Table Of Content

1 We aim at showing that the broad-band SED characteris- tics of our sample of post-AGB stars are best interpreted, as- suming the circumstellar dust is stored in Keplerian rotating passivediscs.Wepresentahomogeneousandsystematicstudy oftheSpectralEnergyDistributions(SEDs)ofasampleof51 post-AGB objects. The selection criteria to define the whole sample were tuned to cover the broad-bandcharacteristics of knownbinarypost-AGBstars. The whole sampleincludes20 dusty RVTauri stars from the General Catalogue of Variable Stars(GCVS).WesupplementedourownGenevaopticalpho- tometry with literature data to cover a broad range of fluxes from the UV to the far-IR. All the SEDs display very simi- larcharacteristics:alargeIRexcesswithadustexcessstarting near the sublimation temperature,irrespectiveof the effective 6 0 temperatureof the centralstar. Moreover,when available, the 0 longwavelengthfluxesshowa black-bodyslope indicativeof 2 thepresenceofacomponentoflargemmsizedgrains.Wear- n gue that in all systems, gravitationally bound dusty discs are a present.Thediscsmustbepuffed-uptocoveralargeopening J angleforthecentralstarandwearguethatthediscshavesome 5 similaritywiththepassivediscsdetectedaroundyoungstellar 2 objects.Weinterpretthepresenceofadisctobeasignaturefor binarity of the central object, but this will need confirmation 1 v bylong-termmonitoringoftheradialvelocities.Wearguethat 8 dusty RVTauri stars are those binarieswhich happento be in 7 thePopulationIIinstabilitystrip. 5 1 Keywords.stars:AGBandpost-AGB–stars:binaries:gen- 0 6 eral–stars:circumstellarmatter 0 / h p - o r t s a : v i X r a Astronomy&Astrophysicsmanuscriptno.4062sdr February3,2008 (DOI:willbeinsertedbyhandlater) Keplerian discs around post-AGB stars: ⋆ a common phenomenon? S.DeRuyter1,H.VanWinckel2,T.Maas2,3,T.LloydEvans4,L.B.F.MWaters2,5,andH.Dejonghe1 1 SterrenkundigObservatorium,UniversiteitGent,Krijgslaan281S9,9000Gent,Belgium 2 InstituutvoorSterrenkunde,K.U.Leuven,Celestijnenlaan200B,3001Leuven,Belgium 3 DepartmentofAstronomy,UniversityofTexas,Austin,TX78712 4 SchoolofPhysicsandAstronomy,UniversityofSt.Andrews,NorthHaugh,St.Andrews,Fife,ScotlandKY169SS 5 SterrenkundigInstituut‘AntonPannekoek’,UniversiteitAmsterdam,Kruislaan403,1098Amsterdam,TheNetherlands Received¡date¿/Accepted¡date¿ Abstract. 1. Introduction belinkedtothepole-to-equatordensitycontrast,asdetermined on the basis of high spatial resolution mid-infrared images Post-AGB stars are low and intermediate initial mass (≤ 8- (Meixneretal.1999).RecentlyalsoGledhill(2005)foundev- 9M⊙) stars which have suffered a large and dusty mass-loss idence for axi-symmetry in the dust density in his polarimet- phaseattheendoftheAsymptoticGiantBranch(AGB)during ricimagingsurveyofcandidatepost-AGBstars.Thedetached whichalmostthewholestellarenvelopewasexpelled.Theyare shellscorrespondtostarswithanopticallythinexpandingcir- evolvingatconstantluminosityonafastevolutionarytrackin cumstellarenvelopewhereasthebipolarandunresolvedtargets whichthecentralstarcrossestheHR-diagramfromacoolAGB have optically thick dust structures, probably in the form of photospheretotheionizingtemperatureofthecentralstarofa discs. It is suggested once again that this bifurcation in mor- PlanetaryNebula(PN).Giventhesmallevolutionarytimescale phologyisrootedinthepresenceorabsenceofabinarycom- of about104years, post-AGBstars are rare and not manyare panion,whichdetermineswhetherornotadiscforms. known(e.gSzczerbaetal.2001;VanWinckel2003). DiscussionsonthemorphologyofPNeusuallystartwitha Stunningkinematicinformationresultedfromtheextensive displayoftheaestheticpicturesoftheHubbleSpaceTelescope COsurveyofBujarrabaletal.(2001):itappearstobeafunda- (HST) showing the complex geometry and structure of the mental property of the omnipresentfast molecular outflow in nebulae,immediatelyfollowedbythepuzzlingandcontradic- PPNe, that it carries a huge amount of linear momentum, up tory finding that, on the AGB, the mass-loss is found to be to 1000 times the momentum available for a radiation driven sphericallysymmetric.Duringthetransitiontime,thestarand wind. Clearly, other momentum sources have to be explored. circumstellar envelope must undergo fundamental and rapid Some molecular jets are resolved by high spatial resolution changesinstructure,mass-lossmodeandgeometrywhichare imaging (Sahai 2004). The formation process of the strongly still badly understood. The debate on which physical mecha- collimatedjetsis,however,stillbadlyunderstood.Anintrigu- nisms are driving the morphology changes gained even more ingsuggestionisthattheprocessessimilartothejetformation impetus from the finding that also resolved cooler post-AGB inlow-massyoungstellar objectsoperateandthatthejetsare starsorProto-PlanetaryNebulae(PPNe)displayasurprisingly borninaccretiondiscs.Thismechanismrequiresasignificant widevarietyinshapesandstructures,veryearlyintheirpost- amountofmassorbitingthepost-AGBstar.Suchadisccould AGBevolution(Balick&Frank2002,andreferencestherein). be present, but likely only in binary stars. Testing of such an hypothesis is severely hampered by the lack of observational In a survey of 27 PPNe, 21 were found to be resolved informationonbinarityinPNeandPPNebutalsoonourpoor (Uetaetal. 2000). Moreover, the degree of asymmetry could theoreticalunderstandingofAGBevolutioninbinarysystems. Send offprint requests to: S. De Ruyter, e-mail: NotethatmostobjectswhichweredetectedinCOarestrongly [email protected] embedded and the sample is probably biased towards more ⋆ BasedonobservationswiththeP7photometeratthe0.7mSwiss massivePPNe. Telescope,LaSilla(Chile)andattheFlemishMercatorTelescope,La Palma(Spain)andonobservationswiththeSubmillimetreCommon- ProbingbinarityinPNeandembeddedPPNedirectlywith User Bolometer Array (SCUBA) at the James Clerk Maxwell radialvelocitymonitoringisnoteasy.Foropticallybrightpost- Telescope(JCMT),MaunaKea,Hawaii. AGBstars,thisisdifferentandsomefamousexamplesexistin S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 3 whichitisclearthatthebinarynatureofthecentralobjectmust Our sample and the selection criteria are presented in haveplayedakeyroleintheevolutionofthesystem. Sect. 2. In Sect. 3 we present our photometric data comple- mented with the literature values and the results of the large ThemostfamousexampleiscertainlyHD44179,thecen- all sky surveys. In Sect. 4 we present the detailed Spectral tral star of the Red Rectangle nebula. It displays a huge and Energy Distribution (SED) construction of all individual ob- broad IRexcess. The IR luminosity is 33 times stronger than jects. Distances to the objects are estimated in Sect. 5. We the optical luminosity (Leinert&Haas 1989; Waelkensetal. end this extensive study with a detailed discussion in Sect. 6. 1996).VanWinckeletal.(1995)haveshownthatHD44179is ConclusionsaresummarizedinSect.7. aspectroscopicbinarywithanorbitalperiodof298days.The eccentricityisaremarkablyhighe=0.37.Sinceitsdiscovery by Cohenetal. (1975), HD44179 has often been used as an 2. Programmestars archetypicalexampleofaC-richpost-AGBobject,butitisnow Our total sample of 51 stars was defined on several observa- generally accepted that many of the remarkable phenomena tional criteria inspired by the characteristics of known post- andthepeculiarmorphologyofthenebula(foranoverviewsee AGB stars in binary systems. The whole sample is given Cohenetal.2004)arecloselyrelatedtothepresenceofastable in Table 1. Besides these proven binaries, we discuss also circumbinarydiscaroundthebinarycentralstar.Thelongevity RVTauristarswithIRASdetectionsofgoodqualityandasam- ofthediscwasdramaticallyconfirmedbythedetectionofcool ple of newly identified objects, originally found by one of us O-richcrystallinesilicate dustgrainsin thedisc (Watersetal. (T.LloydEvans)insearchofnewcandidateRVTauristars. 1998).Themixedchemistryisbestexplainedassumingthefor- mationoftheO-richdiscpredatedthemorerecentC-richtran- sitionofthecentralobject.RecentSpitzerdatashow,however, that also far from the central star, the nebula appears to have a dust component which is O-rich (Markwick-Kemperetal. 2005). The (chemical) history of the binary, disc and neb- ula is therefore far from understood. The disc is resolved in ground-based high spatial-resolution imaging at optical and near-IRwavelengths(Men’shchikovetal.2002,andreferences therein)aswellasinHSTopticalimages(Cohenetal.2004). The disc was also resolved in interferometric CO(2-1) maps, andtheKepleriankinematicsofthediscweredirectlydetected (Bujarrabaletal.2003,2005).HD44179showsaconsiderable amount of dust processing in the disc with indications of the presence of very large grains (Juraetal. 1997) and possibly evenmacro-structures(Jura&Turner1998). Anotherremarkableevolvedobjectwithalong-liveddiscis HR4049.Itisabinarywithanorbitalperiodof430dayswith aremarkablyhigheccentricityofe=0.30.Alsointhisobject, thecircumstellarmaterialshowsa mixedchemistrywith both carbon rich and oxygenrich features. The SED of the dust is also very peculiar as it can be fitted with a single black-body ofabout1150K,from1µmdownto850µm.TheseSEDchar- acteristicsareveryconstrainingandthebestmodelforthecir- cumstellarmaterialis,thatthedustistrappedinaveryopaque dusttorusatKeplerianrotation(Dominiketal.2003).Itisclear that also in this object the dusty disc plays a lead role in the (future) evolution. Other examplesexist and there is substan- tial observationalevidence that the systems are all likely sur- roundedby a circumstellar orbitingdisc (VanWinckel 2003). NotethatsofaronlyfortheRedRectangletheKepleriandisc is spatio-kinematically resolved by CO interferometric maps (Bujarrabaletal.2005). Togaininsightintheevolutionofbinarysystemsandtheir circumstellar material, we reportin this contributionon a ho- mogeneous and systematic study of a sample of objects with similarIRcharacteristicsastheknownbinarypost-AGBstars. Themainaimistostudythebroad-bandSEDsinasystematic way, in order to gain insight in the possible evolutionarylink betweenthedifferentobjects. Table 1. The IRASnumber,the HDnumber or the name from the GCVS, the spectral type, the equatorialcoordinatesα and δ(J2000), the effective temperature T , the 4 eff surfacegravitylogg,themetallicity[Fe/H],thereferenceforthemodelparameters,thetypeofobject(post-AGB,RVTauriorNewSample)andareferencewheretheorbital motionoftheobjectcanbefound,aregiven.NotethatIRAS11472−0800isastronglydepletedobjectaddedtooursamplestars. 1 2 3 4 5 6 7 8 9 10 11 12 No IRASnumber HDnumberor Spectral α(J2000) δ(J2000) Teff logg [Fe/H] ReferenceModelParameters Type ReferenceBinarity GCVSname type (hms) (◦′′′) (K) (cgs) 1 IRAS04166+5719 TWCam G3I 042048.1 +572626 4800 0.0 −0.5 Giridharetal.(2000) RVTauri 2 IRAS04440+2605 RVTau G2I 044706.8 +261044 4500 0.0 −0.4 Giridharetal.(2000) RVTauri 3 IRAS05208−2035 M0e+F 052259.423 −203253.03 4000 0.5 0.0 fromspectraltype NewSample 4 IRAS06034+1354 DYOri G0I 060612.3 +135309 6000 1.5 −2.0 Gonzalezetal.(1997b) RVTauri 5 IRAS06072+0953 CTOri F9I 060957.4 +095235 5500 1.0 −2.0 Gonzalezetal.(1997a) RVTauri S. 6 IRAS06108+2743 SUGem G5I 061400.8 +274212 5750 1.125 −0.7 Wahlgren(1992) RVTauri D 7 IRAS06160−1701 UYCMa G0V 061816.367 −170234.72 5500 1.0 0.0 fromspectraltype RVTauri eR 8 IRAS06176−1036 HD44179 F1I 061958.2160 −103814.691 7500 0.8 −3.3 Waelkensetal.(1992) post-AGB VanWinckeletal.(1995) u 9 IRAS06338+5333 HD46703 F3I 063752.4253 +533101.957 6250 1.0 −1.5 fromspectrum post-AGB Hrivnak,privatecommunication yte 10 IRAS06472−3713 STPup F7I 064856.4131 −371633.332 5750 0.5 −1.5 Gonzalez&Wallerstein(1996) RVTauri Gonzalez&Wallerstein(1996) r 11 IRAS07008+1050 HD52961 F6I 070339.6314 +104613.067 6000 0.5 −4.8 Waelkensetal.(1991b) post-AGB Waelkensetal.(1992) et 1132 IIRRAASS0077218440−−02934201 USAMOo1n73329 FG50II 007713600487..2571 −−203924760317.61 75000000 10..50 −−00..88 VGairnidWhainrcekteall.(1(2909070)) pRoVstT-AauGriB VPoalnlaWrdin&ckCeoltettrealll.((11999955)) al.: K 14 IRAS08011−3627 ARPup F0I 080301.1 −363547 6000 1.5 −1.0 Gonzalezetal.(1997b) RVTauri e 1156 IIRRAASS0089504640−−42483017 FF35 008950681140..1182 −−424841391100.73 76255000 11..55 −−00..55 MMaaaasseettaall..((22000035)) NNeewwSSaammppllee Maasetal.(2003) pler 17 IRAS09144−4933 G0 091609.1 −494606 5750 0.5 −0.5 Maasetal.(2005) NewSample ian 18 IRAS09256−6324 IWCar F7I 092653.4 −633748 6700 2.0 −1.0 Giridharetal.(1994) RVTauri d 19 IRAS09400−4733 M0 094152.9 −474703 NewSample is c 20 IRAS09538−7622 G0 095358.5 −763653 5500 1.0 −0.5 Maasetal.(2005) NewSample s 21 IRAS10158−2844 HR4049 A6I 101807.5903 −285931.201 7500 1.0 −4.5 Dominiketal.(2003) post-AGB Waelkensetal.(1991b) ar 22 IRAS10174−5704 K:rr 101918.1 −571936 NewSample ou 23 IRAS10456−5712 HD93662 K5 104738.3965 −572802.679 4250 0.5 0.0 fromspectraltype NewSample nd 24 IRAS11000−6153 HD95767 F0I 110204.314 −620942.84 7600 2.0 0.1 VanWinckel(1997) post-AGB VanWinckeletal.(1995) p 25 IRAS11118−5726 GKCar M0I 111401.3 −574309 RVTauri os 26 IRAS11472−0800 F5I 114948.5 −081721 5750 1.0 −2.5 fromspectrum t-A 27 IRAS12067−4508 RUCen F6I 120923.7 −452535 6000 1.5 −2.0 Maasetal.(2002) RVTauri G 28 IRAS12185−4856 SXCen G3V 122112.6 −491241 6000 1.0 −1.0 Maasetal.(2002) RVTauri Maasetal.(2002) B 2390 IIRRAASS1123222528−−48615023 HD108015 FG35IbDe 113232140573..1501 −−841710893007.51 7000 1.5 −0.1 VanWinckel(1997) Npoeswt-ASaGmBple star 31 ENTrA F2Ib 145700.6847 −685022.879 6000 1.0 −0.5 VanWinckel(1997) RVTauri VanWinckeletal.(1995) s: 32 IRAS15469−5311 F3 155044.0 −532044 7500 1.5 0.0 Maasetal.(2005) NewSample ac 33 IRAS15556−5444 F8 155932.1 −545318 NewSample o m 34 IRAS16230−3410 F8 162620.29 −341712.3 6250 1.0 −0.5 Maasetal.(2005) NewSample m 35 IRAS17038−4815 G2p(R)e 170736.3 −481908 4750 0.5 −1.5 Maasetal.(2005) NewSample o 36 IRAS17233−4330 G0p(R) 172657.5 −433313 6250 1.5 −1.0 Maasetal.(2005) NewSample n 37 IRAS17243−4348 LRSco G2 172756.1 −435048 6250 0.5 0.0 Maasetal.(2005) NewSample ph 38 IRAS17534+2603 89Her F3I 175525.1889 +260259.966 6500 1.0 0.0 Watersetal.(1993) post-AGB Watersetal.(1993) en 39 IRAS17530−3348 AISco G4I 175618.5 −334847 5000 0.0 0.0 fromspectrum RVTauri om 40 IRAS18123+0511 G5 181449.4 +051255 5000 0.5 0.0 fromspectraltype NewSample e 41 IRAS18158−3445 F6 181913.6 −344432 6500 1.5 0.0 fromspectraltype NewSample no 42 IRAS18281+2149 ACHer F4I 183016.2 +215200 5500 0.5 −1.5 VanWinckeletal.(1998) RVTauri VanWinckeletal.(1998) n? 43 IRAS18564−0814 ADAql G8I 185908.1 −081014 6300 1.25 −2.1 Giridharetal.(1998) RVTauri 44 IRAS19125+0343 F2 191500.8 +034841 7750 1.0 −0.5 Maasetal.(2005) NewSample 45 IRAS19163+2745 EPLyr A4I 191817.5 +275038 7000 2.0 −1.5 Gonzalezetal.(1997b) RVTauri 46 IRAS19157−0247 F3 191822.5 −024209 7750 1.0 0.0 Maasetal.(2005) NewSample 47 IRAS20056+1834 QYSge G0De 200754.8 +184257 5850 0.7 −0.4 Raoetal.(2002) NewSample 48 IRAS20117+1634 RSge G0I 201403.8 +164335 5750 0.0 −0.5 Gonzalezetal.(1997b) RVTauri 49 IRAS20343+2625 VVul G8I 203631.8 +263617 5250 1.0 0.0 fromspectraltype RVTauri 50 IRAS22327−1731 HD213985 A2I 223527.5259 −171526.889 8250 1.5 −1.0 fromspectrum post-AGB VanWinckeletal.(1995) 51 BD+39◦4926 A9I 224611.2273 +400626.294 7500 1.2 −2.9 VanWinckeletal.(1995) post-AGB Kodairaetal.(1970) S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 5 2.1. Confirmedbinarypost-AGBstars WiththeirhighluminosityandoftenlargeIRexcessesdue to thermal radiation from circumstellar dust, there is general Thegroupofbinarypost-AGBstarsinoursamplewasassem- agreement that RVTauri stars are low-mass objects in tran- bled in a rather coincidentalway on the basis of independent sition from the Asymptotic Giant Branch (AGB) to white detailedstar-to-staranalyses.We includedtheprovenbinaries dwarfs (Jura 1986). It was noted already in the seventies that in our programme star sample and in Col. 12 of Table 1 we manyRVTauristarsshowaconsiderablenear-IRexcesscaused giveareferencewheretheorbitalmotionoftheobjectswasdis- by a hot dust component (Gehrz&Ney 1972; Gehrz 1972; cussed.TheorbitalelementsthemselvesarelistedinTableB.1. LloydEvans1985),whichwasattributedtothepossiblepres- The photospheres of several binary post-AGB stars enceofadustydisc.Morris(1987)suggestedthatcircumstel- (e.g. HR4049 (Lambertetal. 1988; Waelkensetal. lar dust could indeed exist in a disc structure in binary sys- 1991a), HD44179 (Waelkensetal. 1992), HD52961 tems, either in a circumbinary disc or in a disc around one (VanWinckeletal. 1992) and BD+39◦4926 (Kodairaetal. ofthecomponents.TheweakcircumstellarCOemission(e.g. 1970)) are strongly affected by a poorly understood selective Bujarrabaletal. 1988) together with the black-body spectral depletion process. The basic scenario of this process is that index at long wavelengths observed in some RVTauri stars circumstellar gas is separated from the dust and subsequently (DeRuyteretal.2005)corroboratetheconclusionthatthecir- re-accretedontothestar(Mathis&Lamers1992;Watersetal. cumstellardustisnotfreelyexpandingbutconfined. 1992).Sincethatgasisdevoidofrefractories,thephotosphere In recent years it has become clear that also many willbealteredchemically,andthismayresultinveryFe-poor RVTauri photospheres show chemical anomalies pointing to stars which are rich in non-refractories like Zn and S. This an efficient depletion of refractory elements (Giridharetal. process is very efficient in the four named binaries where 1994; Giridhar&Ferro 1995; Giridharetal. 1998, 2000; there is observational evidence for the presence of a dusty Gonzalezetal. 1997b,a; VanWinckeletal. 1998; Maasetal. stablereservoir(VanWinckeletal.1995).Efficientseparation 2002).Themanyaffectedstarsshowthatthedepletionprocess of circumstellar gas and dust is not evident and Watersetal. isaverycommonphenomenoninevolvedstars.Giridharetal. (1992) argued that the most favourable circumstances may (2005)showthatdepletioninRVTauristarsislessstrongwhen occur if the circumstellar dust is indeed trapped in a stable disc.Notethatforonestronglydepletedobject,BD+39◦4926, the centralstar is cooler,which is interpretedas pointingto a stronger dilution due to a deeper convective envelope in the therewasnoIRexcessdetectedbyIRAS. coolerstars. Thedepletionpatternsarenotseeninobjectsfor In the few cases where ISO spectra are available,the dust which the intrinsic metallicity is smaller than aboutone tenth processingisstrong,whichresultsinalargecrystallinityfrac- solar([Fe/H]≤ −1.0). tionofthegrains(e.g.Molsteretal.2002).Alsothedustgrain sizedistributionisdifferentindiscsthaninoutflowsandwhere It is clear that the observational restriction that depletion long wavelength fluxes are available, they indicate the pres- isonlyactiveinbinarystars,whichwasformulatedwhenonly ence of large, µm sized and even cm sized dust grains (e.g. fourextremelydepletedobjectswereknown,hasbecomemuch Shentonetal.1995). less evident with the many new detections. Direct evidence Obviouslysuchadiscmustplayanimportantroleinthese for binarity,from radial velocity measurements,is difficultto systems and, although the indirect observationalevidence for obtain since RVTauristars have large pulsationalamplitudes. the presence of a stable disc is well established in the binary Moreover,theyoftenshowthepresenceofshocksintheline- stars,theactualstructureletalonetheformation,stabilityand forming region of the photosphere making the very determi- evolutionarenotwellunderstood.NotethatonlyforHD44179 nation of the radial velocity difficult (e.g. Gilletetal. 1990). isthisdiscspatiallyresolved.Forallotherobjects,thepresence Nonethelessorbitalelementshavebeendeterminedforquitea ofthediscwaspostulated. fewclassicalRVTauristars:UMon(Pollard&Cottrell1995), ACHer(VanWinckeletal.1998),ENTrA(VanWinckeletal. 1999), SXCen (Maasetal. 2002) and orbital motion is also 2.2. Classical RVTauristars from the General foundforIWCar(Pollardetal.1997),EPLyr(Gonzalezetal. CatalogueofVariableStarswith strong 1997b)andRUCen(Maasetal.2002). IRexcess Inthiscontributionweanalysethebroad-bandSEDsofthe RVTauri stars form a class of classical pulsating variables. RVTauristarsoftheGCVSwithareliableIRASmeasurement They are luminous (I-II) supergiant pulsating variables, the at60µm(e.g.Jura1986)andcomparethemwiththeSEDsof lightcurveofwhichshowsalternatingdeepandshallowmin- similarobjects,whicharenotinthePopIICepheidinstability ima with a formal period (measured between one deep mini- strip. mumandthenext)of30to150daysandabrightnessrangeof uptofourmagnitudes.ThespectraltypeistypicallyFtoGat minimum and G to K at maximum. There are two main pho- 2.3. New SampleofcandidateRVTauristars tometric varieties of RVTauri stars: the RVa types maintain a roughlyconstant mean brightness;RVb types show on top of LloydEvans (1985) and Raveendran (1989) showed that the theirpulsationalperiodalongerterm(600to1500days)peri- RVTauri stars are located in a well-defined and relatively odicity. thinly-populatedpartoftheIRAScolour-colourdiagram.The 6 S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? definingrectangleis: Genevaopticalphotometry WeacquiredGenevaopticalphotometryatrandomepochswith [12]−[25]=1.56+2.5log[F(25)/F(12)]=1.0−1.55 the 0.7m Swiss Telescope at La Silla and with the Flemish (cid:26)[25]−[60]=1.88+2.5log[F(60)/F(25)]=0.20−1.0 MercatorTelescopeatLaPalma,usingtherefurbishedGeneva (1) photometer P7 (Raskinetal. 2004). Our total data set was scanned for the maximum and minimum magnitudes (see LloydEvans(1997)searchedfornewexamplesofRVTauri Table C.1). Observation dates, the number of measurements stars, as stars with dusty discs, in the IRAS Catalogue, and andthetotaltimebasesofthesemaximaandminimaaregiven foundthestarsofthe‘newsample’discussedinthispaper.The aswell.AdditionalGenevaopticalphotometrywasfoundinthe presenceinthissampleofstarswiththeinfraredpropertiesof Geneva database: the General Catalogue of Photometric Data theGCVSsampleofRVTauristars,butwhichfelloutsidethe (GCPD,http://obswww.unige.ch/gcpd/gcpd.html). instability strip as judgedfromtheir spectral typesand which Notethesedataareonlygivenforcompleteness,butwerenot lackedthelarge-amplitudeRVTaurivariability,suggestedthat used in our analysis; except for IRAS10456 for which the RVTauri stars are those stars with dusty discs which are cur- GCPDdataareouronlyGenevafluxes. rentlylocatedwithintheinstabilitystrip(LloydEvans1999). Maasetal.(2005)presentedachemicalabundanceanalysis onthebasisofhighsignal-to-noiseandhighresolutionspectra UBVRI photometry for12starsofthenewlydefinedsample.Theyfoundthat9stars UBVRI photometry for the newly-identified objects was areaffectedbythedepletionprocess.Inadetailedstudyofone obtained from the South African Astronomical Observatory object, IRAS08544 (Maasetal. 2003); orbital elements were (SAAO) service observing programme(Table C.2). This uses found which show that this star must have undergone severe the 0.5m Telescope and the 0.75m Automatic Photometric binaryinteractionwhenitwasan(asymptotic)giant. Telescope(APT)atSutherland.IftheUBVRI dataofthestar Inthiscontributionweanalysethebroad-bandSEDsofall show large variations mainly due to a large variability of the 20newlycharacterizedstarsofthissample. star, we plotboth minima and maxima.If the numberof data pointsavailableistooloworiftherearenobigvariations,we useameanUBVRI. 2.4. TotalSample WealsosearchedforJohnsonandCousinsbroad-bandpho- tometryin the literatureandthese dataare givenin TableC.2 The 51 objects of our sample are presented in Table 1. aswell. In Col. 11 we differentiate between the several subtypes. Statisticalstudiesontheoccurrenceandcharacteristicsofpost- AGBstarswithrespecttoAGBstarsand/orPlanetaryNebulae Near-IR data (PNe)donotexistyet.Inthemostextensivecataloguedescrip- JHKL photometry of the newly-identified objects was ob- tion found in the literature, Szczerbaetal. (2001) consider a tainedbyT.LloydEvanswiththe0.75mreflectoroftheSouth Galactic sample of about 220 post-AGB stars known at that AfricanAstronomicalObservatory. time. Although the overlap between the specific sample dis- These data points were complemented with near-IR data cussedhere,andthetotalsampleofSzczerbaetal.(2001)isfar from the 2MASS and DENIS projects and with other fromcomplete,thesampleof51objectsdiscussedhere,forms asignificantpopulationofpost-AGBstarsknowntodate.Note JHKLM datapointsfromtheliterature(TableC.3). thatoursamplewasdefinedonthebasisofaspecificsetofcri- teriaexplainedaboveandoursampleisnotmeanttocoverall Far-IR data knownevolvedstarswithsimilarSEDcharacteristics. For the characterization of the longer wavelength part of the SEDs we use far-IR data from the IRAS Point Source Catalogue (Beichmanetal. 1988, Table C.4); for some 3. Broad-bandphotometricdata stars data from the MSX Infrared Point Source Catalogue (Eganetal. 2003) are also available (Table C.5). The IRAS In order to reconstructthe complete SEDs of our programme satellitehadfourpassbands,at12,25,60and100µm,respec- stars, we combined different sets of broad-band photomet- tively. The SPIRITIII instrument on board the MSX satellite ric data: our own Geneva optical photometry, supplemented had6passbands,at4.29,4.35,8.3,12.1,14.7and21.3µm. with UV, optical, near-IR, far-IR and submillimetre photom- etryfoundintheliterature. Submillimetredata The main difficulty in constructing the SEDs of pulsating stars with large amplitudes, like the RVTauri objects, is the For six stars (TWCam, RVTau, SUGem, UYCMa, UMon acquisitionofequallyphaseddataoverawidespectraldomain. and ACHer) submillimetre data (see Table C.6) are available Sincethesedataarenotavailable,welimitedourstudyofthe from DeRuyteretal. (2005), for HR4049 and IRAS20056 broad-bandenergetics to the phases of minimal and maximal weusedatafromtheotherliterature.ForHR4049,HD52961 coveredbrightness. and89Her,wehavesomenewlydeterminedSCUBA850mi- S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 7 croncontinuummeasurements.Theobservationswerecarried AnappropriatephotosphericKuruczmodel-basedonT , eff out with the 15m James Clerk Maxwell Telescope (JCMT) logg and [Fe/H]- is used for the unattenuated stellar photo- atMaunaKea,Hawaii,duringMarch-April1999(programme spheres(Kurucz1979). M99AN08). SCUBA (Hollandetal. 1999) was operated in photometrymodetosimultaneouslyobtaindataat850µmand 4.2. Colourexcess E(B −V) determination 450µm. For the data reduction, the standard software SURF wasusedwithMarsasafluxcalibrator.Table2detailsthenew Lightcomingfromastarisattenuatedandreddenedbymate- 850µmcontinuummeasurements,whilefor450µm,nosignif- rialintheline-of-sight.Notethatthetotalline-of-sightextinc- icant signal was detected. Note that the new 850µm SCUBA tionislikelytocontainbothacircumstellarandaninterstellar datapointforHR4049is,withintheerrors,thesameastheone component. inDominiketal.(2003). WeestimatethetotalcolourexcessE(B−V)byusingthe average interstellar extinction law given by Savage&Mathis (1979) to deredden the observed maximal UV-optical fluxes. Table2.Newobservationaldataat850µmfromSCUBA. E(B−V)isfoundbyminimizingthedifferencebetweenthe dereddenedobservedfluxesin the UV-optical,and the appro- No Name F850(mJy) priate Kurucz model (Kurucz 1979). We scale to the J filter 11 HD52961 2.8±1.9 whichisthereddestfilterwherenodustexcesscanbeexpected. 21 HR4049 8.7±2.8 Weassumethecircumstellarreddeninglawtobesimilartothe 38 89Her 40.9±2.4 ISMlaw.DeterminingtheE(B−V)inthiswayimpliesthat wedon’tcorrectforthecontributionofthegreyextinction. Results are in Table A.1. The erroron E(B −V) is typi- For ACHer we have a flux point at 1.1mm from cally0.1.Butchangingthestellarmodelsby±250Kineffec- Shentonetal.(1995). tive temperature causes a change in E(B −V) of about 0.2. Thus,togetherwiththeerrorof0.2inducedbytheuncertainty Other ofthe temperatureoftheunderlyingphotosphere,we havean uncertaintyof0.3onthetotalextinctionduringmaximallight. At the short wavelength side, we use the IUE data Adistributionofthetotalcolourexcessesfoundforoursample (InternationalUltravioletExplorer,0.115µm-0.320µm)from starsisshownin Fig.1. Notethatthetotalreddeningissmall the newly extractedspectral data release (INES Archive Data formostofthestars.Wemaythusassumethatasituationwith Server).Ifthereweremultiplespectraavailable,weonlycon- onlygreyextinctionwouldberatherexceptional. sideredthemaximalflux. 4. SpectralEnergyDistributions We analyse the SEDs of our sample stars in a homogeneous andsystematicway. 4.1. Thestars photosphere:modelparameters ThedeterminationofthemodelatmosphericalparametersT eff and logg, and the overall metallicity [Fe/H] is based on the analysisofhighresolutionspectrausedinthechemicalanaly- sesofthestars.Thosespectraarepreferentiallytakenatmax- imum light, because of the minimal molecular veiling during thehotterphaseinthelightcurve(Giridharetal.2000). For most stars we use model parameters deduced from our own spectra (e.g VanWinckel 1997; Maasetal. 2003, Fig.1. The distribution of the total E(B −V) found for our 2005), for others we use values found in the literature. For sample.Remarkthatmostcolourexcessesarelessthan1.0. someofoursamplestars(IRAS05208,UYCMa,IRAS10456, IRAS18123, IRAS18158, VVul), however, we don’t have a spectrum nor do we find any estimates for the photospheric parameters in the literature. Here we deduce the parame- 4.3. A first-order approximation:an opticallythin ters on the basis of the spectral type of the star. And for dust fit someothers(IRAS09400,IRAS10174,GKCar,IRAS13258, IRAS15556)welackopticalphotometry.Themodelparame- Abroad-bandSEDhaslimiteddiagnosticvalueforconstrain- tersaregiveninTable1,Cols.7to9.Thereferenceswherewe ing the chemistry and spatial distribution of the circumstellar foundtheparametersareshowninCol.10. dust. We apply an optically thin dust modelas a simple first- 8 S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? order approximation to fit the SEDs with the prime goal to and IRAS20056) are omitted in the figure. Like HD44179 quantifythedifferencesinSEDcharacteristicsbetweenthedif- (Cohenetal. 2004) and IRAS20056 (Menzies&Whitelock ferentobjects.We use theopticallythindustmodeldescribed 1988; Raoetal. 2002) we suspect that we only see the pho- bySopkaetal.(1985).Inthismodelthedustisassumedtobe tosphere of those sources throughscattered light. This means distributedsphericallysymmetricaroundthestarfrominnerra- thatweseethesesystemsnearlyedge-on. diusr to outer radiusr . Since we applyan opticallythin in out model, the geometry of the dust is not constrained as we ob- servethewholevolumeandalldustparticlescontributetothe observedflux. For moredetails of the modelwe refer to DeRuyteretal. (2005).ThefluxF isdeterminedbyfiveparameters:thenor- ν malizationtemperatureT ,theinnerradiusofthedustshellr , 0 in theouterradiusofthedustshellr ,thespectralindexpand out thedensityparameterm.Applyingaleastsquareminimization thefreeparametersr ,r ,pandmaredetermined.Results in out aregiveninTableA.1. 4.4. SEDs InFig 2 the SEDsof a selectionofpost-AGBstarsare given. The other SEDs are givenin Fig. A.1. The dereddenedfluxes areplottedtogetherwiththescaledphotosphericmodelrepre- senting the unattenuated stellar photosphere. We plotted both Fig.3.ThedistributionoftheenergyconversionratioLIR/L∗ minimal(opentriangles) and maximal(openrectangles)pho- foundforoursample. tometry. HD44179andARPuparetworatherexceptionalstars,for For those stars where we have a submillimetre data point whichweadoptedanotherstrategytodeterminetheSEDs.The (see Table C.6) the SED followsa Rayleigh-Jeansslope from dust excess for HD44179 and ARPup starts even at shorter the 60-100µm flux point redwards. Assuming that the dust wavelengths making it impossible to scale the model photo- emissivity at far-IR wavelengths follows a power law (Q ∼ ν spheretotheJ filter.WethereforescaletotheGenevaGband νp),thespectralindexpasdeterminedfromtheslopebetween datapointinourleastsquaresprocedure. the 60-100µmemission andthe 850µmflux pointisclose to TheIRexcessduetothepresenceofcircumstellardust,is zero.Thesmallspectralindicesintherangefrom0.0to0.5for inallcasesverysignificantwithBD+39◦4926anoticeableex- most objectsare consistentwith the presence of a component ception for which no IRexcess was found. A generalcharac- oflargegrains(radius&0.1mm)inthecircumstellarenviron- teristic is that the dust energy distribution peaks at very high mentofthestars.Forthosestarswherewelackasubmillimetre temperaturesandthatthereisnoevidenceforlargeamountsof data point, the spectral index of the long wavelength slope is cool(T ≤100K)dust:thepeakofthedustSEDsliesaround notwellconstrained. d 10µm and in some cases even bluer. In most cases, the dust Note that the 100µm IRAS fluxes for some stars are af- excessstartsnearthedustsublimationtemperature. fected by Galactic cirrus, so they present upper limits. In the To determine the amount of energy reprocessed by the figuresanarrowisdrawn.Thismakesitstillmoredifficulttobe dust grains in the circumstellar environment,we compute the sureabouttheslopefromthe60-100µmfluxpointredwards. energy ratio LIR/L∗. We first calculate the stellar flux L∗ From all SED characteristicsmentioned,we infer thatthe by numerically integrating the scaled Kurucz model between dustexcessin the SEDs isclearly differentfromwhatwe ex- 145nm and 850µm. This gives a good estimate of the unat- pecttoobserveinapost-AGBstarwheretheexcessrepresents tenuated stellar flux. The energy radiated at even shorter and the expandingand cooling relic of the strong AGB mass-loss longer wavelengths is only a negligible fraction (. 10−4) of episode(s). the total stellar flux so we evaluate our wavelength integral boundaries as adequate. Integrating over the IRexcess model 5. Distanceestimates described in Section 4.3 yields L . We note that the ratio IR LIR/L∗(TableA.1)ishighfor78%oftheobjects(largerthan The typical luminosity of a lower mass post-AGB star is ex- 20%).Assumingthepresenceof20%greyornon-selectiveex- pectedtobebetween1000and10000L⊙.Accurateluminosi- tinction and a mean E(B −V) of 0.5, reduces LIR/L∗ with tiesofindividualpost-AGBobjectsarestilllargelyunavailable 25%, which remains, nevertheless, still large. The absorption sincetheyaregenerallytoofartoobtainreliableparallaxes. andthermalre-radiationofthestellarradiationbythecircum- In the few cases where reliable Hipparcos parallaxes are stellardustisonaverageveryefficient.InFig.3thedistribution availablewehaveadirectprobeofthedistanceandhencelu- oftheenergyconversionratiosisshown.Notethatratioslarger minosity.ResultsaregiveninTableA.2.Notethelargeerrors than 1.0 (for HD44179, ARPup, IRAS17233, IRAS18158 ontheparallaxestranslateinalargeerrorboxforthedistance S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 9 Fig.2.TheSEDsofaselectionofoursampleofpost-AGBstars:HD52961andHD213985areexamplesofconfirmedbinary post-AGBstars,RVTauandIWCararegenuineRVTauristarsandIRAS17233andIRAS18123areRVTaurilikeobjectsfrom thenewlistofT.LloydEvans.ThedereddenedfluxesaregiventogetherwiththescaledphotosphericKuruczmodelrepresenting the unattenuatedstellar photosphere(solid line). An optically thin dust modelwas used to fit the IRexcess (dottedline). Data foundintheliteraturetogetherwithour7bandGenevaphotometry(onlythemaxima)areplottedastriangles.Theminimaldata points(squares)werenotusedforthedeterminationofE(B−V).Theyareshowntogiveanindicationoftheamplitudeofthe pulsations.Whereavailable,crossesrepresentour850µmSCUBAdatapoint.Errorbarsonthe850µmSCUBAdatapointare plottedaswell,forsomeobjectsthoughsmallerthanthesymbols.Thearrowatthe100µmfluxpointsignifiesanupperlimit. and the luminositydeterminations.The parallaxof HD44179 seeonlyscatteredlight(Cohenetal.2004).Thesameisprob- cannotbeusedtoinferaluminositybecauseweknowthatwe ablytrueforARPup(seeSect.4.4). 10 S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? Alcocketal. (1998) reported the discovery of RVTauri against the effectivetemperaturefor all stars and in all cases, stars in the Large Magellanic Cloud (LMC). In light- and thereisdustatorverynearsublimationtemperature.Withtyp- colour-curve behaviour, the classical RVTauri stars appeared icalluminosityestimates,thissublimationtemperatureedgeis to be a direct extension of the type II Cepheids to longer pe- atadistancesmallerthanabout10AUfromthecentralsource. riods.WeusetheP-LrelationofAlcocketal.(1998)-derived Moreover,theSEDsrevealthatthepresenceofdust,veryclose for the LMC RVTauri stars- to derive the luminosity for the totheobject,isirrespectiveoftheeffectivetemperatureofthe pulsatingobjectsinoursample.ForvariableswithP/2>12.6 centralstar.Notethatnoneoftheobjectsshowsevidencefora daysweuse present-daydustymass-loss.Wethereforeinferthatpartofthe dust must be gravitationally bound:any typical AGB outflow M =2.54(±0.48)−(3.91(±0.36))log(P/2), velocitywouldbringthedusttocoolerregionswithinyears. V σ =0.35. (2) Withinour sampleof post-AGBobjectsconsidered- con- firmedbinarypost-AGBstars,classicalRVTauristarsandnew Notethattheintrinsicscatterislarge,implyingverysignif- RVTauri-likeobjects-thereisawiderangeinthestrengthof icantuncertaintiesintheestimatedabsolutemagnitudesofthe the total IRexcess, but the shape of the IRexcess thus indi- objects.InEq.2weusethefundamentalperiodP/2,definedas cates that in all systems, the circumstellar shell is not freely the timebetweena deepand a shallowminimum.Theformal expandingbutstoredinthesystem.Wearguethereforethatthe periodP isshowninTableA.2.Weapplythisextensionofthe same inner geometryas in the resolvedsystem HD44179ap- period-luminosityrelationknownfortypeIICepheids,forthe plies to the whole sample: the objects seem to be surrounded RVTauri stars from the GCVS to derive an estimate for their byaKepleriandisc. luminosities.Results,togetherwiththebolometriccorrections From the dust modelling fit it is also clear that the outer BCV (Besselletal.1998)used,arelistedinTableA.2.Theer- radiiarenotverylargeeither.Objectssimilartotheenigmatic rorpropagationstartingfromtheerrorsintheP-Lrelationare extreme HR4049 are UYCMa, UMon and IRAS17233 for giveninthetableaswell.Theobjectsareindeedluminous,as whichthedusttemperaturegradientobservedisverysmall. canbeexpectedfrompost-AGBobjects,butthelargescatterin theP-Lrelationpreventsmoreaccurateluminosityestimates. Fortheotherobjects,wherewedon’thaveapulsationpe- riod determination,we take L=5000 ± 2000L⊙. Note that itisnotknownwhethertheP-LrelationoftheLMCstarsisdi- rectlyapplicabletoGalacticstars.Ingeneral,theluminosities foundare smaller than,or equalto, about5000L⊙ indicating thatthepopulationisofratherlowinitialmass. By comparing the integrated fluxes of the scaled Kurucz model with the luminosity deduced from the P-L relation or the default assumed value of L=5000L⊙, we calculate the distanceD tothestar. Thedetermineddistancesareshownin Table A.2. The propagated error of the poorly calibrated P-L relation, yields very uncertain luminosities and therefore also uncertaindistances. Remark that the luminosities determined by the P-L re- lation will not be appropriate to stars seen with the dusty Fig.4.Theinnerradiiareverysmall.Thedustmustbelocated disc nearly edge-on. Besides for the well-known example veryclosetothestar:thedustexcessstartsfornearlyallobjects HD44179, this is likely the case for ARPup, IRAS17233, nearthedustsublimationtemperature.AllSEDsshowa clear IRAS18158andIRAS20056forwhichthedistanceslistedin near-IRexcess. TableA.2areclearlyupperlimits. Other indirect indications for the presence of gravitionally 6. Discussion bounddustare: We presented the SEDs of all stars in our sample based on literature data supplemented with our own Geneva photome- - Despite the significant IRexcess, the total E(B −V) for tryand-whereavailable-continuummeasurementsat850µm moststarsisverysmall.Efficientredistributionofthestel- (Fig. 2 and Fig. A.1). Despite the differentcriteria for the se- larfluxisclearlyincontradictionwith thelackofline-of- lectionofthethreesubsamples,itisclearthatthebroad-band sight reddening.This implies a significant grey extinction SEDs display a high degree of uniformity.Between extremes and/oranon-sphericallysymmetricdustdistribution.Since likeHR4049(mostcompactSEDwithasingledusttempera- theobjectsallhavearathersmalltotalextinction,thelatter ture)andHD44179(largestLIR/L∗),theshapeoftheSEDsis ismoreplausible.AgoodindependentestimateoftheISM verysimilarforallstars. contributiontothetotalextinctionintheline-of-sightofall Oneofthemostremarkablefeaturesisthestartofthedust individualobjectsisstilllacking.Itmust,however,beare- excess.InFig.4weplottheinnerradiiofthedustcomponents wardingexperimenttomeasurethosesinceitwouldhelpin

ebook img

Keplerian discs around post-AGB stars: a common phenomenon? PDF

1.4 MB·English
by  S. De Ruyter
#additional_collections #journals #arxiv
Save to my drive
Quick download
Download
Upgrade Premium
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Keplerian discs around post-AGB stars: a common phenomenon?

1 We aim at showing that the broad-band SED characteris- tics of our sample of post-AGB stars are best interpreted, as- suming the circumstellar dust is stored in Keplerian rotating passivediscs.Wepresentahomogeneousandsystematicstudy oftheSpectralEnergyDistributions(SEDs)ofasampleof51 post-AGB objects. The selection criteria to define the whole sample were tuned to cover the broad-bandcharacteristics of knownbinarypost-AGBstars. The whole sampleincludes20 dusty RVTauri stars from the General Catalogue of Variable Stars(GCVS).WesupplementedourownGenevaopticalpho- tometry with literature data to cover a broad range of fluxes from the UV to the far-IR. All the SEDs display very simi- larcharacteristics:alargeIRexcesswithadustexcessstarting near the sublimation temperature,irrespectiveof the effective 6 0 temperatureof the centralstar. Moreover,when available, the 0 longwavelengthfluxesshowa black-bodyslope indicativeof 2 thepresenceofacomponentoflargemmsizedgrains.Wear- n gue that in all systems, gravitationally bound dusty discs are a present.Thediscsmustbepuffed-uptocoveralargeopening J angleforthecentralstarandwearguethatthediscshavesome 5 similaritywiththepassivediscsdetectedaroundyoungstellar 2 objects.Weinterpretthepresenceofadisctobeasignaturefor binarity of the central object, but this will need confirmation 1 v bylong-termmonitoringoftheradialvelocities.Wearguethat 8 dusty RVTauri stars are those binarieswhich happento be in 7 thePopulationIIinstabilitystrip. 5 1 Keywords.stars:AGBandpost-AGB–stars:binaries:gen- 0 6 eral–stars:circumstellarmatter 0 / h p - o r t s a : v i X r a Astronomy&Astrophysicsmanuscriptno.4062sdr February3,2008 (DOI:willbeinsertedbyhandlater) Keplerian discs around post-AGB stars: ⋆ a common phenomenon? S.DeRuyter1,H.VanWinckel2,T.Maas2,3,T.LloydEvans4,L.B.F.MWaters2,5,andH.Dejonghe1 1 SterrenkundigObservatorium,UniversiteitGent,Krijgslaan281S9,9000Gent,Belgium 2 InstituutvoorSterrenkunde,K.U.Leuven,Celestijnenlaan200B,3001Leuven,Belgium 3 DepartmentofAstronomy,UniversityofTexas,Austin,TX78712 4 SchoolofPhysicsandAstronomy,UniversityofSt.Andrews,NorthHaugh,St.Andrews,Fife,ScotlandKY169SS 5 SterrenkundigInstituut‘AntonPannekoek’,UniversiteitAmsterdam,Kruislaan403,1098Amsterdam,TheNetherlands Received¡date¿/Accepted¡date¿ Abstract. 1. Introduction belinkedtothepole-to-equatordensitycontrast,asdetermined on the basis of high spatial resolution mid-infrared images Post-AGB stars are low and intermediate initial mass (≤ 8- (Meixneretal.1999).RecentlyalsoGledhill(2005)foundev- 9M⊙) stars which have suffered a large and dusty mass-loss idence for axi-symmetry in the dust density in his polarimet- phaseattheendoftheAsymptoticGiantBranch(AGB)during ricimagingsurveyofcandidatepost-AGBstars.Thedetached whichalmostthewholestellarenvelopewasexpelled.Theyare shellscorrespondtostarswithanopticallythinexpandingcir- evolvingatconstantluminosityonafastevolutionarytrackin cumstellarenvelopewhereasthebipolarandunresolvedtargets whichthecentralstarcrossestheHR-diagramfromacoolAGB have optically thick dust structures, probably in the form of photospheretotheionizingtemperatureofthecentralstarofa discs. It is suggested once again that this bifurcation in mor- PlanetaryNebula(PN).Giventhesmallevolutionarytimescale phologyisrootedinthepresenceorabsenceofabinarycom- of about104years, post-AGBstars are rare and not manyare panion,whichdetermineswhetherornotadiscforms. known(e.gSzczerbaetal.2001;VanWinckel2003). DiscussionsonthemorphologyofPNeusuallystartwitha Stunningkinematicinformationresultedfromtheextensive displayoftheaestheticpicturesoftheHubbleSpaceTelescope COsurveyofBujarrabaletal.(2001):itappearstobeafunda- (HST) showing the complex geometry and structure of the mental property of the omnipresentfast molecular outflow in nebulae,immediatelyfollowedbythepuzzlingandcontradic- PPNe, that it carries a huge amount of linear momentum, up tory finding that, on the AGB, the mass-loss is found to be to 1000 times the momentum available for a radiation driven sphericallysymmetric.Duringthetransitiontime,thestarand wind. Clearly, other momentum sources have to be explored. circumstellar envelope must undergo fundamental and rapid Some molecular jets are resolved by high spatial resolution changesinstructure,mass-lossmodeandgeometrywhichare imaging (Sahai 2004). The formation process of the strongly still badly understood. The debate on which physical mecha- collimatedjetsis,however,stillbadlyunderstood.Anintrigu- nisms are driving the morphology changes gained even more ingsuggestionisthattheprocessessimilartothejetformation impetus from the finding that also resolved cooler post-AGB inlow-massyoungstellar objectsoperateandthatthejetsare starsorProto-PlanetaryNebulae(PPNe)displayasurprisingly borninaccretiondiscs.Thismechanismrequiresasignificant widevarietyinshapesandstructures,veryearlyintheirpost- amountofmassorbitingthepost-AGBstar.Suchadisccould AGBevolution(Balick&Frank2002,andreferencestherein). be present, but likely only in binary stars. Testing of such an hypothesis is severely hampered by the lack of observational In a survey of 27 PPNe, 21 were found to be resolved informationonbinarityinPNeandPPNebutalsoonourpoor (Uetaetal. 2000). Moreover, the degree of asymmetry could theoreticalunderstandingofAGBevolutioninbinarysystems. Send offprint requests to: S. De Ruyter, e-mail: NotethatmostobjectswhichweredetectedinCOarestrongly [email protected] embedded and the sample is probably biased towards more ⋆ BasedonobservationswiththeP7photometeratthe0.7mSwiss massivePPNe. Telescope,LaSilla(Chile)andattheFlemishMercatorTelescope,La Palma(Spain)andonobservationswiththeSubmillimetreCommon- ProbingbinarityinPNeandembeddedPPNedirectlywith User Bolometer Array (SCUBA) at the James Clerk Maxwell radialvelocitymonitoringisnoteasy.Foropticallybrightpost- Telescope(JCMT),MaunaKea,Hawaii. AGBstars,thisisdifferentandsomefamousexamplesexistin S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 3 whichitisclearthatthebinarynatureofthecentralobjectmust Our sample and the selection criteria are presented in haveplayedakeyroleintheevolutionofthesystem. Sect. 2. In Sect. 3 we present our photometric data comple- mented with the literature values and the results of the large ThemostfamousexampleiscertainlyHD44179,thecen- all sky surveys. In Sect. 4 we present the detailed Spectral tral star of the Red Rectangle nebula. It displays a huge and Energy Distribution (SED) construction of all individual ob- broad IRexcess. The IR luminosity is 33 times stronger than jects. Distances to the objects are estimated in Sect. 5. We the optical luminosity (Leinert&Haas 1989; Waelkensetal. end this extensive study with a detailed discussion in Sect. 6. 1996).VanWinckeletal.(1995)haveshownthatHD44179is ConclusionsaresummarizedinSect.7. aspectroscopicbinarywithanorbitalperiodof298days.The eccentricityisaremarkablyhighe=0.37.Sinceitsdiscovery by Cohenetal. (1975), HD44179 has often been used as an 2. Programmestars archetypicalexampleofaC-richpost-AGBobject,butitisnow Our total sample of 51 stars was defined on several observa- generally accepted that many of the remarkable phenomena tional criteria inspired by the characteristics of known post- andthepeculiarmorphologyofthenebula(foranoverviewsee AGB stars in binary systems. The whole sample is given Cohenetal.2004)arecloselyrelatedtothepresenceofastable in Table 1. Besides these proven binaries, we discuss also circumbinarydiscaroundthebinarycentralstar.Thelongevity RVTauristarswithIRASdetectionsofgoodqualityandasam- ofthediscwasdramaticallyconfirmedbythedetectionofcool ple of newly identified objects, originally found by one of us O-richcrystallinesilicate dustgrainsin thedisc (Watersetal. (T.LloydEvans)insearchofnewcandidateRVTauristars. 1998).Themixedchemistryisbestexplainedassumingthefor- mationoftheO-richdiscpredatedthemorerecentC-richtran- sitionofthecentralobject.RecentSpitzerdatashow,however, that also far from the central star, the nebula appears to have a dust component which is O-rich (Markwick-Kemperetal. 2005). The (chemical) history of the binary, disc and neb- ula is therefore far from understood. The disc is resolved in ground-based high spatial-resolution imaging at optical and near-IRwavelengths(Men’shchikovetal.2002,andreferences therein)aswellasinHSTopticalimages(Cohenetal.2004). The disc was also resolved in interferometric CO(2-1) maps, andtheKepleriankinematicsofthediscweredirectlydetected (Bujarrabaletal.2003,2005).HD44179showsaconsiderable amount of dust processing in the disc with indications of the presence of very large grains (Juraetal. 1997) and possibly evenmacro-structures(Jura&Turner1998). Anotherremarkableevolvedobjectwithalong-liveddiscis HR4049.Itisabinarywithanorbitalperiodof430dayswith aremarkablyhigheccentricityofe=0.30.Alsointhisobject, thecircumstellarmaterialshowsa mixedchemistrywith both carbon rich and oxygenrich features. The SED of the dust is also very peculiar as it can be fitted with a single black-body ofabout1150K,from1µmdownto850µm.TheseSEDchar- acteristicsareveryconstrainingandthebestmodelforthecir- cumstellarmaterialis,thatthedustistrappedinaveryopaque dusttorusatKeplerianrotation(Dominiketal.2003).Itisclear that also in this object the dusty disc plays a lead role in the (future) evolution. Other examplesexist and there is substan- tial observationalevidence that the systems are all likely sur- roundedby a circumstellar orbitingdisc (VanWinckel 2003). NotethatsofaronlyfortheRedRectangletheKepleriandisc is spatio-kinematically resolved by CO interferometric maps (Bujarrabaletal.2005). Togaininsightintheevolutionofbinarysystemsandtheir circumstellar material, we reportin this contributionon a ho- mogeneous and systematic study of a sample of objects with similarIRcharacteristicsastheknownbinarypost-AGBstars. Themainaimistostudythebroad-bandSEDsinasystematic way, in order to gain insight in the possible evolutionarylink betweenthedifferentobjects. Table 1. The IRASnumber,the HDnumber or the name from the GCVS, the spectral type, the equatorialcoordinatesα and δ(J2000), the effective temperature T , the 4 eff surfacegravitylogg,themetallicity[Fe/H],thereferenceforthemodelparameters,thetypeofobject(post-AGB,RVTauriorNewSample)andareferencewheretheorbital motionoftheobjectcanbefound,aregiven.NotethatIRAS11472−0800isastronglydepletedobjectaddedtooursamplestars. 1 2 3 4 5 6 7 8 9 10 11 12 No IRASnumber HDnumberor Spectral α(J2000) δ(J2000) Teff logg [Fe/H] ReferenceModelParameters Type ReferenceBinarity GCVSname type (hms) (◦′′′) (K) (cgs) 1 IRAS04166+5719 TWCam G3I 042048.1 +572626 4800 0.0 −0.5 Giridharetal.(2000) RVTauri 2 IRAS04440+2605 RVTau G2I 044706.8 +261044 4500 0.0 −0.4 Giridharetal.(2000) RVTauri 3 IRAS05208−2035 M0e+F 052259.423 −203253.03 4000 0.5 0.0 fromspectraltype NewSample 4 IRAS06034+1354 DYOri G0I 060612.3 +135309 6000 1.5 −2.0 Gonzalezetal.(1997b) RVTauri 5 IRAS06072+0953 CTOri F9I 060957.4 +095235 5500 1.0 −2.0 Gonzalezetal.(1997a) RVTauri S. 6 IRAS06108+2743 SUGem G5I 061400.8 +274212 5750 1.125 −0.7 Wahlgren(1992) RVTauri D 7 IRAS06160−1701 UYCMa G0V 061816.367 −170234.72 5500 1.0 0.0 fromspectraltype RVTauri eR 8 IRAS06176−1036 HD44179 F1I 061958.2160 −103814.691 7500 0.8 −3.3 Waelkensetal.(1992) post-AGB VanWinckeletal.(1995) u 9 IRAS06338+5333 HD46703 F3I 063752.4253 +533101.957 6250 1.0 −1.5 fromspectrum post-AGB Hrivnak,privatecommunication yte 10 IRAS06472−3713 STPup F7I 064856.4131 −371633.332 5750 0.5 −1.5 Gonzalez&Wallerstein(1996) RVTauri Gonzalez&Wallerstein(1996) r 11 IRAS07008+1050 HD52961 F6I 070339.6314 +104613.067 6000 0.5 −4.8 Waelkensetal.(1991b) post-AGB Waelkensetal.(1992) et 1132 IIRRAASS0077218440−−02934201 USAMOo1n73329 FG50II 007713600487..2571 −−203924760317.61 75000000 10..50 −−00..88 VGairnidWhainrcekteall.(1(2909070)) pRoVstT-AauGriB VPoalnlaWrdin&ckCeoltettrealll.((11999955)) al.: K 14 IRAS08011−3627 ARPup F0I 080301.1 −363547 6000 1.5 −1.0 Gonzalezetal.(1997b) RVTauri e 1156 IIRRAASS0089504640−−42483017 FF35 008950681140..1182 −−424841391100.73 76255000 11..55 −−00..55 MMaaaasseettaall..((22000035)) NNeewwSSaammppllee Maasetal.(2003) pler 17 IRAS09144−4933 G0 091609.1 −494606 5750 0.5 −0.5 Maasetal.(2005) NewSample ian 18 IRAS09256−6324 IWCar F7I 092653.4 −633748 6700 2.0 −1.0 Giridharetal.(1994) RVTauri d 19 IRAS09400−4733 M0 094152.9 −474703 NewSample is c 20 IRAS09538−7622 G0 095358.5 −763653 5500 1.0 −0.5 Maasetal.(2005) NewSample s 21 IRAS10158−2844 HR4049 A6I 101807.5903 −285931.201 7500 1.0 −4.5 Dominiketal.(2003) post-AGB Waelkensetal.(1991b) ar 22 IRAS10174−5704 K:rr 101918.1 −571936 NewSample ou 23 IRAS10456−5712 HD93662 K5 104738.3965 −572802.679 4250 0.5 0.0 fromspectraltype NewSample nd 24 IRAS11000−6153 HD95767 F0I 110204.314 −620942.84 7600 2.0 0.1 VanWinckel(1997) post-AGB VanWinckeletal.(1995) p 25 IRAS11118−5726 GKCar M0I 111401.3 −574309 RVTauri os 26 IRAS11472−0800 F5I 114948.5 −081721 5750 1.0 −2.5 fromspectrum t-A 27 IRAS12067−4508 RUCen F6I 120923.7 −452535 6000 1.5 −2.0 Maasetal.(2002) RVTauri G 28 IRAS12185−4856 SXCen G3V 122112.6 −491241 6000 1.0 −1.0 Maasetal.(2002) RVTauri Maasetal.(2002) B 2390 IIRRAASS1123222528−−48615023 HD108015 FG35IbDe 113232140573..1501 −−841710893007.51 7000 1.5 −0.1 VanWinckel(1997) Npoeswt-ASaGmBple star 31 ENTrA F2Ib 145700.6847 −685022.879 6000 1.0 −0.5 VanWinckel(1997) RVTauri VanWinckeletal.(1995) s: 32 IRAS15469−5311 F3 155044.0 −532044 7500 1.5 0.0 Maasetal.(2005) NewSample ac 33 IRAS15556−5444 F8 155932.1 −545318 NewSample o m 34 IRAS16230−3410 F8 162620.29 −341712.3 6250 1.0 −0.5 Maasetal.(2005) NewSample m 35 IRAS17038−4815 G2p(R)e 170736.3 −481908 4750 0.5 −1.5 Maasetal.(2005) NewSample o 36 IRAS17233−4330 G0p(R) 172657.5 −433313 6250 1.5 −1.0 Maasetal.(2005) NewSample n 37 IRAS17243−4348 LRSco G2 172756.1 −435048 6250 0.5 0.0 Maasetal.(2005) NewSample ph 38 IRAS17534+2603 89Her F3I 175525.1889 +260259.966 6500 1.0 0.0 Watersetal.(1993) post-AGB Watersetal.(1993) en 39 IRAS17530−3348 AISco G4I 175618.5 −334847 5000 0.0 0.0 fromspectrum RVTauri om 40 IRAS18123+0511 G5 181449.4 +051255 5000 0.5 0.0 fromspectraltype NewSample e 41 IRAS18158−3445 F6 181913.6 −344432 6500 1.5 0.0 fromspectraltype NewSample no 42 IRAS18281+2149 ACHer F4I 183016.2 +215200 5500 0.5 −1.5 VanWinckeletal.(1998) RVTauri VanWinckeletal.(1998) n? 43 IRAS18564−0814 ADAql G8I 185908.1 −081014 6300 1.25 −2.1 Giridharetal.(1998) RVTauri 44 IRAS19125+0343 F2 191500.8 +034841 7750 1.0 −0.5 Maasetal.(2005) NewSample 45 IRAS19163+2745 EPLyr A4I 191817.5 +275038 7000 2.0 −1.5 Gonzalezetal.(1997b) RVTauri 46 IRAS19157−0247 F3 191822.5 −024209 7750 1.0 0.0 Maasetal.(2005) NewSample 47 IRAS20056+1834 QYSge G0De 200754.8 +184257 5850 0.7 −0.4 Raoetal.(2002) NewSample 48 IRAS20117+1634 RSge G0I 201403.8 +164335 5750 0.0 −0.5 Gonzalezetal.(1997b) RVTauri 49 IRAS20343+2625 VVul G8I 203631.8 +263617 5250 1.0 0.0 fromspectraltype RVTauri 50 IRAS22327−1731 HD213985 A2I 223527.5259 −171526.889 8250 1.5 −1.0 fromspectrum post-AGB VanWinckeletal.(1995) 51 BD+39◦4926 A9I 224611.2273 +400626.294 7500 1.2 −2.9 VanWinckeletal.(1995) post-AGB Kodairaetal.(1970) S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 5 2.1. Confirmedbinarypost-AGBstars WiththeirhighluminosityandoftenlargeIRexcessesdue to thermal radiation from circumstellar dust, there is general Thegroupofbinarypost-AGBstarsinoursamplewasassem- agreement that RVTauri stars are low-mass objects in tran- bled in a rather coincidentalway on the basis of independent sition from the Asymptotic Giant Branch (AGB) to white detailedstar-to-staranalyses.We includedtheprovenbinaries dwarfs (Jura 1986). It was noted already in the seventies that in our programme star sample and in Col. 12 of Table 1 we manyRVTauristarsshowaconsiderablenear-IRexcesscaused giveareferencewheretheorbitalmotionoftheobjectswasdis- by a hot dust component (Gehrz&Ney 1972; Gehrz 1972; cussed.TheorbitalelementsthemselvesarelistedinTableB.1. LloydEvans1985),whichwasattributedtothepossiblepres- The photospheres of several binary post-AGB stars enceofadustydisc.Morris(1987)suggestedthatcircumstel- (e.g. HR4049 (Lambertetal. 1988; Waelkensetal. lar dust could indeed exist in a disc structure in binary sys- 1991a), HD44179 (Waelkensetal. 1992), HD52961 tems, either in a circumbinary disc or in a disc around one (VanWinckeletal. 1992) and BD+39◦4926 (Kodairaetal. ofthecomponents.TheweakcircumstellarCOemission(e.g. 1970)) are strongly affected by a poorly understood selective Bujarrabaletal. 1988) together with the black-body spectral depletion process. The basic scenario of this process is that index at long wavelengths observed in some RVTauri stars circumstellar gas is separated from the dust and subsequently (DeRuyteretal.2005)corroboratetheconclusionthatthecir- re-accretedontothestar(Mathis&Lamers1992;Watersetal. cumstellardustisnotfreelyexpandingbutconfined. 1992).Sincethatgasisdevoidofrefractories,thephotosphere In recent years it has become clear that also many willbealteredchemically,andthismayresultinveryFe-poor RVTauri photospheres show chemical anomalies pointing to stars which are rich in non-refractories like Zn and S. This an efficient depletion of refractory elements (Giridharetal. process is very efficient in the four named binaries where 1994; Giridhar&Ferro 1995; Giridharetal. 1998, 2000; there is observational evidence for the presence of a dusty Gonzalezetal. 1997b,a; VanWinckeletal. 1998; Maasetal. stablereservoir(VanWinckeletal.1995).Efficientseparation 2002).Themanyaffectedstarsshowthatthedepletionprocess of circumstellar gas and dust is not evident and Watersetal. isaverycommonphenomenoninevolvedstars.Giridharetal. (1992) argued that the most favourable circumstances may (2005)showthatdepletioninRVTauristarsislessstrongwhen occur if the circumstellar dust is indeed trapped in a stable disc.Notethatforonestronglydepletedobject,BD+39◦4926, the centralstar is cooler,which is interpretedas pointingto a stronger dilution due to a deeper convective envelope in the therewasnoIRexcessdetectedbyIRAS. coolerstars. Thedepletionpatternsarenotseeninobjectsfor In the few cases where ISO spectra are available,the dust which the intrinsic metallicity is smaller than aboutone tenth processingisstrong,whichresultsinalargecrystallinityfrac- solar([Fe/H]≤ −1.0). tionofthegrains(e.g.Molsteretal.2002).Alsothedustgrain sizedistributionisdifferentindiscsthaninoutflowsandwhere It is clear that the observational restriction that depletion long wavelength fluxes are available, they indicate the pres- isonlyactiveinbinarystars,whichwasformulatedwhenonly ence of large, µm sized and even cm sized dust grains (e.g. fourextremelydepletedobjectswereknown,hasbecomemuch Shentonetal.1995). less evident with the many new detections. Direct evidence Obviouslysuchadiscmustplayanimportantroleinthese for binarity,from radial velocity measurements,is difficultto systems and, although the indirect observationalevidence for obtain since RVTauristars have large pulsationalamplitudes. the presence of a stable disc is well established in the binary Moreover,theyoftenshowthepresenceofshocksintheline- stars,theactualstructureletalonetheformation,stabilityand forming region of the photosphere making the very determi- evolutionarenotwellunderstood.NotethatonlyforHD44179 nation of the radial velocity difficult (e.g. Gilletetal. 1990). isthisdiscspatiallyresolved.Forallotherobjects,thepresence Nonethelessorbitalelementshavebeendeterminedforquitea ofthediscwaspostulated. fewclassicalRVTauristars:UMon(Pollard&Cottrell1995), ACHer(VanWinckeletal.1998),ENTrA(VanWinckeletal. 1999), SXCen (Maasetal. 2002) and orbital motion is also 2.2. Classical RVTauristars from the General foundforIWCar(Pollardetal.1997),EPLyr(Gonzalezetal. CatalogueofVariableStarswith strong 1997b)andRUCen(Maasetal.2002). IRexcess Inthiscontributionweanalysethebroad-bandSEDsofthe RVTauri stars form a class of classical pulsating variables. RVTauristarsoftheGCVSwithareliableIRASmeasurement They are luminous (I-II) supergiant pulsating variables, the at60µm(e.g.Jura1986)andcomparethemwiththeSEDsof lightcurveofwhichshowsalternatingdeepandshallowmin- similarobjects,whicharenotinthePopIICepheidinstability ima with a formal period (measured between one deep mini- strip. mumandthenext)of30to150daysandabrightnessrangeof uptofourmagnitudes.ThespectraltypeistypicallyFtoGat minimum and G to K at maximum. There are two main pho- 2.3. New SampleofcandidateRVTauristars tometric varieties of RVTauri stars: the RVa types maintain a roughlyconstant mean brightness;RVb types show on top of LloydEvans (1985) and Raveendran (1989) showed that the theirpulsationalperiodalongerterm(600to1500days)peri- RVTauri stars are located in a well-defined and relatively odicity. thinly-populatedpartoftheIRAScolour-colourdiagram.The 6 S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? definingrectangleis: Genevaopticalphotometry WeacquiredGenevaopticalphotometryatrandomepochswith [12]−[25]=1.56+2.5log[F(25)/F(12)]=1.0−1.55 the 0.7m Swiss Telescope at La Silla and with the Flemish (cid:26)[25]−[60]=1.88+2.5log[F(60)/F(25)]=0.20−1.0 MercatorTelescopeatLaPalma,usingtherefurbishedGeneva (1) photometer P7 (Raskinetal. 2004). Our total data set was scanned for the maximum and minimum magnitudes (see LloydEvans(1997)searchedfornewexamplesofRVTauri Table C.1). Observation dates, the number of measurements stars, as stars with dusty discs, in the IRAS Catalogue, and andthetotaltimebasesofthesemaximaandminimaaregiven foundthestarsofthe‘newsample’discussedinthispaper.The aswell.AdditionalGenevaopticalphotometrywasfoundinthe presenceinthissampleofstarswiththeinfraredpropertiesof Geneva database: the General Catalogue of Photometric Data theGCVSsampleofRVTauristars,butwhichfelloutsidethe (GCPD,http://obswww.unige.ch/gcpd/gcpd.html). instability strip as judgedfromtheir spectral typesand which Notethesedataareonlygivenforcompleteness,butwerenot lackedthelarge-amplitudeRVTaurivariability,suggestedthat used in our analysis; except for IRAS10456 for which the RVTauri stars are those stars with dusty discs which are cur- GCPDdataareouronlyGenevafluxes. rentlylocatedwithintheinstabilitystrip(LloydEvans1999). Maasetal.(2005)presentedachemicalabundanceanalysis onthebasisofhighsignal-to-noiseandhighresolutionspectra UBVRI photometry for12starsofthenewlydefinedsample.Theyfoundthat9stars UBVRI photometry for the newly-identified objects was areaffectedbythedepletionprocess.Inadetailedstudyofone obtained from the South African Astronomical Observatory object, IRAS08544 (Maasetal. 2003); orbital elements were (SAAO) service observing programme(Table C.2). This uses found which show that this star must have undergone severe the 0.5m Telescope and the 0.75m Automatic Photometric binaryinteractionwhenitwasan(asymptotic)giant. Telescope(APT)atSutherland.IftheUBVRI dataofthestar Inthiscontributionweanalysethebroad-bandSEDsofall show large variations mainly due to a large variability of the 20newlycharacterizedstarsofthissample. star, we plotboth minima and maxima.If the numberof data pointsavailableistooloworiftherearenobigvariations,we useameanUBVRI. 2.4. TotalSample WealsosearchedforJohnsonandCousinsbroad-bandpho- tometryin the literatureandthese dataare givenin TableC.2 The 51 objects of our sample are presented in Table 1. aswell. In Col. 11 we differentiate between the several subtypes. Statisticalstudiesontheoccurrenceandcharacteristicsofpost- AGBstarswithrespecttoAGBstarsand/orPlanetaryNebulae Near-IR data (PNe)donotexistyet.Inthemostextensivecataloguedescrip- JHKL photometry of the newly-identified objects was ob- tion found in the literature, Szczerbaetal. (2001) consider a tainedbyT.LloydEvanswiththe0.75mreflectoroftheSouth Galactic sample of about 220 post-AGB stars known at that AfricanAstronomicalObservatory. time. Although the overlap between the specific sample dis- These data points were complemented with near-IR data cussedhere,andthetotalsampleofSzczerbaetal.(2001)isfar from the 2MASS and DENIS projects and with other fromcomplete,thesampleof51objectsdiscussedhere,forms asignificantpopulationofpost-AGBstarsknowntodate.Note JHKLM datapointsfromtheliterature(TableC.3). thatoursamplewasdefinedonthebasisofaspecificsetofcri- teriaexplainedaboveandoursampleisnotmeanttocoverall Far-IR data knownevolvedstarswithsimilarSEDcharacteristics. For the characterization of the longer wavelength part of the SEDs we use far-IR data from the IRAS Point Source Catalogue (Beichmanetal. 1988, Table C.4); for some 3. Broad-bandphotometricdata stars data from the MSX Infrared Point Source Catalogue (Eganetal. 2003) are also available (Table C.5). The IRAS In order to reconstructthe complete SEDs of our programme satellitehadfourpassbands,at12,25,60and100µm,respec- stars, we combined different sets of broad-band photomet- tively. The SPIRITIII instrument on board the MSX satellite ric data: our own Geneva optical photometry, supplemented had6passbands,at4.29,4.35,8.3,12.1,14.7and21.3µm. with UV, optical, near-IR, far-IR and submillimetre photom- etryfoundintheliterature. Submillimetredata The main difficulty in constructing the SEDs of pulsating stars with large amplitudes, like the RVTauri objects, is the For six stars (TWCam, RVTau, SUGem, UYCMa, UMon acquisitionofequallyphaseddataoverawidespectraldomain. and ACHer) submillimetre data (see Table C.6) are available Sincethesedataarenotavailable,welimitedourstudyofthe from DeRuyteretal. (2005), for HR4049 and IRAS20056 broad-bandenergetics to the phases of minimal and maximal weusedatafromtheotherliterature.ForHR4049,HD52961 coveredbrightness. and89Her,wehavesomenewlydeterminedSCUBA850mi- S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 7 croncontinuummeasurements.Theobservationswerecarried AnappropriatephotosphericKuruczmodel-basedonT , eff out with the 15m James Clerk Maxwell Telescope (JCMT) logg and [Fe/H]- is used for the unattenuated stellar photo- atMaunaKea,Hawaii,duringMarch-April1999(programme spheres(Kurucz1979). M99AN08). SCUBA (Hollandetal. 1999) was operated in photometrymodetosimultaneouslyobtaindataat850µmand 4.2. Colourexcess E(B −V) determination 450µm. For the data reduction, the standard software SURF wasusedwithMarsasafluxcalibrator.Table2detailsthenew Lightcomingfromastarisattenuatedandreddenedbymate- 850µmcontinuummeasurements,whilefor450µm,nosignif- rialintheline-of-sight.Notethatthetotalline-of-sightextinc- icant signal was detected. Note that the new 850µm SCUBA tionislikelytocontainbothacircumstellarandaninterstellar datapointforHR4049is,withintheerrors,thesameastheone component. inDominiketal.(2003). WeestimatethetotalcolourexcessE(B−V)byusingthe average interstellar extinction law given by Savage&Mathis (1979) to deredden the observed maximal UV-optical fluxes. Table2.Newobservationaldataat850µmfromSCUBA. E(B−V)isfoundbyminimizingthedifferencebetweenthe dereddenedobservedfluxesin the UV-optical,and the appro- No Name F850(mJy) priate Kurucz model (Kurucz 1979). We scale to the J filter 11 HD52961 2.8±1.9 whichisthereddestfilterwherenodustexcesscanbeexpected. 21 HR4049 8.7±2.8 Weassumethecircumstellarreddeninglawtobesimilartothe 38 89Her 40.9±2.4 ISMlaw.DeterminingtheE(B−V)inthiswayimpliesthat wedon’tcorrectforthecontributionofthegreyextinction. Results are in Table A.1. The erroron E(B −V) is typi- For ACHer we have a flux point at 1.1mm from cally0.1.Butchangingthestellarmodelsby±250Kineffec- Shentonetal.(1995). tive temperature causes a change in E(B −V) of about 0.2. Thus,togetherwiththeerrorof0.2inducedbytheuncertainty Other ofthe temperatureoftheunderlyingphotosphere,we havean uncertaintyof0.3onthetotalextinctionduringmaximallight. At the short wavelength side, we use the IUE data Adistributionofthetotalcolourexcessesfoundforoursample (InternationalUltravioletExplorer,0.115µm-0.320µm)from starsisshownin Fig.1. Notethatthetotalreddeningissmall the newly extractedspectral data release (INES Archive Data formostofthestars.Wemaythusassumethatasituationwith Server).Ifthereweremultiplespectraavailable,weonlycon- onlygreyextinctionwouldberatherexceptional. sideredthemaximalflux. 4. SpectralEnergyDistributions We analyse the SEDs of our sample stars in a homogeneous andsystematicway. 4.1. Thestars photosphere:modelparameters ThedeterminationofthemodelatmosphericalparametersT eff and logg, and the overall metallicity [Fe/H] is based on the analysisofhighresolutionspectrausedinthechemicalanaly- sesofthestars.Thosespectraarepreferentiallytakenatmax- imum light, because of the minimal molecular veiling during thehotterphaseinthelightcurve(Giridharetal.2000). For most stars we use model parameters deduced from our own spectra (e.g VanWinckel 1997; Maasetal. 2003, Fig.1. The distribution of the total E(B −V) found for our 2005), for others we use values found in the literature. For sample.Remarkthatmostcolourexcessesarelessthan1.0. someofoursamplestars(IRAS05208,UYCMa,IRAS10456, IRAS18123, IRAS18158, VVul), however, we don’t have a spectrum nor do we find any estimates for the photospheric parameters in the literature. Here we deduce the parame- 4.3. A first-order approximation:an opticallythin ters on the basis of the spectral type of the star. And for dust fit someothers(IRAS09400,IRAS10174,GKCar,IRAS13258, IRAS15556)welackopticalphotometry.Themodelparame- Abroad-bandSEDhaslimiteddiagnosticvalueforconstrain- tersaregiveninTable1,Cols.7to9.Thereferenceswherewe ing the chemistry and spatial distribution of the circumstellar foundtheparametersareshowninCol.10. dust. We apply an optically thin dust modelas a simple first- 8 S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? order approximation to fit the SEDs with the prime goal to and IRAS20056) are omitted in the figure. Like HD44179 quantifythedifferencesinSEDcharacteristicsbetweenthedif- (Cohenetal. 2004) and IRAS20056 (Menzies&Whitelock ferentobjects.We use theopticallythindustmodeldescribed 1988; Raoetal. 2002) we suspect that we only see the pho- bySopkaetal.(1985).Inthismodelthedustisassumedtobe tosphere of those sources throughscattered light. This means distributedsphericallysymmetricaroundthestarfrominnerra- thatweseethesesystemsnearlyedge-on. diusr to outer radiusr . Since we applyan opticallythin in out model, the geometry of the dust is not constrained as we ob- servethewholevolumeandalldustparticlescontributetothe observedflux. For moredetails of the modelwe refer to DeRuyteretal. (2005).ThefluxF isdeterminedbyfiveparameters:thenor- ν malizationtemperatureT ,theinnerradiusofthedustshellr , 0 in theouterradiusofthedustshellr ,thespectralindexpand out thedensityparameterm.Applyingaleastsquareminimization thefreeparametersr ,r ,pandmaredetermined.Results in out aregiveninTableA.1. 4.4. SEDs InFig 2 the SEDsof a selectionofpost-AGBstarsare given. The other SEDs are givenin Fig. A.1. The dereddenedfluxes areplottedtogetherwiththescaledphotosphericmodelrepre- senting the unattenuated stellar photosphere. We plotted both Fig.3.ThedistributionoftheenergyconversionratioLIR/L∗ minimal(opentriangles) and maximal(openrectangles)pho- foundforoursample. tometry. HD44179andARPuparetworatherexceptionalstars,for For those stars where we have a submillimetre data point whichweadoptedanotherstrategytodeterminetheSEDs.The (see Table C.6) the SED followsa Rayleigh-Jeansslope from dust excess for HD44179 and ARPup starts even at shorter the 60-100µm flux point redwards. Assuming that the dust wavelengths making it impossible to scale the model photo- emissivity at far-IR wavelengths follows a power law (Q ∼ ν spheretotheJ filter.WethereforescaletotheGenevaGband νp),thespectralindexpasdeterminedfromtheslopebetween datapointinourleastsquaresprocedure. the 60-100µmemission andthe 850µmflux pointisclose to TheIRexcessduetothepresenceofcircumstellardust,is zero.Thesmallspectralindicesintherangefrom0.0to0.5for inallcasesverysignificantwithBD+39◦4926anoticeableex- most objectsare consistentwith the presence of a component ception for which no IRexcess was found. A generalcharac- oflargegrains(radius&0.1mm)inthecircumstellarenviron- teristic is that the dust energy distribution peaks at very high mentofthestars.Forthosestarswherewelackasubmillimetre temperaturesandthatthereisnoevidenceforlargeamountsof data point, the spectral index of the long wavelength slope is cool(T ≤100K)dust:thepeakofthedustSEDsliesaround notwellconstrained. d 10µm and in some cases even bluer. In most cases, the dust Note that the 100µm IRAS fluxes for some stars are af- excessstartsnearthedustsublimationtemperature. fected by Galactic cirrus, so they present upper limits. In the To determine the amount of energy reprocessed by the figuresanarrowisdrawn.Thismakesitstillmoredifficulttobe dust grains in the circumstellar environment,we compute the sureabouttheslopefromthe60-100µmfluxpointredwards. energy ratio LIR/L∗. We first calculate the stellar flux L∗ From all SED characteristicsmentioned,we infer thatthe by numerically integrating the scaled Kurucz model between dustexcessin the SEDs isclearly differentfromwhatwe ex- 145nm and 850µm. This gives a good estimate of the unat- pecttoobserveinapost-AGBstarwheretheexcessrepresents tenuated stellar flux. The energy radiated at even shorter and the expandingand cooling relic of the strong AGB mass-loss longer wavelengths is only a negligible fraction (. 10−4) of episode(s). the total stellar flux so we evaluate our wavelength integral boundaries as adequate. Integrating over the IRexcess model 5. Distanceestimates described in Section 4.3 yields L . We note that the ratio IR LIR/L∗(TableA.1)ishighfor78%oftheobjects(largerthan The typical luminosity of a lower mass post-AGB star is ex- 20%).Assumingthepresenceof20%greyornon-selectiveex- pectedtobebetween1000and10000L⊙.Accurateluminosi- tinction and a mean E(B −V) of 0.5, reduces LIR/L∗ with tiesofindividualpost-AGBobjectsarestilllargelyunavailable 25%, which remains, nevertheless, still large. The absorption sincetheyaregenerallytoofartoobtainreliableparallaxes. andthermalre-radiationofthestellarradiationbythecircum- In the few cases where reliable Hipparcos parallaxes are stellardustisonaverageveryefficient.InFig.3thedistribution availablewehaveadirectprobeofthedistanceandhencelu- oftheenergyconversionratiosisshown.Notethatratioslarger minosity.ResultsaregiveninTableA.2.Notethelargeerrors than 1.0 (for HD44179, ARPup, IRAS17233, IRAS18158 ontheparallaxestranslateinalargeerrorboxforthedistance S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? 9 Fig.2.TheSEDsofaselectionofoursampleofpost-AGBstars:HD52961andHD213985areexamplesofconfirmedbinary post-AGBstars,RVTauandIWCararegenuineRVTauristarsandIRAS17233andIRAS18123areRVTaurilikeobjectsfrom thenewlistofT.LloydEvans.ThedereddenedfluxesaregiventogetherwiththescaledphotosphericKuruczmodelrepresenting the unattenuatedstellar photosphere(solid line). An optically thin dust modelwas used to fit the IRexcess (dottedline). Data foundintheliteraturetogetherwithour7bandGenevaphotometry(onlythemaxima)areplottedastriangles.Theminimaldata points(squares)werenotusedforthedeterminationofE(B−V).Theyareshowntogiveanindicationoftheamplitudeofthe pulsations.Whereavailable,crossesrepresentour850µmSCUBAdatapoint.Errorbarsonthe850µmSCUBAdatapointare plottedaswell,forsomeobjectsthoughsmallerthanthesymbols.Thearrowatthe100µmfluxpointsignifiesanupperlimit. and the luminositydeterminations.The parallaxof HD44179 seeonlyscatteredlight(Cohenetal.2004).Thesameisprob- cannotbeusedtoinferaluminositybecauseweknowthatwe ablytrueforARPup(seeSect.4.4). 10 S.DeRuyteretal.:Kepleriandiscsaroundpost-AGBstars:acommonphenomenon? Alcocketal. (1998) reported the discovery of RVTauri against the effectivetemperaturefor all stars and in all cases, stars in the Large Magellanic Cloud (LMC). In light- and thereisdustatorverynearsublimationtemperature.Withtyp- colour-curve behaviour, the classical RVTauri stars appeared icalluminosityestimates,thissublimationtemperatureedgeis to be a direct extension of the type II Cepheids to longer pe- atadistancesmallerthanabout10AUfromthecentralsource. riods.WeusetheP-LrelationofAlcocketal.(1998)-derived Moreover,theSEDsrevealthatthepresenceofdust,veryclose for the LMC RVTauri stars- to derive the luminosity for the totheobject,isirrespectiveoftheeffectivetemperatureofthe pulsatingobjectsinoursample.ForvariableswithP/2>12.6 centralstar.Notethatnoneoftheobjectsshowsevidencefora daysweuse present-daydustymass-loss.Wethereforeinferthatpartofthe dust must be gravitationally bound:any typical AGB outflow M =2.54(±0.48)−(3.91(±0.36))log(P/2), velocitywouldbringthedusttocoolerregionswithinyears. V σ =0.35. (2) Withinour sampleof post-AGBobjectsconsidered- con- firmedbinarypost-AGBstars,classicalRVTauristarsandnew Notethattheintrinsicscatterislarge,implyingverysignif- RVTauri-likeobjects-thereisawiderangeinthestrengthof icantuncertaintiesintheestimatedabsolutemagnitudesofthe the total IRexcess, but the shape of the IRexcess thus indi- objects.InEq.2weusethefundamentalperiodP/2,definedas cates that in all systems, the circumstellar shell is not freely the timebetweena deepand a shallowminimum.Theformal expandingbutstoredinthesystem.Wearguethereforethatthe periodP isshowninTableA.2.Weapplythisextensionofthe same inner geometryas in the resolvedsystem HD44179ap- period-luminosityrelationknownfortypeIICepheids,forthe plies to the whole sample: the objects seem to be surrounded RVTauri stars from the GCVS to derive an estimate for their byaKepleriandisc. luminosities.Results,togetherwiththebolometriccorrections From the dust modelling fit it is also clear that the outer BCV (Besselletal.1998)used,arelistedinTableA.2.Theer- radiiarenotverylargeeither.Objectssimilartotheenigmatic rorpropagationstartingfromtheerrorsintheP-Lrelationare extreme HR4049 are UYCMa, UMon and IRAS17233 for giveninthetableaswell.Theobjectsareindeedluminous,as whichthedusttemperaturegradientobservedisverysmall. canbeexpectedfrompost-AGBobjects,butthelargescatterin theP-Lrelationpreventsmoreaccurateluminosityestimates. Fortheotherobjects,wherewedon’thaveapulsationpe- riod determination,we take L=5000 ± 2000L⊙. Note that itisnotknownwhethertheP-LrelationoftheLMCstarsisdi- rectlyapplicabletoGalacticstars.Ingeneral,theluminosities foundare smaller than,or equalto, about5000L⊙ indicating thatthepopulationisofratherlowinitialmass. By comparing the integrated fluxes of the scaled Kurucz model with the luminosity deduced from the P-L relation or the default assumed value of L=5000L⊙, we calculate the distanceD tothestar. Thedetermineddistancesareshownin Table A.2. The propagated error of the poorly calibrated P-L relation, yields very uncertain luminosities and therefore also uncertaindistances. Remark that the luminosities determined by the P-L re- lation will not be appropriate to stars seen with the dusty Fig.4.Theinnerradiiareverysmall.Thedustmustbelocated disc nearly edge-on. Besides for the well-known example veryclosetothestar:thedustexcessstartsfornearlyallobjects HD44179, this is likely the case for ARPup, IRAS17233, nearthedustsublimationtemperature.AllSEDsshowa clear IRAS18158andIRAS20056forwhichthedistanceslistedin near-IRexcess. TableA.2areclearlyupperlimits. Other indirect indications for the presence of gravitionally 6. Discussion bounddustare: We presented the SEDs of all stars in our sample based on literature data supplemented with our own Geneva photome- - Despite the significant IRexcess, the total E(B −V) for tryand-whereavailable-continuummeasurementsat850µm moststarsisverysmall.Efficientredistributionofthestel- (Fig. 2 and Fig. A.1). Despite the differentcriteria for the se- larfluxisclearlyincontradictionwith thelackofline-of- lectionofthethreesubsamples,itisclearthatthebroad-band sight reddening.This implies a significant grey extinction SEDs display a high degree of uniformity.Between extremes and/oranon-sphericallysymmetricdustdistribution.Since likeHR4049(mostcompactSEDwithasingledusttempera- theobjectsallhavearathersmalltotalextinction,thelatter ture)andHD44179(largestLIR/L∗),theshapeoftheSEDsis ismoreplausible.AgoodindependentestimateoftheISM verysimilarforallstars. contributiontothetotalextinctionintheline-of-sightofall Oneofthemostremarkablefeaturesisthestartofthedust individualobjectsisstilllacking.Itmust,however,beare- excess.InFig.4weplottheinnerradiiofthedustcomponents wardingexperimenttomeasurethosesinceitwouldhelpin

See more

The list of books you might like

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