Table Of Content

Astronomy&Astrophysicsmanuscriptno.Gimenez-Garcia (cid:13)cESO2015 January21,2015 α An XMM-Newton view of FeK in HMXBs A.Giménez-García1,4,6,J.M.Torrejón1,2,W.Eikmann3,S.Martínez-Núñez2,L.M.Oskinova4,J.J.Rodes-Roca1,2,5, andG.Bernabéu1,2 1 UniversityInstituteofPhysicsAppliedtoSciencesandTechnologies,UniversityofAlicante,P.O.Box99,E03080Alicante,Spain e-mail:[email protected] 2 X-rayAstronomyGroup.DepartamentodeFísica,IngenieríadeSistemasyTeoríadelaSeñal,UniversityofAlicante,P.O.Box 99,E03080Alicante,Spain 3 Dr.KarlRemeis-SternwarteFAUErlangen-Nürnberg,D96049Bamberg,Germany 5 4 InstituteforPhysicsandAstronomy,UniversityofPotsdam,D-14476Potsdam,Germany 1 5 MAXIteam,InstituteofPhysicalandChemicalResearch(RIKEN),2-1Hirosawa,Wako,Saitama,351-0198,Japan 0 6 SchoolofPhysics,FacultyofScience,MonashUniversity,Clayton,VIC3800,Australia 2 n Accepteddate:25thDecember2014 a J ABSTRACT 0 2 We present a comprehensive analysis of the whole sample of available XMM-Newton observations of High Mass X-ray Binaries (HMXBs)untilAugust,2013,focusingontheFeKαemissionline.Thislineisakeytooltobetterunderstandthephysicalproperties ] ofthematerialsurroundingtheX-raysourcewithinafewstellarradii(thecircumstellarmedium).Wehavecollectedobservations E from46HMXBs,detectingFeKαin21ofthem.WehaveusedthestandardclassificationofHMXBstodividethesampleindifferent H groups. We find that: (1) different classes of HMXBs display different qualitative behaviour in the FeKα spectral region. This is . speciallyvisibleinSGXBs(showingubiquitousFefluorescencebutnotrecombinationFelines),andγCassanalogs(showingboth h fluorescentandrecombinationFelines).(2)FeKαiscentredatameanvalueof6.42keV.Consideringtheinstrumentalandfitsuncer- p tainties,thisvalueiscompatiblewithionizationstateslowerthanFeXVIII.(3)Thefluxofthecontinuumiswellcorrelatedwiththe - o fluxoftheline,asexpected.EclipseobservationsshowthattheFefluorescenceemissioncomesfromanextendedregionsurrounding r theX-raysource.(4)WeobserveaninversecorrelationbetweentheX-rayluminosityandtheequivalentwidthofFeKα(EW).This t phenomenon is known as X-rays Baldwin effect. (5) FeKα is narrow (σ < 0.15 keV), reflecting that the reprocessing material s line a doesnotmoveathighspeeds.Weattempttoexplainthebroadnessofthelineintermsofthreepossiblebroadeningphenomena:line [ blending,ComptonscatteringandDopplershifts(withvelocitiesofthereprocessingmaterialV ∼1000km/s).(6)Theequivalenthy- drogencolumn(N )directlycorrelateswiththeEWofFeKα,displayingclearsimilaritiestonumericalsimulations.Ithighlightsthe H 2 stronglinkbetweentheabsorbingandthefluorescentmatter.(7)TheobservedN inSupergiantX-rayBinaries(SGXBs)isingeneral H v higherthaninSupergiantFastX-rayTransients(SFXTs).Wesuggesttwopossibleexplanations:differentorbitalconfigurations,or 6 adifferentinteractioncompactobject-wind.(8)Finally,weanalysedinmoredetailthesourcesIGRJ16320-4751and4U1700-37, 3 covering several orbital phases. The observed variation of N between phases is compatible with the absorption produced by the H 6 windoftheiropticalcompanions.Theobtainedresultsclearlypointtoaveryimportantcontributionofthedonor’swindintheFeKα 3 emissionandtheabsorptionwhenthedonorisasupergiantmassivestar. 0 . Keywords. Stars:binaries,circumstellarmatter-X-rays:binaries-Surveys 1 0 5 11. Introduction studyoftheFecomplex,andmotivatesacomprehensiveanaly- : sisinHMXBs. vSincetheearlystagesofX-rayastronomy,Felinesinthespec- In particular, FeKα has been proven as a fundamental Xitral regionof ∼6-7 keV (the Fe complex) have been studied in tool in the study of HMXBs (Martínez-Núñezetal. 2014; a largenumberof X-ray sourcesgivenits fruitfulnessas a tool Rodes-Rocaetal.2011;vanderMeeretal.2005).Theoriginof r aforplasma diagnostics.Theywerereportedforthe first timein the fluorescence emitting region has been discussed by many thesupernovaremnantCasA(Serlemitsosetal.1973),andonly authors in the past. Nagase (1989) considered accretion disks two years later in a High Mass X-ray Binary (HMXB) using and the matter stagnated in the accretion and ionization wakes the Ariel 5 satellite (Sanfordetal. 1975). The most recent X- in the stellar wind as plausible areas of FeKα production. rayspacemissions(Swift,Suzaku,ChandraandXMM-Newton) Watanabeetal. (2006) analysedthe classicalHMXB Vela X-1, have triggered a notable improvement in the attainable spec- andproposedtheextendedstellarwind,reflectionoffthestellar tral resolution and effective area, permitting to distinguish be- photosphereandanaccretionwake,asthemostlikelycandidates tweendifferentemissionfeaturesintheFecomplex:narrowand forfluorescencereprocessingregions.Inanycase,FeKαisvery broadfluorescencelines(FeKαandFeKβ),Comptonshoulders sensitive to the physicalconditionsof the vicinity of the X-ray andrecombinationlines(Fe XXV andFe XXVI)(Torrejónetal. source,andprovidesremarkableinformationthatmustbeanal- 2010b).Thisimprovementhasgivenaremarkableimpetusinthe ysed. Fluorescenceis producedas a consequenceofthe X-ray il- lumination of matter. When an Fe atom absorbs a photon car- Sendoffprintrequeststo:A.Giménez-García rying sufficient energy to remove an electron from its K-shell Articlenumber,page1of30 A&Aproofs:manuscriptno.Gimenez-Garcia (E>7.2 keV), the vacancy can be occupied by another electron (clumps) are present as an intrinsic feature of the radiatively froman outershell. If the electroncomesfromthe L-shell, the driven winds of hot stars (Lucy&White 1980; Oskinovaetal. transitionproducesFeKαemission.FeKβemissionisproduced 2012). Second, hydrodynamical simulations show that the X- when the vacancyis filled by a formerM-shell electron.When rayradiationandthegravityfieldofthecompactobjectdisturb Fe ismoreionizedthanFe XIX,the fluorescenceyieldstartsto the wind of the donor, inducing the formation of denser struc- decreasewith theionizationstate (Kallmanetal.2004). There- tureslikefilaments,bowshocksandwakes(Blondinetal.1990, fore,FeKαisafootprintofnotextremelyionizedFe (lessthan 1991). Fe XX). On the other hand, recombination lines Fe XXV and Inthelastdecadeandahalf,newdiscoverieshaveledtothe Fe XXVI unveil the presence of very hot gas, where Fe atoms additionofnewgroupstothepreviouspictureofHMXBs,stress- arealmostcompletelystripped. ing the value of grasping the different features of the sources Previous comprehensive surveys of the Fe complex in such as geometry,compactobjectproperties,opticalstar pecu- HMXBs were carried out by Gottwaldetal. (1995) using EX- liarities and wind clumpiness. The new groups are Supergiant OSATandTorrejónetal.(2010b)usingtheHighEnergyTrans- Fast X-ray Transient systems (SFXTs), γ Cassiopeae analogs mission Gratings (HETGS) on board Chandra. The high spec- andγ-rayBinaries. tralresolutionprovidedbyChandragratingsprovedtobeinstru- SFXTs are systems with a supergiant optical star, as in mental in disentangling the differentionization species present SGXBs,butdefinedbyanextremelytransientbehaviour.During in the Fe complex. However, the relatively low throughput of quiescencetheyexhibitlow luminosity(∼ 1032 erg/s),butthey the instrument allowed to study only the brightest binaries. In spend most of their time in an intermediate level of emission this study we increase significantly the previoussample by us- (∼ 1033−34 erg/s). They display short outbursts (∼few hours), ingthehighthroughputofXMM-NewtonEPN.Thishasallowed reachingluminosities up to 1036−37 erg/s (Sidolietal. 2009). It ustoincludefaintersystems(likeBeX-raybinaries(BeXBs)or islikelythattheclumpinessofthewindplaysamainroleinthe SFXTsinquiescence)whilethemoderateresolutionoftheEPN variability of these sources. Other mechanisms involving cen- CCDs has allowed us to test previous correlations based on a trifugalandmagneticbarrierscouldenhancetheobservedlumi- smallsample. nosity swings, relaxing the needed variation amplitudes in the HMXBs are specially susceptible to be studied using the physical conditions of the wind (Bozzoetal. 2008). Neverthe- Fe complex,onaccountofthesignificanceofthe circumstellar less,otherauthorsexplainthevariabilityappealingtothequasi- medium in the observable phenomena. These systems consists spherical accretion model (Draveetal. 2013; Paizis&Sidoli of a compact object, either a neutronstar (NS) or a black hole 2014). (BH), accreting matter from a massive OB star (usually called γCassiopeaeanalogsarecharacterizedbythethermalnature optical or normal star of the system). In HMXBs the observed of the X-ray emission, with plasma temperatures of ∼ 108 K luminosity is commonly powered via accretion. Consequently, (∼ 10keV),anX-rayluminosityof1032−33 erg/s,andhighflux the way that matter is accreted from the donor directly defines variabilityonvarioustimescales.However,theydonotdisplay theobservableluminosityfeaturesofeverysource. giant outbursts as observed in BeXBs (LopesdeOliveiraetal. WhentheopticalstarisaBestar,thesystemisaBeXB.Be 2010).Presently,itisnotclearthattheX-rayemissionisemitted stars are fast-rotating BIII-V stars, which have shown spectral byaccretionprocesses(ontoaneutronstarorawhitedwarf),or emission lines at some point of their lives. They also show an alternativelygeneratedfromtheinteractionbetweenthesurface excessofinfraredemission,whentheyarecomparedtononBe ofthestar,thecircumstellardiskanditsmagneticfield. starsofthesamespectraltype.Theseobservablesareexplained High Mass γ-ray Binary systems (HMGBs) are HMXBs appealing to an extended circumstellar decretion disk. BeXBs wheretheemissionpeaksabove1MeV.Nowadays,itisthought are usually transient in the X-rays, although some systems ex- thattheemissioniscausedbyacceleratedparticlesintheshock hibit a persistent quiescence emission (L≤ 1034−35 erg/s). The thatisproducedwhenthepulsarwindcollidesthemassivestar outburstshavebeentraditionallyclassified intwotypes.TypeI wind. Therefore, they are powered by the rotational energy of outburst(L≤1037erg/s)arerelatedtoperiastronpassages.Type theneutronstar,inoppositiontotherestofHMXBs,whichare IIoutburstarenotrelatedtotheorbitalphaseandimplyaneven accretionfed.TherearecurrentlyfiveconfirmedHMGBs,allof higherincreaseinluminositythanTypeIoutbursts,reachingthe themwithamainsequenceopticalstar(forareviewonHMGBs Eddingtonluminosity(forareviewonBeXBsseeReig(2011)). seeDubus(2013)). InthecaseofclassicalSupergiantX-rayBinaries(SGXBs), Finally,therearesourceswhich,followinganumberofrea- thecompactobjectisembeddedinthedenseandpowerfulwind sons,cannotbeclassifiedinanyofthealreadymentionedclasses of a OB supergiantcompanion,swallowing everythingthat en- ofHMXBs.Particularly,amongthesetofsourcesstudiedinthis ters its gravitational domain. The mass loss rate of the donor paper, they are 4U 2206+54, Centaurus X-3 and Cygnus X-1. is of the order of & 10−7M⊙yr−1, and the compact object is The optical star in 4U 2206+54 is a O9.5V (Blayetal. 2006), usually found at a close distance of a ∼ 1.5 − 2R . In such neither a supergiant nor a Be star. The system may be part of ⋆ a close orbit, the captured matter is able to fuel a persistent a new groupof wind-fedHMXBs with a main sequencedonor X-ray emission of ∼ 1033−39 erg/s. Flares and off-states are (Ribóetal.2006).CentaurusX-3andCygnusX-1aretheonly often observed in SGXBs, indicating an abrupt transition in systems here collected where accretion is persistently driven the accretion rate. They might be produced either by sudden by an accretion disk (Tjemkesetal. 1986; Shapiroetal. 1976), variations of density in the medium transited by the compact whatisreflectedinthespectraofbothsources. object (Martínez-Núñezetal. 2014; Kreykenbohmetal. 2008), In thispaper,we study the FeKαline forthe wholesample or either by instabilities above the magnetosphere of the neu- ofHMXBsavailablewithXMM-NewtonuntilAugust,2013.In tronstar,asproposedinthequasi-sphericalaccretiontheoryby Section2wepresentthesetofobservations,thereductionpro- Shakuraetal.(2012). cessandthemoreimportantdetailsconcerningthespectralfits. Themediumtransitedbythecompactobjectthroughtheex- In Section 3 we show the obtainedresults: an spectralatlas in- tended atmosphere of an OB supergiant star is not smooth be- cluding every fit and different plots relating fit parameters. In causeof,atleast,twophenomena.First,densityinhomogeneities Section 4 we interpret the obtained results and we summarize Articlenumber,page2of30 A.Giménez-García:Title themostimportantconclusionsinSection5.InAppendixAwe presenta set of tables describingthe obtainedparametersfrom thespectralfits.InAppendixBweshowthespectralatlas,con- tainingtheplotofeveryspectrumthatwehaveanalysedinthis survey.Weshowtheobservationsandthemodels,togetherwith theratiobetweenthem. 2. Observationsanddatatreatment TheXMM-Newtonobservatory(Lumbetal.2012)isfittedwith three X-ray telescopes of 1500 cm2 and a coalignated optical telescope.Spectroscopyandphotometryisdonebythe6instru- mentsonboard:threeX-rayimagingcamerasEPIC(European PhotonImagingCamera),twogratingXrayspectrometersRGS (ReflectionGratingSpectrometer)andanopticalmonitor(OM). EPIC cameras(0.1-15keV) are the only instrumentsat XMM- Newton covering the energy range of the Fe complex. Among EPIC, one camera uses PN CCDs, andthe other two use MOS CCDs. EPIC PN cameras(EPN) surpass by a factor ∼3 the ef- fectiveareaoftheMOScamerasat6-7keV,makingEPNmore Fig. 1: Light curve of the observation of 4U 1538-522 (Ob- suitableforourpurposes.Comparedtoothermissions,theHigh sID:0152780201). Wehavesplit theobservation intwoparts, onefor theingressintheeclipse,andanotheronefortheeclipse. EnergyTransmissionGratingSpectrometer(HETGS)onboard ofChandraprovidesbetterenergyresolutionintheenergyrange of the Fe complex,butthe effectiveareaavailable with EPN is observation2.ThesourcesnotincludedintheLiucatalogue,but significantly higher. EPN provides the adequate conditions to here considered, are: HD 119682, SS 397, IGR J16328-4726, perform the study here presented, on account of the moderate (butsufficient)spectralresolution(∆E/E ∼40),andgreateffec- HD45314,HD157832,SwiftJ045106.8-694803,IGRJ16207- 5129andXTEJ1743-363. tivearea(∼1000cm2),enablingustoanalyzealargeamountof sourcesinanhomogeneousandconsistentway. SinceHMXBsareusuallyvariable,weoftenobserveinthe 2.1. Datareduction same observation a dramatic change in luminosity, remarkably affecting the spectral parameters. In these cases, an averaged WehavereducedthedatausingScienceAnalysisSystem(SAS), spectrumdoesnotreproducethe actualemissionof the source, version12.0.1.Sincethesampleofobservationscontainanhet- anditisadvisabletosplittheobservationinmorethanonetime erogenous group of HMXBs, we found different observation interval.Wehaveconsideredfivedifferentstates1ofthesystems modes (timing and imaging) to account for the different prop- in order to define the time intervals: dips, quiescence, flares, erties of the sources. In the brightestsystems, the observations eclipse ingress/egress, and eclipse. We have used the follow- wereusuallyperformedusingthetimingmode,whilethefaintest ingcriteria.Whenluminositydropsafactor&2inthetimescale sourceswereobservedusingimagingmodes. of .1 hour, we have tagged the time interval as a dip. Analo- Timing modes permit to process the arrival of photons at gously,whenluminosityrises&2inthetimescaleof.1hour,we high rate, since only one CCD operates and the information is havelabelledthetimeintervalasaflare.Forobservationscover- collapsedintoonedimension,allowingafastreadout.Thetime ingeclipsingphases,wehavedefinedtimeintervalsforeclipse resolutionisashighas30µs (7 µs inburstmode,Kirschetal. ingress/egressandeclipse.Therestoftimeintervalsaretagged (2006)). Even though the high timing resolution reached with asquiescentstates. theseobservationmodes,pile-upisstillpresentinseveralcases, In Figure 1 we see the light curve of an observation of specially when the count rate is & 800 counts s−1. We have 4U 1538-522,as an exampleof how we havesplit the time in- checkedineveryobservationifpile-upisaffectingthedata,us- tervalsintheobservations.Thesourcewasobservedduringthe ingthe SAS task epatplot,andwe have excisedthe coreof the ingress in an X-ray eclipse, which is clearly noticeable in the source’spointspreadfunctioninthepertinentcases.Thesizeof lightcurve.Wehaveseparatedtheobservationintwotimeinter- theexcisedregionhasbeenchosenwide enoughto removethe vals,onecoveringtheingressineclipse,andanotheronecover- unwantedpile-upeffects(seeexamplesoftheuseofepatplotin ingtheeclipse. Ngetal.(2010)). Summarizingthesampleofobservations,wehavecollected Background subtraction process is also dependent on the data from 46 HMXBs. 21 of them exhibit FeKα emission. We brightness of the source. In the EPN timing mode, the PSF of note that some sources have more than one available observa- the sources displaying & 200 counts s−1 will span across the tion.Takingeverythingintoaccount(46sources,temporalsplit- whole CCD. Therefore, any area selected as a background re- ting depending on the state of the source, and more than one gion will be contaminated by source photons. Since this effect observation per source in some cases), we end up with a total isstronglyenergydependent,forthebrightestsourceswe have numberof108spectrathatwehaveanalysed. chosen a method of background subtraction similar to the one WehavefollowedthecatalogueofLiuetal.(2006),inaddi- performedintheanalysisofVelaX-1byMartínez-Núñezetal. tiontolaterdiscoveriesorconfirmations,toidentifythecurrently (2014),whereablankskyspectrumtakenintimingmodeisused knownHMXBs,andusedeveryavailableXMM-Newtonpublic astherealbackgroundforenergiesbelow2.5keV,whiletherest ofthespectrumcorrespondstotheoutermostpixelsoftheCCD. 1 Thestatesconsideredinthisworkandthealsocalledstatesinblack holebinarysystemsmustnotbemislead. 2 http://xmm.esac.esa.int/xsa/ Articlenumber,page3of30 A&Aproofs:manuscriptno.Gimenez-Garcia are thermal, we classify the modelas thermal (analogouslyfor non-thermal).Wehavealsousedhybridmodels,combiningther- malandnon-thermalcomponents.Thethermalcomponentsused inthisworkarethefollowing: – bbody:blackbodyemission. – diskbb:modelofanaccretiondiskemissionmadeofmultiple blackbodycomponents. – bremss: thermal bremsstrahlung emission (electrons dis- tributedaccordingtotheMaxwell–Boltzmanndistribution). – mekal:emissionfromopticallythinhotgas,includingspec- trallinesfromseveralelements(Meweetal.1985). – cemekl: built from the mekal model, incorporating multi- temperatureemission. Ontheotherhand,theonlynon-thermalcomponentusedin thisworkis: – powerlaw: phenomenologicalmodel consisting of a simple Fig. 2: Number of accepted models depending on the Reduced-χ2 inverse power law profile (∝ E−Γ). This profile is a foot- value. print of Inverse-Compton scattering by hot electrons (non- thermallydistributed)ofaseedradiationfield. Meanwhile,forcommonobservations,wehaveusedsource-free For the photoelectric absorption, we have used tbnew4, the regions to extract a background spectrum and subtract it from improvedversionoftheWilmsetal.(2000)modeltbabs,setting theformersourceplusbackgroundenergydistribution. thecrosssectionstotheVerneretal.(1996)onesandtheabun- Ancillaryresponse files were generatedusing the SAS task dancesaccordingtoWilmsetal.(2000).Themostimportantpa- arfgen.Forobservationstakenintimingmodeaffectedbypile- rameter of this model is the total equivalent hydrogen column up,wehavefollowedtherecommendationsoftheXMM-Newton N , which is the integrated amount of hydrogen atoms in the H SAS User Guide in order to generate the appropriate ancillary lineofsightfromthe observerto the source,percm2. We have response files. Response matrices were created using the SAS also added the modelcabsto accountforthe Comptonscatter- taskrmfgen. ing,whichisnotcomprisedinthetbnewmodelandisespecially significativeforN &1024 cm−2. H The emission lines are fitted using Gaussian profiles. We 2.2. Spectralfitting havecategorizedasFeKα anyemission linethatfulfilsthe fol- ForthespectralanalysiswehaveusedXSPEC,version12.8.03. lowingconditions: We have rebinnedthe spectra to have a minimum of 20 counts 1) ThecentroidenergyoftheGaussiancomponentliesinthein- perbin,andabinsizeofatleast1/3oftheFWHMoftheintrinsic terval[6.3,6.65]keV.Theintervalincludestheexpecteden- energyresolution,inordertobeallowedtoapplyχ2statisticsin ergyofFeKαemissionfromFeII(∼6.395keV)toFeXXIII thefittingofasetofpoissoniandata(Cash1979). (∼6.63keV)(Kallmanetal.2004).Thisconditionexcludes In Table 1 we present the sample of models employed for the detectionof anyhypotheticalfluorescentemissionfrom the continuumin the fits. Everymodel is a combinationof ad- FeXXIV-XXVat∼6.67−6.7keV,therebyexcludinganycon- ditive and multiplicative models. An additive model stands for fusionbetweenFeKαandtherecombinationlineFeXXV at asourceofX-rays(e.g.bremsstrahlungradiation),andamulti- similarwavelength.ThefluorescenceyieldsofFeXXIV-XXV plicativemodelrepresentsaenergy-dependentchangeofanad- arelowcomparedtolowerionizationstates. ditivemodel(e.g.photoelectricabsorption). 2) The statistical significance (σ ) of the Gaussian com- The models presented in Table 1 have been tested in every sign ponent is greater than 2σ. We have calculated σ from owbitshernvatthioenn,uamndbearcocfepbtiendsdaenpdenmditnhge nounmthbeerRoefdfiuctteedd-χp2ar(anmχ−2me-, χ2k1 − χ2k2, assuming χ2k1 − χ2k2 ∼ χ2k1−k2 5; where χs2kig1narises fromafitusingcertainmodelwiththeGaussiancomponent ters).Everyobservationhasparticularcharacteristics,andthere- included, and χ2 arising from a fit using the same model fore the decision of which Reduced-χ2 value is acceptable has k2 withouttheGaussiancomponent. beentakenonebyone.InFigure2wecansee thatmostofthe fits result in a Reduced-χ2 ≃ 1, as expected for a suitable fit. In some cases, FeKα line is clearly noticeable,butFeKβ is ThehighestvalueofReduced-χ2foranacceptedmodelhasbeen notprominentenoughtopermiterrorestimationofits parame- 1.82.TheparametersarisingfromthefitsarelistedinTablesA.2 ters.Inthesecases,wehaveconstrainedthecentroidenergyand andA.3. thenormofFeKβaccordingtoKallmanetal.(2004): We can classify the additive components of the models as thermal or non-thermal. A component is called thermal when – Energy(FeKβ)=Energy(FeKα)+0.652keV radiation is produced as a consequence of the thermal motion – Norm(FeKβ)=Norm(FeKα)×0.13photons/cm2/s of the plasma particles (e.g. blackbody radiation). Otherwise, the emitted radiation is non-thermal (e.g. non-thermalInverse- 4 http://pulsar.sternwarte.uni-erlangen.de/wilms/research/tbabs/ Compton emission). If all the additive components of a model 5 Thisassumptionisnotstrictlytrue,sinceχ2 andχ2 arenotindepen- k1 k2 dent.However,itprovidesanestimationontheimpactoftheGaussian 3 http://heasarc.nasa.gov/xanadu/xspec/ componentinthemodel. Articlenumber,page4of30 A.Giménez-García:Title The estimated parameters, like the EW, are very sensitive to Group #Sources Fecomplex Models N H the fit of the continuum. Therefore, although the Fe complex appears in the ∼6-7 keV energies, we broadened the spectral BeXB 10 TypeIII T,TN Low scope to an energy range of 1-10 keV to perform the analysis. It also allows us to consider possible calibration inaccuracies SGXB 12 TypeI N High in the charge transfer inefficiency (CTI) and the X-ray loading (XRL), an issue reportedin previousanalysis of EPN observa- SFXT 10 TypeIII T,N,TN High tions(seeMartínez-Núñezetal.(2014)andFürstetal.(2011)). (inquiescence) InthefewcasesofpossibleCTIorXRL,weappliedanartificial gainE = Eold +offset(seeTableA.2). γCasslike 8 TypeII T Low new slope Theestimationoftheparameterconfidenceregions(at90% level) have been calculated with a Markov Chain Monte Carlo HMGB 2 TypeIII T,N,TN Low (MCMC) technique,implementedin XSPEC, whereN genera- tionsofthesetoffreeparametersareusedtodeterminethebest- Table 2: Description of the features observed in this work, for the fitvaluesandtheconfidenceregions.WehavesetN =1.5×104 different groups of HMXBs. We have analyzed data of 46 sources. inourcalculations.Thesechainsarealsovalidtoestimatefluxes However, those classified as peculiars (3 sources) or non classified andequivalentwidths. (AX J1749.1-2733) are not included in this table. We define NH of a groupashigh,whenthetypicalvalueobservediswellovertheestima- tionsoftheinterstellarN inthelineofsightofthesourcesfollowing H Willingaleetal.(2013).Thatis,wesaythattheN ofagroupishigh H whentheabsorptionistypicallyintrinsicofthesystems. 3. Results 3.1. SpectralAtlas donotdetectFeKα.The12SGXBscanbewellfittedusing InAppendixB, wepresentthefullsampleofanalysedspectra. non-thermalmodels,althoughthermalcomponentsare also ThefiguresinAppendixBshowthesetofanalysedobservations plausibleinsomesources.Ingeneral,SGXBsarecharacter- (crosspoints),themodelemployed(solidline),thecomponents izedbyhighabsorptionandthepresenceofFefluorescence ofthemodel(dottedline)andtheratiobetweenobservationand emissionlines. model(lowerboxineachspectrumplot). – SFXTs. We have collected data from 10 sources. Three of We showa listofthe sourcesinTableA.1,givingthe class themshowFeKα:AXJ1841.0-0536,IGRJ11215-5952and wherewehavegroupedthemandthereferenceforsuchaclas- IGRJ16479-4514.TheEWupperlimitintherestofsources sification. We can see that the differentclasses of HMXBs be- ishigh.Therefore,FeKαwouldbeprobablydetectablewith havequalitativelydifferentintheregionoftheFecomplex(∼6- abettersignal-to-noise.Themodelsemployedforfittingthe 7keV),reflectingthedistinctaccretionregimesthatcharacterize SFXT systems are very heterogeneous, with no preference them. We have observed three patterns in the Fe complex, that ofthermal,non-thermal,neitheracombinationofbothkinds wedefineasTypeI,IIandIII(seeFigure3).WedefineTypeI, ofmodels. when fluorescence lines FeKα and FeKβ are observed,but not – γ Cassiopeae analogs.We have gatheredobservationsfrom recombination lines Fe XXV and Fe XXVI. We define Type II, 8 sources. Five of them exhibitFeKα. The EW upper limit whenfluorescencelinesare detected,togetherwith recombina- in the other 3 sources is very high. Again, it implies a tionlinesFeXXVandFeXXVI.Finally,wedefineTypeIII,when very likely presence of fluorescence in the case of better Felinesarenotdetected. signal-to-noise.In addition,recombinationlines of Fe XXV and Fe XXVI, are always present in the set of γ Cassio- peaeanalogs.TheselinesareincludedintheXSPECmodel mekal. For most of the observations we have achieved a The general features observed in this work, for the different goodfitusingacombinationofmekalcomponents.Inafew groupsofHMXB,aresummarizedinTable2,andexplainedbe- cases we have used other components: diskbb and power- lowinmoredetails: law,butmekalisbyfarthemostemployedoneinγCassio- peaelikesystems,inagreementwithpreviousX-rayanalysis – BeXBs.Wehavecollecteddatafrom10sources.Alltheob- (LopesdeOliveiraetal.2010,2006). servationswereperformedinquiescence.We havedetected FeKαemissioninonlyoneBeXB(SAXJ2103.5+4545).The – HMGBs. We have collected data from two HMGBs: upperlimitoftheFeKαEWintherestofBeXBsisingen- LSI+61303andLS5039.NoneofthemshowFefeatures. eral higher than the observed value in SAX J2103.5+4545, However, the signal-to-noise in these observations is poor implyingthat the lack of detectionsmightbe dueto a poor and the upper limits of the FeKα EW are high enough to signal-to-noise. The spectra can be modeled by thermal or do notrule outthe presenceof the line. We haveused both acombinationofthermalandnon-thermalcomponents,ex- thermalandnon-thermalcomponentsinthefits. cept for Swift J045106.8-694803(fitted using an absorbed – Peculiars. Set of sources that do not accommodate in any powerlaw).Sevensourcesaccepta thermalmodel,and6a of the aforementioned classes of HMXBs, as explained in combinationmodel(4ofthemacceptboth). theintroduction.We havecollecteddataof3such systems: – SGXBs.Wehavegathereddatafrom12sources.Tenofthem 4U2206+54,CentaurusX-3andCygnusX-1: show detectableFe fluorescenceemission. The onlyexcep- – 4U 2206+54does not show any detectable Fe emission tionsareIGRJ16465-4507andSAXJ1802.7-2017,themost line,andtheupperlimitintheEWofFeKαislow(itis distantSGXBsincludedinthiswork,at12.5and12.4kpcre- comparabletotheupperlimitsintheBeXBs). Itcanbe spectively.TheEWupperlimitsinthesetwosourcesishigh, fitted by means of an hybrid model (thermal plus non- implyingthattheirfaintnessisverylikelythereasonwhywe thermalcomponents). Articlenumber,page5of30 A&Aproofs:manuscriptno.Gimenez-Garcia Model Continuummodels N powerlaw×tbnew×cabs 1 Non-thermal N powerlaw ×tbnew ×cabs +powerlaw ×tbnew ×cabs 2 1 1 1 2 2 2 N powerlaw ×tbnew ×cabs +powerlaw ×tbnew ×cabs +powerlaw ×tbnew ×cabs 3 1 1 1 2 2 2 3 3 3 T mekal×tbnew×cabs 1 T (mekal+mekal)×tbnew×cabs 2 T (mekal+mekal+mekal)×tbnew×cabs 3 T (cemekl)×tbnew×cabs 4 T bbody×tbnew×cabs 5 T (bbody+bbody)×tbnew×cabs 6 Thermal T (bbody +bbody )×tbnew×cabs 7 1 2 T bbody ×tbnew ×cabs +bbody ×tbnew ×cabs 8 1 1 1 2 2 2 T diskbb×tbnew×cabs 9 T (diskbb+bbody)×tbnew×cabs 10 T bremss×tbnew×cabs 11 T bbody×tbnew×cabs+bremss×tbnew×cabs 12 T (bremss+bbody)×tbnew×cabs 13 TN (powerlaw+bbody)×tbnew×cabs 1 TN powerlaw×tbnew ×cabs +bbody×tbnew ×cabs 2 1 1 2 2 Both TN (powerlaw+diskbb)×tbnew×cabs 3 TN powerlaw×tbnew ×cabs +diskbb×tbnew ×cabs 4 1 1 2 2 TN (powerlaw+mekal)×tbnew×cabs 5 Table 1: Listofmodelsusedtofitthecontinuum, describedinXSPECnotation.Thebasiccomponents arepowerlaw, bbody, diskbb, bremss, mekalandcemekl,togetherwithtbnewtoaccountfortheabsorptionandcabsforthenon-relativisticComptonscattering.Theemployedmodels areacombinationofthesecomponents, inadditiontoGaussianprofilesmodellingemissionlines.Wedividethemodelsinthreetypes: N for # non-thermal,T forthermalandTN formodelscontainingboththermalandnon-thermalcomponents. # # – Centaurus X-3 presents a rich emission lines spectrum. TheoriginofbroadFe featuresin X-rayBinariesisstill an Concretely, in the Fe complex we are able to identify opened question, but the most likely alternatives are related to FeKα,FeKβ,FeXXVandFeXXVI.Wehaveusedeither the presence of an accretion disk (see e.g. Hankeetal. (2009); anhybridmodeleitheranon-thermalmodel. Ngetal. (2010); Duroetal. (2011)). However, narrow features – CygnusX-1exhibitsabroadFefeature,sometimescom- arenotcompatiblewithmaterialrotatingathighvelocitiesorrel- binedwithafaintandnarrow,butstatisticallysignificant, ativisticallybroadened.GiventhatbroadandnarrowFeKαhave FeKαline.Wehavemostlyusednon-thermalmodels,oc- a clearly differentorigin, they must be analysed in a separated casionallycombinedwithabbodyordiskbbcomponent. way.Then,itraisesthequestionofhowtodefinetheseparation markbetween them.In Figure4 we can see thatin oursample – AX J1749.1-2733. In this system, the optical member has thenumberofdetectedsourcesdecreaseswhenincreasingσ . beenclassifiedasaB1-2(Karasevetal.2010),butthelumi- line Moreover,mostofthedetectionsaregroupedatσ <0.15keV. nosityclass remainsunknown,preventingusto incorporate line Hence,σ < 0.15keV seems a naturalthresholdforthe def- the source in a defined group. Although it does not exhibit line initionof narrowlinesin the sample.In addition,we mustpay detectableFeemission,thehighabsorptionclearlypointsto attentionto theplausiblecontaminationof FeKαwith Fe XXV, asupergiantcompanion.Inaddition,the EWupperlimitof whichislocatedat∼6.7keV.Thechosencriterionseparatesthe FeKαiscompatiblewiththevaluesobservedinSGXBsand sourceswhereitis veryunlikelythatFeKαis contaminatedby SFXTs.Itcanbewellfittedusinganabsorbedpowerlawor Fe XXV (narrowlines),fromthesourceswhichprobablysuffer ablackbody. fromthisissue(broadlines).Amoredetailedanalysisofbroad Insummary,itisverylikelythatFeKαisanubiquitousfea- Fe featuresin HMXBs is outof the scope of this paper and its tureinHMXBs,anditsdetectionhighlydependsonthequal- discussionwillbefullydevelopedinaforthcomingwork. ity of the observations. In this regard, the EW of the line Hereafter, when FeKαis mentioned,we refer to the narrow is very affected by the level of intrinsic absorption present feature.Fromthetotalnumberof108analysedobservationswe onthesources(seealsoSection3.4.3).SGXBsandSFXTs, havedetected(narrow)FeKαin60ofthem. whichshowhigherabsorption,tendtoexhibitamorepromi- nentFefluorescence. 3.3. Centroidenergy 3.2. FeKαwidth InFigure5wecanseeanhistogrampresentingthecentroiden- ergy of FeKα. A Gaussian fit of the data reveals a mean value In Table A.3 we show the parameters of every detected FeKα, for the centroid energy of 6.42 keV. There are not significant including the width of the line (σ ). We have made a dis- differencesintheaveragedvaluesobtainedneitherfordifferent line tinction between narrow and broad FeKα. We have defined classesofHMXBs,neitherfordifferentstates.Thestandardde- narrow FeKα when σ < 0.15 keV, and broad FeKα when viationis0.02keV,comparabletothetheerrorthatwetypically line σ >0.15keV.Thisseparationisbothphysicallyandobserva- obtain in the estimation of the centroid energy in the fits (see line tionallymotivated: Table A.3). Taking into accountthe standard deviation and the Articlenumber,page6of30 A.Giménez-García:Title Fe complex: Type I Fe complex: Type II Fe complex: Type III 15 1.5 −−normalized counts skeV11 150 −−normalized counts skeV11 0.51 −−normalized counts skeV11 000...246 0 0 0 1.2 1.2 1.05 1.1 1.1 ratio 1 ratio 1 ratio 1 0.9 0.9 0.95 0.8 0.8 6 7 6 7 6 7 Energy (keV) Energy (keV) Energy (keV) Fig.3:PatternsfoundintheFecomplexofHMXBs:TypeI(leftpanel),TypeII(centralpanel),andTypeIII(rightpanel). K e F w o r r a N Broad Fe features Fig.5:CentroidenergyofFeKα,withaGaussianfitoverplotted(blue Fig. 4: Histogram of theFeKαwidth. Thebulkof thedetections are profile).Themeanvalueis6.42±0.02keV,compatiblewithFeI-XVIII. grouped showing σ < 0.15 keV. Wedefinethem asnarrow FeKα. Eventhoughwehave60detectionsof FeKα,inthisplotweonlysee line TherestaredefinedasbroadFefeatures.Eventhoughwehavedetected 55.Fivecasesfulfiltherequirementsofdetection,butthelowsignal-to- 60 narrow FeKα, this plot only includes 38. The reason is that 22 of noisedonotpermittofindanaccuratecentroidenergy.Theyhavenot them are very narrow (or the signal-to-noise very low) and they have beenincludedinthisplot.Inthesefivecaseswehavesetthecentroid beentreatedinthefitsasdeltafunctions(inTableA.3wepresenttheir energyofFeKαto∼6.4keV. widthasσ =0). line 3.4. Correlatedparameters uncertainties in the CTI corrections in EPN6, the centroid en- One of the goals of this work is to study plausible correlations ergyofFeKαconstrainstheionizationstateofFetolessionized involvingtheparametersofFeKα(position,width,intensityand thanFeXVIII(Kallmanetal.2004),inagreementwithprevious EW,andotherparametersinthefits,suchastheabsorbingcol- studiesinHMXBs(Torrejónetal.2010b;Gottwaldetal.1995; umnandtheintensityofthecontinuum.Notethat,evenwhena Nagase1989).Inthisregard,thestudyofTorrejónetal.(2010b) goodfitis reached,the confidenceregionofa parametermight usingHETGS(moreaccurateinwavelengththanEPN),givesa beoccasionallydifficulttofindduetothedependenceofthepa- narrowerconstrainin the ionizationstate (Fe I-X). Our present rameterwithotherparametersofthemodel.Wespecifyineach worksupportthisresultaddingmoresourcestothesample. of the following subsections the number of cases where a suc- In the rightside of Figure 5 we can see seven FeKα detec- cessfulestimationofthe90%confidenceregionhasbeendone. tionsemergingoutoftheGaussianprofile.Fourofthemareun- likelytobedescribedbysucha Gaussianprofile,sincetheylie morethanthreetimesthestandarddeviationawayfromthemean 3.4.1. ContinuumFluxvsFeKαFlux energy.AllfourbelongtoCygnusX-1. InFigure6wehaverepresentedtheunabsorbedfluxofthecon- 6 Pleasefindmoreinformationaboutlong-termCTIcorrectioninthe tinuumbetween1-10keVcancellingFeKαemission(F1−10keV), againstthefluxofFeKα(F ).Wehavesuccessfullyfounda releasenoteEPIC-pnLong-TermCTI,byM.J.SSmithetal.(2014),at FeKα 90%confidenceregionofthefluxofFeKαin56cases. http://xmm2.esac.esa.int/docs/documents/CAL-SRN-0309-1-0.ps.gz; andEPICstatusofcalibrationanddataanalysisbyGuainazzi(2008), In a logarithmicscale, we identifytwo differentpatternsof at http://xmm2.esac.esa.int/external/xmm_sw_cal/calib/CAL-TN- correlation.First, forasubsetincludingalltheeclipseobserva- 0018.pdf. tionsandIGRJ16318-4848,wefindacorrelationwithPearson Articlenumber,page7of30 A&Aproofs:manuscriptno.Gimenez-Garcia Fig. 6: F1−10keV versus FFeKα.Bluedashed linemarksthecorrelation Fig. 7: EWof FeKαagainst L1−10keV. γ Cassiopeae analogs (circles) observed for IGR J16318-4848 jointly with eclipse observations, and lieat L1−10keV < 1033 erg/s.Opensymbolsindicatethateitherthedis- the red solid line follow the bulk of the observations. The color map tanceeithertheerrorintheestimationofthedistanceisunknown.The indicatestheσ oftheline(definedinSection2.2). solidlinecorrespondstoalinearfitinlogarithmicscaleofthefilleddia- sign monds,thatis,thesourceswithavailabledistancewitherrorestimation andL1−10keV >1033erg/s. Coefficient(PC)of0.98(bluedashedline).Second,fortherest of observations, we find a correlation with PC=0.89 (red solid line). Baldwin (1977) observed an inverse correlation in the EW Alinearfitoftheparameters(inlogarithmicscale)intheout- of CIV and the UV luminosity in AGNs. Analogously,the de- of-eclipseobservations(redlineinFigure6),givesthefollowing creaseoftheEWofFeKαwhenincreasingtheX-rayluminos- dependence: ityiscalledX-raysBaldwineffect.Thedependencethatweob- serve is compatible within the error with the one observed by F1−10keV(erg/s/cm2)=FF1.e0K0±α0.08(erg/s/cm2)×102.18±0.87 (1) Terogryrerjaónngeet:aEl.W(2∝01L0−b0).29in X.-ray Binaries using a narrower en- 1.6−2.5Å Theerrorsshowthestandarddeviationoftheparametersinthe linearfit. 3.4.2. FeKαWidthvsCentroidEnergy The observeddivergenceamongeclipse (plusIGR J16318- 4848)andout-of-eclipseobservationssuggeststhatthecompan- In Figure 8 we present the centroid energy of this feature ver- ion star blocks in a differentproportionthe continuumand the sus its width (σ ). We have successfully found a 90% confi- line FeKαemission.Therefore,animportantcontributionofthefluo- dence region of σ in 20 cases. We can see a moderate cor- line rescenceemissionisproducedinanextendedregionofR&R . relation (black line in Figure 8, PC=0.55), indicating a pos- ⋆ This is consistent with previous analysis of eclipse observa- sible blending of lines. Two observations (uppermost side of tionsofHMXBs(e.g.Rodes-Rocaetal.(2011)andAudleyetal. Fig. 8) do not follow the correlation. They correspond to ob- (2006)).Inparticular,Audleyetal.(2006)estimatesthat20%of servationsof4U1700-37(Obs.ID0083280201)andEXO1722- FeKαinOAO1657-415isemittedfrom19light-secondsoffthe 363 (Obs.ID 0405640201)where the Fe complex is hardly re- X-raysource. solved, and therefore it is very likely that a contribution of We have also the luminosity of the continuum, in order to Fe XXV and Fe XXVI in the model of FeKα is increasing the faceitwiththeEWofFeKα.Fortheflux-to-luminosityconver- measuredwidthoftheFeKαline. sionwe haveusedthe estimationsofthedistanceshowninTa- Coloured squares correspond to the expected width from ble A.1. We haveexcludedeclipse and IGRJ16318-4848from the contribution of three different broadening phenomena:line thisanalysis,giventhattheEWisstronglyaffectedbythehigh blending,DopplershiftsandComptonbroadening.Adiscussion obscurationofthecontinuumthattheysufferfrom.InFigure7 onthedifferentbroadeningcontributionsisgiveninSection4. we plot the EW of FeKα against the unabsorbed luminosity of the continuumbetween1-10keV cancellingFeKα emission 3.4.3. Curveofgrowth (L1−10keV).Weobservetwodifferentgroupsofsources:1)γCas- siopeae analogslie at low luminosities(L1−10keV < 1033 erg/s); InFigure9weshow,foroutofeclipseobservations,theNH ver- 2)therestofsourceswhichexhibitFeKα.γCassiopeaeanalogs sus the EW of FeKα (what is generally known as the curve of don’t show any evident correlation (they are very few points), growth). We have successfully found a 90% confidence region whiletherestpresentamoderateinversecorrelation(PC=-0.25, of both N and EW in 46 cases. We want to take into account H and PC=-0.39 using only the sources with an available estima- observationswereN reflectstheintrinsicabsorptionofthesys- H tion of distance with error, marked as filled diamonds in Fig- tem. Hence, we have set N > 2 as a condition to safely ex- H ure 7). A linear fit using the filled diamondsin Figure 7 gives: ceed the typical N of the interstellar medium for the sources H here studied (checked using the online application following EW = L−0.52±0.27(erg/s)×1017.45±9.83 (2) Willingaleetal. (2013)). The use of this criterion excludes the 1−10keV Articlenumber,page8of30 A.Giménez-García:Title Fig. 8: Width of FeKα (σ ) versus the centroid energy (black Fig. 9: Curve of growth observed for FeKα, that is, EW against line squares). Black solid line traces a linear fit. We have marked in N . The turquoise band marks the expected correlation using nu- H colour the expected width from the contribution of three differ- merical simulations. The sources are identified by different symbols ent broadening phenomena: line blending, Compton broadening and when more than one observation is included: 4U 1700-37 (open cir- Doppler shifts, considering velocities of V(km/s) = 1000 (red) and cle), 4U 1907+09 (open upward triangle), Cygnus X-1 (open down- V(km/s)=2000(green).Everysingleblacksquarehasassociatedasin- wardtriangle),EXO1722-363(opendiamond),IGRJ16318-4848(open gleredsquareandasinglegreensquarecorrespondingtotheexpected square) and IGR J16320-4751 (plus). Only one observation for Cen- valuesofσ forthatobservation(seeSection4). taurus X-3, GX 301-2, Vela X-1 and XTE J0421+560 (all four a star line symbol). BeXBSAXJ2103.5+4545,theγCassiopeaeanalogs:γCassio- peae and HD 110432;andthe SFXT IGRJ11215-5952.More- over,eclipseobservationsshowhigherEWandtheyarenotcom- parabletoout-of-eclipseobservations.Therefore,eclipseobser- vationshavenotbeenplottedin Figure9. Asa consequenceof the chosen criteria, we end up with a set of 36 observations, whereallthedonorsaresupergiants. N and the EW of FeKα are expected to correlate in H HMXBs,asshownbyTorrejónetal.(2010b),sincethespectral linesareusuallystrongerwhentheopticaldepthincreases.Our sampleconfirmsthese expectations,showinga notablecorrela- tion(PC=0.85). We have determined the theoretical curve of growth using numericalsimulations.InthissimulationsthereisaninputofX- rayradiationwithapowerlawprofile,thatistransmittedthrough acloudofsphericallydistributedneutralmatter(Eikmann2012). We have taken into accountthe power law index (Γ) in the simulations, since steeper profiles entail less photons available abovetheFeK-shellthresholdenergy,decreasingtheEW.That is,forthesameN ,thehigherisΓ,theloweristheEWofFeKα. H InFigure9theturquoisebandtracestheresultsfromthesimula- tionswithΓ ∈ [0.5,2],whichisthetypicalrangewherewefind Fig.10:HistogramsshowingacomparisonoftheNH valuesobserved Γinourfits. inSGXBs(filledred)andSFXTs(emptyblue). 3.5. N :SGXBsandSFXTs H In Figure 10 we have plotted histogramsfor the N valuesob- have10N valuesforSGXBsand9forSFXTs.Wehavemerged H H servedinSGXBs(red)andSFXTs(blue).Inthecaseswherewe theminasetof19elements,andconsideredeverypossiblecom- have more than one observation for the same source, we have binationof2groupsof10and9elements(92378possibilities). averagedthevaluesinordertoobtainoneN representativefor We have compared the median of the two subsets, and calcu- H eachsystem.TheorbitalphasecriticallyaffectstheobservedN , latedtheabsolutedifference.99.7%ofthecaseshaveproduceda H and therefore ingress/egress and eclipse phases have not been lowerabsolutedifferencethantheobservedone.Usingthemean takenintoaccount. insteadofthemedianthepercentageisalsoveryhigh(98.8%). We find that SGXBs are in general more absorbed than In conclusion,it is very likely that the discrepanciesin the ob- SFXTs.Wehaveperformedapermutationtestinordertoquan- servedN valuesforSGXBsandSFXTsareproducedbyphys- H tify if the observed disparity in the N is a random effect. We icalreasonsratherthantheyarisebychance. H Articlenumber,page9of30 A&Aproofs:manuscriptno.Gimenez-Garcia Fig.11:EWofFeKαagainstN ,inIGRJ16320-4751.Theturquoise Fig. 12: Orbital modulation of N in the system IGR J16320-4751. H H bandmarksthenumericalsimulationsresults,withΓ∈[0.5,2]. Solidlinescorrespondtotheexpectedabsorptionfromasmoothwind and a non eccentric orbit, assuming an orbital separation a = 1.6R , ⋆ R⋆ = 20R⊙, M˙ = 10−5M⊙/yr,3∞ = 700km/s, β = 0.8;anddifferent 3.6. Individualsourcesanalysis:IGRJ16320-4751and orbitalinclinationsi=0,π/10, π/6rad(red,greenandblue). 4U1700-37 3.6.1. IGRJ16320-4751 ables, since they are much more commonly used than M˙/3∞ in the literature. This way, we constrain our parameters to the IGRJ16320-4751wasdetectedbyASCAin1994and1997(cor- observedand predicted range of values in O supergiants: M˙ = responding to AX J1631.9-4752), and by INTEGRAL in 2003 10−7−10−5M⊙/yrand3∞ =500−3000km/s(Kudritzki&Puls (Tomsicketal. 2003). It is a HMXB composed by an O8I op- 2000;Vinketal.2001). tical star and a neutron star (Rahouietal. 2008). It shows a Fora nullorbitalinclination,weobviouslyobtaina flat N modulation of 8.96 days, that it is considered its orbital pe- H riod (Corbetetal. 2005), and a pulsation period of ∼1300s modulation(redlineinFig.12),thatdescribestheobservedNH (Lutovinovetal. 2005). The ESA archives permits to collect (except at φ = 0), assuming a = 1.6R⋆, R⋆ = 20R⊙, M˙ = eleven observations of IGR J16320-4751,enabling us to study 10−5M⊙/yr, 3∞ = 700 km/s and β = 0.8. Graduallyincreasing the orbitalinclination (i = π/10, π/6 rad; green and blue lines in more detail the curve of growth, as well as tracking the ab- respectively), we are able to describe a high N at φ = 0, but sorptionvariationalongtheorbitalphase. H loosingsimilaritiesaroundotherorbitalphases.Anyhow,given In Fig. 11 we can see the curve of growth, as shown in thesimplicityofthemodel,theobtainedparametersarecertainly Fig. 9, restricted to IGR J16320-4751. We clearly see the de- justindicatives. pendencebetween N and EW of FeKα, as stated for the bulk H of the sources in Section 3.4.3, and the general trend follow- ing the numericalsimulations marked with the turquoise band. 3.6.2. 4U1700-37 However,the agreementwiththe simulationsisnotcompletely fulfilled, given that the spectral fits of IGR J16320-4751 have 4U 1700-37 was detected for the first time by Uhuru in 1970 broughtout a power law index Γ ∼ 0.5 (see Table A.2). Since (Jonesetal.1973).TheopticalstarisHD153919,aO6.5Iaflo- fromalowerpowerlawindexweexpectmoreEWofFeKα,the catedatadistanceof1.9kpc(Ankayetal.2001).Theorbitalpe- pointsforIGRJ16320-4751areexpectedtobelocatedintheup- riodis3.41days.SinceX-raypulsationshavenotbeendetected peredgeoftheturquoiseband,correspondingtothesimulations so far,the compactobjectcan be either a neutronstar or either with Γ = 0.5.We considerthatthe generaltrendiscorrect,but ablackhole.ThedatabaseofESAcontainsfiveobservationsof therearestillsomeuncertaintiesinthefitsand/orthetheoretical 4U1700-37,thatwesplitinninespectrainordertodistinguish hypothesis(sphericalgeometryandneutralmatter). differentstatesofthesource. From 14th Augustto 17th September,in 2008,a campaign InFigure13wecanseethecurveofgrowth,for4U1700-37. consisting of nine observations of IGR J16320-4751 was per- AlthoughsevenoftheninespectrashowFeKα,wewereableto formedbyXMM-Newton.Wehaveusedthissetofdatatoplot constraintheboundariesofN inonlyfiveoftheanalysis.One H theN modulationdependingontheorbitalphase(Fig.12).We of them correspondsto an eclipse observation (filled circle). It H havesetφ=0attheN maximum.Wehavealsocalculatedthe showsmuchmoreEWbecausethecontinuumfluxisblockedby H theoreticalabsorptionexpectedfromasmoothwindinanonec- theopticalstar,whereasFeKαcomesfromamoreextendedre- centric orbitusing a β velocitylaw (Castoretal. 1975) and the gionthatisnotcompletelyhiddenduringeclipse.Wedonotsee motionequation,takingintoaccountvariationsintheorbitalin- anobviousdependencebetweenN andEW,althoughthepoints H clinationi,orbitalseparationa,starradiusR ,masslossrateM˙, lieinaregionclosetotheexpectedvalues(turquoiseband).Any- ⋆ parameterβandtheterminalvelocityofthewind3∞. way, a set of four observations (excluding the eclipse), is too Indeed, M˙ and 3∞ cannot be disentangled in this simple smalltoperformanstatisticalanalysis. model, and the actual parameter used is M˙/3∞. However, here- From 17th to 20th of February 2001, 4U 1700-37 was ob- after we give values of M˙ and 3∞ as if they were free vari- servedbyXMM-Newtonfourtimes,inacampaigncoveringdif- Articlenumber,page10of30

ebook img

An XMM-Newton view of FeK{\alpha} in HMXBs PDF

3.3 MB·English
by  A. Giménez-García
#journals #arxiv
Save to my drive
Quick download
Download
Upgrade Premium
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview An XMM-Newton view of FeK{\alpha} in HMXBs

Astronomy&Astrophysicsmanuscriptno.Gimenez-Garcia (cid:13)cESO2015 January21,2015 α An XMM-Newton view of FeK in HMXBs A.Giménez-García1,4,6,J.M.Torrejón1,2,W.Eikmann3,S.Martínez-Núñez2,L.M.Oskinova4,J.J.Rodes-Roca1,2,5, andG.Bernabéu1,2 1 UniversityInstituteofPhysicsAppliedtoSciencesandTechnologies,UniversityofAlicante,P.O.Box99,E03080Alicante,Spain e-mail:[email protected] 2 X-rayAstronomyGroup.DepartamentodeFísica,IngenieríadeSistemasyTeoríadelaSeñal,UniversityofAlicante,P.O.Box 99,E03080Alicante,Spain 3 Dr.KarlRemeis-SternwarteFAUErlangen-Nürnberg,D96049Bamberg,Germany 5 4 InstituteforPhysicsandAstronomy,UniversityofPotsdam,D-14476Potsdam,Germany 1 5 MAXIteam,InstituteofPhysicalandChemicalResearch(RIKEN),2-1Hirosawa,Wako,Saitama,351-0198,Japan 0 6 SchoolofPhysics,FacultyofScience,MonashUniversity,Clayton,VIC3800,Australia 2 n Accepteddate:25thDecember2014 a J ABSTRACT 0 2 We present a comprehensive analysis of the whole sample of available XMM-Newton observations of High Mass X-ray Binaries (HMXBs)untilAugust,2013,focusingontheFeKαemissionline.Thislineisakeytooltobetterunderstandthephysicalproperties ] ofthematerialsurroundingtheX-raysourcewithinafewstellarradii(thecircumstellarmedium).Wehavecollectedobservations E from46HMXBs,detectingFeKαin21ofthem.WehaveusedthestandardclassificationofHMXBstodividethesampleindifferent H groups. We find that: (1) different classes of HMXBs display different qualitative behaviour in the FeKα spectral region. This is . speciallyvisibleinSGXBs(showingubiquitousFefluorescencebutnotrecombinationFelines),andγCassanalogs(showingboth h fluorescentandrecombinationFelines).(2)FeKαiscentredatameanvalueof6.42keV.Consideringtheinstrumentalandfitsuncer- p tainties,thisvalueiscompatiblewithionizationstateslowerthanFeXVIII.(3)Thefluxofthecontinuumiswellcorrelatedwiththe - o fluxoftheline,asexpected.EclipseobservationsshowthattheFefluorescenceemissioncomesfromanextendedregionsurrounding r theX-raysource.(4)WeobserveaninversecorrelationbetweentheX-rayluminosityandtheequivalentwidthofFeKα(EW).This t phenomenon is known as X-rays Baldwin effect. (5) FeKα is narrow (σ < 0.15 keV), reflecting that the reprocessing material s line a doesnotmoveathighspeeds.Weattempttoexplainthebroadnessofthelineintermsofthreepossiblebroadeningphenomena:line [ blending,ComptonscatteringandDopplershifts(withvelocitiesofthereprocessingmaterialV ∼1000km/s).(6)Theequivalenthy- drogencolumn(N )directlycorrelateswiththeEWofFeKα,displayingclearsimilaritiestonumericalsimulations.Ithighlightsthe H 2 stronglinkbetweentheabsorbingandthefluorescentmatter.(7)TheobservedN inSupergiantX-rayBinaries(SGXBs)isingeneral H v higherthaninSupergiantFastX-rayTransients(SFXTs).Wesuggesttwopossibleexplanations:differentorbitalconfigurations,or 6 adifferentinteractioncompactobject-wind.(8)Finally,weanalysedinmoredetailthesourcesIGRJ16320-4751and4U1700-37, 3 covering several orbital phases. The observed variation of N between phases is compatible with the absorption produced by the H 6 windoftheiropticalcompanions.Theobtainedresultsclearlypointtoaveryimportantcontributionofthedonor’swindintheFeKα 3 emissionandtheabsorptionwhenthedonorisasupergiantmassivestar. 0 . Keywords. Stars:binaries,circumstellarmatter-X-rays:binaries-Surveys 1 0 5 11. Introduction studyoftheFecomplex,andmotivatesacomprehensiveanaly- : sisinHMXBs. vSincetheearlystagesofX-rayastronomy,Felinesinthespec- In particular, FeKα has been proven as a fundamental Xitral regionof ∼6-7 keV (the Fe complex) have been studied in tool in the study of HMXBs (Martínez-Núñezetal. 2014; a largenumberof X-ray sourcesgivenits fruitfulnessas a tool Rodes-Rocaetal.2011;vanderMeeretal.2005).Theoriginof r aforplasma diagnostics.Theywerereportedforthe first timein the fluorescence emitting region has been discussed by many thesupernovaremnantCasA(Serlemitsosetal.1973),andonly authors in the past. Nagase (1989) considered accretion disks two years later in a High Mass X-ray Binary (HMXB) using and the matter stagnated in the accretion and ionization wakes the Ariel 5 satellite (Sanfordetal. 1975). The most recent X- in the stellar wind as plausible areas of FeKα production. rayspacemissions(Swift,Suzaku,ChandraandXMM-Newton) Watanabeetal. (2006) analysedthe classicalHMXB Vela X-1, have triggered a notable improvement in the attainable spec- andproposedtheextendedstellarwind,reflectionoffthestellar tral resolution and effective area, permitting to distinguish be- photosphereandanaccretionwake,asthemostlikelycandidates tweendifferentemissionfeaturesintheFecomplex:narrowand forfluorescencereprocessingregions.Inanycase,FeKαisvery broadfluorescencelines(FeKαandFeKβ),Comptonshoulders sensitive to the physicalconditionsof the vicinity of the X-ray andrecombinationlines(Fe XXV andFe XXVI)(Torrejónetal. source,andprovidesremarkableinformationthatmustbeanal- 2010b).Thisimprovementhasgivenaremarkableimpetusinthe ysed. Fluorescenceis producedas a consequenceofthe X-ray il- lumination of matter. When an Fe atom absorbs a photon car- Sendoffprintrequeststo:A.Giménez-García rying sufficient energy to remove an electron from its K-shell Articlenumber,page1of30 A&Aproofs:manuscriptno.Gimenez-Garcia (E>7.2 keV), the vacancy can be occupied by another electron (clumps) are present as an intrinsic feature of the radiatively froman outershell. If the electroncomesfromthe L-shell, the driven winds of hot stars (Lucy&White 1980; Oskinovaetal. transitionproducesFeKαemission.FeKβemissionisproduced 2012). Second, hydrodynamical simulations show that the X- when the vacancyis filled by a formerM-shell electron.When rayradiationandthegravityfieldofthecompactobjectdisturb Fe ismoreionizedthanFe XIX,the fluorescenceyieldstartsto the wind of the donor, inducing the formation of denser struc- decreasewith theionizationstate (Kallmanetal.2004). There- tureslikefilaments,bowshocksandwakes(Blondinetal.1990, fore,FeKαisafootprintofnotextremelyionizedFe (lessthan 1991). Fe XX). On the other hand, recombination lines Fe XXV and Inthelastdecadeandahalf,newdiscoverieshaveledtothe Fe XXVI unveil the presence of very hot gas, where Fe atoms additionofnewgroupstothepreviouspictureofHMXBs,stress- arealmostcompletelystripped. ing the value of grasping the different features of the sources Previous comprehensive surveys of the Fe complex in such as geometry,compactobjectproperties,opticalstar pecu- HMXBs were carried out by Gottwaldetal. (1995) using EX- liarities and wind clumpiness. The new groups are Supergiant OSATandTorrejónetal.(2010b)usingtheHighEnergyTrans- Fast X-ray Transient systems (SFXTs), γ Cassiopeae analogs mission Gratings (HETGS) on board Chandra. The high spec- andγ-rayBinaries. tralresolutionprovidedbyChandragratingsprovedtobeinstru- SFXTs are systems with a supergiant optical star, as in mental in disentangling the differentionization species present SGXBs,butdefinedbyanextremelytransientbehaviour.During in the Fe complex. However, the relatively low throughput of quiescencetheyexhibitlow luminosity(∼ 1032 erg/s),butthey the instrument allowed to study only the brightest binaries. In spend most of their time in an intermediate level of emission this study we increase significantly the previoussample by us- (∼ 1033−34 erg/s). They display short outbursts (∼few hours), ingthehighthroughputofXMM-NewtonEPN.Thishasallowed reachingluminosities up to 1036−37 erg/s (Sidolietal. 2009). It ustoincludefaintersystems(likeBeX-raybinaries(BeXBs)or islikelythattheclumpinessofthewindplaysamainroleinthe SFXTsinquiescence)whilethemoderateresolutionoftheEPN variability of these sources. Other mechanisms involving cen- CCDs has allowed us to test previous correlations based on a trifugalandmagneticbarrierscouldenhancetheobservedlumi- smallsample. nosity swings, relaxing the needed variation amplitudes in the HMXBs are specially susceptible to be studied using the physical conditions of the wind (Bozzoetal. 2008). Neverthe- Fe complex,onaccountofthesignificanceofthe circumstellar less,otherauthorsexplainthevariabilityappealingtothequasi- medium in the observable phenomena. These systems consists spherical accretion model (Draveetal. 2013; Paizis&Sidoli of a compact object, either a neutronstar (NS) or a black hole 2014). (BH), accreting matter from a massive OB star (usually called γCassiopeaeanalogsarecharacterizedbythethermalnature optical or normal star of the system). In HMXBs the observed of the X-ray emission, with plasma temperatures of ∼ 108 K luminosity is commonly powered via accretion. Consequently, (∼ 10keV),anX-rayluminosityof1032−33 erg/s,andhighflux the way that matter is accreted from the donor directly defines variabilityonvarioustimescales.However,theydonotdisplay theobservableluminosityfeaturesofeverysource. giant outbursts as observed in BeXBs (LopesdeOliveiraetal. WhentheopticalstarisaBestar,thesystemisaBeXB.Be 2010).Presently,itisnotclearthattheX-rayemissionisemitted stars are fast-rotating BIII-V stars, which have shown spectral byaccretionprocesses(ontoaneutronstarorawhitedwarf),or emission lines at some point of their lives. They also show an alternativelygeneratedfromtheinteractionbetweenthesurface excessofinfraredemission,whentheyarecomparedtononBe ofthestar,thecircumstellardiskanditsmagneticfield. starsofthesamespectraltype.Theseobservablesareexplained High Mass γ-ray Binary systems (HMGBs) are HMXBs appealing to an extended circumstellar decretion disk. BeXBs wheretheemissionpeaksabove1MeV.Nowadays,itisthought are usually transient in the X-rays, although some systems ex- thattheemissioniscausedbyacceleratedparticlesintheshock hibit a persistent quiescence emission (L≤ 1034−35 erg/s). The thatisproducedwhenthepulsarwindcollidesthemassivestar outburstshavebeentraditionallyclassified intwotypes.TypeI wind. Therefore, they are powered by the rotational energy of outburst(L≤1037erg/s)arerelatedtoperiastronpassages.Type theneutronstar,inoppositiontotherestofHMXBs,whichare IIoutburstarenotrelatedtotheorbitalphaseandimplyaneven accretionfed.TherearecurrentlyfiveconfirmedHMGBs,allof higherincreaseinluminositythanTypeIoutbursts,reachingthe themwithamainsequenceopticalstar(forareviewonHMGBs Eddingtonluminosity(forareviewonBeXBsseeReig(2011)). seeDubus(2013)). InthecaseofclassicalSupergiantX-rayBinaries(SGXBs), Finally,therearesourceswhich,followinganumberofrea- thecompactobjectisembeddedinthedenseandpowerfulwind sons,cannotbeclassifiedinanyofthealreadymentionedclasses of a OB supergiantcompanion,swallowing everythingthat en- ofHMXBs.Particularly,amongthesetofsourcesstudiedinthis ters its gravitational domain. The mass loss rate of the donor paper, they are 4U 2206+54, Centaurus X-3 and Cygnus X-1. is of the order of & 10−7M⊙yr−1, and the compact object is The optical star in 4U 2206+54 is a O9.5V (Blayetal. 2006), usually found at a close distance of a ∼ 1.5 − 2R . In such neither a supergiant nor a Be star. The system may be part of ⋆ a close orbit, the captured matter is able to fuel a persistent a new groupof wind-fedHMXBs with a main sequencedonor X-ray emission of ∼ 1033−39 erg/s. Flares and off-states are (Ribóetal.2006).CentaurusX-3andCygnusX-1aretheonly often observed in SGXBs, indicating an abrupt transition in systems here collected where accretion is persistently driven the accretion rate. They might be produced either by sudden by an accretion disk (Tjemkesetal. 1986; Shapiroetal. 1976), variations of density in the medium transited by the compact whatisreflectedinthespectraofbothsources. object (Martínez-Núñezetal. 2014; Kreykenbohmetal. 2008), In thispaper,we study the FeKαline forthe wholesample or either by instabilities above the magnetosphere of the neu- ofHMXBsavailablewithXMM-NewtonuntilAugust,2013.In tronstar,asproposedinthequasi-sphericalaccretiontheoryby Section2wepresentthesetofobservations,thereductionpro- Shakuraetal.(2012). cessandthemoreimportantdetailsconcerningthespectralfits. Themediumtransitedbythecompactobjectthroughtheex- In Section 3 we show the obtainedresults: an spectralatlas in- tended atmosphere of an OB supergiant star is not smooth be- cluding every fit and different plots relating fit parameters. In causeof,atleast,twophenomena.First,densityinhomogeneities Section 4 we interpret the obtained results and we summarize Articlenumber,page2of30 A.Giménez-García:Title themostimportantconclusionsinSection5.InAppendixAwe presenta set of tables describingthe obtainedparametersfrom thespectralfits.InAppendixBweshowthespectralatlas,con- tainingtheplotofeveryspectrumthatwehaveanalysedinthis survey.Weshowtheobservationsandthemodels,togetherwith theratiobetweenthem. 2. Observationsanddatatreatment TheXMM-Newtonobservatory(Lumbetal.2012)isfittedwith three X-ray telescopes of 1500 cm2 and a coalignated optical telescope.Spectroscopyandphotometryisdonebythe6instru- mentsonboard:threeX-rayimagingcamerasEPIC(European PhotonImagingCamera),twogratingXrayspectrometersRGS (ReflectionGratingSpectrometer)andanopticalmonitor(OM). EPIC cameras(0.1-15keV) are the only instrumentsat XMM- Newton covering the energy range of the Fe complex. Among EPIC, one camera uses PN CCDs, andthe other two use MOS CCDs. EPIC PN cameras(EPN) surpass by a factor ∼3 the ef- fectiveareaoftheMOScamerasat6-7keV,makingEPNmore Fig. 1: Light curve of the observation of 4U 1538-522 (Ob- suitableforourpurposes.Comparedtoothermissions,theHigh sID:0152780201). Wehavesplit theobservation intwoparts, onefor theingressintheeclipse,andanotheronefortheeclipse. EnergyTransmissionGratingSpectrometer(HETGS)onboard ofChandraprovidesbetterenergyresolutionintheenergyrange of the Fe complex,butthe effectiveareaavailable with EPN is observation2.ThesourcesnotincludedintheLiucatalogue,but significantly higher. EPN provides the adequate conditions to here considered, are: HD 119682, SS 397, IGR J16328-4726, perform the study here presented, on account of the moderate (butsufficient)spectralresolution(∆E/E ∼40),andgreateffec- HD45314,HD157832,SwiftJ045106.8-694803,IGRJ16207- 5129andXTEJ1743-363. tivearea(∼1000cm2),enablingustoanalyzealargeamountof sourcesinanhomogeneousandconsistentway. SinceHMXBsareusuallyvariable,weoftenobserveinthe 2.1. Datareduction same observation a dramatic change in luminosity, remarkably affecting the spectral parameters. In these cases, an averaged WehavereducedthedatausingScienceAnalysisSystem(SAS), spectrumdoesnotreproducethe actualemissionof the source, version12.0.1.Sincethesampleofobservationscontainanhet- anditisadvisabletosplittheobservationinmorethanonetime erogenous group of HMXBs, we found different observation interval.Wehaveconsideredfivedifferentstates1ofthesystems modes (timing and imaging) to account for the different prop- in order to define the time intervals: dips, quiescence, flares, erties of the sources. In the brightestsystems, the observations eclipse ingress/egress, and eclipse. We have used the follow- wereusuallyperformedusingthetimingmode,whilethefaintest ingcriteria.Whenluminositydropsafactor&2inthetimescale sourceswereobservedusingimagingmodes. of .1 hour, we have tagged the time interval as a dip. Analo- Timing modes permit to process the arrival of photons at gously,whenluminosityrises&2inthetimescaleof.1hour,we high rate, since only one CCD operates and the information is havelabelledthetimeintervalasaflare.Forobservationscover- collapsedintoonedimension,allowingafastreadout.Thetime ingeclipsingphases,wehavedefinedtimeintervalsforeclipse resolutionisashighas30µs (7 µs inburstmode,Kirschetal. ingress/egressandeclipse.Therestoftimeintervalsaretagged (2006)). Even though the high timing resolution reached with asquiescentstates. theseobservationmodes,pile-upisstillpresentinseveralcases, In Figure 1 we see the light curve of an observation of specially when the count rate is & 800 counts s−1. We have 4U 1538-522,as an exampleof how we havesplit the time in- checkedineveryobservationifpile-upisaffectingthedata,us- tervalsintheobservations.Thesourcewasobservedduringthe ingthe SAS task epatplot,andwe have excisedthe coreof the ingress in an X-ray eclipse, which is clearly noticeable in the source’spointspreadfunctioninthepertinentcases.Thesizeof lightcurve.Wehaveseparatedtheobservationintwotimeinter- theexcisedregionhasbeenchosenwide enoughto removethe vals,onecoveringtheingressineclipse,andanotheronecover- unwantedpile-upeffects(seeexamplesoftheuseofepatplotin ingtheeclipse. Ngetal.(2010)). Summarizingthesampleofobservations,wehavecollected Background subtraction process is also dependent on the data from 46 HMXBs. 21 of them exhibit FeKα emission. We brightness of the source. In the EPN timing mode, the PSF of note that some sources have more than one available observa- the sources displaying & 200 counts s−1 will span across the tion.Takingeverythingintoaccount(46sources,temporalsplit- whole CCD. Therefore, any area selected as a background re- ting depending on the state of the source, and more than one gion will be contaminated by source photons. Since this effect observation per source in some cases), we end up with a total isstronglyenergydependent,forthebrightestsourceswe have numberof108spectrathatwehaveanalysed. chosen a method of background subtraction similar to the one WehavefollowedthecatalogueofLiuetal.(2006),inaddi- performedintheanalysisofVelaX-1byMartínez-Núñezetal. tiontolaterdiscoveriesorconfirmations,toidentifythecurrently (2014),whereablankskyspectrumtakenintimingmodeisused knownHMXBs,andusedeveryavailableXMM-Newtonpublic astherealbackgroundforenergiesbelow2.5keV,whiletherest ofthespectrumcorrespondstotheoutermostpixelsoftheCCD. 1 Thestatesconsideredinthisworkandthealsocalledstatesinblack holebinarysystemsmustnotbemislead. 2 http://xmm.esac.esa.int/xsa/ Articlenumber,page3of30 A&Aproofs:manuscriptno.Gimenez-Garcia are thermal, we classify the modelas thermal (analogouslyfor non-thermal).Wehavealsousedhybridmodels,combiningther- malandnon-thermalcomponents.Thethermalcomponentsused inthisworkarethefollowing: – bbody:blackbodyemission. – diskbb:modelofanaccretiondiskemissionmadeofmultiple blackbodycomponents. – bremss: thermal bremsstrahlung emission (electrons dis- tributedaccordingtotheMaxwell–Boltzmanndistribution). – mekal:emissionfromopticallythinhotgas,includingspec- trallinesfromseveralelements(Meweetal.1985). – cemekl: built from the mekal model, incorporating multi- temperatureemission. Ontheotherhand,theonlynon-thermalcomponentusedin thisworkis: – powerlaw: phenomenologicalmodel consisting of a simple Fig. 2: Number of accepted models depending on the Reduced-χ2 inverse power law profile (∝ E−Γ). This profile is a foot- value. print of Inverse-Compton scattering by hot electrons (non- thermallydistributed)ofaseedradiationfield. Meanwhile,forcommonobservations,wehaveusedsource-free For the photoelectric absorption, we have used tbnew4, the regions to extract a background spectrum and subtract it from improvedversionoftheWilmsetal.(2000)modeltbabs,setting theformersourceplusbackgroundenergydistribution. thecrosssectionstotheVerneretal.(1996)onesandtheabun- Ancillaryresponse files were generatedusing the SAS task dancesaccordingtoWilmsetal.(2000).Themostimportantpa- arfgen.Forobservationstakenintimingmodeaffectedbypile- rameter of this model is the total equivalent hydrogen column up,wehavefollowedtherecommendationsoftheXMM-Newton N , which is the integrated amount of hydrogen atoms in the H SAS User Guide in order to generate the appropriate ancillary lineofsightfromthe observerto the source,percm2. We have response files. Response matrices were created using the SAS also added the modelcabsto accountforthe Comptonscatter- taskrmfgen. ing,whichisnotcomprisedinthetbnewmodelandisespecially significativeforN &1024 cm−2. H The emission lines are fitted using Gaussian profiles. We 2.2. Spectralfitting havecategorizedasFeKα anyemission linethatfulfilsthe fol- ForthespectralanalysiswehaveusedXSPEC,version12.8.03. lowingconditions: We have rebinnedthe spectra to have a minimum of 20 counts 1) ThecentroidenergyoftheGaussiancomponentliesinthein- perbin,andabinsizeofatleast1/3oftheFWHMoftheintrinsic terval[6.3,6.65]keV.Theintervalincludestheexpecteden- energyresolution,inordertobeallowedtoapplyχ2statisticsin ergyofFeKαemissionfromFeII(∼6.395keV)toFeXXIII thefittingofasetofpoissoniandata(Cash1979). (∼6.63keV)(Kallmanetal.2004).Thisconditionexcludes In Table 1 we present the sample of models employed for the detectionof anyhypotheticalfluorescentemissionfrom the continuumin the fits. Everymodel is a combinationof ad- FeXXIV-XXVat∼6.67−6.7keV,therebyexcludinganycon- ditive and multiplicative models. An additive model stands for fusionbetweenFeKαandtherecombinationlineFeXXV at asourceofX-rays(e.g.bremsstrahlungradiation),andamulti- similarwavelength.ThefluorescenceyieldsofFeXXIV-XXV plicativemodelrepresentsaenergy-dependentchangeofanad- arelowcomparedtolowerionizationstates. ditivemodel(e.g.photoelectricabsorption). 2) The statistical significance (σ ) of the Gaussian com- The models presented in Table 1 have been tested in every sign ponent is greater than 2σ. We have calculated σ from owbitshernvatthioenn,uamndbearcocfepbtiendsdaenpdenmditnhge nounmthbeerRoefdfiuctteedd-χp2ar(anmχ−2me-, χ2k1 − χ2k2, assuming χ2k1 − χ2k2 ∼ χ2k1−k2 5; where χs2kig1narises fromafitusingcertainmodelwiththeGaussiancomponent ters).Everyobservationhasparticularcharacteristics,andthere- included, and χ2 arising from a fit using the same model fore the decision of which Reduced-χ2 value is acceptable has k2 withouttheGaussiancomponent. beentakenonebyone.InFigure2wecansee thatmostofthe fits result in a Reduced-χ2 ≃ 1, as expected for a suitable fit. In some cases, FeKα line is clearly noticeable,butFeKβ is ThehighestvalueofReduced-χ2foranacceptedmodelhasbeen notprominentenoughtopermiterrorestimationofits parame- 1.82.TheparametersarisingfromthefitsarelistedinTablesA.2 ters.Inthesecases,wehaveconstrainedthecentroidenergyand andA.3. thenormofFeKβaccordingtoKallmanetal.(2004): We can classify the additive components of the models as thermal or non-thermal. A component is called thermal when – Energy(FeKβ)=Energy(FeKα)+0.652keV radiation is produced as a consequence of the thermal motion – Norm(FeKβ)=Norm(FeKα)×0.13photons/cm2/s of the plasma particles (e.g. blackbody radiation). Otherwise, the emitted radiation is non-thermal (e.g. non-thermalInverse- 4 http://pulsar.sternwarte.uni-erlangen.de/wilms/research/tbabs/ Compton emission). If all the additive components of a model 5 Thisassumptionisnotstrictlytrue,sinceχ2 andχ2 arenotindepen- k1 k2 dent.However,itprovidesanestimationontheimpactoftheGaussian 3 http://heasarc.nasa.gov/xanadu/xspec/ componentinthemodel. Articlenumber,page4of30 A.Giménez-García:Title The estimated parameters, like the EW, are very sensitive to Group #Sources Fecomplex Models N H the fit of the continuum. Therefore, although the Fe complex appears in the ∼6-7 keV energies, we broadened the spectral BeXB 10 TypeIII T,TN Low scope to an energy range of 1-10 keV to perform the analysis. It also allows us to consider possible calibration inaccuracies SGXB 12 TypeI N High in the charge transfer inefficiency (CTI) and the X-ray loading (XRL), an issue reportedin previousanalysis of EPN observa- SFXT 10 TypeIII T,N,TN High tions(seeMartínez-Núñezetal.(2014)andFürstetal.(2011)). (inquiescence) InthefewcasesofpossibleCTIorXRL,weappliedanartificial gainE = Eold +offset(seeTableA.2). γCasslike 8 TypeII T Low new slope Theestimationoftheparameterconfidenceregions(at90% level) have been calculated with a Markov Chain Monte Carlo HMGB 2 TypeIII T,N,TN Low (MCMC) technique,implementedin XSPEC, whereN genera- tionsofthesetoffreeparametersareusedtodeterminethebest- Table 2: Description of the features observed in this work, for the fitvaluesandtheconfidenceregions.WehavesetN =1.5×104 different groups of HMXBs. We have analyzed data of 46 sources. inourcalculations.Thesechainsarealsovalidtoestimatefluxes However, those classified as peculiars (3 sources) or non classified andequivalentwidths. (AX J1749.1-2733) are not included in this table. We define NH of a groupashigh,whenthetypicalvalueobservediswellovertheestima- tionsoftheinterstellarN inthelineofsightofthesourcesfollowing H Willingaleetal.(2013).Thatis,wesaythattheN ofagroupishigh H whentheabsorptionistypicallyintrinsicofthesystems. 3. Results 3.1. SpectralAtlas donotdetectFeKα.The12SGXBscanbewellfittedusing InAppendixB, wepresentthefullsampleofanalysedspectra. non-thermalmodels,althoughthermalcomponentsare also ThefiguresinAppendixBshowthesetofanalysedobservations plausibleinsomesources.Ingeneral,SGXBsarecharacter- (crosspoints),themodelemployed(solidline),thecomponents izedbyhighabsorptionandthepresenceofFefluorescence ofthemodel(dottedline)andtheratiobetweenobservationand emissionlines. model(lowerboxineachspectrumplot). – SFXTs. We have collected data from 10 sources. Three of We showa listofthe sourcesinTableA.1,givingthe class themshowFeKα:AXJ1841.0-0536,IGRJ11215-5952and wherewehavegroupedthemandthereferenceforsuchaclas- IGRJ16479-4514.TheEWupperlimitintherestofsources sification. We can see that the differentclasses of HMXBs be- ishigh.Therefore,FeKαwouldbeprobablydetectablewith havequalitativelydifferentintheregionoftheFecomplex(∼6- abettersignal-to-noise.Themodelsemployedforfittingthe 7keV),reflectingthedistinctaccretionregimesthatcharacterize SFXT systems are very heterogeneous, with no preference them. We have observed three patterns in the Fe complex, that ofthermal,non-thermal,neitheracombinationofbothkinds wedefineasTypeI,IIandIII(seeFigure3).WedefineTypeI, ofmodels. when fluorescence lines FeKα and FeKβ are observed,but not – γ Cassiopeae analogs.We have gatheredobservationsfrom recombination lines Fe XXV and Fe XXVI. We define Type II, 8 sources. Five of them exhibitFeKα. The EW upper limit whenfluorescencelinesare detected,togetherwith recombina- in the other 3 sources is very high. Again, it implies a tionlinesFeXXVandFeXXVI.Finally,wedefineTypeIII,when very likely presence of fluorescence in the case of better Felinesarenotdetected. signal-to-noise.In addition,recombinationlines of Fe XXV and Fe XXVI, are always present in the set of γ Cassio- peaeanalogs.TheselinesareincludedintheXSPECmodel mekal. For most of the observations we have achieved a The general features observed in this work, for the different goodfitusingacombinationofmekalcomponents.Inafew groupsofHMXB,aresummarizedinTable2,andexplainedbe- cases we have used other components: diskbb and power- lowinmoredetails: law,butmekalisbyfarthemostemployedoneinγCassio- peaelikesystems,inagreementwithpreviousX-rayanalysis – BeXBs.Wehavecollecteddatafrom10sources.Alltheob- (LopesdeOliveiraetal.2010,2006). servationswereperformedinquiescence.We havedetected FeKαemissioninonlyoneBeXB(SAXJ2103.5+4545).The – HMGBs. We have collected data from two HMGBs: upperlimitoftheFeKαEWintherestofBeXBsisingen- LSI+61303andLS5039.NoneofthemshowFefeatures. eral higher than the observed value in SAX J2103.5+4545, However, the signal-to-noise in these observations is poor implyingthat the lack of detectionsmightbe dueto a poor and the upper limits of the FeKα EW are high enough to signal-to-noise. The spectra can be modeled by thermal or do notrule outthe presenceof the line. We haveused both acombinationofthermalandnon-thermalcomponents,ex- thermalandnon-thermalcomponentsinthefits. cept for Swift J045106.8-694803(fitted using an absorbed – Peculiars. Set of sources that do not accommodate in any powerlaw).Sevensourcesaccepta thermalmodel,and6a of the aforementioned classes of HMXBs, as explained in combinationmodel(4ofthemacceptboth). theintroduction.We havecollecteddataof3such systems: – SGXBs.Wehavegathereddatafrom12sources.Tenofthem 4U2206+54,CentaurusX-3andCygnusX-1: show detectableFe fluorescenceemission. The onlyexcep- – 4U 2206+54does not show any detectable Fe emission tionsareIGRJ16465-4507andSAXJ1802.7-2017,themost line,andtheupperlimitintheEWofFeKαislow(itis distantSGXBsincludedinthiswork,at12.5and12.4kpcre- comparabletotheupperlimitsintheBeXBs). Itcanbe spectively.TheEWupperlimitsinthesetwosourcesishigh, fitted by means of an hybrid model (thermal plus non- implyingthattheirfaintnessisverylikelythereasonwhywe thermalcomponents). Articlenumber,page5of30 A&Aproofs:manuscriptno.Gimenez-Garcia Model Continuummodels N powerlaw×tbnew×cabs 1 Non-thermal N powerlaw ×tbnew ×cabs +powerlaw ×tbnew ×cabs 2 1 1 1 2 2 2 N powerlaw ×tbnew ×cabs +powerlaw ×tbnew ×cabs +powerlaw ×tbnew ×cabs 3 1 1 1 2 2 2 3 3 3 T mekal×tbnew×cabs 1 T (mekal+mekal)×tbnew×cabs 2 T (mekal+mekal+mekal)×tbnew×cabs 3 T (cemekl)×tbnew×cabs 4 T bbody×tbnew×cabs 5 T (bbody+bbody)×tbnew×cabs 6 Thermal T (bbody +bbody )×tbnew×cabs 7 1 2 T bbody ×tbnew ×cabs +bbody ×tbnew ×cabs 8 1 1 1 2 2 2 T diskbb×tbnew×cabs 9 T (diskbb+bbody)×tbnew×cabs 10 T bremss×tbnew×cabs 11 T bbody×tbnew×cabs+bremss×tbnew×cabs 12 T (bremss+bbody)×tbnew×cabs 13 TN (powerlaw+bbody)×tbnew×cabs 1 TN powerlaw×tbnew ×cabs +bbody×tbnew ×cabs 2 1 1 2 2 Both TN (powerlaw+diskbb)×tbnew×cabs 3 TN powerlaw×tbnew ×cabs +diskbb×tbnew ×cabs 4 1 1 2 2 TN (powerlaw+mekal)×tbnew×cabs 5 Table 1: Listofmodelsusedtofitthecontinuum, describedinXSPECnotation.Thebasiccomponents arepowerlaw, bbody, diskbb, bremss, mekalandcemekl,togetherwithtbnewtoaccountfortheabsorptionandcabsforthenon-relativisticComptonscattering.Theemployedmodels areacombinationofthesecomponents, inadditiontoGaussianprofilesmodellingemissionlines.Wedividethemodelsinthreetypes: N for # non-thermal,T forthermalandTN formodelscontainingboththermalandnon-thermalcomponents. # # – Centaurus X-3 presents a rich emission lines spectrum. TheoriginofbroadFe featuresin X-rayBinariesisstill an Concretely, in the Fe complex we are able to identify opened question, but the most likely alternatives are related to FeKα,FeKβ,FeXXVandFeXXVI.Wehaveusedeither the presence of an accretion disk (see e.g. Hankeetal. (2009); anhybridmodeleitheranon-thermalmodel. Ngetal. (2010); Duroetal. (2011)). However, narrow features – CygnusX-1exhibitsabroadFefeature,sometimescom- arenotcompatiblewithmaterialrotatingathighvelocitiesorrel- binedwithafaintandnarrow,butstatisticallysignificant, ativisticallybroadened.GiventhatbroadandnarrowFeKαhave FeKαline.Wehavemostlyusednon-thermalmodels,oc- a clearly differentorigin, they must be analysed in a separated casionallycombinedwithabbodyordiskbbcomponent. way.Then,itraisesthequestionofhowtodefinetheseparation markbetween them.In Figure4 we can see thatin oursample – AX J1749.1-2733. In this system, the optical member has thenumberofdetectedsourcesdecreaseswhenincreasingσ . beenclassifiedasaB1-2(Karasevetal.2010),butthelumi- line Moreover,mostofthedetectionsaregroupedatσ <0.15keV. nosityclass remainsunknown,preventingusto incorporate line Hence,σ < 0.15keV seems a naturalthresholdforthe def- the source in a defined group. Although it does not exhibit line initionof narrowlinesin the sample.In addition,we mustpay detectableFeemission,thehighabsorptionclearlypointsto attentionto theplausiblecontaminationof FeKαwith Fe XXV, asupergiantcompanion.Inaddition,the EWupperlimitof whichislocatedat∼6.7keV.Thechosencriterionseparatesthe FeKαiscompatiblewiththevaluesobservedinSGXBsand sourceswhereitis veryunlikelythatFeKαis contaminatedby SFXTs.Itcanbewellfittedusinganabsorbedpowerlawor Fe XXV (narrowlines),fromthesourceswhichprobablysuffer ablackbody. fromthisissue(broadlines).Amoredetailedanalysisofbroad Insummary,itisverylikelythatFeKαisanubiquitousfea- Fe featuresin HMXBs is outof the scope of this paper and its tureinHMXBs,anditsdetectionhighlydependsonthequal- discussionwillbefullydevelopedinaforthcomingwork. ity of the observations. In this regard, the EW of the line Hereafter, when FeKαis mentioned,we refer to the narrow is very affected by the level of intrinsic absorption present feature.Fromthetotalnumberof108analysedobservationswe onthesources(seealsoSection3.4.3).SGXBsandSFXTs, havedetected(narrow)FeKαin60ofthem. whichshowhigherabsorption,tendtoexhibitamorepromi- nentFefluorescence. 3.3. Centroidenergy 3.2. FeKαwidth InFigure5wecanseeanhistogrampresentingthecentroiden- ergy of FeKα. A Gaussian fit of the data reveals a mean value In Table A.3 we show the parameters of every detected FeKα, for the centroid energy of 6.42 keV. There are not significant including the width of the line (σ ). We have made a dis- differencesintheaveragedvaluesobtainedneitherfordifferent line tinction between narrow and broad FeKα. We have defined classesofHMXBs,neitherfordifferentstates.Thestandardde- narrow FeKα when σ < 0.15 keV, and broad FeKα when viationis0.02keV,comparabletothetheerrorthatwetypically line σ >0.15keV.Thisseparationisbothphysicallyandobserva- obtain in the estimation of the centroid energy in the fits (see line tionallymotivated: Table A.3). Taking into accountthe standard deviation and the Articlenumber,page6of30 A.Giménez-García:Title Fe complex: Type I Fe complex: Type II Fe complex: Type III 15 1.5 −−normalized counts skeV11 150 −−normalized counts skeV11 0.51 −−normalized counts skeV11 000...246 0 0 0 1.2 1.2 1.05 1.1 1.1 ratio 1 ratio 1 ratio 1 0.9 0.9 0.95 0.8 0.8 6 7 6 7 6 7 Energy (keV) Energy (keV) Energy (keV) Fig.3:PatternsfoundintheFecomplexofHMXBs:TypeI(leftpanel),TypeII(centralpanel),andTypeIII(rightpanel). K e F w o r r a N Broad Fe features Fig.5:CentroidenergyofFeKα,withaGaussianfitoverplotted(blue Fig. 4: Histogram of theFeKαwidth. Thebulkof thedetections are profile).Themeanvalueis6.42±0.02keV,compatiblewithFeI-XVIII. grouped showing σ < 0.15 keV. Wedefinethem asnarrow FeKα. Eventhoughwehave60detectionsof FeKα,inthisplotweonlysee line TherestaredefinedasbroadFefeatures.Eventhoughwehavedetected 55.Fivecasesfulfiltherequirementsofdetection,butthelowsignal-to- 60 narrow FeKα, this plot only includes 38. The reason is that 22 of noisedonotpermittofindanaccuratecentroidenergy.Theyhavenot them are very narrow (or the signal-to-noise very low) and they have beenincludedinthisplot.Inthesefivecaseswehavesetthecentroid beentreatedinthefitsasdeltafunctions(inTableA.3wepresenttheir energyofFeKαto∼6.4keV. widthasσ =0). line 3.4. Correlatedparameters uncertainties in the CTI corrections in EPN6, the centroid en- One of the goals of this work is to study plausible correlations ergyofFeKαconstrainstheionizationstateofFetolessionized involvingtheparametersofFeKα(position,width,intensityand thanFeXVIII(Kallmanetal.2004),inagreementwithprevious EW,andotherparametersinthefits,suchastheabsorbingcol- studiesinHMXBs(Torrejónetal.2010b;Gottwaldetal.1995; umnandtheintensityofthecontinuum.Notethat,evenwhena Nagase1989).Inthisregard,thestudyofTorrejónetal.(2010b) goodfitis reached,the confidenceregionofa parametermight usingHETGS(moreaccurateinwavelengththanEPN),givesa beoccasionallydifficulttofindduetothedependenceofthepa- narrowerconstrainin the ionizationstate (Fe I-X). Our present rameterwithotherparametersofthemodel.Wespecifyineach worksupportthisresultaddingmoresourcestothesample. of the following subsections the number of cases where a suc- In the rightside of Figure 5 we can see seven FeKα detec- cessfulestimationofthe90%confidenceregionhasbeendone. tionsemergingoutoftheGaussianprofile.Fourofthemareun- likelytobedescribedbysucha Gaussianprofile,sincetheylie morethanthreetimesthestandarddeviationawayfromthemean 3.4.1. ContinuumFluxvsFeKαFlux energy.AllfourbelongtoCygnusX-1. InFigure6wehaverepresentedtheunabsorbedfluxofthecon- 6 Pleasefindmoreinformationaboutlong-termCTIcorrectioninthe tinuumbetween1-10keVcancellingFeKαemission(F1−10keV), againstthefluxofFeKα(F ).Wehavesuccessfullyfounda releasenoteEPIC-pnLong-TermCTI,byM.J.SSmithetal.(2014),at FeKα 90%confidenceregionofthefluxofFeKαin56cases. http://xmm2.esac.esa.int/docs/documents/CAL-SRN-0309-1-0.ps.gz; andEPICstatusofcalibrationanddataanalysisbyGuainazzi(2008), In a logarithmicscale, we identifytwo differentpatternsof at http://xmm2.esac.esa.int/external/xmm_sw_cal/calib/CAL-TN- correlation.First, forasubsetincludingalltheeclipseobserva- 0018.pdf. tionsandIGRJ16318-4848,wefindacorrelationwithPearson Articlenumber,page7of30 A&Aproofs:manuscriptno.Gimenez-Garcia Fig. 6: F1−10keV versus FFeKα.Bluedashed linemarksthecorrelation Fig. 7: EWof FeKαagainst L1−10keV. γ Cassiopeae analogs (circles) observed for IGR J16318-4848 jointly with eclipse observations, and lieat L1−10keV < 1033 erg/s.Opensymbolsindicatethateitherthedis- the red solid line follow the bulk of the observations. The color map tanceeithertheerrorintheestimationofthedistanceisunknown.The indicatestheσ oftheline(definedinSection2.2). solidlinecorrespondstoalinearfitinlogarithmicscaleofthefilleddia- sign monds,thatis,thesourceswithavailabledistancewitherrorestimation andL1−10keV >1033erg/s. Coefficient(PC)of0.98(bluedashedline).Second,fortherest of observations, we find a correlation with PC=0.89 (red solid line). Baldwin (1977) observed an inverse correlation in the EW Alinearfitoftheparameters(inlogarithmicscale)intheout- of CIV and the UV luminosity in AGNs. Analogously,the de- of-eclipseobservations(redlineinFigure6),givesthefollowing creaseoftheEWofFeKαwhenincreasingtheX-rayluminos- dependence: ityiscalledX-raysBaldwineffect.Thedependencethatweob- serve is compatible within the error with the one observed by F1−10keV(erg/s/cm2)=FF1.e0K0±α0.08(erg/s/cm2)×102.18±0.87 (1) Terogryrerjaónngeet:aEl.W(2∝01L0−b0).29in X.-ray Binaries using a narrower en- 1.6−2.5Å Theerrorsshowthestandarddeviationoftheparametersinthe linearfit. 3.4.2. FeKαWidthvsCentroidEnergy The observeddivergenceamongeclipse (plusIGR J16318- 4848)andout-of-eclipseobservationssuggeststhatthecompan- In Figure 8 we present the centroid energy of this feature ver- ion star blocks in a differentproportionthe continuumand the sus its width (σ ). We have successfully found a 90% confi- line FeKαemission.Therefore,animportantcontributionofthefluo- dence region of σ in 20 cases. We can see a moderate cor- line rescenceemissionisproducedinanextendedregionofR&R . relation (black line in Figure 8, PC=0.55), indicating a pos- ⋆ This is consistent with previous analysis of eclipse observa- sible blending of lines. Two observations (uppermost side of tionsofHMXBs(e.g.Rodes-Rocaetal.(2011)andAudleyetal. Fig. 8) do not follow the correlation. They correspond to ob- (2006)).Inparticular,Audleyetal.(2006)estimatesthat20%of servationsof4U1700-37(Obs.ID0083280201)andEXO1722- FeKαinOAO1657-415isemittedfrom19light-secondsoffthe 363 (Obs.ID 0405640201)where the Fe complex is hardly re- X-raysource. solved, and therefore it is very likely that a contribution of We have also the luminosity of the continuum, in order to Fe XXV and Fe XXVI in the model of FeKα is increasing the faceitwiththeEWofFeKα.Fortheflux-to-luminosityconver- measuredwidthoftheFeKαline. sionwe haveusedthe estimationsofthedistanceshowninTa- Coloured squares correspond to the expected width from ble A.1. We haveexcludedeclipse and IGRJ16318-4848from the contribution of three different broadening phenomena:line thisanalysis,giventhattheEWisstronglyaffectedbythehigh blending,DopplershiftsandComptonbroadening.Adiscussion obscurationofthecontinuumthattheysufferfrom.InFigure7 onthedifferentbroadeningcontributionsisgiveninSection4. we plot the EW of FeKα against the unabsorbed luminosity of the continuumbetween1-10keV cancellingFeKα emission 3.4.3. Curveofgrowth (L1−10keV).Weobservetwodifferentgroupsofsources:1)γCas- siopeae analogslie at low luminosities(L1−10keV < 1033 erg/s); InFigure9weshow,foroutofeclipseobservations,theNH ver- 2)therestofsourceswhichexhibitFeKα.γCassiopeaeanalogs sus the EW of FeKα (what is generally known as the curve of don’t show any evident correlation (they are very few points), growth). We have successfully found a 90% confidence region whiletherestpresentamoderateinversecorrelation(PC=-0.25, of both N and EW in 46 cases. We want to take into account H and PC=-0.39 using only the sources with an available estima- observationswereN reflectstheintrinsicabsorptionofthesys- H tion of distance with error, marked as filled diamonds in Fig- tem. Hence, we have set N > 2 as a condition to safely ex- H ure 7). A linear fit using the filled diamondsin Figure 7 gives: ceed the typical N of the interstellar medium for the sources H here studied (checked using the online application following EW = L−0.52±0.27(erg/s)×1017.45±9.83 (2) Willingaleetal. (2013)). The use of this criterion excludes the 1−10keV Articlenumber,page8of30 A.Giménez-García:Title Fig. 8: Width of FeKα (σ ) versus the centroid energy (black Fig. 9: Curve of growth observed for FeKα, that is, EW against line squares). Black solid line traces a linear fit. We have marked in N . The turquoise band marks the expected correlation using nu- H colour the expected width from the contribution of three differ- merical simulations. The sources are identified by different symbols ent broadening phenomena: line blending, Compton broadening and when more than one observation is included: 4U 1700-37 (open cir- Doppler shifts, considering velocities of V(km/s) = 1000 (red) and cle), 4U 1907+09 (open upward triangle), Cygnus X-1 (open down- V(km/s)=2000(green).Everysingleblacksquarehasassociatedasin- wardtriangle),EXO1722-363(opendiamond),IGRJ16318-4848(open gleredsquareandasinglegreensquarecorrespondingtotheexpected square) and IGR J16320-4751 (plus). Only one observation for Cen- valuesofσ forthatobservation(seeSection4). taurus X-3, GX 301-2, Vela X-1 and XTE J0421+560 (all four a star line symbol). BeXBSAXJ2103.5+4545,theγCassiopeaeanalogs:γCassio- peae and HD 110432;andthe SFXT IGRJ11215-5952.More- over,eclipseobservationsshowhigherEWandtheyarenotcom- parabletoout-of-eclipseobservations.Therefore,eclipseobser- vationshavenotbeenplottedin Figure9. Asa consequenceof the chosen criteria, we end up with a set of 36 observations, whereallthedonorsaresupergiants. N and the EW of FeKα are expected to correlate in H HMXBs,asshownbyTorrejónetal.(2010b),sincethespectral linesareusuallystrongerwhentheopticaldepthincreases.Our sampleconfirmsthese expectations,showinga notablecorrela- tion(PC=0.85). We have determined the theoretical curve of growth using numericalsimulations.InthissimulationsthereisaninputofX- rayradiationwithapowerlawprofile,thatistransmittedthrough acloudofsphericallydistributedneutralmatter(Eikmann2012). We have taken into accountthe power law index (Γ) in the simulations, since steeper profiles entail less photons available abovetheFeK-shellthresholdenergy,decreasingtheEW.That is,forthesameN ,thehigherisΓ,theloweristheEWofFeKα. H InFigure9theturquoisebandtracestheresultsfromthesimula- tionswithΓ ∈ [0.5,2],whichisthetypicalrangewherewefind Fig.10:HistogramsshowingacomparisonoftheNH valuesobserved Γinourfits. inSGXBs(filledred)andSFXTs(emptyblue). 3.5. N :SGXBsandSFXTs H In Figure 10 we have plotted histogramsfor the N valuesob- have10N valuesforSGXBsand9forSFXTs.Wehavemerged H H servedinSGXBs(red)andSFXTs(blue).Inthecaseswherewe theminasetof19elements,andconsideredeverypossiblecom- have more than one observation for the same source, we have binationof2groupsof10and9elements(92378possibilities). averagedthevaluesinordertoobtainoneN representativefor We have compared the median of the two subsets, and calcu- H eachsystem.TheorbitalphasecriticallyaffectstheobservedN , latedtheabsolutedifference.99.7%ofthecaseshaveproduceda H and therefore ingress/egress and eclipse phases have not been lowerabsolutedifferencethantheobservedone.Usingthemean takenintoaccount. insteadofthemedianthepercentageisalsoveryhigh(98.8%). We find that SGXBs are in general more absorbed than In conclusion,it is very likely that the discrepanciesin the ob- SFXTs.Wehaveperformedapermutationtestinordertoquan- servedN valuesforSGXBsandSFXTsareproducedbyphys- H tify if the observed disparity in the N is a random effect. We icalreasonsratherthantheyarisebychance. H Articlenumber,page9of30 A&Aproofs:manuscriptno.Gimenez-Garcia Fig.11:EWofFeKαagainstN ,inIGRJ16320-4751.Theturquoise Fig. 12: Orbital modulation of N in the system IGR J16320-4751. H H bandmarksthenumericalsimulationsresults,withΓ∈[0.5,2]. Solidlinescorrespondtotheexpectedabsorptionfromasmoothwind and a non eccentric orbit, assuming an orbital separation a = 1.6R , ⋆ R⋆ = 20R⊙, M˙ = 10−5M⊙/yr,3∞ = 700km/s, β = 0.8;anddifferent 3.6. Individualsourcesanalysis:IGRJ16320-4751and orbitalinclinationsi=0,π/10, π/6rad(red,greenandblue). 4U1700-37 3.6.1. IGRJ16320-4751 ables, since they are much more commonly used than M˙/3∞ in the literature. This way, we constrain our parameters to the IGRJ16320-4751wasdetectedbyASCAin1994and1997(cor- observedand predicted range of values in O supergiants: M˙ = responding to AX J1631.9-4752), and by INTEGRAL in 2003 10−7−10−5M⊙/yrand3∞ =500−3000km/s(Kudritzki&Puls (Tomsicketal. 2003). It is a HMXB composed by an O8I op- 2000;Vinketal.2001). tical star and a neutron star (Rahouietal. 2008). It shows a Fora nullorbitalinclination,weobviouslyobtaina flat N modulation of 8.96 days, that it is considered its orbital pe- H riod (Corbetetal. 2005), and a pulsation period of ∼1300s modulation(redlineinFig.12),thatdescribestheobservedNH (Lutovinovetal. 2005). The ESA archives permits to collect (except at φ = 0), assuming a = 1.6R⋆, R⋆ = 20R⊙, M˙ = eleven observations of IGR J16320-4751,enabling us to study 10−5M⊙/yr, 3∞ = 700 km/s and β = 0.8. Graduallyincreasing the orbitalinclination (i = π/10, π/6 rad; green and blue lines in more detail the curve of growth, as well as tracking the ab- respectively), we are able to describe a high N at φ = 0, but sorptionvariationalongtheorbitalphase. H loosingsimilaritiesaroundotherorbitalphases.Anyhow,given In Fig. 11 we can see the curve of growth, as shown in thesimplicityofthemodel,theobtainedparametersarecertainly Fig. 9, restricted to IGR J16320-4751. We clearly see the de- justindicatives. pendencebetween N and EW of FeKα, as stated for the bulk H of the sources in Section 3.4.3, and the general trend follow- ing the numericalsimulations marked with the turquoise band. 3.6.2. 4U1700-37 However,the agreementwiththe simulationsisnotcompletely fulfilled, given that the spectral fits of IGR J16320-4751 have 4U 1700-37 was detected for the first time by Uhuru in 1970 broughtout a power law index Γ ∼ 0.5 (see Table A.2). Since (Jonesetal.1973).TheopticalstarisHD153919,aO6.5Iaflo- fromalowerpowerlawindexweexpectmoreEWofFeKα,the catedatadistanceof1.9kpc(Ankayetal.2001).Theorbitalpe- pointsforIGRJ16320-4751areexpectedtobelocatedintheup- riodis3.41days.SinceX-raypulsationshavenotbeendetected peredgeoftheturquoiseband,correspondingtothesimulations so far,the compactobjectcan be either a neutronstar or either with Γ = 0.5.We considerthatthe generaltrendiscorrect,but ablackhole.ThedatabaseofESAcontainsfiveobservationsof therearestillsomeuncertaintiesinthefitsand/orthetheoretical 4U1700-37,thatwesplitinninespectrainordertodistinguish hypothesis(sphericalgeometryandneutralmatter). differentstatesofthesource. From 14th Augustto 17th September,in 2008,a campaign InFigure13wecanseethecurveofgrowth,for4U1700-37. consisting of nine observations of IGR J16320-4751 was per- AlthoughsevenoftheninespectrashowFeKα,wewereableto formedbyXMM-Newton.Wehaveusedthissetofdatatoplot constraintheboundariesofN inonlyfiveoftheanalysis.One H theN modulationdependingontheorbitalphase(Fig.12).We of them correspondsto an eclipse observation (filled circle). It H havesetφ=0attheN maximum.Wehavealsocalculatedthe showsmuchmoreEWbecausethecontinuumfluxisblockedby H theoreticalabsorptionexpectedfromasmoothwindinanonec- theopticalstar,whereasFeKαcomesfromamoreextendedre- centric orbitusing a β velocitylaw (Castoretal. 1975) and the gionthatisnotcompletelyhiddenduringeclipse.Wedonotsee motionequation,takingintoaccountvariationsintheorbitalin- anobviousdependencebetweenN andEW,althoughthepoints H clinationi,orbitalseparationa,starradiusR ,masslossrateM˙, lieinaregionclosetotheexpectedvalues(turquoiseband).Any- ⋆ parameterβandtheterminalvelocityofthewind3∞. way, a set of four observations (excluding the eclipse), is too Indeed, M˙ and 3∞ cannot be disentangled in this simple smalltoperformanstatisticalanalysis. model, and the actual parameter used is M˙/3∞. However, here- From 17th to 20th of February 2001, 4U 1700-37 was ob- after we give values of M˙ and 3∞ as if they were free vari- servedbyXMM-Newtonfourtimes,inacampaigncoveringdif- Articlenumber,page10of30

See more

The list of books you might like

Upgrade Premium
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.
DMCA & Copyright: Dear all, most of the website is community built, users are uploading hundred of books everyday, which makes really hard for us to identify copyrighted material, please contact us if you want any material removed.
Copyright @ PDFdrive.to, 2022 | We love our users