3D Spectroscopy with VLT/GIRAFFE - IV: Angular Momentum and Dynamical Support of Intermediate Redshift Galaxies

0.41 MB·English

by  M. Puech

Astronomy&Astrophysicsmanuscriptno.momentum˙v9 (cid:13)c ESO2008
February5,2008
3D Spectroscopy with VLT/GIRAFFE - IV: Angular Momentum &
⋆
Dynamical Support of Intermediate Redshift Galaxies
7
0 M.Puech1,F.Hammer1,M.D.Lehnert2,andH.Flores1
0
2
n 1 LaboratoireGalaxiesEtoilesPhysiqueetInstrumentation,ObservatoiredeParis,5placeJulesJanssen,92195MeudonFrance
a
J 2 Max-Planck-Institutfu¨rextraterrestrischePhysik,Giessenbachstraße,D-85748Garching,Germany

5
2 Received...............;accepted...............

1
ABSTRACT
v
3
Context.Oneofthemostoutstandingproblemsrelatedtonumericalmodelsofgalaxyformationistheso-called“angularmomentumcatastro-
2
phe”,i.e.,theinabilitytoexplaintheoreticallythehighangularmomentumobservedinlocaldiskgalaxies.
7
Aims.Westudytheevolutionoftheangularmomentumfromz∼0.6toz=0tofurtherour understanding ofthemechanismsresponsiblefor
1
0 thelargeangular momentaofdiskgalaxiesobservedtoday. Thisstudyisbasedonacompletesampleof32, 0.4≤z≤0.75galaxiesobserved
7 withFLAMES/GIRAFFEattheVLT.Theirkinematicshadbeenclassifiedasrotatingdisks(11galaxies),perturbedrotators(7galaxies),or
0 complexkinematics(14galaxies).
/ Methods.Wehavecomputedthespecificangularmomentumofdisks(j )andthedynamicalsupportofrotatingdisksthroughtheV/σratio.
h disk
Tostudyhowangularmomentumcanbeacquireddynamically,wehavecomparedthepropertiesofdistantandlocalgalaxies,asafunctionof
p
theirkinematicalclass.
-
o Results.Wefindthatdistantrotatingdiskshaveessentiallythesameproperties(j andR )aslocaldisks,whiledistantgalaxieswithmore
disk d
r complexkinematicshaveasignificantlyhigherscatterinthe j –V andR –V planes.Onaverage,distantgalaxiesshowlowervaluesof
t disk max d max
s V/σthanlocalgalaxies,thelowestV/σvaluesbeingreachedbydistantgalaxiesshowingperturbedrotation.Thiscanprobablybeattributedto
a
heatingmechanismsatworkindistantdisks.
:
v Conclusions.Wefoundobservationalevidenceforanon-linearrandomwalkevolutionoftheangularmomentumingalaxiesduringthelast8
i Gyr.Theevolutionrelatedtogalaxieswithcomplexkinematicscanbeattributedtomergers,butnotto(smooth)gasaccretionalone.Ifgalaxies
X
observedatintermediateredshiftarerelatedtopresent-dayspirals,thenourresultsfitquitewellwiththe“spiralrebuilding”scenarioproposed
r byHammeretal.(2005).
a
Keywords.Galaxies:evolution;Galaxies:formation;Galaxies:kinematicsanddynamics;Angularmomentum;3Dspectroscopy.
1. Introduction as∼40%offieldgalaxiesarenotinequilibrium,i.e.,galaxies
that are not suitable to establish a proper T-F relation. They
The evolution and the origin of the Tully-Fisher (T-F) rela-
thenderivedaT-Frelationthatdoesnotappeartohaveevolved
tion is still a matter of intense debate. A strong evolution
sincez∼0.6,intheM (orstellarmass)versusV plane.On
K max
both in slope and dispersion has been found in B band (e.g.,
the otherhand,the stellar massdensity increasesfromz=1to
Ziegleretal.2002;Bo¨hmetal.2004),fromz∼1toz=0.More
z=0 (e.g., Droryetal. 2005): assuming a rough30% increase
recently, Conseliceetal. (2005) derived the T-F relation in K
in stellar mass fromz=0.6 to z=0, we shouldthen see a ∼0.1
band,K-bandabsolutemagnitudebeingabettertracerofstellar
dex shift along the M axis between the local and distant T-F
∗
mass,andfoundsuchalargedispersionthatonecouldwonder
relations. However, one should keep in mind that typical un-
whetherornottheT-Frelationevenexistsatz∼1.Floresetal.
certainties usually associated with M are of the same order,
∗
(2006, hereafterPaper I) derivedthe first T-F relation fordis-
i.e.,0.1-0.2dex(seePaperI).Anotherpossibilityistoassume
tantgalaxiesusingintegralfieldspectroscopy.Thetwodimen-
somegasaccretionfromtheintergalacticmediumwhichwould
sional spatial coverage allowed them to properly identify the
be directly converted into stellar mass: while M increases,
∗
dynamicalnatureofdistantgalaxies,andestablishthatasmuch
V could then increase in the same time, because the total
max
mass of the system mainly depends on the rotational veloc-
Sendoffprintrequeststo:[email protected]
ity. Then, both quantities could evolve such as the resulting
⋆ Based on FLAMES/GIRAFFE Paris Observatory Guaranteed
evolution in the T-F plane operates along the relation (e.g.,
Time Observations collected at the European Southern Observatory,
Portinari&Sommer-Larsen 2006). More statistics and/or fu-
Paranal,Chile,ESONo.71.A-0322(A)and72.A-0169(A)
2 Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies
ture studies of the T-F relation at higher redshifts (where the Peiranietal. 2004). For a given galaxy, it is in this way that
stellar mass density was much smaller than at z=0.6) should major mergersare the main source of either positive or nega-
providea decisive test of this point. Whether the T-F relation tivechangeinitsangularmomentum.
evolveswithtimewillnodoubtprovideimportantcluesabout Relating the angular momenta of the halo and the disk is
relationshipbetweenthegrowthofmassandthecharacteristics notatallstraightforward.Areasonableassumptionoftenmade
ofthestellarpopulations,and,ultimately,aboutgalaxyforma- isthatthe specificangularmomentum j(i.e.,theangularmo-
tionandevolutioningeneral. mentum per unit mass), is conserved during the collapse of
The ideas concerning the origin of the T-F relation can the gas (Mestel 1963; Fall&Efstathiou 1980). If we also as-
be divided into two broad categories. In the first one, the T- sume that both gas and dark matter are well mixed in the
F relation originates from the cosmological equivalence be- proto-galaxy(Fall&Efstathiou1980),thenthisleadsto j ∼
disk
tween the halo mass and the circular velocity (e.g., Moetal. j . Making this assumption allows models and simulations
halo
1998).Therelationthencomesfromthefactthatthefiniteage to reproduce several properties of local disk galaxies (e.g.,
of the Universe imposes a maximal radius from where mat- Moetal. 1998; Syeretal. 1999; vandenBosch 2001, 2002a;
ter can be accreted to form a galaxy. The second broad cat- Toninietal. 2006). However, some problems remain, such as
egories of models invoked to explain the T-F relation is self- the “mismatch of angular momentum profiles” between the
regulated star formation in disks of different mass (e.g., Silk dark matter and the disk (Bullocketal. 2001; vandenBosch
1997). However, numerical simulations taking into account 2002a;Maller&Dekel2002),andtheinabilityofmodelstosi-
both ingredientsof gas accretion and self-regulationhave not multaneouslymatchcharacteristicssuchastheslopeandzero-
been able to reproduce all aspects of the T-F relation, such point of the T-F relation, slope and zero-point of the radius-
asthezeropoint(e.g.,Steinmetz&Navarro 1999).Manyau- luminosityrelation,theluminosityfunctionofspiralswithrea-
thors suggested that feedback from massive star formation or sonable values for the disk masses, halo structural parame-
active galactic nuclei could help to solve these discrepancies ters, and circular velocity relative to virial velocity (see, e.g.,
(see, e.g.,Eisenstein&Loeb1996;Heavens&Jimenez 1999; Duttonetal.2006).
Steinmetz&Navarro 1999). Thebuild-upofangularmomentuminrotatingdiskscould
Feedback has also been suggested to solve the so-called be better understood through the comparison of theoretical
“angular momentum catastrophe” of the Λ-CDM model, i.e., models and simulations with observations of distant galax-
the inability of simulations to reproduce disk galaxies with ies. In a first attempt at estimating the angular momenta of
sufficient angular momentum in comparison with what is distantgalaxies,Fo¨rsterSchreiberetal.(2006)studiedseveral
observed (e.g., Steinmetz&Navarro 1999). Feedback has z∼2galaxies,andfoundthat,∼10Gyrago,galaxiesappearto
been proposed as a potential solution to this problem, which have approximately the same specific angular momentum as
has been investigated in detail (e.g., Maller&Dekel 2002; today’sspirals,withvaluesroughlysimilartothatexpectedfor
D’Onghiaetal. 2006;Governatoetal. 2006).State-of-the-art theirhaloes(seealsoNesvadbaetal. 2006).Theyarguedthat
numericalsimulations,includingtheeffectsofAGNfeedback thisconfirmsthe hypothesisthatbaryonslikelyacquiredtheir
driven through accretion of gas onto a super-massive central angular momentum during the collapse of their parent dark
black hole, show how a disk can re-form after the merging matter halos. However, as they pointed out, the spatial reso-
of two rotatinggas-richdisks, with a sufficientand consistent lution of their observationsdoesnot allow to uniquelydistin-
amountofangularmomentum(Robertsonetal. 2005). guish between rotating disks and merger-inducedkinematics,
The mechanisms through which galaxies may have ac- thustheoriginofhighangularmomentuminz∼2diskgalaxies
quired their angular momentum has been discussed for many stillremainsunclear.
decades (e.g., Stro¨mberg 1934; Hoyle 1951; Mestel 1963). At lower redshift, a sample of z∼0.6 galaxies has
The so-called “gravitational instability paradigm”, indepen- been observed using the multi-integral field spectrograph
dent of the details of the cosmology, predicts that the angu- FLAMES/GIRAFFE at VLT (see Floresetal. 2006;
lar momentum of a protogalaxy should grow linearly with Puechetal. 2006, hereafter Paper I and Paper II, respec-
time due to tidal torques from interactions with neighboring tively). GIRAFFE observations are confronted with similar
structures, until it decouples from the Hubble flow (Peebles difficulties in term of spatial resolution as any other study of
1969; White 1984). In a more modern picture, galaxies high redshift galaxy dynamics. However, to mitigate against
form from infalling baryonic gas embedded into dark mat- these effects, we developed a classification scheme based on
ter haloes, and their angular momentum is then inherited the kinematics and morphologies of the galaxies, separating
from the halo (White&Rees 1978; Fall&Efstathiou 1980; them into rotating disks and galaxies with complex or dis-
Barnes&Efstathiou 1987). After the end of the epoch dur- turbed kinematics and morphologies(see Paper I for details).
ing which tidal torquing is effective, subsequent evolution of Givenalloftheothercomparisonsmadeinsubsequentpapers,
the angular momentum takes place non-linearly, through a e.g., Paper II and Puechetal. (2006b, hereafter Paper III),
random walk process associated with mergers events and/or this classification method appears very robust (see detailed
mass accretion (Vitvitskaetal. 2002; Maller&Dekel 2002; discussionsinbothPaperIandPaperII).Thegoalofthispaper
Peiranietal. 2004). This random walk leads to a change in is to derive the specific angular momentum in these z∼0.6
the angular momentum of the haloes, with a more signifi- galaxies. The kinematics and emission line properties of this
cant change (increase or decrease, depending on the geome- sample hasbeenstudiedin detailin thethree previouspapers
tryofthemerger)duringmajormergers(Vitvitskaetal.2002; of this series (dedicated to the GIRAFFE Guaranteed Time
Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies 3
Observationsample;seePaperI,PaperII,andPaperIII).This onR weredirectlytakenas1-sigmauncertaintiesreturnedby
d
paperistheforthofthisseriesandisorganizedasfollows:§2 GIM2D,withamedianvalueof0.12kpc.
summarizes the observations and introduce the methodology.
§3 presents the specific angular momentum of the GIRAFFE
sample.§4discussesthedynamicalsupportofdistantrotating
disks,while§5and6discussthe implicationsandsummarize
ourresults.
2. Observations&Methodology
As part of the Guaranteed Time Observations (GTO) of the
Paris Observatory, we obtained observations with the multi-
integralfieldspectrographFLAMES/GIRAFFE,ofacomplete
sample of 32 galaxies,with redshiftsrangingfrom0.4 to 0.7,
EW (OII)≥ 15Å and I ≤22.5. Briefly, we used the LR04
o AB
andLR05grating,changingthesetupstospecificallytargetthe
[OII]λλ3726,3729doublet(withR∼ 10000);integrationtimes
rangedfrom8to13hours;theseeingwastypically∼0.6−0.7
arcsec during all the observations. Data cubes were reduced
usingtheGIRBLDRSv1.12package(Blechaetal. 2000),in-
cluding a flat-fielding. Sky was carefully subtracted with our
ownIDLprocedures.Wederivedforthesegalaxiesbothveloc-
ityfieldsandvelocitydispersionmapsinPaperIandPaperII.
Thesemaps,aswellasHSTmorphology,wereusedtodivide
thesampleintothreedistinctclassesbasedmainlyontheirdy-
namicalcharacteristics:rotatingdisks, perturbedrotators, and
Fig.1. Comparison of half-light radii (in kpc) derived using
galaxies with complex kinematics. A complete description of
GIM2D and using the modeled results from fitting isophotal
theGTOsampleandofthemethodsandanalysisusedtoclas-
ellipsestothegalaxylightprofile.Theresultinghalflightradii
sifythegalaxiesfromoursamplearegiveninPaperI.
derived from both methods agree very well, with the galax-
Thespecificangularmomentum jofarotatingsystemcan
ies with disturbed or complex kinematics showing additional
beestimatedas(see,e.g.,Moetal.1998):
scatter.Bluedotsrepresentgalaxiesclassifiedasrotatingdisks,
green squaresrepresentperturbedrotations,and red triangles
j=βRV ,
max
representgalaxieswithacomplexkinematics.
whereβis a dimensionlessparameterthatdependsonthe ge-
ometry and spatial distribution of mass, R a characteristic ra-
Unfortunately,forfivegalaxies,weonlyhadground-based
diusofthemassdistribution,andV isthemaximalrotation
max images available which were taken at the CFHT (see, e.g.,
velocity.ForathinexponentialdiskofscalelengthR ,thisre-
d Schadeetal. 1996, ; 0.207arcsec pixel−1). Because these im-
lationbecomes:
agesaresignificantlyaffectedbytherelativelylargeseeingdisk
j =2R V .
disk d max (largecomparedtotheR ofthedisks),wedidnotattemptto
d
To estimate a disk scale length, R , it is necessary to de- use GIM2D to fit their disk light profiles since certainly the
d
convolvethediskfromthebulgecomponent.WeusedGIM2D solution would be highly degenerate, and thus unlikely to be
(Simard&Pritchet1998;Simardetal.2002)tomeasureR in reliable.Instead,weusedthefactthatforanexponentialdisk,
d
thegalaxiesobservedwithGIRAFFEusingHST/WFPC2im- R cantheoreticallybeobtainedfromthehalflightradiusr
d half
ages(0.1arcsecpixel−1)orwithpreference,ACSimages(0.05 usingR =r /1.68.InFigure2,wecomparedR asmeasured
d half d
arcsecpixel−1)whentheywereavailable.InPaperIandPaper by GIM2Dwith r /1.68as deducedfrom PaperI and Paper
half
II,wederivedthehalflightradiusfromthemodelingofisopho- II. This figure shows that this is a reasonable assumption (at
talellipses(seealsoHammeretal.2001).Figure1showsthat least for rotating quiescent disk galaxies). We also used this
both half light radii derived using this method and GIM2D methodfortwogalaxiesforwhichGIM2Dfailedtoprovidea
agreeverywell.Unfortunately,GIM2Ddoesnotestimateany statisticallyrobustfittothelightprofiles.
uncertainty for the half light radius, but it is noteworthy that BothFigures1and2showthatforgalaxiesexhibitingmore
a linear fit, between the half lightradii derivedusing GIM2D complexkinematicsthanasimplerotatingdisk(i.e.,thoseclas-
and the modelling of isophots, returns a median standard de- sified as perturbed rotation or complex), the estimates of R
d
viation of ∼0.29 kpc, in close agreementwith the typical un- and/or r have a larger relative dispersion. In general, how-
half
certaintyof∼0.34kpconthehalflightradiusderivedfromthe ever,eventhegalaxieswithcomplexkinematicshaveR and/or
d
isophotalellipses modelling,as claimed in Paper I. Giventhe r consistent with the average results of the rotating disks.
half
fittingmethodsareindependent,thiscomparisonsuggeststhat Therelativelyhighdispersioncanbeeasilyunderstoodasare-
our R measurementsusing GIM2D are robust. Uncertainties sult of their unrelaxed dynamical state: it is likely that these
d
4 Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies
BecauseofthecoarsesamplingoftheGIRAFFEIFUdata
(pixel size of 0.52 arcsec, ∼ 3.5 kpc at z∼0.6), a mean cor-
rectionof20%hastobeappliedonV .Thiscorrectionfac-
max
tor is appropriate for z∼0.6 rotating disks (see discussion of
thisinPaperI).Thiscorrectionfactorwasdeterminedbysim-
ulating GIRAFFE observationsfrom hydrodynamicalsimula-
tions of a Sbc-like rotating disk. As explained in Paper I and
PaperII,weareprobablyseverelyunderestimatingthiscorrec-
tionfactorforobjectsshowingmorecomplexkinematics(see
PaperII),andthiscouldintroducearelativeoffsetbetweenro-
tating disks and objects with more complex kinematics. The
absoluteuncertaintyof thecorrectionfactorfor rotatingdisks
was determined to be ±4% because of possible variations in
size andinclination(see PaperI).Anadditionalsourceofun-
certainty on this correction factor could be due to variations
of the dynamicalpropertiesalong, e.g., the Hubble sequence.
Wecheckedthat,usinganotherhydrodynamicalsimulationof
a “mean” SDSS galaxy,a mean 20% correction factor is still
appropriate,within± ∼2%relativelyto20%correctionfactor.
Weemphasizethatthisuncertaintyisverydifficulttoestimate,
given the high number of parameters which have to be taken
intoaccount.
Fig.2.Comparisonofdiskscalelengths(inkpc)derivedusing Finally, uncertainties on j were estimated using usual
disk
GIM2Dandusingthemodeledresultsfromfittingisophotalel- methods,i.e.propagatingtheindividualuncertaintiesonV ,
max
lipsestothegalaxylightprofile.Theresultingdiskscalelength R , and on the correction factor on V , as detailed above.
d max
from both methods agree very well, with the galaxies with Thisgivesamedian1-sigmauncertaintyof0.09dexand0.10
disturbed or complex kinematics showing additional scatter. dexinV and j ,respectively.Thisistobecomparedwith
max disk
Bluedotsrepresentgalaxiesclassifiedasrotatingdisks, green typicaluncertaintiesforlocalgalaxiesof0.03and0.06dexre-
squaresrepresentperturbedrotations,andredtrianglesrepre- spectively,asestimated fromCourteau(1997). We emphasize
sent galaxies with a complex kinematics. The median uncer- thatsuchanuncertaintyismeaningfulonlyforrotatingdisks,
tainties,i.e.,0.12kpcforGIM2D,and0.34/1.68forthemod- since the accuracy and appropriateness of the model used to
elling of the isophots respectively,are indicated in the upper- reproducethe data is nottaken into account.This uncertainty
leftcorner. shouldthenbeviewedasarandomuncertaintyassociatedwith
j ,iftherotatingdiskmodeliscorrect,butdoescertainlynot
disk
include the systematic uncertainty when such a model is not
appropriate.Indeed,inPapersIandII,weemphasizedthefact
galaxies are not in rotational equilibrium, and an exponential
that objects showing complex kinematics are probably merg-
disk model is then probably an inadequate representation of
ers or mergerremnants.For these objects,the j derivedas
disk
theirtruelightprofile.However,togethomogeneousandcon-
aboveshouldthenbeviewedastheorbitalangularmomentum
sistent estimates, we choose to treat the whole sample, what-
due to the relative motion of the two progenitors, rather than
ever the kinematical class is, as if all galaxies were rotating
the intrinsic spin angularmomentumof a single rotating disk
disks. We also note thatdespite the increasedscatter between
(seealsoFo¨rsterSchreiberetal.2006).Wewilldiscussthisin
the variousestimatesof the scale lengthand half-lightradius,
moredetailsubsequently.
eachkinematicsubsamplegenerallyfallsalongtheone-to-one
line. Itis simply thatthe individualestimatesare less reliable
for the galaxies with complex kinematics but not for the en-
3. AngularMomentumofDistantGalaxies
sembleofeachclass.
V wasestimatedbyfittingadoublegaussiantothe[OII] To compare distant with local galaxies, we used two local
max
doublet(seePaperI).Thisintroducesarelativerandomuncer- samples from Courteau (1997) and Mathewsonetal. (1992).
tainty on the measurement of V that is estimated to be ∼ For the first sample, we used the compilation made by
max
10 km/s, from the comparisonof independentfits to the data. Steinmetz&Navarro (1999), and kindly provided by M.
Thisuncertaintyis relativelylow becauseof the highspectral Steinmetz. For the second sample, we retrieved the electron-
resolution of GIRAFFE (R∼10000) which allows a good ve- ically available data from the CDS1. For this sample, no Rd
locitymeasurement.Moreover,thelimitedsizeoftheIFUcan measurementweredirectlyavailable:we deriveditfromtheir
introduceanobservationalsourceofuncertainty,ifthissizeis (I band) 23.5 magnitude isophotal radius dividing by a mean
toosmalltoreachtheflatpartofrotationscurves.Howeverwe correctionfactorof3.5(Palunas&Williams2000).
showed in Paper I that the size of the GIRAFFE IFU is well-
suitedtomeasureV forallthegalaxiesofthesample. 1 http://cdsweb.u-strasbg.fr/
max
Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies 5
Figure 3 shows the disk scale length versus the maximal
rotationvelocity,bothforlocalanddistantsamples.Bothlocal
samplesagreequitewell,inspiteofthedifferentproxiesused
for the maximal rotational velocity and the disk scale length.
Distantrotatingdisksfallclosetolocalrotatingdisks,although
some of them appear to have a relatively lower disk scale
length. Those are associated to galaxieswith half light radius
slightly lower than that of compact galaxies (i.e., R ≤ 4.7
half
kpc,seePaperII).Ontheotherhand,morekinematicallydis-
turbeddistantgalaxiesshowaveryhighdispersionaroundthe
localR −V relation.Itcanbeexplainedintwoways.First,
d max
morekinematicallycomplexdistantgalaxieshavealargerun-
certaintyinthedeterminationoftheirradius.Thisuncertainty
can be estimated from the scatter of kinematically complex
galaxiesinFigure2,andisσ∼1.45kpc.Relativelytotheme-
dianR ofkinematicallycomplexgalaxies,thistranslatesinto
d
a 0.2 dex scatter in Figure 3. It is thus clear that this is not
sufficienttoexplaintheextremelylargescatterofthekinemat-
ically complex galaxiesin the R −V plane (i.e., σ ∼0.45
d max
dex relatively to the fit of the local galaxies R − V rela-
d max
tion).Thesecond(andmostimportant)sourceofscatteristhat
rotationalvelocitiesundergoabruptspatialandtemporalvari-
ations during dynamical processes such as minor and/or ma- Fig.3. Disk scale length versus maximal rotational velocity,
jor mergers. This effect is illustrated by the black pentagons for the sample of 32 distant galaxies presented in Paper I
track:theyrepresentahydrodynamicalsimulationofamerger and Paper II. Blue dots represent galaxies classified as rotat-
of two Sbc galaxiesfrom Coxetal. (2006). GIRAFFE obser- ingdisks,greensquaresthoseclassified asperturbedrotators,
vationsweresimulatedalongthemergingsequence,aswellas andredtrianglesgalaxieswithcomplexkinematics.Thesmall
HST images (Puech et al., in preparation). From these simu- black dots represent the local sample of Courteau (1997) and
lateddata,bothrotationvelocitiesanddiskscalelengthswere Mathewsonetal.(1992).Themedian1-sigmauncertaintiesare
extractedandcorrectedfollowingsimilarmethodsusedforreal indicatedintheupper-leftcorner,forbothdistantrotatingdisks
distantgalaxies(seeprevioussection).Duringthissimulation, (bluedots),andlocaldisks(blackdots).Forlocalgalaxies,Rd
themergeroccursat∼1.8Gyrafterthebeginningofthesimu- havebeendeterminedusing1Dfitswhereas2Dfitswereused
lation,andissimulatedforupto0.5Gyrsafterthemerger.At fordistantgalaxies:thisexplainswhytheuncertaintyalongthe
theendofthesimulation,theremnantlooksrelativelysimilar ordinateislargerforlocalthanfordistantgalaxies.Blackpen-
to an elliptical, with a small inner disk at the center. It is be- tagonsrepresentsimulatedGIRAFFEobservationsusingahy-
yondthescopeofthispapertoexploreforeverypossibletrack drodynamicalsimulationofamajormergeroftwoSbcgalax-
in Figure3, varying,e.g.,the gasfraction,orbitalparameters, ies.Thebeginningandtheendofthesequence,aswellasthe
or the amount of feedback. What is important to note here is mergeritself,areindicatedbysolidsymbols.
that this simulation, which includes possible observational or
instrumentalbiases,illustrateshowthescatterofthemostkine-
maticallydisturbeddistantgalaxiescouldbeeasilyreproduced we are likely measuring an orbital angular momentum rather
bysuchevents.Thisstrengthenstheideathattheseobjectsare than an intrinsic spin angular momentum of a single disk. In
probablyongoingmergerand/ormergerremnants. sucha case, the shapefactorβ=2is probablynotappropriate,
Figure 4 shows the specific angular momentumof the ro- andcouldbeansourceofadditionaldispersion.Toinvestigate
tatingdisksinthesampleofgalaxiesobservedwithGIRAFFE, this effect, we also over-plottedin this Figure the same simu-
j ,versustheirmaximumrotationvelocity,V .Inthisfig- lation of GIRAFFE observationsof a mergerof Sbc galaxies,
disk max
ure, z∼0.6 rotating disks show a specific angular momentum as described above. Note that the beginning of the sequence
comparable to that of local galaxies with same V (i.e., ∼ correspondstoonlyoneofthetwoprogenitors,whichshowsa
max
same total mass). Objects with a complex kinematics have a deficitinspecificangularmomentumcomparedtolocalgalax-
lowermedianspecificangularmomentum,withlogj ∼2.7, ies. This could be related to the difficulty for hydrodynami-
disk
than in rotating disks, which have a median of logj ∼ calsimulationstoreproducethehigherangularmomentumob-
disk
3.0. This is probably again due to the fact that V can be served in local disks. The position of most of galaxies with
max
severelyunderestimatedforgalaxieswithcomplexkinematics complexkinematicsinthe j −V planecanhereagainbe
disk max
(seePaperIandPaperII,andabove):theirderivedspecifican- easily explained as a result of mergers(see black pentagons).
gular momentumshould then be consideredas lower bounds. As thesesimulateddata includemanypossiblesourcesofun-
As in Figure 3, the dispersion of distant galaxies with com- certainty, it clearly suggests that the larger scatter of galaxies
plexkinematicsissignificantlyhigher.Astheseobjectsarehy- showing complex kinematics can be associated with mergers
pothesizedtobeundergoingamergeroraremergerremnants, ormergerremnants.
6 Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies
sicV/σismorerobustlyestimatedusingtheratiobetweenthe
meansquaredrotationalvelocityandthemeansquaredvelocity
dispersion,bothmeasureddirectlyalongtheline-of-sight.
We choosetoestimate V usingthemaximalrotationalve-
locity(correctedfrominclinationandspatialresolutioneffect,
see§2),asitisprobablythemostaccuratequantityderivable
from GIRAFFE velocity fields. Because the velocity gradient
of the rotationcurveof z ∼0.6 rotatingdisk galaxiesfalls ap-
proximativelyin only one GIRAFFE IFU pixel (or aboutone
resolutionelementgiventhe seeing, see §2),the center ofthe
GIRAFFE velocity dispersion maps show a peak that is due
to shear and/orlarge-scale motionsin velocity,and cannotbe
usedtoestimatetheintrinsicvelocitydispersionofthedisk(see
PaperIandPaperII).Onthecontrary,theregionssurrounding
the peak of the velocity dispersion are much less affected by
the shear and/or large-scale motions in velocity, since the ro-
tationcurveisapproximativelyflatintheseregions(i.e.,hasa
constant velocity with radius). These outer regions of the ve-
locitydispersionmapscanthenbeusedtoconstructareliable
estimateoftheintrinsicvelocitydispersionofthedisk.Tocon-
structsuchreliableestimatesoftheintrinsicvelocitydispersion
ofthedisks,wefirstremovedtheσpeakduetorotationinthe
Fig.4. Specific angular momentum of the disk, jdisk, versus velocitydispersionmap,guidedbythemodelingofthedisper-
their maximum rotation velocity, Vmax, for the sample of 32 sionwemadeinPaperI.Wethenestimatedσbyderivingthe
distant galaxies presented in Paper I and Paper II. Blue dots signal-to-noiseweightedmeanoftheremainingpixels.
represent galaxies classified as rotating disks, green squares Unfortunately, there are very few published velocity dis-
thoseclassifiedasperturbedrotators,andredtrianglesgalaxies persion maps of local galaxies that could be used for direct
with complex kinematics. The small black dots represent the comparison. However, several dozen velocity dispersion pro-
localsampleofCourteau(1997)andMathewsonetal.(1992). files of the gas in local disks that were obtained with long-
The median 1-sigma uncertainties are indicated in the upper- slitspectroscopyhavebeenpublished.We combinedthesam-
left corner, for both distant rotating disks (blue dots), and lo- ples of VegaBeltra´netal. (2001), Corsinietal. (2003), and
cal disks (black dots). Black pentagons represent simulated Pizzellaetal. (2004), which are composed of spiral galaxies
GIRAFFEobservationsusingahydrodynamicalsimulationof with morphological type earlier than Sc. Note that V/σ ratio
amajormergeroftwoSbcgalaxies.Thebeginningandtheend does not seem to depend on the morphological type, since a
of the sequence, as well as the mergeritself, are indicated by similar rangeof valuesare foundfor bothspiralandirregular
solidsymbols. galaxies (see Hunteretal. 2005). We checked that both local
and distant samples have similar distribution in the absolute
B magnitude, M , and we keptonly galaxieswith absolute B
4. DynamicalSupportofDistantDiskGalaxies B
magnitudebrighterthanthelowervaluefoundintheGIRAFFE
Inthissection,wecomparethedynamicalsupport(rotationvs. sample (i.e., M (AB) ≤ -19.26), to ensure that both samples
B
velocitydispersion)ofdistantandlocalrotatingdisks.Suchan are probinggalaxieswith comparablestellar masses and star-
analysiscouldhelpustounderstandhowrotatingdisksacquire formationrates.Toexcludethecentraldynamicallyhotregion,
andlosetheirangularmomentum. weperformedasigmaclippingonthevelocitydispersionpro-
Themostrevealingquantitytoestimatethedynamicalsup- file of local galaxies, keeping only the points along the curve
portofgalaxiesistheratioofrotationvelocitytovelocitydis- thatwerebelow2σaroundthemedian.Wethentookthefinal
persion of the disk, V/σ. V is a circular velocity (quantifying medianoftheremainingpointsasameasureofσinthedisk.
the amount of rotation), and σ is an estimate of the intrinsic Whenmeasuredprojectedontoinclineddisks,theobserved
velocity dispersion in the disk (i.e., turbulent or peculiar mo- velocitydispersionisacombinationofthethreespatialcompo-
tions). In elliptical galaxies, V/σ is usually estimated via the nentsofthevelocitydispersion–theradialcomponentσ ,the
r
ratio of the maximal rotational velocity to the mean veloc- azimuthalcomponentσ ,andtheverticalcomponentσ (e.g.,
φ z
ity dispersion within 0.5r , where r is the effective radius of Binney&Merrifield 1989):
e e
the system (e.g., Daviesetal. 1983; Benderetal. 1994). For
σ2 =(σ2sin2η+σ2cos2η)sin2i+σ2cos2i
spiral galaxies, there is no general consensus: different spa- obs r φ z
tial components of σ, estimated following different methods, where η is the angle between the observed PA and the ma-
havebeenused(e.g.,Bottema1993; VegaBeltra´netal. 2001; jor axis of the galaxy, and i is the inclination angle of the
Hunteretal. 2005). Binney (2005) demonstratedthatwith in- disk.In localspiralgalaxies,σ ∼ σ ∼ 0.7σ forstars (e.g.,
z φ r
tegral field spectroscopy (i.e., a 2-dimensional spatial cover- Hunteretal.2005).Inthecaseoflocalgalaxies,ifweassume
agecombinedwithsimultaneousspectralcoverage),theintrin- bothawellalignedslitandthatgasandstarsdynamicsarewell
Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies 7
coupled(seeVegaBeltra´netal.2001;Pizzellaetal.2004),one
gets σ ∼ σ . In the case of distant galaxies, observed us-
obs z
ing integral field spectroscopy, it is necessary to correct for
the different η corresponding to the different IFU pixels. We
can directly correct for this effect on the mean sigma by av-
eragingthecos2ηandsin2ηterms,andthenmultiplyingbya
(1 + sin2i/2)−1/2 correction factor. Finally, we obtain an es-
timate of the spatially averaged σ , that can be used in the
z
V/σratio.TocheckthatthefinalestimateofV/σdoesnotde-
pendsignificantlyonthemethodusedtoestimateσ,wecom-
paredwiththevelocitydispersionmeasuredatr /4asgiven
half
by VegaBeltra´netal.. Using this alternative method does not
significantlyaffectthegeneraltrendofV/σestimatedinlocal
disks.
Inthesampleoflocalgalaxies,errorshavebeenestimated
asfollows.TheuncertaintyonVwasdirectlytakenasthemean
of the velocity measurement uncertainty on V and V as
max min
given by VegaBeltra´netal. (2001), Corsinietal. (2003), and
Pizzellaetal. (2004). The uncertainty on σ was estimated as
the median of measurement uncertainties of the points of the
velocity dispersion curve kept during the sigma clipping (see
above). A conservative uncertainty of three degrees was as-
sumed for the inclination. We found a median uncertainty of Fig.5.V/σvs.uncertaintyinV/σforthesampleoflocalgalax-
1.6.FigureA.1showshowthisuncertaintyevolveswithV/σ. iesusedforcomparison.ThegalaxieswithsignificantV/σes-
This Figure shows that high values of V/σ have large associ- timates(V/σ≥3∆(V/σ))areshowninred.
ated uncertainties;includingonly sourceswith V/σ estimates
greater than three times their uncertainty,left us with a range
of local V/σ values ranging between 2 and 9, with a median
uncertainty in V/σ of 1.1. In the following, we will consider can we interpret this in terms of angular momentum growth
these points only, as others galaxies have V/σ ratios that are withtimeingalaxies?
too uncertain.For distantrotatingdisks, we estimated the un- Oneofthefirsttheoriesproposedforthebuildupofangu-
certaintyofσtobelowerthan15km/s(seePaperI).Weadded lar momentum in galaxies is the tidal torque theory (Peebles
anadditionaluncertaintytermduetothecorrectionfactorused 1969; White 1984). The basic idea in the tidal torque theory
todeconvolveV (see§2). is that most of the angular momentum is being gradually ac-
max
Figure6showsthecomparisonofV/σasafunctionofthe quired in the linear regime of growth of the density fluctua-
ellipticity inbothdistantandlocalrotatingdisk galaxies.The tions, duetotidaltorquesfromneighboringfluctuations.This
median V/σ is 3.8±2 for distant rotating disks, and 6.1±1.1 process continues relatively efficiently until the halo reaches
for local disks. Recall that highest local V/σ have been dis- its turn around time, i.e., roughly when the protogalaxy de-
carded in this comparison because of a too high uncertainty: couplesfromthe Hubbleflow.Moreover,thistheorysuggests
this could result in an underestimation of the local median thatangularmomentumgainisminimalwhenhaloesaregrow-
V/σ. We also plotted the V/σ ratio for the distant perturbed ing non-linearly after they have decoupled from the expand-
rotatingdisks, with a medianof 2.4±2.5.Figure 6 showsthat ingbackgroundandhaveformedvirializedsystems.Thisthe-
distant disks have lower V/σ ratios than local ones. This is ory then predicts rapid growth of angular momentum in the
consistent with the fact these galaxies are likely undergoing earlyevolutionofmassivehalos,andsubsequentlylittlegrowth
a minor merger and/or a gas accretion event which is heat- (Porcianietal.2002).Atfirstsight,ourresults,combinedwith
ing their disks (Walkeretal. 1996; Velazquez&White 1999; those of Fo¨rsterSchreiberetal. (2006) and Nesvadbaetal.
Abadietal.2003),aswassuggestedinPaperII.Moredatawill (2006)wouldmeanthatthespecificangularmomentumofro-
beneededtoconfirmthistrend,however. tatingdisksdoesnotevolvewithredshiftfromz∼2topresent.
This would indicate that this scenario for the growth of an-
gular momentum is plausible. However, we have to caution
5. Discussion
that current samples are relatively small and incomplete in
terms of galaxy mass. Moreover, there are some issues with
5.1.TheGrowthofAngularMomentuminGalaxies
the physical resolution of the data, especially at z>2 (but see
withTime
Nesvadbaetal. 2006). Even though these results are all for-
Weshowedin§3thatz∼0.6andlocaldiskofsametotalmass mallyconsistentwithnooverallevolutioninthespecificangu-
show comparablespecific angular momentum.More dynami- larmomentum,giventhelimitationsofallstudies,itisdifficult
callycomplexobjectsfallaroundthesame j −V relation to concludeanythingaboutevolutionarytrendsin thespecific
disk max
but with much larger dispersion than the local relation. How angularmomentuminrotatingdiskgalaxies.
8 Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies
orientationofitsangularmomentumvectorrelativetotheper-
hapspre-existingdisk,etc.
Returning to both Figs. 3 and 4, it is clear that perturbed
rotation and kinematically complex galaxies are dynamically
differentfromdynamicallyrelaxedlocaldisks.Thedispersions
inthe j -V areindeedsimilarfortherotatingdisks,higher
disk max
for the perturbedrotators,andveryhighfor the galaxieswith
complex dynamics. We can also note that some of the dis-
tant galaxieswhich are notrotating disks have offsets to both
lower and higher j relative to local galaxies for the same
disk
V .Becausewearelikelyunderestimating j inkinemati-
max disk
callycomplexgalaxies(see§3.1),wecannotconcludespecif-
ically about any difference on median j between the dif-
disk
ferent kinematical classes. However, the higher dispersion in
the V -j plane of kinematically complex galaxies rela-
max disk
tively to distant rotating disks is significant. This dispersion
is caused by abrupt variations of the angular momentum in
galaxieswith complexkinematics,andare likely the resultof
a random walk during a non-linear phase of evolution of the
angularmomentum.Thisisthefirstobservationalevidencefor
such a non-lineargrowth of the angular momentum in galax-
ies, as expectedfromtheoreticalmodels(e.g.,Gardner 2001;
Fig.6. Comparison of V/σ and ellipticity, ǫ, for galaxies at
Vitvitskaetal.2002;Peiranietal.2004).
z∼0.6 compared to a matched sample of local galaxies. The
Vmax/σratioforgalaxiesatz∼ 0.6thathavebeendetermined Animportantquestioniswhatisthedrivingmechanismof
toberotatingdisks(bluedots)andtohaverotationcurvesthat this random walk. Theoreticalmodels show that major merg-
areperturbed(greensquares)arecomparedwithlocalgalaxies erscausethemostabruptvariationsoftheangularmomentum,
whosemorphologyandvelocitycurvesareconsistentwithro- while minor mergersand/orgas accretion are associated with
tatingdisks(blackcrosses).The mediansareindicatedbythe smoother variations (e.g., Vitvitskaetal. 2002; Peiranietal.
positionofthemedianuncertaintiesontherightsideofthedi- 2004; Hetznecker&Burkert2006). If the gasis heatedto ap-
agram(eachbeingrepresentedbymatchingcolors) proximatelythevirialtemperatureofthehalobeforecollapsing
. toformorontoapre-existingdisk(Dekel&Birnboim 2006),
we would expect the gas to have a specific angular momen-
tum similar or greater than that of the dark matter halo (e.g.,
Beyond this, the results presented here can constrain Chenetal.2003;Okamotoetal.2005).Thisiswhywemight
how angular momentum and indeed how galaxies grow af- thenexpectthatthespecificangularmomentumofthediskto
ter the epoch when massive halos (like the ones studies remainconstantortomildlyincreaseordecreaseatlatetimes
here) acquired their angular momenta through tidal torquing. (Peiranietal. 2004). It is hard to believe that smooth gas ac-
The basic ideas that have emerged from theoretical research cretionalonecouldexplainthecomplexorperturbedkinemat-
are that the specific angular momentum can be increased icsand/ordiscrepantvaluesofthespecificangularmomentum.
or decreased in a halo through a random walk process of Weinsteadsuggestthatmergers(bothmajorandminormerg-
interactions and merging with other halos (e.g., Gardner ers,possiblyassociatedwithcomplexkinematicgalaxies,and
2001; Vitvitskaetal. 2002; Porcianietal. 2002) or through perturbedrotations,respectively,seePaperIandII)playasig-
gas accretion onto the halo (e.g., White&Rees 1978). In nificant role in changing the angular momentum of galaxies
the merger process, angular momentum is gained or lost with time. While overall consistent with the tidal torque the-
depending on the geometry, dynamics, the detailed mass ory,theincreaseddispersioninspecificangularmomentumand
distribution of the merging halos, strength of the feed- spinoftheperturbedrotatorsandgalaxieswithcomplexkine-
back, the relative masses of the merging halos and galaxies, matics is consistent with the merger scenario. Vitvitskaetal.
etc. (e.g., Robertsonetal. 2005; Springel&Hernquist 2005; (2002)showthatwithinthecontextofmerging,wewouldfind
D’Onghiaetal. 2006). Alternatively, the angular momentum both increases and decreases in the specific angular momen-
of disk galaxies could grow through the accretion of cooling tum,resultinginanincreaseofthedispersionoftheirangular
halogasorsmoothaccretionofmassintothedarkmatterhalo momenta. This is consistent with what we observe when we
(e.g., White&Rees 1978; Peiranietal. 2004; vandenBosch comparedistantrotatingdisksto galaxieswith complexkine-
2001, 2002a; vandenBoschetal. 2002b, 2003; Chenetal. matics. This interpretation is strongly supported by a simula-
2003; Okamotoetal. 2005). The amount of angular momen- tion of GIRAFFE observations of a major merger. This sim-
tumchangethendependsontherelativeangularmomentumof ulation illustrates how the higher dispersion of kinematically
thegasanddarkmatter,howthegaswasaccreted,howmuch complexgalaxiescan arise from mergers,taking into account
mechanicaldissipation the gasundergoesduringcollapse,the allpossibleobservationaluncertainties.
Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies 9
5.2.TheBuildingofLocalDiskGalaxies Given the above discussion (see § 5.1) about the specific
angular momentum,it is unlikely that (smooth) gas accretion
In the following discussion, we will assume that most of the aloneplaysanimportantroleinthegeneralgrowthofrotating
observed distant galaxies with emission lines are progenitors disks.Itisalsoequallyunlikelythatgasaccretionalonecould
of localdisks. Thisshouldbe the case evenformostgalaxies explain the disturbed kinematicsof perturbedrotation, as this
showing complexkinematicsand perturbedrotations. Indeed, processisexpectedtobedynamicallysmooth(seePaperIand
thosegalaxiesrepresent40%ofgalaxiesatz∼0.6,andif they II) and only lead to small changesin the angularmomentum.
wereE/S0progenitors,therewouldbeamuchhigherfraction Note,however,that,toourknowledge,thereisnocleartheoret-
ofE/S0thatitisobservedtoday(see,e.g.,Hammeretal.2005, icalpredictionaboutthekinematicsofgasaccretionbygalax-
andalsoLotzetal.2006). ies.Thus,ourresultsshowthatitislikelythatmergers(and,a
priori, both major and minormergers)play an importantrole
Because we are comparing samples of galaxies span-
inthegeneralbuild-upofrotatingdisks.Thelinkbetweendy-
ning different total mass (V ), linking distant to local
max namicalprocesses such as mergers,and the dynamicalclassi-
disks from Figure 4 alone is not straight-forward. Can z∼0.6
ficationsadoptedinthisseriesofpaperwillbeaddressedina
galaxies evolve towards local disks through major merg-
forthcomingarticle.
ers? Hammeretal. (2005) claim that 75% of local spiral in
Theresultspresentedinthispaperfitquitenicelywithinthe
the 1010.5-1011.5M range (the so-called “intermediate-mass”
⊙ “spiralrebuilding”scenarioproposedbyHammeretal.(2005),
range, see also Hammeretal. 2006) have experienced a ma-
where∼75±25%oflocalspirals(thoseofearlytype)haveun-
jor merger since z=1. Assuming an evolution rate as (1+z)2.7
dergonea mergersince z=1 and have rebuilt a disk thanksto
(LeFe`vreetal. 2000), one can derive that ∼ 29% of local
gas accretion. This scenario is composed of 3 major phases:
spiral galaxies have experienced a major merger since z=0.6.
a “pre-mergerphase”duringwhichtwo distantspiralsmerge,
Similarly, Lotzetal. (2006) estimated that between 33% and
the “LCG phase” where all material from the progenitorsfall
66%of L ≥ 0.4L∗ galaxieshad a major mergersince z=1.1
B B into the mass barycenterof the system and form a bulge, and
with an evolution rate of (1+z)1.12. Combining the above es-
the “disk growing phase” where subsequently accreted mate-
timates implies that ∼ 15 to 30% of local spirals could have
rial forms a rotating disk (see Fig. 8 of Okamotoetal. 2005,
experienced a major merger since z=0.6. Then, the majority
for how this might look). As the two progenitors are merg-
of z∼0.6 rotating disks cannot evolve towards local spirals
ing, their disks, and thus, their spin angular momentum, are
throughmajormergers.Anotherpossibility,isthatz∼0.6galax-
destroyed during the collision (Coxetal. 2004). At the same
iescouldevolvedtoz=0spiralsthroughminormergersand/or
time, the encounterhas a significantquantity of orbital angu-
gas accretion: their rotating disks would then have survived,
larmomentumwhichcanbeprogressivelyconvertedintospin
and their specific angular momentum should statistically in-
angular momentum as a new disk is re-built around the rem-
crease by only ∼ 0.1 dex (Peiranietal. 2004). A third possi-
nant.Moreover,wesawthattheincreaseofV/σindiskgalax-
bility is that distant disks would evolve in a “closed box”. In
ies could be due to heating mechanismssuch as minor merg-
absence of externaltorques,their specific angularmomentum
ersandgasaccretionevents,asifsomeofthedisk-rebuilding
wouldthenremainconstant.Finally,thelasttwopossibleevo-
weretriggeredbyaccretionofthegasleftoverfromthemerg-
lutiontracksfromz∼0.6toz=0areequallyviable.
ingevent.Yoachim&Dalcanton(2006)suggestthattheprop-
This picture is also supported by the fact that distant ertiesofthickandthindisksareconsistentwithgasrichmerg-
disks (we consider here both rotating disks and perturbed ersplayingasignificantroleintheirformationwherethestars
rotating disks together) seem to be heated relatively to lo- in these mergers formed the thick disk while the settling gas
cal disks (i.e., have systematically higher V/σ), possibly formedmuchof the thin disk. The results presentedhere am-
through minor mergers and/or gas accretion (Walkeretal. plifytheseideasandshowtheircredibilityindirectlyexplain-
1996; Velazquez&White 1999; Abadietal. 2003). A diffi- ingthedynamicsofintermediateredshiftdiskgalaxies.
culty to quantify uncertainty in this result could arise from
the difference observational strategies used in the local and
6. Conclusions
distant samples (slit vs. integral field spectroscopy, respec-
tively).This,however,isunlikelygiventhefactthatmostspi- We have studied the angular momentum and the dynamical
rals today are stronglydominated by rotationalmotions(e.g., supportofa sampleof z∼0.6galaxiesobservedwith the inte-
Binney&Merrifield 1989).An interestingadditionalcompo- gralfieldspectrographFLAMES/GIRAFFE.Wefoundthatthe
nent of the evolution in V /σ we observed is the possi- classificationofdistantgalaxiesbasedontheirkinematicprop-
max
bility of “angular momentum mixing” during minor mergers erties (mainly) and morphologies into three distinct classes,
and/orgasaccretionevents(seeOkamotoetal.2005).During i.e.,rotatingdisks,perturbedrotators,andkinematicallycom-
suchevents,thedirectionoftheangularmomentumvectorcan plex, is apparently robust. This classification appears to also
change, and newly accreted gas may then settle to a differ- selectgalaxieswithangularmomentaconsistentwithlocalspi-
entorientationfromthepre-existingdisk.Duringthisprocess, ralgalaxiesbutshowvaryingdegreesofdispersionrelativeto
theobservationalmanifestationsarelikelytoberelativelyhigh thelocalvalues,increasingfromrotatingdiskstokinematically
velocity dispersion of the gas caused by the gas being shock complexgalaxies.
heatedbyturbulencegeneratedbytheoverlappinganddynam- Thiscanbeinterpretedasanevidenceforanon-linearran-
icallydifferentdisks(whatwecouldobserveinFigure6). domwalk evolutionofthe angularmomentumduringthe last
10 Puechetal.:AngularMomentumofIntermediateRedshiftGalaxies
8 Gyr. A natural driver for this random walk is provided by Bo¨hm,A.,etal.2004,A&A,420,97
majormergers,since the dispersionofkinematicallycomplex Bottema,R.1993,A&A,275,16
galaxiesinthe j −V plane,aswellasthecomplexityof Bullock,J.S.,Dekel,A.,Kolatt,T.S.,Kravtsov,A.V.,Klypin,
disk max
their velocity fields itself (see Puech et al. 2007, in prepara- A.A.,Porciani,C.,&Primack,J.R.2001,ApJ,555,240
tion),can bothbe reproducedby hydrodynamicalsimulations Chen,D.N.,Jing,Y.P.,&Yoshikaw,K.2003,ApJ,597,35
of such events. Major mergers also explains how the angular Conselice, C. J., Bundy,K., Ellis, R. S., Brichmann,J., Vogt,
momentumoflocalrotatingdiskscouldbeacquiredandbeas N.P.,&Phillips,A.C.2005,ApJ,628,160
highasobserved,namelythroughtheconversionoforbitalan- Corsini, E. M., Pizzella, A., Coccato, L., & Bertola, F. 2003,
gular momentum to spin momentum. Moreover,distant disks A&A,408,873
appearto be moreturbulent(lowerV/σ).This is likelyan in- Courteau,S.1997,AJ,114,2402
dication that local disks could grow through the accretion of Cox,T.J.,Primack,J.,Jonsson,P.,&Somerville,R. S.2004,
gas through mergers or in discrete clouds. In a refinement of ApJ,607,L87
this general picture, the higher velocity dispersions could be Cox, T. J., Jonsson,P., Primack P., & Somerville,R. S. 2006,
dueto“angularmomentummixing”(seeOkamotoetal.2005) MNRAS,373,1013
wherebythe relative orbitof the infallinggasisskewed com- D’Onghia,E.,Burkert,A.,Murante,G.,&Khochfar,S.2006,
paredtothepreviousdisk.Thetorqueprovidedbytheaccreting MNRAS,submitted(astro-ph/0602005)
gascausesa changein the angularmomentumvectorthereby Davies, R. L., Efstathiou, G., Fall, S. M., Illingworth, G., &
increasingthedispersionobservedinthegas. Schechter,P.L.1983,ApJ,266,41
Finally,thesefindingsareconsistentwiththeobservational Dekel A. & Birnboim Y., 2006, MNRAS, in press
scenarioproposedbyHammeretal.(2005),where∼75%oflo- (astro-ph/0412300)
calspirals(thoseofearlytype)haveundergoneamajormerger Drory, N., Salvato, M., Gabasch, A., Bender, R., Hopp, U.,
sincez=1,andhavethenrebuilttheirdisksthankstogasaccre- Feulner,G.,&Pannella,M.2005,ApJ,619,L131
tion(possiblyfromhighangularmomentumgasleftoverfrom Dutton,A.A.,vandenBosch,F.C.,Dekel,A.,&Courteau,S.
themergingevent)and/orminormergers.Unfortunately,draw- 2006,ApJsubmitted(astro-ph/0604553)
ingrobustconclusionsastothephysicalprocessesdrivingthe Eisenstein,D.J.,&Loeb,A.1996,ApJ,459,432
observationalmanifestationswehavediscussedinthispaperis Fall,S.M.,&Efstathiou,G.1980,MNRAS,193,189
difficultgiventhesmallnumbersofgalaxieswe havestudied. Flores, H., Hammer,F., Puech, M., Amram,P., & Balkowski,
To increase the statistical robustness of these results, we are C.2006,A&A,inpress(astro-ph/0603563,PaperI)
currently analyzing similar data for a much larger sample of Fo¨rsterSchreiberetal.2006,ApJ,inpress(astro-ph/0603559)
severalhundreddiskgalaxiesaspartoftheVLTLargeProgram Gardner,J.P.2001,ApJ,557,616
IMAGES(P.I.:F.Hammer).Thegalaxiesareselectedbyabso- Gilmore, G., Wyse, R. F. G., & Norris, J. E. 2002, ApJ, 574,
lute J-band magnitude, to have redshifts of or less than 0.9, L39
andtohave[OII]equivalentwidthcomparabletothegalaxies Governato,F.,Willman,B.,Mayer,L.,Brooks,A.,Stinson,G.,
studyhere (Ravikumaretal. 2006). With the robustselection Valenzuela,O.,Wadsley,J.,&QuinnT.2006,MNRAS,sub-
andlargenumbers,weshouldbeabletomakemoredefinitive mitted(astro-ph/0602351)
statements aboutthe physicalbehind the dynamicaland mor- Hammer,F.,Gruel,N.,Thuan,T.X.,Flores,H.,&Infante,L.
phologicalevolutionofspiralgalaxiesoverthelast7Gyrs. 2001,ApJ,550,570
Hammer,F.,Flores,H.,Elbaz,D.,Zheng,X.Z.,Liang,Y.C.,
Acknowledgements. WewouldliketothankM.Steinmetzforhaving
&Cesarsky,C.2005,A&A,430,115
providinguswiththeircompilationofdataonthekinematicsoflocal
Hammer,F.,Lehnert,M.D.,Puech,M.,Flores,H.,Liang,Y.C.
galaxies,andforinterestingcommentsonrecentsimulationwork.We
2006,Messenger,123,41
are especially indebted to T.J. Cox who provided us with hydrody-
Heavens,A.F.,&Jimenez,R.1999,MNRAS,305,770
namicalsimulationsofaSbcgalaxy.WealsowouldliketothankC.
Balkowski,P.Amram,andL.Cheminforveryinterestingdiscussions Hernandez,X.,&Cervantes-Sodi,B.2006,MNRAS,368,351
wehadaboutthesubjectofthispaper. Hetznecker,H.,&Burkert,A. 2006,MNRAS,370,1905
Hoyle, F. 1951, in Problems of Cosmical Aerodynamics,
Proceedings of the Symposium on the Motion of Gaseous
References
Masses of Cosmical Dimensions held in Paris, France,
Abadi,M.G.,Navarro,J.F.,Steinmetz,M.,&Eke,V.R.2003, August 1949, ed. J. M. Burgers & H. C. van de Hulst
ApJ,591,499 (Dayton:CentralAirDocumentsOffice),195
Barnes,J.,&Efstathiou,G.1987,ApJ,319,575 Hunter, D. A., Rubin, V. C., Swaters, R. A., Sparke, L. S., &
Bender,R.,Saglia,R.P.,&Gerhard,O.E.1994,MNRAS,269, Levine,S.E.2005,ApJ,634,281
785 Keresˇ, D., Katz, N., Weinberg, D. H., & Dave´, R. 2005,
Binney,J.2005,MNRAS,363,937 MNRAS,363,2
Binney, J. & Merrifield, M. Galactic Astronomy, Princeton LeFe`vre,O.,etal.2000,MNRAS,311,565
UniversityPress(Princeton,NJ,USA),1989. Lemson,G.,&Kauffmann,G.1999,MNRAS,302,111
BlechaA.,CayatteV.,NorthP.etal.,OpticalandIRTelescope Lotz, J. M., Madau, P., Giavalisco, M., Primack, J., &
Instrumentation and Detectors, Masanori Iye & Alan F. Ferguson,H.C.2006,ApJ,636,592
MoorwoodEds.,2000,SPIEproc.vol.4008,467