Table Of Content

Astronomy&Astrophysicsmanuscriptno.serx1_ref_mod_2 (cid:13)cESO2017 January5,2017 A re-analysis of the NuSTAR and XMM-Newton broad-band spectrum of Ser X-1 M.Matranga1,T.DiSalvo1,R.Iaria1,A.F.Gambino1,L.Burderi2,A.Riggio2,andA.Sanna2 1 UniversitádegliStudidiPalermo,DipartimentodiFisicaeChimica,viaArchirafi36,90123Palermo,Italy e-mail:[email protected] 2 UniversitádegliStudidiCagliari,DipartimentodiFisica,SPMonserrato-SestuKM0.7,09042Monserrato,Italy 7 January5,2017 1 0 2 ABSTRACT n Context.HighresolutionX-rayspectraofneutronstarLowMassX-rayBinaries(LMXBs)intheenergyrange6.4-6.97keV,areoften a characterizedbythepresenceofKαtransitionfeaturesofironatdifferentionizationstages.Sincetheselinesarethoughttooriginate J byreflectionoftheprimaryComptonizationspectrumovertheaccretiondisk,thestudyofthesefeaturesallowsustoinvestigatethe 4 structureoftheaccretionflowclosetothecentralsource.Thus,thestudyofthesefeaturesgivesusimportantphysicalinformationon thesystemparametersandgeometry.SerX-1isawellstudiedLMXBwhichclearlyshowsabroadironline.Severalattemptstofit ] thisfeatureasasmearedreflectionfeaturehavebeenperformedonXMM-Newton,Suzaku,NuSTAR,and,morerecently,onChandra E data,findingdifferentresultsfortheinnerradiusofthediskandotherreflectionorsmearingparameters.Forinstance,Milleretal. H (2013) have presented broad-band, high quality NuSTAR data of Ser X-1. Using relativistically smeared self-consistent reflection . models, theyfind avalue of R close to1.0 R (corresponding to6R , where R istheGravitational radius, defined as usual h in ISCO g g R =GM/c2),andalowinclinationangleoflessthan∼10◦. p g Aims.Theaimofthispaperistoprobetowhatextentthechoiceofreflectionandcontinuummodels(anduncertaintiestherein)can - o affecttheconclusionsaboutthediskparametersinferredfromthereflectioncomponent.Tothisaimwere-analyzealltheavailable r publicNuSTARandXMM-NewtonwhichhavethebestsensitivityattheironlineenergyobservationsofSerX-1.SerX-1isawell t studiedsource,itsspectrumhasbeenobservedbyseveralinstruments,andisthereforeoneofthebestsourcesforthisstudy. s a Methods.Weuse slightlydifferent continuum and reflection models withrespect tothose adopted in literaturefor thissource. In [ particularwefittheironlineandotherreflectionfeatureswithself-consistentreflectionmodelsasreflionx(withapower-lawillumi- natingcontinuummodifiedwithahighenergycutofftomimictheshapeoftheincidentComptonizationspectrum)andrfxconv.With 1 thesemodelswefitNuSTARandXMM-Newtonspectrayieldingconsistentspectralresults. v Results.OurresultsareinlinewiththosealreadyfoundbyMilleretal.(2013)butlessextreme.Inparticular,wefindtheinnerdisk 9 radiusat∼13R andaninclinationanglewithrespecttothelineofsightof∼27◦.Weconcludethat,whilethechoiceofthereflection g 6 model haslittleimpact onthediskparameters,assoonasaself-consistent model isused,thechoiceof thecontinuum model can 0 beimportantintheprecisedeterminationofthediskparametersfromthereflectioncomponent.Hencebroad-bandX-rayspectraare 1 highlypreferabletoconstrainthecontinuumanddiskparameters. 0 . Keywords. line:formation,line:identification,stars:individual:SerpensX-1,stars:neutron,X-rays:binaries,X-rays:general 1 0 7 11. Introduction photons,comingfrom the neutronstar surface and/orthe inner : accretiondisk,byhotelectronspresentinacoronapossiblylo- v cated in the inner part of the system, surroundingthe compact iX-ray spectra emitted by Low Mass X-Ray Binaries (LMXBs) X object(D’Aìetal.2010). of the atoll class (Hasinger&vanderKlis 1989) are usually rcharacterized by two states of emission: the soft and the hard In addition to the continuum, broad emission lines in the a state.Duringsoftstatesthespectrumcanbewelldescribedbya range6.4-6.97keVareoftenobservedinthespectraofLMXBs soft thermal component, usually a blackbody or a disk multi- (seee.g.Cackettetal.2008;Pandeletal.2008;D’Aìetal.2009, color blackbody, possibly originated from the accretion disk, 2010; Iariaetal. 2009; DiSalvoetal. 2005, 2009; Egronetal. and a harder component, usually a saturated Comptonization 2013;DiSalvoetal.2015).TheselinesareidentifiedasKαtran- spectrum. In some cases, a hard power-law tail has been de- sitions of iron at different ionization states and are thought to tected in the spectra of these sources during soft states both originate from reflection of the primary Comptonization spec- in Z sources (DiSalvoetal. 2000), and in atoll sources (e.g., trum overthe accretion disk. These featuresare powerfultools Pirainoetal. 2007), usually interpreted as Comptonization off toinvestigatethestructureoftheaccretionflowclosetothecen- a non-thermalpopulation of electrons. On the other hand, dur- tralsource.In particular,importantinformationcanbe inferred ing hard states the hard componentof the spectrum can be de- fromthelinewidthandprofile,sincethedetailedprofileshapeis scribed by a power law with high energy cutoff, interpreted as determinedbytheionizationstate,geometryandvelocityfieldof unsaturatedComptonization,andaweakersoftblackbodycom- the emitting plasma (see e.g. Fabianetal. 1989). Indeed,when ponent(e.g.,DiSalvoetal.2015).Thehardcomponentisgen- the primary Comptonization spectrum illuminates a colder ac- erally explainedin terms of inverse Comptonscattering of soft cretion disk, other low-energy discrete features (such as emis- Articlenumber,page1of13 A&Aproofs:manuscriptno.serx1_ref_mod_2 sion lines and absorption edges) are expected to be created by elsusedonthesameobservation.Forsakeofclaritytheseresults photoionizationand successive recombinationof abundantele- aresummarizedinTable1. mentsindifferentionizationsstatesaswellasacontinuumemis- Milleretal.(2013)analyzedtwoNuSTARobservationscar- sion caused by direct Compton scattering of the primary spec- ried out on July 2013. They fitted the continuum emission us- trumofftheaccretiondisk.Allthesefeaturestogetherformthe ing a modelconsisting of a blackbody,a disk blackbodyand a so-calledreflectionspectrum,andthewholereflectionspectrum power-law.Withrespecttothiscontinuummodel,evidentresid- is smeared by the velocity-field of the matter in the accretion uals were present around 6.40-6.97 keV, suggesting the pres- disk. enceofaFeline.Thereforetheyaddedakerrdiskcomponent SerX-1isapersistentaccretingLMXBclassifiedasanatoll to the continuumto fit the emission line, taking into account a source,that showstype I X-ray bursts. The source was discov- possiblenon-nullspinparameterfortheneutronstar.Theyalso ered in 1965 by Friedmanetal. (1967). Lietal. (1976) firstly tried to fit the reflection spectrum (i.e. the iron line and other discoveredtype-I X-ray bursts fromthis source that was there- expected reflection features) with the self-consistent reflection foreidentifiedasanaccretingneutronstar.Besidestype-Ibursts modelreflionx,amodifiedversionofreflionxcalculatedfora with typical duration of few seconds (Balucinska&Czerny blackbodyilluminatingspectrum,convolvedwiththekerrconv 1985), a super-burst of the duration of about 2 hours has also component.Theadditionofthereflectioncomponentgaveasig- beenreported(Cornelisseetal.2002).RecentlyCornelisseetal. nificantimprovementofthefit.Inmostcasesthebestfitgavelow (2013), analyzing spectra collected by GTC, detected a two- inclinationangles(lessthan∼10◦),inagreementwithrecentop- hoursperiodicity.They tentatively identified this periodicity as ticalobservations(Cornelisseetal.2013),innerdiskradiicom- theorbitalperiodofthebinaryandhenceproposedthatthesec- patiblewiththeInnermostStableCircularOrbit(ISCO),corre- ondarystarmightbeamainsequenceK-dwarf. spondingtoabout6Rgforsmallvaluesofthespinparameter,a Church&Balucin´ska-Church(2001)haveperformeda sur- ionizationparameterlogξ ∼2.3−2.6,andaslightpreferencefor veyofLMXBscarriedoutwiththeASCAsatellite. Thebest-fit anenhancedironabundance.Thefitresultedquiteinsensitiveto modelusedbytheseauthorstofitthespectrumofSerX-1was the value of the adimensionalspin parameter,a, of the neutron a blackbodyplusa cutoff power-lawwith a Gaussian iron line. star. Oosterbroeketal.(2001)haveanalyzedtwosimultaneousobser- More recently, Chiangetal. (2016) analysed a recent 300 vationsofthissourcecollectedwithBeppoSAXandRXTE.The ksChandra/HETGSobservationofthesourceperformedinthe authors fitted the broad-band (0.1-200 keV) BeppoSAX spec- "continuousclocking"modeandthusfreeofphotonpile-upef- trum with a model consisting of a disk blackbody,a reflection fects. They fitted the continuum with a combination of multi- componentdescribedbytheXSPECmodelpexrav,andaGaus- color disk blackbody, blackbody and power-law. The iron line sian line.However,inthatcase theimprovementinχ2 with re- was found significantly broader than the instrumental energy spect to a model consisting of a blackbody, a Comptonization resolution and fitting this feature with a diskline instead of a spectrum modeled by compST, and a Gaussian was not signif- broadGaussiangavea significantimprovementofthe fit. They icant, and therefore it was not possible to draw any definitive also tried self-consistentreflection models,namely the reflionx conclusionaboutthepresenceofareflectioncontinuum. modelwith a power-law continuumas illuminating source and Bhattacharyya&Strohmayer(2007)carriedouttheanalysis xillver(seee.g.Garcíaetal.2013),todescribetheironlineand of three XMM-Newton observations of this system. They man- otherreflectionfeatures,yieldingconsistentresults.Inparticular, agedtofittheEPIC/pnspectrumwithamodelconsistingofdisk thisanalysisgaveainnerradiusof∼7−8Rgandaninclination blackbody,aComptonizationcontinuummodeledwithcompTT angleofabout30deg. andadiskline,i.e.aGaussianlinedistortedandsmearedbythe As described above,differentcontinuummodels were used Keplerianvelocityfieldintheaccretiondisk(Fabianetal.1989). tofitthespectrumofSerX-1observedwithvariousinstruments Theyfoundstrongevidencethatthe Fe line hasan asymmetric at different times. In Table 1 we summarize the results of the profileandthereforethatthelineoriginatesfromreflectioninthe spectralanalysis of this source obtainedfrom previousstudies, innerrimoftheaccretiondisk.Fitted withaLaorprofile(Laor andinparticulartheresultsobtainedfortheironlineandthere- 1991), the line shape gave an inner disk radius of 4− 5R or flectionmodel.Quitedifferentvalueshavebeenreportedforthe g 16R (depending from the observation) and an inclination an- inclinationangle(fromlessthan10degtoabout40deg),forthe g gletothebinarysystemof40−50◦.Cackettetal.(2008),from innerdiskradius(from4tomorethan100R )andfortheiron g data collected by SUZAKU, performed a study of the iron line line centroidenergyand/orthe ionization parameter logξ indi- profiles in a sample of three LMXBs including Ser X-1. From catingthatthedisk is formedbyneutralorveryhighlyionized theanalysisofXISandPINspectra,theyfoundagoodfitofthe plasma. broad-bandcontinuumusingablackbody,adiskblackbodyand Inthispaperwere-analyzedalltheavailablepublicNuSTAR a power-law. Two years later Cackettetal. (2010) re-analyzed observationsofSerX-1,fittingtheironlineandotherreflection XMM-NewtonandSUZAKUdataofasampleof10LMXBsthat featureswithbothphenomenologicalandself-consistentreflec- includes Ser X-1, focusing on the iron line - reflection emis- tion models. These data were already analysed by Milleretal. sion. In particular, for Ser X-1, they analyzed 4 spectra: three (2013) using a different choice of the continuum and reflec- Epic-PNspectra obtainedwith XMM-Newtonandoneobtained tion models. We compare these results with those obtained with the XIS and the PIN instruments on board of SUZAKU. from three XMM-Newton observations (already analyzed by Initially, they fitted the spectra of the continuum emission us- Bhattacharyya&Strohmayer 2007) fitted with the same mod- ing a phenomenological model, consisting of a blackbody, a els. We choose to re-analyse NuSTAR and XMM-Newton spec- disk-blackbodyandapower-law.Then,theystartedthestudyof tra because these instruments provide the largest effective area theFe lineaddingfirsta disklinecomponentandaftera reflec- available to date, coupled with a moderately good energy res- tioncomponentconvolvedwithrdblur(thattakesintoaccount olution, at the iron line energy, and a good broad-band cover- smearing effects due to the motion of the emitting plasma in a age.Moreover,thesourceshowedsimilarfluxesduringtheNuS- Kepleriandisk).Theyobtaineddifferentresultsforthesmearing TARandXMM-Newtonobservations.NotealsothatNuSTARis parametersbothfordifferentobservationsandfordifferentmod- notaffectedbypile-upproblemsinthewholeenergyrange.The Articlenumber,page2of13 M.Matrangaetal.:Are-analysisoftheNuSTARandXMM-Newtonbroad-bandspectrumofSerX-1 spectralresultsobtainedforNuSTARandXMM-Newtonarevery fore,we extractedthe source spectra from a rectangularregion similartoeachotherandthesmearingparametersofthereflec- (RAW X≥26) and (RAW X≤46) including all the pixels in the tioncomponentarelessextremethanthosefoundbyMilleretal. ydirectionbutexcludingthebrightestcolumnsatRAWX=35 (2013),andingoodagreementwiththeresultsobtainedfromthe andRAWX=36.Thisreducedsignificantlythepileup(pileup Chandraobservation(Chiangetal. 2016). In particularwe find fractionbelowafewpercentintheconsideredenergyrange). an inner disk radius in the range 10−15R and an inclination We selected only eventswith PATTERN ≤ 4 and FLAG=0 g anglewithrespecttothelineofsightof25−30◦. that are the standard values to remove spuriousevents. We ex- tractedthebackgroundspectrafroma similarregionto theone used to extract the source photons but in a region away from 2. ObservationsandDataReduction the source included between (RAWX≥1) and (RAWX≤6 ). Fi- In this paper we analyze data collected by the NuSTAR satel- nally,foreachobservation,usingthetask’rgscombine’wehave lite. Ser X-1 has been observed twice with NuSTAR, obsID: obtained the added source spectrum RGS1+RGS2, the relative 30001013002 (12-JUL-2013) and obsID: 30001013004 (13- addedbackgroundspectrumalongwiththerelativeresponsema- JUL-2013). The exposure time of each observation is about trices.WehavefittedRGSspectruminthe0.35-1.8keVenergy 40ksec.ThedatawereextractedusingNuSTARDAS(NuSTAR range,whereastheEpic-PNinthe2.4-10keVenergyrange. DataAnalysisSoftware)v1.3.0.Sourcedatahavebeenextracted SpectralanalysishasbeenperformedusingXSPECv.12.8.1 fromacircularregionwith120"radiuswhereasthebackground (Arnaud 1996). For each fit we have used the phabs model in has been extracted from a circular region with 90" radius in a XSPEC to describe the neutral photoelectric absorption due to region far from the source. First, we run the "nupipeline"with the interstellar medium with photoelectric cross sections from default values of the parameters as we aim to get "STAGE 2" Verneretal. (1996) and element abundances from Wilmsetal. eventsclean.Thenspectraforbothdetectors,FPMAandFPMB, (2000). For the NuSTAR spectrum, which lacks of low- energy were extracted using the "nuproducts" command. Correspond- coverageup to 3 keV, we fixed the value of the equivalenthy- ing responsefiles were also createdas outputof nuproducts.A drogen column, NH, to the same value adopted by Milleretal. comparisonof the FPMA and FPMB spectra, indicated a good (2013),namelyNH =4×1021cm−2(Dickey&Lockman1990), agreementbetweenthem.Tocheckthisagreement,wehavefit- whilefortheXMM-Newtonspectrumweleftthisparameterfree tedthetwoseparatespectrawithallparameterstiedtoeachother tovaryinthefit,findingaslightlyhighervalue(seeTab.2and butwith a constantmultiplicationfactorleftfreeto vary.Since 3).Asafurthercheck,wehavefittedtheNuSTARspectrumfix- the value of this parameter is 1.00319±0.00145, our assump- ingNH tothesamevaluefoundfortheXMMspectrum,butthe tionisbasicallycorrect.Followingthesameapproachdescribed fitparametersdidnotchangesignificantly. inMilleretal.(2013),wehavethereforecreatedasingleadded spectrumusingthe"addascaspec"command.Asingleresponse 3. SpectralAnalysis filehasbeenthuscreatedusing"addrmf",weightingthetwosin- gle response matrices by the corresponding exposure time. In 3.1.NuSTARspectralanalysis this way, we obtain a summed spectrum for the two NuSTAR observationsandthetwoNuSTARmodules.Wefittedthisspec- The NuSTAR observations caught the source in a high- truminthe3-40keVenergyrange,wheretheemissionfromthe luminosity (∼ 1038 erg/s, Milleretal. (2013)) state, therefore sourcedominatesoverthebackground. most probably in a soft state. As seen in other similar atoll We have also used non-simultaneous data collected with sources,thespectrumofSerX-1ischaracterizedbyasoftcom- XMM-Newton satellite on March 2004. The considered obsID ponent (i.e. blackbody), interpreted as thermal emission from are 0084020401, 0084020501 and 0084020601. All observa- the accretion disk, a hard component (i.e. a Comptonization tions are in Timing Mode and each of them has a duration of spectrum), interpreted as saturated Comptonization from a hot ∼22ksec.Weextractedsourcespectra,backgroundspectraand corona,andoftenbythepresenceofabroadironemissionline response matrices using the SAS (Science Analysis Software) at 6.4 − 6.97 keV depending on the iron ionization state. We v.14 setting the parameters of the tools accordingly. We pro- usedtheComptonizationmodelnthComp(Z˙yckietal.1999)in ducedacalibratedphotoneventfileusingreprocessingtools"ep- XSPEC,withablackbodyinputseedphotonspectrum,tofitthe proc"and"rgsproc"forPNandRGSdatarespectively.Wealso hard component. We used a simple blackbody to describe the extracted the MOS data; these were operated in uncompressed soft component. Substituting the blackbody with a multicolor timing mode. However, the count rate registered by the MOS disk blackbody, diskbb in XSPEC, gives a similar quality fit wasintherange290−340c/s,whichisabovethethresholdfor andthebest-fitparametersdonotchangesignificantly. avoidingdeterioratedresponseduetophotonpile-up.TheMOS To fit the iron line we first tried simple models such as a spectraindeedshowclearsignsofpile-upandwepreferrednot Gaussianprofileoradiskline(Fabianetal.1989).Thebest-fit to include them in our analysis, since these detectorscover the parameters, obtained using alternatively a Gaussian or diskline sameenergyrangeofthePN. profile,areingoodagreementwitheachother(seeTab.2).Using Beforeextractingthespectra,wefilteredoutcontaminations adisklineinsteadofaGaussianprofilewegetanimprovement due to backgroundsolar flares detected in the 10-12 keV Epic of the fit corresponding to ∆χ2 = 54 for the addition of two PN light-curve.Inparticularwe havecutoutabout600sec for parameters.Spectra,alongwiththebest-fitmodelandresiduals obsID0084020401,about800secforobsID0084020501andfi- are shown in Fig.1. In both cases, the fit results are poor (the nallyabout1600secforobsID0084020601.Inordertoremove relative null hypothesis probability is 2.8 × 10−8; the reduced theflares,weappliedtimefiltersbycreatingaGTIfilewiththe χ2 are still relatively large, and evident residuals are present, task "tabgtigen". In order to check for the presence of pile-up especiallyabove10keV,seeFig.1). we have run the task "epatplot" and we have found significant In order to fit the residuals at high energy, we added a contaminationineachobservation.Thecount-rateregisteredin powerlaw component(a hard tail) to all the models described the PN observations was in the range 860-1000c/s that is just above.Ahardpower-lawtailisoftenrequiredtofithigh-energy above the limit for avoiding contamination by pile-up. There- residualsofatollsourcesinthe softstate (seee.g.Pintoreetal. Articlenumber,page3of13 A&Aproofs:manuscriptno.serx1_ref_mod_2 2015, 2016; Iariaetal. 2001, 2002), and this component may flectiontables.Thisisaconvolutionmodelthatcanbeusedwith also be present in the spectrum of Ser X-1 (see Milleretal. anyinputcontinuumandhasthereforetheadvantagetotakeas 2013). Unless it is specified otherwise, for every fit, we froze illuminating spectrum the given Comptonization continuum. It the power-law photon index to the value found by Milleretal. dependson5parameters:therelativereflectionfraction(rel-refl (2013)forSerX-1,thatis3.2.Thenewmodelsarenowcalled defined as Ω/2π, namely as the solid angle subtended by the gauss-planddiskline-pl,respectively.Thenewbestfitparame- reflecting disk as seen from the illuminating corona in units ters are reportedin Tab 2. While the best-fit parametersdo not of 2π), the cosine of the inclination angle, the iron abundance changesignificantlywiththeadditionofthiscomponent,weget relativetotheSolarvalue,theionizationparameterLogξofthe animprovementofthefitcorrespondingtoareductionoftheχ2 accretiondisksurface,andtheredshiftofthesource. by ∆χ2 = 123(forthe modelwith a Gaussian line profile)and Due to its high velocities, the radiationre-emitted fromthe ∆χ2 =113(forthemodelwithadisklineprofile)fortheaddition plasmalocatedintheinneraccretiondiskundergoestoDoppler of one parameter,respectively.The probabilitiesof chance im- andrelativisticeffects(whichsmearsthewholereflectionspec- provementofthefitare8.5×10−24and8.6×10−23,respectively. trum). In order to take these effects into accountwe have con- Someresidualsarestillpresentbetween10and20keVprobably volved the reflection models with the rdblur component (the causedby the presenceof an unmodeledComptonhump.Note kernel of the diskline model), which depends on the values of thatthe softblackbodycomponentremainssignificantevenaf- the inner and outer disk radii, in units of the Gravitational ra- ter the addition of the power-law component. If we eliminate dius (R = GM/c2), the inclination angle of the disk (that was g this componentfrom the fitting model we get a worse fit, cor- kept tied to the same value used for the reflection model), and respondingto a decreaseby∆χ2 = 245forthe additionoftwo the emissivity index, Betor, that is the index of the power-law parameterswhenthesoftcomponentisincludedinthefitanda dependence of the emissivity of the illuminated disk (which probabilityofchanceimprovementofthefitof∼3×10−44. scales as rBetor). Finally, we have also considered the possibility that neutron star has a spin. In this case, the reflection 3.2.Reflectionmodels componenthas beenconvolvedwith the Kerrconvcomponent (Brenneman&Reynolds 2006) that through its adimensional We havealso triedto fittheNuSTARspectrumofSerX-1with spinparameter’a’allowedustoimplementagridofmodelsex- more sophisticated reflection models, performing a grid of fit ploringdifferentvaluesof’a’(seeAppendixA).Forthismodel withself-consistentmodelssuchasreflionxorrfxconv.Re- thereisalsothepossibilitytofittheemissivityindexoftheinner flionxandrfxconvmodelsbothincludethereflectioncontinuum, andouterpartofthediskindependently,althoughinourfitswe the so called Compton hump caused by direct Compton scat- usedthesameemissivityindexforthewholedisk.Forallthefits teringofthereflectedspectrum,anddiscretefeatures(emission wehavefixedthevaluesofR to2400R ,theironabundance out g linesandabsorptionedges)formanyspeciesofatomsatdiffer- to solarvalue,Fe/solar = 1, andthe redshiftofthe sourceto 0. ent ionization stages (Ross&Fabian 2005; Kolehmainenetal. ThebestfitparametersarereportedinTab2–A.2. 2011). We started to fit the data adding a reflection component, Thereflionxmodeldependson5parameters,thatarethe reflionxorrfxconv,convolvedwiththeblurringcomponent abundance of iron relative to the solar value, the photon index rdblur, to the continuum model given by the blackbody and oftheilluminatingpower-lawspectrum(Γ,rangingbetween1.0 thenthcompcomponents(modelsarecalledrdb-reflioandrdb- to 3.0), the normalizationof reflected spectrum, the redshift of the source, and the ionization parameter ξ = L /(n r2) where rfxconv,respectively).Fitresultsforbothmodelsareacceptable, X e withχ2 closeto1.09.Thereareafewdifferencesbetweenthe LX is the X-rayluminosity of the illuminatingsource, ne is the best-fitrepdarametersoftherdb-refliomodelwithrespecttothose electron density in the illuminated region and r is the distance of the rdb-rfxconv model. In particular the rdb-rfxconv model oftheilluminatingsourcetothereflectingmedium.Whenusing gives a lower value of R , while the rdb-reflio model gives a reflionx,whichusesapower-lawasilluminatingspectrum,in in higherionizationparameter(althoughwithalargeuncertainty). ordertotakeintoaccountthehigh-energyrolloveroftheComp- Spectra,alongwiththebest-fitmodelandresidualsarereported tonizationspectrum,wehavemultiplieditbyahigh-energycut- in Fig.1. The residuals that are very similar for the two mod- off, highecut, with the folding energy E set to 2.7 times fold els,apartforthe8-10keVenergyrangewhererdb-refliomodel theelectronstemperaturekT andthecutoffenergyE tied e cutoff showsflatterresidualsthanrdb-rfxconvmodel(seeFig.1). to 0.1 keV. In this way we introduce a cutoff in the reflection continuum,whichotherwiseresemblesapower-law.Thecut-off As before, we also tried to add a power-law component to energyfixedat2.7timestheelectrontemperatureoftheComp- the modelsobtainedbythe convolutionof the blurringcompo- tonizationspectrum(assumedtobesimilartoablackbodyspec- nent(rdblur)with the two differentreflection components(rfx- trum), is appropriate for a saturated Comptonization (see e.g. convorreflionx).Thetwonewmodelsarecalledrdb-rfxconv-pl Egronetal.2013).TofittheComptonizationcontinuumweused and rdb-reflion-pl, respectively. In both cases we get a signif- thenthCompmodel.Moreoverwefixedthephotonindexofthe icant improvement of the fit, with ∆χ2 = 90 for the addition illuminatingspectrum,Γ,tothatofthenthCompcomponent.We of two parameters and ∆χ2 = 66 for the addition of one pa- stress that in our analysis we use a different reflionx reflec- rameter,respectively.Inthesecases,anF-testyieldsaprobabil- tion model with respect to that used by Milleretal. (2013). In ityofchanceimprovementof3.1×10−15 forrdb-reflion-pland factweusedamodelthatassumesaninputpower-lawspectrum 6.1×10−19forrdb-rfxconv-plmodel,respectively.Spectra,along asthesourceoftheirradiatingfluxmodified,inordertomimic withbest-fitmodelandresidualsarereportedinFig.2,whereas the nthcomp continuum, by introducing the model component values of the best-fit parameters are listed in Tab. 3. Residuals highecut.Milleretal. (2013) instead used a modifiedversion arenowflat(seeplotsreportedinupperpanelsofFig.2).Note ofreflionxcalculatedforablackbodyinputspectrum,sincethat alsothatinthiswaywegetmorereasonablevaluesofthebest-fit componentdominatestheirphenomenologicalcontinuum. parameters,especiallyfortheionizationparameter,logξ,which rfxconv is an updated version of the code in is around 2.7 for both models, in agreement with the centroid Done&Gierlin´ski (2006), using Ross&Fabian (2005) re- energyof the iron line at about6.5 keV, and well below 3.7 (a Articlenumber,page4of13 M.Matrangaetal.:Are-analysisoftheNuSTARandXMM-Newtonbroad-bandspectrumofSerX-1 ionizationparameterlogξ ∼ 3.7wouldimplythatthematterof component ’rdblur’ (model called rdb-rfxconv-pl-xmm, results theaccretiondiskwouldbefullyionized). arereportedinTable3). Insummary,thebestfitoftheNuSTARspectrumofSerX-1 We have performed the fit of the spectrum obtained from isobtainedfittingthecontinuumwitha softblackbodycompo- thesethreeobservationssimultaneously,tyingparametersofthe nent,aComptonizationspectrum,andahardpower-lawtailand RGSwith theallparametersofthePN fromthesame observa- fitting the reflection featureswith the rfxconvmodelsmeared tion.ThespectraofthethreeXMMobservationsareverysimilar bytherdblurcomponent,sincethefittingresultsarequitein- witheachother,exceptforthesoftblackbodytemperaturethat sensitivetothevalueofthespinparametera(seeAppendixA). was left free to vary in differentdatasets. Values of the best-fit Thisfit,correspondingtoaχ2(dof)=912.5(911),givesablack- parameters of the model diskline-pl-xmm result to be in good bodytemperatureof≃ 0.54keV,atemperatureoftheseedpho- agreementwithwhatwehavefoundfromthefitoftheNuSTAR tonsfortheComptonizationof≃0.93keV,anelectrontempera- spectrawiththesamemodel. tureoftheComptonizingcoronaof≃2.70keVandaphotonin- We have also performed the fit with a model including the dexoftheprimaryComptonizedcomponentof≃2.17,whereas reflection component rfxconv, called rdb-rfxconv-pl-xmm.As the photon index of the hard power-law tail is steeper, around before,inordertotakeintoaccountstructuresvisibleintheRGS 3.2.Thereflectioncomponentgivesareflectionamplitude(that spectra,wehaveaddedthreegaussianstothemodel.Asbefore is the solid angle subtendedby the accretion disk as seen from wehavetiedparametersoftheRGStothecorrespondingparam- theComptonizingcorona)of≃ 0.24andaionizationparameter etersofthePNfromthesameobservationexceptfortheparam- oflogξ ≃ 2.7.Thesmearingofthe reflectioncomponentgives eterkT thatwasleftfreetovaryamongthethreeobservations. bb an inner disk radius of R ranging between 10 and 16 R , and Note also that for the these fits the inclinationangle is fixed to in g inclinationangle ofthe disk with respectto the line of sightof the corresponding values we found from the NuSTAR spectra. i ≃ 27◦, and the emissivity of the disk scaling as ∝ r−2.6±0.2. ResultsarereportedinTable3,andareingoodagreementwith Note that the Compton hump is highly significant. To evaluate thoseobtainedfortheNuSTARspectrum. its statistical significance we can compare the best fit obtained with the modeldiskline-plwith the best fit given by the model 4. Discussion rdb-rfxconv-pl(themaindifferencebetweenthetwomodelsisin factthatrfxconvcontainsthereflectioncontinuumanddiskline SerX-1isawellstudiedLMXBshowingabroademissionline doesnot).Usingrfxconvinsteadofdisklinewe getadecreases at 6.4 − 6.97 keV interpreted as emission from iron at differ- of the χ2 by ∆χ2 = 87 for the addition of 1 parameter and an entionizationstatesandsmearedbyDopplerandrelativisticef- F-testprobabilityofchanceimprovementof8×10−20,whichis fectscaused bythe fast motionof matter in the inneraccretion statisticallysignificant. disk. Moderately high energy resolution spectra of this source have been obtained from XMM-Newton, Suzaku, NuSTAR, and Chandra. However, as described in Sec. 1, spectral results for 3.3.XMM-NewtonSpectralAnalysis the reflection componentare quite differentfor differentobser- We have also carried out the analysis of XMM-Newton obser- vationsorfordifferentmodelsusedto fitthecontinuumand/or vations of Ser X-1. A previous study, based only on the PN thereflectioncomponent.Whilespectraldifferencesindifferent dataanalysis,hasbeenreportedbyBhattacharyya&Strohmayer observations may be in principle justified by intrinsic spectral (2007).WeupdatedtheanalysisbyperformingthefitoftheRGS variationsofthe source,differencescaused bydifferentcontin- spectra in the 0.35–1.8 keV energy range and the PN spectra uumorreflectionmodelsshouldbeinvestigatedindetailinorder in the 2.4–10 keV energy range. Following the same approach togiveareliableestimateoftheparametersofthesystem.Forin- used for the analysis on NuSTAR data, we assumed a contin- stance,inarecentNuSTARobservationanalyzedbyMilleretal. uum model composed of a blackbody, a hard power-law and (2013),assumingamodifiedversionofreflionxcalculatedfor the nthCompcomponent.In addition to the continuumcompo- ablack-bodyinputspectrum,theauthorsreportasignificantde- nents described above, we have also detected several discrete tection of a smeared reflection componentin this source, from features present in all RGS spectra, both in absorption and in which they derive an inner radius of the disk broadly compat- emission that were supposed to be of instrumental origin by ible with the disk extending to the ISCO (corresponding to 6 Bhattacharyya&Strohmayer (2007). The energies of the most Rg in the case a = 0) and an inclination angle with respect to intensefeaturesdetectedinourspectraliebetween0.5keVand thelineofsight< 10◦.Ontheotherhand,Chiangetal.(2016), 0.75keV.Tofitthesefeatureswehavethereforeaddedthreead- analysing a recent 300 ks Chandra/HETGS observation of the ditional gaussians to our model: two absorption lines at 0.528 sourceobtainedahigh-resolutionX-rayspectrumwhichgavea keVandat0.714keV,respectively,andoneinemissionat0.541 innerradiusofR ∼7−8R andaninclinationangleof∼30◦. in g keV.Theidentificationoftheselinesisnotstraightforward.The In this paper we analyzed all the available NuSTAR 0.528keVenergyisclosetotheneutralOKαline,expectedat and XMM-Newton observations of Ser X-1. These observa- arestframeenergyof0.524keV,whilethe0.541keVemission tions have been already analyzed by Milleretal. (2013) and lineisclosetotheexpectedenergyoftheOIedgeat0.538keV. Bhattacharyya&Strohmayer(2007),respectively,whouseddif- These two lines may be therefore instrumental features caused ferentcontinuumand reflection models and reportdifferentre- byamiscalibrationoftheneutralOedgeintheRGS.Theother sultsforthereflectioncomponent.ThesameXMM-Newtonob- absorption line at 0.714 keV is close to the O VII absorption servationshavealsobeenanalyzedbyCackettetal.(2010)who edgeexpectedatarest-frameenergyof0.739keV.Giventhatthe also report different results for the reflection component, with identificationoftheselinesisuncertain,wewillnotdiscussthem higher inner disk radii (between 15 and more than 45R ) and g furtherinthepaper.Tothiscontinuumwefirstaddedadiskline quite low inclinations angles (< 10◦) when using a blurred re- (model called diskline-pl-xmm, see Table 2) to fit the iron line flectionmodel,andinclinationanglebetween10and35◦ when profile.Thenwefittedthespectrasubstitutingthedisklinewith using a diskline componentto fit the iron line profile (see Tab. the self-consistent reflection model that gave the best fit to the 1formoredetails).WehaveshownthatwecanfittheNuSTAR NuSTAR data, that is ’rfxconv’, convolved with the smearing andXMM-Newtonspectraindependentlywith the samecontin- Articlenumber,page5of13 A&Aproofs:manuscriptno.serx1_ref_mod_2 uum model and with a phenomenologicalmodel (i.e. diskline) Comptonization continuum and the disk (a value of 0.3 would oraself-consistentreflectionmodel(i.e.reflionxorrfxconv)for be compatible with a spherical geometry of a compact corona thereflectioncomponent,findinginallourfitsimilar(compati- inside an outer accretion disk). For the smearing parameters blewithintheassociateduncertainties)smearingparametersfor of the reflection component we find values of the emissivity thereflectioncomponent. index of the disk ranging from -2.8 to -2.48, an inner radius To fit these spectra we have used a continuummodelcom- of the disk from 10.6 to 16.2R , and an inclination angle of g posed by a blackbody component (bbody) and a comptoniza- the system with respect to the line of sight of 25 − 30◦. In tioncontinuum(nthcomp),whichhasbeenwidelyusedinliter- our results the inclination angle is higher than that found by ature to fit the spectra of neutron star LMXBs both in the soft Milleretal. (2013) (who report an inclination angle less than andinthehardstate(seee.g.Egronetal.2013).Withrespectto 10◦),butisverysimilartothatestimatedfromChandraspectra thecontinuummodelusedbyMilleretal.(2013)we havesub- (25 − 35◦, see Chiangetal. (2016). Moreover, the inner disk stituted one of the two blackbodycomponents,the hottestone, radius we find is not compatible with the ISCO. Assuming a withaComptonizationspectrum.Sincethiscomponentgivesthe 1.4M⊙fortheneutronstar,theinnerradiusofthediskislocated mostimportantcontributiontothesourceflux,especiallyabove at22−34kmfromtheneutronstarcenter.Notethatthisvalue 5keV,wehavesubsequentlyusedthiscomponentasthesource is compatible to the estimated radius of the emission regionof of the reflection spectrum. In all our fit the addition of a hard the soft blackbody component, which is in the range 19 − 31 power-lawcomponent,withaphotonindex∼3significantlyim- km.We interpretthiscomponentastheintrinsicemissionfrom provedthe fit. The presence of a hard power-lawcomponentis the inner disk since this is the coldest part of the system and oftenfoundinthespectraofbrightLMXBsinthesoftstate(see becausethetemperatureoftheblackbodycomponentappearsto e.g.Pirainoetal.2007;Pintoreetal.2015,2016),andhasbeen betoolowtorepresentaboundarylayer. interpretedascomptonizationofsoftphotonsoffanon-thermal populationofelectrons(seee.g.DiSalvoetal.2000). To fit the reflection component, which is dominated by a 5. Conclusions prominent iron line, we have first used a phenomenological modelconsistingofaGaussianlineoradiskline,withadiskline Themain aim of this paperis to test the robustnessofdisk pa- providing a better fit than a Gaussian profile (cf. fitting results rameters inferred from the reflection componentin the case of reportedin Table 2). All the disklineparametersobtainedfrom neutron star LMXBs; to this aim we used broad-band, mod- thefittingoftheNuSTARandXMM-Newtonspectraarecompat- erately high resolution spectra of Serpens X-1, a neutron star iblewitheachother,exceptforthelinefluxwhichappearstobe LMXBoftheatolltypewithaveryclearreflectionspectrumthat lowerduringtheXMM-Newtonobservations. hasbeenstudiedwithseveralinstruments.Inparticular,wehave In order to fit the reflection spectrum with self-consistent carried out a broad-band spectral analysis of this source using models, which take into account not only the iron line but data collected by NuSTAR and XMM-Newton satellites, which alsootherreflectionfeatures,wehaveusedbothreflionxand havethebestsensitivityattheiron-lineenergy.Thesedatahave rfxconvreflectionmodels.Inboththesemodels,emissionand been already analyzed in literature. In particular Milleretal. absorption discrete features from the most abundant elements (2013) have analyzed the NuSTAR spectra and have obtained a are included,as well as the reflected continuum.We have con- lowinclinationangleofabout8◦,aninnerdiskradiuscompati- volved the reflection spectrum with the relativistic smearing blewiththeISCO,aionizationparameterlogξbetween2.3and modelrdblur,taking into accountDopplerand relativistic ef- 2.6alongwithanironabundanceofabout3. fectscaused by the fast motionof the reflecting materialin the In thefollowingwe summarizethe resultspresentedin this inner accretion disk. We have also investigated the possibility paper: that the neutron star has a significant spin parameter. We have thereforeperformedagridoffitsusingthekerrconvsmearing – We have performed the fitting using slightly different con- model,insteadofrdblur,freezingthespinparameter’a’atdiffer- tinuum and reflection models with respect to that used by entvalues:0,0.12,0.14andlettingitfreetovaryinanadditional otherauthorstofittheX-rayspectrumofthissource.Ourthe case (see AppendixA for more details). In agreementwith the best fit of the NuSTAR spectrum of Ser X-1 is obtained fit- resultsreportedbyMilleretal. (2013)we findthatthefitisal- tingthecontinuumwithasoftblackbody,aComptonization mostinsensitivetothespinparameterbutpreferslowvaluesof spectrum,ahardpower-lawtailinadditiontothereflection thespinparameter(a<0.04). features.Tofitthereflectionfeaturespresentinthespectrum The results obtained using reflionx or rfxconv are weusedbothempiricalmodelsandself-consistentreflection somewhatdifferentin thefitsnotincludingthe hardpower-law componentsasreflionxandrfxconv,aswellastwodif- component. However, the reflection and smearing parameters ferent blurringcomponentsthat are rdblurandkerrconv. becomeverysimilarwhenweaddthiscomponenttothecontin- Fromtheanalysiscarriedoutusingkerrcovwehaveobtained uummodel(cf.resultsinTabs.3,A.1,A.2).Theadditionofthis thatourfitisinsensitivetothevalueassumedbytheadimen- componentalso significantly improvesall the fits. We consider sional spin parameter ’a’, in agreement with what is found as our best fit model the one including the hard power-law byMilleretal.(2013)intheiranalysis. component, rfxconv as reflection component smeared by the – With regardthe reflectionfeatures,we obtainconsistentre- rbdblur component (model named rdb-rfxconv-plin Tab. sults using phenomenologicalmodels (such as diskline) or 3). The fit of the XMM-Newton spectra with the same model self-consistent models to fit the NuSTAR spectrum of the gave values of the parameters that overall agree with those source. In particular, the reflection component gives a re- obtainedfittingtheNuSTARspectra.Inthiscase,wehavefound flection amplitude of Ω/2π ∼ 0.2 − 0.3 (where Ω is the valuesof the ionizationparameterlog(ξ) rangingbetween 2.58 solid angle of the disk as seen from the corona in units of and 2.71(a bithigher,around3, forthe XMM-Newtonspectra) 2π) and a ionization parameter of log(ξ) ∼ 2.6−2.7. The and reflection amplitudes between 0.2 and 0.3, indicating a smearingofthereflectioncomponentgivesaninnerdiskra- relatively low superposition between the source of the primary diusofR ∼ 10.6−16.2R ,anemissivityindexofthedisk in g Articlenumber,page6of13 M.Matrangaetal.:Are-analysisoftheNuSTARandXMM-Newtonbroad-bandspectrumofSerX-1 in the range −(2.5− 2.8), whereas the inclination angle of Cornelisse,R.,Casares,J.,Charles,P.A.,&Steeghs,D.2013,MNRAS,432, thediskwithrespecttothelineofsightresultsintherange 1361 25−29◦.Wenotethattheinnerdiskradiusderivedfromthe reflection componentresults compatible with the radius in- Cornelisse,R.,Kuulkers,E.,in’tZand,J.J.M.,Verbunt,F.,&Heise,J.2002, A&A,382,174 ferredfromthesoftblackbodycomponent,whichresultsin therange19−31km. D’Aì,A.,diSalvo,T.,Ballantyne,D.,etal.2010,A&A,516,A36 – TheanalysisofXMM-Newtonspectra,carriedoutusingthe samemodelsadoptedtofittheNuSTARspectra,gavevalues D’Aì,A.,Iaria,R.,DiSalvo,T.,Matt,G.,&Robba,N.R.2009,ApJ,693,L1 of the parameters compatible to those described above, al- thoughthetwoobservationsarenotsimultaneous.Theonly DiSalvo,T.,D’Aí,A.,Iaria,R.,etal.2009,MNRAS,398,2022 differencesarethereflectionamplitude,Ω/2π∼0.18−0.19, whichresultsslightlylower,althoughstillmarginallyconsis- DiSalvo,T.,Iaria,R.,Matranga,M.,etal.2015,MNRAS,449,2794 tentwithintheerrors,andtheionizationparameter,log(ξ)∼ 2.9−3.1,whichresultssomewhathigherwithrespecttothe DiSalvo,T.,Iaria,R.,Méndez,M.,etal.2005,ApJ,623,L121 non-simultaneousNuSTARobservations. DiSalvo,T.,Stella,L.,Robba,N.R.,etal.2000,ApJ,544,L119 In conclusion, in this paper we performed an investigation of to which extent the disk parameters inferred from reflection Dickey,J.M.&Lockman,F.J.1990,ARA&A,28,215 fittingdependonthechosenspectralmodelsforboththecontin- uumandthereflectioncomponent.Despitethefactthatseveral Done,C.&Gierlin´ski,M.2006,MNRAS,367,659 authors in previous work have used basically the same continuum model, the resulting reflection parameters, such as the in- Egron,E.,DiSalvo,T.,Motta,S.,etal.2013,A&A,550,A5 nerdiskradius,R ,andtheinclinationanglearescatteredover in a large range of values. In this paper we have re-analyzed all Fabian,A.C.,Rees,M.J.,Stella,L.,&White,N.E.1989,MNRAS,238,729 theavailablepublicNuSTARandXMM-Newtonobservationsof Ser X-1, fitting the continuum with a slightly different, physi- Friedman,H.,Byram,E.T.,&Chubb,T.A.1967,Science,156,374 callymotivatedmodelandtheironlinewithdifferentreflection models. By performinga detailed spectral analysis of NuSTAR Galloway,D.K.,Muno,M.P.,Hartman,J.M.,Psaltis,D.,&Chakrabarty,D. 2008,ApJS,179,360 and XMM-Newton data of the LMXB Ser X-1 using both phe- nomenologicaland self-consistent reflection models, and using García,J.,Dauser,T.,Reynolds,C.S.,etal.2013,ApJ,768,146 acontinuummodelsomewhatdifferentfromthatusedinlitera- tureforthissource,thebestfitparametersderivedfromthetwo Hasinger,G.&vanderKlis,M.1989,A&A,225,79 spectra are in good agreement between each other. These are alsobroadagreementwiththefindingsofMilleretal.(2013)al- Iaria,R.,D’Aí,A.,diSalvo,T.,etal.2009,A&A,505,1143 thoughwefindvaluesoftheinnerdiskandtheinclinationangle thatarelessextreme.Hence,theuseofbroad-bandspectraand Iaria,R.,DiSalvo,T.,Burderi,L.,&Robba,N.R.2001,ApJ,548,883 of self-consistentreflection models,togetherwith an investiga- tionofthecontinuummodel,arehighlydesirabletoinferreliable Iaria,R.,DiSalvo,T.,Robba,N.R.,&Burderi,L.2002,ApJ,567,503 parametersfromthereflectioncomponent. Kolehmainen,M.,Done,C.,&DíazTrigo,M.2011,MNRAS,416,311 Acknowledgements. We thank the anonymous referee for useful suggestions whichhelpedtoimprovethemanuscript.TheHigh-EnergyAstrophysicsGroup ofPalermoacknowledgessupportfromtheFondoFinalizzatoallaRicerca(FFR) Laor,A.1991,ApJ,376,90 2012/13,projectN.2012-ATE-0390,foundedbytheUniversityofPalermo.This workwaspartiallysupportedbytheRegioneAutonomadellaSardegnathrough Li,F.,Lewin,W.H.G.,&Doxsey,R.1976,IAUCirc.,2983 POR-FSESardegna2007-2013,L.R.7/2007,ProgettidiRicercadiBaseeOri- entata,ProjectN.CRP-60529.Wealsoacknowledgefinancialcontributionfrom theagreementASI-INAFI/037/12/0. Miller,J.M.,Parker,M.L.,Fuerst,F.,etal.2013,ApJ,779,L2 Oosterbroek,T.,Barret,D.,Guainazzi,M.,&Ford,E.C.2001,A&A,366,138 References Pandel,D.,Kaaret,P.,&Corbel,S.2008,ApJ,688,1288 Arnaud,K.A.1996,inAstronomicalSocietyofthePacificConferenceSeries, Pintore,F.,DiSalvo,T.,Bozzo,E.,etal.2015,MNRAS,450,2016 Vol. 101, Astronomical Data Analysis Software and Systems V, ed. G. H. Jacoby&J.Barnes,17 Pintore,F.,Sanna,A.,DiSalvo,T.,etal.2016,MNRAS,457,2988 Balucinska,M.&Czerny,M.1985,ActaAstron.,35,291 Piraino,S.,Santangelo,A.,diSalvo,T.,etal.2007,A&A,471,L17 Bhattacharyya,S.&Strohmayer,T.E.2007,ApJ,664,L103 Brenneman,L.W.&Reynolds,C.S.2006,ApJ,652,1028 Ross,R.R.&Fabian,A.C.2005,MNRAS,358,211 Cackett,E.M.,Miller,J.M.,Ballantyne,D.R.,etal.2010,ApJ,720,205 Verner,D.A.,Ferland,G.J.,Korista,K.T.,&Yakovlev,D.G.1996,ApJ,465, 487 Cackett,E.M.,Miller,J.M.,Bhattacharyya,S.,etal.2008,ApJ,674,415 Wilms,J.,Allen,A.,&McCray,R.2000,ApJ,542,914 Chiang,C.-Y.,Cackett,E.M.,Miller,J.M.,etal.2016,ApJ,821,105 Church,M.J.&Balucin´ska-Church,M.2001,A&A,369,915 Z˙ycki,P.T.,Done,C.,&Smith,D.A.1999,MNRAS,309,561 Articlenumber,page7of13 A r tic le n u m b e r, p a g e 8 o f 1 3 Table1.ResultsofSpectralAnalysisofSerX-1fromPreviousStudies A Instrument ContinuumModel ReflectionModel LineModel LineEnergy(keV) Equivalentwidth Rin(Rg) Incl(deg) Emissivityindexlog(ξ) Flux(ergs/cm2/sec) Reference & ASCA bbody+cutpowerlaw — gaussian 6.6±0.17 81eV — — — — Ref(1) A RXTE bbody pexrav gaussian – – — — — — Ref(2) p r BeppoSAX bbody+compTT — gaussian 6.46+−00..1124 2757−555eV — — — — Ref(2) oo XSUMZMA-KNUewton dbibsokdbyb++dciosmkbpbT+Tpowerlaw —— dislakolirne 66..8430+−+−0000....10005680 18362-1±0152eeVV 7.47-±106.5 4206-±520 —— 0.25--1100kkeeVV::(53..93±-40..29))××1100−−99a RReeff((43)) fs:m SUZAKU bbody+diskbb+powerlaw — diskline 6.97+0.15 98eV 8.0±0.3 24±1 — 0.5-25keV:(1.19±0.01)×10−8 Ref(5) a −0.02 n SUZAKU bbody+diskbb+powerlaw reflionx — — — 6±1 16±1 2.6±0.1 0.5-25keV:(1.32±0.08)×10−8 Ref(5) u s XMM-Newton bbody+diskbb+powerlaw — diskline 6.66-6.97 38-50eV 14-26 13-32 — 0.5-25keV:(0.6-0.7)×10−8 Ref(5) cr XMM-Newton bbody+diskbb+powerlaw reflionx — — — 15-107 3-9 2.6-2.8 0.5-25keV:(0.6-0.7)×10−8 Ref(5) ip NuSTAR bbody+diskbb+powerlaw — kerrdisk 6.97±0.01 91±2eV 10.6±0.6 18±2 — (0.5-40keV:1.5×10−8 Ref(6) tn NuSTAR bbody+diskbb+powerlaw reflionx — — — 6-8.3 <10 2.30-2.60 — Ref(6) o. Chandra bbody+diskbb+powerlaw — diskline 6.97±0.02 149±15eV 7.7±0.1 24±1 — — Ref(7) se Chandra bbody+diskbb+powerlaw reflionx — — — 7.1+1.1 29±1 2.5+0.9 — Ref(7) rx Chandra bbody+diskbb+powerlaw xillver — — — 8.4+−10..16 33±1 2.2+−00..76 — Ref(7) 1_ −0.3 −0.5 r e f Notes.Ref(1):Church&Balucin´ska-Church(2001)-Ref(2):Oosterbroeketal.(2001)-Ref(3):Bhattacharyya&Strohmayer(2007)-Ref(4):Cackettetal.(2008)-Ref(5):Cackettetal.(2010)- _m Ref(6):Milleretal.(2013)-Ref(7):Chiangetal.(2016) o d _ 2 a Estimatedonlyforthecontinuumcomponent M.Matrangaetal.:Are-analysisoftheNuSTARandXMM-Newtonbroad-bandspectrumofSerX-1 Table2.ResultsofthefitofNuSTARandXMM-NewtonspectraofSerX-1usingGaussianandDisklinemodels Component Parameter gauss diskline gauss-pl diskline-pl diskline-pl-xmm phabs NH(×1022cm−2) 0.4(f) 0.4(f) 0.4(f) 0.4(f) 0.863±0.008 bbody kTbb(keV) 0.47±0.03 0.54±0.06 0.44±0.04 0.47±0.05 0.47±0.02 RBB (km) 46.1±6.3 34.3±7.7 45.5±9.5 39.2±8.7 35.1±3.2 bbody Norm(×10−3) 22.6±2.3 21.8±0.8 16.9±3.4 16.3±2.2 13.1±0.9 gaussian E(keV) 6.57±0.05 — 6.56±0.05 — gaussian Sigma(keV) 0.37±0.04 — 0.39±0.04 — gaussian Norm(×10−3) 4.03±0.35 — 4.48±0.34 — diskline lineE(keV) — 6.54±0.04 — 6.54±0.03 6.48±0.06 diskline Betor — -2.59±0.12 — -2.54±0.13 -2.58±0.18 diskline Rin (Rg) — 18.6±4.9 — 19.2±4.7 22.0+−25..72 diskline Rout (Rg) — 2400(f) — 2400(f) 2400(f) diskline Incl(deg) — 40.1±3.6 — 41.5±3.9 46.1±5.6 diskline Norm(×10−3) — 4.38±0.47 — 4.54±0.35 2.89±0.28 nthComp Gamma 2.41±0.04 2.43±0.04 2.26±0.04 2.27±0.04 2.10+0.14 −0.06 nthComp kTe(keV) 2.95±0.05 2.98±0.04 2.75±0.05 2.76±0.05 2.27±0.16 nthComp kTbb(keV) 0.96±0.03 0.99±0.04 0.90±0.04 0.92±0.04 0.92±0.06;0.82±0.05;0.88±0.06 nthComp Norm(×10−3) 219±11 200±15 229±12 217±18 160±13 powerlaw Index_pl — — 3.20(f) 3.20(f) 3.20(f) powerlaw Norm — — 0.84±0.12 0.82±0.13 0.72±0.04 gau-rgs E(keV) — — — — 0.528(f) gau-rgs Sigma(×10−3keV) — — — — 2.19(f) gau-rgs Norm(×10−3) — — — — -18.4(f) gau-rgs E(keV) — — — — 0.541(f) gau-rgs Sigma(×10−3keV) — — — — 1.36(f) gau-rgs Norm(×10−3) — — — — 57.1(f) gau-rgs E(keV) — — — — 0.714±0.02 gau-rgs Sigma(×10−3keV) — — — — 5.8±0.6 gau-rgs Norm(×10−3) — — — — -12.1±0.7 - Eq.W(eV) 76±6 85±7 84±6 89±9 72±16;93±18;79±16 - Obs.Flux 5.25±0.03 5.27±0.03 5.27±0.02 5.27±0.02 3.68±0.24 - Luminosity 3.72±0.02 3.72±0.02 3.73±0.02 3.73±0.02 2.62±0.17 χ2 (d.o.f.) - 1.2750(915) 1.2186(913) 1.14134(914) 1.0961(912) 1.3521(4546) red Notes.Fluxandluminosityareobtainedforthe3–40keVenergyband.Fluxesunitsare10−9 (ergs/cm2/sec),whereasluminositiesunitsare1037 (ergs/sec).Theseed-photontemperaturewasleftfreetovaryamongthethreedifferentXMM-Newtonobservations,thisiswhywereportthree valuesforthisparametersintheXMM-Newtonfittingresults(seetextformoredetails).ThevaluesoftheparameterEq.W.referstotheequivalent widthofthetheironlineat6.48keVdetectedineachobservation.Errorsarereportedwitha90%confidence.R andluminositiesareestimated BB assumingadistanceof7.7kpc(Gallowayetal.2008) AppendixA: Modelsincludingkerrconv componentsconvolvedwithkerrconvinsteadof rdblur.Ker- rconvconvolvesthespectrumwiththesmearingproducedbya FromthespectralanalysisdescribedinSec.3.1,wefindthatour kerrdiskmodel.Itfeaturesthedimensionless’a’parameterthat best fit obtained using rdbluras smearing componentgives a characterize the spin of the system. We have performedour fit softblackbodytemperatureof0.54±0.06keVandaradiusofthe firstleaving’a’asafreeparameterandthenfixingittothefol- emittingregionof25±6km,atemperatureoftheseedphotons lowingthreevalues,0,0.12,0.14.Themodelwithreflionxand fortheComptonizationof0.93±0.07keV,anelectrontempera- ’a’ treated as free parameter is called ker-reflio-af, whereas for ture of the Comptonizingcorona of 2.70±0.04keV and a pho- a = 0, a = 0.12anda = 0.14the modelsare called ker-reflio- tonindexoftheprimaryComptonizedcomponentof2.17±0.04, a0,ker-reflio-a012,andker-reflio-a014,respectively.Inthesame whereasthe photonindexof the hardpower-lawtailis steeper, way,themodelwithrfxconvand’a’treatedasfreeparameteris around 3.2. The reflection component gives a reflection am- calledker-rfxconv-af,whereasfora = 0,a = 0.12anda = 0.14 plitude of 0.24±0.04 and a ionization parameter of log(ξ) = themodelsarecalledker-rfxconv-a0,ker-rfxconv-a012,andker- 2.69+0.02.Finally,thesmearingofthereflectioncomponentgives rfxconv-a014,respectively.All the models fit the data well; re- −0.11 an innerdisk radiusof R = 13.4±2.8R , compatiblewith the ducedχ2 arebetween1.08and1.18andresidualsarebasically in g radiusinferredfromtheblackbodycomponent,andanemissiv- identical.Moreoverthebest-fitvaluesofallparametersarevery ity index of the disk equal to -2.64±0.16,whereas the inclina- similar to the case with a = 0 and to the values we get using tion angle of the disk with respect to the line of sight results rdblurinsteadof kerrconv.Thefitisthereforeinsensitiveto equalto 27.1±1.9◦. The analysis of XMM-Newton spectra, car- thespinparameter,althoughthereisaslightpreferenceofthefit riedoutusingthesamemodelsadoptedtofittheNuSTARspec- towardslowvalues(a < 0.04).Itisworthnotingthatinallbest tra,gavevaluesoftheparameterscompatibletothosedescribed fitresidualsafeatureispresentatabout3.9keVthatcouldbethe above, although the two observations are not simultaneous. In resonancelineofCaXIX(3.9keV).Moreover,againweobserve particular in this case we find R 14.2+9.5 R , a reflection am- highenergyresiduals(above30keV)indicatingthepresenceof in −4.6 g plitude of 0.183±0.003 and an ionization parameter of log(ξ) a hard power-law component. Also in this case, we get a very = 3.04±0.11, a temperature of the seed photons in the range largeionizationparameterusingreflionx. 0.76 − 0.85 keV, a photon index of the primary Comptonized Toavoidthisproblem,wethereforeaddedapower-lawcom- componentof2.45±0.22keV.Inotherwords,theXMM-Newton ponentto the model obtained by the convolutionof kerrconv spectraindependentlyconfirmtheresultsobtainedfortheNuS- with the two different reflection components (reflionx or rfx- TARspectra. conv).weconsidered’a’freetovaryorfixedittothreedifferent Inordertocheckthepresenceofa non-nullspinparameter values(0,0.12,0.14).Inallthecasesthefitsarequitegoodwith oftheneutronstar,wefittedtheNuSTARspectrausingreflection valuesofthereducedχ2 from1.0to1.01.Againtheadditionof Articlenumber,page9of13 A&Aproofs:manuscriptno.serx1_ref_mod_2 Table3.ResultsofthefitofNuSTARandXMM-NewtonspectraofSerX-1usingrdblurcombinedwithrfxconvorreflionx Component Parameter rdb-rfxconv rdb-reflio rdb-rfxconv-pl rdb-reflio-pl rdb-rfxconv-pl-xmm phabs NH(×1022cm−2) 0.4(f) 0.4(f) 0.4(f) 0.4(f) 0.896±0.005 bbody kTbb(keV) 0.71±0.02 0.80±0.02 0.54+−00..0052 0.54±0.06 0.39±0.04 RBB (km) 23.6±1.3 15.9±0.8 24.7±7.9 19.2±4.6 49.4±10.6 bbody Norm(×10−3) 30.9±0.5 22.5±0.6 11.3+3.3 6.8±1.2 12.3±1.6 −6.1 highecut Ecut(keV) — 0.1(f) — 0.1(f) — highecut Efold(keV) — 8.61±0.19 — 5.04±0.09 — rdblur Betor -3.02±0.36 -2.49±0.15 -2.64±0.16 -2.53±0.14 -2.46+0.56 −0.42 rdblur Rin (Rg) 7.7±1.3 15.5±4.6 13.4±2.8 13.2±3.1 14.2+−94..56 rdblur Rout (Rg) 2400(f) 2400(f) 2400(f) 2400(f) 2400(f) rdblur Incl(deg) 29.2±1.8 32.2±1.7 27.1±1.9 28.8±2.4 27(f) reflionx Gamma — 2.88±0.08 — 1.51±0.03 — reflionx ξ — 4990+695 — 490+21 — −2350 −98 reflionx Norm(×10−5) — 1.97±0.59 — 10.7±3.5 — rfxconv rel_refl 0.55±0.04 — 0.24±0.04 — 0.183±0.022 rfxconv cosIncl 0.88(f) — 0.88(f) — 0.891(f) rfxconv log(ξ) 2.68±0.05 — 2.69+0.02 — 3.04±0.11 −0.11 nthComp Gamma 3.55±0.18 2.88±0.08 2.17±0.04 1.51±0.03 2.45±0.22 nthComp kTe(keV) 4.36+−00..5273 3.19±0.08 2.70±0.04 5.05±0.09 3.83+−11..9012 nthComp kTbb(keV) 1.51±0.04 1.43±0.05 0.93±0.07 1.04±0.18 0.85±0.05;0.76±0.06;0.82±0.06 nthComp Norm(×10−3) 71.2±7.2 69.7±4.2 192±24 286+18 205±21 −22 powerlaw Index_pl — — 3.21±0.24 3.20(f) 3.98±0.31 powerlaw Norm — — 1.08+1.12 0.82±0.13 0.68±0.05 −0.72 - Obs.Flux 5.26±0.15 5.27±0.17 5.27±0.62 5.27±0.55 4.12±0.38 - Luminosity 3.73±0.11 3.74±0.12 3.74±0.44 3.74±0.39 2.93±0.27 χ2 (d.o.f.) - 1.0983(913) 1.0838(913) 1.0017(911) 1.0123(912) 1.33762(4546) red Notes.Foreachfit,theabundanceofironinthereflectionmodelswaskeptfrozen:Fe/solar=1.Fluxandluminosityareobtainedforthe3–40 keVenergyband.Fluxesunitsare10−9(ergs/cm2/sec),whereasluminositiesunitsare1037(ergs/sec).Theseed-photontemperaturewasleftfree tovaryamongthethreedifferentXMM-Newtonobservations,thisiswhywereportthreevaluesforthisparametersintheXMM-Newtonfitting results(seetextformoredetails).Errorsarereportedwitha90%confidence.R andluminositiesareestimatedassumingadistanceof7.7kpc BB (Gallowayetal.2008) data and folded model data and folded model 10 10 V−1 1 V−1 1 normalized counts s ke−1 0.00.11 normalized counts s ke−1 0.00.11 10−3 10−43 model)/error 02 model)/error 02 (data− −2 (data− −2 5 10 20 5 10 20 Energy (keV) Energy (keV) data and folded model data and folded model 10 10 V−1 1 V−1 1 normalized counts s ke−1 0.00.11 normalized counts s ke−1 0.00.11 10−3 10−43 2 model)/error 0 model)/error 02 (data− −2 (data− −2 5 10 20 5 10 20 Energy (keV) Energy (keV) Fig.1.NuSTARspectraofSerX-1andbest-fittingmodeltogetherwithresidualsinunitsofsigmaforthecorrespondingmodel.Theseare:Top left:’gauss’—Topright:’diskline’—Bottomleft:’rdb-reflio’—Bottomright:’rdb-rfxconv’.Dashedlinesindicatetheblack-bodycomponent, dottedlinesindicatethereflectioncomponents(i.e.theGaussianorDisklineprofilefortheironline,toppanels,ortheself-consistentreflection component,bottompanels,respectively),andthedashed-dottedlinesindicatethecomptonizedcomponent. Articlenumber,page10of13