NASA Technical Reports Server (NTRS) 20140005250: A Suzaku X-ray Observation of One Orbit of the Supergiant Fast X-ray Transient IGR J16479-4514

0.22 MB·English

by  NASA Technical Reports Server (NTRS)

Mon.Not.R.Astron.Soc.000,1–10(2011) Printed5December 2012 (MNLATEXstylefilev2.2)
A Suzaku X–ray observation of one orbit of the supergiant fast
X–ray transient IGRJ16479–4514
L. Sidoli,1⋆ P. Esposito,1 V. Sguera,2 A. Bodaghee,3 J.A. Tomsick,3 K. Pottschmidt,4,5
J. Rodriguez,6 P. Romano,7 and J. Wilms,8
1INAF,IstitutodiAstrofisicaSpazialeeFisicaCosmica,ViaE.Bassini15,I-20133Milano,Italy
2INAF,IstitutodiAstrofisicaSpazialeeFisicaCosmica,ViaGobetti101,I-40129Bologna,Italy
3SpaceSciencesLaboratory,7GaussWay,UniversityofCalifornia,Berkeley,CA94720,USA
4CRESSTandNASAGoddardSpaceFlightCenter,AstrophysicsScienceDivision,Code661,Greenbelt,MD20771,USA
5CenterforSpaceScienceandTechnology,UniversityofMarylandBaltimoreCounty,1000HilltopCircle,Baltimore,MD21250,USA
6LaboratoireAIM,CEA/IRFU-Universite´ParisDiderot-CNRS/INSU,CEADSM/IRFU/SAp,CentredeSaclay,F-91191Gif-sur-Yvette,France
7INAF,IstitutodiAstrofisicaSpazialeeFisicaCosmica,ViaU.LaMalfa153,I-90146Palermo,Italy
8Dr.KarlRemeis-SternwarteandErlangenCentreforAstroparticlePhysics,Freidrich-Alexander-Universita¨tErlangen-Nu¨rnberg,Sternwartstraβe7,
96049Bamberg,Germany
ABSTRACT
We report on a 250 ks long X–ray observation of the supergiant fast X–ray transient
(SFXT) IGRJ16479–4514 performed with Suzaku in 2012 February. During this observa-
tion,about80%oftheshortorbitalperiod(P (cid:24)3.32days)wascoveredascontinuouslyas
orb
possibleforthefirsttime.Thesourcelightcurvedisplaysvariabilityofmorethantwoorders
ofmagnitude,startingwithaverylowemissionstate(10−13ergcm−2s−1;1–10keV)lasting
thefirst46ks,consistentwithbeingduetotheX–rayeclipsebythesupergiantcompanion.
ThetransitiontotheuneclipsedX–rayemissionisenergydependent.Outsidetheeclipse,the
source spends most of the time at a level of 6-7(cid:2)10−12 erg cm−2 s−1 punctuated by two
structured faint flares with a duration of about 10 and 15 ks, respectively, reaching a peak
flux of 3-4(cid:2)10−11 erg cm−2 s−1, separated by about 0.2 in orbital phase. Remarkably, the
firstfaintflareoccursatasimilarorbitalphaseofthebrightflarespreviouslyobservedinthe
system. This indicates the presence of a phase-locked large scale structure in the supergiant
wind,drivingahigheraccretionrateontothecompactobject.TheaverageX–rayspectrumis
hardandhighlyabsorbed,withacolumndensity,N ,of1023 cm−2,clearlyinexcessofthe
H
interstellarabsorption.Thereisnoevidenceforvariabilityoftheabsorbingcolumndensity,
exceptthatduringtheeclipse,wherealessabsorbedX–rayspectrumisobserved.Anarrow
FeK emissionlineat6.4keVisviewedalongthewholeorbit,withanintensitywhichcor-
(cid:11)
relates with the continuum emission above 7 keV. The scattered component visible during
theX–rayeclipseallowedustodirectlyprobethewinddensityattheorbitalseparation,re-
sultinginρ =7(cid:2)10−14 gcm−3.Assumingasphericalgeometryforthesupergiantwind,the
w
derived wind density translates into a ratio M_w/v∞ = 7(cid:2)10−17 M⊙/km which, assuming
terminal velocities in a large range 500–3000 km s−1, implies an accretion luminosity two
ordersofmagnitudehigherthanthatobserved.Asaconsequence,amechanismshouldbeat
workreducingthemassaccretionrate.Differentpossibilitiesarediscussed.
Keywords: X–rays:individual(IGRJ16479–4514)
1 INTRODUCTION 50hr)andpeakfluxes(from20toabout600mCrab,20–60keV;
Sgueraetal.2005,2006,2008,Walter&ZuritaHeras2007,Ducci
IGRJ16479–4514 is a hard X–ray transient discovered by
etal.2010).RecurrentoutburstswerealsoobservedbytheSwift
INTEGRAL on 2003, August 8–10 (Molkov et al. 2003) in
satellite(Kenneaetal.2005,Markwardt&Krimm2006,Romano
the energy range 18–50 keV. Several X–ray flares were caught
etal.2008,LaParolaetal.2009,Bozzoetal.2009).
byINTEGRAL/IBIS,displayingvariabledurations(from0.5to
The proposed counterpart (2MASS J16480656-4512068,
Walteretal.2006)wasconfirmedbyanaccuratelocalizationob-
⋆ E-mail:[email protected] tained with Chandra (Ratti et al. 2010) and classified as a late
⃝c 2011RAS
2 L.Sidoli,etal.
O-typesupergiantwithaspectraltypeO8.5Ilocatedatadistance whichthespacecraftwaspassingthroughthelowcut-offrigidity
of 4.9 kpc (Rahoui et al. 2008, Chaty et al. 2008) or a O9.5 Iab ofbelow6GV.
starlocatedat2.8+4.9kpc(Nespolietal.2008).Theopticalidenti- FortheXISweconsideredonlyeventswithGRADE=0,2–
−1.7
ficationconfirmedtheinitialclassificationofIGRJ16479–4514as 4,6andremovedhotandflickeringpixelsusingSISCLEAN;forthe
a member of the new sub-class of high mass X–ray binaries, the spectralanalysis,theresponsematricesweregeneratedwith XIS-
supergiant fast X–ray transients (SFXTs), at first suggested only RMFGEN andusingray-tracingsimulationswith XISSIMARFGEN.
basedontheshortdurationofitshardX–rayactivity(Sgueraetal. Withalltheaforementioneddataselectioncriteriaapplied,there-
2006). sultingtotaleffectiveexposureisroughly136ksforeachXIS.The
IGRJ16479–4514isaneclipsingSFXT(Bozzoetal.2008b) XISeventsofIGRJ16479–4514wereaccumulatedineachofthe
andtheonewiththenarrowestorbit,showinganorbitalperiodof threeXIScameraswithinacircularregion(3arcminradius)cen-
3.32 days (Jain et al. 2009), later refined to 3.3193(cid:6)0.0005days tredonthetarget,whilethebackgroundswereestimatedfroman
(Romanoetal.2009). annuluswithradii5and7arcmin.Duringtheobservation,thestan-
X–ray long-term monitoring outside outbursts revealed that dardcriterionfortheattitudedeterminationrequiringthattheresid-
thesourcespendsmostofitstimeatareducedlevelofX–rayemis- ualsarelessthan0.005◦couldnotbefullfilled,withresidualsstill
sion(1033-1034ergs−1),from2to3ordersofmagnitudelessthan remainingatabout0.008◦forabout20hours,inthetemporalwin-
the flare peaks, while for the remaining (cid:24)19% of the time, it is dowfromFebruary2501:00through22:00,butstillsmallerthan
undetected, compatibly with being in eclipse (Sidoli et al. 2008, theSuzakupointspreadfunction(PSF)forusualpoint-likesource
Romanoetal.2009). analysis.
Thenatureofthecompactobjectisunclear(asinaboutahalf In the HXD-PIN (no significant emission was detected in
ofthemembersoftheSFXTsclass;forarecentreviewseeSidoli the GSO) the net exposure time, after dead-time correction (live
2010)butthereisindirectevidencebasedonthespectralsimilarity time:92.8%)is141.9ks.Tosubtractnon-X-raybackground(NXB)
withaccretingpulsarssuggestingthattheX–raysourceisaneutron eventsthatareincludedintheHXD-PINweusedthe‘tuned’NXB
star(NS). synthesizedbackground(Fukazawaetal.2009).Aftersubtracting
HerewereportonthefirstX–rayobservationwhichcontinu- thesynthesizedNXBfromtheHXD-PINdata,wealsosubtracted
ously(exceptthatduringtheinterruptionsbecauseofthesatellite contributions of the cosmic X-ray background (CXB) using the
orbit) covers most of an orbital cycle of IGRJ16479–4514. Our modelreportedinGruberetal.(1999)andoftheGalacticRidge
main goal is to investigate the variability of the X–ray properties X–rayemission(GRXE;Valinia&Marshall1998).
alongasingleorbitalcycle. Toensureapplicabilityoftheχ2statistics,thenetXISspectra
wererebinnedsuchthatatleast20countsperbinwerepresent.All
spectraluncertaintiesandupper-limitsaregivenat90%confidence
for one interesting parameter. Data were analysed using FTOOLS
2 OBSERVATIONSANDDATAREDUCTION
version6.11.1andXSPECversion12.Inthespectralfittingweused
IGRJ16479–4514 was observed by Suzaku (Mitsuda et al. 2007) thephotoelectricabsorptionmodelPHABSinXSPECwiththeinter-
between 2012 February 23 and 26. The observation (Obs ID stellar abundances of Wilms et al. (2000) and cross section table
406078010) was performed with the source located at the X–ray set of Balucinska-Church & McCammon (1992). The three joint
Imaging Spectrometer (XIS; Koyama et al. 2007) nominal posi- XIS0,XIS1andXIS3spectrawerefittedtogether,includingcon-
tion.TheXISconsistsoffourtelescopeswithaspatialresolutionof stant factors to allow for normalization uncertainties between the
about2arcmincoupledtofourCCDcamerasoperatinginthe0.2– instruments.Forthetiminganalysisphotonarrivaltimeswerecor-
12keVenergyrangeandwith18′(cid:2)18′fieldofview(1024(cid:2)1024 rectedtotheSolarSystembarycenter.
pixels; 1.05 arcsec pixel−1; Koyama et al. 2007). At the time of
ourobservationonlythetwofrontilluminated(FI)XIS0andXIS3
(effectivearea:330cm2at1.5keV),andtheback-illuminated(BI) 3 ANALYSISANDRESULTS
XIS1(effectivearea:370cm2at1.5keV)wereoperating.Thethree
3.1 Lightcurves
detectorswereoperatedinNormalModewithnowindoworburst
option(allthepixelsontheCCDarereadoutevery8seconds).The The XIS barycentred and background-subtracted light curve of
otherinstrumentaboardSuzakuistheHardX-rayDetector(HXD; IGRJ16479–4514 in the energy range 0.2–10 keV, is shown in
Takahashietal.2007),whichconsistsofPINsilicondiodes(HXD- Fig.1,whilethehardnessratio(HR)betweennetcountsextracted
PIN; 10–70 keV) and Gd SiO Ce (GSO; 50–600 keV) scintilla- inthehard(above3keV)andthesoft(below3keV)energybands
2 5
tors.TheHXDwasoperatedinthestandardmode,withatimeres- is reported in Fig. 2, both versus time (upper panel) and in de-
olutionof61µs.TheHXDisacollimatedinstrumentandthePIN pendenceofthesourceintensity(lowerpanel).NumbersinFig.1
inparticularhasa67.6′(cid:2)67.6′fieldofview(FWHM:34′(cid:2)34′). indicate time intervals which show different source intensity be-
Data reduction and analysis employed version 19 of the haviour. At the beginning of the observation a very low intensity
SuzakuSoftwareincludedintheHEASOFTsoftwarepackageand stateispresent(n.1),consistentwithbeingtheX–rayeclipsepro-
followed the procedures described in the Suzaku ABC Guide.1 duced by the companion star. Indeed, folding the light curve on
Following standard practices, we excluded times within 436 s of therefinedorbitalperiodP =286792(cid:6)43s(Romanoetal.2009)
orb
SuzakupassingthroughtheSouthAtlanticAnomalyandwealso andassumingtheepoch54547.05418(MJD)asorbitalphaseϕ=0
excluded the data when the line of sight was elevated above the (Bozzo et al. 2009), we obtain the orbital phases reported as top
Earth’slimbbylessthan5◦,orwaslessthan20◦fromthebright- x-axisinFig.1.Theuncertaintyontheepochofϕ=0canbeesti-
Earth terminator. Moreover, we excluded time windows during matedinaboutϕ=(cid:6)0.065,obtainedextrapolatingtheerroronthe
orbitalperiodtothetimeoftheSuzakuobservation(432orbital
cyclesbetweentheepoch54547.05418MJDandtheSuzakuob-
1 heasarc.gsfc.nasa.gov/docs/suzaku/analysis/abc/. servations).
⃝c 2011RAS,MNRAS000,1–10
SuzakuobservesoneorbitofIGRJ16479–4514 3
0.0 0.2 0.4 0.6
0
1
7
4
3 6
2 5 8
1
1
1
−
s
1
nt 0.
u
o
C
1
0
.
0
3
−
0
1 0 5×104 105 1.5×105 2×105
Time (s)
Figure1.Background-subtractedlightcurveofIGRJ16479–4514inthe0.2–10keVenergyrange,obtainedcombiningnetsourcecountratesobservedby
thethreeXISunits.Intheupperx-axis,theorbitalphaseisindicated,assumingtheorbitalperiodPorb=286792s(Romanoetal.2009),assumingepoch
54547.05418MJDasorbitalphaseϕ=0(Bozzoetal.2009).Thebinsizeis2048s.Verticaldashedlinesandnumbersindicatetheintervalsoftheeight
time-selectedspectrareportedinTable2.
TheSuzakuobservationdidnotcovertheingresstime,sowe two ramp-and-step functions, characterized by seven parameters:
areunabletodeterminetheexacteclipseduration.Wecanatleast thecountratesofthethree“steps”inthelightcurveandthefour
constrainittobebetween46ks(thedurationoftheeclipseathard times,t ,displayedinFig.3.Thefitresultedinthefollowingtimes
i
X–rays)and143ks(thatis,0.5inphase,assumingaperfectlyedge- measured from the beginning of the observation: t =46 (cid:6) 1 ks,
1
onsystemwithatightorbit).Aneclipsedurationofabout0.6days t =47(cid:6)1ks,t =85(cid:6)2ks,t =88(cid:6)2ks.Thetimet corresponds
2 3 4 1
(52ks)asproposedbyJainetal.2009couldimplythatwemight tothedate55981.5131(cid:6)0.0116(MJD).
beseeingalmostthefulldurationoftheeclipse.Aftertheeclipse
(timeintervaln.2)anintermediatelevelofemissionisobserved,
lastingabout38ks,duringwhichthesoftX–rayintensitybelow3 3.2 TimingAnalysis
keVisindistinguishablefromtheemissionduringeclipse(Fig.2),
Barycentredlightcurveswith8secondsbintimewereproducedfor
while at harder energies the source appears uneclipsed, although
eachunitoftheX-rayImagingSpectrometer(XIS0,XIS1,XIS3)
with a reduced intensity with respect to the uneclipsed emission
in three energy bands (0.2–3, 3–10, 0.2–10 keV). We considered
displayedintheintervalsn.5andn.8.
two time ranges: i) the entire duration of the Suzaku observa-
Two flares are present peaking at 2 counts s−1 and
tion (interval from 1 to 8 in Fig. 1) and ii) the part of the obser-
4 counts s−1 in the time intervals n. 4 and n. 7, respectively. In-
vation excluding the eclipse and the transition to the uneclipsed
tervalsn.3andn.6canbeconsideredtherisetothesetwoflare
emission(timeintervalsfrom3to8inFig.1).Inordertosearch
peaks,whileintervalsn.5andn.8arecharacterizedbyaveryvari-
forspinperiodicitiesintheXISlightcurves,weusedtheLomb-
able, intra-flare emission, which covers about 40% of the orbital
Scargleperiodogrammethodbymeansofthefastimplementation
cycle.
ofPress&Rybicki(1989)andScargle(1982),whichisgenerally
The X–ray light curve shows two sharp transitions from the preferredfordatasetwithgapsandunequalsampling.Periodici-
eclipsetowardstheuneclipsedX–rayemission.Tobetterestimate tiesweresearchedinthefrequencyrangefromi)0.000013Hzor
the time of the eclipse, we simply modeled the light curve with ii)0.000022Hz(afterwhichthesensitivityisreducedduetothe
⃝c 2011RAS,MNRAS000,1–10
4 L.Sidoli,etal.
IGRJ16479−4514
Bin time: 2048. s
1
T 1
F 0.
O
S 1
0
0.
0−3
1
1
D 1
R 0.
A
H 1
0
0.
0−3
1
0
1
R
H
1
1
0.
0 5×104 105 1.5×105 2×105
Time (s)
0
0
1
0
1
R
H 1
1
0.
1
0
0. 0.01 0.1 1
Count rate (0.1−10 keV)
Figure2.Hardnessratio(hard/soft)variationswithtime(Upperpanel)andwiththesourceintensity(Lowerpanel).Thebinsizeis2048s.Softandhard
energyrangesarebelowandabove3keV,respectively.
finitelengthofthelightcurves)to0.0625Hz(correspondingtothe 3.3 Spectroscopy
Nyquistfrequencyofthedataset).Nosignificantandunambiguous
evidenceforcoherentmodulationwasfoundintheperiodograms. We first analysed the joint XIS spectra extracted from the entire
observation,correspondingtoanetintegrationtimeof136ks.The
netsourcecountrates(0.2–10keV)inthethreeXISwerethefol-
lowing: 0.157(cid:6)0.001 count s−1 (XIS0), 0.144(cid:6)0.001 count s−1
(XIS1)and0.165(cid:6)0.001counts−1(XIS3).Adoptinganabsorbed
⃝c 2011RAS,MNRAS000,1–10
SuzakuobservesoneorbitofIGRJ16479–4514 5
0.02
1 t4 Counts s keV−1−1 5×01.00−13
Count/sec 0.1 t2 2×10−3
t3
4
t1 2
χ
0.01 0
−2
0 2×104 4×104 6×104 8×104 105 1.2×105 2 5 10
Time (s) Energy (keV)
Fligighutrceur3v.eD.Touhbelfeoruarmtipm-aensdd-isstceupssfuedncitniotnhefiStteinctg.3th.1eainreitiianldpicaartteodf.theXIS 2×10−4
rtpooeofswit1dhe.ue3ra-5lpl)aso,wwaareomsru-torlnaoddwnelg6.cr4aoebnskusteiolVntreupd(utFimoiinng,.wa(4Ne)fl.HaoA(cid:24)tbdt1sda0piin2en3cgatrcaubmmne−attr2e(rp)orhwafionttG,donareiunsisnuspdlitaoeinnsxig,ltiinv(cid:0)inee, Photons cm s keV−2−1−1 105×10−4−5
a line energy of 6.37 (cid:6) 0.03 keV, compatible with being pro-
duced by Fe K fluorescence. The spectral parameters of the av-
erage X–ray spαectrum are listed in Table 1. The average flux, F, ×10−5
of1.16(cid:2)10−11ergcm−2s−1(correctedfortheabsorption),trans- 2
latesintoanX–rayluminosityof1034ergs−1(assumingadistance 2 5 10
Energy (keV)
of2.8kpc,Nespolietal.2008).Notehoweverthat,giventhelarge
uncertainty in the distance determination (Nespoli et al. 2008), Figure4. XIS spectrum extractedfrom the whole Suzaku observation.
this luminosity can range from 1.7(cid:2)1033 erg s−1 (at 1.1 kpc) to Count spectrum is shown (Upper panel), together with the residuals (in
8.2(cid:2)1034ergs−1(at7.7kpc). unitsofstandarddeviations)ofthedatatotheabsorbedpower-lawmodel
ThehardX-rayspectrumextractedfromtheHXD-PINfield (seeTable1,firstcolumn,forthespectralparameters).Lowerpanelshows
the photon spectrum, graphically rebinned to better show the residuals
of view (FOV) is much brighter than the extrapolation at higher
around6.4keV.
energiesoftheXISpower-lawbest-fit.Indeed,theHXD-PINflux
((cid:24)1.5(cid:2)10−10 erg cm−2 s−1, after taking into account CXB and
GRXE)intheenergyrange15–25keVis(cid:24)15timesgreaterthan
Table 1. Spectral results of the time averaged spectrum from the
that calculated extrapolating at high energies the power-law best Suzaku/XISobservation.(cid:0)isthepower-lawphotonindex.AGaussian
fit to the XIS spectrum. Note however that this latter extrapo- lineisrequiredat∼6.4keV(seeFig.4).Fluxisinthe1–10keVenergy
latedfluxshouldbeconsideredasaconservativeupperlimit,since rangeinunitsof10−11ergcm−2s−1andiscorrectedfortheabsorption,
IGRJ16479–4514typicallyshowsaspectralcut-offabove10keV. NH(inunitsof1022cm−2).Thefluxoftheironline,Fline,isinunitsof
This implies a strong contamination by other X–ray sources in 10−6 phcm−2 s−1.LX istheX–rayluminosity(1–10keV),inunitsof
the FOV, probably a residual contamination by the Z-track low- 1034ergs−1,foranassumeddistanceof2.8kpc.
mass X-ray binary GX340+0, located at about 34 arcmin from
IGRJ16479–4514,and/oranunknownbrighthardX–raytransient. Parameter Power-law Power-law+Gaussianline
Forthisreason,wewillnotdiscussitfurther.
The hardness ratio is variable (Fig. 2) especially during NH 9.5±0.3 9.5±0.3
eclipse,wheretheemissionappearssofterand/orlessabsorbed,so (cid:0) 1.33±0.04 1.35±0.04
wenextexploredthevariabilityofthespectralparametersindepen- Fline − 7.4±1.5
denceoftheorbitalphaseperformingatime-selectedspectroscopy EW(eV) − 70±10
extractingcountsfromthesameintervalsindicatedinFig.1.Dur- Eline(keV) − 6.37±0.03
Unabs.Flux 1.16 1.16
ingtheeclipsearemainingdetectablefluxispresent,astypicalin
eclipsing high mass X–ray binaries, mainly produced by Thom- χLX2/dof 11..00198/2172 10..09991/2169
ν
son scattering into the line of sight of the direct component by
freeelectronsresidinginthesupergiantwind(Haberl1991).Wefit
thespectrawithanabsorbedpower-lawmodel,resultinginaless
absorbed and softer spectrum during the eclipse compared to the
other time intervals (Table 2). The absorption during the eclipse
is consistent with the interstellar reddening observed to the opti-
⃝c 2011RAS,MNRAS000,1–10
6 L.Sidoli,etal.
absenceofsignaturesfromhighlyionizingilluminatingcontinuum
furtherconfirmsalowsourceX–rayluminosity.
We can distinguish several types of variability on different
timescales, especially evident in the energy range 3–10 keV. On
longtimescalesoftensofksweobservethreemainintensitystates:
theendoftheeclipsebythesupergiantcompanion(whichwecov-
eredforabout46ks),atransitionphasetotheuneclipsedemission
(timeintervaln.2inFig.1,lasting38ks)andtheintra-flarelowlu-
minosityemissionwhichcoversmostoftheremainingorbit(40%),
lastingaround110ks(timeintervalsn.5andn.8).
X–ray flares with different durations are present: two main
lowluminosityflares,lasting10–15ks,withacomplexandstruc-
turedshape,punctuatetheorbit,reachingpeakluminositiesof4.4
and 6.4(cid:2)1034 erg s−1 (at 2.8 kpc), separated by about 0.2 in or-
bitalphase.Their0.2–10keVspectraareharderthantheintra-flare
X–rayemission,asusuallyobservedinSFXTsandaccretingpul-
sars. Even in the part of the orbit in-between the two main low
luminosity flares, the source exhibits a large intensity variability
Figure6.FeKαemissionlinefluxversusthefluxintheenergyrange7–10
ontimescaleof1000s.Thislowluminosityflaringactivityhadal-
keV,correctedfortheabsorption.
readybeenunveiledinSFXTs(andinIGRJ16479–4514)thanksto
Swift/XRTlongtermmonitoringobservations(Sidolietal.2008).
InthesofterlightcurvetheX–rayeclipseappearslongerand
calcounterpart(Nespolietal.2008),whilethatmeasuredoutside thereisnoevidenceforthepresenceofthetransitionphase(time
theeclipseissignificantlyhigher,andconsistent,withintheuncer- intervaln.2)clearlypresentinthehardlightcurve.Thustheeclipse
tainties,withbeingconstant,ataboutN (cid:24)1023cm−2.TheX–ray egress appears to be energy dependent. This behaviour is simi-
H
emissionisharderwhenthesourceisbrighter(X–rayflares),atyp- lar to that reported by Jain et al. (2009) from the shape of the
icalbehaviourobservedinSFXTsandinaccretingpulsars. IGRJ16479–4514foldedlightcurvesobservedwithRXTE and
WenextaddedanarrowGaussianlinetothesinglepower-law withSwift/BAT,wheretheX–rayeclipseseemsmoreevidentat
model,resultingintheparametersreportedinTable2andshown higherenergiesandwithsharpertransitions.Thereisalsoasimilar-
inFig.5.Theresultingcontinuumpower-lawparametersarefully ityoftheIGRJ16479–4514temporalvariabilityneareclipseegress
compatible,withintheuncertainties,withthosederivedwithoutthe with what has been observed in the HMXB 4U1700–37 (van der
Gaussianline,sowedonotreportthemagaininTable2.Theen- Meeretal.2005andreferencestherein),showingavariableeclipse
ergyofthenarrowlineindicatesK emissionfromneutral(oral- durationandasimilaregressphase.Thelongduration(38ks)of
α
mostneutral)iron.Itsequivalentwidth(EW)hasbeencalculated this phase roughly corresponds to a region as large as the com-
withrespecttothepower-lawcontinuum,anditslargevalueduring panionradius(althoughaprecisedeterminationofboththedonor
theeclipseisduetothefactthatithasbeencalculatedwithrespect radiusandofthesysteminclinationislacking),probingtheinner-
tothescatteredcomponent.ThedirecteclipsedX–rayemissionis moststructureofthesupergiantwind.
unknown,giventhelargesourcevariability.Theintensityoftheflu- Onshortertimescalesofaround1000sweobservetwosharp
orescenceFeKαlineiscorrelatedwiththeunabsorbedhardX–ray transitions,indicatedinFig.3bythefourcontactsti:(1)fromthe
flux(above7keV),atleastoutsidetheeclipse(Fig.6). eclipsetotimeintervaln.2,whichshowsanintermediatelevelbe-
Finally,weadoptedapower-lawmodeltogetherwithaGaus- tweentheeclipseremainingfluxandtheun-eclipsedflux,and(2)
sianline,butnowfixingthepower-lawphotonindexto1.35,the fromtimeintervaln.2tothefirstflare.Giventhatthecompactob-
valueobtainedfromtheaveragespectrum.Thefitsareacceptable jectispoint-likewithrespecttothesupergiantcompanion,thefirst
andtheresultingvaluesoftheabsorbingcolumndensityshownow shorttransitionlikelyprobesthetransitionoutoffbehindthestellar
somevariability,againwiththelessabsorbedspectrumseendur- atmosphere,which,assuminganorbitalvelocityofafewhundreds
ingtheeclipse.Morecomplexmodelsarenotrequiredbythecount kms−1,resultsinalinearsizeofaroundafew1010cm.Thesecond
statistics of the data and result in unconstrained spectral parame- sharptransitionisinterestinglycoincidentwiththerisingphaseto
ters. thefirstX–rayflare,suggestingapossiblephysicalconnection(see
thediscussionbelow).
Anironlineisdetectedalongtheorbit,withalineenergyof
6.4keV,indicativeofneutralironoranionizationstatelowerthan
4 DISCUSSION
FeXVIII(Kallman&McCray1982).Acorrelationisobservedbe-
The long Suzaku observation we have reported on here, allows tweentheironfluorescentlinefluxandtheilluminatingunabsorbed
ustoinvestigatethepropertiesofaSFXTascontinuouslyaspos- flux above 7 keV, as expected (Inoue 1985) outside the eclipse
sible within one single orbital cycle for the first time. This has (Fig.6).Thelinefluxobservedduringeclipsedoesnotshowev-
beenmadepossiblebytheshortorbitalperiodofIGRJ16479–4514 idenceforareduction,withinthelargeuncertainties,withrespect
whichistheSFXTwiththeshortestorbitknowntodate.TheX–ray to the value detected during the intra-flare emission, so most of
source shines with an average low luminosity L (cid:24)1034 erg s−1, thefluorescingmatterislocatedfarawayfromthecompactobject.
X
(ifadistanceof2.8kpcisassumed)andnobrightflares(exceeding Notethatwecalculatetheequivalentwidthofthelinewithrespect
L (cid:24)1036ergs−1)havebeencaughtduringthisobservation.Nev- toasinglepower-lawcontinuum.Ifthefluorescingmatterisopti-
X
ertheless,ahighamplitudevariabilityisobserved,spanningmore callythintotheilluminatingcontinuum,theluminosityoftheiron
than two orders of magnitude (including the X–ray eclipse). The K linedependsonthespectrumandX–rayluminosityofthecen-
α
⃝c 2011RAS,MNRAS000,1–10
SuzakuobservesoneorbitofIGRJ16479–4514 7
Table2.Suzaku/XISresultsofthetime-selectedspectroscopy(numbersmarkthesametimeintervalsdisplayedinFig.1)withanabsorbedpower-law
model.Fluxesareinunitsof10−11ergcm−2s−1.Bothobserved(notcorrectedfortheabsorption)andunabsorbed(correctedfortheabsorption)fluxesare
reported.LXistheX–rayluminosity(1–10keV),inunitsof1034ergs−1,foranassumeddistanceof2.8kpc.Theabsorbingcolumndensity,NH,isinunits
of1022cm−2.
1 2 3 4 5 6 7 8
Power-law
NH 4.6+−32..41 8.5+−22..61 9.9+−11..96 10.6−+11..10 9.8+−00..77 9.9+−00..87 9.9+−00..76 9.8+−00..76
(cid:0) 3.1+−11..50 1.46+−00..3352 1.07+−00..2221 1.16−+00..1133 1.42+−00..0099 1.28+−00..1009 1.21+−00..0098 1.49+−00..0088
Obs.Flux(1-10keV) 0.014 0.13 1.23 2.9 0.7 3.6 4.2 0.6
Unabs.Flux(1-10keV) 0.045 0.22 1.9 4.7 1.2 5.9 6.8 1.1
LX 0.042 0.21 1.8 4.4 1.1 5.5 6.4 1.0
χ2/dof 1.672/48 1.047/64 0.836/88 1.092/227 0.922/634 1.001/435 0.887/518 1.130/656
ν
Power-law+Gaussianline
Fline(10−5phcm−2s−1) 0.32+−00..2129 0.57±0.27 2.7+−11..55 <1.9 0.65+−00..3344 4.9+−22..51 5.4+−22..22 0.54+−00..2277
EW(eV) 5500+−43570000a 280±160 130±80 <40 60±30 80±40 80±30 50±30
Eline(keV) 6.57+−00..2226 6.33±0.20 6.41+−00..0087 6.4(fixed) 6.38±0.06 6.34±0.06 6.37+−00..0045 6.32±0.09
χ2/dof 1.576/45 0.934/61 0.771/85 1.097/226 0.910/631 0.973/432 0.861/515 1.118/653
ν
Unabs.Flux(7–10keV) 0.0009 0.05 0.60 1.40 0.30 1.63 1.97 0.26
Power-law+Gaussianline
NH 1.3+−00..65 7.3+−11..31 12±1 11.9±0.6 9.2±0.3 10.2±0.4 10.6±0.4 8.8±0.2
(cid:0) 1.35fixed 1.35fixed 1.35fixed 1.35fixed 1.35fixed 1.35fixed 1.35fixed 1.35fixed
χ2/dof 1.920/46 0.923/62 0.799/86 1.118/227 0.913/632 0.963/433 0.870/516 1.132/654
ν
aThelargeEWduringeclipseiscalculatedwithrespecttothescatteredpower-lawcontinuum.
tralsource,onthecoveringfractionoftheilluminatedmatterand emission(timeintervalsfromn.3ton.8).Apossibilityisthatmost
ironabundance(e.g.Sakoetal.1999).Giventhelargeuncertain- ofthedirectX–rayemissioninthisorbitalregionisblockedbya
tiesontheEW(especiallyhugeduringtheeclipsephase)andthe large scale wind structure, as that invoked to explain the period-
variabilityofthecoveringfactoralongtheorbitalphase,alsocon- ically recurrent outbursts in the SFXT IGR J11215–5952 (Sidoli
sideringthelikelypresenceofdifferentlarge-scalestructuresinthe et al. 2007), or by dense shells or other kind of inhomogeneities
supergiantwindwhichcanbeobvious,althoughunknown,sitesof presentinhotmassivestarswinds(see,e.g.,Oskinovaetal.2012
fluorescingmaterial,itisnotpossibletogainfurtherconstraintson or Lobel & Blomme 2008). This implies that most of the X–ray
thepropertiesofthereprocessingwind(alsothesupergiantsurface emissioninthistimeintervalisscatteredX–rayemission,whichis
becomesareprocessoronlyvisibleatcertainorbitalphases). lessabsorbedbecausemainlyproducedbylessdensewindmaterial
Consideringthespectroscopy,amorephysicaldescriptionof locatedofftheorbitalplaneandfartherawayfromthecompanion
theX–rayspectruminaneclipsingHMXBinvolvesthepresence star.Thiscouldexplainwhytheobservedcolumndensityintime
oftwopower-lawcontinua,accountingforthedirectandscattered intervaln.2isnotlargerthanintheremaining(fullyout-of-eclipse)
componentwiththelatterproducedbyThomsonscatteringbyfree partoftheorbit.
electron in the companion wind and/or on the supergiant surface Thelikelypresenceoflargescalestructuresinthesupergiant
(Haberl1991).Boththedirectandthescatteredcomponentsshould windisalsosuggestedbytheorbitalphaseofthefirstflare,which
havethesamespectralslope,sinceThomsonscatteringisenergy iscompatible(withintheuncertainties,seeSect.3.1)withthelo-
independent. This implies that the only way to disentangle these cationalongtheorbitofotherbrightflaresobservedinthepastin
two components is the different absorptions, the scattered com- IGRJ16479–4514(ϕ(cid:24)0.36;Bozzoetal.2009).Thissuggeststhat
ponentbeinglessabsorbedthanthedirectone.DuringtheX–ray the flares are triggered by a higher accretion rate during the pas-
eclipse,wherethedirectcomponentisunseen,theX–rayemission sage inside a phase-locked denser wind component, like the one
shouldonlybeduetoscattering.Indeed,fittingthisspectrumwitha predictedbySidolietal.(2007).Thisseemstobealsoindicatedby
power-lawwithaphotonindexfixedattheaverageslopeof(cid:0)=1.35, the fact that, interestingly, the egress X–ray emission (n. 2) ends
theeclipseshowsthelowestabsorption,consistentwiththeinter- with the sharp transition to the first X–ray flare (see Sect. 3.1).
stellar value (Nespoli et al. 2008). Unfortunately, given the high Ifthisexplanationiscorrect,afterthefirstflare,thecompactob-
absorption(localandinterstellar)andthelowstatistics,wecould jectliesin-betweentheobserverandthedensewindstructure,with
notfindbettercontraintstothespectralparametersadoptingmore thedirectX–rayemissionfinallydominatingtheobservedX–rays
complexmodels. alongtheremainingpartoftheorbit.
AlsotheegressX–rayemission(n.2)inIGRJ16479–4514is Thesecondflare,spacedbyabout0.2inphase,couldberec-
likelydominatedbythescatteredcomponent,giventhehighEWof onciledbythecrossingofasimilarwindcomponent,iftheorbitis
theironlinecomparedtothatmeasuredduringtheout-of-eclipse eccentric(asinthelightcurvessimulatedbyDuccietal.2009),or
⃝c 2011RAS,MNRAS000,1–10
8 L.Sidoli,etal.
Figure5.Variabilityofthebest-fitspectralparametersadoptinganabsorbedpower-lawmodeltogetherwithaGaussianlineat6.4keV(seeTable2).The
observed(notcorrectedfortheabsorption)fluxisintheenergyrange1–10keVandinunitsofergcm−2 s−1.Thefluxoftheironlineisinunitsof
photonscm−2s−1.Theorbitalphaseisindicatedintheupperx-axis(seetext).
bythecompactobjectapproachingtheperiastronpassage.Unfortu- abilityispresent,withmoreabsorptionduringthelowluminosity
nately,welackanyinformationabouttheotherorbitalparameters flares.
(e.g.eccentricity)tobeabletoconfirmthishypothesis.
Theonlyknownorbitalparameteristheperiodof3.32days.
Theabsorbingcolumndensityisalwaysinexcessoftheinter- Wedidnotobservetheeclipseingress,sowecannotconstrainthe
stellarreddeninganddoesnotshowevidenceforvariabilityalong companionradius,R .Ontheotherhand,R canbeestimated
opt opt
theorbit(ifalsotheslopeofthepower-lawismantainedfree),ex- fromthesupergiantspectraltypeandrequiringthatthestardoesnot
cept that during the eclipse. On the other hand, if the power-law overflowitsRochelobe.Thespectraltypeofthesupergiant(O8.5I
slopeisfixedtothevalueoftheaveragespectrum((cid:0)=1.35),vari- or O9.5I) suggests a mass in the range Mopt=30–34 M⊙ and a
⃝c 2011RAS,MNRAS000,1–10
SuzakuobservesoneorbitofIGRJ16479–4514 9
radiusof22-23R⊙ (Martinsetal.2005).Afurtherconstrainton 5 CONCLUSION
thestellarsizecomesfromthefactthatthecompanionshouldnot
ThenewSuzakuobservationswehavereportedhereallowedus
overlflow its Roche lobe radius (Eggleton 1983). This suggests a
toperformanin-depthinvestigationofthepropertiesoftheSFXT
donormassaroundMopt=35 M⊙andastellarradiusRopt=20R⊙. IGRJ16479–4514alongoneorbitalcycle,andtoobtainthefollow-
Adoptingthesevalues,theorbitalperiodtranslatesintoanorbital
ingresults:
separationofabout2.2(cid:2)1012cm.
(cid:15) the source spends most of the observation with an average
We can use the X–ray eclipse to probe the supergiant prop-
out-of-eclipse(intra-flares)X–rayluminosityof1034ergs−1,with
erties, as follows (see, e.g., Lewis et al. 1992). The average total
several kinds of variabilities on different timescales, but with no
intensityduringtheeclipseisabout5%oftheout-of-eclipse,intra-
flare X–ray intensity, so the wind density can be obtained as n brightflares(exceeding1036 ergs−1)duringtheobservedorbital
w
=0.05/(aσ ),whereσ istheThomsonscatteringcrosssection cycle. Given long-term monitoring performed with previous mis-
T T
andaistheorbitalseparation,whichweassumeasthecharacteris- sions,thisisverylikelythemosttypicalappearenceofanorbital
cycleinthisSFXT;
ticpathlenghtthroughthesystem.Thisresultsinawinddensityat
theorbitalseparation,ρ (a),of7(cid:2)10−14gcm−3.Themasscon- (cid:15) the absorbing column density does not show evidence for
w
tinuityequationcanbeusedtoderivetheratiobetweenthewind variability,withintheuncertainties,exceptfromduringtheX–ray
mRass)-loρss(raa)t,eaasnsdumthiengteramsipnhaelrvicealolcgiteyo,masetMry_wfo/rv∞th=e4oπutflao(wain(cid:0)g ecl(cid:15)ipsthe;e remaining flux during the X–ray eclipse, produced by
opt w
wind and a velocity gradient for the wind velocity, β, of 1 (Cas- Thomsonscattering,allowedustoestimatethewinddensityatthe
Ftoorrettearlm.1in9a7l5v).elTohciistireessuinltsthienararnagtieo5M0_0w–/3v0∞00=7k(cid:2)m1s0−−11,7tMhe⊙m/kamss. orb(cid:15)itaalsssuepmairnagtioan,crirecsuullatirngoribnitρwan(da)a=7s(cid:2)ph1e0r−ic1a4lggcemom−e3t;ry for the
lroecsst arcactereitsioninfrtohme rtahnegewiMn_dw,=th1e–7a(cid:2)cc1re0t−io6nMra⊙te/ycra.nAbsesuemstiinmgatdeid- sMu_pwe/rgvi∞ant=w7i(cid:2)nd1,0t−h1e7dMer⊙iv/ekdmw.Sinindcedetnhseitaycctrreatnisolnatleusmiinntoosiatyraimtio-
asM_ =(πR2 /4πa2)(cid:1)M_ ,whereR istheaccretionradius. pliedbythisratio,assumingterminalvelocitiesintherange500–
This aaccccretionacrcate translatwes into anaXcc–ray luminosity L =3– 3000kms−1,isatleasttwoordersofmagnitudehigherthanthat
X
15(cid:2)1036 erg s−1 for the range of parameters estimated before, observed,wecanconcludethatamechanismmediatingtheaccre-
whichistwoordersofmagnitudegreaterthanthatweobserve. tionontotheputativeneutronstarinthesystemislikelytobeat
worktoreducethemassaccretionrate.
Giventhehighwinddensitywehavecalculated,itisunlikely
that the low luminosity in IGRJ16479–4514 is due to the direct
wind accretion. Our findings seems to agree with the recent re-
sultsbyOskinovaetal.(2012),whohavedemonstratedthatsimple ACKNOWLEDGMENTS
Bondi-Hoyleaccretionfromaclumpywindover-predictstheob-
ThisworkisbasedondatafromobservationswithSuzaku.This
servedX–rayvariabilityinHMXBs.Thissuggeststhepresenceof
work was supported by the grant from PRIN-INAF 2009, “The
somemechanismabletodampthestrongX–rayvariationsimplied
transientX–raysky:newclassesofX–raybinariescontainingneu-
bythestructuredsupergiantwinds(Oskinovaetal.2012).Thiscan
tronstars”(PI:L.Sidoli).ABreceivedfundingfromNASAgrant
beduetothedetails(stillpoorlyknown)oftheinteractionofthe
11-ADAP11-0227.LSisgratefultoAngelaBazzanoandTimOost-
accretionflowwiththeshocksintheaccretionwake,leading,for
erbroekforveryhelpfulcommentsandsuggestions.PEthanksSara
example, to a transitional case of accretion regime, intermediate
Motta for fruitful discussions. This research has made use of the
betweenRocheLobeOverflowanddirectaccretion(asoriginally
IGR Sources page maintained by J. Rodriguez & A. Bodaghee
proposed for the SFXT IGR J16418–4532, given its short orbital
(http://irfu.cea.fr/Sap/IGR-Sources/).
period(Sidolietal.2012),whichdisplaysasimilarX–rayintensity
variability).
Theroleofthemagnetosphericsurfaceinreducingtheaccre-
REFERENCES
tionrateontotheNSinSFXTshavebeenstudiedbyseveralauthors
(Bozzoetal.2008a,Duccietal.2010,Shakuraetal.2012b).Bozzo Balucinska-ChurchM.,McCammonD.,1992,ApJ,400,699
etal.(2008)haveinvokedmagnetar-likeNSwithslowpulsations, BozzoE.,FalangaM.,StellaL.,2008a,ApJ,683,1031
to halt the accretion most of the time and allow only a residual Bozzo E., Giunta A., Stella L., Falanga M., Israel G., Campana
flow of matter (producing (cid:24)1034 erg s−1) by means of Kelvin- S.,2009,A&A,502,21
Helmholtzinstability.Shakuraetal.(2012a)haveproposedthatthe Bozzo E., Stella L., Israel G., Falanga M., Campana S., 2008b,
lowX–rayluminositystate((cid:24)1034 ergs−1)sometimesobserved MNRAS,391,L108
inpersistentaccretingpulsarsandinSFXTscouldbeduetosub- CastorJ.I.,AbbottD.C.,KleinR.I.,1975,ApJ,195,157
sonicquasi-sphericalaccretionontoslowlyrotatingpulsars,where ChatyS.,RahouiF.,FoellmiC.,TomsickJ.A.,RodriguezJ.,Wal-
theaccretionofmatterismediatedbyaquasi-staticshellabovethe terR.,2008,A&A,484,783
NSmagnetosphere.Inthiscase,theaccretionratefromthequasi- Ducci L., Sidoli L., Mereghetti S., Paizis A., Romano P., 2009,
staticshelldependsontheabilityoftheplasmatoenterthemag- MNRAS,398,2152
netospherebymeansofplasmacooling.Theseauthorssuggestthat DucciL.,SidoliL.,PaizisA.,2010,MNRAS,408,1540
transitionsbetweenlow-luminositystatestobrightflaringactivity EggletonP.P.,1983,ApJ,268,368
inSFXTscouldbeduetotransitionsbetweentwodifferentregimes FukazawaY.etal.,2009,PASJ,61,17
ofplasmacooling:fromthermalplasmacoolingtoComptoncool- Gruber D. E., Matteson J. L., Peterson L. E., Jung G. V., 1999,
ingdominatedregime.Unfortunately,noneofthesemodelscanbe ApJ,520,124
confirmedtodate,sinceboththepulsationperiodandthemagnetic HaberlF.,1991,A&A,252,272
fieldinIGRJ16479–4514(andinmostofSFXTs)areunknown. InoueH.,1985,SpaceScienceReviews,40,317
⃝c 2011RAS,MNRAS000,1–10
10 L.Sidoli,etal.
JainC.,PaulB.,DuttaA.,2009,MNRAS,397,L11
KallmanT.R.,McCrayR.,1982,ApJS,50,263
KenneaJ.A.,PaganiC.,MarkwardtC.,BlustinA.,CummingsJ.,
NousekJ.,GehrelsN.,2005,Astron.Tel.,599,1
KoyamaK.etal.,2007,PASJ,59,23
LaParolaV.etal.,2009,Astron.Tel.,1929,1
Lewis W., Rappaport S., Levine A., Nagase F., 1992, ApJ, 389,
665
LobelA.,BlommeR.,2008,ApJ,678,408
MarkwardtC.B.,KrimmH.A.,2006,Astron.Tel.,816,1
MartinsF.,SchaererD.,HillierD.J.,2005,A&A,436,1049
MitsudaK.etal.,2007,PASJ,59,1
MolkovS.,MowlaviN.,GoldwurmA.,StrongA.,LundN.,Paul
J.,OosterbroekT.,2003,Astron.Tel.,176,1
NespoliE.,FabregatJ.,MennickentR.E.,2008,A&A,486,911
Oskinova L. M., Feldmeier A., Kretschmar P., 2012, MNRAS,
421,2820
Rahoui F., Chaty S., Lagage P.-O., Pantin E., 2008, A&A, 484,
801
RattiE.M.,BassaC.G.,TorresM.A.P.,KuiperL.,Miller-Jones
J.C.A.,JonkerP.G.,2010,MNRAS,408,1866
RomanoP.etal.,2009,MNRAS,399,2021
RomanoP.etal.,2008,ApJL,680,L137
SakoM.,LiedahlD.A.,KahnS.M.,PaerelsF.,1999,ApJ,525,
921
SgueraV.etal.,2005,A&A,444,221
SgueraV.etal.,2008,A&A,487,619
SgueraV.etal.,2006,ApJ,646,452
ShakuraN.,PostnovK.,HjalmarsdotterL.,2012a,ArXive-prints
Shakura N., Postnov K., Kochetkova A., Hjalmarsdotter L.,
2012b,MNRAS,420,216
SidoliL.,2010,in25thTexasSymposiumonRelativisticAstro-
physics
SidoliL.,MereghettiS.,SgueraV.,PizzolatoF.,2012,MNRAS,
420,554
SidoliL.etal.,2008,ApJ,687,1230
Sidoli L., Romano P., Mereghetti S., Paizis A., Vercellone S.,
ManganoV.,Go¨tzD.,2007,A&A,476,1307
TakahashiT.etal.,2007,PASJ,59,35
ValiniaA.,MarshallF.E.,1998,ApJ,505,134
vanderMeerA.,KaperL.,diSalvoT.,Me´ndezM.,vanderKlis
M.,BarrP.,TramsN.R.,2005,A&A,432,999
WalterR.,ZuritaHerasJ.,2007,A&A,476,335
WalterR.etal.,2006,A&A,453,133
WilmsJ.,AllenA.,McCrayR.,2000,ApJ,542,914
ThispaperhasbeentypesetfromaTEX/LATEXfilepreparedbythe
author.
⃝c 2011RAS,MNRAS000,1–10