Hot Stars: Old-Fashioned or Trendy?

38 Pages·1.8 MB·

by  A. W. A. Pauldrach

Hot Stars:
Old-Fashioned or Trendy?
A.W.A.Pauldrach
Institutfu¨rAstronomieundAstrophysikderUniversita¨tMu¨nchen
Scheinerstraße1,81679Mu¨nchen,Germany
[email protected],
http://www.usm.uni-muenchen.de/people/adi/adi.html
3
0
0
Abstract
2
SpectroscopicanalyseswiththeintentionoftheinterpretationoftheUV-spectra
n
a ofthebrighteststarsasindividuals–supernovae –orascomponents ofstar-
J formingregions–massiveOstars–provideapowerfultoolwithgreatastro-
6 physical potential for the determination of extragalactic distances and of the
1 chemicalcompositionofstar-forminggalaxiesevenathighredshifts.
Theperspectivesofalreadyinitiatedworkwiththenewgenerationoftools

2 forquantitativeUV-spectroscopyofHotStarsthathavebeendevelopedduring
v the last two decades are presented and the status of the continuing effort to
6 constructcorrespondingmodelsforHotStaratmospheresisreviewed.
2
Since thephysicsof theatmospheres ofHot Starsare stronglyaffectedby
2
velocityexpansion dominating thespectraat all wavelength ranges, hydrody-
2
namicmodelatmospheresforO-typestarsandexplosionmodelsforSupernovae
1
ofTypeIaarenecessaryasbasisforthesynthesisandanalysisofthespectra.
2
0 Itisshownthatstellarparameters,abundances,andstellarwindpropertiescan
/ bedetermined by themethodsof spectral diagnostics already developed. Ad-
h
ditionally,itwillbedemonstratedthatmodelsandsyntheticspectraofTypeIa
p
Supernovae ofrequiredqualityarealreadyavailable. Thesewillmakeitpos-
-
o sibletotacklethequestion ofwhetherSupernovae Iaarestandard candlesin
r acosmologicalsense, confirmingordisprovingthatthecurrentSN-luminosity
t
s distancesindicateacceleratedexpansionoftheuniverse.
a Indetailwediscussapplicationsofthediagnostictechniquesbyexampleof
:
v twoofthemostluminousOsupergiantsintheGalaxyandastandardSupernova
i ofTypeIa.Furthermore,itisdemonstratedthatthespectralenergydistributions
X
providedbystate-of-the-artmodelsofmassiveOstarsleadtoconsiderablybet-
r teragreementwithobservationsifusedfortheanalysisofHIIregions.Thus,an
a
excellentwayofdeterminingextragalacticabundancesandpopulationhistories
isoffered.
Moreover,theimportanceofHotStarsinabroadastrophysicalcontextwill
bediscussed. Astheydominatethephysical conditionsoftheirlocalenviron-
mentsandthelifecycleofgasanddustoftheirhostgalaxies,specialemphasis
will be given to the corresponding diagnostic perspectives. Beyond that, the
relevanceofHotStarstocosmologicalissueswillbeconsidered.
2 A.W.A.Pauldrach
1 Introduction
Itis wellknownthatHotStars are nota single groupof objectsbutcomprisesub-
groups of objects in different parts of the HR diagram and at different evolution-
arystages. Themostimportantsub-groupsaremassive O/Bstars, CentralStarsof
PlanetaryNebulae,andSupernovaeofTypeIaandII.Allthesesub-groupshavein
commonthattheyarecharacterizedbyhighradiationenergydensitiesandexpand-
ingatmospheres,andduetothisthestate oftheoutermostpartsofthese objectsis
characterizedbynon-equilibriumthermodynamics.Inordertocoverthebest-known
fundamentalstagesoftheevolutionofHotStarsinsufficientdepththisreviewwill
berestrictedtothediscussionofOstarsandSupernovaeofTypeIa;wewillnotdis-
cussobjectslikeWolf-Rayetstars, LuminousBlueVariables,Be-stars, Supernovae
of Type II, and others. Furthermore, Hot Stars play an important role in a broad
astrophysical context. This implies that a complete review covering all aspects in
theoryandobservationisnotonlyimpossible,butalsobeyondthescopeofthisre-
view.Thus,thisreviewwillfocusonaspecialpartoftheoveralltopicwiththeintent
toconcentrateonjustonesubject;thesubjectchosenisUVspectraldiagnostics. It
willbeshown,however,thatthissubjecthasimportantimplicationsforastrophysical
topicswhicharepresentlyregardedasbeing“trendy”.
ButfirstofallwehavetoclarifywhatUVspectraldiagnosticsmeans.Thisisbest
illustratedbythereallyold-fashioned(1977,MortonandUnderhill)UVspectrumof
one of the brightest massive O stars, the O4I(f) supergiant ζ Puppis. As can be
seeninFigure1,expandingatmosphereshaveapronouncedeffectontheemergent
spectraofhotstars–especiallyintheUV-part.Thesignaturesofoutflowareclearly
recognizedbytheblue-shiftedabsorptionandred-shiftedemissionintheformofthe
well-knownP Cygniprofiles. Itis quite obviousthatthese kindof spectra contain
informationnotonlyaboutstellarandwindparameters,butalsoaboutabundances.
Thus, in principle,all fundamentalparametersofa hotstar can bededucedfroma
comparisonofobservedandsyntheticspectra.
AlthoughspectraofthequalityshowninFigure1havebeenavailableformore
than25years,mostoftheworkdonetodatehasconcentratedonqualitativeresults
andarguments. Inviewoftheeffortputintothedevelopmentofmodern–state-of-
the-art– instruments, it is certainly not sufficient to restrict the analyses to simple
lineidentificationsandqualitativeestimatesofthephysicalproperties. Theprimary
objective must be to extract the complete physical stellar information from these
spectra. Such diagnosticissues principallyhave beenmade possible by the superb
qualityandspectralresolutionofthespectraavailable.
For this objective, the key is to produce realistic synthetic UV spectra for Hot
Stars. Such powerful tools, however, are still in development and not yet widely
used.Theyoffertheopportunitytodeterminethestellarparameters,theabundances
ofthelightelements–He,C,N,O,Si–andoftheheavyelementslikeFeandNi,
quantitatively,asisindicatednotonlybyIUE,butalsomorerecentHST,ORFEUS,
and FUSE observations of hot stars in the Galaxy and Local Group galaxies. All
theseobservationsshowthatthespectraintheUVspectralrangearedominatedby
adenseforestofslightlywind-affectedpseudo-photosphericmetalabsorptionlines
HotStars:Old-FashionedorTrendy? 3
Figure1: MergedspectrumofCopernicusandIUEUVhigh-resolutionobservationsofthe
O4I(f)supergiantζPuppis(900–1500A˚:MortonandUnderhill1977;1500–1800A˚:Walborn
etal.1985). Themostimportantwindlinesofthelightelementsareidentifiedandmarked.
Alsomarkedarethelargenumberofwind-contaminatedlinesoftheirongroupelements(e.g.,
FeV)whichareespeciallypresentbetween1250and1500A˚.(FigurefromPauldrachetal.
1994b).
overlaid by broad P Cygni line profiles of strong, mostly resonance lines, formed
in different parts of the expanding atmosphere. Thus, the obvious objective is to
investigatetheimportanceoftheselineswithrespecttothestructureoftheexpanding
atmospheresthat are characterized by the strength and the velocity of the outflow,
andthroughwhichtheshapeofthespectrallinesismainlydetermined.Overthelast
30yearsitturnedoutthattheachievementofthisobjective,whichwillalsobethe
primaryaimofthispapertoreview,remainsadifficulttask.
Before we discuss in detail the status quo of the diagnostictool required(Sec-
tion 6), we will first examine whether such a tool has been made obsolete by the
generaldevelopmentin astrophysics or whether it is still relevant to currentastro-
nomicalresearch.
For this purpose we first discuss the diagnostic perspectives of galaxies with
pronouncedcurrentstar formation. Duetothe impactofmassivestarsontheiren-
vironmentthe physics underlying the spectral appearance of starburst galaxies are
rooted in the atmospheric expansion of massive O stars which dominate the UV
wavelength range in star-forming galaxies. Therefore, the UV-spectral features of
massiveOstarscanbeusedastracersofageandchemicalcompositionofstarburst
galaxiesevenathighredshift(Section2).
4 A.W.A.Pauldrach
Withrespecttothepresentcosmologicalquestionofthereionizationoftheuni-
verse–whichappearedtohavehappenedataredshiftofaboutz ∼6–theionization
efficiencyofatop-heavyInitialMassFunctionforthefirstgenerationsofstarsisdis-
cussed(Section3).
Startingfromtheimpactofmassivestarsontheirenvironmentitisdemonstrated
thatthespectralenergydistributionsprovidedbystate-of-the-artmodelsofmassive
O stars lead to considerable improvementsif used for the analysis of HII regions.
Thus,thecorrespondingmethodsfordeterminingextragalacticabundancesandpop-
ulationhistoriesarepromising(Section4).
Regardingdiagnosticissues,theroleofSupernovaeofTypeIaasdistanceindi-
cators is discussed. The contextof this discussion concernsthe currentand rather
surprisingsurprisingresultthatdistantSNe Ia appearfainterthanstandardcandles
inanemptyFriedmannmodeloftheuniverse(Section5).
Finally,wediscussapplicationsofthediagnostictechniquesbyexampleoftwo
ofthemostluminousOsupergiantsintheGalaxy;additionally,basicstepstowards
realisticsyntheticspectraforSupernovaeofTypeIaarepresented(Section7).
2 The impact of massivestars on their Environment –
UV Spectral Analysis of Starburst Galaxies
Theimpactofmassivestarsontheirenvironmentinthepresentphaseoftheuniverse
isofmajorimportancefortheevolutionofmostgalaxies.Althoughrarebynumber,
massive stars dominate the life cycle of gas and dust in star forming regions and
areresponsibleforthechemicalenrichmentoftheISM,whichinturnhasasignifi-
cantimpactonthechemicalevolutionofthehostgalaxy. Thisismainlyduetothe
shortlifetimesofmassivestars,whichfavourstherecyclingofheavyelementsinan
extremely efficient way. Furthermore, the large amountof momentum and energy
input of these objects into the ISM controls the dynamical evolution of the ISM.
This takes place in an extreme way, because massive stars mostly group in young
clusters,producingvoidregionsaroundthemselvesandwind-andsupernova-blown
superbubbles around the clusters. These superbubbles are ideal places for further
star formation, as numerous Hubble Space Telescope images show. Investigation
of these superbubbleswill finally yield the required information to understand the
variousprocessesleadingtocontinuousstarformationregions(cf.OeyandMassey
1995).Thecreationofsuperbubblesisalsoresponsibleforthephenomenonofgalac-
ticenergeticoutflowsobservedinstarbursts(Kunthetal.1998)andstarburstgalaxies
evenathighredshift(Pettinietal.1998). Itisthusnotsurprisingthatspectroscopic
studiesofgalaxieswithpronouncedcurrentstarformationrevealthespecificspec-
tralsignaturesofmassivestars,demonstratinginthiswaythattheunderlyingphysics
forthespectralappearanceofstarburstgalaxiesisnotonlyrootedintheatmospheric
expansionofmassiveOstars,butalsodominatedbytheseobjects(cf.Figure2from
Steidel etal. 1996, for star-forming galaxies at high redshift see also Pettini etal.
2000, and for UV line spectra of local star-forming galaxiessee Conti etal. 1996;
notethatthesimilarityofthespectraatnone/lowandhighredshiftssuggestsasimi-
larstellarcontent).
HotStars:Old-FashionedorTrendy? 5
Figure 2: UV spectrum of a z > 3 galaxy (upper part). For comparison, a recent HST
spectrumof thecentral starburst regionintheWolf-Rayetgalaxy NGC4214 isalsoshown
(lowerpart).NotethecharacteristicP-Cygnilines,especiallyofCIVandSiIV,pointingtothe
dominatinginfluenceofmassiveOstars.FigurefromSteideletal.1996.
ThecharacteristicP-Cygnilinesobservedasbroadstellarwindlines,especially
thoseoftheresonancelinesofCIVandSiIV,integratedoverthestellarpopulations
in the spectra of starbursting galaxies, allow quantitative spectroscopic studies of
themostluminousstellar objectsindistantgalaxiesevenathighredshift. Thus,in
principle, we are able to obtain important quantitative information about the host
galaxiesoftheseobjects,butdiagnosticissuesofthesespectrarequireamongother
things synthetic UV spectra of O-type stars as input for the population synthesis
calculationsneededforacomparisonwiththeobservedintegratedspectra.
The potentialof these spectra forastrophysicaldiagnosticscan neverthelessbe
investigatedin a first step by usingobserved UV spectra of nearbyO-typestars as
inputforthecorrespondingpopulationsynthesiscalculationsinstead.Intheframeof
thismethodstarsaresimulatedtoformaccordingtoaspecifiedstar-formationhis-
toryandinitialmassfunctionandthenfollowpredefinedtracksintheHR-diagram.
The integrated spectra are then built up from a library of observed UV spectra of
hotstarsintheGalaxyandtheMagellanicClouds. Theoutputofthisprocedureare
semi-empiricalUVspectrabetween1000and1800A˚ at0.1to0.7A˚ resolutionfor
populationsofarbitraryage,star-formationhistories,andinitialmassfunction. The
computationaltechniqueof thismethodisdescribedcomprehensivelyinthe litera-
ture and we refer the reader to one of the latest papers of a series (Leitherer etal.
2001).
Asanexampleoftheanalysesperformedinthiswayacomparisonoftheaverage
spectrumof8clustersinNGC5253tosyntheticmodelsatsolar(top)and1/4solar
(bottom)metallicityisshowninFigure3(Leithereretal. 2001). TheFigureshows
clearly thata representativevalue of the overallmetallicity of this starburstgalaxy
can be determined, since the modelspectrum in the lower part fits the observation
6 A.W.A.Pauldrach
Figure 3: Average spectrum of 8 clusters in NGC 5253 compared to that of a population
modelatsolar(top)and1/4solarmetallicity(bottom). Itisclearlyshownthatarepresenta-
tivevalueoftheoverallmetallicityofthisstarburstgalaxycanbedetermined. Figurefrom
Leithereretal.,2001.
almostperfectly. Thisresultisespeciallyconvincingasbothmodels(dashedlines)
are based on the same parameters, except for the metallicity. A standard Salpeter
IMF between 1 and 100 M⊙ was used and the starburst has been assumed to last
6Myr,withstarsformingcontinuouslyduringthistime. Theworsefittotheobser-
vationsproducedby the solar metallicity modelspectrum– particularlydiscrepant
aretheblueabsorptionwingsinSiIVandCIVwhicharetoostronginthemodels–
couldbeimprovedbyreducingthenumberofthemostmassivestarswithasteeper
IMF,butatthecostofthefitqualityintheemissioncomponents. Asanimportant
result the ratio of the absorption to the emission strengths is therefore a sensitive
indicatorforthemetallicity.
Moreover, the strong sensitivity of the emission parts of the P-Cygni lines of
NV, SiIV, and CIV on the evolutionary stage of the O stars makes these spectra
quitesuitable asage tracers, whichis shownin Figure4, wherethe time evolution
of the integrated spectrum following an instantaneous starburst is presented. The
line profiles gradually strengthen from a main-sequence-dominated population at
0–1 Myr to a population with luminous O supergiants at 3–5 Myr; and the lines
weakenagainatlaterageduetoterminationofthesupergiantphase–thisbehavior
isespeciallyvisibleintheshapeoftheCIVλ1550resonanceline.
Theconclusionfromtheseexamplesisthatthiskindofanalysisisverypromis-
ing,butreliesonobservedUVspectra,whicharejustavailableforasmallnumber
ofmetallicityvalues,namelythoseoftheGalaxyandtheMagellanicClouds. Inor-
dertomakeprogressinthedirectionoutlinedbefore,realisticsyntheticUVSpectra
ofO-type stars are needed. Thisis in particularthe case for high-redshiftgalaxies
(whichareobservablespectroscopicallywhenthefluxisamplifiedbygravitational
lensingthroughforegroundgalaxyclusters)sinceinthesecasestheexpectedmetal-
licitiesofstarburstinggalaxiesintheearlyuniverse(cf.Pettinietal.2000)aremost
probablydifferentfromlocalones.
HotStars:Old-FashionedorTrendy? 7
Figure 4: Timeevolutionoftheintegratedspectrumforthe10Myrfollowinganinstanta-
neousstarburst.ThestrongsensitivityoftheemissionpartsoftheP-CygnilinesofNV,SiIV,
and CIV on the evolutionary stage of theO starsmakes these spectra quite suitable asage
tracers.FigurefromLeithereretal.2001.
3 First generations of Stars – Ionization Efficiency
of a Top-Heavy Initial Mass Function
Apartfromtheshortevolutionarytimescaleofmassivestarsitisobviouslythemetal-
licity, and, connected to this, the steepness of the Initial Mass Function, which is
responsiblefortherarityoftheseobjectsinthepresentphaseoftheuniverse. Ithas
tobethemetallicity,becauseveryrecentlystrongevidencehasbeenfoundthatthe
primordialIMFhasfavoredmassivestarswithmasses> 102M⊙ (cf.Brommetal.
1999). Thus, in the early universe, when only primordial elements were left over
fromthe Big Bang, natureobviouslypreferredto formmassive stars. Thispredic-
tionisbasedonthemissingmetallicitywhichleadstoacharacteristicscaleforthe
densityandtemperatureoftheprimordialgas,whichinturnleadstoacharacteristic
JeansmassofMJ ∼103M⊙ (Larson1998).
Finally,duetothesephysicalconditionstheInitialMassFunctionbecomestop-
heavyandthereforedeviatessignificantlyfromthestandardSalpeterpower-law(see,
for instance, Bromm etal. 1999, 2001). Such an early populationof very massive
starsatverylowmetallicities(i.e.,PopulationIIIstars)whichhavealreadybeenthe-
oreticallyinvestigatedbyElEidetal.(1983),recentlyturnedouttobealsorelevant
tocosmologicalissues,themostimportantbeingthecosmologicalquestionofwhen
8 A.W.A.Pauldrach
Figure 5: The normalized spectral energy distribution in the continuum of Population III
stars–massrange100–1000M⊙atZ =0.Notethatthespectraattainanuniversalformfor
stellarmassesM >300M⊙FigurefromBrommetal.2001.
and how the cosmic “dark ages” ended (Loeb 1998). We know that the dark ages
endedbecausetheabsenceofaGunn-Petersontrough(GunnandPeterson1965)in
thespectraofhigh-redshiftquasarsimpliesthattheuniversewasreionizedagainat
aredshiftofz > 5.8(Fanetal.2000). Foralongtime,Carretal.(1984)suspected
that the first generationsof stars have been relevantto control this process. Thus,
Population III stars could contribute significantly to the ionization history of the
intergalacticmedium (IGM), but the contributionof the first generationof stars to
the ionization history of the IGM depends crucially on their initial mass function.
Withregardtothe firstgenerationsofstarstheionizationefficiencyofa top-heavy
InitialMassFunctionwill,therefore,havetobeinvestigated.
ItistheenormousamountofUVandEUVradiationoftheseverymassivestars
which could easily change the status of the cold and dark universe at that time to
becomereionizedagain.ThisisindicatedinFigure5,whichalsoshowsthatthetotal
spectral luminosity dependssolely on the total amountof mass, if the mass of the
mostmassivestarsexceeds300M⊙ (cf.Brommetal.2001). Duetothistop-heavy
InitialMassFunctionthetotalspectralenergydistributiondeviatessignificantlyfrom
thatobtainedwiththestandardSalpeterpower-law,asisshowninFigure6,andthe
fluxobtainedcan contributethe decisivepartto the unexplaineddeficitof ionizing
photonsrequiredforthereionizationoftheuniverse(cf.Brommetal.2001).
However,inordertobeabletomakequantitativepredictionsabouttheinfluence
ofthis extremelymetalpoorpopulationof verymassivestars ontheir galacticand
intergalacticenvironmentoneprimarilyneedsobservationsthatcanbecomparedto
thepredictedfluxspectrum.
HotStars:Old-FashionedorTrendy? 9
Figure 6: PredictedfluxfromaPopulationIIIstarclusteratz = 10. Aflatuniversewith
Ω = 0.7isassumed. Thecutoffbelowλ = 1216A˚ isduetocompleteGunn-Peterson
Λ obs
absorption, and the observable flux is larger by an order of magnitude for the case of the
top-heavyIMFwhencomparedtothecaseofthestandardSalpeterpower-law. Figurefrom
Brommetal.2001.
Future observations with the Next Generation Space Telescope of distant stel-
lar populationsat high redshifts will principally give us the opportunityto deduce
the primordialIMF from these comparisons. Kudritzki(2002)has recently shown
thatthisisgenerallypossible,bycalculatingstate-of-the-artUVspectraformassive
O stars in a metallicity range of 1...10−4 Z⊙. From an inspection of his spec-
tra he concluded that significant line features are still detectable even at very low
metallicities;thus,therewillbediagnosticinformationavailabletodeducephysical
propertiesfromstarburstingregionsathighredshiftsthatwilleventuallybeobserved
withtheNextGenerationSpaceTelescopeintheinfraredspectralregion.
AsasecondrequirementweneedtodeterminethephysicalpropertiesofPopu-
lationIIIstarsduringtheirevolution. Akeyissueinthisregardistoobtainrealistic
spectralenergy distributionscalculatedfor metallicitiesdifferentfromzerofor the
most massive objects, since the assumption of a metallicity of Z = 0 is certainly
onlycorrectfortheveryfirstgenerationofPopulationIIIstars.
4 Theoretical Ionizing Fluxes of O Stars
Although less spectacular, we will now investigate the impact of massive stars on
theirenvironmentinamoredirectmanner. Apartfromthechemicalenrichmentof
the ISM, the large amount of momentum and energy input into the ambient inter-
stellarmediumoftheseobjectsisprimarilyofimportance. Especiallytheradiative
energy input shortward of the Lyman edge, which ionizes and heats the Gaseous
Nebulae surroundingmassive Hot Stars, offers the possibility to analyze the influ-
ence of the EUV radiationof the photoionizingstars on the ionizationstructure of
theseexcitedHIIregions.
10 A.W.A.Pauldrach
Figure7: [ArIII]8.99µm/[ArII]6.99µmversus[NeIII]15.6µm/[NeII]12.8µmdiag-
nosticdiagramofobservedandpredictednebularexcitationofHIIregions.Boldfacenumbers
indicatemodels whicharedesignated bytheireffectivetemperaturein103K.Trianglesare
NLTEmodelsandsquaresareLTEmodels;notethattheNLTEmodelsrepresentanimpres-
siveimprovementinreproducingthe[NeIII]emission.Arepresentativeerrorbarforthedata
isshowninthelowerrightcorner.FigurefromGiveonetal.2002.
The primary objective of such investigations are studies of theoretical models
of starburst regions, which for instance are used to determine the energy source
in ultra-luminousinfrared galaxies – ULIRGs – (cf. Lutz etal. 1996; Genzel etal.
1998).Theinterpretationofthecorrespondingextra-galacticobservationsobviously
requiresunderstandingthepropertiesofthespectralenergydistributions(SEDs)of
massivestarsandstellarclusters. Thus,thequalityoftheSEDshastobeprobedin
afirststepbymeansofinvestigationsofGalacticHIIregions.
4.1 The ExcitationofGalacticHIIRegions
Giveonetal.(2002)recentlypresentedacomparisonofobserved[NeIII]15.6µm/
[NeII]12.8µmand[ArIII]8.99µm/[ArII]6.99µmexcitationratios,obtainedfor
asampleof112GalacticHIIregionsand37nearbyextragalacticHIIregionsinthe
LMC, SMC, and M33 observed with ISO-SWS, with the correspondingresults of
theoreticalnebularmodels.
The authors have chosen infrared fine-structure emission lines for their inves-
tigation because these lines do not suffer much from dust extinction and the low
energies of the associated levels make these lines quite insensitive to the nebular
electrontemperature. Moreover,therelativestrengthsofthefine-structureemission
lines chosen are ideal for constraining the shape of the theoretical ionizing fluxes,