198 9ApJ. . .340. .518B 6 , 1 measurable nonthermalfluxes shouldalsohavebeendetectedfromtheOBstarsoflower luminosity,iftheypossessmagneticfields preliminary reportofthisresult wasgivenbyAbbott,Bieging,andChurchwell(1985). ThemodelofWhite(1985)predictsthat Lq, andwithdeclinationsnorth of—45°.Wefoundthatsome25%thesemostluminous starsexhibitnonthermalemission.A program starsformsadistance-limited sampledefinedasallOBstarswithin2.5kpcof theSun,withluminositygreaterthan10 measurements of88OBstarsselectedtoprovidecomplete coverageoftheOandearly-BstarregionH-Rdiagram.Onegoal indicate anonthermalmechanismfortheemission.Strom and Harris(1977)reportedaradiosourcewhichtheyidentifiedwiththe of thissurveywastosearchfor more“surprises,”i.e.,additionaltypesofradioemittersamong thehotstars.Nonewerefound. angular size~0.'004(Felli,Massi,andChurchwell1988). ThehighimpliedbrightnesstemperatureandlackofX-rayemission star OOriA(B0.5Veclipsingbinary)wasdetectedbyGaray, Moran,andReid(1985).Thissourceishighlyvariablehasan any) betweentheradioemissionandastellarwindisunclear. more characterizethesestars(Bohlenderetal1987),andthe closedfieldlinesmustextendouttoatleast15-20stellarradii,forming emission ofrelativelylowbrightnesstemperature,butlarge emittingvolume,whichaccountsfortheobservedradioflux. the magnetosphere(Drakeetal1987).Thesestarsarethus “surface”or“magnetospheric”nonthermalemitters.Therelation(if a magnetosphere.Thenonthermalemissionisthoughttobe opticallythicksynchrotronemissionfromenergeticparticlestrappedin population ofrelativisticelectronswhichareacceleratedbyFermiprocessesinthestrongshocksthatthoughtto permeatehot star winds.Inthismodel,onlyveryweak(fewgauss)surfacemagneticfieldsarenecessarytoproduceopticallythin synchrotron stars as“stellarwind”nonthermalsources.White(1985)hasproposedthatthisemissionresults from asmall radii. Theobservednonthermalradiationmustthereforebeemittedbyenergeticparticlesintheouterstellarwind.We refertothese peculiar, Bp-typestars(Drakeetal1987). O-type orWolf-Rayet(WR)stars(e.g.,Abbott,Bieging,andChurchwell1984;Abbottetal1986),thehighlymagnetic, helium known tobestrongsourcesofnonthermalemission.Thedetecteddatehaveeitherbeenvery luminous volume ofdense,ionizedgas(WrightandBarlow1975;PanagiaFelli1975).Quiteunexpectedly,someearly-type starsarenow star HD26676(B8Vn),butlittleelseisknownaboutthissource. The AstrophysicalJournal,340:518-536,1989May1 © 1989.TheAmericanAstronomicalSociety.Allrightsreserved.PrintedinU.S.A. A secondgoalwastodetermine thefrequencyandcharacterof“stellarwind”nonthermal sources.Asubsetof25ofour The detectionoftwootherearly-typestarsmayindicatethe existenceofothertypesradioemissionamongthehotstars.The Many typesofOBstarshaveneverbeenobservedwithhigh sensitivityatradiowavelengths.Wepresentherecontinuum The veryluminousOandWRstarsallhavewindssodensethattheyareopaquetofree-freeradiationoutatleast 100stellar The Bpstarspossessaratherweakstellarwindatbest(see, e.g.Barker1986).Orderedsurfacemagneticfieldsofakilogaussor Early-type starsareexpectedtobedetectableassourcesoffree-freeradioemission,becausetheirstellarwindsform alarge 1 51 19- 6 in theknownfree-freesourcesPCygni(B1la)andCygOB2No.12(B8la). free-free interpretationforthepreviouslydetectedstarÇPup(04f)andprovidefurtherevidencevariability Subject headings:starsearly-type—masslossradioradiationwinds (08 If),HD152408Ifp),aCam(09.5la),ÍSco(B1Ia+),and169454allofwhichare order of10“Myr“,avalueconsistentwithpreviousresults.Multifrequencyobservationsconfirmthe thought tobefree-freeradiosources.AlloftheseveryluminousOBstarsarelosingmassatarateonthe characterized byanegativespectralindex,luminositiesat5GHzontheorderof10ergssHz,mild free emissioninsix(24%)ofthecases.ThenonthermalsourcesarespatiallyunresolvedbyVLAand stars, suggestingthatthenonthermalmechanismislessefficientatlowerluminosities. variability, andnomeasurablepolarization.Nonthermalradiationwasnotdetectedfromthelessluminous L. Amongtheseveryluminousstars,strongnonthermalradioemissioniscommon,clearlyexceedingfree- coverage ofthisportiontheH-Rdiagram.Asubsetthesestarsformsadistance-limitedsampleall25 OB starswithin2.5kpcoftheSun,withdeclinationsnorth—45°,andluminositiesgreaterthan10 0 q © American Astronomical Society • Provided by the NASA Astrophysics Data System We derivemasslossratesfromtheradiofluxesofHD15570(04f),166734(07If+09I),151804 Radio continuumobservationsweremadeof88O-typeandearlyB-typestarsselectedtoprovidecomplete A SURVEYOFRADIOEMISSIONFROMGALACTICOBSTARS Joint InstituteforLaboratoryAstrophysics,UniversityofColorado Radio AstronomyLaboratory,UniversityofCalifornia,Berkeley Astronomy Department,UniversityofWisconsin-Madison Received 1988June13;acceptedOctober3 E. B.Churchwell David C.Abbott I. INTRODUCTION J. H.Bieging ABSTRACT 518 AND 198 9ApJ. . .340. .518B 1 6 6 comparable tothoseinferredforthehigh-luminositysources.Infact,wedetectedemissionfromonlythreeoflowerluminosity mass-loss ratesforthefollowingstarswhoseradioemissionisthoughttobefree-free:CSco,HD151804,152408,166734, OB starssurveyed,anditisunclearwhetheranyoftheseastellarwindnonthermalsource. and wederiveanewmass-lossrateforthisstarontheassumptionthatitisfree-freesource. and HD15570.ApreliminaryversionoftheserateswasgivenbyAbbott(1985).AmongthelessluminousstarswedetectedaCam contain atotalof90stars.Ofthese,onewasnotobservedfortechnicalreasons,andbadlyconfusedbynearbyHnregion. Table 2werechosentofillinfourcategoriesofspectraltypes:early-O,mid-O,late-O,andearly-B;threeluminosityclasses:V and Chiosi(1982),withsomeadditionsdescribedbyAbbott(1982a).ThestarsinTable1formthedistance-limitedsampleofallOB from theSun.Forbinaries,parametersrefertoprimary.starswithinrecognizedclusters,absolutevisualmagnitudes Thus, thereare88starsforwhichwehavereliableradiodata. distance, exceptwherelimitationsinthescheduledobservingtimepreventedusfromcloseststars.Tables1and2 or dwarf,II-IIIgiant,andIfsupergiant,giving12categoriesinall.Withinacategorythestarswereselectedaccordingto stars within2.5kpcoftheSun,northdeclination—45°,withbolometricstellarluminositiesgreaterthan10L.Thein distances camefromthespectraltypecalibrationofContietal(1983)forO-typestarsandLesh(1968) B-typestars. were derivedfromthedistancemoduligivenbyHumphreys(1978).Forfieldstars,absolutevisualmagnitudesand theimplied bolometric correctionsforthespectraltypesassignedbyConti(1975)O-typestarsandfromappropriate model and Böhm-Vitense(1981)fortheB-typestars.Bolometricluminositieswerecalculatedusingabsolutevisualmagnitude andthe Effective temperatureswereassignedonthebasisofspectraltypesandcalibrationsgivenbyConti(1975)for O-typestars beginning ofthisproject,sosomethestarsinTable1nowhaveadoptedstellarluminositiesthatareslightlybelow ourcut-off the evolutionarytracksofMaederandMeynet(1987).Notethatthesecalibrationsdiffersomewhatfromthoseadopted atthe atmosphere ofKurucz(1979)fortheB-typestars.Stellarmasseswereinterpolatedfromstar’slocationinH-R diagramand value of10L. the windhasbeenmeasuredfromUVspectroscopy,andthesevaluesarealsosummarizedintablesreferences givenby (1982). to escapevelocityfoundinotherstarsofsimilarspectralclassification(e.g.,Abbott1978;GarmanyandConti 1984).These Abbott (1982a).ForstarswithoutUYmeasurements,theterminalvelocitieshavebeenestimatedfromratioof velocity rotational broadeningofthephotosphericlines,takenfromContiandEbbets(1977),Hutchings(1981),Uesugi andFukuda estimates areindicatedbyparenthesesinTables1and2.Wealsolistvaluesofvsiniforallstarswithmeasurements from 0 o 23. 22. 21. 20. 25. 24. 10. 14. 13. 12. 11. 16. 15. 19. 18. 17. 4. 2. 9. 7. 6. 3. 8. 5. 1. A finalgoalwastoprovidefurtherhigh-qualityratesofmasslossforOBstars.Fromthedistance-limitedsurveywederivenew The stellarsamplesaregiveninTables1and2.dataforeachstarweretakenfromtheOBcatalogofGarmany,Conti, Also giveninTables1and2aretheadoptedstellarparametersofluminosity,eifectivetemperature,radius,mass,distance Every starinTables1and2isexpectedtopossessastrongstellarwind.Inmanycasesthemaximum,orterminal,velocity of No. © American Astronomical Society • Provided by the NASA Astrophysics Data System HD orBD + 41°3807 + 40°4220 66811 164794 15558 168076 15570 192281 168112 155913 14947 167971 166734 152236 169454 152408 151804 193237 1 9 Sgr Cyg OB2No.22AB Cyg OB2No.7 Cyg OB2No.8B £ Pup Cyg OB2No.9 Cyg OB2No.11 Cyg OB2No.8C Cyg OB2No.5 Cyg OB2No.8A Cyg OB2No.12 Ç Sco PCyg Other names + + 04 V 03 If 04 f 04 V 04 f 04 (f) 05 (f) 05 Vp 05 III 05 III 05 III 05 f 05 f 05 f 05 III 07.5f +O 07If +091 06f+07f 06 lb 08 Ifp 08 If B1 Ia B1 la B8 la B1 Ia Spectral A Distance-LimitedSampleofLuminousOBStars Type SURVEY OFRADIOEMISSION519 Cyg OB2 Gum Neb Cas OB6 Cyg OB2 Sgr OBI Ser OBI Cyg OB8 Cas OB6 Cyg OB2 Cyg OB2 Cyg OB2 Cyg OB2 Ser OB2 Cyg OB2 Cyg OB2 Per OBI Ser OB2 Cyg OB2 Cyg OBI Sco OBI Set OB3 Sco OBI Sco OBI OB Assn. a) TheStellarSample II. THESURVEY TABLE 1 Distance (kpc) 0.45 2.40 2.18 2.18 2.18 2.29 2.29 2.30 1.82 1.58 1.82 1.99 1.82 1.82 1.82 1.82 1.82 1.82 1.82 1.90 1.66 1.90 1.90 1.99 1.82 -10.12 (mag) -6.26 -5.82 -6.76 -6.80 -6.12 -7.31 -7.47 -6.19 -6.68 -6.40 -6.17 -6.49 -6.10 -7.02 -7.30 -6.60 -7.00 -7.62 -6.10 -8.32 -8.74 -8.02 -6.11 -6.42 M v log TMassRadius eff 4.72 4.68 4.70 4.70 4.67 4.67 4.67 4.62 4.70 4.60 4.65 4.65 4.65 4.67 4.67 4.67 4.54 4.55 4.55 4.60 4.05 4.31 4.31 4.31 4.54 (K) x (L) (M(Rq)(kms^ 6.19 6.40 6.07 6.44 6.46 6.19 6.06 6.06 6.22 6.61 6.09 6.29 6.18 6.09 5.94 0 6.32 6.43 6.30 6.07 6.24 6.10 5.86 6.21 5.95 5.99 133 138 134 103 118 160 98 93 96 61 98 91 78 78 85 45 78 81 66 50 83 71 73 57 56 338 103 25 23 20 24 28 22 30 34 74 31 38 34 14 16 85 16 16 19 18 18 17 19 19 (2900) (2500) (2700) (2700) (2700) (2500) (2700) (2700) (2200) (2200) (2700) 2700 2400 3400 3000 2000 2600 3500 3100 3200 1800 1400: 220 500 850 240 205 270 210 111 130 128 145 115 145 130 115 130 105 150 180 90 45 65 60 75 50 198 9ApJ. . .340. .518B 41. 26. 24. 23. 22. 21. 20. 49. 48. 47. 46. 45. 44. 43. 42. 40. 29. 28. 27. 25. 37. 35. 33. 31. 52. 50. 39. 38. 36. 34. 32. 30. 65. 64. 63. 62. 61. 60. 57. 55. 53. 51. 16. 14. 13. 12. 11. 10. 59. 58. 56. 54. 19. 18. 17. 15. 4. 9. 7., 6. 3. 2. 8. 5. 1. No. © American Astronomical Society • Provided by the NASA Astrophysics Data System + 66°1675 + 39°4168 + 66°1661 HD orBDOtherNames 47839 46150 46233 46573 217086 206267AB 229196 44743 207198 214680 24431 203064 206165 218376 24760 22951A 205196 25638 207538 228779 209975 36486 209481 36861 34656 26676 2905 24398 213087 5394 30614 38666 34078 37737 35921 153919 52089 159176 18326 165052 199579 19820 155806 108 163800 154368 149038 149757 193322 17603 195592 156292 194279 180968 184915 198781 144217 149438 143275 197345 187879 147165 € Per y Cas ß Seo 40 Per ¡i Col AE Aur À Orí 9 Cep € CMa EZ Cam T Seo Ô Seo a Cam Ô Orí ÇOph 68 Cyg a Cyg V380 Cyg 2 Vul ic Aql LY Aur k Cas c Per ßCUa (T Sco 26 Cep 15 Mon 1 Cas 19 Cep 10 Lac 14 Cep Nor 07 III 07 V 07 V 07 06.5 If 06.5 V 06 III 05 V 04 V 09.5 I 09.5 la 09.5 lab 09.5 II 09.5 V 09.5 III 09 1 09 V 09 V 09 V 09 Ve 09 IV 09 V 08.5 III 08 (f) 08 III 07.5 lile 07.5 IV 07 If 07 II 07 III 07 V+ 07 V 07 06.5 V+09 06 V A2 la B2 II B1 la B1 Ib Bl II Bl III Bl III Bl IV B0.5 Ib B0.5 V B0.5 V B0.5 V BO Ib BO III BO V 09.5 Ib 09.5 lab 09.5 la 09.5 V 09.5 III 09.5 III B8 Vn B2 Ib B1 la B0.5 III B0.5 III B0.5 III B0.5 IVe BO V BO V Spectral A SampleofOBStarsRepresentativeSpectralTypes Type Aur OBI NGC 6322 Aur OBI Cep OB2 Cam OBI Cam OBI Ori OBI Sgr OBI Cep OB3 NGC 6383 Cam OBI Cep OB4 Cyg OB9 Cep OB2 Sgr OBI Cyg OB7 Mon OB2 Cep OB2 Cep OB2 Cep OB2 Ori OBI Ori OBI Ori OBI Aur OBI Cep OB2 Cam OBI Cyg OB7 Cas OB5 Mon OB2 Mon OBI Mon OB2 Cyg OB7 Cep OB2 Seo OB2 Per OB2 Seo OB2 Seo OB2 Lac OBI Cyg OB9 Cas OB14 Per OB2 OB Assn. TABLE 2 520 Distance (kpc) 0.83 0.83 0.50 0.84 0.87 0.69 0.83 0.80 0.64 0.21 0.16 0.92 0.45 0.50 0.85 0.60 0.15 0.66 0.79 0.79 0.13 0.83 0.15 0.39 0.24 0.80 0.14 0.40 0.86 0.63 0.31 0.66 0.39 0.83 0.33 0.83 0.16 0.16 0.83 0.87 0.50 0.83 2.51 1.72 1.60 1.00 1.00 1.31 1.58 1.51 1.38 1.00 1.20 1.20 1.50 1.50 1.31 1.00 1.20 1.03 1.15 1.10 1.25 1.31 1.09 -5.24 -5.42 -5.44 -5.21 -5.20 -5.02 -4.77 -4.92 -4.72 -6.50 -4.90 -5.40 -4.69 -5.70 -5.57 -5.32 (mag) -6.19 -4.30 -3.59 -4.71 -5.90 -4.40 -4.40 -4.50 -5.50 -6.12 -4.86 -5.91 -4.96 -5.36 -4.35 -8.56 -0.3 -4.50 -7.16 -7.09 -6.14 -5.00 -4.20 -4.20 -3.50 -6.00 -4.90 -4.90 -4.90 -3.60 -3.81 -4.90 -4.15 -3.23 -4.18 -6.00 -5.54 -6.70 -6.00 -6.00 -6.70 -3.37 -5.05 -4.79 -4.49 -5.34 -5.47 -4.40 -4.37 M v log ^eff 4.54 4.55 4.56 4.57 4.59 4.59 4.59 4.58 4.60 4.62 4.67 4.70 4.23 4.46 4.44 4.41 4.48 4.50 4.52 4.52 4.52 4.52 4.54 4.54 4.54 4.56 4.57 4.56 4.56 4.59 4.59 4.59 4.59 4.59 4.60 4.62 4.08 4.35 4.31 4.31 4.31 4.42 4.42 4.43 4.36 4.46 4.46 4.46 4.44 4.44 4.49 4.48 4.48 4.50 4.50 4.50 4.50 4.48 4.52 4.52 4.52 4.54 4.54 4.42 3.98 (K) log L ho (L) 4.90 4.90 4.98 4.96 4.68 4.40 4.80 4.48 4.47 4.56 4.82 4.78 4.41 4.79 4.45 4.90 4.90 4.89 4.64 4.68 5.34 5.26 5.40 5.28 5.67 5.17 5.21 5.36 5.19 5.11 5.90 5.26 5.46 5.66 5.87 2.26 5.32 5.54 5.12 5.02 5.33 5.24 5.31 5.59 5.23 5.13 5.25 5.85 5.60 5.58 5.20 5.00 5.22 5.04 5.04 5.04 5.10 5.50 5.78 5.50 5.50 5.82 5.24 5.46 5.04 q (M) Mass 0 46 29 41 27 26 43 41 24 21 28 22 27 22 29 22 30 24 31 30 29 30 28 54 37 63 69 22 27 35 22 26 22 33 39 31 33 32 33 22 22 21 21 21 32 32 27 19 30 32 31 16 13 18 14 14 15 17 18 15 18 16 14 19 16 3 Radius (^o) 199 21 48 26 26 30 31 38 50 11 10 17 10 14 11 11 13 13 11 11 14 12 13 12 12 13 12 19 15 19 19 18 10 13 14 15 10 10 13 13 13 10 9 9 9 9 9 8 7 9 9 8 8 8 8 8 6 5 7 8 8 8 3 (km s^ (2400) (1800) (2500) (2500) (2500) (2600) (2800) (2800) (2800) (2600) (2800) (2700) (2900) (3000) (2800) (2100) (2000) (2200) (2300) (1300) (1400) (1700) (1600) (1300) (1400) (1400) (1700) (1700) (2100) (2100) (2100) (2100) (2200) (2200) (1100) (1500) (1400) (1200) (1400) (1700) (1700) (1300) (1900) 2850 2650 2300 2000 3290 2200 3100 2000 2280 2100 2300 2100 3400 3000 3200 1640 1650 1500 1890 Ko 900 270 1 (km s") v smi 295 270 260 260 350 264 295 225 305 341 155 180 115 110 145 105 175 119 115 100 106 100 170 140 172 118 140 130 180 170 175 155 45 40 45 97 27 94 25 70 20 80 63 35 55 70 30 75 50 85 85 80 32 53 55 55 15 198 9ApJ. . .340. .518B hms configuration, themeasuredrmsnoiselevelwassignificantlyhigherthanfordataobtainedinnormalconfigurations. Webelieve map withashiftedcenterwasmadetopermitcleaningtheregioncontainingstar.Thisprocedureusually allowedusto luminosity obtainedbyGarmanyandConti(1984);(2)the theoreticalMimpliedbythemodelofFriendandAbbott(1986);(3) interpretation oftheradioemission(discussedbelow)to: (1)theratederivedfromempiricalrelationbetweenMandstellar corresponding uncertainty.Column(9)liststhetypicalmonochromaticluminosityforstarat6cmwavelength thedistance level greaterthanorequalto3timesthermsnoiseforasource-freeregioninmap.Uncertaintiescolumns (5)-(7)are (1981, 1984).Allofthestarshaverelativelyaccurateopticalpositions.Toclaimaradiodetection,werequirethat therebea frequency andtendtobecurveddownward,i.e.,thespectral indexsteepensforfrequenciesabove5GHz.Althoughthesestarsshow White (1985)fornonthermalradioemission(discussedbelow). given inTable1.Incolumns(10)-(12),wegivethelogarithm oftheratiomass-lossrate(M)derivedfromafree-free at twoormorefrequenciesanddetectedonefrequencies,thespectralindex(orlimit)isgivenincolumn (8),withthe given asthermsnoiselevelnearstellarposition.Upperlimitsare3timesnoise.Whenastarwasobserved atoneepoch pointlike radiosourcecoincidentwiththeopticalpositiontowithincombinedanduncertainties, andata hyphenated entryindicatesahybridorreconfigurationarray).Themeasuredfluxdensitiesanduncertainties(orupper limits)at2,6, give thenameandspectraltypeofstar.Columns(3)(4)listdateobservationconfiguration oftheVLA(a about pointsources. since theinner(u,t;)-planecontainsinformationmainlyaboutlargestsizescalesinmap,andonlyverylittle information arrays. Theveryunevendistributionofthevisibilitydatareduceseffectivetotalintegrationtimefordetection pointsources, the excessnoiseisaneffectofconcentrationvisibilitydatanearoriginin(u,t;)-planewhichoccurs nonstandard of 1.5).Wefoundthat,whenobservinginahybridarrayorduringreconfiguration,theantennaswerenot in astandard achieve anrmsnoiselevelinthevicinityofstellarpositionwhichwasclosetotheoretical(typically within afactor sidelobes. Sometimes(especiallyinC-array)astrongsourcewasfoundnearorbeyondtheedgeofmap,which casealarger M derivedfromtheHaflux(Leitherer1988).Finally,column (13)givesthesurfacemagneticfieldofstarimpliedbymodel and 20cmaregivenincolumns(5),(6),(7).OurcriteriafordetectionswereasdescribedAbbott,Bieging, Churchwell emission ishighlyvariableforthisstar.Atoneepoch(1983 May9)theemissionwasdetectedat2cmand6(butnot20cm) component disappearsandreveals thefree-freeemissionfromastellarwind. density atthattimegaveamass-loss rateconsistentwithotherdeterminations(seeAbbott, Bieging,andChurchwell1984).Thusit exception isCygOB2No.9,whichhasamaximumat5 GHz andaratherflatspectrumatthefrequenciesobserved.Theradio which areclassifiedasdefinitenonthermalemitters.Except forCygOB2No.9,thespectraaresteeplydecliningwithincreasing appears thatmostofthetime theradioemissionfromthisstarisdominatedbynonthermal mechanism,butoccasionallythis an atypicallylowlevel,atwhich timethespectralindexwas+0.6,asexpectedforathermal stellarwindsource.Theradioflux some variabilityintheirradioemission,thegeneralshape ofthespectrumseemstobesimilarfromepochepoch.Anotable observed radioemissionisnonthermalthespectrum. InFigure1,weshowspectrafromthedatainTable3forfivestars and accordingtothedegreeofcertaintywithwhichwecan makethisclassification.Animportantcriterionforestablishingthatthe Although thisweightingproducesasynthesizedbeamwithlargersidelobesand,generally,wide,low-level(typically ~10%)skirt NRAO AIPSsoftwarepackage.Mapswerealwaysmadewithnaturalweightingofthevisibilitydatatomaximize sensitivity. cleaning withthesoftwarepackageinVLADEC-10computer.Afterthatdate,allmappingandfurtherprocessingwere stars. to thecentralresponsethanwouldoccurforuniformweighting,beamshapeisnotcriticalindetectingpointsources suchas (i.e., withintheexpectednoise),imageswerecombinedbeforecleaningandfurtheranalysis.Observationsatstandard (1950) =203150,decl.+40°58'54'.'0. 1984). InstrumentalcalibrationwasperfromedwiththestandardVLAsoftware.Priorto1981January,wedidmappingand “ 6cm”to4860MHz;and201490MHz. VLA frequencies.Forconvenience,inthispaperwewillrefertothedatabywavelengthband:“2cm”corresponds14940MHz; maps separatelyandcomparedforconsistencyinthenoiselevelanydetectedradioemission.Ifagreementwasacceptable Thereafter, thetotalbandwidthusedwas100MHz.WhenbothIFpairswereavailable,dataforeachpairtransformedto circular polarizationchannels.Priortotheintroductionof“BD”IFpair,totalbandwidthusedwas50MHzperchannel. various epochsoveraperiodofseveralyearsandinVLAconfigurations.Theobservationswerealwaysmadewithboth reported inanearlierpaper(Abbott,Bieging,andChurchwell1981),exceptforCygOB2No.22AB,whichthepositionisR.A. and heliumabundancefromtheseanalyses. Cam andôOri(Voelsetal1988).Forthesestarsweadopttheeffectivetemperature,bolometriccorrection,intrinsicB—Vcolor, FK4 catalogs.TheCygOB2associationmemberswerekindlymeasuredbyA.KlemolaoftheLickObservatoryandhavebeen 1 1 In thosecaseswhereastrongsourcewaspresentinthefield,mapswerecleanedwithAIPSsoftwareto remove the Our VLAobservationsofthedistance-limitedsample25OBstars(Table1)aresummarizedinTable3.Columns (1)and(2) TheNationalRadioAstronomyObservatory isoperatedbyAssociatedUniversities,Inc.,undercontractwith theNationalScienceFoundation. Table 3isdividedintosectionsaccordingtothenatureof theradioemissionmechanism,i.e.,nonthermalorthermal(free-free); Calibration andreductionofthedatawerealwaysdonewithsamebasicprocedures(seeAbbott,Bieging,Churchwell The stellarpositionsusedfortheradioobservationsweregenerallyfromSAOcatalog,orwhenavailable,AGK3and A detailedanalysisofthestellarphotosphericspectrumhasbeenmadeforstar(Puppis(Bohannanetal1986),aswella All oftheobservationspresentedhereweremadewithNRAOVeryLargeArray(VLA).Theat © American Astronomical Society • Provided by the NASA Astrophysics Data System SURVEY OFRADIOEMISSION521 b) TheObservations 198 9ApJ. . .340. .518B « J W í H C/3 S 2 g Ë N © American Astronomical Society • •5' o ^ Q 2S o p o Co Os OSOs. <<<<<<<<
S N ÍN ^ Tt UO CO ■P-H CO T-H <« o O o o o d d Pi X • +1 +1 ÖSS • +1 • +1 • +1 . +1 . +1 pt|F ¿HH o in A . +1 w CO »-ht^ o o O o 1 1 +l 1 S — :. +1 . +1 +1 : :. +i 5 »o : + ': Tt+ o •: ^ : ON+ o o d d V U U PQ PQ U U U < < PQ U <ï o ^ Tf H ^ uo ^ ^ ON cn cd C O o cx 'i' 00 c3 è Q è On rf rt T-H o ’ u + a!SS H o O O o o © American Astronomical Society • Provided by the NASA Astrophysics Data System ooPQ r- o Ö -h' of o; V V X S3 < o O >n -O h) T3 x .a ft o -o X) O X © American Astronomical Society • Provided by the NASA Astrophysics Data System 198 9ApJ. . .340. .518B (if any),andthespectraltype. Columns(5)-{8)inTable4Aareallupperlimits,listing turn the3alimiton:6cmfluxdensity; (4) ÔOriAalsoon1980December 5(Aarray)and1985June12(B-Chybridarray).Our resultsaresummarizedforall65starsin the monochromaticradioluminosity (forthedistancesinTable2);ratioofproduct vLforÀ6cm(v=4.86GHz),tothe definite detections(threestars). Columns(l)-(4)inTable4Agivethelinenumberfrom 2,theHDorBDnumber,othernames Tables 4Aand4B,whichgive, respectively,theupperlimitsfornondetections(62stars) andfluxdensitiesforthepossible stellar bolometricluminosity (also fromTable2);andthesurfacemagneticfieldderived fromthemodelofWhite(1985),discussed array); (2)HD108,14947,andkOrion1984March4 (B-Chybridarray);(3)aCamalsoon1985September18(Cand array), exceptforafewotherstarsobservedondates, namely:(1)COph,aCyg,(Per,and<5OriAon1980December5(A v The representativesampleoflowerluminositystarsinTable 2wassurveyedforemissionat6cmon1985June21(B-Chybrid Fig. 1—RadiospectraofthefivedefinitenonthermalemittersinTable3. Aninvertedtraingleconnectedbyadashedlinedenotesanupperlimit. Q • r-H X 09 a in a !>, © American Astronomical Society • Provided by the NASA Astrophysics Data System 1 SURVEY OFRADIOEMISSION 10 Frequency (GHz) 10 525 ooPQ TABLE 4A Upper Limits for the Undetected Lower Luminosity Stars 1 ^0 3 a Upper Limits log M(Mq yr“ ) a c a ^6 cm log L* B* No. HD or BD Other Names Spectral Type (mJy) (ergs s-1 Hz- log (vLJLj (gauss) f-f* F&Ae G&Cf (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) 1. 46223 04 V 0.24 17.82 -11.93 0.89 <-5.10 -5.46 -5.62 2. 46150 05 V 0.34 17.97 -11.77 0.94 <-4.97 -5.49 -5.60 3. 229196 06 III 0.34 17.80 -11.74 0.53 <-5.16 -5.80 -5.81 4. 199579 06 V 0.36 17.44 -11.70 0.53 <-5.40 -6.46 -6.43 5. 165052 06.5 V 0.27 17.91 -11.43 0.79 <-5.04 -6.12 -6.08 6. 206267AB 06.5 V 0.39 17.51 -11.63 0.77 <-5.35 -6.45 -6.40 7. 153919 06.5 1 0.60 18.33 -11.45 0.68 <-4.84 -5.41 -5.37 8. + 39°4168 07 0.42 17.87 -11.13 0.96 <-5.12 -6.70 -6.63 9. + 66° 1675 07 0.58 17.69 -11.38 0.73 <-5.24 -6.56 -6.50 10. 18326 07 V 0.27 17.51 -11.34 0.50 <-5.40 -5.92 -6.86 11. 47839 15 Mon 07 V 0.33 17.28 -11.73 1.38 <-5.65 -6.66 -6.60 12. 159176 07V + 07V 1.05 18.38 -10.86 1.27 <-4.71 -6.28 -6.22 13. 217086 07 V 0.60 17.74 -11.37 0.35 <-5.20 -6.50 -6.44 14. 46573 07 III 0.42 18.07 -11.02 1.94 <-4.96 -6.54 -6.47 15. 163800 07 III 0.21 17.80 -11.67 1.52 <-5.18 -5.92 -5.86 16. 34656 07 II 0.21 17.64 -11.41 1.11 <-5.32 -6.60 -6.50 17. 108 07 If 0.81 18.79 -10.77 1.49 <-4.32 -5.79 -5.71 18. 155806 07.5 IV 0.34 17.50 -11.67 0.85 <-5.37 -6.42 -6.34 19. 203064 68 Cyg 07.5 Hie 0.51 17.59 -11.70 0.34 <-5.24 -6.22 -6.14 20. 36861 À Ori 08 III 0.19 16.77 -12.38 0.81 <-5.95 -6.44 -6.35 21. 17603 08 (f) 0.33 17.60 -11.26 1.19 <-5.37 -6.91 -6.79 22. 19820 08.5 III 0.34 17.62 -11.60 0.48 <-5.34 -6.32 -6.22 23. 193322 09 V 0.42 17.35 -11.43 1.07 <-5.68 -7.03 -6.89 24. 24431 09 IV 0.36 17.64 -11.55 0.80 <-5.35 -6.37 -6.25 25. 149757 COph 09 V 0.30 15.93 -12.85 0.06 <-6.78 -7.03 -6.89 26. 209481 14 Cep 09 V 0.40 17.53 -11.59 0.57 <-5.43 -6.49 -6.36 27.. 214680 10 Lac 09 V 0.60 17.42 -11.35 3.00 <-5.58 -7.05 -6.91 28. + 66°1661 09 V 0.43 17.58 -11.20 0.80 <-5.43 -7.03 -6.89 29. 207198 09 1 0.33 17.44 -11.78 0.74 <-5.53 -6.34 -6.20 30.. 35921 LY Aur 09.5 III 0.19 17.61 -11.51 0.44 <-5.40 -6.49 -6.34 31.. 37737 09.5 III 0.24 17.70 -11.08 0.77 <-5.37 -7.04 -6.87 32.. 156292 09.5 III 0.43 17.92 -10.84 0.79 <-5.19 -6.85 -6.68 33.. 34078 AE Aur 09.5 V 0.36 17.04 -U.96 1.58 <-5.84 -6.68 -6.52 34.. 38666 fi Col 09.5 V 0.33 17.00 -11.33 1.31 <-5.87 -7.76 -7.58 37.. 149038 fi Nor 09.5 lab 0.24 17.40 -11.99 0.48 <-5.58 -6.06 -5.92 38.. 154368 09.5 lab 0.36 17.52 -11.87 0.40 <-5.48 -6.06 -5.92 39.. 195592 09.5 la 0.48 17.88 -11.78 0.53 <-5.20 -5.61 -5.49 40.. 209975 19 Cep 09.5 lb 0.34 17.46 -11.74 1.24 <-5.49 -6.36 -6.20 41.. 228779 09.5 I 0.27 17.55 -11.84 0.30 <-5.46 -6.06 -5.92 42.. 143275 ô Sco BOV 0.45 16.14 -12.54 0.13 <-6.61 -7.21 -7.00 43.. 149438 i Sco BOV 0.36 16.04 -12.26 1.38 <-6.61 -7.82 -7.59 44.. 207538 BOV 0.40 17.53 -11.13 1.98 <-5.56 -7.23 -7.01 45.. 25638 EZ Cam BO III 0.45 16.77 -12.22 0.13 <-6.01 -6.71 -6.53 46.. 205196 BO IB 0.39 17.51 -11.19 0.51 <-5.66 -7.16 -6.78 47.. 22951A 401 Per BO. 5 V 0.34 16.82 -11.63 1.06 <-6.19 -7.58 -7.26 48.. 144217 ß Sco BO. 5 V 0.45 16.14 -12.22 0.21 <-6.72 -7.72 -7.40 49.. 198781 BO. 5 V 0.52 17.45 -10.92 1.40 <-5.74 -7.71 -7.39 50.. 5394 y Cas B0.5 IVe 0.52 16.47 -12.22 0.11 <-6.38 -7.19 -6.94 51.. 24760 € Per BO.5 III 0.42 16.69 -12.24 0.02 <-5.82 -7.19 -6.56 52.. 184915 K Aql BO.5 III 0.43 17.34 -11.58 0.18 <-5.69 -6.81 -6.56 53.. 218376 1 Cas BO. 5 III 0.33 17.21 -11.71 1.00 <-5.79 -6.81 -6.56 54.. 213087 26 Cep B0.5 lb 0.36 17.51 -11.59 0.24 <-5.63 -6.51 -5.98 55.. 180968 2 Vul Bl IV 0.54 17.04 -11.25 0.54 <-6.08 -7.84 -7.49 56.. 147165 G Sco B1 III 0.30 15.90 -12.66 0.30 <-6.87 -7.39 -7.04 57.. 187879 V380 Cyg B1 III 0.48 17.57 -11.00 1.26 <-5.62 -7.39 -7.04 58.. 44743 ß CMa Bl II 0.45 16.50 -12.39 0.59 <-6.38 -6.87 -6.52 59.. 24398 C Per B1 lb 0.24 16.64 -12.45 0.15 <-6.25 -6.55 -5.88 60.. 2905 k Cas B1 la 0.33 17.68 -11.78 0.31 <-5.45 -5.93 -5.18 61.. 194279 B1 la 0.40 17.85 -11.64 0.26 <-5.44 -5.89 -5.13 62.. 52089 e CMa B2 II 0.41 16.07 -12.45 0.34 <-6.84 -7.45 -6.92 63.. 206165 9 Cep B2 lb 0.37 17.40 -11.43 0.58 <-5.85 -6.81 -6.35 65.. 197345 a Cyg A2 la 0.30 17.36 -11.99 0.09 <-6.45 -6.12 -5.40 a From col. (1) of Table 2. b For 26 cm (v = 4.86 GHz). c Stellar surface magnetic field (see text). d Mass-loss rate inferred from a free-free interpretation of the radio data. e Mass-loss rate from the theoretical model of Friend and Abbott 1986. f Mass-loss rate from the empirical relation of Garmany and Conti 1984. 526 © American Astronomical Society • Provided by the NASA Astrophysics Data System 198 9ApJ. . .340. .518B 16 ultraviolet observationsofthisstarwouldclearlybeinterest, giventheradioemission. to haveasignificantwind.Consequentlywedonottabulate valuesforF^,logM,orinTables2and4B.Furtheroptical projected distanceof1.4x10cm.Anonthermalemission mechanismislikely,sinceadwarfwithspectraltypeB8notexpected associated withthestar.Atadistanceof130pc,angularseparation7"betweentworadiopeakscorresponds toa ruled out.Theexcellentagreementbetweenthenorthernradiosourceandstellarpositionstronglyimpliesthat thissourceis about one-thirdthevaluetheyobtained,whichsuggeststhatthissourceisvariableoveraseveralyeartimescale. agreement, theirdataevidentlyrefermainlytothesoutherncomponent.Ourmeasured6cmfluxdensityfor sourceis given asupperlimits);thetheoreticalmodelofFriendandAbbott(1986);empiricalrelationshipderivedby Garmany and epochs. ForHD167971,thedegreeofpolarizationislessthan1.5%at20cm.Theotherstarswithstrongemission typically have where Iisthetotalintensity.The3alimitsondaregiveninTable5forfourofdefinitenonthermalsourcesobserved attwo Conti (1984).Table4Bcontainssimilarentriesforthedetectedstarsandincludesanadditionalcolumnwithdateof observation. below. Finally,columns(9)-(ll)givemass-lossratesderivedfrom:afree-freeinterpretationofthe6cmradioemission(theseare limits ondofafewpercent. noise levelsinthepolarizedmapsareidentical.Thedegreeofpolarization,d,isdefinedusualwayas parameters /,Q,U,andVforthesestars.Innocasedotheormapsshowemissionatstellarposition.general, therms about the1olevelaccordingtotheirstateduncertainties,from1977December28February6.From the positional stellar position)is0.48mJypercleanbeamarea.ThefluxdensitiesreportedbyStromandHarris(1977)at6cmwere variableat Harris andtheweakeronecoincidenttowithinerrorswithopticalpositionofstar.Thetworadiopeaksare separatedby There aretwowell-resolvedradiocomponents,thebrighteronecoincidentwithsourcepositionreported by Stromand As aB8Vnstaritdoesnotmeetourotherselectioncriteriaforthesurvey.InFigure2weshow6cmVLAmap ofthisobject. 7" onanearlynorth-southline.Thesouthernpeakis1.46mJypercleanbeamarea,whilethenorthern(coincident withthe ff 64...... 26676B8Vn1985Jun210.48±0.1315.99-10.17...-8.91-8.29 36 30614aCam09.511985Jun210.37±0.0717.73-11.920.47-5.27-5.64-5.49 35 36486<50riA09.5111980Dec50.37±0.1216.96-12.660.22-5.85-5.82-5.70 a1xbef g d c b a f e No. HDorBDNamesTypeDate(mJy)(ergssHz)log(vL/L(gauss)f-F&AG&C ô OriA.—ThislateOstarwasoriginallydetectedonlyat the3olevel,thoughdetectionappearedconvincing.Subsequent The proximityofthetworadiopeakssuggeststhattheyarephysicallyrelated,thoughcourseachancecoincidence cannotbe vstt None ofthedetectedstarsinTables3or4Bshowsevidenceforpolarizedradioemission.Wehavemademaps theStokes ObservationbyS.Drake(privatecommunication). Mass-lossrateinferredfromafree-freeinterpretationoftheradiodata. HD 26676.—ThisstarwasobservedbecauseStromandHarris(1977)reportedthedetectionofradioemissionat6cm and21cm. Stellarsurfacemagneticfield(seetext). For/16cm(v=4.86GHz). Fromcol.(1)ofTable2. Mass-lossratefromtheempiricalrelationofGarmanyandConti1984. Mass-lossratefromthetheoreticalmodelofFriendandAbbott1986. Cyg OB2No91984Apr 44.2±0.2<186.00.4<7 9Sgr 1984Nov 27...2.0±0.2<33 HD 1679711984Apr 47.0+0.2<1313.80.1<3.9 HD 168112...1984Apr 41.2+0.2<801.90.1<24 © American Astronomical Society • Provided by the NASA Astrophysics Data System Star Date(mJy)(%) (%)(mJy) 1984Nov278.0 ± 1.0<917.0+1.0<2.125.0+2.0 <1.5 1984 Nov276.3±0.2<10 8.1±0.1<46.20.6<5 g 3 ¿ 528 BIEGING, ABBOTT, AND CHURCHWELL Vol. 340 Fig. 2.-6 cm cleaned map of HD 26676. Contours are linearly spaced at 0.135 mJy per clean beam area, zero omitted. The cross marks the optical position and the filled square the 6 cm position reported by Strom and Harris (1977). Inset shows the 50% contour of the clean beam. observations by us on 1985 June 12 and by S. Drake (private communication) on 1986 July 7 failed to detect the star at lower flux density limits. If the earlier observation can be trusted, the star has apparently varied, in which case the emission may be nonthermal (but cf. the case of P Cyg—van den Oord et al. 1985). However, the inferred value of M for the detected flux density is in good agreement with values expected for the theoretical and empirical result (see Table 4B). Thus, free-free emission cannot be ruled out. HD 167971.—This remarkable star has the strongest nonthermal radio emission detected in our survey. It is believed to be a triple system comprising an G7.5f primary in a wide orbit about two less luminous main-sequence O stars which form an eclipsing pair (Leitherer et al. 1987). Observations of HD 167971 obtained in A array under excellent observing conditions on two dates (1984 Dec. 21 and 1985 Feb. 16) permitted an analysis of the visibility function at 2, 6, and 20 cm. The visibility data were vector-averaged in annuli in the (u, p)-plane, binned for equal numbers of data points in each annulus. Phases were zero within the uncertainties, which at least confirms the symmetry of the source. Results for the two epochs were essentially identical. Amplitudes are plotted in Figure 3 for the 1985 February 16 data. The emission appears to be unresolved at all three wavelengths. The strongest limit on the source size is from the 2 cm visibility data. A least-squares fit to a uniform circular disk model implies, at the 90% confidence level, a 14 radius less than 0?023 and a brightness temperature Tb > 21,000 K. At the stellar distance of 2 kpc, 0'.'023 corresponds to 7.0 x 10 cm = 5051?^ (for R* = 22 RQ see Table 1). Although the limit on Tb is not especially high, the emission is clearly not thermal (e.g., from a very extended corona), from the shape of the radio spectrum (Fig. 1). Note that the free-free photosphere, as defined by equation (11) of Wright and Barlow (1975), is at roughly 1.7 x 1014 cm for this star. Thus the nonthermal emission is highly concentrated at radii close to the star but is still not obscured by free-free absorption. The best radio position for HD 167971, from the A array maps at 2 cm, is R.A. (1950) = 18h15m17s635 + 0S00L Decl (1950) = — 12°15'46'.T5 ±0':02. HD 168112.—The radio emission is clearly nonthermal in nature from the shape of the radio spectrum (cf. Fig. 1). The averaged visibility function is shown in Figure 4, as for HD 167971 (discussed above). The emission is unresolved, within the uncertainties, at all wavelengths. At 2 cm, the average visibility may be biased by the noise (see White and Becker 1982), since the flux density is low. This effect may account for the apparent increase in average flux at the longest spacings at 2 cm. For a uniform disk model of radius 0 and brightness temperature Tft, the 6 cm visibility implies 6 < 0':08 and Tb > 17,000 K at the 90% confidence level. At 20 cm, the corresponding limits are 0 < 0"23 and Tb > 35,000 K. The best radio position for HD 168112, from the A array maps at 6 cm, is R.A. (1950) = 18h15m52s774 + 0S002 Decl (1950) = -12°07'38'.73 ± O'm • _ . , -v , Cyg OB2 No. 9.—Figure 5 shows the annular average visibility as for HD 168112. The source appears unresolved at all three wavelengths. The 2 cm data imply for a uniform disk model that 9 < 0".032 and Tb > 12,000 K. At 6 cm the visibility function sets limits for a uniform disk of 0 < 0'.'058 and T* > 41,000 K; at 20 cm, the limits are 0 < 0'.'27 and Tb > 13,000 K (all at the 90% confidence level). © American Astronomical Society • Provided by the NASA Astrophysics Data System ooPQ 529 ° No. 1, 1989 SURVEY OF RADIO EMISSION ^0 ft QJ "a e (0 öß cö 0) > Projected Baseline (A) Projected Baseline (A) Fig. 3 Fig. 4 Fig 3 —Visibility functions for HD 167971, at epoch 1985 Feb 16, showing annular vector averages of the fringe amplitude as a function of average projected baseline. Top: data for ¿2 cm with 1 a error bars; middle: for 26 cm; bottom: for 220 cm. For 6 cm and 20 cm, the 1 III. ANALYSIS a) Statistics of Nonthermal Emission The discovery of nonthermal emission from the O-type stars 9 Sgr (04 V) and Cyg OB2 No. 9 (05f) by Abbott, Bieging, and Churchwell (1984) greatly complicated the interpretation of radio radiation from early-type stars. Abbott (1985) has discussed our empirical criteria to discriminate free-free emission from nonthermal emission. Our primary indicators are : . . , i. Does the mass-loss rate implied by a free-free interpretation disagree strongly with the mass-loss rate of other observational diagnostics? In Table 3 we give the ratio of the mass-loss rate derived on the assumption that the observed fluxes are tree-tree © American Astronomical Society • Provided by the NASA Astrophysics Data System 198 9ApJ. . .340. .518B made withtheabilitytoresolve spatiallythefree-freeemission. free-free emission.Todate,however, theseprovisionalclassificationshavebeenverified whensubsequentobservationshavebeen foolproof. Forexample,White (1985)hasshownhownonthermalemissioncouldmimic the0.6spectralindexexpectedfrom any free-freestellarwindradio source(WhiteandBecker1982). show thepresenceorabsenceofvariability. Table 3wegivethespectralindexformultifrequencyradio observations. emission, comparedto:(1)themass-lossratefromempirical relationshipofGarmanyandConti(1984)(col.[10]);(2)the theoretical mass-lossratepredictedbythemodelofFriend andAbbott(1986)(col.[11]);(3)theHamass-lossrategivenby Leitherer (1988)(col.[12]).Alargevalueforanyofthesethree ratiosindicatesaprobablenonthermalradiosource. 530 6 We stressthatunlesstheflux densityofthesourceisgreatenoughtobespatiallyresolved bytheVLA,thesecriteriaarenot iv. Istheradioemissionresolved bytheVLA(seeFigs.3,4,and5)?Withasufficientsignal-to-noise ratio,theVLAcanresolve iii. Istheradioemissionstronglyvariable?Wereproduce previousradioobservationswehavemadeoftheseprogramstarsto ii. Doesthespectralindexofradiocontinuumobey v°powerlawexpectedfromfree-freeemission?Incolumn(8)of © American Astronomical Society • Provided by the NASA Astrophysics Data System c > (D CD e Oh 0) BIEGING, ABBOTT,ANDCHURCHWELL .002 .004 .006 .008 .002 .004 .006 .008 .002 .004 .006 .008 .01 0 0 0 56 5 5 0 5000010l.SxlO 0 102xl03xl04xl0SxlQ 0 5x10101.5x10 :í *iIïï: ÿ J ^^^^^^^^^1L_ Fig. 5.—Asfor4,CygOB2No.9 Projected Baseline(A) n—^^^—r~i—^^ ^ir 1 Vol. 340 198 9ApJ. . .340. .518B 6 1 5-1 _1 14 102 -1 companion inawideorbit.Such systemsmustbeveryrareinthegeneralstellarpopulation, however,andthestatisticspresented exceptionally largeluminosities, L>10. correlation betweennonthermal radioemissionandstrongerthanaverageX-rayluminosity reportedbyChlebowski(1988),andthe which aretypicalforluminousOstars.Thestellarwindterminal velocitiesmeasuredfromUVobservationsof9Sgr,HD167971, well-studied andisthoughttobesingle(Garmany,Conti, andMassey1980).HD168112exhibitsradialvelocityvariationsof either onthesamelineofsightorisinaverywideorbit withtheeclipsingpair(Leithereretal1987).Thestar9Sgrhasbeen and HD15558alsofollowtypicalrelationsforearly-typestars. Twodistinctivefeaturesofthenonthermalsourcesareapreliminary not strongevidenceforacompanion.CygOB2No.9and 8aaretoofainttohavebeenobservedforradialvelocityvariations. low-amplitude (Conti,Leep,andLorre1977);however,such variabilityischaracteristicofatmospheresluminousOstarsand date suggeststhatthestatisticsofTable3willnotchangedramatically iffurthermonitoringofthesourcesiscarriedout. flat orpositivespectrum.The0.6spectralindexfoundforCygOB2No.9on1983Maycorrespondedtoadramatic decreasein binarity. HD15558isasingle-line,longperiod,spectroscopic binarywithatypicalmassfunction(GarmanyandMassey1981).HD sources inTable3couldalsoexhibitnonthermalemission atothertimes.However,therarityofsuchbehaviorinobservationsto absent onthisoccasion,andtheobservedfluxwasunderlyingfree-freeemissionfromstellarwind.Someof “free-free” absolute fluxdensitylevels,whichledAbbott,Bieging,andChurchwell(1984)tohypothesizethatthenonthermal emissionwas nonthermal emissionisstillhigh(White1985). of thesymmetrysourcesonplaneskyandFaradaydepolarization,sincewinddensityin theregionof radio sourcecouldhelpclarifythenatureofnonthermalemissionmechanism. the sourceofnonthermalemissionthanisprovidedforbycurrentmodels.VLBIobservationsCygOB2No.9 toresolvethe outward. Theapparentcorrelationintheradiofluxdensitiesatthreefrequencies(Fig.6)seemstorequireamuchsmaller sizefor fields are“frozen”intothewind,sothistimescaleshouldalsobecharacteristicforvariationsinmagneticfield topropagate i is120kms“,whichtypicalforOstarsofthesespectral types.Thenonthermalsourcesalsoappearnormalintheirincidenceof equation (11)ofWrightandBarlow(1975)assumingamass-lossrate2x10“Myr,thefree-freephotosphere forCyg No. 1,1989SURVEYOFRADIOEMISSION531 s, thetimeforstellarwindtotravelfromfree-freephotosphereatonefrequencynextis~1yr.Note that magnetic OB2 No.9isroughly1.2x10cmat14.9GHz,2.45and5.01.5GHz.Atavelocity of3000km which coveraperiodof6yr.Thereiscleartendencyforthefluxesatthreefrequenciestovarytogether. emission. greatest variabilitytodate.WeshowinFigure6alloftheobservationsthisstarfromTable3andWhite Becker (1983), The timescaleandamplitudeofthevariabilityisprobablylimitedbylargevolumewhichpresumablyresponsible forthe intrinsic nonthermalluminositiesbutsomeofthesmallestratiosenergytobolometricradiativeenergy. vary. Thevariabilityistypicallyaboutafactorof2,whichmildcomparedtotheobservedinotherstellar radiosources. bolometric energyrangesfrom10“to.Incomparisonotherstellartypes(e.g.,Gibson1985),hotstarshave thelargest emission. Hz. IfweapproximatethetotalnonthermalemissionbyproductvLat5GHz,thenratioofenergyto total of24%.Clearly,allstarscouldemitnonthermalradiationatlevelstoosmalltobedetectedagainstthebackgroundfree-free of highestluminosity.Thenonthermalemissionisdefinitelydetectedinfive(20%)thestars,andprobableonemore,fora third stellarobject. the northeast(p.a.55°)whichhasnoopticalcounterpart.Theyarguedthatthissecondradiopeakprobablycorrespondstoyeta spherically symmetricmodel.Abbott,Bieging,andChurchwell(1981)notedthepresenceofasecondpeakradioemission0"9to wavelengths (e.g.,Abbott,Telesco,andWolff1984).CygOB2No.5isaclosebinarywhichthoughttobeinteractingVreux wind (vandenOordetal1985).ThestarCygOB2No.5isalsoverypeculiarinthattheradioemissionvariablewithanearlyflat explanations forthevariabilityincludevariablemass-lossepisodesfromstarand/orchangesinionizationbalance each hasbeenspatiallyresolvedwiththeVLA(BeckerandWhite1985),free-freeinterpretationisbeyonddoubt.Possible mass-loss rates.Themaininconsistencyisvariability.PCygniandCygOB2No.12arevariableinfluxspectralindex,yet 167971 isatriplesystemcomposedofaneclipsingpairmain-sequence, earlyO-typestars,andamoreluminousOfstarwhichis spectral index(Persietal.1985),yetthegenerallevelofradioemissionisconsistentwithmass-lossdiagnosticsatother internally consistentinthesensethatstarswhichhaveanegativeradiospectralindexarealsovariableanddiscrepantimplied bol0 0 1985), anditmaybethatvariablemassexchangeiscontributingsignificantfree-freeemissionnotaccountedforbytheusual v 19- No measurablepolarizationwasfoundforanyofthesources,asindicatedinTable5.Thelacknetmay bearesult Abbott, Bieging,andChurchwell (1984)suggestedthatthenonthermalemissionmight arisefromanundetectedcompact The nonthermalsourceshaveallbeenobservedatBalmer a.Leitherer(1988)derivesmass-lossratesfromtheseobservations None ofthenonthermalsourcesisspectroscopicallypeculiar. Allaremembersofwell-studiedOBassociations.TheirmeanVsin All ofthesourcesexhibitedanegative,ornonthermal,spectralindex.CygOB2No.9isonlysourcewhichhasever exhibiteda This tendencyisquitesurprising,becausetheradiiatwhichradioemissionoriginatingmustbedifferent. Again using Cyg OB2No.9isthenonthermalsourcewithgreatestnumberofobservations,andalsostarwhichhasexhibited the All ofthenonthermalsourcesarevariableontimescalesmonths.Bothabsolutefluxlevelandspectralindex appearto The nonthermalsourceshaveseveralcharacteristicsincommon.intrinsicluminosityat5GHzisontheorder10ergss From thestatisticsofourdistance-limitedsample,wefindthatstellarwindnonthermalemissioniscommonamongOBstars Using thesecriteriawehaveclassifiedallofthestarsinourdistance-limitedsampleasshownTable3.Thearegenerally © American Astronomical Society • Provided by the NASA Astrophysics Data System c) ImplicationsforModelsofthe NonthermalRadioEmission b) CharacteristicsoftheNonthermalEmission 198 9ApJ. . .340 . .518B 6 6 and particleenergyfalloffwith radius,theemergentluminosityofsynchrotronemission hasacharacteristicbehaviorinWhite’s nonthermal fluxesimplysurface magneticfieldstrengthsofafewgauss. emission. Ifthegeometryof the magneticfieldsisdescribedbymodelofWeber and Davis(1967),thentheobservedradio which arecarriedbythewind outtothelargeradiibeyondfree-freephotosphere,where theyproduceobservablesynchrotron star (LucyandWhite1980;Lucy 1982).Fermiaccelerationbythestrongshocksproduces asmallpopulationofrelativisticelectrons the factthatallstellarwindsofearly-typestarsarethought tobepermeatedbystrongshocksintheregionofaccelerationnear sources. within oursampleweexpectthatnotevenoneisanOBstar withacompactcompanion.Infact,wedetectedatleastfivenonthermaî Conti, andMassey(1980)weestimatethatsuchsystemscomprise lessthan1%ofallOBstarsmoreluminous10L.Thus, ratio of2ormore,timesthefractionthesesystemswhich arelongperiod(i.e.,haveawideorbit).UsingthestatisticsofGarmany, OB starsisthus0.07timesthefractionofmassivewhich arebinaries,timesthefractionofmassivestarswhichhaveamass because thestellarlifetimebecomesasymptoticallyconstant withincreasingmass.Theexpectedfractionofsuchsystemsamongall the fractioniscorrespondinglysmaller.Forstarswithaprimary massgreaterthan120Mthefractionisvirtuallyunchanged, For thecaseofa120Mprimaryand60secondary, thisfractionisroughly0.07.Forstarswithmoresimilarinitialmasses fraction ofthetimespentasanOBstarwithacompactcompanionistherefore comprising anormalOBstarandcompactcompanionmayoccurbecausethestellarlifetimedecreaseswithincreasing mass.The star duringitsheliumandheavierelementburningphase,possiblyasacompactcompanionbeyondthistime. Abinary are fromTable3.Errorbars1a;whennotshown,errorscomparabletothesizeofsymbol.Linesconnectingpointsonlyintended toguidetheeye. recognized asaspectroscopicallynormalOBstarduringitshydrogen-burningphase,luminousbluevariableor Wolf-Rayet star presentlymoreluminousthan10Lhadaninitialstellarmassgreaterroughly60M.Suchawill generallybe above virtuallyruleoutthisinterpretation.Forexample,accordingtotheevolutioncalculationsofMaederandMeynet (1987),a 0 0 0 0 61ens,tofOB2 No9 532 Because theradiusoffree-free photospherevarieswithwavelength,whiletheparticle numberdensity,magneticfieldstrength, White (1985)hasproposedadifferentmodeltoexplainnonthermal radioemissionfromearly-typestars.Hismodelisbasedon 9°' ~I“!^yCyg-asafunctionoftime,for2cm,6and20cmwavelengths.OpencirclesarefromWhiteBecker(1983); filledcircles © American Astronomical Society • Provided by the NASA Astrophysics Data System (duration ofH-burninginsecondary)—(lifetimeprimary) BIEGING, ABBOTT,ANDCHURCHWELL (duration ofH-burninginsecondary)’ ^ Vol. 340 198 9ApJ. . .340. .518B 12 -5_1 1 -1 predicted byevolutionarytracks, adiscrepancywhichindicatesthatthemeanmolecular weightmayalsobeunderestimatedforthe in column(8)ofTable6.Heliumenrichmenthasbeenverified spectroscopicallyfortheprogramstars(Pup(Bohannanetal.1986) other starsinTable6. and aCam(Voelsetal.1988); inbothcasestheobservedheliumnumberfractionis 0.17. Thisvalueissomewhatgreaterthan important quantity.Themeanmolecularweightforourprogram starswascalculatedfromthestar’slocationinH-Rdiagram and thesurfacehydrogenheliumabundancesofevolution calculationsofMaederandMeynet(1987).Thesevaluesaregiven the productsofnuclearburning.Forcalculation mass-lossratesbyequation(4)theabundanceofheliumisonly made inrelativelycompactarrayconfigurations,sothe derivedvaluesofMshouldnotsufferfromtheproblemsspatial resolution discussedbyWhiteandBecker(1981). where multifrequencyobservationsareavailable,thequoted mass-lossrateistheaveragevalue.TheseVLAobservationswereall where Tisthelocaltemperature(inK)atradio“photosphere.” ThederivedratesofmasslossaregiveninTable6.Forcases calculations ofDrew(1988), andwiththefactthatHenrecombinationlinesare not observedinthenebulaesurrounding We furtherassumethathelium issinglyionizedforallstarssothatZ=y1in cases. Thisisinaccordwithrecentmodel (Scheuer 1960) in GHz,yisthemeannumberofelectronsperion,Zrmschargeandgfree-freeGauntfactor takentobe where /zisthemeanionicweight(inamu),F^terminalvelocityinkms~\Ddistancetostarkpc,v thefrequency Felli 1975) wind implied byWhite’s(1985)modelareexceedinglylow,perhapsunrealisticallyso. luminosity. Wealsonotethat,althoughsurfacefieldstrengthshavenotgenerallybeenmeasurablefortheOBstars, thevalues larger magneticfieldstrengths,orthemodelofWhiteisdeficientinthatitpredictstoomuchsynchrotronemissionfor starsoflesser {f are afewtenthsofgauss.Wethusleftwiththeconclusionthateithermostmassiveandluminousstarshave systematically limit onthemagneticfieldstrengthimpliedbyupperradioluminosityandequation(2).Manyof upperlimits detected, withthepossibleexceptionof<5OriAdiscussedabove.ThisdiscrepancyisapparentinTable4A,whichgives theupper then manyofthelessluminousstarsshouldhavebeendetectedasnonthermalsourcesaccordingtoWhite’smodel. Yetnonewas difficult toaccountforinamodelofthetypeproposedbyWhite. with frequency.Asnotedabove,thetemporalcoherenceofvariabilityCygOB2No.9atdifferentfrequencies mayalsobe strongly negativespectralindex.Yet,accordingtothemodelofequation(2),spectrumluminosityiseitherflat orincreases observed spectralindex.WiththeexceptionofstarCygOB2No.9allidentifiednonthermalsourcesalways exhibita with measuredrotationalbroadening,andV=forstarswithoutvaluesofFsiniinTable1or 2.Thefield strengths ofthedetectednonthermalsourcesareinrange0.8to1.6gauss. a shockvelocityamplitudeof=(laV^)(whereisthesoundspeed)androtationalF1.43siniforstars derived byassumingamass-lossratefromtheempiricalcalibrationofGarmanyandConti(1984),strongshocklimitß=1.6, observed radioluminosity,equations(2)and(3)implyasurfacemagneticfieldstrength.InTables34wegivevaluesfor where Misthemass-lossjateinunitsof2x10yr,Vrotationalvelocityatstellarsurface, surface magneticfieldstrengthingauss.Thedimensionlessfunctiona(x)hasbeentabulatedbyTucker(1975). shocks inunitsof12.5kms~,Risthestellarradius35Twindtemperature40,000K,andB* amplitude oftheshock,F^isterminalstellarwindvelocityinunits2500kms,wseparationsuccessive where ß=1.6fortheexpectedlimitofstrongshocks,andconstantsLvaregivenby model, whichisdescribedbyhisequation(42): No. 1,1989 roi rot 0rot 0 t During thelatterpartoftheirevolution,mostluminous starsareexpectedtohavesurfacecompositionswhichenrichedby We assumethatthewindsof all programstarsarefullyionized.Thisassumptionmaybe questionableforthestaraCyg(A2la). The free-freeradiofluxdensity,S(inJy),isrelatedtothestellarmass-lossrate,M,by(WrightandBarlow1975;Panagia and Another discrepancyislessdirect.Ifallstarspossesssurfacemagneticfieldsofstrengthcomparabletothosethedetected stars, x There areseveralpossiblepointsofconflictbetweenourobservationsandWhite’smodel.Themoststrikingdiscrepancy isthe Our conclusionthatnonthermalstellarwindemissioniscommoninluminousearly-typestarssupportsWhite’smodel.Foran © American Astronomical Society • Provided by the NASA Astrophysics Data System Vi447168752HZ(Jb) (2w)-(F)(R)(r)'B,(F/L/[l+(V/UJr^’ œJrot0 20 L =2.5x10 t c) Free-FreeEmissionandMass-LossRates 32 g =-1.66+1.27log[T//(Zv)], (5) fiwind 26A/ß) SURVEY OFRADIOEMISSION 5 x10“ti*[a(l)/a(l+ß)r 6 3.01 xIO" 1/2 z(yg(i 9 (0317 {l +OJlT/Hn^/v,),v>V,’ 2 i(v/v)-^ ,V