S.-W., Wheeler, J. C., & Mineshige, S. 1996, in preparation
Kitamoto, S., Tsunemi, H., Miyamoto, S., Yamashita, K., & Mizobuchi, S. 1989, Nature, 342, 518
Kitamoto, S., Tsunemi, H., Miyamoto, S., & Hayashida, K. 1992, ApJ, 394, 609
Kusunose, M., & Mineshige, S. 1996, ApJ, submitted
Lasota, J.-P. 1996, in IAU Colloquium 158, Cataclysmic Variables and Related Objects, ed. J.
Wood & A. Evans (Dordrecht: Kluwer), in press
Lasota, J.-P., Narayan, R., & Yi, I. 1996, A&A, in press
Liang, E. P., & Nolan, P. L. 1984, Sp Sci Rev, 38, 353
Liang, E. P., & Price, R. H. 1977, ApJ, 218, 247
Meyer, F., & Meyer-Hofmeister, E. 1994, A&A, 288, 175
 – 10 –

Mineshige, S. 1994, ApJ, 431, L99

Miyamoto, S., Kitamoto, S., Hayashida, K., & Egoshi, W. 1995, ApJ, 442, L13
Narayan, R. 1996, ApJ, in press
Narayan, R., Kato, S. & Honma, F. 1996a, ApJ, submitted
Narayan, R., McClintock, J. E., & Yi, I. 1996b, ApJ, 457, 821
Narayan, R., & Yi, I. 1994, ApJ, 428, L13
Narayan, R., & Yi, I. 1995, ApJ, 452, 710
Nowak, M. 1995, PASP, 107, 1207
Paczyński, B., & Wiita, P. J. 1980, A&A, 88, 23
Piran, T. 1978, ApJ, 221, 652
Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337
Shapiro, S. L., Lightman, A. P., & Eardley, D. M. 1976, ApJ, 204, 178
Sunyaev, R., et al. 1993, A&A, 280, L1
Svensson, R. & Zdziarski, A. 1994, ApJ, 436, 599
Tanaka, Y. & Lewin, W. H. G., 1995, in X-Ray Binaries, ed. W. H. G. Lewin, J. van Paradijs &
E. P. J. van den Heuvel (Cambridge: Cambridge University Press), 126
Tsunemi, H., Kitamoto, S., Okamura, S., & Roussel-Dupré, D. 1989, ApJ, 337, L81
van der Klis, M. 1995, in X-Ray Binaries, ed. W. H. G. Lewin, J. van Paradijs & E. P. J. van den
Heuvel (Cambridge: Cambridge University Press), 252
Vikhlinin, A., et al. 1995, ApJ, 441, 779
Webster, B. L., & Murdin, P. 1972, Nature, 235, 37
Wheeler, J. C., Kim, S.-W., Moscoso, M. D., Kusunose, M., & Mineshige, S. 1996, in Basic Physics
of Accretion Disks, ed. S. Kato, S. Inagaki, J. Fukue, & S. Mineshige (New York: Gordon
& Breaches), in press

This preprint was prepared with the AAS LATEX macros v4.0.
 – 11 –

Fig. 1.— (a) The thermal equilibria of an accretion disk in the mass accretion rate-column density
plane. Two critical mass accretion rates are denoted by Ṁ1 and Ṁ2 , corresponding to the maximum
rate for an optically thin disk and the first turning point of the optically thick disk respectively. (b)
The radial variations of Ṁ1 and Ṁ2 (solid lines). The dotted line is for Ṁc1 = 2 from equation (3).
Note that Ṁ2 has a minimum of Ṁc2 ≈ 0.23 at about rc ≈ 8 − 9rg .

Fig. 2.— A schematic description of the disk configurations and the spectral states of black hole
candidates (see text).