Discovery of a broad iron line in the black-hole candidate Swift J1753.5-0127, and the disk emission in the low/hard state revisited

0.37 MB·English

by  Beike Hiemstra

Mon.Not.R.Astron.Soc.000,1–9(2008) Printed21January2009 (MNLATEXstylefilev2.2)
Discovery of a broad iron line in the black-hole candidate
−
Swift J1753.5 0127, and the disk emission in the low/hard
state revisited.
9
0
0 Beike Hiemstra1⋆, Paolo Soleri2, Mariano M´endez1, Tomaso Belloni3,
2
n Reham Mostafa4 and Rudy Wijnands2
a 1KapteynAstronomical Institute, Universityof Groningen, P.O. Box 800, 9700 AV Groningen, The Netherlands
J 2Astronomical Institute Anton Pannekoek, Universityof Amsterdam, Kruislaan 403, 1098 SJ Amsterdam, The Netherlands

5 3INAF – Osservatorio Astronomico di Brera, Via E. Bianchi 46, I-23807, Merate (LC), Italy
1 4Department of Physics, Fayoum University,P.O. Box 63514, Fayoum, Egypt

]
E
H
.
h ABSTRACT
p We analyzed simultaneous archival XMM-Newton and RXTE observations of the X-
- ray binary and black hole candidate Swift J1753.5−0127. In a previous analysis of
o
the same data a soft thermal component was found in the X-ray spectrum, and the
r
t presenceofanaccretiondiskextendingclosetotheinnermoststablecircularorbitwas
s
a proposed. This is in contrast with the standard picture in which the accretion disk is
[ truncatedatlargeradiiin the low/hardstate.We tested a number of spectralmodels

and we found that severalof them fit the observedspectra without the need of a soft
1
disk-like component. This result implies that the classical paradigm of a truncated
v
accretion disk in the low/hard state can not be ruled out by these data. We further
5
5 discovereda broad iron emissionline between 6 and 7 keV in these data. From fits to
2 thelineprofilewefoundaninnerdiskradiusthatrangesbetween∼6–16gravitational
2 radii, which can be in fact much larger, up to ∼250 gravitationalradii, depending on
. the modelusedtofitthe continuumandthe line.We discusstheimplicationsofthese
1
results in the context of a fully or partially truncated accretion disk.
0
9 Key words: accretion disks, disk states – stars: individual (Swift J1753.5−0127) –
0
X-rays: binaries
:
v
i
X
r
a 1 INTRODUCTION ofthispaperwetreatSwiftJ1753 asablackholecandidate
(BHC).
Swift J1753.5 0127 (hereafter Swift J1753) was discovered
−
with the Swift Burst Alert Telescope on June 30 2005 as a Rossi X-ray Timing Explorer (RXTE) and INTE-
hard X-ray source (E > 15 keV; Palmer et al. 2005). Swift GRAL observations showed that Swift J1753 never left the
J1753 was also detected at energies below 10 keV with the low/hard state (LHS) during the whole outburst (Cadolle
Swift X-RayTelescope(XRT),andintheUVwiththeSwift Bel et al. 2007; Zhang et al. 2007). The LHS of X-ray bi-
UV/OpticalTelescope(Morrisetal.2005; Stilletal.2005). nariesischaracterized byahardpower-lawX-rayspectrum
Usingopticalspectroscopy,Torresetal.(2005)founddouble (photon index, Γ 1.7; e.g., M´endez & van der Klis 1997;
∼
peakedHαemission lines,whichindicatethatthesystemis Remillard&McClintock2006),whichisusuallyinterpreted
a low-mass X-ray binary. The source was also detected in astheresultofComptonization ofthermalphotonsfroman
radio,whichmightevidencejetactivity(Fenderetal.2005; optically thick accretion disk by hot electrons (e.g., Esin et
Cadolle Bel et al. 2006). Based on thehard spectral behav- al. 1997). Observations of sources with low absorption pro-
ior,Cadolle Beletal.(2005) proposedthatSwift J1753 isa videevidencethatintheLHStheaccretion diskiscooland
blackholecandidate.Although,itisgenerallyassumedthat istruncatedatalarge radius(e.g.,McClintock etal.2001).
SwiftJ1753containsablackhole,thereisyetnodynamical In the last years, a contribution of a jet to the X-ray emis-
confirmation of its nature.Bearing thisin mind,in therest sion in the LHS has been considered (Markoff et al. 2001;
Fender,Belloni & Gallo 2004).
During an outburst, the mass accretion rate onto the
black hole increases and a transition to the high/soft state
⋆ E-mail:[email protected] (HSS) can occur (for reviews, see Homan & Belloni 2005;
2 B. Hiemstra et al.
McClintock &Remillard 2006).Thehigh/soft stateisdom- for background flares. No flares are present and we used
inated by thermal emission below 5 keV, likely produced thewholeexposureforouranalysis.Wecheckedthefiltered
∼
by an optically thick and geometrically thin accretion disk eventfilesforphotonpile-upbyrunningthetaskepatplot.
with an inner radius at, or close to, the innermost stable Nopile-upisapparentinthePNandMOS1datasinceboth
circularorbit(ISCO;Shakura&Sunyaev1973). Duetothe cameras were operated in ‘timing’ mode. The MOS1 data,
increase of thermal photons, the Comptonizing plasma in however,showasharpdropinthehistograminwhichcounts
the system cools down and the spectrum of the source be- versus RAWX1 are plotted. This feature is due to a physi-
comes soft. A part of the Comptonized photons is reflected cal damage of some pixels in the MOS1 camera as a conse-
offtheaccretiondisk,andproducesanironKαlinebetween quenceof amicro-meteorite impact. Forthisreason, weex-
6and7keV(e.g.,Fabianetal.1989;Laor1991),wherethe cludedtheMOS1datafromouranalysis.TheMOS2camera
width of the emission line is related to the size of the inner was operated in ‘full-frame’ mode and suffered from mod-
disk radius. In this interpretation, if the disk is truncated eratephotonpile-up.Wethereforeexcludedthecentral22′′
at large radii in the LHS, no or narrow iron-emission lines region of the source point spread function (PSF) and con-
are expected, whereas the iron line would be broad in the firmed, using the task epatplot, that the remaining MOS2
HSS where the inner disk radius is close to the ISCO and dataarenotaffected bypile-up.TheODFfilesalsoinclude
therelativistic effects are strong. high resolution spectra at energies below 2 keV,taken with
Recent high-resolution spectral data of the BHC X- the Reflection Grating Spectrometer (RGS). Since the PN
ray binaries Swift J1753, GX 339 4 and XTE J1817 330, andMOScamerascovermostoftherangeoftheRGSspec-
− −
showed a soft thermal component in the spectra while the tra, we do not use theRGS data in thispaper.
sourceswereintheLHS(Milleretal.2006a,hereafterM06; We extracted source and background spectra applying
Miller et al. 2006b; Rykoff et al. 2007). It was suggested the standard procedures. The source spectrum for the PN
thatthesoftcomponentiscausedbythermalemissionfrom data is extracted using RAWX in [28:48], and the back-
an accretion disk which, even in the LHS, extends down to ground in [5:25]. For MOS2, we generated the source spec-
the ISCO. The same data set of XTE J1817 330 was re- trum using an annulus with inner radius of 22′′ and outer
analyzed by Gierlin´ski et al. (2008), who sho−wed that the radiusof68′′ centered on thesourcelocation. Weextracted
spectrum could also be fit without the need of a thermal theMOS2backgroundspectrumfromacircularregionwith
component. aradiusof160′′ located 10.5′ away from theposition of the
Here we analyzed a simultaneous archival XMM- source. We created photon redistribution matrices and an-
Newton/RXTE (the same data as reported in M06) ob- cillaryfilesforthesourcespectrausingtheSAStoolsrmfgen
servation of Swift J1753, and we tested whether a thermal and arfgen, respectively.
disk-likecomponentisrequiredtofittheX-rayspectra.We We rebinned thesource spectra using thetool pharbn2
explored a range of continuum models, beyond the models (M.Guainazzi,privatecommunication),suchthatthenum-
described in M06, with and without a disk component. We ber of bins per resolution element of the PN and MOS2
alsoreportonthepresenceofabroadFelineinthespectra. spectra is 3. We checked that after rebinning, all channels
In 2 we describe the extraction and reduction of the data, haveat least 15 countsin theenergy range used in ourfits.
§
in 3wepresenttheresultsoffittingthedatawithdifferent Wenotethatdifferentlyfromus,M06groupedthePNdata
§
continuum models, and in 4 we discuss the results of the to have at least 10 counts per bin, and they therefore over-
§
continuummodels and theline components with respect to sampled the intrinsic resolution of the PN data by a factor
theissue of a radially truncated accretion disk. of 8. The cross calibration between PN and MOS agrees
∼
to7% from 0.4 keVto12keV3,andwefoundthatonly be-
low 2 keV the systematic difference in the relative effective
areas of these cameras went up to 7%. We therefore added
2 DATA REDUCTION
a7%systematicalerrortothePNandMOS2spectrabelow
We analyzed archival observations of Swift J1753 simulta- 2 keV.
neouslytakenwithXMM-Newton andRXTE onMarch24,
2006, i.e. 270 days after the start of the outburst. XMM-
Newton st∼artedits 42ksobservationat16:00:31(UT),and 2.2 RXTE
∼
the RXTE observation started at 17:32:16 (UT). In Fig. 1
WereducedtheRXTE Proportional CounterArray(PCA)
we show the date of this observation on top of the All Sky
andHighEnergyX-rayTimingExperiment(HEXTE)data
Monitor (ASM) RXTE light curve of Swift J1753, starting
using the HEASOFT tools version 6.4. We applied stan-
from its outburst in 2005. In the next subsections we de-
dard filtering and obtained PCA and HEXTE exposures
scribetheprocedurethatwefollowed toextractthespectra
for each of thedifferent telescopes.
1 In timing mode, one of the spatial coordinates of the CCD is
compressed in order to increase the time resolution. For XMM-
2.1 XMM-Newton
Newtonthespatialcoordinatethatisnotcompressedinthismode
iscalledRAWX.
We reduced the XMM-Newton Observation Data Files
2 Thepharbntoolallowstorebintheenergychannelsofthespec-
(ODF) using SAS version 7.1.2 and we applied the latest
trafordifferentinstrumentstakingintoaccounttheinstrument’s
calibration files. We extracted the event files for the EPIC-
spectral resolution. It allows torebin both on minimumnumber
pn (hereafter PN) and both EPIC-MOS cameras (MOS1
ofbinsperresolutionelementandonminimumnumberofcounts
andMOS2)usingthetasksepprocandemproc,respectively. perbin
Weappliedstandardfilteringandexaminedthelightcurves 3 http://xmm2.esac.esa.int/docs/documents/CAL-TN-0018.pdf
Line and disk emission in Swift J1753.5−0127 3
16
14
12
c
e
s/s 10
nt
u 8
o
c
M 6
S
A 4
2
0
53600 53800 54000 54200 54400 54600 54800
Time (MJD)
Figure 1. RXTE/ASM light curve of Swift J1753 with daily points, starting at the outburst in 2005 (MJD 53551). The XMM-
Newton/RXTE observation, ∼270 days after the start of the outburst, is represented by the vertical bar. Note that 3 years after the
startoftheoutburst,thesourceisstillactive.
which are 2.3 ks and 0.8 ks, respectively. Using thetool multiplicative constants, were linked between the different
∼ ∼
saextrct, we extracted PCA spectra from PCU2 only, be- instruments.
ingthisthebestcalibrateddetectorandtheonlyonewhich
isalwayson.FortheHEXTEdatawegeneratedthespectra
InthefitresidualsofthePNandMOS2data,twonar-
usingcluster-Beventsonly,sinceafterJanuary2006cluster
rowabsorptionlinesnear1.8keVand2.2keVwerepresent.
A stopped rocking and could no longer measure the back-
This is most likely due to a mismatch in the calibration of
ground. We estimated the PCA background and measured
the edges in the EPIC instruments. To reduce the impact
the HEXTE background using the standard RXTE tools
of these features in our fit results, we fit Gaussian absorp-
pcabackest and hxtback, respectively. We further built in-
tion lines, which we found to be at an energy of 1.82 0.01
strumentresponsefilesforthePCAandHEXTEdatausing ±
keV and 2.25 0.03 keV, with widths of 0.23 0.15 eV and
pcarspandhxtrsp,respectively.Theenergychannelsofthe ± ±
2.0 1.9 eV, respectively.
RXTE spectra havemore than 20 countsperenergy bin so ±
thatχ2 statisticscanbeapplied,thereforenochannelrebin-
ningisrequired.FollowingM06,weaddeda0.6%systematic Thereappeartobeabroademissionlinebetween6and
errortothePCAspectrausingtheFTOOLgrppha,andwe 7 keV in the XMM-Newton/RXTE spectra. (This emission
did not add any systematic error to theHEXTE data. line is best seen in the PN spectra; MOS2 effective area is
a factor of 2 smaller than that of PN, and in this case
∼
the effective area has been reduced due to the excision of
thecentral part of the source PSF to avoid pile-up. On the
3 DATA ANALYSIS AND RESULTS
other hand, the spectral resolution of RXTE is 1 keV at
∼
For our spectral analysis of Swift J1753 we used XSPEC 6 keV.) Similar emission lines, probably due to iron, have
v11(Arnaud1996).Wesimultaneouslyfitthespectraofthe been reported in other BHC (e.g., Miller et al. 2002b) and
different instruments, where we restricted the PN and the activegalactic nuclei(e.g., Tanakaet al.1995). Weinitially
MOS2spectratoanenergyrangeof0.6–10keVand0.6–9.4 added a Gaussian component with a variable line width to
keV,respectively.ForthePCAandHEXTEspectraweused thespectrumtofitthisline,buttherealwaysremainedresid-
the energy ranges 3–20 keV and 20–100 keV, respectively. ualsaround7keVwhichwecouldonlyeliminatebyadding
To investigate the nature of the putative disk emission in at least another Gaussian to our model. On the contrary,
the LHS (as reported in M06), we tested several models to a Laor profile (Laor 1991) yields a good fit of the line and
describe thecontinuum spectrum of Swift J1753. therefore,intherestofthepaperweusethisrelativisticline
profilewheneverwefitalinetothespectra,althoughwealso
tried different relativistic line profiles as described in 3.8.
§
3.1 General fit procedure For the Laor component we allowed all the parameters to
befreetovary,excepttheouterdiskradius,whichwefixed
Eachofthespectralcontinuummodelswepresentbelowin-
to the default value of 400 R (R is the gravitational ra-
cludestheeffectofinterstellarabsorptioninthedirectionof g g
diusof theblack hole, definedasR =GM/c2, with G and
SwiftJ1753,usingthephabscomponentwithcross-sections g
c the common physical constants, and M the mass of the
from Bal ucin´ska-Church & McCammon (1992) and abun-
black hole). Further, we constrained the line energy to the
dances from Anders & Grevesse (1989). The total column
rangebetween6.4keVand7keV,whichistheenergyrange
density in the direction of the source, as derived from HI
data, is N =1.7 1021 cm−2 (Dickey & Lockman, 1990). spannedbytheKαlinesfromneutraltohydrogen-likeiron.
H
×
Sincethiscolumndensityprovidesanupperlimittothein-
terstellar absorptiontoSwift J1753, weleft N freetovary Forallfitsofthedifferentcontinuummodels,thevalues
H
duringourfits.Foranoverallnormalizationbetweenthedif- of the parameters and the 90% confidence errors are given
ferentinstruments,wemultipliedthecontinuummodelbya in Tab. 1. In thetextwe only give approximate numbersof
constant component. All model parameters, except these theparameters.
4 B. Hiemstra et al.
3.2 A power-law model givesareducedχ2 of1.14,wherethebreakenergyandpho-
ton indices are consistent with those in the fit without the
The first continuum model we fit was a power-law model
line. We note that the bknpl+Laor model fits the spec-
(pl). This model yields a poor fit (χ2/ν = 845/502). Fol-
traequallywell asthe pl+Laor+diskbb modeldoes, with
lowing M06, weadded adisk blackbodycomponent diskbb
theparametersoftheLaorcomponentbeingconsistentbe-
(Mitsuda et al. 1984) with a best-fit inner disk tempera-
tweenthetwomodels.Thenormalizationofthelineismore
ture of 0.4 keV; the reduced χ2 in this case is 1.21 (χ2/ν
∼ than 3σ different from zero.
are given in Tab. 1). The spectra and fit residuals of the
To see whether a soft thermal component is required,
pl+diskbb model are shown in the left panel of Fig. 2. A
we included a diskbb component to the model. The fit
broad emission line between 6–7 keV is apparent in this
with a bknpl+Laor+diskbb model gives a reduced χ2 of
figure. We added a Laor line profile to the pl+diskbb
1.12, which is a marginal improvement of the fit without
model and thefit improved significantly, with a reduced χ2
the diskbb component. In contrast with the Laor com-
of 1.14. To check whether a fit to the spectrum with only
ponent, the diskbb component is not significant, with a
a line and a power-law component provides a good fit, we
normalization that is < 3σ different from zero. The nor-
fit apl+Laor model, without thediskbb component.The
malization of the line is 6σ different from zero for the
pl+Laor model gives a reduced χ2 of 1.30. In the best-fit
bknpl+Laor+diskbbmodel.Thelineinthebest-fitmodel
model (i.e. the pl+Laor+diskbb model), the line is re-
(i.e. bknpl+Laor) has an equivalent width of 67.5 eV.
quired at high confidence as indicated by the fact that the
Wenotethatthepartofthepowerlawbeforethebreak
normalization of the line is 10σ different from zero. The
at 3keVpossiblyaffectsthecontributionofthediskblack-
line in the best-fit model has an equivalent width of 90.3 ∼
body,and that therefore the diskbb component in the bro-
eV, and the diskbb component is required at a confidence
kenpower-lawmodelisnotassignificantcomparedwiththe
level of more than 8σ. The flux of the diskbb component
continuummodel based on a power law without a break.
is 7.2 10−12 erg cm−2 s−1 (0.6–10 keV), where the unab-
sorbed×fluxofthebest-fitmodelis3.9 10−10 ergcm−2 s−1
×
(0.6–10 keV). We note that the diskbb normalization that
3.5 Comptonization model
weobtainedismuchsmallerthanthevaluethatM06found,
andthecontributionfrom thedisktothetotalfluxthatwe We then replaced the bknpl component with a model that
found is a factor of 3 less than reported in M06. describes thehard emission as Comptonization of soft pho-
∼
In the case when the distance, inclination angle of the tons in a hot plasma (comptt; Titarchuk 1994). Using a
disk, and mass of the black hole are known, the inner disk disk geometry for the Comptonizing material, the comptt
radius, Rin can be obtained from the diskbb normalization model gives a poor fit (χ2/ν = 753/500). A spherical ge-
(ND).However,forthecaseofSwiftJ1753theseparameters ometry does not provide a good fit either. Adding a Laor
arecurrentlynotknown.Assumingthatthesourceiscloser component gives a reduced χ2 of 1.34, with the normaliza-
than 10 kpc and has an inclination lower than 63◦, since tion of theline being more than 7σ different from zero.
eclipsesarenotyetreported,Rin.1.48√NDkm.Assuming Thecomptt+Laor+diskbbmodelgivesareducedχ2
a black hole mass >3 M⊙, Rin <0.33 √ND Rg. From the of 1.15, with an inner disk temperature of 0.4 keV and a
∼
diskbbnormalization wefound anon-truncateddisk,while diskbb normalization that is > 3σ different from zero. The
inferred from the line profile we found Rin 15 Rg. For an lineinthebest-fitmodelhasanequivalentwidthof59.9eV,
inclination angle>85◦,asdeterminedfrom∼thelineprofile, withalinenormalizationthatis>4σdifferentfromzero.We
Rin can become much larger. notethattheseedphotontemperatureofthecompttcom-
ponentbecomesmorethan10timessmallerthaninthecase
whennodiskbbisaddedtothecontinuum.The0.6–10keV
3.3 Cut-off power-law model flux of this three-component model is 3.9 10−10 erg cm−2
s−1,whereasthefluxofthediskbbcompo×nentis8.9 10−12
Next,weaddedahigh-energycutofftothepower-lawcom- erg cm−2 s−1. ×
ponent.Inthecase whenwelet alltheparametersfree, the
high-energy cut off became larger than 200 keV. Since this
value is outside the energy range of the spectra, this com- 3.6 Reflection models: PEXRAV and PEXRIV
ponent is effectively the same as the simple power law. We
therefore did not explore this model further. ThepresenceofabroadenedFe-emissionlinebetween6and
7keVsuggeststhattheX-rayemissionmaypartlybedueto
reflectionofhardX-raysbyarelativelycooldisk.Wethere-
foreexploredfitstothedatathatincludethisreflection.We
3.4 Broken power-law model
first used the pexrav and pexriv models, which describe
The next continuum model we tried is based on a broken theCompton reflection byneutraland ionized material, re-
power law (bknpl) instead of a power law. The fit of the spectively (Magdziarz & Zdziarski 1995).
bknpl model gives already a reasonable fit with a reduced The incident spectrum is assumed to be a power law
χ2 of 1.20, which is similar to the value for the fit with a with a high-energy cut off. Initially, we allowed this cut-
pl+diskbb model. The break energy is at 2.9 keV, with a off energy to be free to vary, but it always converged to
photonindexof1.7beforethebreak,and1.6afterthebreak. valuesmuch higher than 200 keV, outside theenergy range
Thelatterphotonindexissimilartothephotonindexofthe coveredbyourinstruments,andwethereforeusednocutoff.
power-law component in thepl+diskbb model. Further,we fixed all abundances to solar and the reflection
A Laor component added to this continuum model factor to 1. Compared to pexrav, the pexriv model has
Line and disk emission in Swift J1753.5−0127 5
0 0
V 1 V 1
e e
ec/k ec/k
nts/s 1 nts/s 1
u u
o o
d c d c
alize 0.1 alize 0.1
m m
or or
n 1 n 1
0 0
0. 0.
4 4
2 2
χ 0 χ 0
2 2
− −
4 4
− −
1 10 1 10
Energy (keV) Energy (keV)
Figure 2.XMM-Newton PN(0.3–10keV), MOS2(0.3–9.4keV), RXTE PCA(3–20 keV),andHEXTE(20–100 keV)spectraofSwift
J1753.Inthelowerpanelsthebestfitresidualsofapl+diskbbmodel(left)andapl+Laor+diskbbmodel(right)areshown.TheLaor
lineprofileisfittedtothebroademissionlinethatisapparentinthespectraataround7keV.
two extra parameters: The ionization parameter ξ, which flion only provides the reflected emission, not the direct
weleftfreetovary,andthedisktemperature,kT ,which component, and we therefore used the reflion+pl model
disk
we fixed tothe default valueof 3 104 K. tofitthespectra,wherewecoupledthephotonindexofthe
×
The fit with the pexrav model gives a reduced χ2 of illuminatingpower-lawcomponentinrefliontothephoton
1.38,andwiththepexrivmodelthereducedχ2=1.30.The index of thepl.
pexrav/pexrivmodelsincludeonlytheeffectofbound-free
transitionsinthereflectedspectrum.Toaccountfortheiron The reflion+pl model gives a reduced χ2 of 1.14.
In this case we do not need to include a component to
emissionlineataround7keV(causedbybound-boundtran-
fit the line, since the line is part of the reflected emis-
sitions),weaddedaseparateLaorcomponenttothereflec-
tion models. The spectral fits improved significantly, with sion. The unabsorbed flux of this two-component model is
a reduced χ2 of 1.19 and 1.18 for the pexriv+Laor and 3.9 10−10 ergcm−2 s−1 (0.6–10keV),wherethefluxofthe
pow×er-law component is 3.0 10−10 erg cm−2 s−1 (0.6–10
pexrav+Laor models, respectively. The normalization of
×
thelineis9σ differentfromzerointhefitwithpexrav,and keV).Thereflection ratioin the0.6–10 keVenergy rangeis
0.3. Statistically, a diskbb component is not required to fit
8σdifferentfromzerointhefitwithpexriv.Theequivalent
thedata; adding this component gives a reduced χ2 of 1.13
widthofthelineis181eVand187eVinthepexrav+Laor
and pexriv+Laor models, respectively. with a best-fit temperature kTin = 0.4 keV and a diskbb
normalization whichislessthan3σ differentfromzero.The
Adding a diskbb component to the reflection plus
line model gives a reduced χ2 of 1.14 for both mod- 95% confidence limits for the diskbb component implies a
disk flux < 6.1 10−12 erg cm−2 s−1 (0.6–10 keV).
els, with a best-fit disk temperature kT = 0.3 keV and
in ×
0.4 keV for the neutral and ionized models, respectively.
Togaininformationaboutthediskradius,weconvolved
The diskbb component is statistically not required in the
reflion with the Kerrconv component (Brenneman &
pexrav+Laor+diskbb model, with a diskbb normaliza-
Reynolds 2005). The Kerrconv kernel is applied to the
tion that is < 3σ different from zero. In contrast, for the
reflectedcomponenttoaccountfor relativisticeffectsin the
pexriv+Laor+diskbb model the diskbb normalization is
disk.Wefixedthetwoemissivityindicesto3(i.e.similarto
7σ different from zero. The flux of the diskbb component
theindexthatwefoundinthelineprofile)andthebreakra-
is 5.1 10−12 erg cm−2 s−1 and 5.5 10−12 erg cm−2 s−1
× × diustoalargevalue(thisisequivalenttoasingleemissivity
(0.6–10 keV) for the neutral and ionized reflection models,
lawforthewholedisk).Wefurtherfixedtheouterradiusof
respectively.
thedisktoitsmaximumvalue.Wefoundthatthespinofthe
blackholeisnotconstrainedbythesedata,andsinceweare
testing for the possibility of a truncated disk at a distance
3.7 Ionized reflection model: REFLION
that is much larger than the radius of the ISCO, we fixed
Sincepexravandpexrivdonotincludethecontributionof the spin to its maximum value of 0.998 and left the inner
bound-boundtransitions to the reflected spectrum, we also radius of the disk free to vary. This allows the disk radius
fitthedatawithamodelthatdescribesthereflectionbyan to float from 1.23 R to its maximum value. We checked
g
∼
ionized, optically thick,illuminated atmosphereof constant that fixing the spin to zero did not change the results sig-
density (reflion; Ross & Fabian 2005). The atmosphere nificantly. In summary, we allowed only two parameters of
isilluminatedbyapower-lawspectrumwithanexponential Kerrconv tobefree:Theinnerradiusofthedisk,andthe
cutoffathighenergy(thecut-offenergyisfixedat300keV, inclination of thedisk with respect totheline of sight. The
and hence outside our energy range). This model includes (reflion+pl) Kerrconv gives areduced χ2 of 1.13, with
∗
ionizationstatesandtransitionsforthemostimportantions an inner disk radius of 256 R , an inclination angle consis-
g
at energies above 1 eV, like O III–VIII, Fe VI–XXVI, and tentwith0◦,andtherestoftheparametersconsistentwith
several others. We fixed the iron abundance to solar. Re- themodel parameters without theKerrconv kernel.
6 B. Hiemstra et al.
2
0
1.
el
d
o
M
a/
Dat 1
8
9
0.
2 4 6 8
Energy (keV)
Figure 3. In the various continuum models fit to the XMM-Newton/RXTE spectra of SwiftJ1753, a broad emissionlineis apparent
between 6 and 7 keV. Here, we show a zoom in of the PN residuals in the fit with the pl+Laor+diskbb model. To show the broad
emissionline,wehavesetthenormalizationofthelinetozero,andappliedanadditionalrebinningtotheplot.
3.8 The broad iron line at around 7 keV The exact values of the line parameters may also de-
pendontheapproximationsdonetocalculatetherelativistic
In previous subsections we included a Laor component to
profileof theline. To test this, we tried threeother compo-
the continuum model to fit a broad line at around 7 keV.
∼ nents,besidestheLaorprofile,torepresentthelineprofile:
The line is significantly required by the fit, as indicated by
diskline (Fabian et al. 1989), Kerrdisk (Brenneman &
the fact that the normalization of the line is different from
Reynolds 2005), and Kyrline (Dovˇciak, Karas & Yaqoob
zeroataconfidencelevelof3–14σ,dependingonthemodel
2004).Weonlyappliedthesecomponentstothepl+diskbb
we used to fit the continuum. As shown in the right panel
model, such that we either fit pl+diskbb+diskline,
of Fig. 3, the line is extending from 5 keV up to 9 keV,
∼ ∼ pl+diskbb+Kerrdisk or pl+diskbb+Kyrline. For all 3
with an equivalent width of 60–187 eV, depending on the
∼ models we left the inner radius of the disk, the inclination
continuum model used to fit the spectrum.
and the line energy free to vary, although the line energy
Thebest-fitparametersofthelinedependonthemodel was constrained to range between 6.4 and 7 keV. We fur-
that we used to represent the underlying continuum emis- ther assumed a disk with a single emissivity index which
sion.Forinstance,forthedifferentbest-fitmodelsinTab.1, wefixedto3.Forthemodelswith KerrdiskandKyrline
the innerdisk radius varies from 5.5 Rg to 14.6 Rg and the we fixed the spin of the black hole to 0.998. The results
inclinationappearstobeconstrainedtohighvalues,i>85◦. are unaffected if we fix the spin of the black hole to zero.
Suchhighinclinationseemstobeincompatiblewiththefact As determined from these relativistic line profiles, the in-
thatnoeclipseshavebeenobservedintheX-raylightcurve clination of the disk is 89.9◦ (with a 95% confidence lower
of Swift J1753. Apossible explanation for thishigh inclina- limit of 69.7◦), 90◦ (with a 95% confidence lower limit of
tion isthat thebroad linethatwe observeis acombination 89◦), and 77.6◦, (with the 95% confidence limits of 64.6◦
of lines from more than one ionization stage of iron. We and86.1◦),for thediskline,Kerrdisk,andKyrlinepro-
testedthisbyaddingthreeLaorprofilestothepl+diskbb files, respectively. For comparison, for the Laor profile we
modelasdescribedin 3.2,withtheenergyofeachlinefixed found a 95% confidence lower limit for the inclination of
§
at 6.4 keV, 6.7 keV and 6.97 keV, respectively, to account 84◦ (see Tab. 1).
forpossibleemissionfromneutralandlowlyionized,He-like, ∼
Thevalueoftheinnerdiskradiusinferredfromtheline
and H-likeiron, respectively. Wefurther coupled all the re-
profile also depends somewhat on the model that we used
maining parameters of these three lines, with only the line
to describe the line. For instance, the pl+diskbb+Laor
normalizations left free to vary. This multiple line model
gives a reduced χ2 of 1.14, with the 95% confidence limits model gives Rin = 14.6 Rg, with a 95% confidence upper
oftheinclinationangle83◦ and86◦,andthelinenormaliza- limitof19.7Rg (seeTab.1).Forthepl+diskbb+diskline
model we find R = 11.7 R , with a 95% confidence range
tionsofthe6.4 keVand6.7keVlinesbeingconsistent with in g
of6.2-15.5 R ,whereasthepl+diskbb+Kerrdiskandthe
zero.Amorerealisticapproachwouldbetolinktherelative g
pl+diskbb+KyrlinemodelgiveR =1.9R ,witha95%
normalizations oftheiron linessuchthattheirratioswould in g
confidence range of 1.7–3.8 R , and R = 15.5 R , with a
bedeterminedbytheionizationbalanceinthedisk.Wenote g in g
95% confidencerange of 11.8–20.1 R , respectively.
thatthereflioncomponent,asdescribedin 3.7,calculates g
§
self-consistentlytherelativeintensitiesoflinesfromdifferent Wenotethatweusedherethesamedataasthoseused
ionizationstagesofallelementsinthereflectedspectrumby by M06, and our analysis is similar to theirs; while here we
fitting theionization parameter of thedisk. Recall that the reportonabroadironemissionlinewithanequivalentwidth
fits of the reflion component convolved with Kerrconv of 90 eV. Using the same continuum model, M06 reported
in 3.7 yield a large inner disk radius ( 250 R ) and a low an upper limit of 60 eV for the Fe line. We repeated the
g
inc§lination angle (i<8◦ at 95% confide∼ncelevel). analysis of Swift J1753 where we applied the same spectral
Line and disk emission in Swift J1753.5−0127 7
resolution used by M06 (a minimum of 10 counts per bin) as due to a cool accretion disk truncated at a large radius
andwesimultaneouslyfitthePN,PCAandHEXTEspectra (e.g.,Chatyetal.2003).Thispictureisstrengthenedbythe
with a diskbb+pl model affected by absorption. We found factthatduringtransitionsfromtheHSStotheLHSthera-
best-fit parameters and reduced χ2 values consistent with diusofthediskincreaseswithadecreasingdisktemperature
those in M06. We then added a Laor component to the (Gierlin´skietal.2008),suggestingthatthediskrecedesand
modelwiththesameparametersettingsasdescribedin 3.1. coolsdownduringthetransition.Othersourcesshowasimi-
§
Thediskbb+pl+Laormodelgivesχ2/ν =2003/1951with larbehavior(e.g.,Kalemcietal.2004),butsincetheinferred
thelineparametersconsistentwiththosethatwefound.The temperatures of the disk are below 0.3–0.5 keV, interstel-
∼
linehasanormalization thatismorethan6σ differentfrom lar absorption and inadequate coverage of the low-energy
zero,andanequivalentwidthof73.5eVwhichis,withinthe partofthespectrummakesomeofthoseresultslesssecure.
confidence limits, consistent with the 90 eV measurement Changes of characteristic frequencies in the power density
that we report. spectra of BHCs during state transitions (e.g., M´endez &
vanderKlis1997;Homan&Belloni2005)alsoseemtosup-
port the idea of a receding disk,although in this case there
isnoclear-cutexplanationabouttheoriginofthevariability
4 DISCUSSION
(see the discussion in M06), and hence the evidence is less
We analyzed simultaneous observations of Swift J1753 in compelling.
thelow/hard statetakenwithXMM-Newton andRXTE in Thepictureofanaccretiondisktruncatedatlargeradii
March 2006. At variance with the results of Miller et al. in the LHS is also expected in some models. For instance,
(2006a), wecansuccessfully fittheX-rayspectrum ofSwift intheadvection-dominatedaccretion-flow(ADAF)modelin
J1753 with continuum models that do not require emission whichtheaccretionflowisdividedintotwozones:Ageomet-
fromaprominentdiskextendingdowntotheinnermoststa- rically thin,optically thickaccretion diskwithalargeinner
ble circular orbit. Models including emission from a disk radius, and a vertically extended inner region, the ADAF
component, as proposed by M06, are also consistent with region, which is hot and optically thin and radiates less ef-
the data, although for those models where the disk compo- ficiently (Narayan,McClintock & Yi1996; Esin et al. 1997;
nentisrequiredwefoundthatthefluxofthediskisafactor see Narayan & McClintock 2008 for a review). This model
of 2 to 3 less than in the analysis of M06. Regardless of hasbeensuccessfullyappliednotonlytosourcesintheLHS,
thecontinuummodel,wefoundabroad iron lineat around buttoquiescentsourcesaswell(Narayanetal.1996).Some
7 keV with an equivalent width of 60–187 eV, depending recent work (Meyer et al. 2007; Liu et al. 2007) suggests,
on themodel. Ifthe line is from a single ionization stage of however,thatapartoftheADAFmaterialmayrecondense
ironandthebroadeningisduetorelativisticeffectsnearthe backintoadisk,butthecontributionofthiscondenseddisk
black hole,thenthelineprofileimplies an innerdisk radius to the total X-ray luminosity would be small (Taam et al.
between 2 R and 16 R , depending on the relativistic 2008).
g g
∼ ∼
line profile (i.e. Laor, diskline, Kerrdisk or Kyrline), or the In the last five years, XMM-Newton and Chandra re-
model used to fit thecontinuum.The innerdisk radius can vealedrelativelybroadironemissionlinesofBHCswithpro-
belargerifthewidthofthelineresultsfromthecombination filesthatareconsistentwithbeingbroadenedbyrelativistic
oflinesfrom differentionization stages ofiron,with each of effects near the black hole (see Miller 2007 for a review).
theselines being broadenedby relativistic effects. In fact, a These broad iron lines are generally detected in the HSS
model of reflection from an ionized disk that calculates the (e.g., GRO J1655 40; Bal ucin´ska-Church & Church 2000;
−
strengthofallrelevantlinesself-consistently,convolvedwith XTE J1650 500; Miller et al. 2002b; GX 339 4; Miller et
− −
akernelthataccountsfortherelativisticeffectsclosetothe al. 2004; Dunn et al. 2008), in which the disk is thought
blackholeyieldsaninnerdiskradiusof256R ,witha95% to extend down to the ISCO (as described before), and the
g
confidencelower limit of 246 R . relativistic effects are therefore strong. Broad iron emission
g
To get a better insight in the accretion disk geometry lines, consistent with a disk extending close to the ISCO,
thereare several factors that must be considered. We know were also detected in GX 339 4 in the LHS (Miller et al.
−
thatblack-holebinariesshowtwobasicstates,thelow/hard 2006b), and in V4641 Sgr when the source was in an inter-
and the high/soft states (see e.g. Tananbaum et al. 1972), mediate state (Miller et al. 2002a). We found a broad iron
with transitions in between these states (see e.g., McClin- line in the XMM-Newton/RXTE spectrum of Swift J1753,
tock &Remillard 2006; Homan &Belloni 2005). TheX-ray making it only the second case of a BHC in the LHS state
emission in the HSS is dominated by a thermal component showingsuchabroadline.Arelativisticbroadenedironline
which is usually identified as emission from the accretion in this black hole state would be in contradiction with the
disk. In this state, the inner radius of the disk, R , ap- standard scenario as described above, in which the disk is
in
pears to remain constant despite large changes of the disk truncated at verylarge distances from theblack hole.
flux(e.g.,Tanaka&Lewin1995;M´endez,Belloni&vander This contradiction appears to be resolved by recent
Klis1998). Tanaka(1992) interpretedthislack ofchangein studies of the BHCs Swift J1753, GX 339 4 and XTE
−
R as a signature of the innermost stable orbit around a J1817 330 (M06; Miller et al. 2006b; Rykoff et al. 2007)
in
−
black hole. Theblack holespectra in theLHSisdominated that reveal thepossible existence of soft emission below 2
∼
byemissionfromapower-lawlikecomponent;inthosecases keVinthesesystems.Thisemissionhasbeeninterpretedby
wheretheinterstellarabsorptiontowardsthesourceislow,a these authors in terms of a prominent accretion disk in the
softcomponentwithalowtemperature(.0.5keV)appears LHS,withtemperaturesof0.2–0.3keV,andinnerdiskradii
to be present in the spectrum (e.g., XTE J1118+480; Mc- of 2–6 R . The idea of a cool disk extending all the way
g
∼
Clintock et al. 2001). This component has been interpreted down to the ISCO in the LHS has recently been challenged
8 B. Hiemstra et al.
by Gierlin´ski et al. (2008) for the case of XTE J1817 330 They found that the line at around 7 keV in the spec-
−
and by D’Angelo et al. (2008) for the cases of Swift J1753 trum of a bright LHS observation can either be fit with a
andGX339 4.Theseauthorspointedoutthatatruncated modelincludingextremerelativisticeffects,orbyresonance
−
disk would beaffected by irradiation from the same Comp- iron K-line absorption from an outflowing disk wind and
tonizedphotonsthatproducethehigh-energyemission,and an emission line that is compatible with a truncated disc.
would therefore appear to be hotter and a have smaller ra- On the other hand, R´oz˙an´ska & Madej (2008) calculated
dius than in the non-irradiated case. This effect was not atmosphere models for an accretion disk around a super-
taken into account in the fits of M06, Miller et al. (2006b), massiveblackholeirradiatedbyahardX-raypowerlaw,and
or Rykoff et al. (2007). Gierlin´ski et al. (2008) fit the Swift theyfoundthattheobservedspectrumcontainsaCompton
spectraofXTEJ1817 330 withasimplifiedmodelthatac- shoulder that can contribute to the asymmetry and equiv-
−
counts for irradiation, and they find that if this effect is alent width of some observed Fe Kα lines in active galactic
not considered, the inner disk radius is underestimated by nuclei. One should also take into account that, as we have
a factor of 2–3. D’Angelo et al. (2008) analyzed the case of shown here, the best-fit profile of the line depends on the
an ADAF flow inside a truncated accretion disk and, from underlying continuum assumed to fit the data, which adds
qualitative fits to theX-ray spectra of Swift J1753 and GX extrauncertainties to theinferred parameters of thedisk.
339 4, they conclude that reprocessing of hard photons by
−
atruncateddiskwouldmimicasoftdiskextendingdownto
very small radii.
5 CONCLUSIONS
Using the same data as analyzed by M06, we found
that there are realistic models that fit the X-ray spectrum WehaveshownthattheX-rayspectrumofSwiftJ1753 can
ofSwift J1753 without theneed ofathermaldisk-likecom- befitwith acontinuummodelthatdoesnotrequireadisk-
ponent extending down to very small radii. We point out like component, and inferred from the line profile the disk
that for our fits we do not consider the effect of possible doesnotextenddowntotheISCO,butitisconsistentwith
irradiationofthediskbyhigh-energyComptonizedphotons a disk truncated at a few to a few hundred gravitational
(Gierlin´ski etal.2008; D’Angeloetal.2008). Wecancalcu- radii. However, as demonstrated by M06, the data of Swift
latethemagnitudeofthiseffectbynotingthatinthecaseof J1753 also allow for fits with models in which the disk is
thereflion+pl model only 30% of thepower-law flux is truncatedatradiiclosetotheISCO,althoughirradiationof
∼
reflected off theaccretion disk, and therefore theremaining thediskbytheComptonizedemissionmayaffectthispicture
70% of this flux must be thermalized and re-emitted by (Gierlin´skietal.2008;D’Angeloetal.2008).Wefoundthat
∼
thedisk.Fromourresultswededucedthatthediskfluxdue in those cases that a disk-like component was required, the
to this irradiation would have to be 2 10−10 erg cm−2 contribution of the disk to the total flux is a factor of 2 to
s−1 (0.6–10 keV).However,in thefit w∼ith×thereflion+pl 3 times less than the flux contribution that M06 found for
model we did not detect emission from the accretion disk, thedisk component.
with an upper limit that is a factor 25 less than what is We have detected a broad iron line in the spectrum of
∼
expected.(Wenotethatthediscrepancyisevenlargergiven SwiftJ1753.Theinnerradiusofthediskdeducedfromfitsto
thatoneshouldalsoincludetheintrinsic-gravitational-disk theline,usingaLaorprofile,seemstosuggestadiskextend-
flux; see Gierlin´ski et al. 2008.) This a challenge to the in- ing down to 5.5–12.5 R , for the best-fit models. Fits with
g
terpretation of the Feemission line as dueto reflection. differentrelativisticlineprofilessuggestsomewhatlargerin-
In the case of GX 339 4, Tomsick et al. (2008) found nerdisk radii,upto 16 Rg (disklineandKyline),orlower
− ∼
that the disk component is significantly required to fit the Rin, down to 2 Rg (Kerrdisk), whereas reflection models
∼
LHSspectraatluminositiesof2.3%and0.8%oftheEdding- smearedbyrelativisticeffectssuggestevenmuchlargerradii,
tonluminosity,afactorof 2and 7belowtheluminosities Rin 250 Rg. The XMM-Newton/RXTE data as analyzed
∼ ∼ ∼
asprobedbyMilleretal.(2006b)forthesamesource.Tom- here, do not provide a definitive answer to the question of
sick et al. (2008) found inner disk radii of 4 R however, theaccretiondiskistruncatedatlargeradiiornot,sincethe
g
∼
followingtheresultsofGierlin´skietal.(2008)andD’Angelo answer strongly depends on the continuum and line model
et al. (2008), theirvalueswould probably increase bya fac- thatareused.Oneshouldhoweverbearinmindthatthein-
tor of 2–3 if irradiation of the disk by the Comptonized terpretation ofthelineprofilein termsofrelativistic effects
∼
emissionwasconsidered.Tomsicketal.(2008)alsodetected closetotheblackholeisnottheonlypossibleone,andother
broadfeaturesduetoironKαintheLHSofGX339 4;their alternatives do not require a disk extending down close to
−
results indicate that if the width of the line is produced by the ISCO to explain the shape of the line. It remains to be
relativistic effects in the disk, then the line must originate seen whether this kind of models can fit the data of Swift
from within 10 R of theblack hole. J1753 and othersources in theLHS.
g
In this work we used relativistic line profiles to fit the
broad emission line at around 7 keV, although one should
bear in mind that there are alternative models that ex-
ACKNOWLEDGMENTS
plain the width and the profile of the iron Kα line with-
out the need of relativistic effects. For instance, Laurent & This work is based on observations obtained with XMM-
Titarchuk(2007)proposedamodelthatexplaintheproper- Newton, an ESAsciencemission with instrumentsandcon-
tiesoftheironlineintermsofdown-scatteringofhardpho- tributions directly funded by ESA Member States and the
tons in a Comptonizing outflow with optical depth greater USA (NASA).We thank Diego Altamirano for useful com-
than1.Done&Gierlin´ski(2006)appliedasimilarmodelto mentsand discussions, and Matteo Guainazzi for providing
fit BeppoSAX data of the galactic BHC XTE J1650 500. thepharbn tool. Weare grateful to the XMM-Newton help
−
Line and disk emission in Swift J1753.5−0127 9
deskfortheirassistanceandadvise.TMBacknowledgessup- Miller,J.M.,Homan,J.,Steeghs,D.,Rupen,M.,Hunstead,R.W.,
port from the International Space Science Institute (ISSI) Wijnands,R.,Charles,P.A.,&Fabian,A.C.2006,ApJ,653,
and from ASIvia contract I/088/06/0. 525
Miller,J.M.2007,ARA&A,45,441M
Mitsuda, K., Inoue, H., Koyama, K., Makishima, K., Matsuoka,
M., Ogawara, Y., Suzuki, K., Tanaka, Y., Shibazaki, N., &
REFERENCES
Hirano,T.1984,PASJ,36,741
Anders,E.,&Grevesse,N.1989,GeCA,53,197 Morris, D.C., Burrows, D.N., Racusin, J., Roming, P., Chester,
Arnaud,K.1996,AstronomicalDataAnalysisSoftwareandSys- M.,LaVerghetta, R.,Markwardt, C.B.,& Barthelmey, S.D.
tems V, A.S.P. Conference series, ed. G.H. Jacoby & J. 2005,Astron.Tel.552
Barnes,101,17 Narayan,R.,McClintock,J.E.,&Yi,I.1996,ApJ,457,821N
Bal ucin´ska-Church, M., & Church, M.J. 2000, MNRAS, 312L, Narayan,R.,&McClintock,J.E.2008, NewAR,51,733N
55B Palmer, D.M., Barthelmey, S.D., Cummings, J.R., Gehrels, N.,
Bal ucin´ska-Church&McCammon1992, ApJ,400,699 Krimm,H.A.,Markwardt,C.B.,Sakamoto, T.,&Tueller,J.
Brenneman,L.W.,&Reynolds,C.S.2005,AAS,207,0606B 2005,Astron.Tel.546
CadolleBel,M.,Rodriguez,J.,Goldwurm,A.,Goldoni,P.,Lau- Remillard,R.A.,&McClintock,J.E.2006,ARA&A,44,49R
rent, P., Ubertini, P., & Mereghetti, S. 2005, Astron. Tel., Ross,R.R.,&Fabian,A.C.2005,MNRAS,358,211
Ro´z˙an´ska,A.,&Madej,J.2008,MNRAS,386,1872
574
Rykoff, E.S., Miller, J.M., Steeghs, D., & Torres, M.A.P. 2007,
Cadolle Bel, M., Ribo´, M., Rodriguez, J., Chaty, S., Corbel, S.,
ApJ,666,1129
Goldwurm,A.,&Malzac,J.2006,ProceedingsoftheVIMi-
Shakura,N.I.,&Sunyaev, R.A.1973, A&A,24,337
croquasarWorkshop:MicroquasarsandBeyond
Still, M., Roming, P., Brocksopp, C., & Markwardt, C.B. 2005,
CadolleBel,M.etal.2007,ApJ,659,549
Astron.Tel.553
Chaty, S.,Haswell,C.A.,Malzac,J.,Hynes, R.I.,Shrader,C.R.,
Taam,R.E.,Liu,B.F.,Meyer,F.,&Meyer-Hofmeister,E.2008,
&Cui,W.2003,MNRAS,346,689C
ApJ,688,527T
Done,C.,&Gierlin´ski,M.2006,MNRAS,367,659
Tananbaum, H.,Gursky, H., Kellogg,E., Giacconi, R.,& Jones,
Dovˇciak,M.,Karas,V.,&Yaqoob, T.2004,ApJS,153,205D
C.1972,ApJ,177L,5T
D’Angelo, C., Giannios, D., Dullemond, C., & Spruit, H. 2008,
Tanaka, Y. 1992, Ginga Memorial Symp., ed. F. Makino & F.
A&A,488,441D
Nagase(Tokyo: Inst.ofSpaceandAstronaut.Sci.),19
Dickey,J.M.,&Lockman,F.J.1990,ARA&A,28,215
Tanaka,Y.,&Lewin,W.H.G.1995,X-RayBinaries,ed.W.H.G.
Dunn,R.J.H.,Fender,R.P.,Ko¨rding,E.G.,Cabanac, C.,&Bel-
Lewin,J.vanParadijs,&E.P.J.vandenHeuvel(Cambridge:
loni,T.2008,MNRAS,387,545D
CambridgeUniv.Press),126
Esin,A.A.,McClintock,J.E.,&Narayan,R.1997,ApJ,489,865
Tanaka, Y., Nandra, K., Fabian, A.C., Inoue, H., Otani, C.,
Fabian,A.C.,Rees,M.J.,Stella,L.,&White,N.E.1989,MNRAS,
Dotani, T., Hayashida, K., Iwasawa, K., Kii, T., Kunieda,
238,729
H.,Makino,F.,&Matsuoka,M.1995,Nature,375,659
Fender,R.,Garrington,S.,&Muxlow,T.2005,Astron.Tel.558
Titarchuk,L.1994,ApJ,434,313
Fender,R.P.,Belloni,T.M.,&Gallo,E.2004,MNRAS,355,1105
Tomsick, J.A., Kalemci, E., Kaaret, P., Markoff, S., Corbel, S.,
Gierlin´ski,M.,Done,C.,&Page,K.2008, MNRAS,388,753
Migliari,S.,Fender,R.,Bailyn,C.D.,&Buxton,M.M.2008,
Homan,J.,&Belloni,T.2005,Ap&SS,300,107
ApJ,680,593
Kalemci,E.,Tomsick,J.A.,Rothschild,R.E.,Pottschmidt,K.,&
Torres,M.A.P.,etal.2005, Astron.Tel.566
Kaaret,P.2004,ApJ,603,231K,
Zhang,G.-B.etal.2007,ApJ,659,1511
Laor,A.1991,ApJ,376,90
Laurent,P.,&Titarchuk,L.2007,ApJ,656,1056L
Liu,B.F.,Taam,R.E.,Meyer-Hofmeister,E.,&Meyer,F.2007,
ApJ,671,695L ThispaperhasbeentypesetfromaTEX/LATEXfileprepared
Magdziarz,P.,&Zdziarski,A.A.1995,MNRAS,273,837 bythe author.
Markoff,S.,Falcke,H.,&Fender,R.2001, A&A,372,L25
McClintock,J.etal.2001,ApJ,555,477
McClintock,J.E.,&Remillard,R.A.2006,CompactStellarX-ray
sources, ed. W.H.G. Lewin & M. van der Klis, (Cambridge:
CambridgeUniv.Press),157
M´endez,M.,&vanderKlis,M.1997,ApJ,479,926
M´endez, M., Belloni, T., & van der Klis, M. 1998, ApJ, 499L,
187M
Meyer, F., Liu, B.F., & Meyer-Hofmeister, E. 2007, A&A, 463,
1M
Miller,J.M.,Fabian,A.C.,in’tZand,J.J.M.,Reynolds,C.S.,Wi-
jnands,R.,Nowak,M.A.,&Lewin,W.H.G.2002,ApJ,577L,
15M
Miller, J.M., Fabian, A.C., Wijnands, R., Reynolds, C.S., Ehle,
M., Freyberg, M.J., van der Klis, M., Lewin, W.H.G.,
Sanchez-Fernandez, C., & Castro-Tirado, A.J. 2002, ApJ,
570L,69M
Miller,J.M.,Fabian,A.C.,Reynolds,C.S.,Nowak,M.A.,Homan,
J., Freyberg, M.J., Ehle, M., Belloni, T., Wijnands, R., van
der Klis, M., Charles, P.A., & Lewin, W.H.G. 2004, ApJ,
606L,131M,
Miller, J.M., Homan, J., & Miniutti, G. 2006, ApJ, 652, L113
(M06)
Table1.Resultsofvariouscontinuummodels.Inthefirstpartofthetableweshowtheresultsofapower-law(pl)basedmodel.Alinecomponent(Laor),adiskblackbody(diskbb)or
bothwereadded(thesecomponentsarealsoaddedtotheothercontinuummodels).Theplparametersarethephotonindex,Γ,andthenormalizationNΓ(inunitsofphotons/keV/cm2/s
at 1 keV). The Laor parameters are the lineenergy, EL, which we constrain to range between 6.4 and 7 keV, the emissivityindex, index, inner disk radius, Rin (with Rg=GM/c2),
inclination,i,andnormalizationoftheline,NL (inunitsofphotons/cm2/s).Thediskbbparametersarethetemperatureatinnerdiskradius,kTin,anddisknormalization,ND,which
is defined as [Rin(km)/d(10 kpc)]2cosi, where d is the distance to the source. Next, we show the results of a broken power-law (bknpl) based model, with Γ1, the photon index for
energies below thebreakenergy, breakpoint, Eb,photon indexafter thebreak, Γ2,andnormalization, NΓ (sameunitsas plnormalization). Next, wegivetheresultsofacontinuum
modelbasedonaComptonization(comptt)component.Thecompttparametersare:softseedphotontemperature,kT0,plasmatemperature,kT,opticaldepthoftheplasma,τ,and
normalization,Nc.Next,weshowtheresultsoftworeflectionmodelspexravandpexriv,withparameters:photonindex,Γ,cosineoftheinclinationangle,cosi,andnormalization,NP
(sameunitsasplnormalization).Anextraparametersofthepexrivcomponentisthediskionizationparameter,ξ.Finally,wegivetheresultsofanotherreflectionmodel:reflion+pl.
Togain informationabout thediskradius,we convolve this model withaKerrconv component. The reflioncomponents arethe ionization parameter, ξi, andnormalization ofthe
reflectedspectrum,NR.Thephotonindexofthereflioncomponentiscoupledtotheplphotonindex.ThefreeKerrconvparametersaretheinclinationandinnerradiusofthedisk.
We fit each of the above models including an absorption model (phabs) with NH the equivalent hydrogen column, and a constant component to normalize between the instruments.
Forallcontinuummodels,theunabsorbedfluxis3.9×10−10 ergcm−2 s−1 (0.6–10keV).Inthemodelswithadiskbbcomponent,thefluxofthiscomponentrangesbetween0.3–1.6×
10−11 ergcm−2 s−1 (0.6–10keV).
pl Laor diskbb phabs
Parameter NΓ EL Rin i NL kTin NH
Model Γ (10−2) (keV) index (Rg) (deg) (10−4) (keV) ND (1021 cm−2) χ2/ν
pl+diskbb 1.60±0.01 5.57±0.05 – – – – – – – 0.38±0.04 36.7±10.4 1.61±0.02 603.7/500
pl+Laor 1.66±0.01 6.05±0.02 – – 6.40+−00..106 3.8±0.5 1.36+−00..3113 90.0+−01..08 5.8±0.7 – – 1.60±0.01 646.1/497
pl+Laor+diskbb 1.62±0.01 5.71±0.04 – – 6.79±0.17 3.9+−00..49 14.6±5.1 86.2+−02..13 1.8±0.3 0.36±0.02 35.2+−32..27 1.63±0.02 564.2/495
bknpl Laor diskbb phabs
Eb NΓ EL Rin i NL kTin NH
Γ1 (keV) Γ2 (10−2) (keV) index (Rg) (deg) (10−4) (keV) ND (1021 cm−2) χ2/ν
bknpl 1.69±0.01 2.9±0.2 1.60±0.01 6.17±0.04 – – – – – – – 1.67±0.02 601.8/500
bknpl+Laor 1.70±0.01 2.7±0.2 1.62±0.01 6.18±0.05 6.75+−00..2152 3.7+−60..39 12.5−+111.7.9 86.2+−02..18 1.8±0.8 – – 1.67±0.02 564.0/495
bknpl+Laor+diskbb 1.71±0.04 2.9±0.2 1.62±0.01 6.36±0.34 6.89+−00..1419 3.9+−61..12 17.4±7.9 86.2+−01.11.3 1.5+−00..74 0.18±0.02 2458.8−[phone]..61 2.09+−00..1067 552.9/493
comptt Laor diskbb phabs
kT0 kT NC EL Rin i NL kTin NH
(keV) (keV) τ (10−2) (keV) index (Rg) (deg) (10−4) (keV) ND (1021 cm−2) χ2/ν
comptt+Laor 0.13±0.01 11.9±2.4 3.2±0.3 1.7±0.2 6.79±0.20 3.8+−20..59 11.1+−43..98 86.2+−10..11 2.7±0.6 – – 1.55±0.03 662.8/495
comptt+Laor+diskbb 0.01+−00..0051 23.1+−32..00 2.2+−00..503 4.1−+00..18 6.75+−00..2355 3.5+−60..57 12.4−+171.4.6 86.2+−02..18 1.6±0.6 0.38±0.04 35.1−+2197..93 1.60±0.04 566.4/493
pexrav/pexriv Laor diskbb phabs
ξ NP EL Rin i NL kTin NH
Γ cosi (ergcm/s) (10−2) (keV) index (Rg) (deg) (10−4) (keV) ND (1021 cm−2) χ2/ν
pexrav 1.67±0.01 0.16±0.04 – 6.06±0.03 – – – – – – – 1.61±0.02 693.7/501
pexrav+Laor 1.68±0.01 0.13±0.05 – 6.11±0.04 6.40+−00..102 3.4+−00..74 5.5±0.8 86.9±2.7 3.4±0.6 – – 1.64±0.02 590.0/496
pexrav+Laor+diskbb 1.65±0.01 0.01+−00..005 – 5.90±0.07 6.40+−00..102 3.2±0.5 5.6+−10..59 84.8+−21.04.8 2.5+−01..63 0.31±0.05 50.7−+4332..74 1.68±0.06 560.9/494
pexriv 1.68±0.01 0.13±0.03 134+−17143 6.09±0.03 – – – – – – – 1.64±0.02 651.6/500
pexriv+Laor 1.69±0.01 0.12±0.03 0.0+0.01 6.12±0.03 6.40+−00..10 3.4±0.6 5.5±0.7 84.8+−12..43 3.5±0.7 – – 1.65±0.02 586.4/495
pexriv+Laor+diskbb 1.64±0.01 0.01+−00..005 317+−121085 5.85±0.07 6.43+−00..5073 10.0+−06..02 5.6±3.2 78.0+−81.83.2 1.0+−00..94 0.35±0.03 31.0−+170.2.1 1.67±0.04 562.2/493
pl reflion Kerrconv diskbb phabs
NΓ ξi NR i Rin kTin NH
Γ (10−2) Γ (ergcm/s) (10−8) (deg) (Rg) (keV) ND (1021 cm−2) χ2/ν
reflion+pl 1.54±0.01 4.10±0.15 – 1.54±0.01 5053+−562580 7.6±0.8 – – – – – 1.75±0.01 568.4/500
reflion+pl+diskbb 1.59±0.01 5.17+−00..2049 – 1.59±0.01 1850+−567268 8.2±0.2 – – – 0.39±0.06 17.7−+1181..80 1.66±0.01 560.3/498
(reflion+pl)∗kerrconv 1.55±0.01 4.18±0.26 – 1.55±0.01 4800+−29363 7.9±0.3 – 0.9+−70..69 256±12 – – 1.75±0.01 562.1/498