stripping, it enhanes the eet and due to the nally high eetive radii, the dark
matterfrationalsogrowsbynearlyafatorof2after10generationsofminormergers.
•
Violent relaxation does not eet the overall dierential energy distributions of the host galaxy.•
Duetodynamialfritionandtidalstrippingthestellarpartilesdevelopapromi-nent bumpat low bindingenergies.
•
The veloity dispersionofthe bulgeonly(one-omponent)models donot hange their shape,keep theirinitialHernquist proleand stay isotropioverthe wholeradial range.
•
For two-omponent aretions the nal oalesene of the bulges always is on radial orbits, the stellar veloities beome radially anisotropi at radiiapproxi-mately larger than the spherialhalf-mass radius
•
Using diuse satellites, the mean square speeds of the remnants derease with eahsubsequentgeneration,whihisonlylimitedbythehighamountofmass-lossand onsequently the gravitational radii inrease muh less than expeted.
•
The head-on minor mergers of ompat one-omponent models evolve nearly homologous, i.e. the observable values like the line-of-sight veloity dispersionand the eetiveradius evolve very lose to those of the whole system.
•
In all other minor merger sequenes, we observe a dramati break of homology,asthe remnantsbuild up anextended envelope ofstars, whilethe entral
ong-uration stays onstant.
•
Therefore the eetive radii inrease rapidly up toafator of4.5
, whihismuhloser to virialexpetations.
•
The rapidsize growthresultsina signiantinrease of thedarkmatterfrationwithin the spherialhalf-mass radius up toa fator of
∼ 1.8
.•
Due to the inreasing dark matter fration, the eetive line-of-sight veloity dispersions donot derease but stay onstant.One important question whih has to be solved for elliptial galaxies is, how the
ompat early-types at a redshift
z ∼ 2
grow with time. As their stellar distribution is already red without signiant star formation, we used dry mergers to explain thisevolution. van Dokkum et al. (2010) nds a size-mass relation of
r e α M 2.04, whih
indiatesa size inrease of a fator of4 asthe galaxy'smass gets doubled sine
z ∼ 2
.The resultingrelationof our minor mergersenarios of two-omponent models iseven
higher
r e α M >2.04 up to a exponent of 2.4
, whih shows that dissipationless minor
mergers are a good way to solve this problem. However, Nipotiet al. (2009a) tried a
similarapproahandndamuhlowersizeinreaseintheirsimulations(
r e ∝ M 1.09).
One reason forthis bigdisrepany is,that they alulated the exponent of thestellar
mass by averagingoverall theirmergerhierarhies. As they havemore major mergers
than minor mergers, this of ourse lowers the size inrease signiantly. Additionally,
they use a steeper slope for the stellar density prole of the host and the satellite
galaxies, where the size inrease an not be that eient, as the areted material
is more onentrated in the satellite's enter ompared to our setup. Finally, their
satellites are even more ompat than our satellites whih lie on an extrapolation of
the
z = 2
mass-size relationof Williamsetal. (2010).Furthermore we nd that the dark matter frations for our idealized simulations
agreewellwithpreviouswork,wherethe darkmatterfrationinreasesindrymergers.
This hanges the ratio of dynamial and stellar mass and might, e.g. help to explain
the tiltof the fundamentalplane (Boylan-Kolhin etal.,2005). Ofourse, that isjust
one possibility to explain the tilt and Grillo & Gobat (2010) suggest that it depends
moreon
M ∗ /L
, butit isnotlear yethowstrongthe single ontributions are. We also agree with Nipoti etal. (2009a),that the inrease of the dark matterfration is moreeientfor minormergersandforthis senarioisdominatedbythe rapidsize growth.
ButinontrasttoNipotietal.(2009a),wend thattheentraldarkmatterfrationof
equal-massmergersillustratesa'real'hangeaused byviolent relaxationand mixing.
Looking at the veloities at dierent radii, our minor merger results are not able
to explain reent observations of very high veloity dispersions at high redshift (van
Dokkum et al., 2009; van de Sande et al., 2011). Our results indiate, that we get a
derease of the mean square speeds of the total system, but the observed line-of-sight
veloity dispersionhardly hanges. This indiates, that simple dissipationlessmergers
arenotabletodereasetheveryhighLOSVDofsomeompatearlytypegalaxies(van
Dokkumet al.,2009). Thisproblemmightbesolved,if we inludesome gas andAGN
feedbak oruse more realisti galaxymodels, whih have dierent orbital properties.
Butaltogether ourwork shows thatdissipationlessdrymergersareable toinrease
the size of aompat earlytypegalaxy. As welieeven abovethe observed preditions
asmallamountofgas,whihisknown tolowerthesizegrowth (Covingtonetal.,2011;
Hopkins et al.,2008), would perhaps not be enoughto destroy this senario.
CHAPTER 7
SIZE AND PROFILE SHAPE
EVOLUTION OF MASSIVE
QUIESCENT GALAXIES
In this hapter, we fous on the evolution of the density struture and the
size evolution of ompat early-type galaxies and try to understand the
im-portane of dark matter. We know that the sizes and mass distributions of
ompat, quiesent, massive galaxies evolve rapidly from
z ∼ 2 − 3
to thepresent. Many of the
∼ 10 11 systems athigh redshift havesizes of ∼
1kp and
surfae brightness proleswith Sersi indies
<
4. Atz = 0
elliptialgalaxiesabove
2 · 10 11solar masses are morethan afator of 4larger, indiatinga size
evolution of r
∝ M α with α ≥ 2
. They also have surfae brigtness proles
with
n ser ≥ 8
. Within a hierarhial galaxy formation senariothis evolutionan be explained under two assumptions. The galaxies predominantly grow
by mergerswith lower massgalaxies and the galaxies have tobe embedded in
massive dark matter halos so that stars of merging satellites are stripped at
large radii inreasing the prole shapeparameter. We drawthese onlusions
fromidealizedsimulationsof thegrowth ofompat spheroidalgalaxies -with
and without dark matter - by repeated ollisionlessmergers with mass ratios
of 1:1, 1:5, and 1:10. In simulations withoutdark matter the sizes evolve less
than the orresponding bulge+halo senarios. If the galaxies are embedded
in darkmatter halos the stars of the lower mass satellites are more eiently
stripped at large radii resultingin a signiantly faster size inrease than
ex-petedfromvirialestimates. Repeated1:5mergersgive
α = 2.3
andafteronlytwomergergenerations theSersiindex has alreadyinreased to
n ser > 8
. Foran assumed mass inrease of the observed galaxies of a fator of two sine z
=2 we onlude that the presene of a massive dark matter halo around the
galaxies duringtheir minor mergerdriven assembly is neessary to explain
si-multaneously their large present day sizes, r
>
4 kp and high Sersi indies,n ser > 6
.7.1 Introdution
Intheurrentlyfavoredosmologial
Λ
CDMmodel,theuniverseonsistsof24%
matterand
76%
darkenergy (Λ
),where only4%
of the total matter is inbaryoni form(e.g.Spergel etal. (2007)). The other 96%onsist of old darkmatter,whihhas not been
deteteddiretly,buthasbeenmostsuessfullyappliedtoexplainmanyobservational
aveats like the rotation urves of spiral galaxies. On large sales the
Λ
CDM model,shows verygoodagreementwithobservationsoftheosmimirowavebakgroundand
the large sale struture of galaxies. In the ontext of the
Λ
CDM model, struture inthe universe forms bottom up (White & Rees, 1978; Davis et al., 1985), where the
rst objets ollapse at high redshifts due to utuations in the bakground density
eld. Theserstobjetsmergeandbuildup thedarkmatterhalosoftoday'sobserved
galaxies.
The baryons assemble in the potential wells of these dark matter halos and form
stars whihbuildthe observablepartsof theuniverse. Thebrightestandmost massive
objets are elliptial galaxies, whih form ata redshift of
z ∝ 2 − 3
in gas-rih majordiskmergers(Davisetal.,1985;Bournaudetal.,2011)anddue togiantoldgas ows,
diretly feeding the entral galaxy (Kere² et al., 2005; Naab et al., 2007, 2009; Joung
et al., 2009;Dekel etal.,2009; Kere² etal.,2009; Oser etal.,2010). Theirsubsequent
evolutionisnotfullyunderstoodyet,astheseelliptialsareafator
∼
4-5smallerthantheir ounterparts in the present day universe. On the other hand, they are already
quiesent, without star formation, and are only a fator of
∼
2 less massive (Daddiet al., 2005; Trujilloet al., 2006; Longhettiet al., 2007; Toft et al., 2007; Zirm et al.,
2007;Trujillo etal.,2007;Zirm et al.,2007; Buitrago et al., 2008; vanDokkum et al.,
2008; Cimatti et al., 2008; Franx et al., 2008; Sarao et al., 2009; Damjanov et al.,
2009; Bezanson et al., 2009). In addition, they have very dierent surfae brightness
proles. In partiular, the ompat, high redshift elliptials always have steep power
lawusps intheirenter whereas themore extended presentday elliptialshaveored
proles. This means, that tting a Sersi prole to elliptial galaxies either results
in entral extra light, where the entral surfae brightness is above the tted prole
or in ase of ore elliptials the prole predits more light than the galaxy has. The
Sersi indies,whih area measurementof theprole'survatureare
∼
4forthe uspygalaxies and
∼
8-10 for the ore elliptials. Furthermore, reent observations of strong gravitationallensinginthe SLACS sample(Koopmansetal.,2006;Boltonetal.,2008;Gavazzietal.,2007,2008;Auger etal.,2009,2010)haverevealedaninreasingentral
dark matterfration with stellar mass (Barnabè etal., 2011).
Inthishapter,weinvestigatetheevolutionofelliptialgalaxieswiththeaidof
high-resolution N-body simulations of idealized one- and two-omponent galaxy models.
Withdierentinitialmassratios andadierenthoie ofmergerorbitswe explorethe
eetoffrequentdissipationlessgalaxymergers. Insetion7.2wegiveashortoverview
of the galaxy properties and the simulation parameters. In Setion 7.3we present the
eieny ofdry mergersfor the size growthof ompat galaxies and inSetion7.4we
lookatthe evolutionofthe surfaedensitiesand themass assemblyinmultiplemerger
generations. InSetion7.5weonvert oursurfaedensitiestoviablesurfaebrightness
prolesand explore the evolutionof the Sersi proleand in Setions 7.6we illustrate
the hange of dark matterfrations. Finally wedraw our onlusions inSetion 7.7.