The Galactic plane at faint X-ray fluxes - II. Stacked X-ray spectra of a sample of serendipitous XMM-Newton sources

0.43 MB·

by  R. S. Warwick

Mon.Not.R.Astron.Soc.000,1–12(2013) Printed15January2014 (MNLATEXstylefilev2.2)
The Galactic plane at faint X-ray fluxes - II. Stacked X-ray spectra
of a sample of serendipitous XMM-Newton sources
4 1⋆ 1 2
R. S. Warwick , K. Byckling , D. Pe´rez-Ram´ırez
1
0 1DepartmentofPhysicsandAstronomy,UniversityofLeicester,UniversityRoad,Leicester,LE17RH,UK
2 2DepartamentodeF´ısica,UniversidaddeJae´n,CampusLasLagunillas,23071Jae´n,Spain

n
a
J Accepted.Received;inoriginalform

4
1
ABSTRACT

] WehaveinvestigatedtheX-rayspectralpropertiesofasampleof138X-raysourcesdetected
E
serendipitouslyinXMM-NewtonobservationsoftheGalacticplane,atanintermediatetofaint
H flux level. We divideour sample into 5 subgroupsaccordingto the spectralhardnessof the
. sources, and stack (i.e. co-add) the individual source spectra within each subgroup. As ex-
h
pectedthesestackedspectrashowasofteningtrendfromthehardesttothesoftestsubgroups,
p
whichisreflectedintheinferredline-of-sightcolumndensity.Thespectraofthethreehard-
-
o est subgroupsare characterized by a hard continuum plus superimpose Fe-line emission in
r the 6–7 keV bandpass. The average equivalent width (EW) of the 6.7-keV He-like Fe-Kα
st lineis170+35eV,whereasthe6.4-keVFe-Kfluorescencelinefromneutralironandthe6.9-
−32
a keVH-likeFe-LyαlinehaveEWsof89+26 eVand81+30 eVrespectively,i.e.roughlyhalf
[ −25 −29
thatofthe 6.7-keVline.The remainingsubgroupsexhibitsoftthermalspectra.Virtuallyall

1 ofthespectrally-softX-raysourcescanbeassociatedwithrelativelynearbycoronally-active
v late-typestars,whichareevidentasbrightnear-infrared(NIR)objectswithintheX-rayerror
6
circles.Onasimilarbasisonlyaminorityofthespectrally-hardX-raysourceshavelikelyNIR
1
identifications.The averagecontinuumandFe-linepropertiesofthe spectrally-hardsources
1
are consistent with those of magnetic cataclysmic variables but the direct identification of
3
largenumbersofsuch systemsin Galactic X-raysurveys,probingintermediatetofaintflux
.
1 levels,remainschallenging.
0
4 Key words: stars: dwarf novae - novae, cataclysmic variables - X-rays: binaries - X-rays:
1 stars
:
v
i
X
r 1 INTRODUCTION (e.g.Hertz&Grindlay1984;Motchetal.2010).Finallyatrelative
a
lowX-rayluminosities(LX < 1032 ergs−1)thelocalpopulation
Fromthegroundbreakingdiscoveriesofmissions,suchasUhuru,
ofcoronally-activestarsandbinariesdominatethesourcestatistics,
SAS-3, Ariel V and HEAO-1, that date back over three decades,
particularlyinthesoftX-rayregime(e.g.Gu¨del2004).
rightthroughtothepresenteraofobservatoryclassmissions,such
asChandra, XMM-Newtonand Suzaku, thestudyofX-raybright Despite thegreat advances inour understanding of the vari-
Galactic sources has been a key theme within high energy as- ous classes of Galactic X-ray source, our knowledge of the pop-
trophysics. As a result we now have extensive knowledge of the ulation properties, such as the number density, the exact form of
most luminous X-ray emitters in our Galaxy (with X-ray lumi- theluminosityfunctionandtheGalacticdistribution,remainsvery
nositytypicallyintherange1035−38 ergs−1),whichinthemain incomplete.ThisisofparticularrelevanceformoderndayGalac-
can be divided into either low-mass or high-mass X-ray binaries tic surveys conducted in the hard X-ray band (i.e. above 2 keV)
(LMXBsorHMXBs)poweredbytheaccretionofmatterontoei- where, forsources of intermediateluminosity, thevisiblevolume
theraneutronstarorblackhole(e.g.Grimm,Gilfanov&Sunyaev can extend beyond the Galactic Centre to encompass the bulk of
2002; Ebisawaetal. 2003). At intermediate X-ray luminosities theGalacticdisc.Ideallyonewouldusesuchsurveystoconstruct
(nominally 1032−35 erg s−1) other types of source in addition large samples of identified sources from which the properties of
to X-ray binaries enter the mix, including bright supernova rem- the various populations might be inferred. In practice the identi-
nants, cataclysmic variables (CVs) powered by accretion onto a fication of the counterparts to hard X-ray sources in the Galac-
white dwarf star, massive stars with shock-heated winds and the ticplane, evenwithsub-arcsecX-raypositions, isextraordinarily
most extreme coronally-active binaries, such as RSCVn systems difficult due to the crowded nature of the star fields and the ef-
fectsofextinctionatopticalandnear-infrared(NIR)wavelengths
(e.g.Ebisawaetal.2005;Laycocketal.2005;vandenBergetal.
⋆ E-mail:[email protected] 2009). Evenwhen likelycounterparts areidentified, their follow-
2 R. S.Warwick, K. Byckling,D. Pe´rez-Ram´ırez
up, including distance determination, can often present signifi- 2010;Yuasaetal.2012).However,asnotedearlier,thecurrentun-
cant challenges (e.g. Motchetal. 2010; vandenBergetal. 2012; certaintiesrelatingtothepopulationpropertiesandstatisticsleave
NebotGo´mez-Mora´netal.2013). manyofthedetailsofthismodeltobeconfirmed.
One way of constraining the properties of low to interme- InWarwick,Pe´rez-Ram´ırez&Byckling(2011;hereafterPa-
diate luminosity, source populations is through their integrated per I), we investigated the serendipitous X-ray source population
emission. In the Galactic context, the spectrally hard Galactic intheGalacticplane,utilisingtheSecondXMM-NewtonSerendip-
RidgeX-rayEmission(GRXE)(Worralletal.1982;Warwicketal. itousSourceCatalogue(2XMMi;Watsonetal.2009).Thecurrent
1985;Koyamaetal.1986a;Yamauchietal.1996)hasbeeninter- paper(PaperII)buildsontheworkinPaperIbyfocusingonthe
preted both in terms of the superposition of faint point sources X-rayspectralpropertiesofasubsetofsourceswithrelativelygood
(Sugizakietal. 2001; Revnivtsevetal. 2006; Yuasaetal. 2012) photon statistics. In Paper I, we showed that the great majority
andasahighlyenergetic,veryhightemperaturephaseoftheinter- (i.e.>90%)ofthespectrally-softX-raysourcesmaybeidentified
stellarmedium(Koyamaetal.1986b;Kanedaetal.1997;Tanaka withrelativelylocalcoronally-activestars;however,thenatureof
2002), amongst other possibilities (e.g. Valiniaetal. 2000). The thespectrally-hardX-raysourcesremainsunclear.Thegoalofthe
GRXE is seen as a narrow ridge of emission extending out to currentpaperisthereforetoinvestigatetheX-rayspectralproper-
|l| ∼ 60◦, but with a surface brightness that peaks towards tiesofasampleofsourcesrepresentativeofthesourcepopulation
theGalacticCentre(Yamauchi&Koyama 1993;Revnivtsevetal. presentintheGalacticplaneatintermediatetofaintX-rayfluxes.A
2006;Koyamaetal.2007).Inthe4–10keVbandthespectrumof keyquestioniswhetherareasonablenumberofsuchsourceshave
the GRXE matches that of 5–10 keV optically-thin CIE thermal thecharacteristicsofCVs,consistentwiththehypothesisthatCVs
plasmawithprominentFe-linesat6.67and6.97keVarisingfrom contributesignificantlytotheGRXE.
K-shellemissioninHe-likeandH-likeions(Koyamaetal.1986a; The remaining sections of this paper are organised as fol-
Kanedaetal.1997).AnFeKαlineat6.4keVresultingfromthe lows. In the next section we discuss how we defined our source
fluorescence of neutral or near-neutral iron is also evident in the sampleandextractedX-rayspectraldataforeachsourcefromthe
X-rayspectrumoftheGRXE(Koyamaetal.1996;Ebisawaetal. XMM-Newton sciencedataarchive.Tocarryoutmeaningfulspec-
2008;Yamauchietal.2009).Hereafterwerefertothethreepromi- tral analysis, we have divided our sample of sources into 5 sub-
nentFeKαlinesasthe6.4-keV,6.7-keVand6.9-keVlines.Below groupsbasedonthesourcespectralhardnessandstacked(i.e.co-
4keV,emissionslinesofabundantelementssuchasMg,Si,S,Ar added)theindividualsourcespectrawithineachsubgroup(§3).In
andCaarealsoevidentinthespectrumoftheGRXE,withthein- §4, we analyse and compare the stacked spectra for the different
tensityratiosoftheK-linesassociatedwiththeHe-likeandH-like subgroups. Wegoontoinvestigatetheincidenceoflongerwave-
ionsofeachelement(includingthoseofFe)indicativeofamulti- lengthcounterpartsbycross-correlatingourX-raysourcepositions
temperature plasma (Kanedaetal. 1997; Tanaka 2002). The tem- withNIRcatalogues(§5).Insection§6wediscussourresultsinthe
peraturestructureappearstobesimilaratdifferentlocationsalong context of thelikelycontribution ofmagnetic CVstotheGRXE.
theGRXE,withtheexceptionoftheregionwithinadegreeorso Finallywebrieflysummarizeourmainconclusions.
oftheGalacticCentre(Uchiyamaetal.2011,2013).
Ithasalsobeenreportedthatabove10keVtheGRXEspec-
trumexhibitsahard powerlaw tail,which extends tothehard X-
ray/γ-ray region (Yamasakietal. 1997; Valinia&Marshall 1998;
2 THESOURCESAMPLEANDDATAREDUCTION
Strongetal.2005;Krivonosetal.2007).
RecentstudieshaveshownthattheGRXEsurfacebrightness In Paper I we discuss the selection and properties of a sam-
followsthatoftheNIRlightassociatedwiththeoldstellarpopula- ple of 2204 serendipitous X-ray sources drawn from 116 XMM-
tionoftheGalaxy(Revnivtsevetal.2006).AlsoverydeepChan- NewtonobservationswithpointingsintheGalacticplanetowards
dra observations have directly resolved over 80% of the GRXE thecentral quadrant of theGalaxy. Herewe focus on asubset of
neartheGalacticCentreintopointsources(Revnivtsevetal.2009). sourcesfromthisoriginalsamplewithaviewtoextendingthespec-
Taken together this is compelling evidence for the origin of the tralanalysisbeyondthehardnessratioand’bandindex’considera-
bulkofGRXEintheintegratedemissionofpointsources,although tionsofPaperI.
there isstillsome debate asto whether theremight remain some Theselectionofobjectsforthisnewstudywasbasedonthe
excessemission attributabletoadistinct veryhot diffusecompo- following requirements: (i) the source was detected in the EPIC
nentwithin2degreesoftheGalacticCentre(Uchiyamaetal.2011, pn camera in an observation in which the medium filter was de-
2013;Heard&Warwick2013;Nishiyamaetal.2013). ployed; (ii) there were a nominal 200-1200 pn counts associated
ThenextkeystepistoidentifytheGalacticsourcepopulation withthesource (morespecifically theproduct of the 0.5–12 keV
or populations that give risetothe GRXE.Thefirst requirement, pncount-ratetimestheeffectivesourceexposure timewasinthe
in this regard, is that the source population should have a suffi- quotedrange);(iii)ifthereweremultipleobservationsofthesame
cientlyhighvolumeemissivity(essentiallytheproductofthemean source,theobservationwiththelongestexposuretimeand/orbetter
spacedensityandmeanX-rayluminosity)toexplaintheobserved qualitydatawasselected.Afterapplyingthesecriteria,thesample
surfacebrightnessoftheGRXE.Asecondconstraintisthatthein- reducedto138sourcesdrawnfrom63differentXMM-Newtonob-
tegratedspectrumofthesourcesshouldmatchtheobservedspec- servations(hereafterwerefertothissetof138sourcesasthecur-
trumoftheGRXE.Inthiscontextithasbeenproposedthatamix rentsample).
ofmagneticCVspluscoronally-activebinariesmayhavesufficient Briefdetailsofthesourcescomprisingthecurrentsampleare
spatialdensityandhardX-rayluminositytoaccountforthebulkof givenintheAppendix. ThedistributioninGalacticlongitudeand
theGRXEanditsextensionintotheGalacticCentre(Munoetal. latitude is shown in Fig. 1. The bulk of the sources are located
2004;Sazonovetal.2006;Revnivtsevetal.2006;Revnivtsevetal. within±1◦oftheGalacticplaneandroughly20percentliewithin
2008). Also, CVs and active binaries have marked spectral simi- ≈1◦oftheGalacticCentre.
laritiestotheGRXE(Revnivtsevetal.2006;Tanaka&Yamauchi In Paper I we categorised sources as spectrally hard or soft
The GalacticplaneatfaintX-rayfluxes 3
Table 1. Thedesignation and hardness ratio range forthe 5 source sub-
groups.Thenumberofsourcescontributing andthetotalcountspersub-
group,netofthebackground,arealsoquoted.
Subgroup Hardness Numberof Totalnet
ratio sources counts
H 1.0–0.8 32 8221
MH 0.8–0.2 25 6729
MD 0.2–(-0.2) 22 7667
MS -0.2–(-0.8) 26 6906
S -0.8–(-1.0) 33 11449
Figure1.(a)ThedistributioninGalacticlongitudeofthesourcescompris-
ingthecurrentsample.(b)ThesameinGalacticlatitude.
Forthegreatmajorityofthesourcesthecountrateisintherange
5–100 pncount ks−1 ineither orbothofthesoftandhardbands
(cf.fig.5inPaperI).Forthesoftsources,thefluxrangesampledis
0.1–2×10−13 ergcm−2s−1(0.5–2keV)andforthehardsources,
0.6–12×10−13ergcm−2s−1(2–10keV).
Thepresentanalysisisbasedonspectraldatasolelyfromthe
EPICpncamera.Weextractedthepnspectrumofeachsourceus-
ingtheXMM-NewtonScienceAnalysisSystem(SAS)andstandard
techniques. Inbrief, each dataset wasreprocessed inorder toap-
plythemostrecent calibrationandscreenedfor highbackground
flares by applying a cut when the full-fieldcount rate in the 10–
12 keV band exceeded 0.4 ct s−1. Only single or double X-ray
eventswerechosenwithPATTERN=0–4andwith FLAG=0(the
latterexcludingeventsclosetochipgapsandbadpixels/columns).
Thesourcespectrawereextractedwithinacircularregionofradius
r=35arcsec(=700pixels)formostofthesources.However,for
13sourcestheradiusofthesourceextractionregionwascalculated
using the SAS task REGION in order to minimise possible con-
taminationfromnearbysources. Thebackground wastakenfrom
acircularextractionregionwitharadiusofr=120arcsec(=2600
pixels); in general, the background region was positioned on the
sameCCDchipasthesource(oroccasionallyonanadjacentchip).
Finally,theredistributionmatrixfiles(RMFs)andtheancillaryre-
Figure2.(a)Thedistributionofthebroad-band(0.5–2keV:2–12keV) sponsefiles(ARFs)werecreatedusingRMFGENandARFGENfor
hardnessratiowithinthecurrentsample.Theverticaldottedlinesdefinethe eachindividualsourcespectrum.
rangesusedtoselectthe5sourcesubgroups(seeTable1).(b)Themedian
photonenergy(keV)inthesourcespectrumplottedversusthebroad-band
hardnessratioforeachsource.
3 STACKINGOFTHESOURCESPECTRA
depending on whether their broad-band hardness ratio,HR1, was After background subtraction, the net number of counts actually
positive or negative. In keeping with the parent sample, the cur- recordedforeachsourcewastypicallyintherange100–800 (see
rent sample splits fairlyevenly into the soft and hard source cat- TableA1).Themeasuredcountsweregenerallyreducedcompared
egories as illustrated in Fig. 2(a). An alternative approach to the tothepredictionusedinsourceselectionduetotwoeffects;theloss
use of hardness ratiosto classify the spectral properties of X-ray ofsignallyinginthewingsofthePSFoutsideofthesourceextrac-
sourceswithlimitedcountstatisticsistoemploy‘quantile’analysis tion circle and the use of a more stringent threshold for the data
(e.g.Hongetal.2004)and,inthatcontext,Fig.2(b)comparesthe filteringthanemployedintheproductionofthe2XMMicatalogue.
medianphotonenergyinthe(background-subtracted)sourcespec- Since individually the sources have insufficient counts for
trum,E ,withthecorrespondingvalueofHRforeachsource. detailed spectral analysis (although see §4.2), the full sample of
50%
Forthecurrentsample,E rangesfrom0.5keVatthesoftex- 138 sources was divided into 5 subgroups (H, MH, MD, MS
50%
tremeuptotoamaximumof≈6keVforthehardestsources. and S) based on the source spectral hardness (see Fig. 2 and Ta-
Giventhatthesourceselectionisbasedonatotalcountcrite- ble1).Withineachofthesefivesubgroupsthesourcespectrawere
rion,thereisawidespreadinthecountratesoftheselectedobjects. stacked. After applying a scaling so as to match the area of the
background region tothat of thesource extractionregion, theas-
sociated background spectra were similarly co-added. The RMF
1 More specifically, HR = (H-S)/(H+S), where H corresponds to the andARFfilesofeachindividualsourcespectrumwerecombined
2XMMiBands4and5(2–12keV)andStoBands2and3(0.5–2keV) byusingtheFTOOLMARFRMFv.2.2.0.Thereafter,theindividual
-seePaperIfordetails. outputfilesfromMARFRMFwerecombinedinADDRMFv.1.21so
4 R. S.Warwick, K. Byckling,D. Pe´rez-Ram´ırez
astogiveatotalresponsefileapplicabletothestackedspectrafor H,MHandMDsubgroups.Thereisalsoevidenceforthe6.9-keV
each source subgroup. Finally, the spectra were binned using the line in the H and MH spectra and weakly in the MD spectrum.
FTOOLGRPPHA;8-channelbinning(ofthe4096inputPHAchan- Similarlythe6.4keVFelineisdetectedinthespectraofthetwo
nels)wasemployedinthecaseofthespectrapertainingtotheH, hardestsubgroups,butnotintheMDspectrum.Fig.4illustratesthe
MHandMDsubgroups,whereasfortheMSandSsubgroups,4- formofthemeasuredspectrumandthecorrespondingbest-fitting
and8-channel binningwasappliedtothePHAranges0–511 and spectralmodelintheregionoftheiron-linecomplexfortheHsub-
512–4095,respectively. group.
Source statisticsfor thefivesubgroups aregiven inTable1. In order to determine the line EWs averaged over the three
Summed over the full set of 138 sources, 40972 pn counts were datasetswerepeatedtheabsorbedpowerlawplusFe-linesspectral
recordedinthe0.5–12keVband,netofthebackground. fittingwiththeEWsofthethreelinestiedacrosstheH,MHand
Thespectraofthefivesubgroupsafterbackgroundsubtraction MDspectra(inadditiontothephotonindexofthepowerlawcon-
areillustratedinFig.3,togetherwiththecompositespectrumfor tinuum).TheresultisreportedinTable2onthelinelabelledJoint-
thefullsample.Averyclearspectralsofteningtrendisevidentfrom EW.InsummarytheaverageEWofthe6.7-keVlinemeasuredin
theHthroughtotheSsubgroups(ratherasexpected).Thespectra the full sample of sources with HR > −0.2 is ≈ 170 eV, with
ofsubgroupsHandMHshowstrongsoftX-rayabsorptionandalso the6.4-keVand6.9-keVlinescominginatapproximatelyhalfthis
evidence forFe-featuresinthe6–7keV band. TheMDsubgroup value.
spectrumretainsahardtailoutto∼7keV,whichisbarelypresent Asimpleabsorbedpowerlawmodelprovidedapoordescrip-
inthespectrumofMSsubgroupandcompletelyabsentfromthat tionofthespectrapertainingtotheMSandSsubgroups.Accord-
oftheSsubgroup. ingly these spectra were fitted using a two-temperature thermal
plasmamodel,whichisacommonlyusedapproximationforsys-
temsexhibitingabroadspanofemissionmeasureversustempera-
4 X-RAYSPECTRALANALYSIS ture.InXSPECthiswasachievedbyemployingtwoapeccompo-
nents witha tiedmetal abundance Z⊙. Thelatter parameter was,
Theanalysisofthe5stackedspectrawascarriedoutusingXSPEC nevertheless, allowed to vary (as a single scale factor applied to
version12.8.1(Arnaud1996).Thespectralfitswerelimitedtothe thestandardsolarabundanceratiosdefinedbyAnders&Grevesse
energyrange2–9keVfortheHsubgroupandto1–9keVforboth 1989).Thesameforegroundcolumndensitywasassumedtoapply
the MH and MD subgroups. In each case, data between 7.8 and toboththermalcomponents.
8.3 keV were excluded in order to eliminate the effect of back- Thefittingofthespectrumofthesoftestsubgroupresultedin
groundsubtractionresidualspertainingtotheKαCufluorescence temperaturescloseto0.5keVand1.0keVforthecoolerandhot-
linesinthedetectorbackground.FortheMSandSsubgroups,we tercomponents, respectively - asreportedinthelower sectionof
usedtheenergyrange0.5–5keV. Table 2. In this model the 1-keV component contributes 68% of
theobserved0.5–2keVflux.Theinferredmetalabundancewithin
thermalplasmaisextremelysubsolar,butthisismostprobablyan
4.1 Thespectralmodels
artefactofthetwo-temperaturespectralapproximation,ashasbeen
The X-ray spectra of the H, MH and MD subgroups were inves- notedbymanyauthors(e.g.Stricklandetal.2000).Thecombina-
tigatedusing aspectral model comprisingapowerlaw continuum tionofthelowderivedNHandatwo-temperaturedescriptionofthe
plusthreeGaussianemissionlinesatfixedenergiesof6.41,6.68, emissionspectrumarefullyconsistentwiththeunderlyingsource
and6.96keV.Asnotedearlier,thesethreelinescorrespondtoKα populationbeingnearby,coronally-activestars.
emissionfromneutral(ornearneutral)Fe-atoms,He-likeFe-ions Wenextattemptedtousethesameemissionmodel(kTvalues
andH-likeFe-ions,respectively. Theintrinsicwidthsof thethree fixedat 0.5 and 1keV and Z⊙ set to0.13) tofitthe spectrum of
lineswerefixedatσ=30eV(e.g.Koyamaetal.2007;Capellietal. theMSdatasetbutfailedtoobtainasatisfactoryresult.Infact,the
2012).Theemissioncomponentsweresubjecttoaline-of-sightab- presenceofa(modest)hardtailwithinthisdatasetrequiresahot-
sorptioncolumndensity,NH,(usingphabsinXSPEC).Ajointfit terthermalcomponent,withtheadditionalNH likelymaskingthe
wasemployed, initially,withonlythepowerlaw photonindex,Γ, soft emission characterized by the 0.5-keV component. The best
tiedacrossthethreedatasets.Theresultsofthisanalysisaresum- fitobtainedwhenthetemperatureofthelattercomponent wasal-
marizedinthefirstthreelinesofTable2,wherethelinestrengths lowedtovaryisreportedinTable2.Inthiscasethe1-keVemission
arequotedintermsoftheirequivalentwidth(EW)withrespectto is the cooler component with the hotter ∼ 3 keV emission con-
theunderlyingcontinuum.Heretheerrorsareat90%confidence, tributing51%oftheobserved0.5-2keVflux.Theinferredcolumn
exceptfortheupperlimitwhichisquotedat3σ. densityfor theMSsubgroup isapproximately threetimesthat of
From the results in Table 2, it isevident that the simple ab- the S subgroup. If we (naively) interpret this in terms of the MS
sorbed powerlaw pluslinesmodel provides areasonable descrip- sourcesbeing typicallyat threetimesthedistanceof thoseinthe
tionofthedata.Thehardnessratioselectioncriterionusedtode- Ssubgroup,thentheimplicationisthattheformerareonaverage
finethethreesourcesamplesisreflectedinthederivedNH which anorder ofmagnitude moreX-rayluminous(sincethesameflux
varies from ≈ 8 × 1022 cm−2 for the H subgroup through to thresholdappliesinthetwocases).ThehotteremissionoftheMS
≈ 3×1021 cm−2 fortheMDsubgroup. Thederived photonin- subgroup might therefore reflect higher levels of coronal-activity
dex, Γ = 1.55±0.07, demonstrates that the sources contribut- andtheprobableinclusionofsomeactivebinaries.
ingtothestackedspectrahave,onaverage,ratherhardcontinuum Giventhatcoronally-activestarsandbinarieslikelydominate
spectra. When the powerlaw continuum is replaced by a thermal theMSandSsubgroups,butcertainlynottheMHandHsubgroups
bremstrahlung continuum, a correspondingly high temperature is (see Paper I and §5), it is safe to assume that the MD subgroup
derived,kT= 19+6 keV,withonlymarginalchangesintheother includesbothsoftandhardpopulationobjects.Whenwereconsider
−4
spectralparametersandtheχ2ofthefit. thespectralfittingoftheMDsubgroup(inthiscaseintherestricted
Thepresenceofa6.7-keVFelineisafeaturecommontothe 1-5keVband),wefindthattheinclusionofa1-keVthermalplasma
The GalacticplaneatfaintX-rayfluxes 5
Figure3.Thebackground-subtracted spectraforthe5sourcesubgroupsH,MH,MD,MSandS,togetherwiththecompositespectrumforall138sources
(lowerrightcorner).Thedatabetween7.8and8.3keV(i.e.theregionoftheKαCufluorescencelinesinthedetector)havebeenexcluded.Thehorizontal
dottedlineindicatesthelevelofzeronetcountsperchannel.
Table2.Topsection:Thebest-fittingparametersfromthespectralfittingofanabsorbedpowerlawplusemissionlinesmodeltothespectraoftheH,MHand
MDsubgroups.Thecolumnsgivethederivedcolumndensity,NH,thephotonindex,Γ,theequivalentwidth,EW,ofthethreeFelinesandthereducedχ2ν
andnumberofdegreesoffreedomνforthefit.Lowersection:Thebest-fittingparametersfromthespectralfitstothespectraoftheS,MSandMDsubgroups,
incorporatingvariouscombinationsofthermalplasmaandpowerlawcomponents.ThecolumnslistNH,Γ,twoplasmatemperatureskTcoolandkThot,the
derivedmetalabundanceZ⊙relativetosolar.Thereducedχ2νandthenumberofdegreesoffreedomνforthefitarealsoquoted.
Group NH Γ EW(6.4keV) EW(6.7keV) EW(6.9keV) χ2ν/ν
×1022cm−2 eV eV eV
H 8.3+−00..65 1.55+−00..0077 94+−3300 149+−3354 55+−3343 1.16/517
MH 1.27+−00..1120 - 152+−5533 172+−5511 192+−6655 -
MD 0.34+−00..0076 - <88 187+−8991 56+−7536 -
Joint-EW - 1.54+−00..0077 89+−2265 170+−3352 81+−3209 1.20/523
NH Γ kTcool kThot Z⊙ χ2ν/ν
×1022cm−2 keV keV
S 0.057+−00..003242 - 0.51+−00..1132 1.01+−00..1046 0.13+−00..0022 1.17/159
MS 0.18+−00..0022 - 1.0(fixed) 3.6+−10..17 0.13(fixed) 0.85/160
MD 0.87+−00..0055 1.55(fixed) 1.0(fixed) - 0.13(fixed) 1.05/96
6 R. S.Warwick, K. Byckling,D. Pe´rez-Ram´ırez
000−3−3−3 theX-rayspectraofthesetwosourcesincomparisontothespec-
111
5×5×5× traoftwo’typical’sourcesdrawnfromtheHsubgroup, onewith
adetectable6.4-keV lineandtheother witha6.7-keV feature.If
weexcludethetwosourcesidentifiedabovefromtheMHsample,
000−3−3−3 theimpactistoreduce theEWvaluestabulatedfor theMH sub-
×1×1×1 groupinTable2byabout30%;however,suchanexclusionwould
222
eVeVeV besomewhatarbitrary.
kkk
s/s/s/
s/s/s/
ountountount 101010−3−3−3
ccc
5 NIRCOUNTERPARTSOFTHESOURCES
000−4−4−4 We have searched for potential NIR counterparts of the X-
111
5×5×5× ray sources in the current sample by cross-matching the XMM-
Newton source positions with the Two Micron All Sky Survey
(2MASS)2 (Cutrietal.2003; Skrutskieetal. 2006). Themethod-
555
EEEnnneeerrrgggyyy (((kkkeeeVVV))) ology for the cross-matching was essentiallythe same as that re-
Figure4.TheHsubgroupspectrumaroundtheiron-linecomplex.Thecon- portedinPaperI,wherefulldetailsareprovided.
tributiontothebest-fittingmodelofthepowerlawcontinuumandeachof Wefoundthatofthe59sourcesintheSandMSsubgroups,
theFelinecomponentsisshownasblackdottedlineswiththecombinedfit
showninred. 53(90%)haveabright(KS <14)2MASSstarwithina3σerror
radiusoftheX-rayposition(i.e.withinanerrorcirclewithradius
equal to 3 times the X-ray position error quoted in the 2XMMi
component(alongwiththeΓ=1.55powerlaw)givesasignificant ’slim’catalogue3).Thereare3furtherlikely2MASScounterparts
improvement to the fit (as measured by ftest routine in XSPEC). iftheerrorcirclesareextendedto3.25σ.Theremaining3sources
Thebest-fitparameters for thispowerlaw plus thermal model are include one instance where the 2MASS catalogue suffers from
provided as the last entry in Table 2. In this model, the thermal theconfusionofabrightnearbystarand,similarly,twoinstances
componentcontributesroughly20%ofthe1–5keVflux.Thusthe wheretheX-raypositionmaybemarginallyoffsetduetothecon-
MDsubgroupdoesseemtocompriseamixofsourcepopulations. fusionof anearby X-raysource. Inshort,virtuallyallof thesoft
Wenote, however, that thepresence of anumber of sources with X-raysourceshaveplausibleNIR-brightcounterparts,themajority
intrinsicallysoftspectraintheMDsubgroupisunlikelytohavean ofwhicharelikelytobenearby,coronally-activestarsandbinaries
undueimpactontheFe-lineEWmeasurements,sincesuchsources (seePaperI).
willcontributelittletoeitherthelinesortheunderlyingcontinuum The NIR associations for the harder spectral subgroups are
inthe6-7keVband. much less complete. For theMD subgroup, 13 sources out of 22
(59%)havebright(KS <14)2MASSobjectsinanominal3σX-
rayerrorcircle,whereasfortheMHandHsubgroupsthestatistics
4.2 Fe-linepropertiesoftheindividualsources are 10 out of 25 (40%) and 8 out of 32 (25%), respectively. The
average(area-weighted)3σerrorcircleradiusforthecurrentsam-
Inordertoinvestigatethecontributionofindividualsourcestothe
pleofsourcesis2.5arcsec.Utilisingasetofpositionsoffsetfrom
Fe-lineemission in the H, MH and MD subgroups, we extracted
eachX-rayposition,wefindthat,fortheseGalacticplanefields,the
thecountsrecordedforeachsourceinanarrowbandpass(ofwidth
correspondingchancecoincidenceratewith2MASSstarsbrighter
240eV)centredoneachFeline.Theleveloftheunderlyingcon-
tinuumappropriatetoeachmeasurementwasalsoestimatedbased thanKS =14 is15%.Fig.7summarisesthesestatistics.Theim-
plicationisthat,atleastwithintheMDandMHsubgroups,asize-
onalinearinterpolationofthecountsrecordedintwo‘continuum’
ableminorityofsourcesmayhaverealidentificationswithbright
bands(eachofwidth600eV)centredat5.94and7.54keV.Fig.5
NIRstars.
showsaplotofthenetlinecountsinthe6.4-,6.7-and6.9-keVlines
We have plotted a NIR two-colour diagram for the 2MASS
versusthehardnessratioforthefullsetofsourcesthatcomprisethe
starscontainedwithintheX-rayerrorcirclesinFig.8(butexclud-
threesubgroups.Inthisfigurethedatapointsplottedinredcorre-
ingsourceswithrelativelypoorphotometryforwhichthe2MASS
spondtothosesourcesforwhichthelinemeasurementrepresents
Qflag=Uinoneormorebands).Inthisdiagramthebulkofthe
a’detection’ at or above the 2.5σ significance level. Thenumber
objectsassociatedwithX-raysourcesintheSandMScategories
ofsourceswithdirectlydetectablelinesat6.4,6.7and6.9keVis
fall on the locus of late-type main sequence stars with relatively
2,6and1,respectively,withjustonesourceexhibitingmorethan
lowreddening.Incontrast,thestarslinkedtotheX-raysourcesin
asingleline(158770 - withtwinfeaturesat 6.7and 6.9keV). A
theMDthroughtoHsubgroupsarecharacterizedbyanincreasing
summaryoftheFe-linepropertiesofthese8sourcesisprovidedin
degreeofreddening.
Table6.
Wehavealsoplottedacolour-magnitudediagramforthesame
Inbroadterms,Fig.5confirmstheresultsofthespectralfitting
set of NIR objects in Fig.9(as the circular symbols). This figure
ofthesourcesubgroups.Inthecaseofthe6.7-keVline,thereisa
also shows the tracks of several different types of stellar objects
positivebiasintheEWmeasurements,particularlyforthesources
inthecolour-magnitudeplaneassumingthatthevisualabsorption,
withintheHandMHcategories(HR>0.2).The6.4and6.9keV
measurementsalsofitthisdescription. AV, in the Galactic plane increases at the rate of 2 mag kpc−1
Thereisasuggestion thattherelativelyhighlineEWsmea-
suredintheMHsubgroup (seeTable2)maybeduetotheinflu- 2 http://www.ipac.caltech.edu/2mass/
enceoftwosources-source168031whichhasarelativelybright 3 Inthecaseofmultipleobservations ofanX-raysource,theX-raypo-
6.4 keV line and, as noted above, source 158770 which exhibits sition and position errors quoted in the 2XMM ’slim’ catalogue is the
detectable line features at both 6.7 and 6.9 keV. Fig. 6 illustrates weightedaveragevalueacrossallthedetectionsofthesource.
The GalacticplaneatfaintX-rayfluxes 7
Figure5.Thenetcountsofthe6.4-keV(toppanel),6.7-keV(middle)and6.9-keV(bottompanel)Felinesversusthehardnessratioforeachsource.Thered
datapointsrepresentthesourceswithasignificant(2.5σ)numberofnetcountsintheline(seeTable3).
Table3.The8individualsourcesinwhichoneormoreoftheFe-lines(at6.4,6.7and6.9keV)wasdetectedabove2.5σ.Thetableliststhenetcountinthe
threelinesforeachsourceandalsoquotestheEWifthelinewasdetectedabove2.5σ.
Source 212279 148374 168031 158770 166494 151081 217813 157004
HR 0.01 0.13 0.48 0.75 0.84 0.95 0.97 1.00
Netcts(6.4keV) -0.8±3.0 3.2±3.0 15.0±6.0 5.3±4.9 -0.8±3.0 2.7±5.3 16.5±6.3 3.3±4.7
EW(eV) – – 674±269 – – – 244±93 –
Netcts(6.7keV) 12.6±4.2 11.3±4.1 5.3±5.0 15.4±5.7 10.8±4.3 15.9±6.2 2.6±4.7 24.7±6.7
EW(eV) 1218±409 2945±1066 – 508±188 1210±477 329±127 – 635±171
Netcts(6.9keV) 0.1±3.0 8.6±3.7 3.4±4.8 18.6±5.7 -1.6±3.0 3.8±4.7 6.2±5.0 2.0±4.4
EW(eV) – – – 686±210 – – – –
8 R. S.Warwick, K. Byckling,D. Pe´rez-Ram´ırez
Figure6.Themeasuredsourcespectrumnetofbackgroundforfoursources.Thespectraldatahavebeengroupedinto40eVchannelsandsmoothedwitha
Gaussianlinespreadfunctionofwidthσ=60eV.Theverticaldashedlinesindicatethe6.4(red),6.7(green)and6.9keV(blue)bandpasses.Thedatabetween
7.8and8.3keV(i.e.,theregionoftheKαCufluorescencelinesinthedetector)havebeenexcluded.Toppanels:TwosourcesfromtheMHsubgroupwith
detectablelinesat6.4keV(168031)and6.7and6.9keV(158770).Bottompanels:Hsubgroupsourceswithlinefeaturesat6.4keV(217813)and6.7keV
(157004).
Figure 7. The fraction of sources with a bright 2MASS star contained Figure8.TheH−KSversusJ−Hcolour-colourdiagramofthebrightest
withintheX-rayerrorcirclefortheMD,MHandHsubgroups.Thechance 2MASSstarcontainedwithintheX-rayerrorcircle.X-raysourcesintheS
rateisindicatedbythedottedline. andMSsubgroupsareshownastheblacksymbols,whereasthoseinthe
MD,MHandHsubgroupsareplottedinred,greenandblue,respectively.
The solid diagonal line illustrates the reddening vector corresponding to
(equivalenttoAJ = 0.56magkpc−1;AK = 0.224magkpc−1). AV = 25,(AJ = 7.0;AH = 4.3;AK = 2.8),fromastartingpoint
WeuseacompilationofNIRmagnitudesforCVs(Aketal.2007, representativeoftheNIRcoloursofanunreddenedMOVstar.
2008)toestimatetheabsolutemagnitudeandintrinsiccolourtypi-
The GalacticplaneatfaintX-rayfluxes 9
Figure9.TheJ−KSversusKScolour-magnitudediagramofthebrightestNIRstarscontainedwithintheX-rayerrorcircle.Theresultsfromthecomparison
with2MASSareshownasthecircularsymbolswithcolourcodingindicativeoftheX-raysubgroupasfollows-SandMS(black),MD(red),MH(green)
orH(blue).ThediamondsymbolsshowthesameinformationwhenthestudyisextendedtoUKIDSSforasubsetofthesources-seethetext.Thecluster
ofbrownpointsrepresentsknownCVswhentheircoloursandmagnitudesaretranslatedtoastandarddistanceof2kpc.Thecurvesillustratethetracksof
differenttypesofstarasthedistanceisvariedfrom1pc-20kpc,assumingthatAV intheGalacticplaneincreasesattherateof2magkpc−1.Thecurves
becomesolidlinesforstellardistancesintherange200pc-2kpc,whichcorrespondstoanX-rayluminosityintherange1030−32 ergs−1,assuminga
limitingX-rayfluxof2×10−13ergcm−2s−1(2–10keV).Browncurve:aluminousCVwithMK=5.0andintrinsicJ−KS=0.5;Yellowcurve:aFOV
dwarfstarwithMK =2.25,J −KS =0.18;Redcurve:aM6VdwarfwithMK =6.74,J −KS =0.98;Purplecurve:aK3IIIgiantwithMK =-2.03,
J−KS=0.90;Bluecurve:aBOIsupergiantwithMK=-6.44,J−KS=-0.15.
calofaCV.Similarly,theNIRpropertiesofthedwarf,giantandsu- caseofmassivebinaries,inthecollidingwinds.Asisevidentfrom
pergiantstarsaretakenfromCoveyetal.(2007).Thisfigureillus- Fig.9,inthiscaseaKS magnitudeof10orfainterimpliesahard-
tratesratherclearlythatiftheassociationofanX-raysourcewitha band X-ray luminosity considerably greater than 1032 erg s−1,
highlyreddened,butrelativelybrightNIRstar(with,say,J−KS > which points to either a binary (or multiple) massive star system
1.0 and KS < 14) isreal, then it isunlikely that we aredealing or the presence of an accreting high-mass X-ray binary (Paper I;
witheitheradistantluminousCVoraverycoronally-activedwarf Mauerhanetal.2010;Andersonetal.2011andreferencestherein).
(F-M)star.Thebasicpointisthatifweinterpretahighdegreeof
reddeningasindicativeofarelativelylargedistance(i.e.assuming Fig.9 also shows a cluster of points representative of the
thatthereddeningisnotafeatureoftheverylocalenvironmentof coloursandmagnitudesofknownCVs,whentranslatedtoacom-
thesource)thenratherfaintmagnitudesareimpliedfordwarfstars. mondistanceof2kpc,basedonthecompilationofCVNIRmag-
It is much more tenable that the association is witheither a late- nitudesanddistancesreportedbyAketal.(2007,2008).Asnoted
typegiantorabrightsupergiant(e.g.Mauerhanetal.2009,2010; above for a distance of 2 kpc, our hard band limiting flux corre-
Motchetal. 2010; NebotGo´mez-Mora´netal. 2013). The former spondstoanX-rayluminosityof≈1032ergs−1.Clearly,inorder
categoryincludesRSCVnbinariescontainingaG-orK-typegi- toidentifysuchapopulationofsourcesamongstoursampleofhard
anttwinnedwithasub-giantormainsequencecompanion,which X-ray sources, one would need to go tosignificantly fainter NIR
arecharacterizedbyenhanced,high-temperaturecoronalemission magnitudesthanareprobedby2MASS.Typicallysuchconterparts
drivenbytidalinteractions(e.g.Agrawal,Riegler,&White1981). wouldhaveKS > 15, butbesubject toonlyamodest degreeof
If,forexample,weobserveanRSCVnsystemcontainingaK3III interstellarreddening.Theproblem,ofcourse,isthatintheGalac-
giantatKS =10,thentheimplieddistanceis≈2kpc.Foralim- ticplaneatthesefaintIRmagnitudes,thechanceincidencerateof
itingfluxinthehardbandof,say,2×10−13 ergcm−2s−1(2–10 stars may become excessive, at least when using the typical few
keV),theimpliedX-rayluminosity,LX,is≈1032ergs−1,which arcsecerrorcirclesavailablefromXMM-Newton.
isplausibleforthemostactivesystems.IfthecounterparttotheX-
raysourceisamassiveearly-typestarsuchasaWolf-Rayetstaror Bywayofanexperiment,wehaveselectedasubsetofthehard
anOB-supergiant,thenasignificantX-rayfluxabove2keVmay X-raysources(withintheMD,MHandHsubgroups)whichhave
beproducedinshockedregionswithinanunstablewindor,inthe relativelysmallX-rayerrorcircles(3σerrorcircleradius6 1.5′′)
andforwhichdeepNIRimagingisavailablefromtheUnitedKing-
10 R. S.Warwick, K. Byckling, D. Pe´rez-Ram´ırez
dom Deep Sky Survey (UKIDSS)4(Lawrenceetal. 2007). Of the complexofiron-KlinesincludingstrongHe-like,H-likeandfluo-
18 X-ray sources meeting this criterion, we have found possible rescentcomponents,thelatterarisingfromtheilluminationofthe
NIRassociationsin9cases.Theseareshownasthediamondsym- surfaceofthewhitedwarfand/ortheouterregionsoftheaccreting
bols in Fig.9. Two of these NIR objects (both linked to H sub- flowbythehardcontinuum.
groupsources) arelocatedintheregionofthediagramwherewe In a study of 20 magnetic CVs observed by ASCA,
might expect to find X-ray luminous CVs. Interestingly a search Ezuka&Ishida(1999)foundthecharacteristictemperatureofthe
ofthecatalogues heldatVizieR5 showsthatoneof theseobjects underlying hard continuum to be typically kT ∼ 20 keV. Al-
isaknownIntermediatePolar(alsoknownasSAXJ1748.2-2808; though with a large source-to-source scatter, the average EW of
Ritter&Kolb 2003; Sidolietal. 2006; Nobukawaetal. 2009). A the 6.4-keV, 6.7-keV and 6.9-keV lines across this CV sample
VizieR search on the full sample of hard X-ray sources also re- were roughly ∼ 100 eV, ∼ 170 eV and ∼ 100 eV, respec-
vealed a handful of other likely identifications, including several tively.SimilaraverageEWswerereportedintheparallelstudyby
high reddened sources near the Galactic Centre previously stud- Hellier,Mukai&Osborne(1998).Morerecently,Hellier&Mukai
iedbyMauerhanetal.(2009),twoLMXB(Liuetal.2007),aBe- (2004)investigatedtheX-rayspectraof5magneticCVsobserved
starHMXB(Ebisawaetal.2003)andapotentialAMHERobject at high spectral resolution with the Chandra High-energy Trans-
(Motchetal.2010). mission Grating (HETG). Withinthis limited sample the average
In summary, our cross-correlation with NIR catalogues EWswere∼ 120 eV, ∼ 160 eV and∼ 110 eVfor the6.4-keV,
demonstrates that virtually all thesoft X-raysources inour sam- 6.7-keV and 6.9-keV lines, respectively. Finally, Bernardinietal.
ple(comprisingtheSand MSsubgroups) canbeassociated with (2012)report thedetection of intenseFe-Kαlineemissioninthe
a bright late-type stellar counterpart possessing an active stellar X-rayspectraof9recentlyidentifiedIPsandquotetheEWofthe
corona. The nature of the Galactic population or populations un- 6.4-keVfluorescentline,inallcases,tobeintherange130-220eV.
derlying the hard X-ray sources (comprising subgroups H, MH Inourearlieranalysis(§4),wefoundthattheemissionspec-
and MD) is less clear-cut, although highly coronally-active sys- traof the three hardest subgroups could be represented as ahard
tems such as RSCVn binaries, wind dominated objects such as continuum plus Fe-lines. We modelled the hard continuum as a
Wolf-Rayet starsand OB-supergiants and accretion-powered sys- powerlaw of photon index Γ = 1.55 but obtained equivalent re-
temsrangingfromCVs,throughtoLMXBandHMXBs,allcon- sultswhenthepowerlawwasreplacedbyathermalbremstrallung
tributetothemix.Extragalacticinterlopers,predominantlyActive continuumwithkT≈ 20keV.TheaverageEWoftheHe-likeFe-
Galactic Nuclei (AGN) seen through the high column density of lineat6.7keVwas170+35 eV.Similarly,theEWsofthe6.4-keV
−32
the Galactic plane may also represent a significant (maybe 50%) and6.9-keV Fe-lineswere89+26 eVand81+30 eV,respectively,
−25 −29
fraction of the hardest subgroup of sources (Paper I; Handsetal. i.e.roughlyhalfthatofthe6.7-keVline.Evidently,thesespectral
2004).However,ouranalysisdoesprovideahintthatanappropri- characteristics are fully consistent with those of known magnetic
atecombinationofsub-arcsecX-raypositionsanddeephighreso- CVs.
lutionNIRimaginghasthepotentialtorevealanincreasingnumber The inference is that a sizeable fraction of our current hard
ofCVcounterparts,atleastinsurveysreachingintermediaterather bandsamplemaybemagneticCVs,albeitwithcounterpartswhich
thanverydeepX-rayfluxthresholds. are in general too faint to pick out reliably from available NIR
catalogues.However,itisnecessarytoconsiderwhatimpactnon-
CVinterlopersmighthaveonthemeasurementsderivingfromthe
stacked (i.e. sample-averaged) spectra. For example, we noted in
6 DISCUSSION
§5 that, based on source count estimates, AGN might contribute
Currently,theleadingcandidatefortheunresolvedsourcepopula- upto50%ofthehardestsubgroupofsources.Compton-thinAGN
tioncontributingtotheGRXEismagneticCVs,nearbyexamples are characterized by powerlaw continuum spectra with Γ ≈ 1.7
ofwhichtypicallyhave2–10keVX-rayluminositiesintherange plusaprominentFefluorescencelinewithanEWtypicallyinthe
from1031–1034ergs−1(e.g.Ezuka&Ishida1999;Sazonovetal. range20-120eV(seeFukazawa2011;Chaudharyetal.2012and
2006;Revnivtsevetal.2008).Inprinciplethisincludesbothpolars referencestherein).ClearlyanydilutionoftheCVcontributionby
(wherethespinandorbitaresynchronizedornearlysynchronized) AGNswould tend toreduce theEW of the6.7-keV and 6.9-keV
and Intermediate Polars (IPs) (where the orbital period is much lines,whilsthavingrelativelylittleimpactonthe6.4-keVmeasure-
longerthanthewhitedwarfspinperiod).However,theplasmatem- ments.Thereisperhapsslightevidenceforthisfromacomparison
peraturesofpolarsaresomewhatlowerthaninIPsduetoenhanced of the results from the three hardest subgroups (on the presump-
cyclotroncooling(e.g.Cropperetal.1998)anditislikelythatIPs tionthatthehighcolumndensitythroughtheGalacticdiscplaces
contributemosttothehardX-ray(>5keV)volumeemissivity. AGNpreferentiallyinthehardestsubgroup);howeveritseemsun-
In magnetic CVs, an accretion shock is generated above the likelythatthishasintroduced astrongoverall biasinour Fe-line
whitedwarfsurface,whichheatstheaccretingmaterialtotemper- EWmeasurements.
atures kT > 15 keV. The resulting highly-ionised, optically-thin Having demonstrated, on spectral grounds, that magnetic
plasma cools in the post-shock flow and, eventually, settles onto CVsvery likelyrepresent asignificant Galacticpopulation at the
thewhitedwarfsurfacethroughanaccretioncolumn.Theresulting intermediate to faint X-ray fluxes encompassed by our XMM-
X-rayspectrum comprisesacomplex blend of contributions gen- Newton source sample, what are the implications for the ori-
eratedatdifferingtemperatures,densitiesandoptical depths(e.g. gin of the GRXE? In a recent study of the distribution of K-
Cropperetal. 1999; Yuasaetal. 2010). Aswell asa hard contin- shelllineemissionalongtheGalacticplaneobservedbySuzaku,
uum, the X-ray spectra of magnetic CVs are characterized by a Uchiyamaetal. (2013) find that the average EW of the 6.4-keV,
6.7-keV and 6.9-keV linesassociated withtheGRXE are≈ 110
eV, ≈ 490 eV and ≈ 110 eV (see also Ebisawaetal. 2008;
4 http://www.ukidss.org/ Yamauchietal. 2009). As noted by Uchiyamaetal. (2013) and
5 http://vizier.u-strabg.fr/viz-bin/Vizier/ other authors, although the 6.4-keV and 6.9-keV line EWs mea-