Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~garcia/kong.bhspect.ps Äàòà èçìåíåíèÿ: Wed Feb 9 00:13:40 2011 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:22:21 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http www.badastronomy.com bad tv foxapollo.html

The X-ray Spectra of Black Hole X-ray Novae in Quiescence as Measured by
Chandra
Albert K.H. Kong, Jerey E. McClintock, Michael R. Garcia, Stephen S. Murray
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138
Didier Barret
Centre d'Etude Spatiale des Rayonnements, 9 Avenue du Colonel Roche, 31028 Toulouse, France
ABSTRACT
We present Chandra observations of black hole X-ray novae V404 Cyg, A0620{00,
GRO J1655{40 and XTE J1550{564 in quiescence. Their quiescent spectra can be well
tted by a power-law model with number slope   2. While a coronal (Raymond-
Smith) model is also a statistically acceptable representation of the spectra, the best t
temperatures of these models is  5 times higher than that seen in active stellar coro-
nae. These four spectra of quiescent X-ray novae are all consistent with that expected
for accretion via an advection-dominated accretion ow (ADAF) and inconsistent with
that expected from a stellar corona. This evidence for continued accretion in quies-
cence further strengthens the case for the existence of event horizons in black holes.
Both A0620{00 and GRO J1655{40 were fainter than in previous observations, while
V404 Cyg was more luminous and varied by a factor of 2 in a few ksec. A reanalysis
of the X-ray data for XTE J1550{564 shows that (like V404 Cyg and A0620{00) its lu-
minosity exceeds the maximum prediction of the coronal model by a large factor. The
0.3{7 keV luminosity of the four sources studied ranges from  10 30 10 33 erg s 1 .
Subject headings: binaries: close | black hole physics | stars: individual (V404 Cyg,
A0620{00, GRO J1655{40, XTE J1550{564) | X-rays: stars
1. Introduction
X-ray Novae (XN) are compact binary systems in which a Roche-lobe-over owing main se-
quence or subgiant star, typically  1 M , transfers matter onto a black hole (BH) or neutron star
(NS) primary (for a review, see van Paradijs & McClintock 1995; Tanaka & Lewin 1995; Tanaka
& Shibazaki 1996). XN are highly variable and undergo rare but dramatic X-ray and optical
outbursts. For most of the time, XN are in a quiescent state and are very faint. During quies-
cence, the mass accretion rate from the disk to the compact object may be very small, producing
a low level (perhaps no) X-ray emission. X-ray observations of quiescent XN have been hindered
due to the limited sensitivity of previous X-ray telescopes. Nonetheless, several of the brightest

{ 2 {
black hole X-ray novae (BHXN) have been detected with ROSAT, ASCA and BeppoSAX. This
quiescent X-ray (and associated non-stellar optical) emission is diôcult to explain using standard
accretion disk models. Narayan, McClintock, & Yi (1996), Narayan, Barret, & McClintock (1997a)
and Narayan, Garcia, & McClintock (2001) showed that the observations can be explained by an
advection-dominated accretion ow (ADAF) model.
An ADAF is an accretion ow in which most of the energy is stored in the accreting gas rather
than being radiated away promptly, as in a thin accretion disk. This thermal energy is advected
with the ow to the center { hence the name ADAF. If the accretor is a BH, the gas with all its
thermal energy will be lost from view as it falls through the event horizon. However, in the case of
a NS, the accretion energy will eventually be radiated from the star's surface. This dierence can
explain the fact that quiescent BHs are much fainter than quiescent NSs. Using pre-Chandra data,
Narayan, Garcia, & McClintock (1997b) showed that BHs display a large variation of luminosity
between their bright and their faint states, while NSs have a much smaller variation. Menou et al.
(1999) subsequently pointed out that in comparing the luminosities of BH and NS systems, it is
important to compare systems with comparable orbital periods. More recently, Garcia et al. (2001;
hereafter G01) presented a comprehensive study of a series of Chandra observations of BHXN in
quiescence; they conrmed that the quiescent X-ray luminosities of BHXN are  100 lower than
those of neutron star X-ray novae (NSXN). Such ndings provide strong evidence that BHs have
event horizons.
Recently, Bildsten & Rutledge (2000) suggested that the rapidly rotating secondaries of BHXN
may generate stellar coronae with suôcient X-ray luminosity to account for the observed quiescent
luminosities of many of these systems. Based on an analogy to the `saturated' coronae in the most
luminous RS CVn stars, the coronae in quiescent BHXN are predicted to have maximum luminosi-
ties of 0.1% of the stellar bolometric luminosity, and X-ray spectra that are typical of moderately
hot (kT  < 1 keV), optically thin thermal plasmas. While the X-ray luminosity of V404 Cyg is
too high to be produced by a stellar corona, previous observations with modest sensitivity have
indicated that the luminosities of other BHXN are consistent with a saturated corona (Bildsten &
Rutledge 2000). For these systems, the high S/N X-ray spectra attainable with Chandra and XMM-
Newton can provide a critical test of the possible coronal origin of the quiescent X-ray luminosity
(Bildsten & Rutledge 2000; Lasota 2001).
In this paper, we report the detailed analysis of Chandra spectra of the brightest three quiescent
BHXN observed under an AO-1 GTO program (V404 Cyg, A0620{00 and GRO J1655{40). We also
reanalyzed the spectrum of a fourth BHXN (XTE J1550{564) observed under a DDT proposal. We
note that three other BHXN (GRO J0422+32, GS 2000+25 and 4U 1543{47) observed under our
AO-1 GTO and GO programs provided insuôcient counts for spectral analysis. We brie y describe
previous quiescent observations of these four sources in x 2. In x 3 we outline our analysis procedure
and report the results in x 4. The results are discussed in x 5.

{ 3 {
2. Previous Quiescent X-ray Observations
All four BHXN have been observed previously in the X-ray, and a summary of previous quies-
cent observations is given in Table 1; we here discuss them brie y.
V404 Cyg | This relatively bright quiescent BHXN has previously been observed by ROSAT,
ASCA and BeppoSAX (see Table 1). In general, the X-ray spectrum can be tted by a power-law
model with photon index   2 and NH  (1 2)  10 22 cm 2 ; the luminosity is  10 33 ergs s 1
(Narayan et al. 1997a). We also note that the quiescent source ux can vary on short time scales.
Wagner et al. (1994) reported that V404 Cyg decreased in intensity by a factor of 10 in < 0.5 day
and showed variability by a factor of  2 on time scales of  30 minutes.
A0620{00 | This source was observed by ROSAT in 1992 during its quiescent state (McClin-
tock et al. 1995; Narayan et al. 1997a). The 398 counts detected allowed only a modest estimate
of the source spectrum. Simple one component models t the spectrum equally well: for example
a power-law with   3:5 and NH = (0:1 1)  10 22 cm 2 or a blackbody with kT = 0:16 +0:10
0:05
keV. The luminosity is  5  10 30 ergs s 1 . An ASCA observation in 1994 March failed to detect
the source; a 3 upper limit on the luminosity was 8  10 30 ergs s 1 (Asai et al. 1998).
GRO J1655{40 | The only quiescent observation of GRO J1655{40 was taken in 1996 March
with ASCA (Ueda et al. 1998; Asai et al. 1998). The spectrum can be tted by a power-law model
with a photon index   0:7 and NH < 3  10 21 cm 2 ; the source luminosity is 3  10 32 ergs s 1
in 0.5{10 keV. However, we note that this observation was taken between two outbursts separated
by  1 years and therefore it may not represent the true quiescent emission.
XTE J1550{564 | This microquasar system was observed as a DDT program on 2000 August
21 and 2000 September 11, which were > 120 d after the peak of the 2000 outburst of the source;
a detailed spectral analysis has already been given by Tomsick, Corbel, & Kaaret (2001). The
energy spectrum can be tted by an absorbed power-law spectrum with  = 2:3 +0:41
0:48 and NH =
(8:5 +2:2
2:4 )  10 21 cm 2 ; the mean luminosity (0.5{7 keV) is about 6:7  10 32 erg s 1 .
3. Chandra Observations and Data Reduction
V404 Cyg | Chandra observed V404 Cyg on 2000 April 26 for a total of 10,295 s. Our
observations cover spectroscopic phases 0.44{0.46 (Casares & Charles 1994), where phase zero
corresponds to the closest approach of the secondary star. The source was positioned on the ACIS-
S3 CCD with an oset of 40 00 from the nominal pointing for the S3. The data were collected using
a 1/4 subarray mode, which boosted the time resolution to 1.14 s. The CCD temperature was
120 ô C. Standard pipeline processed level 2 data were used for the analysis. V404 Cyg was clearly
detected and the source position is  = 20h 24m 03.82s, ô = +33d52m 02.14s (J2000), which is in
good agreement with the optical and radio position of V404 Cyg (Wagner et al. 1991).

{ 4 {
The Chandra detectors are known to experience periods of high background, which are par-
ticularly signicant for the S3 chip (e.g. Garcia et al. 2000). We searched for such background
ares in our data by examining the light curve of the entire S3 chip minus the source regions. We
found that the background was very stable during the whole observation with an average count
rate of 0.13 count s 1 . In order to reduce the background, we only analyzed data from 0.3{7 keV.
We extracted data from a circle of 3 pixels ( 1:5 00 ) centered on V404 Cyg and background from
an annulus with inner and outer radii of 10 and 50 pixels, respectively. There were 1587 counts
in the source region and the expected number of background counts in the source region was only
0.4 counts.
A0620{00 | This source was observed by Chandra on 2000 February 29 for 44,000 s. ACIS-S
was operated in the standard conguration with a time resolution of 3.24 s. A0620{00 was observed
on the S3 chip with a 40 00 oset from the nominal pointing. Background was examined; only intervals
where the source-free count rate was less than 0.15 count s 1 were selected for analysis. The total
net exposure time is 41,189 s. The source position is  = 06h 22m 44.48s, ô = -00d 20m 46.36s
(J2000) which is consistent with the optical position (Liu, van Paradijs, & van den Heuvel 2001).
The observations cover spectroscopic phases 0.09{1.67 (Orosz et al. 1994; Leibowitz, Hemar, &
Orio 1998). Only data from 0.3{7 keV were used for spectral analysis. We extracted data from a
circle of 1:86 00 centered on A0620{00. This relatively large aperture encompasses all of the counts
in the central region that might reasonably be attributed to the source. There were 137 counts in
the source region. The background counts in a 1:86 00 aperture are estimated to be 1.2. This small
background level was not subtracted.
GRO J1655{40 | Chandra observed GRO J1655{40 on 2000 July 1 for 43,000 s, which
corresponds to spectroscopic phases of 0.49{0.68 (van der Hooft et al. 1998). The source was located
on ACIS-S3 with 40 00 oset from the aim-point; standard 3.24 s frame transfer time was employed.
Good data were selected with background count rate < 0:15 count s 1 , resulting in a net exposure
of 42,506 s. GRO J1655{40 was very faint; by ltering the data from 0.3{7 keV and applying a
circular extraction region of 1:41 00 centered on the source, only 66 counts were collected. This choice
of aperture encompasses all of counts in the central region that are attributable to the source. The
estimated background counts in a 1:41 00 aperture is estimated to be 0.7; this background was not
subtracted. The Chandra source position is  = 16h 54m 00.09s, ô = -39d 50m 45.37s (J2000),
which is consistent with the radio and optical position (Hjellming 1994; Bailyn et al. 1995).
XTE J1550{564 | The source was observed on 2000 August 21 for  5,000 s and 2000
September 11 for an additional  5,000 s; the observations cover spectroscopic phases 0.06{0.11
and 0.63{0.68, respectively (Orosz et al. 2001). Technical details of the observations can be found
in Tomsick et al. (2001). We used similar procedures to those outlined in Tomsick et al. (2001) to
reduce the data. However, we extracted data from 0.3{7 keV and used a smaller circular extraction
region with a radius of 2 00 , which is suôcient to encompass all of the counts in the central region.
There are 66 and 109 counts in the rst and the second observations, respectively; we ignored the
background counts in the source region, which we estimated to be 0.2 counts for the rst observation

{ 5 {
and 0.3 counts for the second observation .
4. Spectral Analysis
4.1. V404 Cyg
Spectra were extracted with CIAO v2.1 1 and were analyzed with XSPEC v11 2 and also
SHERPA v2.1.2 3 . The results from both analysis systems were consistent, and we report the
XSPEC results herein. In order to allow  2 statistics to be used, all the spectra were grouped into
at least 30 counts per spectral bin. Response les were selected according to the CCD temperature
with standard CIAO routines. We t the data with several single-component spectral models
including power-law, thermal bremsstrahlung, Raymond-Smith and blackbody models including
interstellar absorption. The best-t parameters determined by these ts are shown in Table 2.
All models except the blackbody model gave statistically acceptable ts to the data ( 2 =
 < 1). The power-law model provides the best t, and yields parameters consistent with previous
observations (e.g.  = 1:81  0:14; see Table 1). This best tting model is shown in Figure 1 and
the corresponding plot of condence regions for column density (NH ) and photon index () are
shown in Figure 2a. The condence bounds for the Raymond-Smith model are shown in Figure
2b. The best t temperature for this model is kT = 7:5 keV, and the 90% lower limit on the
temperature is kT > 6:1 keV.
The hydrogen column density for V404 Cyg from optical observations was estimated to be
5:4  10 21 cm 2 (A V = 3:1; Casares & Charles 1994). The best t values for NH from the power
law and bremsstrahlung models are marginally higher than the optically determined value, but
this does not necessarily argue against these models. X-ray binaries often show absorption in the
X-ray ux which is somewhat higher than that determined by their optical absorption (Garcia
1994; Vrtilek et al. 1991).
In order to test if the optically-determined absorption yields an acceptable X-ray spectral t,
we re-ran the ts with the absorption xed to this value. The results of these ts are also given in
Table 2. Even though this NH value is outside the 99% condence bounds shown in Figure 2, these
ts do yield acceptable values of  2 = (except for the blackbody model). This is a re ection of the
fact that the minimum value of  2 = obtained with NH as a free parameter is slightly less than
one, thereby allowing points outside the  2
min + 9:21 (Lampton, Margon, & Bowyer 1976) contour
to have  2 =  1. For these ts with NH xed, the best t temperature for the Raymond-Smith
model is raised to 8.9 keV, and the 90% lower limit is raised to > 7:2 keV.
1 http://asc.harvard.edu/ciao/
2 http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/index.html
3 http://asc.harvard.edu/ciao/download/doc/sherpa html manual/index.html

{ 6 {
We do not see any signicant Fe-K line emission between 6.4{7 keV, with a 90% condence
upper limit of  800 eV (line width xed at 0.1 keV) on the equivalent width.
4.2. A0620{00
We analyzed the energy spectrum of A0620{00 using procedures similar to those discussed
above for V404 Cyg. We grouped the data into spectral bins containing at least 10 counts and
used both  2 and CASH (Cash 1979) statistics to estimate the best-t parameters and their errors.
We chose to bin the data in order to achieve enough counts per bin to employ the  2 statistic.
However, binning the data heavily can result in a loss of spectral information. One can also apply
the Gehrels approximation (Gehrels 1986) to permit the use of fewer counts ( 5) per bin, but this
approach over-estimates the errors. The CASH statistic is a maximum likelihood method designed
to estimate the best-t parameters using unbinned or slightly binned data. This is particularly
useful when the source yields only very few photons. The disadvantage of the CASH statistic,
relative to the  2 statistic, is that it does not provide a goodness-of-t criterion for comparing
dierent models. It is therefore worthwhile to examine the results obtained using both  2 and
CASH statistics. In Table 2, except for V404 Cyg, all the best-t parameters and errors are based
on the CASH statistic; the reduced  2 values are also shown to indicate the quality of the t. In
order to justify the signicance of the CASH statistic, we performed Monte-Carlo simulations to
estimate the signicant level of the ts; these results are also given in Table 2.
Both methods give very consistent results. We employed the same four single-component
models with interstellar absorption that we used for V404 Cyg; the best-t parameters for the
various spectral models are shown in Table 2. Among all the models, the power-law gives the best
t ( 2 = = 0:71,  = 2:20:5), while the blackbody gives the worst t ( 2 = = 1:58); Monte-Carlo
simulations based on the CASH statistic also show similar results. The condence regions for the
power-law t are shown in Figure 2a, and those for the Raymond-Smith t are shown in Figure
2b. The best t Raymond-Smith temperature is kT = 5:5 keV, and the 90% lower bound on the
temperature is kT > 3:5 keV.
The values of NH determined by the power-law and bremsstrahlung ts are consistent with
the optical value, corresponding to NH = (1:94  0:28)  10 21 cm 2 (Wu et al. 1976, 1983; Predehl
& Schmitt 1995). The value of NH determined by the Raymond-Smith and blackbody models is
lower than the optical value. This conclusion provides marginal evidence that neither the blackbody
nor the Raymond-Smith models are correct descriptions of the source spectrum because X-ray ts
tend to nd NH higher than (or consistent with) the optically determined value. As in the case of
V404 Cyg, we re-ran the ts with NH xed at the optically determined value. The results of these
ts are also shown in Table 2. The derived parameters are consistent (within 1) with the results
obtained by varying NH , except for the case of the blackbody model. The best-t temperature for
the Raymond-Smith model is 4.1 keV, and the 90% lower bound is > 2:8 keV.

{ 7 {
Previously, the best measurement of the X-ray spectrum of A0620{00 was that aorded by
ROSAT (Narayan et al. 1997a), which gave  = 3:5  0:7 with NH xed to the optical value.
This led to the speculation that the quiescent X-ray spectra of BHXN with orbital periods  < 1 day
might be softer than the spectra of longer period systems. However, this result was based on
only 39  8 detected source photons in the presence of a signicant background. The present
result is much more robust because it is based on more than 3 times as many counts, a negligible
background, and a much wider energy band. It is important to note that A0620-00 was also a factor
of  2 fainter in this Chandra observation than it was during the previous ROSAT observation.
The best tting power-law model indicates a 0.4{2.4 keV emitted ux of 1:9  10 14 erg cm 2 s 1 ,
corresponding to a luminosity of 2:110 30 erg s 1 , which is a factor of two below the ROSAT value
(see Table 1).
4.3. GRO J1655{40
The spectrum of GRO J1655{40 was analyzed using the same methods discussed above for
A0620{00. The energy spectrum was grouped into spectral bins containing at least 5 counts and
t using  2 and CASH statistics. Unbinned data was also t using CASH statistic, and the results
were consistent. All simple models give acceptable ts. While the blackbody model gives the
poorest ts, it cannot be rejected on the basis of  2 = and Monte-Carlo simulations. However, the
NH for the blackbody model is slightly lower (1:5) than the optical value of (6:66  0:57)  10 21
cm 2 (Predehl & Schmitt 1995; Hynes et al. 1998), while the other three models indicate values of
NH consistent with the optically-derived value. The relatively low value of NH suggests that the
blackbody model may not be a true representation of the source spectrum.
The best t temperature for the Raymond-Smith model is kT = 12:24 keV, and the 90% lower
limit on the temperature is kT > 3:63. If we x NH to the optical value, these values are raised to
kT = 17:15 keV and kT > 5:8 keV.
As above, we list the best t parameters in Table 2, and show a plot of the condence regions
for power-law and Raymond-Smith ts in Figures 2a and 2b. It is important to note that these
observations show GRO J1655{40 to be a factor of  10 fainter than previous quiescent observations
(see Table 1). The observed 0.4{2.4 keV emitted ux, for the best tting power-law model, is
1:5  10 14 erg cm 2 s 1 ; the observed 0.3{7.0 keV luminosity is 2:4  10 31 erg s 1 . The large
decrease in ux and luminosity indicate that the previous ASCA observations may not have been
taken during the true quiescent state because the observations occurred between two outbursts.
4.4. XTE J1550{564
We combined the two spectra of XTE J1550{564 as shown in Tomsick et al. (2001), grouped
the resulting data into bins containing at least 10 counts each, and t the data to models using  2

{ 8 {
and CASH statistics. The results of the spectral ts are shown in Table 2, and the corresponding
parameter condence regions are shown in Figure 2. All four models yield statistically acceptable
ts, and we see no straightforward way to select one model over the others. With the exception of
the blackbody model, all of the models indicate that NH is somewhat higher than that determined
optically (Sanchez-Fernandex et al. 1999). However, as indicated above, this is only a weak
argument against the blackbody model. Fits with NH xed to the optical value are also statistically
acceptable, and indicate harder ( lower, kT higher) spectra than the ts with NH free.
The Raymond-Smith ts indicate a best t temperature of kT = 4:38 keV, and a 90% lower
limit to the temperature of kT > 2:81 keV. Fits with NH xed to the optical value raise these
values to kT = 10:31 keV and kT > 5:15 keV. The results of power-law t are consistent with those
found by Tomsick et al. (2001).
In order to determine if the quiescent X-ray emission of XTE J1550{564 has a ux consis-
tent with a stellar corona, we calculated the unabsorbed X-ray ux (F X ) and bolometric ux for
XTE J1550{564 using the methods of Bildsten & Rutledge (2000). Based on our best-t power-law
result, the unabsorbed 0.4{2.4 keV ux of XTE J1550{564 is 2:98  10 13 erg cm 2 s 1 . For the
bolometric ux, we used F bol = 10 0:4(Vq +11:51+B:C: A V ) erg cm 2 s 1 (Bildsten & Rutledge 2000),
where B.C. is the bolometric correction for spectral type, V q is the quiescent magnitude, and A V is
the reddening. We adopted a V q of 22  0:2 and a spectral type of K3III from recent VLT observa-
tions (Orosz et al. 2001), which indicates B.C.= 0:8. We computed F bol using the A V determined
from optical observations (A V = 2:17; Sanchez-Fernandez et al. 1999) and nd F bol = 5:1  10 13
erg cm 2 s 1 . We also determined F bol using the A V estimated from our X-ray spectral tting. For
a power-law model, NH = 8:73  10 21 cm 2 implies that A V = 4:88 (Predehl & Schmitt (1995)],
implying F bol = 6:1  10 12 erg cm 2 s 1 . These values of F bol are discussed in Section 6.
5. Time-resolved Spectrum of V404 Cyg
The background-subtracted light curve of V404 Cyg during our observations is shown in Fig-
ure 3. The light curve shows a factor of  2 variability in a few ksec. We do not nd any signicant
peak in the power spectrum on timescales from 2.3 s to 10,000 s and the 3 upper limit on the
semi-amplitude is 39% (0.3{7 keV).
The marked variability led us to search for spectral changes at diering ux levels. The data
was divided into seven segments based on the source intensity (see Figure 3). The spectrum from
each segment contains at least 100 counts. The results of tting each spectrum with a power-law
model are shown in Table 3. The best-t column density varied between (2:91 11:08)  10 21
cm 2 , and the best t photon index  varied between 1.1{2.4. We found no correlation between
either the column density or  and the ux. However, we do nd a positive correlation between
the absorption column and the photon index (see Figure 4) with a correlation coeôcient of 0.93 (>
99%).

{ 9 {
However, we suspect that this correlation is not intrinsic to the source, but is rather an artifact
of the tting process which links  and NH . For example, we note that the slope of the correlation is
nearly (within  5%) the same as the slope of the major axis of the parameter condence contours
(Figure 2a). Also, we extracted and examined two spectra, one for count rates below 0.11 counts s 1
and the other for count rates above 0.18 counts s 1 (see Figure 3), and found them to be identical.
We conclude that the spectral shape does not vary with intensity.
6. Discussion
We analyzed the Chandra ACIS-S X-ray spectra of four BHXN in quiescence by tting the
spectra to simple one component models (power-law, thermal Bremsstrahlung, Raymond-Smith,
blackbody) including interstellar absorption. While the statistics aorded by the Chandra data
surpass that previously available, they are still inadequate to rule out any of these simple model,
except for the blackbody model in the case of V404 Cyg. There is some weak additional evidence
against a few other models: For A0620{00 the Raymond-Smith and blackbody models imply un-
likely values of NH which are lower than the optically-determined values. The same is true for
GRO J1655{40 and XTE J1550{564 in the case of the blackbody model. On the other hand, the
thermal bremsstrahlung model provides a good t to the data in all cases; however, the physical
interpretation of this model is unclear (see Christian & Swank 1997). The model which does t
well in all cases and which has a straightforward physical interpretation is the power-law model
with a photon index of  2. This slope is consistent with the spectra expected for an ADAF.
Bildsten & Rutledge (2000) suggest that much of the X-ray ux observed from quiescent BHXN
may be produced by a rotationally enhanced stellar corona in the secondary star, as seen in tidally-
locked binaries such as the RS CVn systems. Lasota (2000) has criticized this view, suggesting
instead that the physically smaller secondaries of CVs provide a better analog, and that in this
case the expected coronal emission is far below that seen in quiescent BHXN.
The coronal hypothesis of Bildsten & Rutledge (2000) makes two clear, testable predictions.
First, that LX < 10 3 L bol , and second, that the spectrum of the quiescent BHXN should be similar
to that of a stellar corona, i.e., well represented by a Raymond-Smith model with kT < 1:4 keV,
(Dempsey et al. 1993). The luminosity and spectral evidence available for ve of the six BHXN
observed by Chandra rule strongly against these hypotheses, as detailed below. Note that we do not
include in this discussion a seventh BHXN observed with Chandra 4U 1543-47, (see G01) because
it contains a fully radiative secondary (Orosz et al. 1998) and is not expected to possess an X-ray
corona. Consequently, this system is irrelevant to the present discussion.
Figure 5, which is adapted from Bildsten & Rutledge (2000), compares the quiescent uxes of
BHXN with the predictions of the coronal model. The quiescent ux of GRO J0422+32 exceeds
the maximum prediction of the coronal model by a factor of  60, and V404 Cyg exceeds this limit
by a factor of  40. XTE J1550{564 exceeds the coronal limit by a factor  50 (or  400) for the

{ 10 {
highest (or lowest) L bol computed in section 4.4. However, the luminosity of XTE J1550{564 should
be treated with caution because a mini-outburst occurred 120 d after this observation (see Tomsick
et al. 2001). This situation is very similar to the case of the ASCA observation of GRO J1655{40
made between two outbursts which gave a high value of the luminosity (see Section 4.3). Finally,
A0620{00 is a factor of  5 above the coronal prediction, which may be a signicant discrepancy
since the prediction corresponds to the maximum likely level of coronal emission.
Turning to the spectral evidence, we nd herein that the X-ray spectra of V404 Cyg, A0620-
00, GRO J1655{40 and XTE J1550{564 are harder (equivalently hotter) than typical spectra of
stellar coronae. The average temperature for these sources as determined from the NH free (xed)
ts to Raymond-Smith models is 7.4 keV (10.1 keV). The average of the 90% lower limits to the
temperatures is > 4 keV (or > 5:24 keV from the NH xed ts). Coronal sources are often t by
Raymond-Smith models with two separate temperature components. The average of the higher of
these temperatures has a value of 1.4 keV (Dempsey et al. 1993). Thus, in the four systems for
which the data are of suôcient quality to allow us to measure the X-ray spectrum, the temperature
is  5 to 7 times higher than the higher temperature typically seen from stellar corona.
Thus the combination of spectral and luminosity information argue against a coronal source
for the quiescent luminosity in 5 out of the 6 cases for which the coronal mechanism is potentially
relevant (i.e. excluding 4U 1543-47). Only in the case of GS 2000+25, where we are unable to
determine a spectrum due to the very low number of counts, is it possible that coronal emission
from the secondary dominates the quiescent luminosity.
During strong ares, stellar coronae are occasionally seen at temperatures higher than the
1.4 keV average value quoted above. For example, a \super-hot giant are" from Algol was seen
to have a peak temperature of 12.37 keV (Favata & Schmitt 1999). In this regard, it is important
to note that both A0620-00 and GRO J1655{40 were observed with Chandra at lower luminosities
than in previous quiescent observations. Therefore it is unlikely that these two systems were in a
aring state during our observations.
Either the secondaries in BHXN have coronae unlike those seen before, or the source of the
quiescent luminosity is not coronal. This is not to say that these secondaries do not have X-ray
emitting corona, but merely that the luminosity from such a corona is swamped by the accretion
luminosity even during quiescence. An obvious point to note is the following. Emission from a
stellar corona will contribute at some level to the quiescent X-ray luminosity. If in a few cases this
level is signicant, then the accretion luminosities of the black holes must be even lower than our
estimates and the argument for event horizons would be further strengthened.
It is worth noting that the ve BHXN for which coronal emission is ruled out cover the full
range of orbital period and stellar bolometric ux. It therefore seems unlikely that there is some
particular region of parameter space where the coronal model applies. In comparison, the ADAF
model is consistent with all the observations, covering the full parameter space (Narayan et al.
1996, 1997a, 2001; Lasota 2000).

{ 11 {
Results of this paper further constrain the required ADAF model. Quataert & Narayan (1999)
proposed that signicant mass can be lost to an out ow/wind in ADAF models. They also predicted
the spectral shape for ADAF models with and without winds for V404 Cyg. Our observations
indicate that the power-law photon indices of V404 Cyg, A0620{00, GRO J1655{40 and XTE J1550{
564 are consistent with   2. Therefore, models in which Comptonization dominates are favored
(Narayan et al. 1997a); strong-wind models become unlikely unless ô (the fraction of the turbulent
energy which heats the electrons) is large enough (Quataert & Narayan 1999). ADAF models also
predict line emission in X-ray spectra (e.g. Narayan & Raymond 1999). We set an upper limit
on the equivalent width of any line feature between 6.4{7 keV for V404 Cyg and it is much higher
than the theoretical prediction even for model with winds. A larger collecting area instrument like
XMM-Newton is needed to study this kind of feature. Recent RXTE and Chandra observations
of XTE J1550{564 also suggest that the ADAF model can explain the quiescent X-ray emission,
although it does not explain all the behavior observed at other wavelengths (Tomsick et al. 2001).
The similarity of the quiescent spectra of V404 Cyg, XTE J1550{564, GRO J1655{40, and A0620-00
found herein suggests that they may all be described by a similar ADAF model. Detailed broadband
spectral modeling of these systems should be considered in order to further constrain the models.
The sources discussed in this paper have a wide range of luminosities. V404 Cyg is the brightest
quiescent BHXN in our sample, with a 0.3{7 keV luminosity of  5  10 33 erg s 1 . In our Chandra
observations the source was somewhat more luminous than in previous quiescent observations in
which the luminosity was about 10 33 erg s 1 (see Table 1). Wagner et al. (1994) reported that
V404 Cyg exhibited a decrease in intensity by a factor of 10 in < 0.5 day, while our Chandra
observations showed a factor of 2 variability in a few ksec. Wagner et al. (1994) also found that
there may have been a factor  2 variability on timescales of  30 minutes in the highest intensity
bins for the ROSAT observations. Thus V404 Cyg in quiescence shows variability in X-rays on
both short-term (a few ksec) and long-term (years) timescales. Signicant X-ray variability in
quiescence was also seen in 4U 1630{47 (Parmar et al. 1997), A0620{00 (Asai et al. 1998; Menou
et al. 1999; also Table 1) and GX 339{4 (Kong et al. 2000). V404 Cyg and GX 339-4 are similar in
some respects: for example, their quiescent X-ray luminosities are comparable (Kong et al. 2000,
2002) and GX 339{4 has also been observed to undergo X-ray variability by a factor of 3 during its
quiescent or `o' state (Kong et al. 2002). Thus, variability in the quiescent state is common, which
suggests that BHXN in quiescence are not totally turned o. We note that XTE J1550{564 also
varied in luminosity by factor of  2 during the two Chandra observations in quiescence (Tomsick
et al. 2001); only V404 Cyg and GX 339{4 have a quiescent luminosity higher than XTE J1550{564.
In Figure 6, we plot the Eddington-scaled luminosities (based on the best-t power-law model)
as a function of orbital period P orb ; this is an update of the same plot from G01. For the mass
of XTE J1550{564 we assumed M = 10:6M (Orosz et al. 2001); the distance to XTE J1550{564
is estimated to be 2.5{6.3 kpc (e.g. Sanchez-Fernandez et al. 1999; Orosz et al. 2001) and we
have adopted an average distance of 4 kpc. We note that for the three long orbital period systems
(V404 Cyg: 6.47 d, GRO J1655{40: 2.6 d and XTE J1550{564: 1.55 d), the quiescent luminosity is

{ 12 {
higher than for the other systems (Figure 6). This implies that the accretion rate in these systems
is higher than for the others according to the ADAF model (Narayan et al. 1997a; Menou et al.
1999). It is not clear if there is a positive correlation between the luminosities and orbital periods
(see Figure 6); a larger sample of long orbital period systems is required to study this correlation.
In summary, we note that our results conrm the prediction of Lasota (2000), who previously
pointed out that X-ray emission from a quiescent BHXN is unlikely to come from a stellar coronae;
instead he argues that the emission is due to an ADAF. Based on this model, Lasota (2000) predicted
uxes similar to those reported herein. Moreover, he pointed out that detection of GRO J0422+32
by Chandra would rule out the coronal model, and such a detection has been made (G01). However,
our Chandra spectra are able to rule out only a few of the simple, one-component spectral models
we t to the data. With its larger collection area, observations with XMM-Newton should be
able to do a signicantly better job of constraining the source spectra. Additionally, we note that
V404 Cyg is variable on a few ksec timescale, so simultaneous optical and X-ray observations may
shed substantial light on the quiescent accretion processes in this source.
AKHK was supported by a Croucher Fellowship. JEM was supported in part by NASA
grant GO0-1105A. MRG acknowledges the support of NASA LTSA Grant NAG5-10889 and NASA
Contract NAS8-39073 to the CXC. The HRC GTO program is supported by NASA Contract
NAS-38248.
REFERENCES
Asai, K., Dotani, T., Hoshi, R., Tanaka, Y., Robinson, C.R., & Terada, K. 1998, PASJ, 50, 611
Bailyn, C.D., et al. 1995, Nature, 374, 701
Bildsten, L., & Rutledge, R.E. 2000, ApJ, 541, 908
Callanan, P.J., Charles, P.A., Honey, W.B., & Thorstensen, J.R. 1992, MNRAS, 259, 395
Campana, S., Parmar, A.N., & Stella, L. 2001, A&A, 372, 241
Casares, J., & Charles, P.A. 1994, MNRAS, 271, L5
Cash, W. 1979, ApJ, 228, 939
Christian, D.J., & Swank, J.H. 1997, ApJS, 109, 177
Dempsey, R.C., Linsky, J.L., Schmitt, J.H.M.M., & Fleming, T.A. 1993, ApJ, 413, 333
Favata, F., & Schmitt J.H.M.M. 1999, A&A, 350, 900
Garcia, M.R., McClintock, J.E., Narayan, R., Callanan, P., Barret, D., & Murray, S.S., 2001, 553,
L47 (G01)
Garcia, M.R., Murray, S.S., Primini, F.A., Forman, W., McClintock, J.E., & Jones, C. 2000, ApJ,
537, L23

{ 13 {
Garcia, M.R. 1994, ApJ 435, 407
Gehrels, N. 1986, ApJ, 303, 336
Hjellming, R.M. 1994, IAUC 6055
Hynes, R.I., Haswell, C.A., Shrader, C.R., Chen, W., Horne, K, Harlaftis, E.T., O'Brien, K.,
Hellier, C., & Fender, R.P. 1998, MNRAS, 300, 64
Kong, A.K.H., Kuulkers, E., Charles, P.A., & Homer, L. 2000, MNRAS, 312, L49
Kong, A.K.H., Charles, P.A., Kuulkers, E., & Kitamoto, S. 2002, MNRAS, 329, 588
Lampton, M., Margon, B., & Bowyer, S. 1976 ApJ 208, 177
Lasota, J.-P. 2000, A&A, 360, 575
Leibowitz, E.M., Hemar, S., & Orio, M. 1998, MNRAS, 300, 463
Liu, Q.Z., van Paradijs, J., & van den Heuvel, E.P.J. 2001, A&A, 368, 1021
McClintock, J.E., Horne, K., & Remillard, R.A. 1995, ApJ, 442, 358
Menou, K., Esin, A.A., Narayan, R., Garcia, M.R., Lasota, J.-P., & McClintock, J.E. 1999, ApJ,
520, 276
Narayan, R., McClintock, J.E., & Yi, I. 1996, ApJ, 457, 821
Narayan, R., Barret, D., & McClintock, J.E. 1997a, ApJ, 482, 448
Narayan, R., Garcia, M.R., & McClintock, J.E. 1997b, ApJ, 478, L79
Narayan, R., Garcia, M.R., & McClintock, J.E. 2001, to appear in Proc. IX Marcel Grossmann
Meeting, eds. V. Gurzadyan, R. Jantzen and R. Ruôni, Singapore: World Scientic
Narayan, R., & Raymond, J. 1999, ApJ, 515, L69
Orosz, J.A., Bailyn, C.D., Remillard, R.A., McClintock, J.E., & Foltz, C.B. 1994, ApJ, 436, 848
Orosz, J.A., Jain, R.K., Bailyn, C.D., McClintock, J.E., & Remillard, R.A. 1998, ApJ, 499, 375
Orosz, J.A., Groot, P.J., van der Klis, M., McClintock, J.E., Garcia M.R., Zhao P., Jain R.K.,
Bailyn C.D., & Remillard R.A. 2001, ApJ, in press (astro-ph/0112101)
Parmar, A.N., William, O.R., Kuulkers, E., Angelini, L., & White, N.E. 1997, A&A, 319, 855
Predehl, P., & Schmitt, J.H.M.M. 1995, A&A, 293, 889
Quataert, E., & Narayan, R. 1999, ApJ, 520, 298
Sanchez-Fernandez, et al. 1999, A&A, 348, L9
Tanaka, Y., & Lewin, W.H.G. 1995, in X-ray Binaries, p.126, eds Lewin, W.H.G., van Paradijs J.,
ban den Heuvel, E.P.J., Cambridge Univ. Press, Cambridge.
Tanaka, Y., & Shibazaki, N. 1996, ARAA, 34, 607
Tomsick, J.A., Corbel, S., & Kaaret, P. 2001, ApJ, 563, 229

{ 14 {
Ueda, Y., Inoue, H., Tanaka, Y., Ebisawa, K., Nagase, F., Kotani, T., & Gehrels, H. 1997, ApJ,
492, 782
van der Hooft, F., Heemakerk, M.H.M., Alberts, F., & van Paradijs, J. 1998, A&A, 329, 538
van Paradijs, J., & McClintock, J.E. 1995, in X-ray Binaries, p.58, eds Lewin, W.H.G., van Paradijs
J., van den Heuvel, E.P.J., Cambridge Univ. Press, Cambridge.
Vrtilek, S.D., McClintock, J.E., Seward, F.D., Kahn, S.M., & Wargelin, B.J. 1991, ApJS, 76, 1127
Wagner, R.M., Starreld, S.G., Hjellming, R.M., Howell, S.B., & Kreidl, T.J. 1994, ApJ, 429, L25
Wagner, R.M., Bertram, R., Starreld, S.G., Howell, S.B., Kreidl, T.J., Bus, S.J., Cassatella, A.,
& Fried, R. 1991, ApJ, 378, 293
Wu, C.-C., Aalders, J.W.G., van Duinen, R.J., Kester, D., & Wesselius, P.R. 1976, A&A, 50, 445
Wu, C.-C., Panek, R.J., Holm, A.V., Schmitz, M., & Swank, J.H. 1983, PASP, 95, 391
This preprint was prepared with the AAS L A T E X macros v5.0.

{ 15 {
TABLE 1
Previous Quiescent Observations of Black Hole X-ray Novae
Source Date Instrument NH  Luminosity Distance References
(10 22 cm 2 ) (10 33 erg s 1 ) (kpc)
V404 Cyg 1992 Nov ROSAT 2.29 a 6 y 8.1 (0.1{2.4 keV) 3.5 1
2.1 a 4:0 +1:9
1:5 1.1 (0.7{2.4 keV) 2
1994 May ASCA 1:1 +0:3
0:4 2:1 +0:5
0:3 1.20 (1{10 keV) 3
1996 Sept BeppoSAX 1.0 (xed) 1:9 +0:6
0:3 1.04 (1{10 keV) 4
A0620{00 1992 Mar ROSAT 0.16 (xed) 3:5 +0:8
0:7 0.004 (0.4{1.4 keV) 1.0 2
1994 Mar ASCA 1.6 (xed) 2 (xed) < 0:008 (0.5{10 keV) 5
GRO J1655{40 1996 Mar ASCA < 0:3 0:7 +2:1
0:4 0.3 (0.5{10 keV) 3.2 5
XTE J1550{564 2000 Aug & Sep Chandra 0:85 +2:2
2:4 2:3  0:4 0.67 (0.5{7 keV) 2.5{6.3 6,7
NOTES | y Uncertainty not given. For XTE J1550{564, the luminosity is based on a distance of 4 kpc.
(1) Wagner et al. 1994; (2) Narayan et al. 1996; (3) Narayan et al. 1997a; (4) Campana et al. 2001; (5) Asai et al. 1998; (6)
Tomsick et al. 2001; (7) Orosz et al. 2001

{ 16 {
TABLE 2
Best-fitting Spectral Parameters
Source Model NH  kT=kTRS a  2
 =dof (prob) CASH Flux c
(10 21 cm 2 ) (keV) M-C Prob b
V404 Cyg Power-law 6:98  0:76 1:81  0:14 0.92/45 (0.63) 1.42
Bremsstrahlung 6:04 +0:60
0:55 6:68 +2:49
1:50 0.94/45 (0.57) 1.40
Raymond-Smith 5:82 +0:56
0:50 7:54 +2:70
1:43 1.11/45 (0.28) 1.57
Blackbody 2:30  0:42 0:81  0:04 2.09/45 (0.00002) 1.26
Power-law 5.40 (xed) 1:55  0:07 1.20/46 (0.17) 1.47
Bremsstrahlung 5.40 (xed) 8:66  2:13 1.0/46 (0.46) 1.42
Raymond-Smith 5.40 (xed) 8:89  1:57 1.13/46 (0.25) 1.57
Blackbody 5.40 (xed) 0:69  0:03 3.49/46 (10 14 ) 1.15
A0620{00 Power-law 2:37 +1:14
1:04 2:19  0:50 0.71/11 (0.73) 0.78 0.018
Bremsstrahlung 1:52 +0:72
0:67 3:11 +3:59
1:17 0.75/11 (0.69) 0.74 0.018
Raymond-Smith 1:05 +0:57
0:50 5:46 +6:51
2:07 1.03/11 (0.42) 0.48 0.022
Blackbody 0 d 0:57 +0:06
0:07 1.58/11 (0.10) 0.10 0.017
Power-law 1:94  0:28 (xed) 2:07 +0:28
0:19 0.71/12 (0.74) 0.75 0.018
Bremsstrahlung 1:94  0:28 (xed) 2:55 +1:44
0:73 0.78/12 (0.67) 0.68 0.016
Raymond-Smith 1:94  0:28 (xed) 4:15 +2:66
1:30 1.38/12 (0.17) 0.14 0.023
Blackbody 1:94  0:28 (xed) 0:30  0:03 2.39/12 (0.004) 0.00 0.009
GRO J1655{40 Power-law 8:59 +6:19
4:52 1:70 +0:88
0:78 0.83/9 (0.59) 0.66 0.017
Bremsstrahlung 7:72 +5:11
3:46 8:40 +1
5:73 0.83/9 (0.58) 0.66 0.016
Raymond-Smith 7:18 +4:23
2:97 12:24 +1
8:61 0.85/9 (0.56) 0.63 0.019
Blackbody 3:03 +3:47
2:13 0:88 +0:29
0:18 0.94/9 (0.49) 0.57 0.012
Power-law 6:66  0:57 (xed) 1:47  0:40 0.75/10 (0.67) 0.60 0.016
Bremsstrahlung 6:66  0:57 (xed) 13:21 +1
8:98 0.75/10 (0.68) 0.64 0.015
Raymond-Smith 6:66  0:57 (xed) 17:15 +1
11:35 0.77 /10 (0.65) 0.62 0.018
Blackbody 6:66  0:57 (xed) 0:76 +0:14
0:12 1.07/10 (0.38) 0.39 0.012
XTE J1550{564 Power-law 8:73 +2:42
2:93 2:28 +0:47
0:64 1.27/13 (0.22) 0.22 0.16
Bremsstrahlung 6:93 +2:13
1:85 3:36 +4:75
1:33 1.26/13 (0.23) 0.24 0.15
Raymond-Smith 6:50 +1:97
1:62 4:38 +4:31
1:57 1.22/13 (0.25) 0.21 0.18
Blackbody 3:04 +1:80
1:49 0:69 +0:11
0:09 1.39/13 (0.16) 0.18 0.14
Power-law 3:90  0:60 (xed) 1:35  0:25 1.68/14 (0.05) 0.02 0.17
Bremsstrahlung 3:90  0:60 (xed) 12:56 +1
7:18 1.59/14 (0.07) 0.04 0.15
Raymond-Smith 3:90  0:60 (xed) 10:31 +1
5:16 1.61/14 (0.07) 0.03 0.16
Blackbody 3:90  0:60 (xed) 0:65 +0:08
0:06 1.39/14 (0.15) 0.15 0.12
NOTES | All quoted uncertainties are 90% condence.
Except for V404 Cyg, the best-t parameters and uncertainties are based on the CASH statistic. The reduced  2
values were obtained in a separate analysis using the  2 statistic.
a Thermal bremsstrahlung, blackbody or Raymond-Smith temperature (solar abundance)
b For A0620{00, GRO J1655{40, and XTE J1550{564, we list one minus the probability that the best t model
would produce a lower value of the CASH statistic than that calculated from the data, as determined via XSPEC
Monte-Carlo simulations. A low probability indicates a poor t.
c Absorbed ux in 0.3{7 keV (10 12 erg cm 2 s 1 )
d NH hit the minimum value of 0 allowed by XSPEC

{ 17 {
TABLE 3
Time-resolved Spectral Parameters
NH   2
 =dof Luminosity a
(10 21 cm 2 )
1 10:21  2:28 2:41  0:41 1.21/11 8.07
2 5:08  1:19 1:57  0:26 0.88/16 2.81
3 6:14  1:97 1:72  0:40 0.41/7 4.83
4 6:47  1:09 1:60  0:20 0.85/28 2.81
5 2:91  1:29 1:14  0:40 1.17/6 4.26
6 5:86  1:18 1:56  0:24 0.85/21 3.82
7 11:08  2:15 2:22  0:31 1.11/17 5.32
NOTES | a Luminosity in 0.3{7 keV (10 33 erg s 1 ),
assuming a distance of 3.5 kpc

{ 18 {
Fig. 1.| Upper panel: The Chandra spectrum of V404 Cyg with an absorbed power-law model
( = 1:81 and NH = 6:98  10 21 cm 2 ). Lower panel: residuals after subtracting the t from the
data in units of 1.

{ 19 {
V404 Cyg
GRO J1655-40
A0620-00 A0620-00
V404 Cyg
GRO J1655-40 XTE J1550-564
XTE J1550-564
Raymond-Smith model
Power-law model
Fig. 2.| Left: Contour plot for the column density (NH ) and photon index () derived from
the Chandra spectrum of V404 Cyg, A0620{00, GRO J1655{40 and XTE J1550{564. The cross
in the center marks the best t parameters and the contours encompass the 68%, 90% and 99%
condence levels. Vertical dashed lines show the optically determined NH . Right: Contour plot
for the column density (NH ) and Raymond-Smith temperature (kT RS ) derived from the Chandra
spectrum of V404 Cyg, A0620{00, GRO J1655{40 and XTE J1550{564. The cross in the center
marks the best t parameters and the contours encompass the 68%, 90% and 99% condence
levels. Except for V404 Cyg, all of the plots were derived using CASH statistic. Vertical dashed
lines show the optically determined NH .

{ 20 {
0 2000 4000 6000 8000
0.1
0.2
0.3
Elapsed Time (s)
1 2 3 4 5 6 7
Fig. 3.| Chandra ACIS-S 10 ksec light curve of V404 Cyg in the 0.3{7.0 keV band. The time
resolution is 500 s. Also shown are the seven time intervals used for time-resolved spectral analysis.

{ 21 {
Fig. 4.| Plot of power-law photon index () against absorption column density (NH ). A positive
correlation can be seen.

{ 22 {
Fig. 5.| Quiescent X-ray and Bolometric Fluxes of BHXN, after Bildsten & Rutledge (2000).
X-ray uxes are as reported herein or from G01, but in all cases converted to 0.4{2.4 keV emitted
uxes (note that Figure 2 of Narayan, Garcia and McClintock (2001) plotted a 0.5{10.0 keV ux for
V404 Cyg but agrees with this plot in all other respects). In cases where the spectrum cannot be
determined, we have assumed  = 2. Bolometric uxes are from Bildsten & Rutledge (2000) or as
reported herein. An \X" indicates that the X-ray ux is unlikely to be due to a stellar corona, An L
(S) indicates that it is the X-ray luminosity (spectrum) that argues against this coronal hypothesis.
Based on the data herein, in only one (GS 2000+25) of the six BHXN studied could the corona of
the secondary produce a signcant part of the detected X-ray ux.

{ 23 {
Fig. 6.| Quiescent luminosities of BHXN (lled circles and triangles) and NSXN (open circles) after
G01. Data points in triangle are from the results of this work. Only the lowest quiescent detections
(except for V404 Cyg which we show both the lowest detection and the luminosity derived here)
or Chandra upper limits are shown. The non-hashed areas represent common orbital periods for
BH and NSXN. The BHXN shown are, from left to right: GRO J0422+32, A0620{00, GS 2000+25,
4U 1543{47, XTE J1550{564, GRO J1655{40 and V404 Cyg.