Discovery of mHz X-ray Oscillations in a Transient Ultraluminous X-ray Source in M82

0.15 MB·English

by  Hua Feng

Accepted for publicationin The Astrophysical JournalLetters
PreprinttypesetusingLATEXstyleemulateapjv.11/10/09
DISCOVERYOFMHZX-RAY OSCILLATIONSINATRANSIENT ULTRALUMINOUS X-RAYSOURCEINM82
Hua Feng1, Fengyun Rao1, & Philip Kaaret2
Accepted for publication in The Astrophysical Journal Letters
ABSTRACT
We reportthe discoveryofX-rayquasi-periodicoscillations(QPOs)atfrequencies of3-4mHz from
a transient ultraluminous X-ray source (ULX) X42.3+59 in M82. The QPOs are strong and broad
0
and appear with weak or absent red noise, and are detected only in Chandra observations when the
1 source is brighter than 1040 ergs s−1. The QPO behavior is similar to the type A-I QPOs found in
0 XTE J1550−564,which is a subclass of low frequency QPOs with properties in between type A and
2
B. Therefore, we identify the QPOs in X42.3+59 as of type A or B, and rule out the possibility of

n type C. With this identification, the mass of the black hole in X42.3+59 can be inferred as in the
a range of 12,000-43,000 solar masses by scaling the QPO frequency to that of the type A/B QPOs
J in stellar mass black holes. Cool disk emission is detected in one Chandra observation, and the disk

0 inner radius suggests a similar black hole mass range. Black holes of such a high mass are able to
2 produceanenergyoutputinamannersimilartoX42.3+59byaccretingfromtheinterstellarmedium
directly.

] Subject headings: black hole physics — accretion, accretion disks — X-rays: binaries — X-rays:
E
individual (CXOM82 J095551.1+694045=X42.3+59)
H
.
h
1. INTRODUCTION example, the type C QPOs in GRS 1915+105 are de-
p
tected at frequencies of as low as 1 mHz (Morgan et al.
- Ultraluminous X-ray sources (ULXs) are variable X-
o 1997). QPOs have been detected in two ULXs, and are
ray sources which are not coincident with the nucleus of
r arguedto be of type C (Strohmayer & Mushotzky 2003;
t the galaxy and exhibit luminosities over the Eddington
as limit ofa20 M⊙ blackhole (3×1039 ergss−1)assuming Satrmoahnmnaeyrersiemtilaalr. 2t0o07ty)pbeecCauQsePtOhseyinapGpaelaarcttiocvBaHryBisn.
isotropicemission. Theyaretoobrighttobepoweredby
[ Scalingthe compactobjectmasswiththeoscillationfre-
accretionontostellarmassblackholeslikeGalacticblack
quency indicates the presence of IMBHs in these two
1 holebinaries(BHBs)andarethuscandidatesofinterme-
ULXs (Feng & Kaaret2007b; Strohmayer & Mushotzky
v diate mass black holes (IMBHs). However, if the radia-
6 tionissuper-Eddingtonorbeamedalongthelineofsight, 2009).
5 IMBHs are not required (King et al. 2001; Begelman M82 is a starburst galaxy at a distance of 3.63 Mpc
4 2002). Modeling the X-ray spectra of ULXs could shed (Freedman et al. 1994). The X-ray source X42.3+59
3 lightontheir masses. Forexample,the temperatureand was identified as a transient ULX in M82 from multi-
. pleX-rayobservationswithChandra(Kaaret et al.2006;
1 size of the accretion disk can be used to weigh the cen-
Feng & Kaaret2007b;Kong et al.2007). Feng & Kaaret
0 tral compact object (Makishima et al. 2000). However,
(2007b) argued that X42.3+59 is more likely to be an
0 identification and quantification of the disk emission in
intermediate mass black hole than a stellar mass ob-
1 the energy spectrum, especially when it is the domi-
ject accreting from a massive star according to its tran-
: nant component, has confronted difficulties and been
v sient nature (Kalogera et al. 2004). It lies on the sky
provedunreliableinafewcases(Gon¸calves & Soria2006;
i plane at 5′′ to another ULX X41.4+60, which at most
X Feng & Kaaret 2007a, 2009). In contrast, characteristic
times is the brightest source in M82, and thus Chandra
time scales of the X-ray emission, such as quasi-periodic
r
is the only telescope able to spatially resolve them in
a oscillations(QPOs)andthefrequencybreakinthepower
X-rays. Kaaret et al. (2006) reported significant timing
spectrum, could be used to determine the mass of com-
noise near 1 mHz from X42.3+59from one Chandra ob-
pact objects via a model independent calibration (cf.
servation. Here, using new Chandra data, we report the
McHardy et al. 2006; Casella et al. 2008).
discovery of mHz QPOs in X42.3+59 and discuss their
Low frequency QPOs from mHz to several tens Hz
possible nature.
havebeenfoundinGalacticBHBs. Onthe basisoftheir
properties like the coherence, amplitude, phase lag, and 2. OBSERVATIONSANDANALYSIS
harmonic component, low frequency QPOs can be clas-
We checked all archival Chandra and XMM-Newton
sified into three types of A, B and C (Wijnands et al.
observations of M82 to date, as well as three joint
1999; Remillard et al. 2002). Type A/B QPOs appear
Chandra/XMMobservationsthatweperformedrecently
inarelativelynarrowfrequencyrangeatafew Hz,while
(PI: H. Feng), to search for timing noise of X42.3+59
type C QPOs vary in a wide frequency range in re-
at low frequencies. Among these observations, QPO
sponse to the spectral parameters of the source. For
features at a few mHz are found in five observations
(Chandra ObsIDs 6097/8190/10027 and XMM ObsIDs
1Department of Engineering Physics and Center for Astro-
0112290201/0560590101), two out of which, 10027 and
physics,TsinghuaUniversity,Beijing100084, China
2DepartmentofPhysicsandAstronomy,UniversityofIowa,Van 0560590101,were made simultaneously with an over lap
AllenHall,IowaCity,IA52242, USA of 17 ks.
2 Feng, Rao, & Kaaret
TABLE 1
Spectralandtimingpropertiesof X42.3+59.
Instrument XMM Chandra Chandra Chandra XMM
(a) (b) (c) (d) (e)
ObservationID 0112290201 6097 8190 10027 0560590101
Observationdate 2001-05-06 2005-02-04 2007-06-02 2008-10-04 2008-10-03
Exposure(ks) 27 53 53 19 30
RT (counts s−1) 2.59 0.20 0.31 0.29 3.27
RS/RT 0.7 1.0 1.0 1.0 0.2
Spectral Properties
NH (1022 cm−2) 3.31±0.19 3.22±0.15 3.19±0.28
Γ 1.44±0.09 1.31±0.07 1.33±0.13
fX (10−12 ergs−1 cm−2 s−1) 4.49±0.10 7.25±0.12 6.19±0.19
LX (1040 ergss−1) 2.35−2.47 1.13±0.04 1.76±0.05 1.51±0.08
χ2/dof 80.7/74 87.1/74 83.6/74
QPOProperties
Significance(σ) 3.00 3.39 6.51 4.49 2.91
Frequency(mHz) 3.98±0.18 2.89±0.35 3.58±0.21 2.77±0.33 2.97±0.33
FWHM(mHz) 1.0±0.6 1.9±0.6 2.6±0.5 2.0±0.6 1.5±0.4
rms/mean(%) 5.0±0.9 10.2±1.6 14.1±1.1 17.4±2.0 14.4±2.6
Note. —RT isthetotalcountrateinthesourceextraction region,RS/RT isthefractionofsourcephotons neededforthecalculation
oftheQPOamplitude,NH istheabsorptioncolumndensity,Γisthepower-lawphotonindex,fX istheobservedfluxin1-8keV,andLX
is the luminositycorrected forabsorption in1-8 keV. Errorsarequoted at a confidence of 90% for spectral parameters and 68% (1σ) for
timingparameters.
factor1.5amongthethreeobservations. Allspectralpa-
TABLE 2 rameters with 90% errorsare listed in Table 1. We tried
Spectralfittingwith a disk pluspower-law model subject
to add an additional multicolor disk component to the
to absorption forobservation 8190.
power-law spectrum. Pronounced improvement on the
Parameter Value fits is only obtained for observation 8190; the disk com-
ponent is favored at a significance of 4.1 σ. The best-fit
TNiHn 0(4.1.27±±00..40)3×ke1V022 cm−2 model parameters are listed in Table 2 with 90% errors
Rin 3.5+−31..09×104 km quoted. Power spectra were calculated from lightcurves
Γ 1.53±0.12 created using events from the same region for spectral
NPL (2.2±0.4)×10−3 analysis with a time step equal to the CCD frame time
fX (7.14±0.13)×10−12 including the readout time, which is 0.44104 s for the
LX (2.7±0.7)×1040 three observations. Each lightcurve was divided into M
diskfraction 27%
segments with M = 25, 18, and 8 for observations 6097,
χ2/dof 65.6/72
8190, and 10027, respectively. A Fourier transform was
Note. —NH andΓarethe sameas inTable1. Tin isthedisk performed for each segment and the individual power
inner temperature, Rin is the disk inner radius calculated assum- spectra were averagedand then geometricallybinned by
ingaface-ondiskatadistanceof3.63Mpc,NPL isthepower-law
normalizationinphotonskeV−1 cm−2 s−1 at1keV,fX istheab- afactorof1.2,1.1,and1.3,respectivelyforthethreeob-
sorbedfluxinergs−1 cm−2 s−1,LX istheunabsorbedluminosity servations. The binning factor is chosenfor a reasonable
inergss−1. Theflux,luminosity,andthediskfractionarequoted frequency resolutionaroundthe QPOsand a sufficiently
in1-8keV.Allerrorsarequotedat90%confidence.
large binning factor needed for the fitting. We checked
different energy bands and found that the most signifi-
For the three Chandra observations, the location of cant timing noise appears in the 1-8 keV range.
X42.3+59onthechargecoupleddevice(CCD)wasmore For the two XMM observations, the lightcurves were
than 3.5′ off the optical axis – where a point source produced using merged PN and MOS events in their
spreads onmultiple pixels and photon pileup for sources common good time intervals (GTIs) from a half circu-
at the flux level of X42.3+59 is unimportant. Event lar region (region B in Fig. 2 of Feng & Kaaret 2007b,
lists were created using CIAO 4.1.2 with CALDB 4.1.3. where the contaminationfrom the other ULX X41.4+60
Sourceenergyspectrainthe1-8keVbandwereextracted is minimized) with a time step 10 times the PN frame
fromthesourceregionandgroupedbyafactorof4in1-4 time ofabout0.73s. Fourier transformswereperformed
keV, 8 in 4-6 keV, and 16 in 6-8 keV based on the origi- in each GTI with 4096 points, and the power spectrum
nalchannels,with backgroundsubtractedfromanearby wasobtainedbyaveragingallindividualones. Thenum-
source-freeregionatthesamelocationontheskyforthe ber of segments, M, is 8 and 9 and the geometric bin-
three observations. A power-law model subject to inter- ning factor is 1.1 and 1.2, respectively for 0112290201
stellar absorption provides adequate fits to the Chandra and 0560590101. For observation 0560590101, MOS1
spectrum. Theabsorptioncolumndensityisalmostcon- data were not included because they contain too many
stant at about 3.3×1022 cm−2. The source spectrum timing gaps. Energy spectra of X42.3+59 are not avail-
is hard, with a power-law photon index of about 1.3- able from XMM data due to considerable photon con-
1.4. The luminosity correctedfor absorptionvaried by a fusion with nearby sources. Spectral information for
Discovery of mHz QPOs in X42.3+59 in M82 3
For the simultaneous Chandra and XMM observations,
the same QPO feature is detected by both. The XMM
power spectra are substantially contaminated by pho-
tons from the nearby variable ULX X41.4+60, and the
calculated QPO parameters may be affected. The pho-
tonfractionof X42.3+59in the extractionregionis esti-
mated and taken into account when calculating the rms
amplitude of QPOs from the XMM data. All QPO pa-
rametersare listed in Table 1, and the QPO significance
is calculated based on the χ2 variation on the normal-
ization of the QPO component. We checked the power
spectra at even lower frequencies for the three Chandra
observations, and found possible red noise only in 6097
appearing at frequencies well below the QPO frequency.
Forobservation8190,the power spectrumbetween 0.01-
1 mHz is dominated by the white noise. For observation
10027, which is shorter than the other two, the powers
between 0.1-1 mHz are also white noise dominated.
There are several other Chandra Advanced CCD
Imaging Spectrometer (ACIS) observations in which
X42.3+59wasactive(seeFeng & Kaaret2007b)butthe
mHz QPOs were not detected. Observation 5644 has
an exposure of 70 ks and the source luminosity was
9 × 1039 ergs s−1 in 1-8 keV. The number of photons
from X42.3+59 detected in 5644 is as many as in 6097,
andtwicethatin10027;thenondetectionofQPOsisnot
due to low statistics. Observation 2933 and 6361 both
haveanexposureof18ks andthe sourceluminosity was
4× 1039 and 8× 1039 ergs s−1, respectively. Another
four ACIS observations (378, 379, 380-1 and 380-2) are
Fig.1.—X-rayPowerspectraofX42.3+59inthe1-8keVenergy shorter than 10 ks, and the source was slightly dimmer
range. The letters indicates the observation as shown in Table 1.
than in observation 2933. The mHz QPOs were not de-
ThedottedlineindicatestheGaussianplusconstantmodelthatfit
thespectrum. ThepowersarenormalizedafterLeahyetal.(1983) tectedinanyobservationwhenthesourcewasdim,below
andthebaselinesofb,c,anddareshiftedverticallyforclarityby ∼1040 ergs s−1.
5,10,and15,respectively.
3. DISCUSSION
XMM 0560590101 can be obtained from the simultane- Low frequency QPOs in stellar mass black hole X-ray
ous Chandra observation 10027. For XMM 0112290201, binaries have a diverse population distributed in a wide
which does not have a joint Chandra observation, the frequencyrange. Therefore,inferenceofthecompactob-
count rate of X42.3+59 has been resolved from surface jectmassviatheQPOfrequencyshouldbedonebetween
brightnessfitting(Feng & Kaaret2007b). Therearefour QPOs of the same type. The QPO coherence parameter
reliable spectral measurements of X42.3+59 with Chan- Q, defined as the ratio of the central frequency to the
dra (the three ones reported here and observation 5644 full width at half maximum (FWHM), is around 1.4-1.5
reported in Feng & Kaaret 2007b). Taking into account derivedfrom the three Chandra observationsand nearly
the spectral variations of X42.3+59 with an absorption 4 from the XMM observation 0112290201. The rms am-
column density in the range of (3.2−3.4)×1022 cm−2 plitude of the QPOs in the three Chandra observations
anda power-lawindex in1.3-1.5,the 1-8keVluminosity is high, from 10% to 17%, but is relatively low in the
ofX42.3+59inXMM0112290201canbeestimatedusing XMM observation 0112290201 of 5%. We note that the
PIMMS as in the range of (2.35−2.47)×1040 ergs s−1. Chandra results should be more reliable due to less con-
Prominenttiming noiseabovethe Poissonlevelcanbe tamination. As mentioned above, possible red noise is
seen around a few mHz in the power spectra from these only found in Chandra 6097, and the power spectrum is
observations(Figure1),whichweinterpretasQPOs. We totallydominatedbythewhitenoiseintheothertwoob-
fitthepowerspectrainthewholefrequencyrangewitha servations at frequencies down to 0.1 or even 0.01 mHz.
a Gaussianplus constantmodel, which is overplottedon Therefore, the QPOs in X42.3+59 can be summarized
each spectrum with a dotted line. A Lorentzianalso fits as being broadand strong,andoccurringwith absentor
these peaks, but not as well as the Gaussian does. No weak red noise continuum.
obvious broadband red noise such as a (broken) power- The QPOs in X42.3+59 are similar to four QPOs de-
law continuum is found near the QPO feature. We rule tected in XTE J1550−564 on 1999 Mar 18 and 21 and
out the possibility that they are caused by instrumental Apr 02 and 03, respectively, which also have a low Q
oscillationsliketheditheringasfollows. FortheChandra of about 1.0, an rms amplitude of about 10%, and ac-
observations, photons from nearby regions do not show companiedwithaweakrednoise(Remillard et al.2002).
similar oscillations and simulations with MARX using Another similarity is that three out of the four QPOs in
thesameaspectsolutionandbadpixelfilesconfirmsthat XTE J1550−564do not exhibit a Lorentz profile. These
noinstrumentationsignalispresentatthesefrequencies. four QPOs were classified as ‘possible’ type A QPOs by
4 Feng, Rao, & Kaaret
Remillard et al. (2002), or type A-I (a subclass of type (Belloni et al. 2005), also known as the steep power-
A)byHoman et al.(2001)todistinguishthemfromweak law state defined by Remillard & McClintock (2006). In
type A QPOs. Casella et al. (2005) argued that type A- XTE J1550−564,the four type A-I QPOs were found in
I QPOs should be classified as type B based on their the steep power-law state as well (Remillard et al. 2002;
frequency and amplitude. However, type B QPOs are Sobczak et al. 2000). However, the X-ray spectrum of
usually narrow, with a Q value of & 6 concluded from X42.3+59is hard, seemingly inconsistent with the steep
a few BHBs (Casella et al. 2005). Therefore, the type power-lawstate. Diskemissionisdetectedinobservation
A-I could be a specialsubclass amid type A andtype B. 8190, with a moderate fraction of 27% in 1-8 keV. This
In spite of the controversyover the classification of type diskcomponentwillcontribute40%tothetotalemission
A-I, we identify the QPOs found in X42.3+59as of type below 1 MeV if the power-law component is not cut off.
A-I based on their similar profile and the weak/absent In the hard state, the disk contributes less than 25% in
red noise, or conservatively, of type A/B according to 2-20 keV for stellar mass black holes. For more massive
the discussion above. As noted above, the mHz QPOs black holes, though there is no defined energy range, a
weredetectedonlywhenthe sourcewasluminous,above disk fraction of 27%-40% seems too high for being clas-
∼1040 ergs s−1. Type A/B QPOs show identical be- sified as in the hard state. A varying disk component
havior that they appear only when the source flux is and an energetic power-law component suggest that the
high enough (Casella et al. 2004). This further supports source may be in a state similar to the steep power-law
the identification of the mHz QPOs as of type A/B. By state in stellar mass black holes. Interestingly, the best-
any means, the QPOs in X42.3+59are not analogousto fit disk inner radius is consistent with the gravitational
type C, which are narrow, strong, and accompanied by radius of a black hole of 11,000-44,000 M⊙, which is in
a strong red noise continuum; otherwise, the following agreement with the mass estimated from the QPO fre-
discussion would be different as type C QPOs appear at quency.
a frequency range significantly wider than others. X42.3+59 lies about 5′′ from the kinematic center of
The four type A-I QPOs in XTE J1550−564 have a M82 (Weliachew et al. 1984) on the sky plane. Follow-
central frequency varying between 7.1 to 10.3 Hz; all ing the discussion in Kaaret et al. (2001) and references
type A QPOs in XTE J1550−564 are found in a fre- therein, an upper limit of the compact object mass can
quencyrangeof4.88-10.3Hz,whichcoversthefrequency be placed of about 3×104 M⊙ such that the black hole
range of type B QPOs (4.94-6.1 Hz) in the same source will not fall onto the nucleus of the galaxy due to dy-
(Remillard et al. 2002). The QPOs in X42.3+59 have namical friction if the black hole was born at the same
a central frequency between 2.77-3.98 mHz. Comparing time of the galaxy. This upper limit is consistent with
with the type A-I QPOs in XTE J1550−564, the upper our mass estimate via QPOs.
andlowerboundsofthefrequencyrangehaveexactlythe X42.3+59 is coincident with an extended radio source
same ratio but scaled by a factor of about 2600. Based known as 42.21+59.2, which is believed to be an H II
on the assumption of type A-I QPOs, the compact ob- region (McDonald et al. 2002; Kaaret et al. 2006) asso-
ject mass of X42.3+59 can be inferred as 25,000-30,00 ciatedwithastarclusterseenfrominfrared(Kong et al.
M⊙ after adopting a mass of 9.68-11.58 M⊙ for XTE 2007). Therefore, it could be ruled out that the source
J1550−564(Orosz et al. 2002). A more conservative es- is a background AGN. If the disk component is true, its
timate taking into account all type A/B QPOs leads to highbolometricluminosityof1.2×1041ergss−1 andlow
a black hole mass of 12,000-43,000M⊙ in X42.3+59,in- temperature make it a goodcandidate for the ionization
ferred from the possibly smallest and largest ratios be- source of the H II region.
tween QPO frequencies from the two sources. The outburst luminosity of the source is about
Type A/B QPOs are detected in several other BHBs 1040 ergs s−1. In the total 20 observations of the source
besides XTE J1550−564, including GX 339−4, XTE (16 reported in Feng & Kaaret 2007b, plus Chandra
J1859+226, and GS 1124−684 (see Casella et al. 2005, 8190,10027,10025,and10026),X42.3+59wasfoundac-
and references therein). In these sources, the type A/B tivein12ofthem,indicativeofadutycycleǫ=0.6. As-
QPOsshowupinasimilarfrequencyrange,e.g.4.5-7.49 suminganaccretionefficiencyη =0.1,themeanmassac-
HzinGX339−4(Belloni et al.2005)and4.45-8.42Hzin cretionrateofthe sourceisfound tobe M˙ =ǫL /ηc2 =
X
XTEJ1859+226(Casella et al.2004). TypeBQPOsare 7×1019 g s−1. If the source is accreting from the inter-
alsodetectedinGRS1915+105varyinginthe frequency stellar medium directly, the mass accretionrate in g s−1
rangeof2.44-6.84Hz(Soleri et al.2008). Theyarebroad can be written as
but shown above a strong red noise component, par-
tially similar to the QPOs in X42.3+59. These sources M˙ =1.4×1011 M 2 ρ cs −3
also contains a similar massive black hole as in XTE (cid:18)M⊙(cid:19) (cid:18)10−24gcm−3(cid:19)(cid:18)10kms−1(cid:19) ,
J1550−564,but their masses are not as well constrained
as in the latter (Remillard & McClintock 2006). There- where M is the black hole mass, ρ is density and c is
s
fore,inferringthemassofX42.3+59viatypeA/BQPOs the speed of sound of the medium (Frank et al. 2002).
fromthese sourceswillleadto consistentbut loosercon- Adoptingcanonicalvaluesofthedensityandsoundspeed
straints. We note that all above estimates of the mass shown in the equation, a black hole mass of 2×104 M⊙
depend on a hypothesis that the frequency of type A/B isrequiredtopowertheX-rayluminositywithoutacom-
QPOsis linearlyscaledwith the blackhole mass,which, panionstar. Thismeansthatifthe massestimatedfrom
however, has never been tested. More work regarding the QPOs are reliable, the source does not have to be in
the mass dependence of type A/B QPOs is needed. a close binary system, but is massive enough to accrete
In GX 339−4, all type A/B QPOs were detected from the interstellar medium directly. Pellegrini (2005)
when the source was in the soft intermediate state suggests that the Bondi rate may overestimate the true
Discovery of mHz QPOs in X42.3+59 in M82 5
accretion rate for AGNs by 1-2 orders of magnitude. In and the mission planning teams of Chandra and XMM-
this case, a corresponding higher black hole mass would Newton for making the observations possible. HF ac-
be required. Conversely,if the ULX indeed lies within a knowledges funding support from the National Natural
starcluster,thegasdensitycouldbesubstantiallyhigher, Science Foundation of China under grant No. 10903004
whichwoulddecreasetherequiredblackholemass. Any- and 10978001, the 973 Program of China under grant
way, one should be cautious of the high uncertainty on 2009CB824800, and the Foundation for the Author
such an estimate. of National Excellent Doctoral Dissertation of China
under grant 200935. PK acknowledges partial sup-
port from Chandra grant GO9-0034X and NASA grant
Wethanktheanonymousrefereeforhelpfulcomments
REFERENCES
Begelman,M.C.2002,ApJ,568,L97 Leahy,D.A.,Darbro,W.,Elsner,R.F.,Weisskopf,M.C.,Kahn,
Belloni,T.,Homan,J.,Casella,P.,vanderKlis,M.,Nespoli,E., S.,Sutherland,P.G.,&Grindlay,J.E.1983,ApJ,266,160
Lewin,W.H.G.,&Miller,J.M.,&M´endez, M.2005, A&A,, Makishima,K.,etal.2000,ApJ,535,623
440,207 McDonald,A.R.,Muxlow,T.W.B.,Wills,K.A.,Pedlar,A.,&
Casella,P.,Belloni,T.,Homan,J.,&Stella,L.2004,A&A,426, Beswick,R.J.2002,MNRAS,334,912.
587 McHardy,I.M.,Koerding,E.,Knigge,C.,Uttley,P.,&Fender,
Casella,P.,Belloni,T.,&Stella,L.2005, ApJ,629,403 R.P.2006, Nature,444,730
Casella,P.,Ponti,G.,Patruno,A.,Belloni,T.,Miniutti,G.,& Morgan,E.H.,Remillard,R.A.,&Greiner,J.1997,ApJ,482,
Zampieri,L.2008,MNRAS,387,1707 993
Feng,H.,&Kaaret,P.2007a, ApJ,660,L113 Orosz,J.A.,etal.2002,ApJ,568,845
Feng,H.,&Kaaret,P.2007,ApJ,668,941 Pellegrini,S.2005,ApJ,624,155
Feng,H.,&Kaaret,P.2009,ApJ,696,1712 Remillard,R.A.,&McClintock,J.E.2006,ARA&A,44,49
Frank,J.,King,A.R.,&Raine,D.J.2002,AccretionPowerin Remillard,R.A.,Sobczak, G.J.,Muno,M.P.,&McClintock,J.
Astrophysics(3rded.;Cambridge:CambridgeUniv.Press) E.2002, ApJ,564,962
Freedman,W.L.etal.1994,ApJ,427,628 Sobczak, G.J.,McClintock,J.E.,Remillard,R.A.,Cui,W.,
Gon¸calves, A.C.,&Soria,R.2006,MNRAS,371,673 Levine,A.M.,Morgan,E.H.,Orosz,J.A.,&Bailyn,C.D.
Homan,J.,Wijnands,R.,vanderKlis,M.,Belloni,T.,van 2000a, ApJ,544,993
Paradijs,J.,Klein-Wolt,M.,Fender,R.,&M´endez, M.2001, Soleri,P.,Belloni,T.,&Casella,P.2008, MNRAS,383,1089
ApJS,132,377 Strohmayer,T.E.,&Mushotzky, R.F.2003,ApJ,586,L61
Kaaret,P.,Prestwich,A.H.,Zezas,A.,Murray,S.S.,Kim, Strohmayer,T.E.,&Mushotzky, R.F.2009,ApJ,703,1386
D.-W.,Kilgard,R.E.,Schlegel,E.M.,Ward,M.J.2001, Strohmayer,T.E.,Mushotzky, R.F.,Winter,L.,Soria,R.,
MNRAS,321,L29 Uttley,P.,&Cropper,M.2007, ApJ,660,580
Kaaret,P.,Simet,M.G.,&Lang,C.C.2006, ApJ,646,174 Weliachew,L.,Fomalont,E.B.,Greisen,E.W.1984,A&A,137,
Kalogera,V.,Henninger,M.,Ivanova, N.,&King,A.R.2004, 335
ApJ,603,L41 Wijnands,R.,Homan,J.,&vanderKlis,M.1999,ApJ,526,L33
King,A.R.,Davies,M.B.,Ward,M.J.,Fabbiano,G.,&Elvis,
M.2001,ApJ,552,L109
Kong,A.K.H.,Yang,Y.J.,Hsieh,P.-Y.,Mak,D.S.Y.,&Pun,
C.S.J.2007,ApJ,671,349