VLT spectroscopy and non-LTE modeling of the C/O-dominated accretion disks in two ultracompact X-ray binaries PDF
Preview VLT spectroscopy and non-LTE modeling of the C/O-dominated accretion disks in two ultracompact X-ray binaries
Astronomy&Astrophysicsmanuscriptno.aa (cid:13)c ESO2008 February5,2008 VLT spectroscopy and non-LTE modeling of the C/O-dominated ⋆ accretion disks in two ultracompact X-ray binaries 6 0 K.Werner1,T.Nagel1,T.Rauch1,N.J.Hammer2,1,andS.Dreizler3 0 2 n 1 Institutfu¨rAstronomieundAstrophysik,Universita¨tTu¨bingen,Sand1,72076Tu¨bingen,Germany a 2 Max-Planck-Institutfu¨rAstrophysik,Karl-Schwarzschild-Straße1,85741Garching,Germany J 3 Institutfu¨rAstrophysik,Universita¨tGo¨ttingen,Friedrich-Hund-Platz1,37077Go¨ttingen,Germany 4 2 Receivedxxx/Acceptedxxx 1 ABSTRACT v 6 Aims.Wepresentnewmedium-resolutionhigh-S/Nopticalspectraoftheultracompactlow-massX-raybinaries4U0614+091and4U1626-67, 4 takenwiththeESOVeryLargeTelescope.TheyarepureemissionlinespectraandthelinesareidentifiedasduetoC-andO-. 5 Methods. Line identification is corroborated by first results from modeling the disk spectra with detailed non-LTE radiation transfer 1 0 calculations.Hydrogenandheliumlinesarelackingintheobservedspectra. 6 Results.OurmodelsconfirmthedeficiencyofHandHeinthedisks.ThelackofneonlinessuggestsanNeabundanceoflessthanabout10 0 percent(bymass),however,thisresultisuncertainduetopossibleshortcomingsinthemodelatom.Thesefindingssuggestthatthedonorstars / areerodedcoresofC/Owhitedwarfswithnoexcessiveneonoverabundance.ThiswouldcontradictearlierclaimsofNeenrichmentconcluded h fromX-rayobservationsofcircumbinarymaterial,whichwasexplainedbycrystallizationandfractionationofthewhitedwarfcore. p - o r Keywords. Accretion,accretiondisks–Binaries:close–X-rays:binaries–Stars:individual:4U0614+091,4U1626-67 t s a : v 1. Introduction tionforstudyingthesesystemsisthatthestrippeddonorstars i X offerthepossibilitytoprobetheinteriorcompositionofwhite Low-mass X-ray binaries (LMXBs) consist of a neutron star dwarfs,whichdependsontheinterplayofgravitationalsettling r a or black-hole accretor and a low-mass donor star (M<∼1M⊙). andcrystallizationofchemicalelements. Of particular interest are those systems with orbital periods Three of these five Ne-rich systems belong to the above- P <80min, which is the minimum period for LMXBs with orb∼ mentioned eight UCBs with measured orbital periods, while hydrogen-rich main sequence donors. In these ultracompact two of themare believedto beUCBs becauseof theirsimilar binaries (UCBs) the mass donor must be a non-degenerate opticalandX-rayproperties.TheclassofNe-richUCBscon- hydrogen-deficientstar or a white dwarf (e.g. Verbunt & van sistsof: den Heuvel 1995). Currently eight such systems with mea- sured orbital periods (Porb=11-50min) are known (Ritter & (1)4U1626-67 (Porb=41min) Kolb2003). (2)4U0614+091 In the recent past, the existence of a group of five ultra- (3)2S0918-549 compactsystemswithneon-richwhitedwarfdonorshasbeen (4)4U1543-624 (P =18min,Wang&Chakrabarty2004) orb claimedbasedonX-rayspectralproperties(Schulzetal.2001, (5)4U1850-087 (P =20min) orb Juett etal. 2001, Juett & Chakrabarty 2003). The donors are thenC/OorevenO/Ne/Mgwhitedwarfsthathavetransferred And4U1626-67mayberegardedastheprototypeofthisclass. a significantfractionoftheir mass to the neutronstar, in con- Thedonor’sNe-richC/O-WDnatureisderivedfromX-rayand trasttotheusualwisdomthatthedonorsaretheremainsofHe ultraviolet spectra that exhibit double-peaked emission lines whitedwarfs.Thishascausednewexplorationsoftheforma- thatobviouslystemfromtheaccretiondisk(Schulzetal.2001, tionofthesesystems(e.g.Yungelsonetal.2002).Ourmotiva- Homeretal.2002). The close relation of the other four objects (2)–(5) to 4U1626-67wasbasedonextraordinaryhighNe/Oabundance Sendoffprintrequeststo:K.Werner ⋆ BasedonobservationsmadewithESOTelescopesattheParanal ratios (when compared to the ISM value) measured from ObservatoryunderprogrammeID72.D-0013(A). ASCA spectra of (3)–(5)and a Chandraspectrumof (2). The Correspondenceto:[email protected] spectraexhibitphotoelectricabsorptionedgesofneutralOand 2 K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 OIII OIII CIII CIII CII 1.5 i.s. CIII x u i.s. DIB e fl DIB 4U0614+091 v ati el r OII OII OII 1.0 CII 4U1626-67 OII OII OII OII OII 3600 3800 4000 4200 4400 4600 4800 5000 5200 5400 o wavelength / A Fig.1. Blue spectra of 4U0614+091and 4U1626-67.H and He lines are lacking. The emission lines are identified as O- andC-(seealsoTable2).Allabsorptionfeaturesareofinterstellar(i.s.)origin:Ca3934/3968Å,the4300ÅCHband,and twodiffuseinterstellarbands(DIB).Plottedoverthetwoobservedspectraisasyntheticaccretiondiskspectrumreddenedwith E(B-V)=0.55andE(B-V)=0.40,respectively. Ne in the interstellar medium (ISM) along the line-of-sight, In conclusion, the only way to confirm that the four sys- which is suspected to originate from expelled material close tems(2)–(5)–besidestheprototype(1)–indeedcontainC/O- to the binary systems. New X-ray spectroscopic observations rich donors,perhapsenriched with Ne, is by UV and/oropti- of(3)and(4)obtainedwithChandraandXMM-Newtoncon- calspectroscopy.Nelemansetal. (2004)show thatthe optical firmed the ASCA results, although different values for the spectra of (2)–(4) are devoid of hydrogen and helium emis- Ne/O enrichment have been obtained from a Chandra and a sion lines, which are usually seen in H- or He-rich accreting XMM-Newton spectrum of (4) (Juett & Chakrabarty 2003). systems. Their spectra exhibit low-ionisation C and O emis- In contrast to the ASCA measurement of (5), recent XMM- sion lines, most prominent in the brightest of these objects, NewtonandChandraobservationsfoundnoevidenceofanun- 4U0614+091.Inthispaperwe presentnewopticalspectraof usualNe/Oratio(Sidolietal.2005,Juett&Chakrabarty2005). thissystem thatcovera largerwavelengthinterval.Theycon- These apparently contradictory results can be attributed to a firm the earlier conclusion by Nelemans etal. (2004) that the variableNe/Oratioduetochangesintheionisationstructure emission lines arise from the C/O-dominated material that is inthemeasuredabsorptioncolumnsthat,however,arenotun- probably located in the accretion disk and not in the X-ray derstood (Juett & Chakrabarty 2005). Hence, this means that heated face of the white-dwarf donor. We also present a first the measured Ne/O ratio does not reflect the donor composi- detailed opticalspectrumof the prototype4U1626-67,which tion. Although the orbital periods of these systems are below mainly shows weak oxygenemission lines and provesthe H- 80min(oratleastbelievedtobe sosmall), thisdoesnotnec- and He- deficiency in this system, too. Nelemans & Jonker essarily meanthat theycontain C/O donors.For example,the (2005) also performed VLT observations of this system (also X-rayburstpropertiesofthe4U1820-30(P =11min)suggest in spring 2004) with a similar setup. They show a section of orb anHe-WDdonorinthatultracompactsystem(e.g.Strohmayer theirspectrumandemphasizethesimilaritywith4U0614+91. &Brown2002).AnotherexamplearetherelativelyshortX-ray Inadditionwepresentresultsfromfirstattemptstomodel bursts observed from (3) that even suggest H-burning of ma- theobservedspectrawithnon-LTEaccretion-diskmodels.We terialaccretedonto the neutronstar (NS) (Jonkeretal. 2001), derive upper limits for the H and He abundances and inves- whileopticalspectrasuggestaC/OWDdonor(Nelemansetal. tigate the formation of neon lines. Detailed C and O line- 2004and thiswork).A possible explanationis that the heavy formationcalculationscanalreadyqualitativelyexplaintheob- elements(C,O,Ne)undergospallationduringaccretionleav- served emission lines. The ultimate goal is to derive detailed ing H and He nuclei for thermonuclear burning on the NS abundancesand other disk parametersfrom the observedline (Bildstenetal.1992).Ina recentpaperIn’tZandetal.(2005) profiles. concludethattheNe/OratioandtheX-rayburstpropertiesare Thepaperisorganisedasfollows.Wedescribeourobserva- allbestexplainedwithanHe-richdonor. tionsinthefollowingsection.Wethenpresentourlineidenti- K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 3 CIII OIII OIII CII CIV CII OII CII CII 1.0 x DIB CII x u e fl 4U0614+091 v i.s. ati el x r 0.5 OII 4U1626-67 CII CIII OII CIII 5600 5800 6000 6200 6400 6600 6800 7000 7200 7400 o wavelength / A Fig.2. Red spectra; 4U0614+091 shows emission lines from C- and O-, while the 4U1626-67 spectrum is virtually featureless.Allabsorptionfeaturesareeithertelluric(markedbyhorizontalbars)orofinterstellarorigin(Na5890/5896Åand aDIBat5780Å).Thefeaturesat5575Åareartifacts(“x”)duetoanOskyemissionline.Overplottedisthesyntheticaccretion diskspectrumlikeinFig.1. ficationsinSect.3.Section4containsadescriptionofourdisk completeUV/opticalwavelengthrange(1150–10000Å).The modelcalculationsandwesummariseandconcludeinSect.5. FUV spectrum is described in detail by Homer etal. (2002) and a first comparison with synthetic accretion disk spectra was presented by Werner etal. (2004). In Fig.3 we compare 2. Observations the opticalHST spectrumtakenfromthe MAST archivewith We obtained optical medium-resolution long-slit spectra of ourVLTspectrumonanabsolutefluxscale.Theblueandred 4U1626-67and4U0614+091.TheV magnitudeofbothsys- sections of the VLT spectrum were scaled by a factorof 1.65 temsis18.5(Ritter&Kolb2003).WeusedtheFORS1spectro- and 1.45, respectively, to normalise them to the HST flux at graphattachedto UT1ofESO’s VeryLargeTelescope(VLT) 5600Å.ItisobviousthattheVLTspectrumisnotonlyweaker onParanalinChile.Theslitwidthwas1′′.Weusedtwogrisms but also flatter than the HST spectrum.This can be attributed (600Band 600R, the latter in combinationwith ordersepara- eithertofluxcalibrationproblemsortosourcevariability. tion filter GG435) and obtained spectra covering the regions 3600–6000Åand5400–7500Åwithameandispersionof1.20 3. Lineidentification and 1.07 Å pix−1, respectively. Observations were performed in servicemodebetweenNov.2003and Mar. 2004.Eachtar- Emissionlinesfromhydrogenandheliumarecompletelylack- get was observed at least twice, see Table 1 for details. For ing in the VLT spectra of both systems. We detect neither 4U1626-67 the exposure time covers almost one orbital pe- BalmerlinesnorlinesfromHe(e.g.4471or5876Å)orHe riod so that any respective spectral variation is smeared out. (e.g.4686Å), whichare usuallyseen in opticalspectraof in- The data were processed throughthe standard ESO reduction teractingbinaries.Inthefollowingtwosectionswediscussthe pipeline.Inadditionweperformedafluxcalibrationusingob- linespectraofbothsystemsindetail.Lineidentificationisper- servations of the DA white dwarf EG274 with the same in- formed“byeye”withoutanequivalent-widthestimation. strumental setup. Since this flux standard was observed only once per wavelengthrange and since the science targetswere 3.1.4U0614+091 mostly observedat differentdates, this providesonly a rough absolutefluxcalibration.Foreachobjectthespectrawere co- Thespectrumof4U0614+091showsnumerousemissionfea- added to obtain one final spectrum. The blue and red spectra tures that can be assigned to ionized carbon and oxygen, aredisplayedinFigs.1and2,respectively. namely C- and O-. The detected emission features are Archival spectra of 4U1626-67 taken with the Hubble listed in Table2 together with possible line identifications. Space Telescope (HST) and the STIS instrument cover the Mostfeaturesareblendsoflinesfromatleasttwoions,making 4 K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 Table 2.Emissionfeaturesobservedinthe spectraof4U1626-67and4U0614+091andsuggestedline identifications.A plus signinparenthesesdenotesanuncertaindetection. Feature Observedin Ion Transition Wavelength (Å) 4U0614+091 4U1626-67 (Å) 3720 + + O 3p3P–3d3Do 3704–3732 O 3s5P–3p5Do 3695–3735 O 3s4P–3p4So 3713–3749 3815 + O 3p2Po–4s2P 3794–3830 O 3p1D–3d1Po 3817 O 3s3Po–3p3D 3755–3811 3920 (+) C 3p2Po–4s2S 3919,3921 4075 + + O 3p4Do–3d4F 4070–4111 C 4f3Fo–5g3G 4068–4070 O 3s3P–3p3Do 4073–4119 4180 + C 4f1Fo–5g1G 4187 C 2p3p3D–2s5f3Fo 4153–4163 4345 + + O 3s4P–3p4Po 4317–4367 O 3s’2D–3p’2Do 4347–4351 4410 + (+) O 3s2P–3p2Do 4415–4452 4650 + + C 3s3S–3p3Po 4647–4651 C 3s3Po–3p3P 4651–4674 O 3s4P–3p4Do 4639–4696 4710 + + O 3p’2Po–3d’2P 4691–4702 O 3p2Do–3d2F 4699–4742 O 3p’2Do–3d’2F 4698–4703 O 3p2Do–3d4D 4710–4753 4745 (+) C 2s2p22P–2s23p2Po 4735–4747 4940 + (+) O 3p2Po–3d2D 4941–4956 O 3p4So–3d4P 4891–4925 5590 + (+) O 3s1Po–3p1P 5592 5650 + C 3s4Po–3p4S 5641–5662 5695 + C 3p1Po–3d1D 5696 5810 + O 2p42P–3p’2Do 5754–5810 C 3s2S–3p2Po 5801,5812 C 2s4d1D–2p3d1Fo 5826 6100 + C 3p2P–3d2Do 6095–6103 O 3p2Po–3s2S 6081,6103 O 4s4P–3s4So 6113,6153 6150 + C 2s4d3D–2p3d3Do 6148–6164 C 4d2D–6f2Fo 6151 O 4s4P–3s4So 6113,6153,6214 6460 + C 4f2Fo–6g2G 6462 6580 + C 3s2S–3p2Po 6578,6583 O 3d2F–4p2Do 6496,6565,6571 6730 + C 4d2D–6p2Po 6723 C 3p4D–3d4Do 6725–6755 C 3s3Po–2p3D 6727–6773 O 3s2P–3p2So 6641,6721 O 3d2P–4p2Po 6667,6678,6718 O 4d3Do–3s3D 6707–6768 6790 + C 3s4Po–3p4D 6780–6812 C 2p3d3Po–2s6s3S 6775–6793 7235 + C 3p2Po–3d2D 7231–7237 K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 5 HST 3 x 1 o-A 1 -s 2 -m2 c g r x e 6 1 -0 1 ux / VLT x fl1 4000 4500 5000 5500 6000 6500 7000 o wavelength / A Fig.3.CompleteVLTspectrumof4U1626-67comparedtoaspectrumtakenwithHST.TheVLTspectrum,scaledtotheHST spectrumat5600Å,isflatterpossiblyduetoproblemsinabsolutefluxcalibration.TheHST spectrumhasa poorerresolution andS/N-ratio,althoughsomeemissionfeaturesatλ< 5000Åcanberecognizedinbothdatasets.Artificialemissionspikesare markedby“x”.Overplottedisthesyntheticaccretiondiskspectrum,likeinpreviousfigures,butnowreddenedwithE(B-V)=0.20 tofitthecontinuumshapeoftheHSTspectrum. CII CII OIII 1.0 x CII e flu 4U0614+091 CII v ati el r 0.7 OII CIII 6400 6500 6600 6700 6800 6900 7000 7100 7200 7300 o wavelength / A Fig.4.Detailfromthreesinglespectraof4U0614+091.Thickline:SpectrumtakenonNov.03,2003.Twothinlines:Spectra taken consecutivelyon Dec. 15, 2003(see Table 1). The strength of the two strongestemission features, the C multiplets at 6580Åand7235Å,showanincreaseoveratimeintervalofseveralweeks. interpretationquitedifficult,butsomeofthemareprobablyun- anditispossiblethattheemissionstrengthofthestrongestC blended.Thesearethefeaturesat5590Å(O),5650Å(C), lines is variable on a time scale of weeks. In order to assess 5695Å(C),and7235Å(C).Theemissionfeaturesat4710 the significance ofthis variability,we simulatedseveraltimes and 4940 Å are perhaps exclusively due to O, but they are with different noise the strongest emission features observed blends of four and two multiplets from this ion, respectively. onNov.3andDec.15,applyingtheS/NratiooftheVLTspec- Nelemansetal.(2004)presentopticalspectraofthisstartaken tra. We find a 2σ probability that the variability is real. The withVLT+FORS2withslightlydifferentwavelengthcoverage linewidthscorrespondtoaprojectedrotationalvelocityofthe (4620–8620Å). We essentially confirm the detection of their order1200kms−1. emissionfeaturesandlineidentifications,however,ourspectra extendto shorterwavelengths,downto3600Å, whichallows ustodetectsomeadditionalemissionlines.Ontheotherhand, 3.2.4U1626-67 wedonotseefouroftheweakemissionfeaturesidentifiedin theNelemansetal.(2004)spectra(atλλ5140,5190,5280,and The spectrum of 4U1626-67 clearly shows less features than 6070Å).Thiscouldbethe consequenceofline variability.In does that of 4U0614+091. Only five emission lines can be Fig.4wecompareoursinglespectraof4U0614+091indetail, detected in the blue spectrum (see Table2) and they are seen in both binaries. The red spectrum of 4U1626-67is virtually 6 K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 Table1.Logoftheobservations. Table3.Summaryofmodelatomsusedinthediskmodelcal- culations.ForeachionwelistthenumberofNLTElevelsand Date,UT Grism T Airmass Seeing thenumberoflinetransitions.Inbracketswegivethenumber exp (sec) (arcsec) of lines in the wavelength range covered by our optical spec- traafterfine-structuresplittingfordetailedline-profilecalcula- 4U0614+091 tions. 2003Nov.04,07:22:08 600B 1735 1.22 0.73 2003Nov.04,07:54:57 600B 1735 1.20 0.78 Element Ion NLTElevels Lines 2003Nov.03,07:40:17 600R 1735 1.21 1.00 2003Dec.15,06:25:42 600R 1735 1.28 0.52 H I 10 45(8) 2003Dec.15,07:00:31 600R 1735 1.37 0.57 II 1 – 4U1626-67 He I 29 61(16) 2004Mar.23,08:22:00 600B 1735 1.37 0.76 II 14 91(14) 2004Mar.23,09:01:42 600B 1735 1.36 0.77 III 1 – 2004Mar.21,08:48:53 600R 1735 1.36 1.58 C I 7 4(0) 2004Mar.22,08:19:40 600R 1735 1.37 0.87 II 38 160(42) 2004Mar.22,09:00:00 600R 1735 1.36 0.71 III 58 329(143) 2004Mar.23,07:41:51 600R 1735 1.40 0.57 IV 9 17(2) EG274 V 1 – 2004Mar.23,09:48:17 600B 50 1.05 0.67 O I 1 – 2004Mar.23,09:53:25 600R 40 1.05 0.67 II 29 82(56) III 36 42(47) IV 11 5(0) continuous.ThestrongCemissionsseenin4U0614+091at V 6 4(0) 6580Åand7235Åarelackingin4U1626-67.Thatcouldpoint VI 1 – to a higher ionisation of carbon in 4U1626-67, although the Ne I 3 0 C 5810 Å line is lacking, too. The presence of O lines in II 68 232(49) thebluespectrumof4U1626-67(e.g.at4345Å)alsocontra- III 4 0(0) dictsa higherionisation,rather,itappearsthatthedifferences IV 1 – in the spectra of both binaries could be assigned to a differ- entC/O ratio,beinghigherin4U0614+091.Theentiresetof emissionlinesin4U1626-67couldbeduetooxygenalone,but intheUVspectrumCisclearlypresent(Homeretal.2002).In regionsin whichwe believethe physicalpropertiesare repre- addition,thelinewidthsappearbroaderthanin4U0614+091, sentativefortheformationoftheobservedopticalspectra.This however, this is difficult to quantify because of possible line shouldatleastgivearoughideaoftherelativestrengthoflines blendsandanuncertainidentificationofthecontinuum. fromdifferentionisationstagesand,thus,isprimarilythought All absorption features seen in our optical spectra are ei- to put our line identifications on firm ground. At present we therofinterstellaroriginortelluric.TheyaremarkedinFigs.1 neglect irradiation of the disk by the neutron star, because it and2. would introduce new free parameters. We expect that the ra- dial ionisation structure of the disk will be shifted to larger radii when irradiation is taken into account, but we hope that 4. Exploratorydiskmodels therelativelinestrengthsarenotaffectedtotheextentthatour We began the construction of accretion disk modelsto calcu- identifications become completely wrong. Although the stud- latesyntheticspectraandreporthereonthecurrentstateofour ied systemsarestrongX-raysources,we donotsee recombi- work.WeuseouraccretiondiskcodeAcDc,whichisdescribed nation lines fromhighlyionised species in the opticalspectra indetailbyNageletal.(2004).Inessence,itassumesaradial as might be expected.It is conceivable that the outer parts of α-disk structure (Shakura & Sunyaev 1973). Then the disk is thedisksinwhichtheopticalspectrumarisesisshieldedfrom dividedinto concentricannuli.For each annuluswe solve the X-rayirradiationbyaninflatedinner-diskregion. radiationtransferequation(assumingplane-parallelgeometry) Since we do not compute the spectrum of the entire disk, togetherwith the non-LTErate equationsfor the atomic level wecannotexpecttomatchtheoverallobservedcontinuumflux, populations,plusenergy-andhydrostaticequations,inorderto and we have alreadypointedoutthat there are problemswith calculateadetailedverticalstructure.Theintegrateddiskspec- theabsolutefluxcalibration.Weuseinterstellarreddeningasa trumisthenobtainedbyco-addingthespecificintensitiesfrom freeparameterinordertoroughlyfitthemodeltotheobserved theindividualannuli,accountingforinclinationandKeplerian fluxlevel.Theappliedreddeningisgiveninthefigurecaptions. rotation. The synthetic spectra presentedhere are based on the fol- Itisnottheaimofthispapertopresentadetailedfittothe lowingchoiceofdisk-modelparameters.Thecentralobjectisa emissionline spectraofthetwo binaries.Thisrequiresexten- neutronstarwith10kmradiusandamassof1.4M .Themass ⊙ siveparameterstudiesthatareextremelytimeconsuming.We accretion rate is 2·10−10 M /yr. The disk extendsfrom 1000 ⊙ ratherchoosetocalculatesyntheticspectrafromselecteddisk to2000stellarradii.ThecorrespondingKeplerianvelocitiesat K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 7 0 70 C III C IV -2 T / kK50 Temperature -4 C V C II Carbon 30 C I -6 0 ) -7 O IV O III 3 -m c Density -2 g -9 ρ / -4 O II g ( Oxygen lo-11 O V O I -6 2 on 0 cti Ne III 0 a τlog Ross--42 Optical depth nization fr--42 NeoNneN eII IV Ne I o -6 g i-6 o l 6.0 0 ) HII 2 -s m 5.5 -2 Hydrogen g / c Gravity -4 HI ( g 5.0 o l -6 0 2.5 He II ) m -2 He I k (z / 2.0 Height above mid-plane Helium g -4 o l1.5 He III -6 -5 -4 -3 -2 -1 0 1 -5 -4 -3 -2 -1 0 1 log (m / g cm-2) log (m / g cm-2) Fig.5.VerticalstructureoftheC/O-diskmodelatadistanceof Fig.6.Verticalionisationstratificationofchemicalelementsin 20000kmfromtheneutronstar.Theemergentdiskfluxatthis the disk models at a distance of 20000 km from the neutron locationcorrespondstoTeff=28000K.Thephysicalvariables star.Theverticallinedrawnatlogm=−0.9indicatesτRoss =1. areplottedagainstthecolumnmassmeasuredfromthesurface towardsthemid-plane. Fortheopacityandemissivitycalculations,itisessentialto solve the non-LTE rate equations with detailed model atoms. OurmainemphasiswasputontheC-andO-ions.The number of non-LTE levels and radiative line transitions are theseradiiamountto4300and3000kms−1,andtheeffective summarisedinTable3.Levelenergies,oscillatorstrengths,and temperaturesto47000Kand28000K,respectively.Thedisk bound-free cross-sections for photoionisation are taken from is divided into five annulisuch that T decreases almost lin- eff the Opacity Project (Seaton etal. 1994) TOPbase1. Electron earlywithradiussteps.Theinclinationangleissetto10◦.The collisional rates for (de-) excitation, ionisation, and recom- Reynoldsnumberusedtoparametrisethediskviscositywasset bination are computed with the usual approximate formulae. to10000,whichcorrespondstoα≈0.5.Thechemicalcompo- sitionisC=50%andO=50%(bymass). 1 http://vizier.u-strasbg.fr/topbase/topbase.html 8 K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 He-, C-, O-, andNe. Theyessentiallycontainmost of the possible C and O line identifications given in Table 2, 1.3 4U0614+091 plusmanyothersthatturnouttobetooweaktobeseeninthe rotationallybroadenedspectra. DIB Asan examplewe presentthe verticalstructureof the ac- cretiondisk modelat a distanceof 2000stellar radiifromthe ux neutron star. The emergent flux at this location corresponds ve fl1.0 Hβ HeII to Teff=28000K. Figure 5 shows the run of several quanti- ati ties above the disk midplane on a column mass scale m. The el Rosseland optical depth reaches unity at logm = −0.9. Here r the gravity amountsto logg=5.8. Mass andelectrondensities 4U1626-67 are 6·10−8 gcm−3 and5·1015 cm−3, respectively.Hencethe physicalconditionsintheline-formingregionsarecomparable tothoseintheatmosphereofasubluminousBstar.Theverti- 0.7 caldistributionofionicfractionsofCandOisshowninFig.6. TherewealsoshowtheH,He,andNefractionsthatweretaken fromthetestmodelsthatincludethesespeciesinanamountof 4800 5000 5200 5400 wavelength / Ao 10%. The dominant ionisation stages of oxygen at τRoss = 1 are O and O, giving rise to prominent emission lines of Fig.7.Comparisonofmodelswithzeroand10%hydrogenand theseionsinthespectra.Inthecaseofcarbon,Cdominates, helium content to the observations. From the lack of Hβ and closely followed by C and C. This explains why we see He emission lines in the observedspectra, we concludethat lines from three C ions. The dominant neon ions in the line- theaccretiondisksarestronglyHandHedeficient.Themodel formingregionsare Ne and Ne. Forhelium we have He spectrawerenormalisedtothelocalcontinuumflux. dominating,followedbyHe,sowewouldexpectstronglines fromthese ionsif neonandheliumwere abundant.Hydrogen is ionised by about99.9%,but still, prominentemission lines CIII NeII arepredictedifHwereanabundantspecies(seebelow). 1.5 4U0614+091 CII 4.1.CandOlinescomparedtoobservations DIB The synthetic spectrum is plotted together with the observed ux spectra in Figs. 1 and 2. Let us first consider 4U0614+091. e fl Generally, many of the observedfeatures are also seen in the v elati 4U1626-67 mcoorrdoebl,oarlattheosuoguhrtlhineeyiddoenntoifitcmataitocnhs.thTehsetrCengltihn.esTohfisthbeasmicoadlleyl r 1.0 aretooweak(e.g.at6580Åand7235Å),whiletheCemis- sionsareinsomecasestooweak(e.g.4650Å)orinothercases too strong (4180Å). The C 5810Å line matcheswell. The OII OII OII oxygen lines show a similar behaviour. The O lines of the OII modelaretooweak(e.g.at4940Å).SomelinesofOmatch reasonably well (e.g. at 3720 Å), while others do not (e.g. at 4400 4500 4600 4700 5590Å,whichismuchtooweakinthemodel). o wavelength / A Themodelcomparisonwith4U1626-67confirmsourideas from the first inspection of the spectrum. The lines in the Fig.8.Comparisonofmodelswithzeroand10%neoncontent blue region can be explained qualitatively by the mere pres- to the observations. The model including neon exhibits weak Ne lines in the region 4350 – 4450 Å, which cannot be de- ence of oxygen lines. This, and the missing C 5810 Å line mightoriginateinarelativelylowC/Oratiowhencomparedto tectedintheobservedspectra. 4U0614+091. It is disappointing that our NLTE disk model obviously gives a poorer fit to the observedline spectrum than the sim- The neon model atom was taken from Dreizler (1993). For pleisothermal,constant-density,LTEslabmodelpresentedby the final spectrum synthesis, fine structure splitting of atomic Nelemans etal. (2004). However, that slab model can at best levels must be considered;level populationswere taken from fit a a limited spectral range, as it is emitted from a partic- themodelswithappropriatestatisticalweights.Levelenergies wereobtainedfromtheNIST2 database.Theopticalsynthetic ular emission region with an assumed value for temperature spectra finally include a total number of 377 lines from H, anddensityandwillneverbeabletosimultaneouslyfittheob- servedspectrafromtheX-raytotheUVandopticalranges.In 2 http://physics.nist.gov/ goodagreementwiththeobservation,ourmodelexhibitslines K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 9 fromthreeionisationstagesofcarbon,indicatingthattemper- 67.ThiscouldsuggestthatthestrippingprocessoftheWDin ature and density in the line-forming regions are reproduced 4U0614+091ismoreadvanced. reasonably well. The fact that the strength of many emission Itishardtoestimatethesystematicerrorofthederivedup- linesof a particularionis either over-or underestimatedmay perlimitfortheneonabundancesothatthefollowingconclu- stemfromdrawbacksinthemodelatoms.Onereasoncouldbe sionsareatthemomentratheruncertain.FromthelackofNe thatthemodelionsarestilltoosmallandignoretheinterlock- lineswefindthatthe Neabundancecannotexceed≈ 10%or, ing effects of neglected energy levels. A further extension of inotherwords,theNe/Oratioisatmostoftheorder0.2.The themodelionsishamperedbythelackofatomicdata,mostly muchhigherNe/Oratios(≈ 0.7)derivedfromthe ISMX-ray oscillatorstrengths.Anotherreasonforthepoorlinefitscould absorptionedgesofneutralNeandO(Schulzetal.2001,Juett beerrorsintheelectroncollisionalrates.Onlyfewareknown etal. 2001) would produce detectable Ne lines in the disk fromexperimentsorquantummechanicalcalculations. spectra. This confirms the conclusion of Juett & Chakrabarty These problemsalso might affectour neon line-formation (2005),thatthe determinedISM abundancesof Ne and O are calculations.However,theemployedNemodelatomhasbeen affected by ionisation effects and, hence, do not reflect the designedfortheanalysisofsdO starsandgavesuccessfulfits abundancesofthedonorstars. toobservedNelines(Dreizler1993).Thereforewethinkthat For an initial solar metallicity the 22Ne abundance in an thepredictedNeemissionlinesinthemodelaremorereliable Ne-enrichedcrystallizedandfractionatedWDcorecanbeesti- thanthemajorityoftheCandOlines.AtomicdataforHand matedfromtheoreticalmodelsto≈0.07,butthisvalueisvery Heareaccuratelyknownandourmodelatomshavebeenem- uncertain(Yungelsonetal.2002).Itcouldbeevenhigherbya ployedwithsuccesstoanalysemanyclassesofstellarspectra, factorof3(Isernetal.1991).SuchanNe-richcorewouldhave hence, we regardthe computedline strengthsof H and He as amassof≈ 0.06M⊙ (Yungelsonetal.2002).Themassofthe muchmorereliablethanthoseofthemetallines.Anotherrea- WDdonorin4U1626-67ismuchsmaller(0.01M⊙;Yungelson sonforthepoorlinefitsmightbethattheassumedα-diskdoes etal.2002).IfweacceptanupperlimitofNe=0.1fortheob- notdescribethephysicsoftheemittingregionwell. servedneonabundance,thenwemaydrawthefollowingcon- clusion.Eitherthe WD core hascrystallizedand fractionated, then our observation favors a relatively small Ne-enrichment 4.2.Limitsonabundancesofhydrogen,helium,and asaresultofthecrystallizationprocess.Or,ifoneacceptsthat neon fractionationwouldresultinahighNeabundanceof0.2(unob- served),thentheWDcorein4U1626-67hadnotimetocrys- The upper abundance limits we derive for H and He are re- tallize which, depending on various details, lasts several Gyr gardedasrealisticbecauseoftheaboveconsiderations,butthe (e.g.Hernanzetal.1994). limitforneonislesssecure. Futureworkwillconcentrateonthediskmodelingforthe In additionto the pureC/O disk modelwe also computed prototype4U1626-67,forwhichtheobservationaldatabaseis a modelincluding10%hydrogenanda modelincluding10% thebestofallsuchsystems.Ultimately,thespectralproperties heliuminordertoseeifthisallowstheHandHeabundancesto (flux distribution and emission line strengths) over the com- beconstrained.InFig.7wedisplaytheresultingemissionlines plete wavelengthrangecomprisingspectral observationswith of H and He 5412 Å. From the lack of observed emission β Chandra, HST, and VLT, must be explainedby a unique disk lines,itisevidentthatHandHeareatmostpresentatthe10% model. level. This excludesthe possibility that the disks of 4U1626- 67and4U0614+091aredominatedbyhydrogenorheliumand Acknowledgements. T.R. is supported by the DLR under grant confirmsthesuspicionthattheyareinfactC/O-dominated. 50OR0201. We thank the referee for constructive criticism that We also computed a C/O model that includes 10% neon. helpedtoimprovethepaper. It exhibits weak Ne lines, but no such lines are detected in theVLTspectra(Fig.8).Weconcludethattheneonabundance References cannotbelargerthan≈10%. Bildsten,L.,Salpeter,E.E.,&Wasserman,I.1992,ApJ,384,143 Cowley,A.P.,Hutchings,J.B.,&Crampton,D.1988,ApJ,333,906 5. Summary Dreizler,S.1993,A&A,273,212 Hernanz,M.,Garcia-Berro,E.,Isern,J.,Mochkovitch,R.,Segretain, Wehavepresentednewhigh-qualityopticalspectraoftheultra- L.,&Chabrier,G.1994,ApJ,434,652 compactlow-massX-raybinaries4U0614+091and4U1626- Homer,L.,Anderson,S.F.,Wachter,S.,&Margon,B.2002,AJ,124, 67.Theyarepureemissionlinespectraandmostprobablystem 3348 fromtheaccretiondisk.Thespectrallinesareidentifiedasdue InT’Zand,J.J.M.,Cumming,A.,VanderSluys,M.V.,Verbunt,F. to C- and O-. Line identifications are corroborated by &Pols,O.R.2005,A&A,441,675 Isern, J., Hernanz, M., Mochkovitch, R., & Garcia-Berro, E. 1991, firstresultsfrommodelingthediskspectrawithdetailednon- A&A,241,L29 LTEradiationtransfercalculations.Hydrogenandheliumlines Jonker,P.G.,vanderKlis,M.,&Homan,J.2001,ApJ,553,335 arelackingandourmodelsconfirmthedeficiencyofHandHe Juett,A.M.,&Chakrabarty,D.2003,ApJ,599,498 inthedisk.Hence,thedonorstarsinthesesystemsareinfact Juett,A.M.,&Chakrabarty,D.2005,ApJ,627,926 the eroded cores of C/O white dwarfs. There are indications Juett,A.M.,Psaltis,D.,&Chakrabarty,D.2001,ApJ,560,L59 thatthe O/Cratio in 4U0614+091ishigherthan in 4U1626- Nagel,T.,Dreizler,S.,Rauch,T.,&Werner,K.2004,A&A,428,109 10 K.Werneretal.:VLTspectroscopyofLMXBs4U0614+091and4U1626-67 Nelemans, G., & Jonker, P. G. 2005, in Interacting Binaries: Accretion, Evolution, and Outcomes, ed. L. Burderi et al., AIP Conf.Proc.,797,396 Nelemans,G.,Jonker,P.G.,Marsh,T.R.,&vanderKlis,M.2004, MNRAS,348,L7 Ritter,H.,&Kolb,U.2003,A&A,404,301(updateRKcat7.4) Schulz,N.S.,Chakrabarty,D.,Marshall,H.L.,etal.2001,ApJ,563, 941 Seaton,M.J.,Yan,Y.,Mihalas,D.,&Pradhan,A.K.1994,MNRAS, 266,805 Shakura,N.I.,&Sunyaev,R.A.1973,A&A,24,337 Sidoli,L.,Parmar,A.N.,&Oosterbroek, T.2005, TheINTEGRAL Universe, Proc. 5th INTEGRAL Workshop, ed. V. Scho¨nfelder, G.Lichti,C.Winkler,ESASP-552,389 Strohmayer,T.E.,&Brown,E.F.2002,ApJ,566,1045 Verbunt, F., & van den Heuvel, E. P. J., 1995, in X-ray Binaries, ed. W.H.G. Lewin, J. van Paradijs, E.P.J. van den Heuvel, (Cambridge:CambridgeUniversityPress),p.457 Wang,Z.,&Chakrabarty,D.2004,ApJ,616,L139 Werner, K., Nagel, T., Dreizler, S., & Rauch, T. 2004, in IAU Coll. 194: Compact Binaries in the Galaxy and Beyond, ed. G. Tovmassian,E.Sion,RevMexAAConferenceSeries,20,146 Yungelson, L. R., Nelemans, G., & van den Heuvel, E. P. J. 2002, A&A,388,546
The list of books you might like
