Genzel et al., 2011, ApJ, 733, 101 - MPE

30 Pages·2011·2.19 MB·English

TheAstrophysicalJournal,733:101(30pp),2011June1 doi:10.1088/0004-637X/733/2/101
(cid:2)C2011.TheAmericanAstronomicalSociety.Allrightsreserved.PrintedintheU.S.A.
THESINSSURVEYOFz∼2GALAXYKINEMATICS:PROPERTIESOF
∗
THEGIANTSTAR-FORMINGCLUMPS
R.Genzel1,2,S.Newman3,T.Jones3,N.M.Fo¨rsterSchreiber1,K.Shapiro3,4,S.Genel1,S.J.Lilly5,A.Renzini6,
L.J.Tacconi1,N.Bouche´7,15,A.Burkert8,G. Cresci9,P.Buschkamp1,C.M.Carollo5,D.Ceverino10,R.Davies1,
A.Dekel10,F.Eisenhauer1,E.Hicks11,J.Kurk1,D.Lutz1,C.Mancini6,T.Naab12,Y.Peng5,A.Sternberg13,
D.Vergani14,andG.Zamorani14
1Max-Planck-Institutfu¨rextraterrestrischePhysik(MPE),Giessenbachstr.1,D-85748Garching,Germany;[email protected]
2DepartmentofPhysics,LeConteHall,UniversityofCalifornia,Berkeley,CA94720,USA
3DepartmentofAstronomy,CampbellHall,UniversityofCalifornia,Berkeley,CA94720,USA
4AerospaceResearchLaboratories,NorthropGrummanAerospaceSystems,RedondoBeach,CA90278,USA
5InstituteofAstronomy,DepartmentofPhysics,Eidgeno¨ssischeTechnischeHochschule,ETHZu¨rich,CH-8093,Switzerland
6OsservatorioAstronomicodiPadova,Vicolodell’Osservatorio5,Padova,I-35122,Italy
7DepartmentofPhysics&Astronomy,UniversityofCalifornia,SantaBarbara,SantaBarbara,CA93106,USA
8Universita¨ts-SternwarteLudwig-Maximilians-Universita¨t(USM),Scheinerstr.1,Mu¨nchen,D-81679,Germany
9IstitutoNazionalediAstrofisica–OsservatorioAstronomicodiArcetri,LargoEnricoFermi5,I–50125Firenze,Italy
10RacahInstituteofPhysics,TheHebrewUniversity,Jerusalem91904,Israel
11DepartmentofAstronomy,UniversityofWashington,Box351580,U.W.,Seattle,WA98195-1580,USA
12Max-PlanckInstituteforAstrophysics,KarlSchwarzschildstrasse1,D-85748Garching,Germany
13SchoolofPhysicsandAstronomy,TelAvivUniversity,TelAviv69978,Israel
14INAFOsservatorioAstronomicodiBologna,ViaRanzani1,40127Bologna,Italy
Received2010November24;accepted2011February9;published2011May11
ABSTRACT
Wehavestudiedthepropertiesofgiantstar-formingclumpsinfivez∼2star-formingdiskswithdeepSINFONI
AOspectroscopyattheESOVLT.TheclumpsresideindiskregionswheretheToomreQ-parameterisbelowunity,
consistent with their being bound and having formed from gravitational instability. Broad Hα/[Nii] line wings
demonstratethattheclumpsarelaunchingsitesofpowerfuloutflows.Theinferredoutflowratesarecomparableto
orexceedthestarformationrates,inonecasebyafactorofeight.Typicalclumpsmayloseafractionoftheiroriginal
gasbyfeedbackinafewhundredmillionyears,allowingthemtomigrateintothecenter.Themostactiveclumps
may lose much of their mass and disrupt in the disk. The clumps leave a modest imprint on the gas kinematics.
Velocity gradients across the clumps are 10–40kms−1kpc−1, similar to the galactic rotation gradients. Given
beam smearing and clump sizes, these gradients may be consistent with significant rotational support in typical
clumps.Extremeclumpsmaynotberotationallysupported;eithertheyarenotvirializedortheyarepredominantly
pressuresupported.Thevelocitydispersionisspatiallyratherconstantandincreasesonlyweaklywithstarformation
surfacedensity.Thelargevelocitydispersionsmaybedrivenbythereleaseofgravitationalenergy,eitherattheouter
disk/accretingstreamsinterface,and/orbytheclumpmigrationwithinthedisk.Spatialvariationsintheinferred
gasphaseoxygenabundancearebroadlyconsistentwithinside-outgrowingdisks,and/orwithinwardmigration
oftheclumps.
Keywords: cosmology:observations–galaxies:evolution–galaxies:high-redshift–infrared:galaxies
Online-onlymaterial:colorfigures
1.INTRODUCTION ent rotation, especially among the more massive (M∗ (cid:2) a few
1010M(cid:5))andbright(KsAB (cid:3)21.8)systems(Fo¨rsterSchreiber
The rest-frame UV/optical morphologies of most z > 1 et al. 2006, 2009; Genzel et al. 2006, 2008; Weiner et al.
“normal” star-forming galaxies (henceforth “SFGs”; Steidel 2006; Wright et al. 2007; Law et al. 2007, 2009; Shapiro
et al. 1996, 2004; Franx et al. 2003; Noeske et al. 2007; et al. 2008; Bournaud et al. 2008; Cresci et al. 2009; van
Daddi et al. 2007; Cameron et al. 2010) are irregular and of- Starkenburgetal.2008;Epinatetal.2009;Lemoine-Busserolle
ten dominated by several giant (kpc size) star-forming clumps & Lamareille 2010). These kinematic studies also find that
(Cowie et al. 1995; van den Bergh et al. 1996; Elmegreen high-z SFGs as a rule exhibit large local velocity dispersions
et al. 2004, 2007, 2009; Elmegreen & Elmegreen 2005, 2006; oftheirionizedgascomponent,withratiosofrotationvelocity
Fo¨rster Schreiber et al. 2009, 2011). These clumpy, asym- v tolocalintrinsicvelocitydispersionσ rangingfrom1to6.
c 0
metric structures often resemble z ∼ 0 mergers (Conselice Observations of CO rotational line emission indicate that z ∼
et al. 2003; Lotz et al. 2004). However, spatially resolved 1–3SFGshavelarge(∼30%–80%)baryoniccoldgasfractions
studies of the ionized gas kinematics of these clumpy galax- (Daddietal.2008,2010a;Tacconietal.2008,2010).
ies find a surprisingly large abundance of disks with coher- Thesebasicobservational propertiescanbeunderstood ina
simple physical framework, in which gravitational instability
∗ BasedonobservationsattheVeryLargeTelescope(VLT)oftheEuropean and fragmentation in semi-continuously fed, gas-rich disks
SouthernObservatory(ESO),Paranal,Chile(ESOprogramIDs076.A-0527, naturallyleadstolargeturbulenceandgiantstar-formingclumps
079.A-0341,080.A-0330,080.A-0339,080.A-0635,183.A-0781).
15SupportedbytheMarieCuriegrantPIOF-GA-2009-236012fromthe (Noguchi 1999; Immeli et al. 2004a, 2004b; Bournaud et al.
EuropeanCommission. 2007; Elmegreen et al. 2008; Genzel et al. 2008; Dekel et al.
1
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
2009a; Bournaud 2010). A more detailed discussion of these Schreiber et al. (2006) proposed that the gravitational energy
instabilities follows in Section 2.4, where we show that gas- released by the accreting gas (including minor mergers) at
rich,marginallystabledisksshouldhavemuchlargerandmore the interface of the cold streams and the disk may trigger
massivestar-formingcomplexesthanthoseinz∼0SFGsand thelargerandommotions.Asimilarexplanationisfavoredby
that these complexes should be located in regions where the Genzeletal.(2008)andKhochfar&Silk(2009),whileDekel
valueoftheToomre(1964)Q-parameterisbelowunity. et al. (2009a) argue that smoother-than-average streams may
Themostrecentgenerationofcosmologicalgalaxyevolution notbeabletodrivealargelocalvelocitydispersionbutinother
models and simulations finds that the buildup of z > 1 SFGs casesaccretionfromthehalomightdrivethediskintostability
in the mass range of 1010 to 1011M(cid:5) is dominated by smooth (Q > 1). Instead, Immeli et al. (2004a, 2004b), Dekel et al.
accretionofgasand/orminormergers(Keresˇetal.2005,2009; (2009a), and Ceverino et al. (2010) all favor gravitational
Dekel & Birnboim 2006; Bower et al. 2006; Kitzbichler & torquesinthediskandcollisionsbetweenthegiantclumps,or
White2007;Ocvirketal.2008;Dave´2008;Dekeletal.2009b; acombinationofthegravitationaltorquesandstellarfeedback
Oser et al. 2010). In contrast, the overall cosmological mass (Elmegreen & Burkert 2010) as the main drivers of the turbu-
assemblyofgalaxies,especiallyofthemostmassiveonesand lence.Ifthemaindriverofthelargevelocitydispersionsisstellar
at late times, is probably dominated by mergers (Bower et al. feedback,andspecificallyradiationpressureondustgrains,one
2006;Kitzbichler&White2007;Naabetal.2007,2009;Guo mightexpectacorrelationoftheamplitudeofturbulencewith
& White 2008; Dave´ 2008; Genel et al. 2008). The large and star formation rate or surface density (Fo¨rster Schreiber et al.
semi-continuous gas accretion in these “cold flows” or “cold 2006;Genzeletal.2008;Murrayetal.2010a).
streams”mayrapidlybuildupgalaxydisks(Dekeletal.2009b; In this paper we present and analyze new high-quality
Ocvirk et al. 2008; Keresˇ et al. 2009; Oser et al. 2010). If SINFONI/VLT integral field (IFU) spectroscopy (Eisenhauer
the incoming material is gas-rich, then violent gravitational et al. 2003; Bonnet et al. 2004) of five luminous, clumpy
instabilitiesinthesediskscouldleadtothelargestarformation z ∼ 2 SFGs. We employed both laser guide star (LGS) and
ratesderivedfromobservations(Geneletal.2008;Dekeletal. natural guide star (NGS) adaptive optics (AO) to improve the
2009a).Thegiantclumpsareexpectedtomigrateintothecenter angular resolution to an effective ∼0(cid:7).(cid:7)2 FWHM. For all of the
viadynamicalfrictionandtidaltorquesonatimescaleof targets, the quality of the derived spectra is much superior to
previous data, because of long integration times (9 to 19 hr in
tin-spiral ≈(vc/σ0)2tdyn(Rdisk)∼10tdyn(Rdisk)<0.5Gyr, (1) four of the targets) and/or the high surface brightness of the
selectedclumpygalaxies.Withthesedataitisnowpossible,for
wheretheymayformacentralbulgeandaremnantthickdisk
thefirsttime,tostudydetailedlineprofilesonthescaleofthe
(Noguchi 1999; Immeli et al. 2004a, 2004b; Fo¨rster Schreiber
mostmassiveandlargestclumps(1–3kpc).Ourmeasurements
et al. 2006; Genzel et al. 2006, 2008; Elmegreen et al. 2008;
deliver interestingnew constraints onthekinematic properties
Carollo et al. 2007; Dekel et al. 2009a; Bournaud et al. 2009;
andlifetimesofthegiantclumps.WeadoptaΛCDMcosmology
Ceverinoetal.2010).
with Ω = 0.27, Ω = 0.046, and H = 70kms−1 Mpc−1
The efficacy of the “violent disk instability” for forming m b 0
(Komatsuetal.2010),aswellasaChabrier(2003)initialstellar
bulges by the in-spiral of the giant clumps hinges on the
massfunction(IMF).
survival of the clumps in the presence of outflows driven by
stellar winds, supernovae, and radiation pressure, even if sec-
ular bulge growth may also occur directly from the disk 2.OBSERVATIONSANDANALYSIS
without clump migration. This “star formation feedback” is
2.1.SourceSelection,Observations,andDataReduction
widely thought to be a key ingredient in the evolution of star-
forming galaxies (Dekel & Silk 1986; Kauffmann et al. 1993; By selection, the five galaxies we discuss in this paper
Finlator & Dave´ 2008; Efstathiou 2000; Bouche´ et al. 2010; are massive (vc ∼ 250kms−1, M∗ ∼ 1010.6M(cid:5), Mdyn (R ∼
Dutton et al. 2010). Local universe giant molecular clouds 10kpc) (cid:2) 1011M(cid:5)), luminous (star formation rates (SFR) ∼
(aGtiMmCessc)aalreetproneto∼raMpidexp/uMl˙sion,wofhgicahsbpyrotbhaisblfyeeddibssaicpkatoens 31–260.–52k9p0cM).(cid:5)Tyhre−y1)samanpdle tfhaeirluypplearrgreang(eRdoisfkma(HssWaHndMb)olo∼-
expulsion clump out
GMCsonatimescaleofafewtensofmegayears(Murray2010). metric luminosity of the z ∼ 2 SFG “main sequence” (Fo¨rster
High-z clumps may live longer because their ratio of gravita- Schreiber et al. 2009; Noeske et al. 2007; Daddi et al. 2007).
tionalbindingenergytostarformationrateis∼100timeslarger In this subset of the z ∼ 2 SFG population, data cubes with
than in the local universe (Dekel et al. 2009a). Exactly how integrationtimesafewto20hrpergalaxyhavesufficientsig-
stablethehigh-zclumpsareandhowlargetheirgasexpulsion nal to noise ratio (S/N) in a sufficient number of independent
timescalesmightbeisamatterofcurrentdebate.Krumholz& pixels(N ∼50–200)toextractthedetailedpropertiesofgi-
pix
Dekel (2010) find that the high-z clumps only lose a modest ant clumps, each of which have intrinsic FWHM diameters of
fraction(<50%)oftheiroriginalmassbyfeedbackaslongas 0(cid:7).(cid:7)15–0(cid:7).(cid:7)3(Genzeletal.2008).
thestarformationefficiency perfree-falltimedoes notsignif- As part of the SINS GTO survey (Fo¨rster Schreiber et al.
icantlyexceedafewpercent(similartolocalSFGs;Kennicutt 2009) and the SINS/zCOSMOS ESO Large Program (see
1998a). Murray et al. (2010a) and Genel et al. (2010) argue Mancini et al. 2011) of high-z galaxy kinematics carried out
thatthemajorityoftheclumps’initialgasmassisexpelledby with SINFONI at the VLT, we observed the Hα, [Nii], and
feedbackintheformofmomentum-drivenwinds. [Sii]emissionlinesintherest-frameUV-selectedSFGsQ1623-
While it is plausible that the very active high-z SFGs BX599 (z = 2.332) and Q2346-BX482 (z = 2.258; Erb et al.
are naturally driven toward marginal gravitational instability 2006b;Fo¨rsterSchreiberetal.2006,2009),andintherest-frame
(Q∼1)byself-regulation(Quirk1972;Gammie2001;Thomp- optically selected SFGs D3a15504 (z = 2.383), ZC782941
sonetal.2005),thedominantagentsresponsiblefortherequired (z = 2.182; also called ZC407302 in Mancini et al. 2011),
(and observed) high-velocity dispersions are not known, and and ZC406690 (z = 2.196) (Kong et al. 2006; Genzel et al.
possibly multi-factorial (Krumholz & Burkert 2010). Fo¨rster 2006,2008;Fo¨rsterSchreiberetal.2009;Mancinietal.2011;
2
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
Y. Peng et al. 2011, in preparation). The two rest-UV-selected Uncertainties of all fitted parameters are calculated through
sources were photometrically identified in optical imaging by 100 Monte Carlo simulations in which the spectrum of each
their U GR colors (satisfying the “BX” criteria), their redshift spatial pixel is perturbed assuming a Gaussian distribution of
n
confirmed from optical spectroscopy, and first observed in the the rms from the input noise cube. The final integrated line
near-IR with the long-slit spectrometer NIRSPEC on Keck II intensity, velocity, and velocity dispersion maps were then
(Steidel et al. 2004; Adelberger et al. 2004; Erb et al. 2006b). multipliedbyamaskconstructedfromallpixelswithHα line
Therest-optically-selectedtargetswereidentifiedbasedonK - emission at >3σ significance. We compare the line emission
s
bandimagingandviathe“BzK”colorcriterionfor1.4<z< maps to similar resolution (∼0(cid:7).(cid:7)15–0(cid:7).(cid:7)25 FWHM) images of
2.5 star-forming galaxies (Daddi et al. 2004), and followed- the rest-frame UV/optical stellar continuum. In the case of
up with VLT/VIMOS optical spectroscopy to confirm their BX482, we use the Hubble Space Telescope (HST)/NIC2
redshift (Kong et al. 2006; Lilly et al. 2007). Prior to the H-bandimage(throughtheF160Wfilter)ofFo¨rsterSchreiber
SINFONIobservations,noneofthemhadnear-IRspectroscopic etal.(2011a).ForZC782941andZC406690,weusetheHST/
data. ZC782941 and ZC406690 were moreover specifically Advanced Camera for Surveys (ACS) I-band (F814W filter)
drawnfromthe1.7deg2zCOSMOSspectroscopicsurvey(Lilly imagestakenaspartoftheCOSMOSsurvey(Koekemoeretal.
et al. 2007) to be located within 30(cid:7)(cid:7) of G < 16 mag stars 2007). For D3a15504 we have taken and analyzed a 2 hr AO-
suitable for Natural Guide Star adaptive optics (AO) assisted assistedexposureofthegalaxywithVLT/NACOintheK band,
s
observations. aspartofouroriginalSINSsurveyprogram(Fo¨rsterSchreiber
Thefivegalaxiesspantherangeofkinematicpropertiesfound etal.2009).
in the SINS survey of z ∼ 2 SFGs (Fo¨rster Schreiber et al.
2.2.ModelingoftheVelocityFields
2009). BX482 and ZC406690 are large clumpy, rotating disks
with a prominent ∼5kpc ring of star formation. D3a15504 We identified the most prominent clumps from maps of
is a large rotating disk with a central active galactic nucleus individual Hα velocities (“channels”) or, in the case of clump
(AGN).ZC782941isamorecompact,rotating,andasymmetric D in BX482, from the rest-frame optical continuum map. For
disk. The asymmetry is mainly caused by a compact clump identificationasaclump,werequiredthepresenceofanobvious
northofthemainbodyofthegalaxy,whichmaybeasecond, localmaximuminatleasttwoseparatevelocitychannelmaps.
lowermassgalaxyinteractingwiththemaingalaxy(a“minor” Figure 1 gives examples of such velocity channel maps for
merger). BX599 is an example of the compact “dispersion- D3a15504 (top row), BX482 (middle rows), and ZC782941
dominated” systems that tend to be common among less (bottom row), and marks the positions of the most prominent
massive,UV-selectedgalaxies(Erbetal.2006b;Lawetal.2007, clumps by circles/ovals and alphabetical symbols. Our list of
2009).However,ournewLGSAOSINFONIdatanowresolve clumps is meant to identify the brightest obvious clumps, and
BX599 spatially and reveal a substantial velocity gradient of isnotcompleteforthefainterclumpswhoseidentificationcan
150–200kms−1 across ∼3kpc. The observed ratio of half the bemoreambiguous.ForBX482(Figure2,middleleftcolumn),
velocity gradient to the integrated velocity dispersion Δvgrad/ ZC782941 (middle right column), ZC406690 (Figure 2, right
(2σint) ∼ 0.6. This is similar to several rotating disk galaxies column),andBX599(Figure9)thebrightestclumpsalsostand
intheSINSsurvey(Fo¨rsterSchreiberetal.2009).BX599may out in the velocity integrated Hα and the continuum maps. In
thusbeacompactrotatingdisk.Foramoredetaileddescription D3a15504andZC782941(clumpsB–E),someoftheclumpsare
of the SINS and SINS/zCOSMOS surveys, source selection, lessobviousorevenwashedoutintheintegratedmapsbecause
andglobalgalaxyproperties,werefertoFo¨rsterSchreiberetal. of diffuse integrated disk emission. We determined intrinsic
(2009)andMancinietal.(2011). HWHM clump radii from Gaussian fits to the appropriate
Table 1 summarizes integration times and the final FWHM velocitychannelsandsubtractedtheinstrumentalresolutionin
angular resolutions in these galaxies. For a description of squares.
the data reduction methods and analysis tools we refer to Inadditiontothebasicvelocityandvelocitydispersionmaps
Schreiber et al. (2004), Davies (2007), and Fo¨rster Schreiber obtainedfromLINEFIT,wealsoconstructed“residual”mapsby
et al. (2009). With the final data cubes in hand, we median- removingthelarge-scalevelocityfield.Forthispurposeweused
filtered the data by two spatial pixels and fitted Gaussian line “kinemetry”(Krajnovic´etal.2006;Shapiroetal.2008)orsim-
profiles to each pixel with the fitting code LINEFIT (Fo¨rster plerotatingdiskmodelsfittedtotheHαdata(Genzeletal.2006,
Schreiber et al. 2009). LINEFIT performs weighted fits to 2008; Cresci et al. 2009). The resulting velocity/dispersion
the observed line profiles as a function of the two spatial mapscapturethelarge-scalekinematics,whichcanthenbesub-
coordinates based on an input noise data cube and an input tractedfromtheLINEFITmaps,inordertomakelocalresiduals
spectral response function. The instrumental spectral response standoutmoreclearly.Forthepurposesoftheanalysispresented
function as obtained from OH sky lines is shown as a gray below,bothmethodsgiveindistinguishableresults.
dashed curve in the profiles shown in Figures 7–9. For the Toperformakinemetryanalysis,werequireknowledgeofthe
0(cid:7).(cid:7)05 × 0(cid:7).(cid:7)125 pixel scale in the K band we used here it is dynamicalcenter,positionangle,andinclinationofagalaxy.For
fit quite well by a Gaussian of FWHM ∼ 85kms−1 (green thehigh-S/Ndatapresentedhere,weareabletodeterminethe
curve in the upper left panel of Figure 9), with some excess dynamical centers directly from the shapes of the isovelocity
emissioninthelinewingsrelativetothisbestfittingGaussian. contours. Position angles and inclinations are estimated from
For the analysis of the line profiles in our program galaxies, the orientations of the maximum velocity gradients (line of
thesesmalldifferencesarenegligible,however.LINEFITtakes nodes) and the minor to major axis ratios of the line and
this instrumental line profile as inputs to compute intrinsic continuum emission. Using these inputs, we parameterize the
velocitydispersions.Likewise,thevelocitydispersionslistedin observed velocity fields as Fourier expansions along the angle
Table2andshowninFiguresA1andA2areintrinsicvaluesafter ϕ in the plane of the sky. Ideal, thin-disk rotation is described
removaloftheinstrumentalbroadening(andanybeam-smeared byacos(ϕ)term(seeShapiroetal.2008formoredetails).To
rotation). determinethehigher-order(local)variationsofthevelocityfield
3
T
h
e
A
s
t
r
o
p
h
y
s
ic
a
l
J
o
u
r
n
a
l
,
7
3
Table1 3
:1
ObservingLog 0
1
(3
Galaxy Band/PixelScale Mode FWHMResolution(arcsec) IntegrationTime,ObservingDate Reference 0
p
Q1623-BX599(z=2.332) K0(cid:7).(cid:7)05×0(cid:7).(cid:7)1 LGS 0(cid:7).(cid:7)23 2h00 Erbetal.2006b;Fo¨rster p),
2
2010Apr12–13 Schreiberetal.2009 0
Q2346-BX482(z=2.258) K0(cid:7).(cid:7)05×0(cid:7).(cid:7)1 LGS 0(cid:7).(cid:7)25 9h30 Erbetal.2006b;Genzeletal. 11
J
2007Oct27–29 2008;Crescietal.2009; un
e
2007Nov13–15 Fo¨rsterSchreiberetal.2009 1
2008Jul27–31
2009Nov11–13and17
D3a15504(z=2.383) K0(cid:7).(cid:7)05×0(cid:7).(cid:7)1 LGS,NGS 0(cid:7).(cid:7)18 18h40 Kongetal.2006;Genzel
2006Mar16–20 etal.2006;Geneletal.2008;
2009Apr30 Crescietal.2009;Fo¨rster
2009May1and16 Schreiberetal.2009
4 2009Jun16
2010Feb11–13
2010Mar9
2010Apr2
ZC782941(z=2.182) K0(cid:7).(cid:7)05×0(cid:7).(cid:7)1 NGS 0(cid:7).(cid:7)22 10h30 Genzeletal.2008;Cresci
(alsoZC407302) 2007Apr16–23 etal.2009;Fo¨rsterSchreiber
2009Apr18 etal.2009;Mancinietal.
2010Jan9and13 2011;Y.Pengetal.2011,in
2010Feb10 preparation
ZC406690(z=2.196) K0(cid:7).(cid:7)05×0(cid:7).(cid:7)1 NGS 0(cid:7).(cid:7)22 10h00 Mancinietal.2011;Y.Peng
2010Apr17 etal.2011,inpreparation
2010May25
2010Nov30,
2010Dec7,10,29,30,and31,
2011Jan2and3
G
e
n
z
e
l
e
t
a
l
.
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
Table2
DerivedGalaxyProperties
Source RowNumberBX599 BX482 D3a15504 ZC782941(ZC407302) ZC406690 ZC406690 ZC406690
All ClumpA ClumpsA-F ClumpA ClumpA ClumpB ClumpC
Z 1 2.33 2.26 2.38 2.18 2.2 2.2 2.2
DL(Gpc) 2 19.1 18.3 19.6 17.6 17.7 17.7 17.7
kpcarcsec−1 3 8.33 8.38 8.3 8.42 8.41 8.41 8.41
Fobs(Hα)1e-16(ergs−1cm−2) 4 3.3 0.35 0.04 0.2 1.4 0.57 0.4
A(Hα)a 5 0.73 1.1 1.8 2.1 1.1 1.1 1.1
L(Hα)0(ergs−1)b 6 2.8e43 3.8e42 9.4e41 5.2e42 1.4e43 5.7e42 4.0e42
SFR(M(cid:5)yr−1)c 7 66 12 3.3 17 40 11 14
Mmol−gasM(cid:5)c 8 3.3e10 7.8e9 3.0e9 8.7e9 1.6e10 7.8e9 9.6e09
Σmol−gas(M(cid:5)pc−2)c 9 4.4e3 2.1e3 6.9e2 4e3 8.4e3 1.4e3 1.8e3
RHWHM−intrc(kpc) 10 1.5 1 1 0.8 0.8 1.2 1.2
Σstar−form(M(cid:5)yr−1kpc−1)b 11 4.6 2.7 0.72 5.7 13.6 1.6 2.2
fbroad(=broad/total) 12 0.5(0.13) 0.32(0.08) 0.26(0.15) 0.31(0.1) 0.4(0.1) 0.6(0.1) (cid:2)0.25
[Sii]6718/6733 13 ... 0.7(0.2) 0.9(0.3) 1.1(0.35) 0.75(0.07) 1.09(0.1) ...
σclumpkms−1e 14 76(20) 62(3.4) 53(7) 95(7) 81(4) 88(4) 78(3)
n(e)clumpd(cm—3) 15 ... 2000(+∞,–1000)900(+2500,–700) 400(+1100,–350) 1500(+900,–400)420(+230,–140) ...
γredf 16 1 1.5 1.5 2 2 2 2
L(Hα)broad,0(ergs−1) 17 1.4e43 1.8e42 3.7e41 3.2e42 1.1e43 6.9e42 (cid:2)2.0e42
Δvmax(kms−1)g 18 1000 350 ∼400 420 440 810 ...
Mbroad(M(cid:5)) 19 4.5e8 6e7 1.2e7 1.1e8 3.6e8 2.2e8 (cid:2)6.5e7
dMout/dt(case1)h(M(cid:5)yr−1) 20 300 21 6 54 200 150 (cid:2)22
dMout/dt(case2)h(M(cid:5)yr−1) 21 68 5 1.4 13 46 34 (cid:2)5
dMout/dt(case3)h(M(cid:5)yr−1) 22 94 6.5 2.2 17.5 62 49 (cid:2)7
dMout1/2/dt/SFRi 23 2.8 1.0 1.1 2.0 3.1 8.4 (cid:2)0.9
texpulsionj(Myr) 24 360 1.2e3 1.6e3 520 265 170 <1.5e3
t∗(Myr)k 25 ... 30–100 >1e3 80–800 100–3e4 80–800
tZ(closed)(Myr)k 26 360 360 930 350 150 560 400
tZ(leaky)(Myr)k 27 920 480 1600 650 260 2e4 510
texpansion(Myr)k 28 120 310 360 140 86 120 ...
tdiss/torbit 29 10 12 14 7 1.7 1.1 (cid:3)10
M∗,final/Mgas,0l 30 0.27 0.49 0.48 0.34 0.25 0.11 (cid:3)0.52
Δv/(sini2Rclump)(kms−1kpc−1)m 31 ... 19(–10) 30(±12) 42(10) 20(–30) 60(+30) 30(+15)
4.4Mdyn−rot(M(cid:5))n 32 ... 4.3e8 1.1e9 1.1e9 2.2e8 2.1e9 2.1e9
Mmol−gas/4.4Mdyn−rot 33 ... 18 3 8 75 4 5
Mdyn−press(M(cid:5))o 34 ... 2.1e9 1.5e9 3.9e9 2.8e9 5.2e9 3.9e9
Mmol−gas/(Mdyn−rot + Mdyn−press) 35 ... 3.6 1.7 2.1 5.7 1.4 2.2
Frad=L/c(dyn) 36 8.5e34 1.6e34 4.2e33 2.2e34 5.1e34 1.4e34 1.8e34
(ΔvmaxdMout1/2/dt)/Frad 37 14 2 3 4 7 34 2
Notes.
aA(Hα)=7.4E(B–V),withE(B–V)stars=0.44E(B–V)gas(Calzetti2001).
bExtinction-corrected.
cSFR(M(cid:5)yr−1)=L(Hα)0/(2.1e41ergs−1),Mmol−gas(M(cid:5))=1.2e9SFR(M(cid:5)yr−1)0.75R(kpc)0.54(Equation(2),Kennicuttetal.2007).L(Hα)0isextinction-
corrected.Radiihereandelsewhereinthetable(e.g.,row10)are“intrinsic”radii,withtheinstrumentalresolutionremovedinsquares.
dFrom[Sii]6718/6733ratio(Osterbrock1989).
eIntrinsiclocalvelocitydispersion,afterremovalofbeamsmearedrotationandinstrumentalresolution.
fCorrectionforintrinsicdifferentialextinction.
gΔvmax=(cid:9)vbroad(cid:10)−2σbroad.
hEstimatesofoutflowrates(AppendixB)fortwomodelsofphotodissociation/caseBrecombinationandforcollisionalexcitation.
iUsestheaverageoftheestimatedofthetwophotodissociation/caseBmodelsinrows20and21.
jTimescaleforexpulsionofgasbyoutflows:texpulsion=2Mmol−gas/(dMout1/2/dt).Thelifetimeofaclumpisshorter,giventhatinadditiontogasoutflowsthereis
alsostarformation.
kTimeestimatesfromstellaragedating(Section4.2.2),chemicalenrichment(Section4.2.3,AppendixC)andexpansion(Section4.2.4).
lRatiooffinal(stellar)massatthetimewhenallthegasisexpelledbywinds,relativetotheinitialgasmass,M∗,final/Mgas,t=0=1/(1+[(dMout/dt)/SFR]).
mMaximumobservedvelocitygradientacrossclumpsin“raw”velocitymaps(inparentheses“residual”maps);positivesignisprogradeandnegativesignretrograde
withgalaxyrotation.
nMdyn−rot(M(cid:5))=b2.31e5(RHWHM(kpc))3(Δv(kms−1)/(2siniRHWHM(kpc)))2.
oMdyn−press(M(cid:5))=b5.63e5(σclump(kms−1))2RHWHM(kpc).
and/or larger scale, non-axisymmetric deviations from simple we use input models with a ring surrounding a central (ex-
rotationalmotion,wesubtractthiscos(ϕ)mapfromtheobserved tincted)bulgeforthemassdistribution,andforD3a15504and
velocityfield. ZC782941weuseexponentialdiskmodels(Genzeletal.2006,
The disk models compute data cubes from input structural 2008;S.Newmanetal.2011,inpreparation).Dynamicalmodel-
parameters(cf.Crescietal.2009).ForBX482andZC406690 ingandanalysisoftherest-frameopticalmorphologyindicates
5
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
D3a15504
A A A
z=2.39
B C
D
1”
E E
F
-200 -160 -130 -90 +130
G
1” +340 +300 +270 +230 +200
D
A C
B
BX482 +165 +130 +90 +60 +30
z=2.26
v
ZC782941
1” z=2.18
A
C B C
C
D E D E D D
-195 -160 -125 -90 +90
Figure1.Mapsofindividualvelocity“channels”ofwidth∼34kms−1intheHαlineofD3a15504(toprow),BX482(middletworows),andZC782941(bottomrow).
Themapsareresampledto0(cid:7).(cid:7)025perpixelandhavearesolutionofFWHM∼–0(cid:7).(cid:7)18–0(cid:7).(cid:7)25.Velocitiesrelativetothesystemicredshiftindicatedaregiveninkms−1.
Circles/ovalsandsymbolsdenotetheclumpsidentifiedinthesegalaxies.Crossesdenotethekinematiccentersofthegalaxyrotation.Thecolorscaleislinearand
autoscaledtothebrightestemissionineachchannel.
(Acolorversionofthisfigureisavailableintheonlinejournal.)
D3a15504 z=2.38 BX482 z=2.26 ZC782941 z=2.18 ZC406690 z=2.19
A
A
B C D A C
D
C
B
E A B
E
D
C D
F B
K−Hα H−Hα I−Hα I−Hα
A
A A
B C D C
D
C
A B
E B
D E
F C D
B
Hα G Hα Hα Hα
A
A
B C D A C
D
B C
A
E B
D E
C D
F B
K−band NACO H−band NIC I−band ACS I−band ACS
Figure2.FWHM∼0(cid:7).(cid:7)2Hαandrest-frameUV/opticalcontinuumimagesoffourmassiveluminousz∼2SFGs.Allmapshavebeenre-binnedto0(cid:7).(cid:7)025pixels.Top
row:three-colorcompositesofintegratedHαlineemission(red),andcontinuum(blue–green)images,alongwiththemostprominentclumpsidentifiedbylabelsA,
B,....Middle:integratedSINFONIHαemission.Allfourimagesareonthesameangularscale,withthewhiteverticalbarmarking1(cid:7)(cid:7)(∼8.4kpc).Bottom.HSTNIC
H-band,ACSI-band,orNACO-VLTAOKs-bandimagesoftheprogramgalaxies,ataboutthesameresolutionastheSINFONIHαmaps.Thecolorscaleislinear
andautoscaled.
(Acolorversionofthisfigureisavailableintheonlinejournal.)
6
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
thatthiscentralcomponentinBX482has∼20%ofthetotaldisk brightestclumpswheregassurfacedensitiesmaybesomewhat
mass(Genzeletal.2008;Fo¨rsterSchreiberetal.2011a,2011b). overestimated.
In either case the absence and/or weakness of emission from The gas surface densities/masses and star formation rates
the center has no influence on the analysis we discuss in the estimatedfromEquation(2)andlistedinTable2areuncertain
following. Position angles and inclinations are determined as by at least a factor of two to three. In addition to the well-
above.Themodeldataarethenconvolvedwiththeangularand known issue of how to infer molecular gas column densities/
spectral resolution profiles and sampled at the observed pixel masses from the integrated line flux of an optically thick
scales.ThetotaldynamicalmassM isthenvariedtoachieve CO rotational line (see the in-depth discussion in Tacconi
dyn
a best-fit match to the observed rotation velocities. To study et al. 2008 and Genzel et al. 2010), and the question of
the non-axisymmetric motions in a galaxy, the best-fit model whether Equation (2) adequately describes the gas to star
velocity and velocity dispersion maps are subtracted from the formationrelationforthephysicalconditionsonclumpscalesat
respectiveobservedmaps. z∼2,thereistheimportantissueofdifferentialextinction.We
We compare these residual maps with Hα surface bright- willargueinSection3.2thattheasymmetryofbroadHα/[Nii]
nessmapsderivedfromtheobserveddatacubes.Likewise,we lineemissionisdirectevidenceforsuchdifferentialextinction.
constructed[Nii]/Hαratiomapsfromintegratedlineemission Itisunclear,however,whatthegeneralimpactofthedifferential
maps smoothed with a 3 pixel (0(cid:7).(cid:7)15) kernel. We multiplied extinctionwouldbeonclumpscales.Onemightnaivelyexpect
these maps with a mask constructed from all pixels with Hα thattheeffectincreasesgascolumndensities/massesrelativeto
emissionat>3σ significance.Wealsoconstructedpixel–pixel averages on larger scales. However, there are almost certainly
correlation plots of residual velocity dispersion (δσ = σ also evolutionary effects, such that in a given aperture there
(data) − σ (model)) versus Hα surface brightness, and maybeveryhighdustcolumndensitiesinbothneutralclouds
[Nii]/HαlineratioversusHαsurfacebrightness.Beforeinves- and Hii regions with relatively low extinction. Such spatial
tigating possible trends in these correlations, we culled pixels separationsof300pcto>1kpcareseeninnearbyspirals,such
withlargeδσ or[Nii]/Hαuncertainties.Inaddition,inthecase asM51(Rand&Kulkarni1990),aswellasatz∼1(Tacconi
ofD3a15504(whichhasaprominentcentralbulge,AGNanda etal.2010).Asaresult,theKennicutt–Schmidtscalingrelation
narrowlineregion),wealsoremovedthenuclearregion. in Equation (2) may break down or be significantly altered on
smallscales(e.g.,Schrubaetal.2010inM33on(cid:3)80pcscales).
2.3.DeterminationofStarFormationRatesandGasMasses
2.4.SpatialDistributionoftheToomreQ-parameter
Forcalculatingstarformationratesandgassurfacedensities
fromtheHαdata,weusedtheconversionofKennicutt(1998b) A rotating, symmetric and thin gas disk is unstable to
modifiedforaChabrier(2003)IMF(SFR=L(Hα) /2.1×1041 gravitationalfragmentationiftheToomreQ-parameter(Toomre
0
erg s−1). We corrected the observed Hα fluxes for spatially 1964)is(cid:3)1.Foragas-dominateddiskinabackgroundpotential
uniform extinction with a Calzetti (2001) extinction curve (of dark matter and an old stellar component) Q is related
(A (Hα) = 7.4 E(B – V)), including the extra “nebular” to the local gas velocity dispersion σ0 (assuming isotropy),
correction (Agas = Astars/0.44) introduced by Calzetti (2001). circularvelocityvc,epicyclicfrequencyκ(κ2=4(vc/Rdisk)2+
We determined E(B – V) from the integrated UV/optical Rdiskd(vc/Rdisk)2/dRdisk),gassurfacedensityΣgas,andradiusof
photometryofthegalaxies(row5inTable2).Fo¨rsterSchreiber thediskRdiskviatherelation(Binney&Tremaine2008;Escala
et al. (2009) find that including the extra nebular correction &Larson2008;Elmegreen2009;Dekeletal.2009a)
brings Hα- and UV-continuum-based star formation rates of (cid:2) (cid:3)(cid:4) (cid:5) (cid:6)(cid:7)
z∼2SINSgalaxiesintobestagreement. σ κ σ a v2R /G
Q = 0 = 0 c disk
We estimated molecular surface densities (and masses, in- gas πGΣ v πR2 Σ
cluding a 36% helium contribution) from Equation (8) of (cid:2) (cid:3)ga(cid:2)s c(cid:3) (cid:2) (cid:3)di(cid:2)sk gas(cid:3)
Kennicutt et al. (2007), modified for the Chabrier IMF used = σ0 aMtot = σ0 a . (3)
here, v M v f
c gas c gas
(cid:2) (cid:3) (cid:2) (cid:3) √ √
Σ Σ
log mol−gas =0.73log star−form +2.91. (2) Here the constant a takes on the value of 1, 2, 3, and2
M(cid:5)pc−2 M(cid:5)yr−1kpc−2 foraKeplerian,constantrotationvelocity,uniformdensityand
solidbodydisk;f isthegasfractionwithinR .Ifthedisk
gas disk
Equation (2) is based on Hα, 24 μm, and CO observa- consistsofmolecular(H +He),atomic(Hi+He),andstellar(∗)
2
tions of M51 and is similar to results for larger samples of components,Qtot−1=QH2−1+ QHi−1+ Q∗−1ifallcomponents
z ∼ 0 SFGs (e.g., Equation (4) in Kennicutt 1998a, and have similar velocity dispersions. If there is a (young) stellar
Figure 4 of Genzel et al. 2010). It has the added advantage component distributed similarly to the gas, the combined gas
ofbeingbasedonspatiallyresolvedmeasurementsofthegasto + youngstarcomponentwillthushaveaQ thatisinversely
tot
starformationrelationwithasimilarspatialresolution(0.5kpc) proportionaltothesumofthegasandstellarsurfacedensities.
as our high-z data and also covering a similar range of gas Inthatcasef shouldbereplacedbythemassfractionf of
surface densities (10–103M(cid:5)pc−2). Figure 4 in Genzel et al. that“young”gacsomponent.Suchadiskisunstable(orstaybouleng)to
(2010;seealsoDaddietal.2010b)alsoshowsthattowithinthe fragmentationbygravity,dependingonwhetherQ isless(or
tot
uncertainties(ofaboutafactoroftwo),z∼0andz∼1–3SFGs greater)thanunity.Equation(3)canberewrittenas
(with galaxy-integrated measurements of CO luminosities and (cid:2) (cid:3) (cid:2) (cid:3)
SFRs)arefitbythesamerelation,althoughthegasmassesfrom σ z Qf
thebestfitsofGenzeletal.(2010)are∼20%largerthanesti- v0 = R = ayoung, (4)
c disk
matedfromEquation(2).InEquation(2)wedidnotcorrectthe
data for the fraction of Hα emission from outflowing gas (see where z is the z-scale height of the disk. Gas-rich, marginally
Section3.2).Thiscorrectionissmall,withtheexceptionofthe stable disks are thick and turbulent. The largest and fastest
7
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
D3a15504 z = 2.39
Hα
σ
obs
ΔQ < 0.5
Figure3.HαGaussianfitvelocities(topleft),HαGaussianfitdispersion(bottomleft),andinferredToomreQ-parameter(right,Equation(2))forD3a15504.Shown
inthetopcenterisalsothemapofHα-integratedfluxfromFigure2.Thelocationsofthemainclumps(Figure1)foundintheindividualvelocitychannelmapsare
denotedbycircles/ellipses.TheHα,velocity,andvelocitydispersionmaps(resolution0(cid:7).(cid:7)18FWHM)werere-binnedto0(cid:7).(cid:7)025pixels.ForconstructionoftheQ-map,
thedataweresmoothedto0(cid:7).(cid:7)25FWHM.ThetypicaluncertaintiesintheQ-valuesare ±0.05to ±0.3(1σ)throughoutmostofthediskofD3a15504.Pixelswith
ΔQ(cid:3)0.5aremaskedout.
(Acolorversionofthisfigureisavailableintheonlinejournal.)
growing, Jeans-unstable mode not stabilized by rotation is the topcenterpanelofFigure8.Inthiscase,wedonothaveaccessto
“Toomrescale/mass,”givenby(Elmegreen2009;Genzeletal. ahigh-resolutioncontinuumimage.Themostprominentclumps
2008;Escala&Larson2008;Dekeletal.2009a) arelabeledforeachgalaxy(seethemoredetaileddiscussionin
(cid:2) (cid:3) Section 2.2). Tables 2 and 3 summarize the derived physical
σ
R ≈0.8Q−1a−2 0 R properties. A “typical” individual clump within the massive
Toomre (cid:2) (cid:3)(cid:2)vc (cid:3)disk (M∗ ∼ 1010–1111M(cid:5)) BX/BzK galaxies in the SINS survey,
f R σ2 suchasanaverageclumpinD3a15504,ZC782941,andBX482,
≈1 young disk kpc∝ 0 and accountsforafewpercentoftheUV/opticallightoftheentire
0.4 5kpc Σ
(cid:2) (cid:3) gas galaxy, has a current star formation rate of a few solar masses
MToomre ≈0.6Q−2a−4 σ0 2Mdisk (pTearbyleea2r;,Fao¨nrdstearSstcehllraeribmeraestsaol.f20o1n1ebtoanadrfeefweretnimceessth1e0r9eMin(cid:5)).
v
(cid:2) c(cid:3) (cid:2) (cid:3) The most extreme clumps in BX482 and ZC406690 make up
≈5×109 f0yo.u4ng 2 10M11dMisk(cid:5) M(cid:5) ∝ Σσg04as, (5) ∼rat1e0s%o–f2100%–40ofsothlaerimntaesgsreastepderHyαearflaunxdesm, hasasveess∼ta1r0f1o0rMm(cid:5)at.ion
where the numerical factors are for a flat rotation curve 3.1.GiantClumpsaretheLocationsof
(a = 1.4). Gas-rich, marginally stable disks thus should have GravitationalInstability
much larger and more massive star-forming complexes than
thoseinz∼0SFGswith(cold)gasfractionsoflessthan10% AsdiscussedintheIntroductionandSection2.4,aplausible
hypothesis is that the ∼1–2kpc diameter giant star-forming
andlargerfractionsofstabilizingoldstellardisksandbulges.
clumps in z > 1 SFGs represent the largest/most massive
For the four well-resolved disks/rings, we created maps of
gravitationally unstable entities in the high-z disks. If this is
theToomreparameterQ(x,y).Wecombinedthecomputedgas
indeed the case, an empirical determination of the Toomre
surface density for each pixel (Equation (2)), with the best-
parameter(Equation(2))asafunctionofpositionshouldshow
fittingmodelrotationcurvetocomputetheepicyclicfrequency
thatclumpsandtheirsurroundingshaveQ(cid:3)1.
κ and the velocity dispersion map to calculate Q(x, y) from
Following the methods discussed in the last section,
Equation (3). We then used different Monte Carlo realizations
Figures 3–6 give the Q-maps at a resolution of ∼0(cid:7).(cid:7)22–0(cid:7).(cid:7)25
andstandarderrorpropagationstocomputemapsoftheuncer-
taintiesΔQ. FWHM for D3a15504, BX482, ZC782941, and ZC406690,
where we have only retained pixels with an rms uncertainty
ΔQ < 0.3–0.5. As inputs for our calculations we used the ve-
3.RESULTS
locity,velocitydispersion,andHα-integratedfluxmapsshown
Figures1and2showvelocitychannelmapsandtheintegrated in the left and middle panels of Figures 3–6. The central re-
Hα and continuum images for four of the five galaxies. The gionsinallfourgalaxiesshouldbeneglected,forthefollowing
integratedHαimageofthefifthgalaxy(BX599)isshowninthe reasons. The central fewkpc of D3a15504 may be affected by
8
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
BX482 z = 2.26
ΔQ < 0.5
Hα
σ
obs
Figure4.MapsofHαGaussianfitvelocities(topleft),HαGaussianfitdispersion(bottomleft),andtheToomreQ-parameter(right,Equation(2))forBX482.Shown
inthecenterisalsothemapofHα-integratedfluxfromFigure2Thelocationsofthemainclumps(Figure2)aredenotedbycircles/ellipses.TheHα,velocity,and
velocitydispersionmaps(resolution0(cid:7).(cid:7)18FWHM)werere-binnedto0(cid:7).(cid:7)025pixels.ForconstructionoftheQ-map,thedataweresmoothedto0(cid:7).(cid:7)25FWHM.The
typicaluncertaintiesintheQ-valuesare ±0.03to ±0.2(1σ)alongthebrightringofBX482.AllpixelswithΔQ>0.5weremaskedout.
(Acolorversionofthisfigureisavailableintheonlinejournal.)
Table3
AbundanceMeasurements
Source [Nii]/Hα Δ(Nii/Hα) μ=12 + log(O/H)a Δμ
(1) (2) (3) (4) (5)
BX599all 0.19 0.08 8.49 0.18
BX482clumpA 0.14 0.017 8.41 0.05
BX482clumpsB+C 0.11 0.024 8.35 0.09
BX482nucleus 0.22 0.027 8.53 0.05
D3a15504clumpsA–F 0.31 0.02 8.61 0.03
D3a15504interclump 0.33 0.02 8.63 0.03
D3a15504nucleus 0.43 0.04 [8.69]b 0.04
ZC782941clumpA 0.18 0.026 8.48 0.06
ZC782941clumpsB–E 0.28 0.021 8.58 0.03
ZC782941interclump 0.205 0.021 8.51 0.04
ZC406690all 0.097 0.017 8.32 0.08
ZC406690clumpA 0.073 0.015 8.25 0.09
ZC406690clumpB 0.22 0.017 8.53 0.03
ZC406690clumpC 0.14 0.019 8.41 0.06
Notes.
aμ=8.90+0.57log([Nii]/Hα)(Pettini&Pagel2004),withμ(cid:5)=8.66(Asplundetal.2004).
bSuspectbecauseofpossibleinfluenceofcentralAGN(Genzeletal.2006).
acentralAGN,aswellasbylargenon-circularmotions.Both bletofragmentationthroughouttheirdisks.Theclumpsarethus
increase the velocity dispersion there (Figure 3, bottom left; gravitationallyboundornearlyso.Ouranalysisonlyconsiders
Genzeletal.2006).ThecentralregionsofBX482,ZC782941, thegaseouscomponent.AsdiscussedinSection2.4,takinginto
and ZC406690 exhibit elevated velocity dispersions due to an account a stellar component with dispersion similar to that of
additionalcentralmass(withoutmuchHαemission)inthecases the gas will probably lower the Q-values still further. Given
ofBX482(Genzeletal.2008)andZC406690(S.Newmanetal. thetypicalmoleculargasfractionsof∼0.3–0.8(Tacconietal.
2011,inpreparation),andunresolvedbeamsmearingofrotation 2010;Daddietal.2010a),thispushesQtosignificantlybelow
inZC782941. unity in the prominent clumps. These clumps thus appear to
Wefindthatthroughouttheextendedouterdisksandtoward beinthehighlyunstableregime,wherelinearToomre-stability
the clumps of D3a15504, BX482, ZC782941, and ZC406690, analysisisinappropriate.ThefactthattheQ-parameterisbelow
the empirically determined Q-parameter is at or even signifi- unityeveninthemorediffusediskregionssuggeststhatglobal
cantlybelowunity.Aspostulated,theseSFGsareindeedunsta- perturbations are significant in setting the Q-distribution. We
9
TheAstrophysicalJournal,733:101(30pp),2011June1 Genzeletal.
ZC782941 z = 2.18
Hα
σ
obs
C
Δ
Figure5.MapsofHαGaussianfitvelocities(topleft),HαGaussianfitdispersion(bottomleft),andtheToomreQ-parameter(right,Equation(2))forZC782941.
ShowninthecenterisalsothemapofHα-integratedfluxfromFigure2.Thelocationsofthemainclumps(Figure2)aredenotedbycircles/ellipses.TheHα,velocity,
andvelocitydispersionmaps(resolution0(cid:7).(cid:7)18FWHM)werere-binnedto0(cid:7).(cid:7)025pixels.ForconstructionoftheQ-map,thedataweresmoothedto0(cid:7).(cid:7)25FWHM.The
typicaluncertaintiesintheQ-valuesare ±0.06to ±0.4(1σ)formostoftheouterdiskofZC782941.PixelswithΔQ>0.5weremaskedout.
(Acolorversionofthisfigureisavailableintheonlinejournal.)
v
ZC406690
z = 2.19
Hα ΔQ < 0.3
σ
obs
Figure6.MapsofHαGaussianfitvelocities(topleft),HαGaussianfitdispersion(bottomleft),andtheToomreQ-parameter(right,Equation(2))forZC406690.
ShowninthecenterisalsothemapofHα-integratedfluxfromFigure2.Thelocationsofthemainclumps(Figure2)aredenotedbycircles/ellipses.TheHα,velocity,
andvelocitydispersionmaps(resolution0(cid:7).(cid:7)22FWHM)werere-binnedto0(cid:7).(cid:7)025pixels.ThetypicaluncertaintiesintheQ-valuesare ±0.01to ±0.1(1σ)formostof
theouterdiskofZC782941.PixelswithΔQ>0.3weremaskedout.
(Acolorversionofthisfigureisavailableintheonlinejournal.)
concludethattheQ-mapsinFigures3–6areconsistentwiththe 3.2.EvidenceforPowerfulOutflowsonClumpScales
commonlyheldviewthattheclumpsformbygravitationalin-
stability.However,wecannotexcludethealternativepossibility UV spectroscopy of metal absorption lines and of Lyα
thattheinstabilityisdrivenbyalarge-scalecompression,such emissionlinesprovidecompellingevidenceforubiquitousmass
asexperiencedinagalaxyinteractionor(minor)merger(e.g., outflowsin“normal”high-z(Pettinietal.2000;Shapleyetal.
DiMatteoetal.2007). 2003; Steidel et al. 2004, 2010; Weiner et al. 2009). More
10