Shape of the oxygen abundance profiles in CALIFA face-on spiral galaxies

1 MB·

by  L. Sánchez-Menguiano

Astronomy&Astrophysicsmanuscriptno.sanchezmenguiano2016 (cid:13)cESO2016
January8,2016
Shape of the oxygen abundance profiles
in CALIFA face-on spiral galaxies
L.Sánchez-Menguiano1,2,S.F.Sánchez3,I.Pérez2,R.García-Benito1,B.Husemann4,D.Mast5,6,A.Mendoza1,
T.Ruiz-Lara2,Y.Ascasibar7,8,J.Bland-Hawthorn9,O.Cavichia10,A.I.Díaz7,8,E.Florido2,L.Galbany11,12,
R.M.GónzalezDelgado1,C.Kehrig1,R.A.Marino13,14,I.Márquez1,J.Masegosa1,J.Méndez-Abreu15,M.Mollá16,
A.delOlmo1,E.Pérez1,P.Sánchez-Blázquez7,8,V.Stanishev17,C.J.Walcher18,Á.R.López-Sánchez19,20,andthe
CALIFAcollaboration
1 InstitutodeAstrofísicadeAndalucía(CSIC),GlorietadelaAstronomías/n,Aptdo.3004,E-18080Granada,Spain
6 e-mail:[email protected]
1 2 Dpto.deFísicaTeóricaydelCosmos,UniversidaddeGranada,FacultaddeCiencias(EdificioMecenas),E-18071Granada,Spain
0 3 InstitutodeAstronomía,UniversidadNacionalAutónomadeMéxico,A.P.70-264,04510,México,D.F.
2 4 EuropeanSouthernObservatory(ESO),Karl-Schwarzschild-Str.2,85748Garchingb.München,Germany
5 InstitutodeCosmologia,RelatividadeeAstrofísica-ICRA,CentroBrasileirodePesquisasFísicas,RuaDr.XavierSigaud150,
n
CEP22290-180,RiodeJaneiro,RJ,Brazil
a
6 ObservatorioAstronómicodeCórdoba,UniversidadNacionaldeCórdoba,Argentina
J
7 DepartamentodeFísicaTeórica,UniversidadAutónomadeMadrid,Cantoblanco,E28049,Spain
7 8 Astro-UAM,UAM,UnidadAsociadaCSIC
9 SydneyInstituteforAstronomy,SchoolofPhysicsA28,UniversityofSydney,NSW2006,Australia
] 10 InstitutodeFísicaeQuímica,UniversidadeFederaldeItajubá,Av.BPS,1303,37500-903,Itajubá-MG,Brazil
A
11 MillenniumInstituteofAstrophysicsMAS,NuncioMonseñorSóteroSanz100,Providencia,7500011Santiago,Chile
G 12 DepartamentodeAstronomía,UniversidaddeChile,CaminoElObservatorio1515,LasCondes,Santiago,Chile
13 CEICampusMoncloa,UCM-UPM,DepartamentodeAstrofísicayCC.delaAtmósfera,FacultaddeCC.Físicas,Universidad
.
h ComplutensedeMadrid,Avda.Complutenses/n,28040Madrid,Spain
p 14 DepartmentofPhysics,InstituteforAstronomy,ETHZürich,CH-8093Zürich,Switzerland
- 15 SchoolofPhysicsandAstronomy,UniversityofStAndrews,SUPA,NorthHaugh,KY169SSStAndrews,UK
o
16 CIEMAT,Avda.Complutense40,E-28040Madrid,Spain
r
t 17 DepartmentofPhysics,ChemistryandBiology,IFM,LinköpingUniversity,SE-58183Linköping,Sweden
s 18 Leibniz-InstitutfürAstrophysikPotsdam(AIP),AnderSternwarte16,D-14482Potsdam,Germany
a
19 AustralianAstronomicalObservatory,POBox915,NorthRyde,NSW1670,Australia
[
20 DepartmentofPhysicsandAstronomy,MacquarieUniversity,NSW2109,Australia

1
v
Received25September2015/Accepted12December2015
2
4
5
ABSTRACT
1
0 Wemeasuredthegasabundanceprofilesinasampleof122face-onspiralgalaxiesobservedbytheCALIFAsurveyandincludedall
1. spaxelswhoselineemissionwasconsistentwithstarformation.Thistypeofanalysisallowedustoimprovethestatisticswithrespect
0 topreviousstudies,andtoproperlyestimatetheoxygendistributionacrosstheentiredisctoadistanceofupto3-4disceffectiveradii
6 (re).WeconfirmtheresultsobtainedfromclassicalHii regionanalysis.Inadditiontothegeneralnegativegradient,anouterflattening
1 canbeobservedintheoxygenabundanceradialprofile.Aninnerdropisalsofoundinsomecases.Thereisacommonabundance
: gradientbetween0.5and2.0re ofαO/H =−0.075dex/re withascatterofσ=0.016dex/re whennormalisingthedistancestothe
v disceffectiveradius.ByperformingasetofKolmogorov-Smirnovtests,wedeterminedthatthisslopeisindependentofothergalaxy
i properties,suchasmorphology,absolutemagnitude,andthepresenceorabsenceofbars.Inparticular,barredgalaxiesdonotseem
X
todisplayshallowergradients,aspredictedbynumericalsimulations.Interestingly,wefindthatmostofthegalaxiesinthesample
r withreliableoxygenabundancevaluesbeyond∼2effectiveradii(57galaxies)presentaflatteningoftheabundancegradientinthese
a
outerregions.Thisflatteningisnotassociatedwithanymorphologicalfeature,whichsuggeststhatitisacommonpropertyofdisc
galaxies.Finally,wedetectadroportruncationoftheabundanceintheinnerregionsof27galaxiesinthesample;thisisonlyvisible
forthemostmassivegalaxies.
Keywords. Galaxies:abundances–Galaxies:evolution–Galaxies:ISM–Galaxies:spiral–Techniques:imagingspectroscopy–
Techniques:spectroscopic
1. Introduction ingunsolvedfundamentalquestionsarecomprehendingthepro-
cessesthatareinvolvedintheassemblyofgalaxiesofdifferent
masses,therelativeimportanceofmergersversuscontinuousgas
Understandinghowdiscgalaxiesformandevolveisoneofthe
greatest challenges in galactic astronomy. Some of the remain-
Articlenumber,page1of21
A&Aproofs:manuscriptno.sanchezmenguiano2016
accretion infall into the disc, the rate of metal enrichment, and 2002; Pilyugin et al. 2004; Dalcanton 2007), or stellar-to-gas
theangularmomentumtransfersduringtheseprocesses. fraction (e.g. Zahid et al. 2014; Ascasibar et al. 2015). More-
Thedistributionofgaschemicalabundancesandstellarpa- over,studyingtheserelationsatdifferentredshiftscanhelpusto
rameters as well as their variation in space and time are im- understand the assembly history and evolution of galaxies (e.g.
portant tools for answering these questions on the evolution of Kobulnicky&Koo2000;Maiolinoetal.2008;Mannuccietal.
discsinspiralgalaxies.Infallmodelsofgalaxyformationpredict 2009,2010;Bellietal.2013).Theinside-outscenarioisnotonly
that spiral discs build up through accretion of material, which supported by studies focused on the stellar content in galaxies.
leadstoaninside-outgrowth(Matteucci&Francois1989;Molla Gasmetallicitystudieshavealsobeenkeyelementsinfavourof
etal.1996;Boissier&Prantzos1999).Theaccretionbringsgas such disc growth, predicting a relatively quick self-enrichment
into the inner parts of the discs, where it reaches high densi- withoxygenabundancesandanalmostuniversalnegativemetal-
ties that trigger violent and quite efficient star formation. Thus, licity gradient once it is normalised to the galaxy optical size
there is a fast reprocessing of gas in the inner regions, which (Boissier&Prantzos1999,2000).Severalobservationalstudies
leads to a population of old, metal-rich stars surrounded by a have found this radial decrease in the oxygen abundance along
high-metallicity gaseous environment, while the outer regions thediscsofnearbygalaxies(e.g.Vila-Costas&Edmunds1992;
arepopulatedbyyounger,metal-poorstarsformedfrompoorly Zaritsky et al. 1994; van Zee et al. 1998; Bresolin et al. 2009;
enrichedmaterial.Thefirstevidencethatsupportsthisscenario Moustakasetal.2010;Richetal.2012).
for disc evolution comes from studies on stellar colour profiles However, gas metallicity studies have also presented evi-
in nearby galaxies, which find bluer colours in the outer parts denceoftheexistenceofsomebehavioursintheoxygenabun-
(e.g. de Jong 1996; MacArthur et al. 2004; Taylor et al. 2005; danceprofilesthatdeviatefromthepureinside-outscenario:A
Muñoz-Mateos et al. 2007). This blueing can be interpreted on decreaseoranearlyflatdistributionoftheabundanceinthein-
thebasisofachangeinthediscscale-lengthasafunctionofthe nermostregionofdiscs,firstobservedbyBelley&Roy(1992);
observed wavelength band. This result is predicted by models andaflatteninginthegradientintheouterregionsmeasuredin
based on the inside-out growth (Prantzos & Boissier 2000). To several works (Martin & Roy 1995; Vilchez & Esteban 1996;
explain the nature of these colour gradients, recent works have Roy & Walsh 1997, among others). These features have been
analysed the radial change in the star formation history (SFH, theoreticallysuggestedtobemotivated,forinstance,bythepres-
Sánchez-Blázquezetal.2009;Pérezetal.2013). enceofradialmigration(Minchevetal.2011,2012).Neverthe-
Another independent result that is consistent with this sce- less,theiroriginisstillunknown.
nario comes from the weak dependence of disc galaxies with Allthesespectroscopicstudieswerelimitedbystatistics,ei-
redshiftonthestellarmass-sizerelation.Accordingtotheinside- ther in the number of observed Hii regions or in the coverage
outgrowthofdiscs,galaxiesareexpectedtoincreasetheirscale of these regions across the galaxy surface. The advent of inte-
lengthswithtimeastheygrowinmass(Bardenetal.2005;Tru- gral field spectroscopy (IFS) techniques offers astronomers the
jilloetal.2004,2006),resultinginaconstantmass-sizerelation opportunity to overcome these limitations by tracing the distri-
withcosmictime. butionofionisedgasandestimatingspatiallyresolvedchemical
Inthiscontext,thestudyoftheinterstellarmedium(ISM)is abundances for the gas phase. Its two-dimensional spatial cov-
crucial to understand the chemical evolution of galaxies, since erageallowsustoextractseveralhundredsoreventhousandsof
the enriched material is expelled during the evolution of stars, spectraacrosstheentiregalaxyextent.Thisenablesstudyingthe
is mixed with the already existing gas, and condenses to form variationofgaspropertiesthroughoutthewholedisc.
newstars.TheISMisfundamentallygaseous,anditschemical Moreover, IFS surveys offer the opportunity of extending
abundance can be derived by analysing spectroscopic features, thestudytoalargenumberofobjects,allowingformeaningful
thatis,nebularemissionlines.Theseemissionlinesareexcited statistical analysis. However, until recently, this technique was
byphotoionisationoftheinterstellargasbyhotandyoungmas- rarelyusedinasurveymode.Therewereonlyafewexceptions
sivestars(Aller1984;Osterbrock1989),whichformcloudsof such as the SAURON survey (Bacon et al. 2001) and the Disk
ionised hydrogen (Hii regions) where star formation (among MassSurvey(Bershadyetal.2010).
other processes) takes place. As oxygen is the most abundant These pioneering projects were not optimal for a statisti-
heavy element in Universe, this makes it the best proxy for the cal study of the properties of Hii regions because they incom-
totalgasmetallicity. pletelycoveredthefullextentofthegalaxies,amongotherrea-
The study of the Milky Way (MW) is also an exceptional sons.Suchastatisticalstudystartedwiththedevelopmentofthe
tool for our understanding of galaxy evolution, mainly because PINGSproject(Rosales-Ortegaetal.2010),whichacquiredIFS
we can observe both the stellar and the gaseous components mosaicdataforadozenverynearbygalaxies.Thisprojectwas
in greater detail than in other galaxies. Among other chemical followedbytheobservationofalargersampleofface-onspiral
properties of our Galaxy, the gas abundance gradient has been galaxies (Mármol-Queraltó et al. 2011) as part of the feasibil-
extensively studied (e.g. Shaver et al. 1983; Deharveng et al. ity studies for the CALIFA survey (Sánchez et al. 2012a). The
2000; Pilyugin et al. 2003; Esteban et al. 2005; Quireza et al. advent of CALIFA allowed extending the study to much more
2006; Rudolph et al. 2006; Balser et al. 2011); it is still not representative samples of nearby galaxies by covering all mor-
properly traced, however, especially in the inner parts. There- phologies.
fore, complementary information coming from data of external Based on large-statistics samples of Hii regions extracted
galaxieswouldhelpustoovercomethisproblem. from galaxies observed by these programs, Sánchez et al.
Thestudyofthegasmetallicityinexternalindividualgalax- (2012b, 2014) studied the distribution of metals within star-
ies using spectroscopic data allows us to shed light on funda- forminggalaxiesandprovidedthestrongestevidencesofarfor
mentalphysicalpropertiesofgalaxies,suchasSFR(e.g.Ellison acharacteristicgasabundancegradientouttotwoeffectiveradii
etal.2008;Lara-Lópezetal.2010;López-Sánchez2010),mass (r ). These studies also confirmed the behaviours mentioned
e
and luminosity (e.g. Lequeux et al. 1979; Skillman et al. 1989; abovethatdeviatefromthisgradient,aspreviouslyobservedby
Tremontietal.2004;Rosales-Ortegaetal.2012),effectiveyield otherIFSworksonindividualgalaxies(e.g.Bresolinetal.2009;
androtationvelocity(e.g.Vila-Costas&Edmunds1992;Garnett Sánchez et al. 2011; Rosales-Ortega et al. 2011; Bresolin et al.
Articlenumber,page2of21
L.Sánchez-Menguianoetal.:ShapeoftheoxygenabundanceprofilesinCALIFAface-onspiralgalaxies
2012; Marino et al. 2012). However, by selecting Hii regions etal.(2012a),Husemannetal.(2013),andGarcía-Benitoetal.
theydidnottakeadvantageofthefullcapabilityofanIFSstudy (2015).
andrestrictedthestudytoisolatedareasofthegalaxies. After following the standard steps for fibre-based IFS data
CurrentIFSsurveys(e.g.ATLAS3D,Cappellarietal.2011; reduction, the pipeline provides a regular-grid datacube, with x
CALIFA, Sánchez et al. 2012a; SAMI, Croom et al. 2012; andycoordinatesindicatingtherightascensionanddeclination
MaNGA, Bundy et al. 2015) have shown the potential of this ofthetargetandzbeingthestepinwavelengthforallgalaxiesin
kind of data to deliver important insights on this and other key thesample.Aninverse-distanceweightedimage reconstruction
questions about the formation and evolution of galaxies at low scheme was performed as interpolation method to reconstruct
redshifts. Recent articles have proved the power of this tool to the datacube. As a result, we have individual spectra for each
properlymapthespatiallyresolvedpropertiesofgalaxiesbyus- sampledspaxelof1(cid:48)(cid:48) ×1(cid:48)(cid:48) andafinalspatialresolutionforthe
ingthefulltwo-dimensional(2D)informationspaxelbyspaxel datacubesofFWHM∼2.5arcsec.
that is provided by these surveys (e.g. Papaderos et al. 2013;
Singh et al. 2013; Davies et al. 2014; Galbany et al. 2014;
Barrera-Ballesterosetal.2015;Belfioreetal.2015;Gomesetal.
The subset of galaxies used in this work was selected by
2015;Hoetal.2015;Holmesetal.2015;Lietal.2015;Martín-
adoptingthefollowingcriteria:
Navarroetal.2015;Wilkinsonetal.2015).
Inthiswork,wemakeuseoffull2Dinformationinanalysing
(a) Spiral galaxies with morphological types between Sa and
CALIFAdataspaxelbyspaxelwiththegoalofcharacterisingthe
Sm,includingbarredgalaxies.
radialgasabundanceprofileinasampleofface-onspiralgalax-
(b) Face-on galaxies, with i < 60◦, to avoid uncertainties in-
ies.Wenotonlyfocusonthebroadlyanalysedgradientofthese
ducedbyinclinationeffects.
profiles,butalsostudyotherfeaturesthatdeviatefromthesim-
(c) Galaxieswithnoevidentsignaturesofinteractionormerging
plenegativetrend,suchasinnerdropsandouterflattenings.We
(i.e.tails,bridges,rings,etc.).
alsoaimtocomparetheresultswiththoseobtainedfollowingthe
classicalprocedureofanalysingHii regions.Aspaxel-by-spaxel (d) GalaxieswithHαdetectedalongdifferentgalactocentricdis-
studyallowsustoimprovethestatisticswithrespecttoprevious tanceswithasignal-to-noiseratio(S/N)forthespaxelsabove
studies on the topic and also offers the possibility of properly 4onaverage.
estimatingtheoxygendistributionacrosstheentirediscsovera
distanceofupto3-4disceffectiveradii.Aproper2Dstudyofthe The classification according to morphological type and into in-
oxygenabundancedistributionthatanalysespossibleazimuthal teracting or non-interacting galaxy was based on the visual in-
variationswillbepresentedinaforthcomingwork. spection carried out by Walcher et al. (2014, see details in the
article).Afterimposingtheserestrictions,thegalaxysamplewas
The structure of the paper is as follows. Section 2 provides
reducedto204galaxies.Fromthese,weonlyanalysedthe129
a description of the sample and data we use in this study. Sec-
galaxiesthathavefinallybeenobservedbytheCALIFAcollab-
tion3describestheanalysisrequiredtoextractthespaxel-wise
information.WeexplaintheproceduretodetecttheHii regions orationwiththeV500setup.
Figure 1 shows the comparison of the distribution of mor-
analysed for comparison (Sect. 3.3) and derive the correspond-
phologicaltypes,absoluteB-bandmagnitudes,anddisceffective
ing oxygen abundance values using both methods (Sect. 3.4).
radiibetweenthespiralgalaxiesintheCALIFAmothersample
Our results are shown in Sect. 4, where we study the oxygen
andthesampleusedinthisstudy.Thereisacleardeficiencyof
abundanceslopedistribution(Sect.4.1),itsdependenceondif-
earlier (Sa-Sab) and later (Sc-Sdm) spirals; the sample is dom-
ferentpropertiesofthegalaxies(Sect.4.2),andtheexistenceof
inated by intermediate galaxies. This may be due to the impo-
acommonabundanceprofile(Sect.4.3).Finally,thediscussion
sition of ionised gas throughout the discs, which we have pri-
oftheresultsandthemainconclusionsaregiveninSect.5.
oritised to perform a detailed 2D study of the gas metallicity.
The distribution of galaxies according to their absolute mag-
nitude clearly shows an absence of faint galaxies, with values
2. Dataandgalaxysample above -18 mag. This fact is a consequence of a selection effect
in the definition of the CALIFA mother sample (Walcher et al.
Theanalyseddatawereselectedfromthe939galaxiesthatcom-
2014). The CALIFA mother sample was created by applying a
prisetheCALIFAmothersample(Sánchezetal.2012a).These
sizeselectioncriteriondefinedbyaminimumapparentisophotal
galaxieswereobservedusingthePotsdamMultiApertureSpec-
size.Asize-limitedsamplelikethisfavoursinclinedoverface-
trograph(PMAS;Rothetal.2005)atthe3.5mtelescopeofthe
onsytemsbecausetheinclinationincreasestheapparentisopho-
CalarAltoobservatorywithaconfigurationcalledPPAK(Kelz tal size (because the surface brightness increases). This effect
etal.2006).Thismodeconsistsof382fibresof2.7arcsecdiam-
causes these inclined galaxies to dominate the low-luminosity
eter each, 331 of them (the science fibres) covering an hexago-
populationofgalaxies.Becauseweselectedonlyface-ongalax-
nal field of view (FoV) of 74(cid:48)(cid:48)x64(cid:48)(cid:48). To achieve a filling factor ieswithi<60◦,weautomaticallydiscardedallthesefaintgalax-
of 100% along the full FoV and increase the spatial resolution,
ies.Ontheotherhand,acorrelationbetweenthemorphological
a dithering scheme of three pointings was adopted. Two differ-
typeofthegalaxiesandthemass(thus,withtheluminosity,see
entsetupswerechosenfortheobservations:V500,withanom-
e.g.Roberts&Haynes1994;GonzálezDelgadoetal.2015)has
inal resolution of λ/∆λ ∼ 850 at 5000 Å (FWHM ∼ 6Å) and been found, where later types present lower masses (and lumi-
a wavelength range from 3745 to 7500 Å, and V1200, with a nosities).Thiscontributestothedeficiencyoflaterspiralsthatis
better spectral resolution of λ/∆λ ∼ 1650 at 4500 Å (FWHM foundinthesamplebecauseofthelackoflow-luminositygalax-
∼ 2.7Å) and ranging from 3650 to 4840 Å. The data analysed ies and the correlation between these two parameters. For the
here were calibrated with version 1.5 of the reduction pipeline. distribution of galaxies according to their disc effective radius,
More detailed information about the CALIFA sample, obser- we finally obtained similar distributions for the CALIFA spiral
vational strategy and data reduction can be found in Sánchez galaxies and our sample: the sample is dominated by galaxies
Articlenumber,page3of21
A&Aproofs:manuscriptno.sanchezmenguiano2016
Fig.1.Distributionofmorphologicaltypes(topleft),absolute B-bandmagnitudes(topmiddle),anddisceffectiveradii(topright)ofthespiral
galaxiesintheCALIFAmothersample(unfilledblackhistograms)andthegalaxiesselectedinthiswork(filledbluehistograms).Themiddleand
bottompanelsshowthenormaliseddistributionsseparatedaccordingtothepresenceorabsenceofbars:barredgalaxies(redhistograms),unbarred
galaxies(greenhistograms)andnotclearlyidentifiedgalaxies(bluehistograms).Thedashedgreylineandtheredarrowmarkthelocationofthe
MWineachpanel.ThevaluesfortheabsoluteB-bandmagnitude(M =−20.8mag)andthedisceffectiveradius(r =3.6kpc)aretakenfrom
B e
Karachentsevetal.(2004)andBovy&Rix(2013,consideringr ∼1.67r )respectively.
e d
withr between4and10kpc.Furthermore,therearenosignif- mag,excludingthefaintgalaxiesbelowthislimitforthereasons
e
icantdifferencesinthedistributionwhenthegalaxiesaresepa- explainedabove.
rated into barred and non-barred galaxies, galaxies of all sizes
arepresentinbothcases.
We note that with the limitations we mentioned, which are
Figure 2 shows the distribution of our sample (filled dots) linked to the criteria we adopted to define the sample, the se-
and the total sample of CALIFA spiral galaxies (empty dots) lectedsetofgalaxiesiswellsuitabletocarryoutthestudypre-
alongthe(B−V)vsM colour-magnitudediagram.Oursample sentedhere,thatis,adetailed2Dstudyoftheradialgasmetal-
V
coversthesamerangeastheCALIFAspiralsabove M ∼ −17 licitydistributioninspiralgalaxies.
V
Articlenumber,page4of21
L.Sánchez-Menguianoetal.:ShapeoftheoxygenabundanceprofilesinCALIFAface-onspiralgalaxies
eachspectrumbyalinearcombinationoftheSSPtemplatesthat
are collected in the library after correcting for the appropriate
systemic velocity and velocity dispersion (including the instru-
mental dispersion, which dominates the total observed disper-
sion)andtakingintoaccounttheeffectsofdustattenuation.We
adoptedtheCardellietal.(1989)lawforthestellardustextinc-
tionwithR =3.1.
V
To measure the emission line fluxes and after the stellar
component is subtracted, FIT3D performs a multi-component
fitting using a single Gaussian function per emission line plus
a low-order polynomial function. When more than one emis-
sionlinewasfittedsimultaneously(e.g.fordoubletsandtriplets
like the [Nii] lines), the systemic velocity and velocity disper-
sion were forced to be equal to decrease the number of free
parameters and increase the accuracy of the deblending pro-
cess. The measured line fluxes include all lines required in de-
terminingthegasmetallicityusingstrong-linemethods,thatis,
Hα, Hβ, [Oii]λ3727, [Oiii]λ4959, [Oiii]λ5007, [Nii]λ6548,
[Nii]λ6584, [Sii]λ6717, and [Sii]λ6731. FIT3D provides the
intensity,equivalentwidth(EW),systemicvelocity,andvelocity
Fig.2.DistributionofthespiralgalaxiesintheCALIFAmothersample
(emptysmalldots)andthegalaxiesselectedinthiswork(filledlarge dispersionforeachemissionline.Thestatisticaluncertaintiesin
dots)inthe(B−V)vsM colour-magnitudediagram. themeasurementswerecalculatedbypropagatingtheerrorasso-
V
ciatedwiththemulti-componentfittingandtakingintoaccount
theS/Natthespectralregion.
3. Analysis As indicated above, FIT3D fits both the underlying stellar
population and the emission lines. In addition to the parame-
The main goal of this study is to characterise the radial abun-
ters derived for the emission lines, the fitting algorithm there-
dance profiles in the galaxy sample using the full 2D informa-
tionandcompareitwiththeresultsobtainedusingonlytheHii fore provides information related to the stellar population: the
luminosity-weightedagesandmetallicities,theaveragedustat-
regions. In this section we describe the procedure followed to
tenuation,themass-weightedagesandmetallicities,theaverage
selectthespaxels,analysetheirindividualspectra,andtoderive
mass-to-lightratio,andtheindividualweightsofthemulti-SSP
the corresponding oxygen abundance. We also explain how we
havedetectedtheHii regionsusedforcomparisonandthesub- decompositionthatinessencetracetheSFH.
Theentireprocedureoffittingandsubtractingtheunderlying
sequentanalysis.
stellarpopulationandmeasuringtheemissionlinesusingFIT3D
isdescribedinmoredetailinSánchezetal.(2011)andSánchez
3.1. Measurementoftheemissionlines etal.(2015b).
Wenotethatalltheseparameters(bothstellarandgas)were
In the spectrum of a galaxy (or a portion of it), the emission
derivedspaxelbyspaxelforeachindividualspectrumofthedat-
lines are superimposed on the underlying stellar spectrum. To
acubes, providing the sets of 2D maps that are the base of our
accuratelymeasuretheemissionlinefluxes,thestellarcontribu-
analysis.
tionmustbeestimatedandsubtractedfromthegalaxyspectrum
to derive a pure gas spectrum (allowing for the contribution of
noisefromthestellarpopulations)foreachindividualspaxel(or 3.2. Extractinginformationspaxelbyspaxel
Hii region).
As a result of the FIT3D fitting process, we obtained the set of
Several tools have been developed to model the underlying
2Dintensitymapsfortheemissionlinesthatarerequiredtode-
stellar population and decouple it from the emission lines (e.g.
terminethegasmetallicity.Toguaranteerealisticmeasurements
Cappellari&Emsellem2004;CidFernandesetal.2005;Ocvirk
oftheemissionlinefluxesforeachspaxel,weadoptedalower
etal.2006;Sarzietal.2006;Kolevaetal.2009;Sánchezetal.
limitbelowwhichweconsideredthatthefluxesareofthesame
2011). Most of them are based on the assumption that the star
orderasthecontinuumerror.Inthisway,wediscardedthespax-
formation history (SFH) of a galaxy can be approximated as a
elswhoseemissionlinefluxesemployedinthedeterminationof
sumofdiscretestarformationburstsand,therefore,thatthestel-
the oxygen abundance are lower than 1σ over the continuum
larspectrumcanbeconsideredastheresultofthecombination
ofspectraofdifferentsimplestellarpopulations(SSP)withdif- level.Fromallthespaxelswithfluxvaluesabovethislimit,we
nowselectedthosethatareassociatedwithstarformation(SF).
ferentagesandmetallicities.
Inthiswork,wemadeuseofafittingpackagenamedFIT3D1 The intensities of strong lines were broadly used to discern
betweendifferenttypesofemissionaccordingtotheirmainex-
to model both the continuum emission and the emission lines.
citation source (i.e. starburst or AGN) throughout the so-called
This tool uses an SSP template grid that comprises 156 indi-
diagnostic diagrams (e.g. Baldwin et al. 1981; Veilleux & Os-
vidual populations covering 39 stellar ages between 0.001 and
terbrock1987).Inmostcasesthesediagramsareveryusefulin
14.1 Gyr and four metallicities between 0.004 and 0.03. This
distinguishingbetweenstrongionisationsources,suchasclassi-
gridcombinestheGranadamodelsfromGonzálezDelgadoetal.
calHii/SFregionsandpowerfulAGNs.However,theyareless
(2005)fort<63MyrwiththoseprovidedbytheMILESproject
accurateindistinguishingbetweenlow-ionisationsources,such
(Vazdekisetal.2010;Falcón-Barrosoetal.2011)forolderages
asweakAGNs,shocks,and/orpost-AGBsstars(Stasin´skaetal.
(following Cid Fernandes et al. 2013). This way, FIT3D fits
2008;CidFernandesetal.2011).Alternativemethodsbasedon
1 http://www.astroscu.unam.mx/~sfsanchez/FIT3D acombinationoftheclassicallineratiosandadditionalinforma-
Articlenumber,page5of21
A&Aproofs:manuscriptno.sanchezmenguiano2016
tion regarding the underlying stellar population have been pro- largererrorsarethosewithalowS/N,andtheydonotaffectour
posed,forinstance,theso-calledWHANdiagram(CidFernan- conclusionssignificantly.
desetal.2011).ThisdiagramusestheEW(Hα)totakeintoac- The top right panel of Fig. 3 shows the location of the
count weak AGNs and ‘retired’ galaxies, that is, galaxies that selected spaxels in a particular spiral galaxy of the sample,
have stopped forming stars and are ionised by hot low-mass NGC0165,over-plottedtotheH map.Bluedotscorrespondto
α
evolvedstars. thespaxelsclassifiedasSFregionsandreddotsarethosethatlie
higherthantheKewleyetal.(2001)curveandcanthereforebe
Themostcommonlyuseddiagnosticdiagramwasproposed
associatedwithAGNs.Thefigureshowsthattheselectedspax-
byBaldwinetal.(1981,hereafterBPTdiagram).Thisdiagram
makes use of the [Nii]λ6584/Hα and [Oiii]λ5007/Hβ line ra- els follow the Hα emission. The classification of red points as
tios,whicharelessaffectedbydustattenuationbecauseoftheir ionisedbyAGNsinthisgalaxyaswellasinothercasesisclearly
proximity in wavelength space. Different demarcation lines for falsebecauseoftheirdistancetothegalacticcentres.Thismis-
classificationismostprobablyduetotheerrorsintheconsidered
BPTdiagramhavebeenproposedtodistinguishbetweenSFre-
gions and AGNs. The most popular are the Kauffmann et al. emission lines. As in the previous case, they could cause some
(2003)andKewleyetal.(2001)curves.PureHii/SFregionsare spaxels that are dominated by SF ionisation to lie higher than
considered to be below the Kauffmann et al. (2003) curve and theKewleyetal.(2001)curveinasimilarwaythaterrorscould
producetheoppositeeffectwithspaxelsassociatedtoAGNs.As
AGNs above the Kewley et al. (2001) curve. The area between
these spaxels represent only the 2% for the whole sample, in-
thetwocurvesisbroadlyanderroneouslyassignedtoamixture
ofdifferentionisationsources,sincepureSFregionscanalsobe cluding them would not alter our results significantly, and thus
they were discarded from the further analysis. For this galaxy,
foundhere.
1201spaxelswereassociatedwithSFregions.
These two demarcation lines have a different origin. The
Kewleyetal.(2001)curvewasderivedtheoreticallyusingpho-
toionisationmodelsand correspondstothemaximum envelope 3.3. DetectionandselectionofHiiregions
forionisationproducedbyOBstars.TheKauffmannetal.(2003)
We detected the Hii regions and extracted the corresponding
curvehasanempiricalorigin,basedontheanalysisoftheemis-
spectrausingasemi-automaticprocedurenamedHIIEXPLORER2.
sion lines for the integrated spectra of SDSS galaxies. It de-
scribes the envelope for classical Hii/SF regions well that are Theprocedureisbasedontwoassumptions:(a)Hii regionsare
peaky and isolated structures with a strongly ionised gas emis-
found in the discs of late-type spiral galaxies. However, it ex-
sion,particularlyHα,thatissignificantlyhigherthanthestellar
cludescertainkindsofSFregionsthathavealreadybeenfound
continuumemissionandtheaverageionisedgasemissionacross
abovethisdemarcationline(Kennicuttetal.1989;Hoetal.1997
and, more recently, Sánchez et al. 2014). Selecting Hii/SF re- thegalaxy;(b)Hii regionshaveatypicalphysicalsizeofabout
gions based on the Kauffmann et al. (2003) curve may there- one hundred or a few hundred parsecs (González Delgado &
Pérez 1997; Oey et al. 2003; Lopez et al. 2011), which corre-
fore bias our sample towards classical disc regions. Moreover,
spondstoatypicalprojectedsizeofafewarcsecatthestandard
itdoesnotguaranteethatothersourcesofnon-stellarionisation
distanceofthegalaxiesinthesample.
areexcludedthatmightpopulatethisarea,suchasweakAGNs,
shocks, and/or post-AGBs stars. We adopted the Kewley et al. A more detailed description of this algorithm can be found
inSánchezetal.(2012b),withafewmodificationspresentedin
(2001)curvetoexcludestrongAGNsandanEWcriteriontoex-
Sánchez et al. (2015a). Basically, the main steps of the process
clude weak AGNs and ‘retired’ emission (Cid Fernandes et al.
2011).However,weweremorerestrictiveintheEWrangethan areasfollows:(i)Firstwecreateanarrow-bandimageof120Å
width centred on Hα shifted at the redshift of each galaxy. (ii)
Cid Fernandes et al. (2011) and established the limit in 6 Å to
alsoguaranteeabetterS/Noftheemissionlinesforallspaxels. ThisimageisusedasaninputforHIIEXPLORER.Thealgorithm
detects the brightest pixel in the map and then adds all the ad-
Foradetailedanalysisoftheweakemissionprocedurespaxelby
jacentpixelsuptoadistanceof3.5”iftheirfluxesexceed10%
spaxelusingCALIFAdata,seeGomesetal.(2015).
of the peak intensity. After the first region is detected and sep-
In addition to the light from the ionised SF regions, there arated, the corresponding area is masked from the input image
is a background of diffuse nebular emission that extends over andtheprocedureisrepeateduntilnopeakwithafluxexceed-
the whole disc of the galaxies and can blur contribution of the ing the median Hα emission flux of the galaxy is found. The
SF regions, which is the subject of our study. However, most result is a segmentation FITS file describing the pixels associ-
of the diffuse ionised emission has been excluded by the 1σ ated with each detected Hii region. Finally, (iii) the integrated
limit imposed to the flux of the selected spaxels and the EW spectrum corresponding to each segmented region is extracted
criterion,sincethisemissionisdominatedbythestellarcontin- fromtheoriginaldatacube,andthecorrespondingpositiontable
uum. For a comparison of the location in the BPT diagram for ofthedetectedareaisprovided.
low-ionisation emission sources see, for instance, Kehrig et al.
After we extracted the spectra for the detected clumpy
(2012);Papaderosetal.(2013),andGomesetal.(2015).
ionised regions, we applied the same analysis described in
The top left panel of Fig. 3 shows the [Oiii]λ5007/Hβ vs. Sects.3.1and3.2:eachextractedspectrumwasdecontaminated
[Nii]λ6584/Hα diagnostic diagram for the spaxels in all 129 bytheunderlyingstellarpopulationusingFIT3D,andtheemis-
galaxiesofoursampleabovetheconsideredfluxlimitandwith sionlinefluxesweremeasuredbyfittingeachlinewithaGaus-
sianfunction.Theselineratioswereusedtodistinguishbetween
EW(Hα)>6Å.ThesolidanddashedlinesrepresenttheKewley
et al. (2001) and Kauffmann et al. (2003) demarcation curves, the detected ionised regions, the ones associated with star for-
mation. In a similar way as for individual spaxel spectra, using
respectively.SomepointsdominatedbySFionisationmightbe
the BPT diagram Hii/SF regions were considered to be under
present above the Kewley et al. (2001) curve as a result of the
errorsoftheconsideredemissionlines.Theyare,therefore,ex- theKewleyetal.(2001)curveandpresentanEW(Hα)>6Å.
cludedfromfurtheranalysisbyourcriteriaforselectingspaxels
associated with SF activity. However, the spaxels that present 2 http://www.astroscu.unam.mx/~sfsanchez/HII_explorer
Articlenumber,page6of21
L.Sánchez-Menguianoetal.:ShapeoftheoxygenabundanceprofilesinCALIFAface-onspiralgalaxies
Fig.3.Leftpanels:NormaliseddensitydistributionofthespaxelswithEW(Hα)above6Å (top)andofthedetectedHii/SFregions(bottom)along
theBPTdiagram.ThesolidanddashedlinesinbothpanelsrepresenttheKewleyetal.(2001)andKauffmannetal.(2003)demarcationcurves.SF
regionsareconsideredtobebelowtheKewleyetal.(2001)curve.Rightpanels:LocationofthespaxelsclassifiedasSFregions(bluedots)and
AGNs(reddots)accordingtotheBPTdiagramsuperimposedontheIFS-basedHαmapderivedforonegalaxyofthesample,NGC0165(top)
andaHαmapinunitsof(log10)10−16 ergs−1 cm−2 arcsec−1 forNGC0165,togetherwiththedetectedHii regionsshownasblacksegmented
contours(bottom).
Figure 3 (bottom left) shows the [Oiii]λ5007/Hβ vs. 3.4. Measurementoftheoxygenabundances
[Nii]λ6584/HαdiagnosticdiagramfortheHii/SFregions.The
solid and dashed lines represent the Kewley et al. (2001) and A direct procedure to measure abundances from observed
Kauffmannetal.(2003)demarcationcurves,respectively. spectra requires using temperature-sensitive line ratios such
as [Oiii]λλ4959,5007/[Oiii]λ4363. This is known as the
T −method (Peimbert & Costero 1969; Stasin´ska 1978; Pagel
e
Figure3(bottomright)showsanexampleofanHαmapfor etal.1992;Vilchez&Esteban1996;Izotovetal.2006).How-
onespiralgalaxyofthesample,NGC0165,wherethelocation ever, some of these auroral or nebular lines are very faint, and
oftheHii regionsissuperimposedasblacksegmentedcontours. they become even fainter as the metallicity increases (when
Forthisgalaxywedetected72Hii regions. a more efficient cooling mechanism begins to act through the
Articlenumber,page7of21
A&Aproofs:manuscriptno.sanchezmenguiano2016
metallines,whichproducesadecreaseinthetemperature),and
eventually, they are too weak to be detected. Furthermore, as a
resultoftheweaknessoftheinvolvedlines,theT −methodcan
e
onlybeappliedtonearbyandlow-metalobjectsforwhichvery
highS/Nspectraareobservable.
Itisthereforenecessarytolookforindirectmethodsthatal-
low us to estimate the chemical abundances. To do this, abun-
danceindicatorsbasedontherelationsbetweenmetallicityand
the intensity of strong and more readily observable lines have
beendeveloped.ThesemethodshavefirstbeenproposedbyAl-
loin et al. (1979) and Pagel et al. (1979). Since then, several
calibrators based on direct estimations of oxygen abundances
(Zaritskyetal.1994;Pilyugin2000;Denicolóetal.2002;Pettini
&Pagel2004;Pérez-Montero&Díaz2005;Pilyugin&Thuan
2005; Pilyugin et al. 2010; Marino et al. 2013) and photoion-
isation models (Dopita & Evans 1986; McGaugh 1991; Kew-
ley & Dopita 2002; Kobulnicky & Kewley 2004; Dopita et al.
2006, 2013; Pérez-Montero 2014) have been proposed and are
widelyusedtodays.Forarevisionofthedifferentmethods,their
strengthsandtheircaveats,seeLópez-Sánchezetal.(2012). Fig. 4. Radial density distribution of the spaxels in the oxygen
In this work we aim to derive the spatial distribution of abundance-galactocentricdistancespaceforthesamegalaxyasinFig.3
the oxygen abundance across the considered galaxies. For this (right panels). The radial distances are deprojected and normalised to
purpose, we used the emission line intensities derived spaxel the disc effective radius. The diamonds represent the mean oxygen
by spaxel for the sample of Hiiregions described before. We abundance values, with the error bars representing the corresponding
standarddeviations,forbinsof0.25r andtheredsolidlinetheerror-
adopted the empirical calibrator based on the O3N2 index that e
weightedlinearfitderivedforvalueswithintherangebetween0.5and
wasfirstintroducedbyAlloinetal.(1979):
2.0r (yellowdiamonds).Theparametersofthefitareshownintheup-
e
perrightcornerofthepanels,includingthezeropoint(a),theslope(b)
(cid:32)[Oiii]λ5007 Hα (cid:33) andthecorrelationcoefficient(r).Thevioletdotscorrespondtotheoxy-
O3N2=log Hβ × [Nii]λ6584 . (1) genabundancesderivedfortheindividualHii regions,andthedashed
blacklineisthelinearregressionforthesepoints.
This index (i) is only weakly affected by dust attenu-
ation because of the close distance in wavelength between
the lines in both ratios, (ii) presents a monotonic depen- axes. We preferred not to correct for the inclination effects in
dence on the abundance and (iii) uses emission lines cov- galaxies with an inclination below 35◦ because the uncertain-
ered by CALIFA wavelength range. One of the most pop- tiesinthederivedcorrectionexceedtheverysmalleffectonthe
ular calibrations for this index has been proposed by Pet- spatial distribution of the spaxels, even more when an intrinsic
tini & Pagel (2004, hereafter PP04). However, this indica- ellipticityisalsoconsidered.
tor lacks observational points at the high-metallicity regime Wethenderivedthegalactocentricdistanceforeachspaxel,
(12+log(O/H)>8.2) and instead uses predictions from pho- which was later normalised to the disc effective radius, as sug-
toionisation models. Therefore, we here adopted the improved gested in Sánchez et al. (2012b, 2013). This parameter was
calibration proposed by Marino et al. (2013, hereafter M13), derived from the disc scale-length based on an analysis of
where 12+log(O/H)=8.533−0.214 × O3N2. This calibra- the azimuthal surface brightness profile (SBP), explained in
tion uses Te-based abundances of ∼ 600 Hii regions from the AppendixA of S14. Other normalisation length-scales were
literature together with new measurements from the CALIFA used for better comparison with other studies, such as the r
25
survey, providing the most accurate calibration to date for this radius,whichisdefinedastheradiuscorrespondingtoasurface
index.Thederivedabundanceshaveacalibrationerrorof±0.08 brightnesslevelof25mag/arcsec2intheSDSSr-band3,andthe
dex,andthetypicalerrorsassociatedwiththepurepropagation physicalscaleofthegalaxy,thatis,thedistancesinkpc.
oftheerrorsinthemeasuredemissionlinesareabout0.05dex. Finally, we obtained the radial distribution of the oxygen
abundanceforeachgalaxy.Tocharacterisethisprofile,weper-
formedanerror-weightedlinearfittothederivedoxygenabun-
3.5. Oxygenabundancegradients
dance mean values for radial bins of 0.25 r within the range
e
To derive the radial distribution of the oxygen abundance for between 0.5 and 2.0 r . The radial binning was made to min-
e
eachgalaxy,wedeterminedthepositionangleandellipticityof imise possible azimuthal differences in the oxygen abundance
thedisctoobtainthedeprojectedgalactocentricdistancesofthe distribution, and the size of the bins was chosen to match the
selectedspaxels.Thesemorphologicalparameterswerederived seeing value. We excluded the innermost region (r/r <0.5),
e
by performing a growth curve analysis (Sánchez et al. 2014, which sometimes presents a nearly flat distribution or even a
hereafter S14). The inclination was deduced by also assuming droptowardsthecentre(e.g.Belley&Roy1992;Rosales-Ortega
anintrinsicellipticityforgalaxiesofq = 0.13(Giovanellietal. et al. 2011; Sánchez et al. 2012b, 2014). We also omitted the
1994): outer region (r/r >2.0), which it is found to have a flattening
e
(1−(cid:15))2−q2 intheabundancegradientforgalaxiescoveringregionsbeyond
cos2i= , (2) r (e.g. Martin & Roy 1995; Vilchez & Esteban 1996; Roy &
1−q2 25
Walsh1997;vanZeeetal.1998;Bresolinetal.2009;Rosales-
where (cid:15) is the ellipticity provided by the analysis and given by
(cid:15) =1−b/a,withaandbbeingthesemi-majorandsemi-minor 3 Usingtheseventhdatarelease(DR7,Abazajianetal.2009).
Articlenumber,page8of21
L.Sánchez-Menguianoetal.:ShapeoftheoxygenabundanceprofilesinCALIFAface-onspiralgalaxies
Ortega et al. 2011; Bresolin et al. 2012; Marino et al. 2012; form a Student’s t-test to check the significance of the correla-
López-Sánchez et al. 2015). The edges of the range were ob- tion,weobtainthatfor∼ 80%ofthegalaxiestheoxygenabun-
tained based on a visual inspection of each individual galaxy, dance and the radial distance (normalised to the disc effective
discarding the regions where we observed the mentioned fea- radius)aresignificantlycorrelatedtothe95%level(0.05).
tures.ThefittedintervalhaschangedwithrespecttoS14,simply The distribution of zero points ranges between 8.4 and 8.7,
becauseofabetterspacecoveragethatallowedustorefinethe reflecting the mass-range covered by the sample as a conse-
radiallimits.Thelinearfitwasweightedusingthestandarddevi- quenceofthewell-knownM−Zrelation(e.g.Tremontietal.
ationsof(mean)valueswithineachbinandconsideredonlythe 2004;Sánchezetal.2013).Thepresenceofapeakinthedistri-
binsthatcontainedatleastninevaluesoftheoxygenabundance. butionandasmallstandarddeviationistheresultofabiasinthe
Thisminimumnumberofvaluesrequiredforeachbinwasdeter- selection of the sample, explained in Sect. 2, which is due to a
minedtoensureaprecisioninthederivedmeanthatistentimes lackoflow-luminositygalaxies.
better than the dynamic range of abundance values covered in
Finally, the distribution of slopes presents a characteristic
thefit,takingintoaccountthedispersioninthemeasurements. value of α = −0.07dex/r with a standard deviation of
O/H e
Itisimportanttonotethatuncertaintiesinthedetermination
σ∼0.05dex/r .WeperformedaLillieforstest(Lilliefors1967)
e
of the deprojected galactocentric distances can significantly af-
to assess the compatibility of the distribution with a Gaussian,
fectthederivationoftheabundancegradients.Moreover,asthe
obtainingateststatisticsof0.07andaP-valueof0.62,showing
radial galactocentric distances are normalised to the disc effec-
that the distribution of slopes has a clear peak and is remark-
tive radius, the uncertainties in the determination of the r can
e ably symmetric. We ran a Monte Carlo simulation to estimate
alsocontributetothescatterobtainedonthefinaloxygengradi-
thecontributionoftheerrorsinthederivedslopestotheσofthe
ents.Ononehand,performingMonteCarlosimulations,weob-
distribution,obtainingthattheseerrorscanonlyexplain49%of
tainedthatanerrorof5◦intheinclinationandPAofthegalaxies
the width distribution. We may have underestimated the errors
can produce a dispersion in the gradient distribution of at most
involved in the determination of the slopes, particularly the ef-
0.05dex/r (0.02dex/r onaverage).Ontheotherhand,compar-
e e fect of the inclination. Otherwise, the remaining σ must have
ingdifferentmethodstoderivether (asdescribedinS14)and
e anotheroriginthatweinvestigateinmoredetailinSect.4.2.
taking into account both the nominal errors and the differences
TheanalysisfortheindividualHii regionsleadsustovery
amongthem,theoverallcontributiontothedispersioninthegra-
similarresults.Inthiscase,wehaveawiderdistributionforthe
dientdistributioncomingfromthederivationofther isatmost
e correlation coefficients, but we have to note that the number of
0.04dex/r (0.01dex/r onaverage).Alltheseuncertaintiesare
e e points involved in the linear fit is larger using all the individ-
wellaccountedforbyourerrorestimationofthegradient(0.05
ual Hii regions, since we did not apply any kind of radial bin-
dex/r ,seeSect.4.1).
e ningtothedata.Thecorrelationcoefficientislargerthan0.32for
In Fig. 4 we present an example of the abundance gradi-
∼60%ofthegalaxies,whichcorrespondstoasignificancelevel
ent for the same galaxy shown in the right panels of Fig. 3,
of98%(0.02).Thedistributionofzeropointscoversalmostthe
NGC0165,usingboththespaxel-wiseinformation(colourmap)
samerangeasforthespaxels(between8.3and8.8),allowingus
andtheindividualHii regions(violetdots).Thediamondsrep-
todrawthesameconclusions.Finally,thedistributionofslopes
resent the mean oxygen abundance values of the radial bins,
presents a characteristic value of α = −0.05dex/r with a
with the error bars indicating the corresponding standard devi- O/H e
standarddeviationofσ∼0.06dex/r .TheLillieforstestgivesa
ations.Theredsolidlineistheerror-weightedlinearfitderived e
teststatisticof0.08andaP-valueof0.28,verysimilartotheone
for values within the range between 0.5 and 2.0 r (yellow di-
e described for the spaxel-wise analysis. The Monte Carlo simu-
amonds), and the dashed black line is the linear regression for
lation yields a contribution of 44% of the errors in the derived
theindividualHii regions. Thisfigureillustratestheprocedure
slopestothedistributionwidth,againinsufficienttoexplainthe
explainedbeforetoderivetheoxygenabundancegradient.From
σofthedistribution.
theoriginal129galaxies,wewereabletofit122lines,andthe
If we use different scales to normalise the radial distances
remaininggalaxieswerediscardedfromfurtheranalysisbecause
like r and the physical scale of the galaxy (radius in kpc) for
ofthelownumberofspaxelsassociatedwithSFregionsthatare 25
boththespaxel-wiseandtheindividualHii regionanalysis,we
neededtocarryoutthelinearfit.Thisfinalsampleprovidesmore
than185000oxygenabundancevalues,∼8230ofthembeyond obtaininallcasesasimilarandnarrowdistribution,althoughfor
twodisceffectiveradii,andwithmorethan7100Hii regionsto the physical scale the distribution is clearly asymmetric, with a
comparewith,∼605beyond2r . tail towards large slopes. We note that our range of masses is
e
narrow, and consequently, so is the range of r and r , which
e 25
inturncausesthedistributionwhennormalisingtothephysical
4. Results scalenarrowaswell,incontrasttowhatweshouldobtainfora
wider range of masses. The different slope values are given in
With the procedure explained in the previous section we ob-
Table1.
tained the oxygen abundance gradient for the 122 galaxies in
oursample.Wedescribethemainpropertiesoftheseabundance
profilesbelow.
4.1.1. Comparisonwithothercalibrators
It is beyond the purpose of this work to make a detailed com-
4.1. Abundancegradientdistribution
parisonoftheoxygenabundancegradientsderivedusingdiffer-
Figure 5 shows the distribution of the correlation coefficient, entempiricalcalibrators.However,wecompareourresultswith
zero-point, and slope of the abundance gradients for the final thoseprovidedwithothermethodsbyderivingtheoxygenradial
sampleusingboththespaxel-wiseinformation(reddashedhis- distributions using some of the most commonly used empirical
togram)andtheindividualHii regions(bluefilledhistogram). calibrators:(i)thetechniqueproposedbyPP04fortheO3N2in-
We focus first on the spaxel-wise analysis. For almost all dex,(ii)thePilyuginetal.(2010,hereafterP10)calibrationfor
galaxiesthecorrelationcoefficientislargerthan0.75.Ifweper- theONSindex,(iii)andtheDopitaetal.(2013,hereafterD13)
Articlenumber,page9of21
A&Aproofs:manuscriptno.sanchezmenguiano2016
Fig.5.Distributionofcorrelationcoefficients(leftpanel),zeropoints(middlepanel),andslopes(rightpanel)ofthelinearfitsderivedforthe
oxygenabundancegradientsofthefinalsampleusingspaxel-wiseinformation(dashedredbars)andtheindividualHii regions(filledbluebars).
For both the zero point and slope distributions the lines represent the Gaussian distribution of the data (solid line for spaxels, dashed line for
individualHii regions),assumingthemeanandstandard-deviationofthedistributionofeachanalysedparameterandsampledwiththesamebins.
Table1.Oxygenabundancegradientslopesderivedusingdifferentcalibratorsanddistancenormalisations.
Calibrator α −spaxels α −Hii regions
O/H O/H
[dex/r ] [dex/r ] [dex/kpc] [dex/r ] [dex/r ] [dex/kpc]
e 25 e 25
O3N2[M13] −0.07±0.05 −0.08±0.06 −0.009±0.008 −0.05±0.06 −0.06±0.06 −0.008±0.010
O3N2[PP04] −0.11±0.07 −0.12±0.09 −0.014±0.012 −0.08±0.09 −0.09±0.09 −0.011±0.014
ONS[P10] −0.06±0.06 −0.08±0.07 −0.008±0.009 −0.06±0.09 −0.08±0.11 −0.008±0.012
pyqz[D13] −0.14±0.09 −0.15±0.10 −0.019±0.014 −0.11±0.09 −0.13±0.11 −0.015±0.015
calibrationbasedontheMAPPINGSIVcodedevelopedbythe
authors.
The PP04 calibration for the O3N2 index, as already men-
tioned in Sect. 3.4, is one of the most popular calibrations
used for this index and is defined as 12+log(O/H) = 8.73−
0.32 × O3N2.Thiscalibrationisnotvalidinthelow-metallicity
regime (12 + log(O/H) < 8), but as we do not reach this
limit, this effect will not affect our results. The P10 ONS cal-
ibration makes uses of the N /R and S /R ratios (defined
2 2 2 2
as R = [Oii](λ3727+λ3729), N = [Nii](λ6548+λ6584),
2 2
S = [Sii](λ6717+λ6731)) as temperature and metallicity in-
2
dexesandisvalidoverthewholerangeofexploredmetallicities.
Fig.6.Comparisonoftheoxygenabundancedistributionderivedusing
Thederivedrelationstodeterminetheoxygenabundancesusing
thecalibrationproposedbyM13fortheO3N2indexwiththedistribu-
thiscalibrationaregivenbyP10.Finally,theD13calibrationis tion derived using the P10 calibration for the ONS index (left panel)
basedonagridofphotoionisationmodelscoveringawiderange and the calibration based on pyqz code (D13, right panel). The black
ofabundanceandionisationparameterstypicalofHii regionsin contoursshowthedensitydistributionoftheSFspaxels,theoutermost
galaxies.ThiscalibrationcanbeusedthroughaPythonmodule oneincluding80%ofthetotalnumberofspaxelsanddecreasing20%
implemented by the authors, known as pyqz, which is publicly in each consecutive contour. The black dashed lines indicate the 1:1
available4. relationbetweentherepresentedcalibrators.
Table 1 shows a comparison among the oxygen abundance
slopes derived using the different calibrators and the different
We also illustrate this comparison in Fig. 6. The left panel
normalisations for the radial distance described before. In this
represents the distribution of oxygen abundances derived using
tablewepresentthevaluesusingboththespaxelsandtheindi-
vidualHii regions. the M13 calibration for the O3N2 index vs the P10 calibration
fortheONSindex.Intherightpanelweshowthesamedistribu-
tionoftheM13calibration,butthistimevstheD13calibration
4 http://dx.doi.org/10.4225/13/516366F6F24ED based on the pyqz code. Both panels show a tight correlation
Articlenumber,page10of21