Astronomy&Astrophysicsmanuscriptno.6878 (cid:13)c ESO2008 February5,2008 LE ⋆ First detection of photospheric depletion in the LMC M.Reyniers⋆⋆ andH.VanWinckel InstituutvoorSterrenkunde,DepartementNatuurkundeenSterrenkunde,K.U.Leuven,Celestijnenlaan200D,3001Leuven,Belgium e-mail:[email protected] ReceivedDecember5,2006;acceptedDecember31,2006 7 0 ABSTRACT 0 2 Context.RecentphotosphericabundancestudiesofgalacticfieldRVTauristarsshowthatdepletionofrefractoryelementsisrather n commonintheseevolvedobjects. a Aims.Theprocessthatcreatesthischemicalanomalyisnotunderstoodwell,butitprobablyrequiresthepresenceofgravitationally J bounddustinabinarysystem.Wetestforthepresenceofdepletioninextra-galacticobjects. 1 Methods.Adetailedphotosphericabundancestudyonthebasisofhigh-qualityUVESspectrawasperformedontheRVTauristar 1 intheLMC:MACHO82.8405.15.Abundanceswerederivedusingacriticallycompiledlinelistwithaccuratelog(gf)valuesandthe latestKuruczmodelatmospheres. 1 Results.With[Fe/H]=−2.6incombinationwith[Zn/Fe]=+2.3and[S/Ti]=+2.5,MACHO82.8405.15displaysastrongdepletion v abundancepattern.TheeffectofthedepletioniscomparabletothestrongestdepletionsseeninfieldGalacticRVTauristars. 7 Conclusions.ThechemicalanalysisofMACHO82.8405.15provesthatthedepletionprocessalsooccursintheextragalacticmembers 2 oftheRVTauripulsationclass.OurprogramstarisamemberofalargersampleofLMCRVTauriobjects.Thissampleisunique, 3 sincethedistancesofthemembersarewell-constrained.Furtherstudiesofthissamplearethereforeexpectedtogaindeeperinsight 1 intothepoorlyunderstooddepletionphenomenonandoftheevolutionarystatusofRVTauristarsingeneral. 0 Keywords.Stars:AGBandpost-AGB–Stars:abundances–Stars:individual:MACHO82.8405.15–MagellanicClouds 7 0 / h 1. Introduction Theextensivechemicalstudiesof the photospheresof field p RV Tauri stars showed, however, that none of the RV Tauri - o It is now almost one century ago that Seares&Haynes (1908) stars shows evidence of post 3rd dredge-up chemical enrich- r noted how RV Tauri, the prototype of the RV Tauri variables, ments, such as enhanced carbon or enhanced s-process abun- st showsa lightcurvewith subsequentdeepandshallowminima. dances (except for maybe V453Oph, see Derooetal. 2005). a Thisobservation,togetherwiththe requiredlimitsonthespec- Instead, the abundance studies (see e.g. Giridharetal. 2000, v: traltype (F to K), is still used as the definingcharacteristic for 2005;Maasetal.2005,andreferencestherein)revealthatsevere i theRVTaurivariables.Prestonetal.(1963)definedthreespec- abundance anomalies are observed in mainly RVB stars, with X troscopicclasses:ThestarsoftypeRVAhavespectraltypeG-K, underabundancesofelementswithahighcondensationtemper- r and show strong absorptionlines butnormalCN or CH bands. ature. The origin of these depletion patterns probablyinvolves a TheRVBstarsaregenerallysomewhathotter(spectraltypeF), chemicalfractionationdue to dust formationin the circumstel- weak-linedobjectsthatshowenhancedCNandCHbands.RVC lar environment,followed by a decoupling of the gas and dust stars are also weak-lined stars, but they show normal bands of withareaccretionofthecleanedgasonthestellarphotosphere. CHandCN. It was realised that the spectroscopic classes of Prestonetal. The exact evolutionary stage of RV Tauri stars is still not could be attributed to a different degree of depletion, with the veryclear,buttheyareprobablyinthepost-AGBstageofevo- RVB stars the most heavily depleted. The whole process prob- lution,asfirstarguedbyJura(1986).ManyoftheRVTauristars ably requires a stable circumstellar environment, like a stable show an infrared excess, which is caused by the circumstellar disk (Watersetal. 1992). Since a stable disk is only likely to dustthatwas likelyformedby heavymass lossduringthe pre- formin abinarysystem,itisbelievedthatalldepletedRV Tau cedingAGBphase.TheMACHOexperimentinthe90srevealed starsareprobablybinaries(VanWinckel2003;DeRuyteretal. the first extragalactic RV Tauri stars in the Large Magellanic 2006). Note that depletion does not seem to occur in globular Cloud. Alcocketal. (1998) showed the light curvesof abouta clusterRVTauristars,orinfieldRVTauristarsoflowmetallic- dozen RV Tauri stars in the LMC, with “formal periods” (the ity ([Fe/H]<−1). Recently, we (Reyniersetal. 2006) reported timebetweentwosuccessiveminima)between40and120days, ontheintrinsicmetal-poorRVTauristarMACHO47.2496.8in definedasinglePLrelationforthesestars,andstrengthenedthe theLMC,whichisuniquefordisplayingaverystrongs-process post-AGBinterpretationofRVTauristarsbytheirluminosities. enrichment. This corroboratesthe post-AGB nature of this ob- ject,butaddstothechemicaldiversityobservedinthispulsation class. Sendoffprintrequeststo:M.Reyniers ⋆ based on observations collected at the European Southern PreliminaryworkbyLloydEvans&Pollard(2004)showed, Observatory,Chile(programme074.D-0619(A)) on the basis of low-resolution spectra, that at least two RV ⋆⋆ PostdoctoralfellowoftheFundforScientificResearch,Flanders Tauri stars found through the MACHO experiment are RVB 2 M.ReyniersandH.VanWinckel:FirstdetectionofphotosphericdepletionintheLMC Table 1. Log of the high-resolution VLT-UVES observations, with small spectral gaps between 5757Å and 5833Å and be- tween8521Åand8660Åduetothespatialgapbetweenthetwo UVESCCDs. date UT exp.time wavelength S/N start (sec) interval(Å) 2005-02-08 01:43 3600 4780−6808 130 3758−4983 110 2005-02-09 02:37 7200 (cid:26)6705−10084 120 stars,andhencepotentiallydepleted.Inthisletter,wefocuson the brightestofboth objects:MACHO82.8405.15(m =14.3). V Alcocketal. (1998) found a period of 93.1days for this ob- jectanda meanabsolutevisualmagnitudeofM =−4.52.The V amplitude of the pulsation is significant, with a peak-to-peak of 0.89 in the MACHO V magnitude. Here, we present a de- tailed abundance analysis of MACHO82.8405.15, based on high-resolution,highsignal-to-noiseVLT-UVESspectra. 2. Observationsandanalysis Spectra were taken with UVES on the VLT-UT2 (Kueyen) in visitor mode by one of us (HVW). Details of the observations, togetherwithsomeindicativesignal-to-noise(S/N)ratios,canbe foundin Table 1. The spectra were taken just beforeminimum light.Thereductionwasperformedinthededicated“UVEScon- text” of the ESO MIDAS software packet, including all stan- dard reduction steps. Optimal extraction was used, except for Fig.1. The VLT+UVES spectrum of MACHO82.8405.15, to- the spectrum taken in the first night, since this was taken with gether with the spectrum of ACHer, around the 4810Å Zn theslicerandoptimalextractionisnotimplementedforthistype line(upperpanel)andthe6155Åoxygentriplet(lowerpanel). ofdata.AsamplespectrumcanbefoundinFig.1. ACHerisadepletedfieldRVTauristarthatissomewhatcooler Themethodofanalysisissimilartothemethodusedinour (T =5500K) than MACHO82.8405.15.It is extensively dis- previous publications on similar stars (e.g. Derooetal. 2005). eff cussed in VanWinckeletal. (1998). In the upperpanel, the Zn First, a complete line identification was performed, using the lineisremarkablystrong,whilelinesofotherironpeakelements line lists of The´venin (1989, 1990). Then, lines with a clean arealmostabsent.Inthelowerpanel,theoxygentripletaround profile and an equivalent width smaller than 150mÅ were se- 6155Å is clearly visible, indicating the hotter temperature for lected for abundance calculation purposes. The log(gf) values MACHO82.8405.15.Notethatthespectruminthelowerpanel of the selected lines were taken from the line list with accu- hasquitealotoftelluricemission.Correctionforthisemission rate oscillator strengths described in VanWinckel&Reyniers is difficult, since this partof the spectrum is observedwith the (2000).Lineswithoutanaccuratelog(gf)valuewerediscarded. imageslicer,hencenoskywindowisavailable. Forthe abundancecalculation,we usedthe latestKuruczmod- els(http://kurucz.harvard.edu),incombinationwiththeLTE line analysis program MOOG by C. Sneden (April 2002 ver- sion). Since initially only 24 Fe and 3 Fe lines were found as a result of the lower electron density and of the weaker UV withourusuallinelist,11Feand4Felineswereaddedfrom blanketing. The´venin(1989, 1990), althoughthese papersare generallyre- We had to follow an alternative scheme to deduce the at- gardedasalessaccuratesourceforoscillatorstrengths. mosphericparameters.First, we estimated the gravity fromthe In our search for the atmospheric parameters of luminositythatisgiveninAlcocketal.(1998).Thisestimateis MACHO82.8405.15, we realised that the classical “spec- of course dependent on the adopted effective temperature, the troscopic” method of studying the iron lines failed. The usual initial mass, and the reddening. However, for all temperatures techniqueofdemandingexcitationequilibriumfortheFelines compatiblewithanF-starspectrum,alowgravityoflogg≤1.0 resultedinaneffectivetemperatureofT =7000K,butsucha isfound.Afurtherdecreaseintheparameterspacewasobtained eff higheffectivetemperaturegivesunrealisticsupersolarzinc(Zn) throughfittingtheHγandHδBalmerlinesusingKuruczBalmer abundances. Moreover, at T =7000K, ionisation equilibrium profiles. The Hα and Hβ lines were not used for this purpose eff can only be attained with a high gravity of logg=2.5. The since they show strong emission features. The wings and the reasontheclassicalmethodfailsisnotclear,butisprobablyre- coresofthetwolineswerefittedsimultaneously,withmoreem- latedtothedynamicalcharacterofthephotosphere(seeSect.3) phasisonagoodfitforthewings.Wecouldconstraintheparam- and/orto the well-knownnon-LTEeffectof UV overionisation eterspace furthertomodelswithT ≤6000K bythisBalmer- eff in the atmospheres of late-type stars (e.g. Shchukinaetal. profilefitting.Fromacomparisonofphotometrytakenwiththe 2005). This effect is expected to be greater in metal-poor stars Swiss Euler telescope at La Silla (see Reyniersetal. 2006, for M.ReyniersandH.VanWinckel:FirstdetectionofphotosphericdepletionintheLMC 3 Fig.3. The spectral energy distribution (SED) of Fig.2.ThefinalHγandHδBalmerfits. MACHO82.8405.15. The dereddened photometry is shown with triangles, while the Kurucz model is overplotted with a solid line. The photometry includes: U, B, V magnitudes in more details on these observations) with synthetic colours, we the Geneva system; I in the Cousins system; I, J, and K from inferredthatthereddeningiscertainlysmall,withanupperlimit DENIS; and J, H, and K from 2MASS. The optical photom- ofE(B-V)=0.2.Ifwecombinethisinformationwiththegravity etry shown here is taken at a more luminous phase than the fromtheluminosityandtheBalmerfits,wecouldconcludethat UVES spectroscopy on which the atmospheric parameters are theonlyconsistenttemperatureisT =6000K.Thefinalchoice eff based. The SED fits, however, were only used to constrain the betweenthetwoconsistentgravitiesatthistemperature,0.5and reddeningtowardsMACHO82.8405.15. 1.0,wasmadeonthebasisoftheconsistencyintheabundances, resultinginlogg=0.5.TheHγandHδBalmerfitswiththefinal parametersareshowninFig.2;thespectralenergydistribution isshowninFig.3.Notethatthereisaphasedifferenceof0.24 betweentheUVESspectraonwhichtheBalmerfittingwasper- formedand the photometry,implying a high uncertainty in the derivedparameters(atleast±250KinT and±1.0inlogg).We eff wanttostress,however,thatintheparameterdetermination,the photometrywasusedsolelyto constrainthereddeningtowards MACHO82.8405.15. 3. Radialvelocitiesofindividuallines Beforeturningtotheactualabundances,theradialvelocitiesof theindividualspectrallinesneedtobenoted.Duringouranaly- sis, we noticedthatsomelinesweremorevelocity-shiftedthan others. Moreover, a clear correlation exists between the radial velocity and the excitation potential of a line, with lower ve- locities for lines with a high excitation potential. This is illus- trated in Fig. 4. It is indicative of an optical depth effect due Fig.4.Theclearcorrelationbetweentheheliocentricradialve- to a differential velocity field in the line-formingregion of the locitiesofindividuallinesandtheirlowerexcitationpotentials. dynamical photosphere of this pulsating star. In field RVTauri stars,thisshock-wavepassagecanleadtoline-splittingatsome pulsational phases (e.g. Gilletetal. 1990). Very similar veloc- the number of lines used; the third one is the mean equivalent ity effects were also detected in the small-amplitude pulsating width;thefourthcolumngivestheabsoluteabundancesderived post-AGBstarHD56126(Le`breetal.1996;Barthe`setal.2000; logǫ=log(N(el)/N(H))+12;σ ,thefifthcolumn,istheline-to- ltl Fokinetal.2001),andwereagainattributedtothepassageofa linescatter;thesixthcolumngivestheabundancerelativetothe shock wave. Note that the UVES spectra were taken just be- sun [el/H]. For the referencesof the solar abundances(seventh foreminimumlight,i.e.thephaseatwhichanewshockwaveis column)neededtocalculatethe[el/H]values:seeReyniersetal. expectedtodevelop(Gilletetal.1989).Theadoptionofplane- (2006). The dust condensation temperatures (last column) are parallel,LTEKuruczmodelscanthusonlybeafirstapproxima- takenfromLodders(2003). tionoftherealphotosphere. The abundances are also graphically presented in Fig. 5, where the abundance relative to solar [el/H] is plotted against condensationtemperature.Theclearanti-correlationasseenon 4. Resultsanddiscussion Fig. 5 is undoubtfully recognised as a depletion pattern: the The final abundances can be found in Table 2. The first col- slightlysubsolarabundancesofZnandSareexpectedforafield umnofthistablegivesthe actualion;thesecondcolumngives LMC-star, while ironis depletedby morethan a factorof 300. 4 M.ReyniersandH.VanWinckel:FirstdetectionofphotosphericdepletionintheLMC Table2.AbundanceresultsforMACHO82.8405.15. MACHO82.8405.15 T =6000K eff logg=0.5(cgs) ξ =3.5kms−1 t [M/H]=−2.5 ion N W logǫ σ [el/H] sun T λ ltl cond C 13 63 8.00 0.13 −0.57 8.57 40 N 9 46 8.04 0.16 0.05 7.99 123 O 3 26 8.53 0.10 −0.33 8.86 [O] 2 102 9.16 0.08 0.30 8.86 Na 4 52 5.80 0.14 −0.53 6.33 953 Mg 2 38 5.29 0.01 −2.25 7.54 1327 S 3 37 6.99 0.02 −0.34 7.33 655 Ca 3 25 4.24 0.15 −2.12 6.36 1505 Sc 1 17 0.12 −3.05 3.17 1647 Ti 4 22 2.18 0.26 −2.84 5.02 1573 Fig.5. The anti-correlation between the abundance of an el- Cr 2 49 3.26 0.04 −2.41 5.67 1291 ement ([el/H]) and its condensation temperature. This indi- Cr 3 39 3.40 0.07 −2.27 5.67 1291 cates that a very efficient depletion process has taken place Mn 3 28 3.60 0.13 −1.79 5.39 1150 Fe 35 30 5.08 0.25 −2.43 7.51 1328 in MACHO82.8405.15.The errorbar is the line-to-line scatter Fe 7 52 4.87 0.19 −2.64 7.51 1328 listedinTable2. Zn 4 87 4.26 0.06 −0.34 4.60 723 Y 1 24 −0.26 −2.50 2.24 1647 Ba 3 105 0.13 0.15 −2.00 2.13 1447 DeRuyteretal. 2006) that binarity plays a key role in the cre- ationofastabledustycircumstellarenvironmentthatisprobably neededfortheprocesstooccur.High-resolutioninfrareddataof MACHO82.8405.15wouldyieldinvaluableinformationonthe TheCNOelementshavelowcondensationtemperaturesandare physicalconditionsandmineralogyofthecircumstellarenviron- typicallyexpectednottobedepleted.Thelowcarbonabundance ment. of [C/H]=−0.6 makes clear that MACHO82.8405.15 has not Our first extragalactic detection of a depleted photosphere experienced a 3rd dredge-up. One remarkable result is the de- in a luminous star shows that the depletion process may be tectionoftwoforbidden[O]linesat6300.23Åand6363.88Å, rathercommonelsewhere.MACHO82.8405.15isamemberof bothinabsorption.Theabundancesderivedfromtheselinesare a larger sample of luminous LMC RV Tauri objects. Our re- muchhigherthantheabundancesderivedfromtheOtripletat cent analysis of one of the other objects (MACHO47.2496.8, 6155Å (see Table 2). The [O] lines are known notto be very Reyniersetal.2006)shows,however,averydifferentchemical sensitivetonon-LTEeffects,sothislargeabundancedifference resultwith a verystrongs-process-enhancedphotospherein an isnotexpected.Thetwo oxygenabundancescanbe broughtin intrinsically metal-poor environment of [Fe/H]=−1.4. Clearly agreement if one adopts a slightly lower temperature (5750K) the LMC RVTauri stars are also very diverse chemically. A combined with a very low gravity of logg=−0.4. Such a low global study of these objects will enable us to better constrain gravity is, however, not compatible with the luminosity. The boththedepletionprocessitself,aswellasthetrueevolutionary abundances of the s-process elements Y and Ba show an ex- statusofRVTauristars. pectedbehaviourandareprobablynotintrinsicallyenhanced. Acknowledgements. Itis apleasure tothank Clio Gielen forthe construction oftheSED,SophieSaesenfortheEulerobservations, theanonymousreferee forthequickandconstructivereport,andtheGenevastaffforobservationtime 5. Conclusion ontheEulertelescope. MRacknowledges financialsupportfromtheFundfor ScientificResearch-Flanders(Belgium). In this letter, we have found evidence that the photospheric depletion phenomenon, which is frequently observed in field RV Tauri stars, also occurs in the extragalactic members of References this pulsation class. The effect on the observed metallicity of MACHO82.8405.15is very significant ([Fe/H] = −2.6), while Alcock,C.,Allsman,R.A.,Alves,D.R.,etal.1998,AJ,115,1921 Barthe`s,D.,Le`bre,A.,Gillet,D.,&Mauron,N.2000,A&A,359,168 the initial iron content was probably only slightly subsolar, DeRuyter,S.,VanWinckel,H.,Maas,T.,etal.2006,A&A,448,641 as expected in the LMC. While the degree of depletion of Deroo,P.,Reyniers,M.,VanWinckel,H.,Goriely,S.,&Siess,L.2005,A&A, MACHO82.8405.15isbeyondanydoubt,theanalysisitselfcan 438,987 beimprovedbyadoptingamorerealisticmodelatmospherethat Fokin,A.B.,Le`bre,A.,LeCoroller,H.,&Gillet,D.2001,A&A,378,546 accountsfortheobserveddynamicsandthenon-LTEeffectsthat Gillet,D.,Burki,G.,&Duquennoy,A.1990,A&A,237,159 Gillet,D.,Duquennoy,A.,Bouchet,P.,&Gouiffes,C.1989,A&A,215,316 we encountered in our analysis. Additional spectra at different Giridhar,S.,Lambert,D.L.,&Gonzalez,G.2000,ApJ,531,521 pulsationphasesmayrevealinterestingresults,boththosefrom Giridhar,S.,Lambert,D.L.,Reddy,B.E.,Gonzalez,G.,&Yong,D.2005,ApJ, theLTEapproachofadynamicphotosphereandfromtheabun- 627,432 dancedeterminationoflinesandspeciesthatbecomeaccessible Jura,M.1986,ApJ,309,732 atdifferenttemperatures. Le`bre,A.,Mauron,N.,Gillet,D.,&Barthe`s,D.1996,A&A,310,923 LloydEvans,T.&Pollard,K.R.2004,inASPConf.Ser.310:IAUColloq.193: Although detailed scenario for the depletion process is VariableStarsintheLocalGroup,ed.D.W.Kurtz&K.R.Pollard,344 still lacking, there is growing evidence (VanWinckel 2003; Lodders,K.2003,ApJ,591,1220 M.ReyniersandH.VanWinckel:FirstdetectionofphotosphericdepletionintheLMC 5 Maas,T.,VanWinckel,H.,&LloydEvans,T.2005,A&A,429,297 Preston,G.W.,Krzeminski,W.,Smak,J.,&Williams,J.A.1963,ApJ,137,401 Reyniers,M.,Abia,C.,VanWinckel,H.,etal.2006,astro-ph/0610240 Seares,F.H.&Haynes,E.S.1908,LawsObservatoryBulletin, Universityof Missouri,14,215 Shchukina,N.G.,Bueno,J.T.,&Asplund,M.2005,ApJ,618,939 The´venin,F.1989,A&AS,77,137 —.1990,A&AS,82,179 VanWinckel,H.2003,ARA&A,41,391 VanWinckel,H.&Reyniers,M.2000,A&A,354,135 VanWinckel,H.,Waelkens,C.,Waters,L.B.F.M.,etal.1998,A&A,336,L17 Waters,L.B.F.M.,Trams,N.R.,&Waelkens,C.1992,A&A,262,L37