Astronomy&Astrophysicsmanuscriptno.spectralQPO˙AandA˙V3 (cid:13)c ESO2016 January13,2016 A possible imprint of quasi-periodic oscillations in the X-ray spectra of black hole binaries P.Varniere1,R.Mignon-Risse1,andJ.Rodriguez2 1 APC, AstroParticule et Cosmologie, Universite´ Paris Diderot, CNRS/IN2P3, CEA/Irfu, Observatoire de Paris, Sorbonne Paris Cite´, 10, rue Alice Domon et Le´onie Duquet, 75205 ParisCedex13,[email protected] 6 1 2 LaboratoireAIM,CEA/IRFU-CNRS/INSU-Universite´ ParisDiderot,CEADSM/IRFU/SAp, 0 F-91191Gif-sur-Yvette,France. 2 n Received¡date¿/Accepted¡date¿ a J ABSTRACT 2 1 Context. While nobody would deny the presence of Quasi-Periodic Oscillations in the power ] E densityspectrumofBlackholebinariesnortheirimportanceintheunderstandingofthemecha- H nismspoweringtheX-rayemissions,thepossibleimpactonthetime-averageddiskenergyspec- . trumfromthephenomenonresponsiblefortheQuasi-PeriodicOscillationsislargelyignoredin h p modelsofsourcesemission. - o Aims.Hereweinvestigatethepotentialimpactofsuchstructureontheresultantenergyspectrum. r t Methods.UsingdatafromthewelldocumentedoutburstsofXTEJ1550-564welookedatpos- s a siblehintsthatthepresenceofQuasi-PeriodicOscillationsactuallyimpactstheenergyspectrum [ emitted by the source. In particular we look at the evolution of the relation between the inner 1 discradiusandtheinnerdisctemperatureobtainedfromfitstothespectraldata.Wethentestthis v 6 furtherbydevelopingasimplemodelinordertosimulatespectraofadiskwithastructuremim- 1 ickinganincreasinginstrengthQuasi-PeriodicOscillationandcomparethesimulatedresultsto 8 2 thoseobtainedfromrealdata. 0 Results.Wedetectasimilardepartureintheinnerradius-innertemperaturecurvecomingfrom . 1 thestandardfitofoursimulatedobservationsasisseeninXTEJ1550-564data.Weinterpretour 0 6 resultsasevidencethatthestructureattheoriginoftheQuasi-PeriodicOscillationimpactthe 1 energyspectrum. : v Conclusions. Furthermore, in states with a significant disk emission the inaccuracy of the de- i X terminationofthediskparametersincreaseswiththestrengthoftheQuasi-PeriodicOscillation, r an increase which then renders the value given by the fit unreliable for strong Quasi-Periodic a Oscillation. Keywords.X-rays:binaries,stars:individual(XTEJ1550-564),accretiondisks 1. introduction WhenlookingatthePowerDensitySpectrum(PDS)ofmicroquasarsthemoststrikingfeaturesare the presence of narrow peaks, called Quasi-Periodic Oscillations (QPO). These Low-Frequency (< 30Hz, LF) and High-Frequency (> 40Hz, HF) QPOs contain a significant part of the source 1 Varniere,Mignon-Risse&Rodriguez:spectralQPOs rapid variability. Some LFQPOs can have an amplitude of up to 30%. Several distinct models exist to describe them, and many imply a warm/hot structure orbiting the disk causing the X- ray modulation. Such structures include axisymmetric (see e.g. Bursa et al., 2004; Schnittman et al.,2006b;Vincentetal.,2014),orprecessingtori(Schnittmanetal.,2006c;Ingrametal.,2009), hotspots(Karasetal.,1992;Schnittman&Bertschinger,2004;Tagger&Varniere,2006;Pechacek etal.,2013)orspirals(Tagger&Pellat,1999;Varniereetal.,2002). Nevertheless,whenanalyzingthesourcesenergyspectrathediskismostlyconsideredasrela- tivelyhomogeneousandwithasmooth,monotonictemperatureprofile.Ifsuchafeaturelessdisk maybeagoodrepresentationfortheso-calledthermalstatesduringwhichtheenergyspectrare- semble “pure” blackbody and the PDS associated show little or no variability, with no or weak QPO, this model is not appropriate to describe states with prominent QPOs. Here we aim to ex- ploreifthestructureattheoriginofQPOsindeedhasameasurableimpactontheenergyspectrum anditsfits. The shape and the existence of more than one component in the X-ray spectra makes it hard to look at a direct impact from the structure at the origin of the QPO. Nevertheless, we can look at a possible impact of the presence of QPOs on the spectral fit parameters and the correlations between them. Many correlations between spectral parameters and QPO parameters have been presented(Munoetal,1999;Remillardetal.,2002;Rodriguezetal,2002;Vignarcaetal,2003) butnotmanystudieshavebeendonetoinvestigateanychangeinthecorrelationbetweenspectral parametersdependingontheQPOpresence. In section 2 we use data from two outbursts of XTE J1550-564 observed by RXTE, which have the advantage of being long and well observed, to see if some trends are indeed visible in the correlation between spectral parameters depending on the presence of QPOs. In order to test furtherthisidea,wepresentinsection3asimplifiedmodelofadiskhavinganincreasinglystrong QPOthatwethenusetocomputethesource’senergyspectrum.Insection4weusethismodelin conjecture with the module fakeit from XSPEC to test the impact on the energy spectrum of a hotstructureinthedisk.ThisallowsustodirectlymeasuretheimpactoftheQPOasthedifference betweenthemodelparametersandtheresultingfitsbutalsotocomparethosesimulatedfitsresults againstrealcorrelations. 2. LinkbetweenthediskparametersandQPOs IthasbeenassertedearlyonthatthepropertiesoftheLFQPOs,inparticulartheirfrequencies,are related to parameters of the disk (Muno et al, 1999; Rodriguez et al, 2002; Vignarca et al, 2003; Varniere et al., 2002) obtained through spectral fitting. As there is much less data for HFQPOs, nosimilarstudyhasbeendoneyetbutHFQPOsseemtobelinkedwiththepresenceofLFQPOs of type A or B (Remillard et al., 2002). Here we are using data from XTE J1550-564 during the outbursts of 98-99 and 2000, both known to harbor HFQPOs, to see if there is any difference in the behavior of the different spectral parameters depending if there is or not a QPO of any types observed. 2 Varniere,Mignon-Risse&Rodriguez:spectralQPOs 2.1. ObservationsandDataReduction XTEJ1554−564hasundergone5differentoutburstsduringthelifetimeofRXTE.Thelastthreeoc- curredin2001,2002,and2003areconsideredas“failed”sincethesourcedidnotshowanyspectral transitiontosoftflavored(i.e.disk)states.Wethereforeconsiderhereonlythefirsttwooutbursts that also are the two brightest. They respectively occurred from 1998, Sept 6, until roughly the beginning of May 1999 for the discovery one and from 2000, April 10 to 2000, July 16 for the second and fainter outburst. Both outbursts have been extensively described in the literature (e.g Sobczaketal,2000;Remillardetal.,2002;Milleretal.,2001;Rodriguezetal,2003,2004).Inthe followingstudyweconsideronlytheobservationsshowingthepresenceofLFQPOand/orHFQPO as reported in these articles. The QPO parameters are taken from Remillard et al. (2002); Miller et al. (2001); Rodriguez et al (2004). Since the spectral calibration of the Proportional Counter Array (PCA) has significantly evolved since the publication of these articles, we re-reduced and re-analyzedthespectraldatainordertopresentthemostaccuratespectralparametersbutwekept theparameters.Inordertocomparewiththepreviousresults(e.gRemillardetal.,2002)wekept thedistance(D = 6kpc)andinclination(70◦)thatwereusedintheseearlierpapers.Thedistance andinclinationweuseareonlyascalingfactorappliedtotheradius(seefullexpressionifeq1). Aslongasweusethesamefortheobservationsandthesimulatedobservationswecancompare theresults. ThePCAdatawerereducedwiththeHEASOFTv6.16suite.Goodtimeintervalsweredefined as intervals with a satellite elevation angle above the Earth greater than 10◦, an offset between the pointing and the source direction less than 0.02◦, and recommended limitations on propor- tionalcounterunits(PCU)potentialbreakdowns.Spectrawerethenextractedfromthetoplayerof PCU2.Backgroundspectrawereobtainedwiththebrightbackgroundmodelandgeneratedwith pcabackest, while response files were generated through pcarsp. 0.8% systematic errors were addedtoallspectralchannelsbeforefitting. ThePCAspectrawerefittedwithXSPECv12.6.2,between3and30keV.Aswewereinterested in basic parameters we only considered simple spectral models. All spectra are well represented with an absorbed (phabs) disk (diskbb) plus power law (powerlaw) model. The value of the highspectralboundaryofourfitsdidnotallowustosignificantlyconstrainthepotentialpresence andpropertiesofareflectionbump.WeneverthelessnotethataGaussianat6.5keVrepresenting fluorescentemissionofiron,usuallytakenasasignatureforreflection,was,inafewcases,added tothemodeltoobtainagoodfit. TheabsorptionwasfixedtoN =0.8×1022cm−2(Milleretal., H 2003). The spectral fit gives access to the slope of the power-law (i.e. the photon index Γ), its nor- malization,thetemperatureoftheinneredgeofthediskandthedisknormalization.Thelatteris relatedtotheinnerapparentdiskradiusR followingNorm=((R /km)/(D/(10kpc)))2cosiwith in in Dthedistancetothesourceanditheviewingangleofthedisk.Itisknownthattheapparentradius relatestotherealinnerradiusthrougha”hardeningfactor” f (Shimura&Takahara,1995).Here wechoosetokeepthatR andnotaddacorrectionfactorasweareinterestedintherawoutputof in thefitthatwethenintendtocomparewithoursimulatedobservations. 3 Varniere,Mignon-Risse&Rodriguez:spectralQPOs 2.2. Departurefromr -T correlation in in In Fig. 1 we represent the evolution of the disk parameters r vs. T obtained from the spectral in in fitstothedata.Twobehavioremergefromthisplot,atlargeradiusthemajorityofpointsnarrowly Tin (keV) vs Rin (km)(HFQPO in red) 1.2 LFQPOType (black C, blue B, red A) Fit of Tin (keV) vs Rin (km) 1.0 V) 0.8 e k n ( Ti 0.6 0.4 20 50 100 Rin (km) Fig.1. Correlation between the inner edge position and inner edge temperature as directly given by the spectral fits for the outburst of 98-99 and 2000 of XTE J1550-564. Red dots represent observations with HFQPOs detected while there is none in the black dots. The crosses represent thetypeofLFQPO,blackforthecommontypeC,bluefortypeBandredfortypeA.Timingdata fromRemillardetal.(2002);Rodriguezetal(2002). followapower-law,whilethereisanobviousdepartureastheinneredgeofthediskgetssmaller than50km.InterestinglythisdepartureoccurswhenHFQPOsorsimilarlytypeA/BLFQPOsare observed1.Itisalsointerestingtonotethatthisdepartureisalwaysinthesamedirection,namely, atsimilarinnerradius,wegetahotterdiskinthecaseofadetectedHFQPO/typeA/BLFQPOthan in the standard case with a type C (hard state). This may be compatible with the HFQPOs being somekindofhotstructureinthedisk. Atthispoint,onehastowonderif,asallthosedatapointscomefromthespectralfittingwitha smoothmonotonicdisk,itispossiblethatthedeparturefromthecorrelationcouldberootedfrom thepresenceofaQPO/warmstructurewhichcausesthefittofailandgivesinaccuratevalues.In ordertotestthisassumption,thatinthepresenceofalocalizedhotstructureinthediskspectral fitusingasmoothtemperatureprofilewillgiveinaccuratevalue,wewillperformfitsofadisk withsuchstructurebutinacontrolled,hencesimulated,mannerandcomparetheresultsofthefit withtherealvalues. 3. Simplemodelforthetemperatureprofile QPO models often connect the signal modulations to the presence of structures embedded in the disk.Ourmainconcernhereisnottheoriginofthestructure,buttheconsequencesontheemission andifiteverneedstobetakenintoaccountwhendoingaspectralfit.Thismeansthat,ratherthan takingfullmagnetohydrodynamic(MHD)simulationsofthedifferentmodelsproposedtoexplain QPOs we are interested in, we decided to create a simple, analytical, model in order to test in a cleanerandeasierway,itsimpact.Indeed,inafullfluidsimulationchangingoneparameterinthe initialconditioncanhaverepercussionsonseveralobservableparametersandthereforeitisharder 1 Asasidenotewearelookingtoseeifthefewpointsthatdepartfromthecorrelationwithoutapublished HFQPOfrequencyhaveornotsomehighfrequencystructurefineenoughtobeaHFQPO. 4 Varniere,Mignon-Risse&Rodriguez:spectralQPOs tostudythedifferenteffectsseparately. Here we are using a similar perturbative approach as in Varniere & Blackman (2005) and take a regular diskbb disk model with T (r) ∝ r−3/4 on which we add a component mim- 0 icking the structure at the origin of the QPOs that depends on radius and azimuthal angle as T (r,ϕ) = h(r) . s(r −r ,ϕ). For simplicity we choose to decompose T as a height function h 1 s 1 that depends only on r and a shape function s which is finite only near the disk structure we are studyingwhichisinturndefinedbyr =r (t,ϕ).Alsoforsimplicitywetaketheshapefunctionto s s begaussianandtheheightfunctiontoberelatedtotheequilibriumtemperatureT (r). 0 Usingthisstructurewecantakeintoaccountavarietyofshapesmimickingavarietyofmodels, forexample: -atoruswithr (t,ϕ)=cte×sin(νt)andνtheoscillatingfrequencyofthetorus, s -ahotspotwithr (t,ϕ)=cteforthehotspotazimuthalsizeandzerootherwise. s Inthegeneralcasethisconstant,whichdefinesthepositionofthestructureinthedisk,iscalledr , c thecorotationradiusofthestructure.Inthecaseofnon-axisymmetricalstructuressuchasablob Ω(r ),thefrequencyatwhichthestructureisrotatingisalsothefrequencyatwhichthefluxwill c bemodulated,hencetheQPOfrequency. Thisprovidesasimplebutusefulframeworktomodeladiskwithaddedperturbativestructures. Withinthisframeworktheperturbedtemperaturereads T(r,ϕ) = T0(r)×1+γexp−21(cid:32)r−δrrs(ϕ)(cid:33)22 (1) c where r (t) is the position of the temperature maximum in the disk (the center of the torus or c hotspot). δ parametrizes the radial extent of the structure while γ is the maximum amplitude of the perturbation. In the general case r is allowed to be a function of time to take into account a c changeinfrequencyofthemodulation,buthereweareconsideringonlythecasewheretheQPO frequencyisstableduringone‘simulatedobservation’hencekeepingr aconstant.Indeed,inthe c blobmodelthefrequencyofthemodulationisΩ(r ).Inthatcase,thewidthofthepeakinthePDS c representstheradialextentofthestructure,hereparametrizedbyδ. These last two parameters, (γ,δ) are the ones we will be using to mimic a growing flux modulation. Indeed, using similar models we are able to reproduce several timing observables suchas,forexample,thermsamplitudeofQPOsasshowninVarniere&Blackman(2005)fora pseudo-newtonian potential and Varniere & Vincent (2016) in general relativy, hence validating thissimplemodelasarepresentationofadiskgivingrisetoafluxmodulation,henceaQPO.All the simulations presented in Table 1 represent an evolving flux modulation from zero to about ∼20%hencewellinsidetheobservedlimits. Using these parametrized temperature profiles we then can compute the emitted thermal flux fromtheentirediskconsideringitmadeofmultipleblackbodycomponentsinasimilarmanneras 5 Varniere,Mignon-Risse&Rodriguez:spectralQPOs doneinthewidelyusedXSPECmodeldiskbbforamonotonicprofile: r2 2E3 (cid:90) 2π(cid:90) ∞ r f(E)= in cosi (cid:63) dr dφ (2) D2 c2h3 E (cid:63) 0 1 ekbT(r(cid:63),φ) −1 where r is the radius in units of r the inner disk radius, k is Boltzmann’s constant, h is (cid:63) in b Planck’s constant and c the speed of light. The only difference with the diskbb model is that our T(r,φ)isanon-monotonicfunctionofr andalsoofφ,hencewehavetokeepbothintegrals.We thencreatedanXSPECmodelofdiskwiththisstructure(whichwenameddiskblobforthecase ofanelongatedhotspot)whichwillbemadeavailableonceoptimized. This allows two things: first we can fit observations with our non-monotonic disk profile and second,usingtheprocedurefakeitinXSPECwecancreatesyntheticspectraofapower-lawplus thedisktakingintoaccountthepresenceofawarm/hotstructure.Usingthesimulatedspectrawe canthentestifthepresenceofastructurecausingaQPOinthePDSwouldalsohaveadetectable impactontheenergyspectrum.UsingfakeitinXSPECgivesusseverallimitationsaswecannot usetimedependenttemperatureprofiles,hencewecannottakeintoaccountthedopplerboosting onthestructure.Bothofthoselimitationscanbeaddressed.Indeed,thesimulatedspectraaremade asobservationsof3000seconds,henceareaveragedoverseveralthousandsofQPOperiods.For each‘observation’wetookaconstantQPOfrequency(namelyr isconstant)sothetemperature c profile for the case of an elongated blob is conserved by rotation at the QPO frequency, which allowsustousethetime-averagespectrum2. ConcerningtheimpactoftheDopplerboosting,this effectisgreatlydiminishedinthecaseofaface-ondisksothecalculationdoneherewillbevalid foraface-ondiskandrepresentalowerlimitformoreedge-ondisksastheDopplereffectboosts theemissionofthehotspot,henceincreasingitsimpact.Hereweareinterestedtoseeifthislower limitisalreadydetectable. 4. Impactontheenergyspectrumfitting In order to see if the presence of a QPO has any impact on the energy spectrum we computed severals synthetic spectra. We used the parameters of XTE J1550-564 which is extensively observed and against which we could then test our results. For each simulation we will use the procedurefakeitonanabsorbed(phabs)disk(diskblob)pluspowerlaw(powerlaw)model. diskblobwillallowustomimicadiskwithaslowlyincreasinghotstructureaimingtoreproduce the time evolution effect of a growing QPO. The latter is introduced in the model through two parameters(inthatcaseγandδonly).WethenfittedthesyntheticspectrawithXSPECfollowing thestandardprocedurewithdiskbbandpowerlaw. UsingthemodelpresentedinEq.1γrepresentstheamplitude/strengthoftheinstabilitycausing thestructurewhileδparametrizeditsradialextent,whichcan,inturn,relatetotheFWHMofthe QPO we are modeling. The value of γ and δ in Table 1 are coherent with the QPO features one wouldexpectduringanoutburst(asseeninthegeneralcaseinVarniere&Blackman,2005,then 2 We also compared an extended blob with a fully axi-symmetrical (ring) structure with the same tem- peratureprofileandfoundtheresultstobegloballyconsistent.Indeedtheaverageprofileofabloboffinite azimuthalsizewouldbesmallerthantheaxi-symmetric(azimuthof2π)versionofthesameblob.Thiscom- fortedusthatahotterstructurepresentinthediskwillindeedhaveaneffectontheoverallenergy-spectrum. 6 Varniere,Mignon-Risse&Rodriguez:spectralQPOs for a particular outburst in Varniere & Vincent 2016b). Here we took an ‘origin point’ from the curveonFig.1.at(r ,T ) = (45,0.7)whereT isthenormalizationofT inEq.1.Wechosethis in in in 0 pointasitisjustbeforewestartseeingadepartureinthecorrelation.Thismeansthat,inallour simulations,ourdiskhasaninneredgeatr =45kmandanassociatedinneredgetemperatureof in 0.7keV.Thisallowustoseethediscrepancybetweenthe‘real’valueandthevaluegivenbythefit, measuredby∆ =(Xreal−Xfit)/Xreal).Thisdiscrepancyisameasureoftheaccuracyofthefitand X itsreliability.InallthesetsinTable1(r ,T )=(45,0.7)andonlythecouple(γ,δ)ismodified. in in γ δ Tfit Rfit in in 0 0.05 0.71 41.4 0.1 0.05 0.73 40.1 0.1 0.1 0.69 48.2 0.2 0.1 0.70 44.2 0.3 0.1 0.74 40.9 0.4 0.1 0.79 36.8 0.4 0.15 0.80 37.9 0.45 0.1 0.81 35.5 0.45 0.15 0.82 36.1 0.45 0.18 0.86 34.1 0.5 0.15 0.84 35.4 0.5 0.18 0.87 33.9 0.5 0.2 0.89 33.8 0.7 0.1 0.92 28.6 0.7 0.15 0.98 27.9 0.7 0.18 0.99 28.6 0.7 0.2 1.02 27.5 0.9 0.18 1.14 24.0 0.9 0.2 1.16 24.8 1 0.18 1.22 22.8 Table1.Hotstructureparametersandtheresultsfromthefit,asareminderforallofthosethereal inner edge of the disk is at 45km and the temperature is 0.7keV. All those have an amplitude of modulationsbetweenzeroand20%,seeVarniere&Vincent(2016)fordetailsontheamplitudes calculation. Followingtheevolutionofthefitparameters,weseethatassoonastheQPOhasanon-zeroam- plitudethereisadiscrepancybetweenthefittedvalueofthediskparametersgivenbythespectral fitandtherealvalueusedtocreatethespectrum. This discrepancy grows with the QPO amplitude (represented by the parameter γ in our model):alreadywithafeaturehavingparameters(γ,δ)=(0.4,0.15),whichtranslateina5%flux modulation we get ∆ = 14% and ∆ = 12%. It is interesting to note that the fitted value for Tin Rin r are almost always smaller than the actual value of the disk we input while the T are almost in in alwaysoverestimated.Indeedinourcaseofa‘real’inneredgeat45kmwegetsomefitresultsup toabout23kmforourlargestfeatureswhichisequivalenttoanrmsof20%.Whilethediscrepancy is always higher for the temperature, the impact on the determination of the inner radius of the 7 Varniere,Mignon-Risse&Rodriguez:spectralQPOs diskisactuallymoreimportant.Indeed,theminimalpositionoftheinneredgeofthediskisoften usedtoputsomelimitonthespinoftheblackhole.Inthatcaseanunderdeterminationbyalmost factoroftwobetweenthe‘real’valueandthefittedonewouldleadtoamuchhigherinferredspin. In the case here, where the inner edge of the disk is at 45km, which is approximatively 3r for a g 10M ,thefitgivesbackaalueof23kmrepresentingabout1.6r ,whichwouldinturnindicatean (cid:12) g almost maximal spin. As this happens when we have strong QPO, it is better to continue using onlythevalueoftheradiusfromstateswithoutQPOstoputconstraintsonthecentralobjectspin. Furthermore, when we plot the results of the fit from the synthetic spectra together with data pointsfromXTEJ1550-564wesee,inFig.2.,thatthegreystarrepresentingoursimulatedspectra occupy the same space as the HFQPO/type B LFQPO data points from XTE J1550-564 hence strengtheningthelinkbetweenQPO,hotspot,andfit-difficulties. Tin (keV) vs Rin (km)(HFQPO in red) 1.2 LFQPOType (black C, blue B, red A) blob model Fit of Tin (keV) vs Rin (km) 1.0 ) V 0.8 e k ( n Ti 0.6 0.4 20 50 100 Rin (km) Fig.2. Correlation between the inner edge position and inner edge temperature as given by the spectralfitsfortheoutburstof98-99and2000ofXTEJ1550-564.Reddotsrepresentobservations with HFQPOs detected while there is none in the black dots. The crosses represent the type of LFQPO,blackforthecommontypeC,bluefortypeBandredfortypeA.Thegreystarsarethe resultofthesyntheticspectrafit.TimingdatafromRemillardetal.(2002);Rodriguezetal(2002). We see that, even with the RXTE resolution and range, neglecting the QPO in the spectral analysiscanleadtolargediscrepancies.Asasidenote,suchdeparturefromcorrelationcouldbe usedtodetect,purelyfromspectralanalysis,thepossiblepresenceofHFQPOsbutwouldnotallow topredictitsparametersaswecannotdisentangletheimpactofδandγonthespectrum. 5. Conclusion Using a simple model to mimic the emission from a disk with a non-monotonic temperature, as hasbeentheorizedtobethecaseinthepresenceofQPOs,wehavecreatedsyntheticspectraofa system exhibiting an increasingly strong QPOs. This allowed us to study in a clean environment 8 Varniere,Mignon-Risse&Rodriguez:spectralQPOs theimpactofathestructureattheoriginoftheQPOontheenergyspectrumanddetermineifwe canneglecttheminthespectralfit. First, our simulated observations are coherent with the departure from correlation seen in the T -r diagramofXTEJ1550-564inpresenceofHFQPOandLFQPOsBorA.Inthecaseofvery in in smallamplitudeQPOsthereisanegligibleimpactontheenergyspectrumanditcanbeignoredin thespectralfit. Nevertheless,inpresenceofamediumstrength/impactQPO(morethan5%rmsamplitude)we cannotneglectthepresenceofthehotstructureasitleadstosignificantdiscrepancybetweenthe fittedvalueandthephysicalparameters.Especiallythoseerrorstendtoalmostexclusivelygivea smaller inner radius and higher inner temperature. Therefore there is a need to improve the disk fitting by taking into account the structures at the origin of the QPOs if we want to constrain the diskparametersintheirpresence. Acknowledgements. Theauthorthankstheanonymousrefereethathelpedclarifythepapertothisfinalform.Weacknowl- edgethefinancialsupportoftheUnivEarthSLabexprogramatSorbonneParisCite(ANR-10-LABX-0023andANR-11- IDEX-0005-02). References BursaM.,AbramowiczM.A.,KarasV.,KluzniakW.,2004,TheApJL,617,L45 Feroci,M.,etal.,2014,ProceedingsoftheSPIE,9144,2 Ingram,A.,Done,C.,Fragile,P.C.,MNRAS,397,L101 Karas,V.,Vokrouhlicky,D.,Polnarev,A.G.,1992,MNRAS,259,569 Pechacek,T.,Goosmann,R.W.,Karas,V.,Czerny,B.,Dovciak,M.,2013,A&A,556,A77 Miller,J.M.,Wijnands,R.,Homan,J.etal.,2001,ApJ,563,928. Miller,J.M.,Marshall,H.L.,Wijnands,R.etal.,2003,ApJ,38,7. Muno,M.P.;Morgan,E.H.;Remillard,R.A.,1999,ApJ,527,321. SchnittmanJ.D.,BertschingerE.,2004,TheApJ,606,1098 SchnittmanJ.D.,RezzollaL.,2006,TheApJL,637,L113. Schnittman,J.D.,Homan,J.,Miller,J.M.,2004,ApJ,642,420 Remillard,R.A.;Sobczak,G.J.;Muno,M.P.;McClintock,J.E.,2002,ApJ,564,962. Rodriguez,J.;Varniere,P.;Tagger,M.;Durouchoux,Ph.,2002,A&A,387,487. Rodriguez,J.andCorbel,S.andTomsick,J.A.,2003,ApJ,595,1032. Rodriguez,J.andCorbel,S.andKalemci,E.andTomsick,J.A.andTagger,M.,2004,ApJ,612,1018. Shimura,T.andTakahara,F.,1995,ApJ,445,780. Sobczak,G.J.,McClintock,J.E.,Remillard,R.A.etal.,200,ApJ,544,993. Tagger,M.&Pellat,R.1999,A&A,349,1003 Tagger,M.&Varniere,P.,2006,ApJ,652,1457. Varniere,P.;Rodriguez,J.;Tagger,M.,2002,A&A,387,497. Varniere,P.&Blackman,E.G.2005,NewA,11,43. Varniere,P.&Vincent,F.H.,2016,submittedA&A. Varniere,P.&Vincent,F.H.,2016b,tobesubmittedApJ. Varniere,P.&Vincent,F.H.,2015,proceedingdelaSF2A. Vignarca,F.;Migliari,S.;Belloni,T.;Psaltis,D.;vanderKlis,M.,2003,A&A,397,729. Vincent,F.H.;Mazur,G.P.;Straub,O.etal.,2014,A&A,563,A109. 9