Tesi di Dottorato di Ri er a in:
ASTRONOMIA E ASTROFISICA
XXVIII Ci lo
The hidden omplexity of stellar
population properties in
high-redshift early-type galaxies
Supervisore:
Dr. Mar ella Longhetti
Tesi di dottorato di:
Ilaria LONOCE
Matr. 720551
1 Introdu tion 3
1.1 Ellipti al galaxies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Formationand evolutionary s enarios. . . . . . . . . . . . . . . . . . 6
1.3 Observationalapproa h . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Metalli ity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6 This thesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2 Spe tral indi es 21 2.1 D4000 andH+K(CaII) . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2 Li k/IDS systemindi es . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3 UV indi es. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4 Measurementsand errors. . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4.1 MonteCarlo simulations . . . . . . . . . . . . . . . . . . . . . 36
3 Stellar populationsynthesis models 41 3.1 Simple StellarPopulation . . . . . . . . . . . . . . . . . . . . . . . . 41
3.2 Models forellipti al galaxies . . . . . . . . . . . . . . . . . . . . . . . 43
3.3 BC03 models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Maraston models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4.1 TMJmodels . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4 Spe tros opi data 49 4.1 FORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2 VIMOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3 X-Shooter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5 Eviden e of multiple stellar populations in z ∼ 1 ETGs 57 5.1 The sampleand thespe tros opi data . . . . . . . . . . . . . . . . . 58
5.1.1 Spe tros opi data . . . . . . . . . . . . . . . . . . . . . . . . 59
5.1.2 Dataredu tion . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2.1 Double- omponent analysis . . . . . . . . . . . . . . . . . . . 71
5.2.2 Theages ofthetwo stellar omponents. . . . . . . . . . . . . 75
5.3 Dis ussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.4 Con lusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6 Very high stellar metalli ity in a massive z = 1.4 ETG 89 6.1 Thetarget and thespe tros opi data . . . . . . . . . . . . . . . . . 90
6.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.3 Dis ussionand on lusions. . . . . . . . . . . . . . . . . . . . . . . . 99
7 Individual stellar metalli ity measure in z > 1.4 ETGs 101 7.1 Thesample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.1.1 Spe tros opi data . . . . . . . . . . . . . . . . . . . . . . . . 103
7.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.2.1 COSMOS-307881: UV up-turn . . . . . . . . . . . . . . . . . 122
7.3 Dis ussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
7.4 Summary and on lusions . . . . . . . . . . . . . . . . . . . . . . . . 131
8 Con lusions 133
Bibliograa 138
Introdu tion
Theprin ipalaimofthis PhDthesisistheinvestigationoftheformationand evolu-
tionofgalaxiesduringthe osmi epo hs. Thefollowedapproa histheobservational
analysisofthestellarpopulationproperties ofhighredshiftgalaxiesdire tlyintheir
early phases of evolution in order to at h a pi ture of their possible origin and of
thesubsequent pro esseswhi h have broughtthemtoward thelo alUniverse,where
we observe theirdes endants.
In parti ular this thesis is fo used on ellipti al galaxies whi h,as it will be pre-
sented in Se tion 1.1, are ideal andidates to probe the evolution of most of the
stellar massin thelo al Universe, and on their stellar properties: age (Se tion 1.4)
and metalli ity (Se tion 1.5). All the work is based on spe tros opi data oming
from the 8m ESO-VLT teles opes (Se tion 4) and on the potentiality of spe tral indi es(Se tion2) inthe omparisonwiththeupdatedandmostlyusedstellarpop-
ulationmodels (Se tion3).
1.1 Ellipti al galaxies
Ellipti alandlenti ular galaxies,i.e. Early-TypeGalaxies (ETG), onstitute awell-
dened lassofobje tsthankstowhi hwe antra eba ktheevolutionofmostofthe
stellarmassoftheUniverse. Inthisse tionIwill brieydes ribethe hara teristi s
ofthis family ofgalaxiesinthewell-studied lo alUniverse.
First of all, ETGs, for denition, share their regular ellipti al morphology. An
example of the regular morphology of ETGs is shown in Fig. 1.1 with the HST
imagesof thesample of 15 high redshift ETGs thatwill be analyzed in Chapter 5.
Manystudies inthelo alUniverse,asforexample those basedontheSloan Digital
SkySurvey(SDSS), tellus that ETGsin lude about the 80%of thebaryoni mass
belongingtostars,andthatmorethanthe80%ofthestellarmassen losedinETGs
belongs to themost massive galaxieswithM∗ > 3x1010M⊙ (Renzini2006 [96℄). As
it an be seen in Fig. 1.2, indeed, the total stellar mass ontribution of the ETGs
(toppanel)withrespe ttotheirnumber,ismu hmorethanthatofthespiraland/or
star-forminggalaxies ( alledLate-Type Galaxies)(bottom panel).
Figure1.1: HSTimagesof15highredshiftETGsanalyzedin Chapter5.
Figure1.2: Massdistributionsin thelo alUniverseofearly-typegalaxies(toppanel)and
late-typegalaxies(bottompanel). [96℄.
Furthermore, these galaxies onstitute an homogeneous lass in terms of olors
and stellar population ontent, displaying always red olors and evolved (i.e. old)
stellar populations. Generally their spe tra show only absorption lines (see for ex-
ampleFig. 5.1)sin e themost ofthe starformation isalready quen hed.
The homogeneity of ETGs is onrmed also by the numerous s aling relations
observed in the lo al Universe. In parti ular, these s aling relations involve both
their stru tural parameters, as the ee tive radius, and their olors and mass. In
parti ular, the Mass-age, Mass-metalli ity, and Mass-α/Fe will be s aling relations
involvedinthestudiespresentedwithinthisthesis,asitwillbedis ussedinthenext
se tions(inparti ularinSe tion1.5). Alltheserelationsdemonstratethepresen eof
a lear orrelation between themassand thestellar populationproperties implying
important onstraintson their formationand evolution pro esses.
The lass of ETGs and their study thus appear as the more powerful tool to
investigatetheformationandevolutionhistory ofgalaxiesintheUniverse. Wehave
seen that in the lo al Universe on one hand they are a lass of well re ognizable
obje ts, and on the other hand they in lude the most of the baryoni mass, even
en losedinthemost luminous andthus easierto beanalyzed systems.
Mostofthesefavourable hara teristi sareastrongpointalsointhemoredistant
Universe at least until z ∼ 1,even ifwith some dieren e and withsome more ob-
servativedi ulties. Forexampleintheearlier osmi epo hs,galaxieshostyounger
stellar populations, thus their olorsareslightly bluerand/or ouldbe obs ured by
dustduetotheveryre entstarformationevents. Moreover, existstheobservational
problem ofthe morphology (asdis ussed in Tamburri et al. 2014 [113 ℄), due to the
faintness of these distant obje ts whi h prevents a ertain morphologi al lassi a-
tion. Finally,we do not knowiftheobserved s alingrelation aretheprodu tof the
evolution ofETGs duringthe osmi epo hs,or of theformationme hanisms. This
latterpossibilitywouldimply thatthese relationsstill holdat any redshiftwiththe
same hara teristi s.
Forallthesereasons, exploringtheformationandevolutionofETGsbothinthe
lo al andin thedistant Universe is a powerful toolto shedlight on theissueof the
formation of the stru turesas they appearin theUniverse that we observe. In the
nextse tion Iwill introdu e themainproposeds enarios for thestarformationand
assemblyhistoriesofETGstogetherwiththeobservationaleviden essofar olle ted.
1.2 Formation and evolutionary s enarios
Sin ethe1960s,manytheoriesontheformationandevolutionofthisspe ial lassof
galaxies, theellipti als, have been formulated to explain theobservationaleviden e
thatETGsshowtheremarkableuniformityintheirphotometri and hemi alprop-
ertiesdis ussedinthepreviousse tion. Thesemodelsnotonlyhave tofullydes ribe
whatisobservedinthelo alanddistant galaxies, butalso tomat hthe onstraints
imposed bythestudieson theformationof darkmatterhalos stru tures,ina wider
osmologi al framework.
One oftherstmodels proposedto des ribe theformationof theETGs wasthe
so- alledmonolithi ollapse s enario. Inthis framework,ellipti als are assumedto
haveformedathighredshift(z > 5 − 7)asaresultofarapid ollapseofagas loud.
Thisgasisthenqui kly onvertedintostarsbymeansofaverystrongburstfollowed
byarapidquen hingofthestarformationa tivitydue tothegala ti windpowered
by supernovae explosions and stellar windswhi h arryout theresidualgas (Eggen
etal. 1962 [24 ℄, Larson1974 [54 ℄, van Albada1982 [126 ℄).
Thenumerousobservationalworkswhi hhaverevealedthepresen eofoldstellar
populations inETGs and there ent ndingof a populationof massive and passive
galaxies already in pla e at high redshift, when the Universe was only a few Gyr
old(Cimatti etal. 2004 [14 ℄,Sara o et al. 2005 [103 ℄), areinagreement withthis
s enario.
However,boththedete tionofresidualstarformationa tivityinETGsatz < 1
(Kaumann et al. 1993 [43 ℄, Trager et al. 2000 [120 ℄), and the problem that the
monolithi model annotbeeasilyin luded inaglobal osmologi al framework,have
lead to the formulation of a revised monolithi model (Kobayashi 2004 [46 ℄, Pipino
etal. 2010[91℄). Inthis news enario,nowpla edinthe osmologi al ontext ofthe
Cold DarkMatter (CDM)modelwhi hissu essfulinreprodu ingtheanisotropies
ofthe osmi mi rowaveba kground,themassiveETGswouldassemblemostoftheir
mass at z > 2 − 3 by means of merging events with small substru tures omoving in the same potential well. The following evolution is mainly hara terized by the
ageing of the stellar populations, with new small star formation events at z < 1,
onne ted forexample withmerging eventswithsmallersatellite galaxies.
An alternative s enario of galaxy formation is thehierar hi al model, based on
the hierar hi al lustering s enario for the formation of dark matter haloes in the
CDM osmologi al framework. This model suggests that massive galaxies are the
nal produ t of several gas-ri h merging ( alled wet) episodes of dis galaxies at
high redshift (z ∼ 4 − 5) whi h form most of the nal galaxy mass (Kaumann
et al. 1993 [43℄, De Lu ia et al. 2006 [23 ℄). Observational eviden es in favour of
this s enario arethe sharp in reasing of themerging rate with redshift(see Patton
et al. 1997 [86℄) and the observed relation between the olors and the stru tures
omplexityof galaxies,with olorstending to bluerwavelengths withthein reasing
of the omplexity. Finally, this s enario is well supported by numeri al simulations
of the merging intera tion among galaxies whi h su essfully reprodu e also the
kinemati and photometri properties ofmassive ellipti als (Naabetal. 2006 [79℄).
On the other hand, the number density of ellipti al galaxies does not de rease
within reasingredshiftatleastuntilz ∼ 2,withthefra tionofETGsbeing onstant
at 20 − 30% (Tamburri et al. 2014 [113 ℄, Im et al. 1996 [38 ℄), ontrary to what is
foreseenbythehierar hi al s enario.
A further model of galaxy formation, the so- alled dry merging view, predi ts
mergingeventswithoutgas( alleddry)fromwhi hveryluminousETGsareformed
bytheunionofquies entgalaxies,i.e. withoutongoingstarformation. Thiss enario
issupportedbytheeviden eofanalmost-nullevolutionoftheETGsluminosityfrom
z ∼ 1, while a strong de rease would be expe ted due to the ageing of stars. This
ee t ould be thus explained assuming an in reasing of mass not followed by new
starformation, i.e. dry merging(Bell etal. 2004 [4℄).
In the last years, a new s enario alled inside-out growth has be ome under
dis ussion. In this s enario, ETGs formed at high redshift (z ∼ 4 − 5) as single
ompa t spheroidsfromgas-ri hmergingevent. Thisinitialwetmergingeventmust
have happened at high redshift where the progenitors galaxies still have residual
gas,whileat lowerredshiftthe ompa tETGswillexperien e onse utivedryminor
mergingeventswhi hinvolvelow masssystems. Themainee t ofthedrymerging
events is the adding of an external envelope at low density to the ompa t ETG,
whi h is able to leave un hanged the stellar masswhile the total ee tive radius is
enlarged. This model has been formulated be ause of the observation of ellipti al
galaxiesat 1 < z < 2.5withanee tiveradius afa tor∼ 2 − 5smallerwithrespe t
tothemeanradiusoflo alETGswiththesame stellarmass(Daddietal. 2005[20 ℄,
Longhettietal. 2007 [62 ℄, Cimattietal. 2008 [17 ℄, Sara oet al. 2009 [105 ℄). This
me hanism should expand the galaxy dimension, but not the total mass, until the
values oftheee tiveradius mat h thoseoftheETGswiththesame massobserved
inthelo alUniverse (Naabetal. 2009 [80℄, Bezansonetal. 2009 [5℄).
A tually,thiss enarioshowssomelimitationsasforexampletheunlikelynumber
ofminor drymergingeventsne essaryto expandtheETGs andthela kof ame h-
anism able to justify the presen e of a number of non- ompa t ETGs observed at
highredshiftequivalenttothatofthe ompa t ones(Sara o etal. 2010,2011[106 ℄,
[107 ℄). Theselast results suggestthat thepro esses thatdrivethemassassembly of
ETGsareprobablynot universalwithin thewhole lassofgalaxiesandwhi h ould
be thefa torsthatdetermine dierent formationhistories remainan open issue.
In on lusion, theorigin and theassembly history of theETGs still remains an
open issue up to now, and the way to dis riminate among the dierent proposed
s enarios is oered by the studies of high redshift galaxies. For example, the wet
merging s enario predi ts that ETGs are formed by omposite stellar populations
withtheinner populationyoungerthan theoutskirts. Onthe ontrary, drymerging
events, mixingthestellarpopulationsoftheprogenitors shoulddilutetheage radial
variation, forming a rather at stellar population distribution. In the inside-out
growth s enario, instead,thegas-ri h mergingeventswhi h should have formed the
ompa t ETGs, suggests a stellar population highly oeval in the enter, to whi h
there will be added and redistributed in the outer regions the ontributions of the
following drymerging events,preserving theabsen e ofstrong gradients.
This pi ture of formationmodels of ETGsthus highlights thepossible presen e
ofmultiple stellar populationswithin this lassofobje ts but withsome dieren es
dependingonthewaymasshasbeena reted. Uptonow,studiesonETGsbasedon
photometri analysishaveallowed toidentify themalsointhedistantUniverse,but
only re ently theavailable spe tros opi datafor someof these distant obje ts with
highsignalto noise ratioshasopened a newpossibilityofstudying these omposite
populations at high-redshift asitwill be detailedintheworkof Chapter5.
1.3 Observational approa h
Intheprevious se tionthestilldebateds enarios ofETGsformation have been pre-
sented. All the possible tests aimed at verifying whi h is the most realisti model
to des ribetheformationandevolutionofETGsmustpassthroughthe omparison
withtheobservations. From theobservationalpoint of view,twodierent but om-
plementary approa hes existfor thestudying of galaxy formation. In this se tion I
presentthesetwomostlyadoptedapproa hestogetherwiththeirmainobservational
results obtained sofarin thelo aland highredshiftUniverse.
The so- alled ar haeology approa h (Thomas et al. 2005, 2010 [116 ℄, [117 ℄)
studiesindetailsthepropertiesoflo al galaxies,tryingto tra eba ktheirpaststar
formation history. The studieson lo al ETGshave theadvantageof dealing witha
hugequantityofdata(e.g. ompletesamples)withhighsignaltonoiseratioandwith
highspe traland spatialresolution. Inparti ular,thisapproa hhasbeen pioneered
by the analysis of slopes and s atter of the main ETGs s aling relations (Renzini
2006[96 ℄)andfollowedbydetailedstudiesofabsorption-lineindi es(Longhettietal.
2000 [60 ℄,Trageretal. 2000 [120 ℄,Thomasetal. 2005,2010 [116 ℄,[117 ℄). Themain
hallengeinthear haeologyofstellarpopulationsofETGsisthedisentanglement of
ageandmetalli ity,whoseee tsaredegenerateinold(> 5Gyr)stellarpopulations.
Moreover, inthe lo al Universe thestellar properties of oldETGs ould have been