Mon.Not.R.Astron.Soc.000,000–000(0000) Printed16January2017 (MNLATEXstylefilev2.2) The imprints of bars on the vertical stellar population gradients of galactic bulges A. Molaeinezhad1⋆, J. Falco´n-Barroso2,3, I. Mart´ınez-Valpuesta2,3, H.G. Khosroshahi1, A. Vazdekis2,3, F. La Barbera6, R.F. Peletier5, M. Balcells2,4 1SchoolofAstronomy,InstituteforResearchinFundamentalSciences(IPM),POBox19395-5746Tehran,Iran 2InstitutodeAstrof´ısicadeCanarias,E-38200,LaLaguna,Spain 3Depto.Astrof´ısica,UniversidaddeLaLaguna(ULL),E-38206LaLaguna,Tenerife,Spain 7 4IsaacNewtonGroupofTelescopes,Apartado321,38700SantaCruzdeLaPalma,CanaryIslands,Spain 1 5KapteynAstronomicalInstitute,UniversityofGroningen,Postbus800,9700AVGroningen,theNetherlands 0 6INAF OsservatorioAstronomicodiCapodimonte,I-80131Napoli,Italy 2 n a Accepted2017January09.Received2017January09;inoriginalform2016July26 J 2 1 ABSTRACT Thisisthesecondpaperofaseriesaimedtostudythestellarkinematicsandpopulationproper- ] A tiesofbulgesinhighly-inclinedbarredgalaxies.Inthiswork,wecarryoutadetailedanalysis ofthestellarage,metallicityand[Mg/Fe]of28highly-inclined(i>65o)discgalaxies,from G S0 to S(B)c, observedwith the SAURON integral-fieldspectrograph.The sample is divided h. intotwocleansamplesofbarredandunbarredgalaxies,onthebasisofthecorrelationbetween p thestellarvelocityandh3 profiles,aswellasthelevelofcylindricalrotationwithinthebulge - region.Wefindthatwhilethemeanstellarage,metallicityand[Mg/Fe]inthebulgesofbarred o andunbarredgalaxiesarenotstatisticallydistinct,the[Mg/Fe]gradientsalongtheminoraxis r t (awayfromthedisc)ofbarredgalaxiesaresignificantlydifferentthanthosewithoutbars.For s barredgalaxies,stars thatare verticallyfurtheraway fromthe midplaneare in generalmore a [ [Mg/Fe]–enhancedand thus the verticalgradientsin [Mg/Fe] for barredgalaxiesare mostly positive, while for unbarred bulges the [Mg/Fe] profiles are typically negative or flat. This 1 result, together with the old populations observed in the barred sample, indicates that bars v arelong-lastingstructures,andthereforearenoteasilydestroyed.Themarked[Mg/Fe]differ- 6 enceswiththebulgesofunbarredgalaxiesindicatethatdifferentformation/evolutionscenarios 6 arerequiredtoexplaintheirbuild-up,andemphasizestheroleofbarsinredistributingstellar 4 materialinthebulgedominatedregions. 3 0 Keywords: galaxies:bulges–galaxies:kinematicsanddynamics–galaxies:abundances– . 1 galaxies:evolution–galaxies:formation–galaxies:stellarcontent. 0 7 1 : v 1 INTRODUCTION face-on disc galaxies and the stellar population gradients along i X the bars major axis (see Sa´nchez-Bla´zquez 2016, and references IntheeraofIFUspectroscopy,thespatiallyresolvedspectroscopic therein).Moorthy&Holtzman(2006)analysedlong-slitspectraof r information, along with modern stellar population models, have a 38 face-on spirals of type S0–Sbc and reported a trend towards providednewinsightsintoourunderstandingoftheroleofbarson youngeragesforthebulgesofbarredgalaxies.Theyconcludedthat theformationandevolutionofdiscgalaxies. Barredgalaxiesrep- apositiveagegradientisadirectindicatoroftheexistenceofabar. resent aconsiderable fractionoftheentirediscgalaxypopulation Pe´rez&Sa´nchez-Bla´zquez (2011) studied the age and metallicity (e.g. Eskridgeetal. 2000; Knapenetal. 2000; Whyteetal. 2002; gradients in the bulge region of a sample of 20 early-type barred Grosbøletal.2004;Marinova&Jogee2007).Theyareconsidered galaxies.Usingthelong-slitspectraalongthebarmajoraxis,they askeydriversoftheinternalsecular evolutionindiscgalaxiesby foundthat atagiven velocitydispersion (σ),thebulgesof barred re-distributionoftheangularmomentum,triggeringofstarforma- galaxies seem to be more metal rich than the bulges of unbarred tion and the morphological transformation of galaxies in general, galaxies.Theymeasuredmostlynegativemetallicitygradientsfor as demonstrated observationally (e.g. Kormendy 2013) and theo- the bulges in their sample, suggesting a strong link between the retically(e.g.Athanassoula2013). metallicity of the bulge and the presence of a bar. They also in- Most stellar population studies of barred systems focus on dicatedthatbulgepopulationsdistributionsandgradientmeasure- mentscouldbebiasedagainsttheexistenceofthebulgesubstruc- ⋆ Email:[email protected] tures,suchasanuclear ringoraninnerdisc.Similarresultshave (cid:13)c 0000RAS 2 Molaeinezhadetal. beenobtained byCoelho&Gadotti(2011),that claimthecurrent (see Bureau&Athanassoula 2005; Athanassoula 2016, for more star formation is enhanced in the centres of barred galaxies, and details).Thisisfollowedbyintroducingamethodtoquantifycylin- consequently the measured mean stellar age of bulges is strongly dricalrotationthatisrobustagainstinnerbulgesubstructuressuch affectedbythebar. asinner discs, which arelikelymore common inbarred galaxies. A detailed and unbiased study of the extended stellar pop- Our results also confirmed high levels of cylindrical rotation in ulation in the bulges of barred and unbarred galaxies requires barred galaxies withboxy/peanut bulges. In the current work, we high inclinations to ensure minimal contamination by the stel- studythestellarpopulationpropertiesandinparticular,thevertical lar disc. Nonetheless, the number of works, dedicated to study gradientsalongtheminoraxisinthebulgesof28highly-inclined the stellar populations in highly-inclined barred galaxies is still (i > 65o)discgalaxies,observedwiththeSAURONintegral-field verylowandislimitedtoafew nearbycases(e.g.Jablonkaetal. spectrograph.UsingthekinematiccriteriaintroducedinPaperI,we 2007;Williamsetal.2011;Pe´rez&Sa´nchez-Bla´zquez2011).Un- areabletosecureanunbiasedclassificationofbarredandunbarred derstandingthevariationsofstellarpopulationparametersofbulges galaxiesinoursample,andcomparethestellarpopulationproper- atdifferent heights fromthediscprovides stronger constraintson tiesandverticalgradientsineachclassofbulges. theformationandevolutionscenariosofbulges.Morespecifically, The paper is organised as follows. In Section 2 we describe comparing thevertical gradientsofthestellarpopulation parame- thesampleandgiveasummaryofopticalpropertiesandkinemat- tersinbulgeswithandwithoutbarscanprovideveryimportantnew icsofthegalaxiesinoursample.Section3isdedicatedtotheline- informationtoexpandourunderstandingofbarsnurtureanditsrole strengthmeasurementsandpresenttheresultsofourstellarpopula- informationandevolutionofdiscgalaxies. tionanalysis.InSection4webrieflydiscussthepossiblescenarios Using long-slit spectroscopy, Jablonkaetal. (2007) studied to interpret these results. Finally, we summarize our findings and theverticalgradientsofstellarpopulationpropertiesinthebulges drawourconclusionsinSection5. of 32 edge-on galaxies, spanning a wide range of Hubble types, fromS0toSc.Theyfoundthatthecentralpartsofthebulgesare mostly younger and more metal rich than their outer regions and 2 SAMPLE&IFUOBSERVATIONS theouterpartsofthebulgeshavehigher[α/Fe]thantheinnerparts. Theyfoundnomeaningfuldifferencesbetweenthestellarpopula- 2.1 SampleSelection&Properties tionsofnearlyedge-onbarredandunbarredgalaxies,althoughthey Oursampleconsistsof28highly-inclined(i > 65o)galaxieswith were not in the position to precisely distinguish bulges with and highS/NIFUobservations,thatcanbeusedtostudythestellarpop- withoutbars.Ithasbeenwidelyacceptedthatboxy/peanut-shaped ulationgradientsalongtheminoraxis.Thesampleisdrawnfrom bulges (hereafter BP bulges), found in about half of edge-on thecomplete,magnitudelimitedsampleof260nearbygalaxiesin galaxies,arebarsviewedfromtheside(e.g.Kuijken&Merrifield theATLAS3Dsurvey(Cappellarietal.2011),complementedwith 1995; Merrifield&Kuijken 1999; Bureau&Freeman 1999; the sample of highly inclined galaxies introduced in Paper I. For Athanassoula 2005). Williamsetal. (2011) studied the stellar theATLAS3Dsurvey,wefocusedondiscgalaxies,andonlykept kinematicsandpopulationsinfiveedge-ongalaxies,dominatedby those galaxies whose inclination, as measured via the best fitting BPbulges,usinglong-slitspectroscopy.Theymeasuredthevertical mass-follow-light JAMmodels (Cappellarietal. 2013), areabove gradientsinthebulgesandclaimedthatBPbulgesintheirsample 65◦. This criterion allows us to access the clean vertical extent donotformahomogeneousclassofobjects.Later,Williamsetal. of the bulge, lessdisrupted by contamination of dust, disc and/or (2012),analysedthecentralandthemajoraxisprofilesinthestellar central components. We visually inspected the sample to discard populations of 28 highly inclined S0–Sb disc galaxies, presented galaxieswithirregular morphology or low qualityIFUdata(with inBureau&Freeman(1999)andChung&Bureau(2004),mostly poorspatialsamplingintheregionsofinterest).Inadditionwedis- with BP bulges. They found that at a given σ, gradients in BP cardedgalaxieswheretheemissionlinecleaningofthespectrawas bulges are shallower than those in elliptical galaxies, but that the particularlyproblematicintheshortwavelengthrangeprovidedby stellar populations at the very centres of disc galaxies with or theSAURONspectrograph.Ourfinalsamplecomprises20galaxies withoutBPbulgesdonotdifferfromthoseofearly-typegalaxies. from the ATLAS3D sample (hereafter S2), along with 8 galax- As we stated earlier, only a limited number of studies have ies from our original sample of Paper I (hereafter S1). The basic been dedicated to investigate the stellar population properties of photometricpropertiesofthese28galaxiesarepresentedinTable1. bulgesinhighly-inclinedbarredgalaxies,andontopofthat,there- sultsaremostlyinconsistent(seeSa´nchez-Bla´zquez2016,andthe referencestherein).Thismightbe,atleastinpart,duetothelack 2.2 Observations ofareliableandrobustmethodtodistinguishbarredandunbarred galaxies in highly-inclined systems. The presence of a bar is not TheSAURONobservationsofbothsamplesS1andS2aredescribed generallyapparentifthebarisorientedclosetoorexactlyparallel indetailinPaperIandCappellarietal.(2011),respectively.Briefly, totheline-of-sight,eveninedge-ongalaxies,(seee.g.Bureauetal. the spectroscopic observations of the S1 sample were carried out 2004; Athanassoula 2016; Molaeinezhadetal. 2016). Therefore, between October 1999 and 2011 with the SAURON integral-field previousresultsmightbehighlybiased(seePeletieretal.2007). spectrograph (Baconetal. 2001) attached to the 4.2-m William Inthefirstpaperofthisseries(Molaeinezhadetal.2016,here- Herschel Telescope (WHT) of the Observatorio del Roque de los after Paper I) we studied the connection between bulge morphol- MuchachosatLaPalma,Spain.Weusedthelowspatialresolution ogy and kinematics in 12 mid to highly-inclined disc galaxies, mode ofSAURON which givesa33′′ ×41′′ field-of-view (FoV), observed with the SAURON integral-field spectrograph. We sug- with a spatial sampling of 0′.′94 × 0′.′94. This setup produces gested that the strong positive correlation between the stellar ve- 1431 spectra per pointing over the SAURON FoV. Additionally, locity (V) and the h3 Gauss–Hermite moment appears to be the a dedicated set of 146 lenses provides simultaneous sky spectra most reliable indicator for the presence of bars among all other 1.′9 away from the main field. The spectral resolution delivered bar diagnostics, even in cases with end-on orientation of the bar by the instrument is ∼4.2 A˚ (FWHM) and covers the narrow (cid:13)c 0000RAS,MNRAS000,000–000 Barsandthestellarpopulationsofbulges 3 Table1.Propertiesofoursampleofgalaxies Galaxy Sample PA Vhel MK T–type incl. σ0 Dust mcyl Bar zdisc xB zB (deg) (kms−1) (mag) (deg) (kms−1) (arcsec) (arcsec) (arcsec) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) NGC3098 S2 88.5 1397 −22.72 −1.5 90 126.2 N 0.54±0.20 N? 1.5 10 9 NGC4026 S2 177.5 985 −23.03 −1.8 84 158.1 N 0.54±0.13 Y 3.0 12 10 NGC4036 S2 261.2 1385 −24.40 −2.6 75 181.9 F 0.26±0.24 N 2.0 12 9 NGC4179 S2 142.8 13 −23.18 −1.9 86 167.5 N 0.59±0.12 Y 2.0 13 10 NGC4251 S2 99.0 1066 −23.68 −1.9 80 128.8 N 0.73±0.08 Y 2.0 12 9 NGC4270 S2 109.8 2331 −23.69 −2.0 80 139.6 N 0.58±0.12 Y 1.5 13 10 NGC4346 S2 98.8 832 −22.55 −2.0 77 127.0 N 0.63±0.11 Y 3.5 12 12 NGC4425 S2 25.8 1908 −22.09 −0.6 90 82.8 N 0.53±0.15 Y 3.0 14 8 NGC4435 S2 10.0 791 −23.83 −2.1 68 152.8 D 0.51±0.14 Y? 3.0 11 9 NGC4461 S2 8.1 1924 −23.08 −0.8 71 133.0 N 0.58±0.15 Y 3.5 13 9 NGC4474 S2 79.4 1611 −22.28 −2.0 89 87.9 N 0.55±0.17 Y? 2.0 10 9 NGC4521 S2 166.3 2511 −23.92 −0.1 90 185.8 N 0.71±0.08 Y 2.0 10 8 NGC4710 S2 27.4 1102 −23.53 −0.9 88 104.7 D 0.54±0.20 Y 3.0 19 13 NGC4762 S2 29.6 986 −24.48 −1.8 90 133.7 N 0.66±0.13 Y 1.0 10 7 NGC5103 S2 140.6 1273 −22.36 – 90 111.2 N 0.37±0.21 N 2.0 9 8 NGC5326 S1 130.0 2520 −23.77 – 65 144.9 N 0.35±0.20 N 2.5 10 8 NGC5353 S2 140.4 2198 −25.11 −2.1 80 281.2 D 0.57±0.11 Y 3.0 15 9 NGC5422 S1 152.3 1838 −23.69 −1.5 90 161.8 N 0.69±0.08 Y 2.5 16 10 NGC5475 S1 166.2 1671 −22.88 – 79 115.0 N 0.25±0.25 N 2.5 12 8 NGC5574 S2 62.7 1589 −22.30 −2.8 89 81.9 N 0.51±0.23 Y 2.0 11 9 NGC5611 S2 64.6 1968 −22.20 −1.9 74 137.4 N 0.40±0.24 N 1.5 10 9 NGC5689 S1 84.0 2160 −24.00 – 81 157.4 D 0.64±0.09 Y 3.0 17 13 NGC5707 S1 35.0 2212 −23.22 – 80 131.8 N 0.25±0.20 N 1.5 11 8 NGC5746 S1 170.0 1727 −24.99 – 81 202.8 D 0.61±0.09 Y 3.5 22 15 NGC5838 S1 40.1 1341 −24.13 −2.6 72 246.0 N 0.47±0.17 Y 2.0 11 9 NGC5854 S2 54.8 1663 −23.30 −1.1 74 104.7 N 0.38±0.26 Y 3.5 13 10 NGC5864 S2 65.6 1874 −23.62 −1.7 74 110.7 N 0.59±0.12 Y 1.5 13 10 NGC6010 S1 102.9 2022 −23.53 – 90 159.2 D 0.45±0.19 Y 3.5 11 10 NOTES:(1)Galaxyname.(2)Sample–S1:Molaeinezhadetal.(2016)sample.S2:ExtractedfromtheATLAS3Dsurvey.(3)Positionangle(N–E)ofthe dustfreeminoraxis.(4)Heliocentricvelocity,takenfromCappellarietal.(2011).ForgalaxiesnotintheATLAS3Dsample,valuesaretakenfromNED.(5) TotalgalaxyabsolutemagnitudeinK-band,takenfromPeletier&Balcells(1997)andCappellarietal.(2011),forsamplesS1andS2,respectively.(6) MorphologicalT–typefromHyperLeda(Patureletal.2003).(7)Inclinationsderivedfromthebest-fittingmass-follow-lightJAMmodelling(Cappellarietal. 2013),orifnotavailable,derivedfromdiscellipticityinR-bandfromPeletier&Balcells(1997),correctedforfinitediscthickness.(8)Centralvelocity dispersion,definedasthemaximumσwithinthebulgeanalysiswindow,fromourdata.(9)Dustfeatures–D:dustydisc,F:dustyfilament,N:none(Taken fromCappellarietal.2011).(10)Levelofcylindricalrotation(mcyl)withinthebulgeanalysiswindowanditsuncertainty.Valuesofmcylaregenerally between1(purecylindricalrotation)and0(nosignofcylindricalrotation).(11)Detectionofabar,basedonthekinematicanalysisasinPaperI.The Questionmark(?)indicatescaseswherebardetectionisuncertain.(12)Verticalextentoftheregion,closetothediscplanewhichismostlikelydisruptedby contaminationofdustand/orcentralcomponents.(13)and(14)Analysiswindowofthebulgealongthemajorandminoraxisrespectively.See§2formore details. spectral range 4800-5380 A˚. The IFU observing strategy and the easily distinguishable if it is not inclined enough or has not ver- datareductionoftheS2asdescribedinCappellarietal.(2011)is tically thickened enough to be clearly recognized above the disc similartotheS1sample,whichguaranteesthehomogeneityofdata plane, and/or beyond a possible large classical bulge. It is worth foroursample.Thestellarkinematicsdata,publishedhereforS1 notingthatthroughout thisstudy,thebulgeofagalaxyisreferred isthesameaswaspresentedinPaperI.KinematicsmapsoftheS2 totheexcessoflightthatisabovethemainexponentialdisc. samplewerepresentedindetailsinKrajnovic´etal.(2011).Inorder A major concern in such studies is the disc contamination, that to ensure meaningful and accurate stellar population parameters could severely bias the bulge morphology analysis. For this pur- all the data were Voronoi binned, following Cappellari&Copin pose,aswedidinPaperI,wehaveusedGalfit(Pengetal.2002)to (2003),toatargetSNR=60.Wedescribeourprocedurestoextract findthebestexponentialdiscmodelandsubtractitfromtheorig- thepopulationparametersin§3. inalimage.Thesesubtractedimageshavebeenusedtodelimitthe regionofinterestinourgalaxies:the’photometricbulge’.Theex- tentofthebulgeintheradial(x)andvertical(z)directionshasbeen 2.3 Samplemorphologicalandkinematicclassification evaluated by integrating the residual images, along the minor (or major)axis,anddeterminingtheregioninwhichmorethan90per 2.3.1 Morphologicalanalysis centoflightcomesfromthebulge.Wehavemadeuseofi−band imagesfromtheSloanDigitalSkysurveyDR10(Ahnetal.2014) Inourstudyofthestellarpopulationpropertiesofbulges,thechoice for this purpose. In addition, we have produced unsharp masked of highly-inclined galaxiesensures minimal contamination bythe (median filtering)imagestohighlight thehigh frequency features stellardisc.However,attheseinclinationsthebarcomponentisnot (cid:13)c 0000RAS,MNRAS000,000–000 4 Molaeinezhadetal. oftheimagesandrevealpossiblestructureswithnoradialsymme- tries.ThesearepresentedinAppendixA,together withthebulge regionobtainedfromourGalfitanalysis.Theradialandverticalex- tentofourbulges(inarcsec)arelistedinTable1. 2.3.2 Kinematicclassification Bureau&Athanassoula (2005) suggested, using N-body simula- tions, a number of major-axis stellar kinematic features that can beusedasbardiagnosticsinhighlyinclinedsystems:(1)adouble- hump rotation curve; (2) a broad central velocity dispersion peak with a plateau (and possibly a secondary maximum) at moderate radii and (3) an h profile correlated with velocity over the pro- 3 jectedbarlength.Allthesediagnosticshavebeensuccessfullyex- pandedbeyondthediscplanebyIannuzzi&Athanassoula(2015), whostudiedtheimprintsofbarsonthe2Dline-of-sightkinemat- icsofsimulateddiscgalaxies.InPaperI,wetestedthisapproach on12mid-tohighlyinclinedgalaxies,allofthemincludedinthe S1 sample, and confirmed that the correlation between the stellar lainveeroyfrseiglihatblveelboacritdyia(gVn)osatnicdttohoel.h3ThGeausussc–cHesesrmofittehipsarmamethetoedr iiss Figure 1. Distribution of the level of cylindrical rotation (mcyl) for the galaxies with positive V–h3 correlation (red histogram) and those with theabilitytounveilthehiddenbar,whenthebarisnotclearlyvis- mostlynegativeornullcorrelation(bluehistogram). ibleinphotometricdata,duetoitsorientationorstrength.Wehave takenadvantage ofthisapproachtodetect thepresenceofbarsin oursampleofgalaxies.Thediagnosticplotsshowingthedegreeof requiresthedeterminationofoftenuncertainoffsetstoaccountfor correlationcanbefoundinAppendixA. differencesinthefluxcalibrationbetweenmodelsandobservations. Inaddition,atypicalkinematicfeatureofmostbarredgalaxies The wavelength range provided by SAURON (4800–5380 A˚) withBPbulgesiscylindricalrotation(e.g.Kormendy&Illingworth constrains the set of spectral indices that can be used to estimate 1982; Bureau&Freeman 1999; Falco´n-Barrosoetal. 2006; stellar population parameters, i.e. age, metallicity, and [Mg/Fe]. Molaeinezhadetal. 2016). In such systems, the mean stellar Wehaveusedthethreeavailableindicesthataremeasuredacross rotationspeedshowslittledifferenceatdifferentheightabovethe the full field for all galaxies, namely the Hβo age indicator (i.e. disc plane. In Paper I, we introduced a method to quantify the theoptimizedHβ indexofCervantes&Vazdekis(2009)),aswell levelofcylindrical rotationinbulges. Thisquantity(m )shows asthemetallicity-and[Mg/Fe]–sensitiveindicesFe5015andMgb cyl theimportance of thisproperty inbulgesand isgenerally avalue (Trageretal.1998).Noticethatthe[Mg/Fe]abundanceratioisof between+1whichindicatespurecylindricalrotationand0where particularinterestforthepresentstudy,asitmeasures–forafixed there is no sign of cylindrical rotation within the bulge analysis IMF – the different timescale of the ejection of magnesium and window. Figure 1 shows the distribution of m for the galaxies ironintotheinterstellarmedium.Thus,the[Mg/Fe]isbelievedto cyl in our sample, separated into two categories of galaxies with bea”chemicalclock”toestimatestar-formationtimescalesforthe positive V–h3 correlations and those with negative or null V–h3 bulkofastellarpopulation(e.g.Thomasetal.2011;Vazdekisetal. correlations. Asexpected, thegalaxieswithrelativelyhigh values 2015;Mart´ın-Navarro2016). of m are mostly those showing positive correlation between Our analysis relies on α–MILES stellar population mod- cyl V and h3, while those with lower mcyl values do not show a els, from Vazdekisetal. (2015, hereafter V15). The models, con- significant correlation. The m values and related uncertainties structedfromMILESstellarspectra,havebeencomputed,inaself- cyl arepresentedinTable1. consistentmanner,atbothsolarscaleandfor[α/Fe]=+0.4,overa Asourgoalistohavetwocleanclassesofbarredandunbarred rangeofmetallicities,withtheaidoftheoreticalstellarspectrafrom systems, weusethecombination of bothbar diagnostictools,the Coelhoetal. (2005, 2007), and using BaSTi (either scaled-solar correlationV–h3andthelevelofcylindricalrotation,todefinethe or α–enhanced) isochrones from Pietrinfernietal. (2004, 2006). galaxies belonging to each class. In this study, ”barred” galaxies Notice that throughout the present work, we adopt the notation refertothosesystemsthatshowremarkablypositivecorrelationbe- [Mg/Fe], rather than [α/Fe], as our estimate of abundance ratios tweenV andh3 intheareasabovethemid-planewithinthebulge reliesentirelyonMg-andFe-sensitivefeatures,i.e.Mgb5177and analysiswindow,whilethosegalaxieswithnon-positiveV–h3and Fe5015. m <0.55aremarkedas”unbarred”system.Notsurprisingly,all The α–MILES models cover the (optical) MILES spectral cyl galaxieswithBPbulgesinoursamplebelongtothe”barred”class. range (λλ ∼ 3500–7400 A˚), in the metallicity and age range of −2.27 6 [Z/H] 6 +0.4 and 0.03 6 t 6 14 Gyr, respectively. For bothscaled-solar and α–enhanced SSPmodels, wehavepro- duced an interpolated grid of model line-strengths with a step of 3 STELLARPOPULATIONQUANTITIES ∼0.06Gyrinage,and∼0.017dexinmetallicity.Thescaled-solar Besidesstellarkinematics,wehavealsomeasuredline-strengthin- andα–enhancedgridsaretheninterpolated/linearly-extrapolatedto dicesoftheS1andS2galaxiesintherecentlydefinedLineIndex cover the [Mg/Fe] range from −0.1 to +0.8 dex, with a step of System (LIS) LIS-8.4A˚ (Vazdekisetal. 2010, hereafter VAZ10). 0.009dexin[Mg/Fe].Noticethattheextrapolationdoesnotaffect Thismethodhastheadvantageofcircumventingtheuseoftheso- atallourresults,astheintegrated,averageprofilesof[Mg/Fe]for calledLick/IDSfittingfunctionsforthemodelpredictions, which barredandunbarredgalaxiesexceedsonlyslightly(by∼0.05dex; (cid:13)c 0000RAS,MNRAS000,000–000 Barsandthestellarpopulationsofbulges 5 NGC5689 (a) 77 0.0 3.0 6.0 9.012.0 66 z_b (arcsec) 1.00 55 1.50 44 2.00 3.00 33 -0.66 7.1000.00 -0.35-0.25 14.00 0.0 06.1 05.26 22 11 22 33 44 55 (b) 5 14.00 0. 470.00 4 0.26 0.15 0.06 3 -0.25 -0.35 2 -0.66 1 -0.96 -1.26 1 2 3 4 5 6 7 Figure2.Leftcolumn,panel(a):Theindex-indexdiagramofHβoversus[MgFe50]’forallVoronoibins(greyfilledcircles)withinthebulgeanalysiswindow ofNGC5689,arepresentativebarredgalaxywithBPbulgeinoursample.Leftcolumn,panel(b):TheMgbversusFe5015diagramforthisgalaxy.Overlaidare theMILESSSPmodelsfordifferentagesandmetallicities.Thegreenandredgridscorrespondtothescaled-solarandα–enhancedSSPmodels,respectively. ColouredsymbolsindicateaverageVoronoibinvaluesatvariousheightsfromthemajoraxis.Middlecolumn:Theabsorptionline-strengthmapsofHβo,Mgb andFe5015,respectively(fromtoptobottom).RightColumn:computedSSP-equivalentmapsofage,metallicityand[Mg/Fe],respectively. seeFig.7)thevalueof[Mg/Fe]=+0.4forwhichV15α–enhanced green and red grids correspond to the scaled-solar and [Mg/Fe]– modelshavebeencomputed. enhancedSSPmodels,respectively.Themiddlecolumnofthefig- Foreachspectrum,wederivetheSSP-equivalentage,metal- ureshowstheHβo,Fe5015,andMgbindexmaps,whiletheright licity,and[Mg/Fe],byminimizingthefollowingequation: column presents the age, metallicity and [Mg/Fe] maps obtained withtheprocedureoutlinedabove. (EW −EW )2 χ2 = X obs,i mod,i , (1) σ2 i obs,i 3.1 Comparisonofthemeanstellarpopulationparametersof where the index i runs over the available indices (Hβo, Fe5015, bulgesofbarredandunbarredgalaxies and Mgb), EW and EW are observed and model line- obs,i mod,i strengths, while σ denote uncertainties on observed line- Themajorityofstellarpopulationstudiesinbulgesfocusonthein- obs,i strengths. The minimization is performed over the grid of SSP tegratedpropertiesinsideacertainaperture,nearthecentre.These modelpredictions,withvaryingage,metallicity,and[Mg/Fe](see studiescouldthereforestronglydependonthepresence(ornot)of above).Uncertaintyonbest-fitparametersareestimatedfromN = differentsubcomponents (e.g.ayoungstellarcentraldisc,nuclear 1000bootstrapiterations,wherethefittingisrepeatedaftershifting cluster; see Peletieretal. 2007). Thiseffect is particularly impor- observedlinestrengthsaccordingtotheiruncertainties. tantinhighly-inclinedgalaxies,wherethemaindisccontamination Figure 2 shows the line-strength indices and resulting stel- isaproblematicissue.Toavoidthesedifficulties,welimitouranal- lar population parameters for NGC5689, a representative barred ysisofthebulgestellarpopulations toawindow wellbeyond the galaxywithBPbulgeinoursample. Intheleftcolumn, panel (a) discplane.Morespecifically, theregionsof interestinour bulges showstheHβo versus[MgFe50]’ forallVoronoi bins(greyfilled arelimitedtotheclean(e.g.dustfree)sideofthegalaxies,between circles)withinthebulgeanalysiswindow ofNGC5689. Coloured −xBand+xBalongthemajoraxisandzdisctozBalongtheminor symbols indicate representative values of Voronoi bins at various axis of galaxies. The z indicates the vertical extent of the re- disc heightsfromthemajoraxis(see§3.2formoredetails).Overlaidare gion,closetodiscplanethatmightbecontaminatedbydustand/or theMILESSSPmodelsfordifferent agesandmetallicities.Panel centralcomponents(evaluatedbyvisualinspectionofthevelocity (b) presents the Mgb versus Fe5015 diagram for this galaxy. The maps). (cid:13)c 0000RAS,MNRAS000,000–000 6 Molaeinezhadetal. Figure3.DistributionofmeanSSP-equivalentage,metallicityand[Mg/Fe] Figure4.DistributionofmeanSSPequivalentage,metallicityand[Mg/Fe] withinthebulgeanalysis windowofbarred(redhistogram)andunbarred versusthedynamicalmass(MJAM)estimates,withinoneeffectiveradius (bluehistogram) galaxies inoursample.PK−S indicates theprobability (Reff)ofbothclassesofbarredandunbarredgalaxiesinoursample.Over- thatthetwodistributionsaredrawnfromthesamepopulation. plotedasgreyfilledcirclesshowstellarpopulationsofgalaxies,measured inthecentralregionsofthegalaxies,withinacircularaperturewitharadius ofzdisc.Horizontaldottedlinerepresentsthemeanstellarpopulations of thecentralapertures. In order to investigate stellar populations of bars and the in- fluence of bars on bulges of nearby galaxies, we compare the key stellar population parameters within the bulge analysis win- dependentofthepresenceofthebar,aremoremetalrich,younger dowsofoursample.Figure3showsthedistributionsofthemean and less [Mg/Fe]–enhanced in the central regions. Jablonkaetal. SSP-equivalentage,metallicityand[Mg/Fe]forourtwoclassesof (2007)arguedthattheouterregionsofbulges,whicharegenerally bulges.AKolmogorov–Smirnovtest(hereafterK–S)indicatesthat olderandmoremetalpoor thancentralregionsconsistofthefirst the difference between the two populations of bulges is not sig- stars to form, supporting those scenarios in which star formation nificant.ThisisinagreementwithresultsofJablonkaetal.(2007) proceeds fromthe outer partsinwards. Here, weprovide adiffer- whofoundnosignificantdifferencebetweenthestellarpopulations entexplanationandinterprettheobservedstellarpopulationprop- ofearlytype(S0–S(B)b)edge-onbarredandunbarredgalaxies. ertiesinthecentralregionsofbulgesasduetohigher(andpossibly Akeyaspecttoconstrainbulgeformationscenariosisthecor- unsought) contamination level of maindisc components instellar relationofthestellarpopulationpropertiesofbulges,withtheto- populationpatternsofbulges. tal mass of their parent galaxies. Figure 4 shows the distribution of the mean SSP-equivalent age, metallicity and [Mg/Fe] of the 3.2 Verticalgradients bulges (within the bulge analysis window) versus the dynamical mass (M ) estimates from (Cappellarietal. 2013), within one Exploringthepossibleinfluenceofbarontheformationandevo- JAM effective radius (R ) of galaxies. The central stellar populations lution of bulges is not possible without understanding the varia- eff (within a circular aperture with a radius of z ) are also shown tions of stellar population parameters with radius. Here, we have disc (grey filled circles). As this figure shows, no significant correla- takenadvantageofthespatiallyresolvedstellarpopulationmapsto tionsofthestellarpopulationparameterswiththedynamicalmass study thevertical gradients of the SSP-equivalent age, metallicity arefound,exceptformetallicity,wheremoremassivegalaxieshar- and[Mg/Fe]inoursample. bouronaveragemoremetalrichbulges.Ingoodagreementwiththe Forthispurposewehavedefined,ontheclean(i.e.dust-free) resultsbyJablonkaetal.(2007),mostofbulgesinoursample,in- sideofourgalaxies,asetofpseudo-slits(from−xBto+xB),par- (cid:13)c 0000RAS,MNRAS000,000–000 Barsandthestellarpopulationsofbulges 7 alleltothemajoraxisofthebulge,atvariousheightsfromthema- galaxieswithlowestinclinationsinoursample.Wehaveconfirmed joraxis.Theslitscoverthewholeverticalextentofthebulgefrom that theaverageand slope valuesobtained for thefull sampleare z = 0toz = zB.Theouterslits(withrespecttothemajoraxis) consistentwiththosefortheleastinclinedgalaxiesalone. arewidertocompensate forthesmallnumber ofcontributeddata Theseresultsleaveuswithaveryintriguingpicturewithvery points(Voronoibins),sothatthenumberofbinswithinallslitsis similar average stellar population values for barred and unbarred almostthesame.Theverticalprofilesareconstructedbyfindingthe galaxies, consistent mean age profiles, slightlysteeper metallicity meanvalueofstellarpopulationparameterinquestionineachslit gradientsinbarredgalaxies,butmostnotablyadrasticdifferentin atdifferentheightsfromthediscplane,normalisedtozB.There- thevertical[Mg/Fe]gradientbetweenthem.Inthenextsectionwe latederrorsarecalculatedasthestandarderrorofthemeansineach willdescribetheimplicationsoftheseresultsandoutlinepossible pseudo–slit.Inordertoevaluatetheverticalgradientsinage,metal- scenariosthatmightexplainthem.Wewillalsocomparethemwith licityand[Mg/Fe]inourbulges,wefitanerror-weightedstraight thosefromtheclosestbarredgalaxyknown:theMilkyWay. line to the individual profiles excluding the regions close to disc plane (z ), which are most likely affected by contamination of disc dustand/orcentralcomponents.Theslopeofthelines(verticalgra- 4 DISCUSSION dients)andtherelatederrorsforallgalaxiesinoursamplearelisted inTable2. Theresultspresentedhereposeachallengetomanytheoriesabout Figure5demonstratesthemeansandtheverticalgradientsof the formation and evolution of barred galaxies. A common ap- the SSP-age, metallicity and [Mg/Fe] of bulges as a function of proachtoinvestigatetheirinfluenceontheformationandevolution level of cylindrical rotation(mcyl) for all galaxies inour sample. ofbulgesistocomparethestellarpopulationpropertiesofgalaxies Theseresultsrevealatrendtowardssteeperpositiveverticalgradi- withandwithoutbars.Ifbarsfollowadifferentevolutionarypath, entsin[Mg/Fe]forbulgeswithhigherlevelofcylindricalrotation, wewouldexpect different abundance patternsinbulgesof barred whicharemostlybarredsystems(lowerrightpanel).However,as andunbarredgalaxies. noted earlier, cylindrical rotation cannot be considered a reliable Most bulges studies in the literature show similar and rela- kinematicpropertytoidentifythefullpopulationofbarredgalax- tivelybroaddistributionsinmeanSSP-equivalentagesandmostly ies,asthisfeaturestronglydependsonbarorientationandgalaxy no vertical gradients in age, independent of the bulge clas- inclination. sification (e.g. Falco´n-Barrosoetal. 2002; Jablonkaetal. 2007; IntheleftpanelsofFig.6,wepresenttheverticalprofilesof Peletieretal.2007;Morellietal.2008).Thisisinagreementwith theSSPequivalentmetallicityandageforthe2classesofbulgesin earlier findings for early-type galaxies (e.g. Mehlertetal. 2003; oursample,togetherwiththeerror-weightedaveragedprofiles.To Sa´nchez-Bla´zquezetal.2006).Furthermore,theflatnessofthever- derive theaveraged profiles,theindividual data pointsarebinned ticalageprofilesinbarredgalaxiesinour samplecouldinprinci- sothateachnon-overlappingbincontainsthesamenumberofdata plesupportthoseformationscenariosofbulgesinwhichthebulge points.Theerrorweightedaverageandthecorrespondingerrorin formswithverylittleinfluencefromthedisc(e.g.mergersormono- eachbinaredeterminedusingthemaximum-likelihoodmethod(see lithic collapse), as expected in bulges of unbarred galaxies. It is Taylor1982),sothatdatapointswithsmalleruncertaintiesineach worthnotingthatthefactthatoldstellarpopulationsareobservedin bincontributedmoretothecomputedmeanthanthosewithlarger bulgesofgalaxiesdoesnotnecessarilymeanthatbarsareoldstruc- errorbars.Therightpanelsshowthedistributionofaveragevalues tures,asbarmighthaveformedrecentlyoutfromoldstars,coming in each class. The individual SSP-equivalent population profiles, fromthedisc(Sa´nchez-Bla´zquezetal.2011).Thefactthatbarred alongtheminoraxisforallbulgesinour samplearepresented in galaxiescontainoldstellarpopulationsis,however,inmarkedcon- Appendix B. The figure shows that the mean age profiles for our trasttotheinternalsecularevolutionscenarioexpectedinthesesys- sample are rather consistent, in mean values and slope, for both tems,inwhich thebarred bulgesareproduced through radial and barredand unbarred galaxies.Themetallicitygradient isnegative verticaltransportofdiscmaterial(whichistypicallyyoung),dueto in both cases too, although it appears to be somewhat flatter for instabilities and resonances (see Martinez-Valpuesta&Shlosman unbarred systems. Mean values, though, are very similarfor both 2004; Moorthy&Holtzman 2006; Kormendy&Kennicutt 2004). typesofbulges. Originally,itwasthoughtthatifbulgesformedsecularly,starsthat More interestingly, as shown in Fig. 7, the distribution havebeenscatteredfurthestfromthediscaretheoldest starsand and gradients of [Mg/Fe] are remarkably different for barred thereforeapositiveagegradientisexpected(Freeman2008).Con- and unbarred galaxies. [Mg/Fe] gradients are mostly positive in sideringthestellar agepatternof bulgesinour sample, andmore barred galaxies, while for unbarred ones they are mostly flat. A specificallybarredgalaxies,onecouldna¨ıvelyconcludethatapure Kolmogorov-Smirnov (K-S) test gives a probability of 10−4 that secular evolution scenario via the bar buckling process, which is vertical [Mg/Fe] gradients of the barred and unbarred bulges are thewidelyacceptablemechanismtoformbulgesinbarredgalaxies, drawnfromthesamedistribution. cannotexplaintheobservedpropertiesofbulges,becausethenthe starsatlargescaleheightsshouldbeyounger.Theonlysolutionwe A major concern, in this kind of studies is that our gradient see that bars were formed long timeago, and were not destroyed measurements could beseverely biased against the physical scale later on. The disc itself was later rejuvenated with newly formed height of the bulges. To test this, we have computed the correla- starsfrominfallinggasclosetothemidplane,asweareseeingin tion between the [Mg/Fe] gradients scaled to the physical size of our own Milky Way. This leads to these bulges having old ages thebulgealongtheminoraxis,inkpcunitsandnormalisedtozB, overall,exceptinthemidplane,whereingeneralyoungerstarsare inarcsec.Ourresults,notshownhere,areveryconsistent inboth beingseen. caseswithoutmajordifferences,exceptforfewcaseswithrelatively Both sample of bulges show negative metallicity gradients smallbulges.Wehavealsoinvestigatedwhetherthegradientsmea- along the minor axis, a behaviour that have been formerly inter- suredinourgalaxiesdependonthepotentiallevelofcontamination preted as a direct evidence against the secular evolution scenario introducedbytheprojectionofmaindiscbehindthebulgesinthose ofbulges.However, morerecent numericalmodelsshow that,the (cid:13)c 0000RAS,MNRAS000,000–000 8 Molaeinezhadetal. Figure5.Meanandvertical gradients ofthestellar age,metallicity and[Mg/Fe]ofbulgesasafunctionofthelevelofcylindrical rotation(mcyl)forall galaxiesinoursample.Ineachpanel,theverticaldashedlinesindicatethemeanlevelofcylindricalrotationforbarredandunbarredbulges,respectively,while thehorizontaldashedlinesshowtheaveragedstellarpopulationproperties(upperpanels)andverticalgradients(lowerpanels),respectively. Barred Galaxies Unbarred Galaxies 11.0 10.5 yr]) 10.0 ge[ 9.5 a g( 9.0 o l 8.5 8.0 0.2 -0.0 H] M/ -0.2 [ -0.4 -0.6 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 z/z_B z/z_B Figure6.Leftpanels:TheverticalprofilesoftheSSPequivalentmetallicityandageforthe2classesofbulgesinoursample.Thegreylinescorrespondtothe individualprofilesforeachgalaxy.Foreachclassofbulges,theindividualdatapointsarebinnedsothateachnon-overlappingbincontainsthesamenumber ofdatapoints.Hatchedregionsindicatetheerror-weightedaveragedprofilesandthecorrespondinguncertainty,followingthemaximum-likelihoodmethod, ineachbin.Linesrepresentthebestlinearfitstothedatasets,whiledatapointsbelongtotheregions,closetothediscplanearenotincludedinthefitting process.Rightpanels:Distributionoftheverticalgradientsinageandmetallicityforbothclassesofbulgesinoursample.Thegradientshavebeencalculated index/zB,wherethezBistheverticalextentofthebulge/barinarcsec. negative metallicity gradient in secularly formed bulges can be resultappearstorevealthatbothkindsofbulgeshavedifferentori- consistentlyreproduced,dependingontheoriginaldiscradialgra- gins.Bulgesofunbarredgalaxiesareoftendenotedas’classical’or dients(Martinez-Valpuesta&Gerhard2013),verticalgradientsof ’elliptical-like’ bulges based on the remarkable resemblance they thecompositionoftwo-componentdisc(Bekki&Tsujimoto2011) haveinmanyproperties(frommorphology tostellarpopulations) or both (DiMatteoetal. 2014). Other bulge formation scenarios, to elliptical galaxies. If, as suggested in the literature, they have mergersandmonolithiccollapse,alsopredictsuchmetallicitygra- similarformationandevolutionchannels,thenbulgesofunbarred dients (Eggenetal. 1962; Larson 1974; Arimoto&Yoshii 1987). galaxies have been formed through merger of monolithically col- Therefore, thenegative metallicitygradients along theminor axis lapsedcloudsathighredshifts(Baughetal.1996),similartolow- ofbulgesofbarredandunbarredgalaxiesdonotallowustosingle luminosityellipticals(e.g.Moorthy&Holtzman2006).Inthatsit- outthebulgeformationandevolutionaryscenarios. uation the material that forms those bulges comes typically from The most striking feature found in our data is that bulges of low-masssystemswherethe[Mg/Fe]istypicallylow.Whilethere barred galaxies are in general more [Mg/Fe]–enhanced at higher issomelevelofenrichmentinthisprocess(thatiswhythetypical latitudesthan similar regions of bulges in unbarred systems. This [Mg/Fe]valuesarehigherthanindiscmaterial),thefinalmixtureof (cid:13)c 0000RAS,MNRAS000,000–000 Barsandthestellarpopulationsofbulges 9 lutionof thebar fractionindiscgalaxiesasafunctionofredshift (seeShethetal.2008;Melvinetal.2014). TheMilkyWay(MW)isaparticularcaseofabarredsystem wherewecandirectlyresolvethedistributionofthedifferentkine- maticandchemicalstellarpopulations.Thebulgeisknowntocon- sistof predominantly oldstars(∼10Gyr)withaverysmall frac- tion (up to 3.5%) of the stars younger than 5Gyr (Clarksonetal. 2008,2011).Theoriginoftheseintermediate-agestarsisstillun- derdebateanddifferentmechanismshavebeensuggestedtointer- prettheobservational results(seeGonzalez&Gadotti2016,fora detailedreviewofallarguments).Themetallicitydistributionfunc- tionofstartsintheGalacticbulgeisbroad,asymmetric(Rich1990; Ibata&Gilmore1995;Zoccalietal.2003,2008)withaclearver- tical gradient of ∼0.6dex/kpc decreasing towards high latitudes (e.g. Zoccalietal. 2008; Nessetal. 2013). The vertical metallic- itygradientoftheMWbulgeisaround-0.6dex/kpc(Zoccalietal. 2008), -0.06 dex/deg=-0.43 dex/kpc (Gonzalezetal. 2013) and - Figure7.Theintegrated,errorweightedaverageprofilesofthevariationsof 0.45 dex/kpc (Nessetal. 2013). In any case, the three measure- the[Mg/Fe],withincreasingheight(z)fromthediscplaneforbothbarred mentsarehigherthantheaveragevaluewemeasuredinbulgesof andunbarredgalaxiesinoursample.Theinsethistogramshowsthedistri- barredgalaxies(∇[M/H]=-0.3dex/kpc).Inprinciple,itcouldseem butionoftheverticalgradientsin[Mg/Fe]forbothclassesofbulgesinour thattheMWissomehowdifferenttotherestofthegalaxies.This sample.Gradientshavebeenmeasuredintherangezdisc <z <zB,nor- malizedbytheverticalextentofthebulge(zB)inarcsec.PKS givesthe differencecouldbepartiallyduetodifferentmethodweusetomea- probabilitythatthetwodistributionsaredrawnfromthesamepopulations, suretheverticalgradients.IncaseofMW,wearealwaysusingthe asderivedfromaK-Stest. minoraxis,butinexternalgalaxiesweareaveragingovertheentire extent of the bulge (including those bulge regions away from the minoraxis).Inordertounderstandthedifferencewehaveusedthe simulations by Martinez-Valpuesta&Gerhard (2013) to compute allthatmaterialislowerthantheonetheycouldhaveiftheywere thegradients along theminor axisand alsoaveraging withinlon- made of primordial stars (those that formed quickly at very early gitude−5◦ < l < +5◦.Theverticalgradientsdecreaseby28%, epochs). from -0.46 dex/kpc to -0.33 dex/kpc respectively. Moreover, it is Forbarredgalaxies, thehigh [Mg/Fe]abundance ratiofound worthnotingthat,whenlookingatbulgesofexternalgalaxieswe at high latitudes implies that stars at higher z in barred galaxies are integrating foreground and background light, and in the MW formed much faster compared to those in the inner regions. The wealwayshaveamagnitudelimitavoidingmanybackgroundstars. reasonwhythisisnotclearlyseenintheagegradients,whereone ThesetwoexplanationscanexplainthedifferenceswiththeMW. mayexpectamorepositivegradientinbarred,relativetounbarred Finally,theα-elementabundancesoftheMWbulgeshowthat systems, is that beyond ∼6 Gyr it is more and more difficult to stars with [Fe/H]< −0.3 are more α-enhanced than metal-rich distinguishformationeventsontime-scalesoftheorderoflessthan stars([Fe/H]> −0.3)(Cunha&Smith2006;Lecureuretal.2007; 1Gyr.Ouragediagnostic,theHβoindex,willnotgiveusthatlevel Richetal.2007).WhilethemeasurementsintheMWcomefrom of precision with the typical observational uncertainties we have individual stars,theyseemtobeconsistent withthosefromunre- in the Hβo equivalent width (see Cervantes&Vazdekis 2009, for solvedstellarpopulationsinourgalaxysample.Inthisrespect,as details). shownforthestellarkinematicsinPaperI,theMWisnodifferent Sinceyoungpopulationsarenotobservedathighzinbarred fromthetypicalbarredgalaxyinthenearbyUniverse. galaxies, it is clear that the bar was made of stars that formed On the theoretical side, there has been mounting evidence longtimeago.Thealternativeisthatthebarwillselectivelypush supporting the long-lasting nature of bars (e.g. Athanassoula old material vertically from a disc made of young and old stars 2003; Martinez-Valpuestaetal. 2006; Debattistaetal. 2006; (like the ones we observed in the nearby Universe), which is not Athanassoulaetal. 2013). More recently Debattista et al. (2016, easytoexplain.Thehigh[Mg/Fe]observedinthoseouterregions private communication) studied, in the context of the MW anal- seem to imply that whatever process that formed these stars with ysis, the influence of an evolving long lasting bar on the stellar such high [Mg/Fe] must have formed very quickly. We are thus populations in the bulge dominated region, offering a somewhat saying that the bars we observed today (at least in our sample) alternative scenario. By means of a pure N–body simulation, were bars that formed long time ago. Numerical simulations of they show that the stellar populations of disc material that have barred galaxies suggest they are made from disc instabilities and differentinitialin-planekinematicsseparatewhenabarforms.The gothroughaparticularlyimportant bucklingphasewherealotof populationthathashotterradialkinematicsalsohashottervertical material gets pumped-up to higher latitudes above the disc plane kinematics and becomes the vertically thick and box-shaped part (see Martinez-Valpuesta&Shlosman 2004; Athanassoula 2005; of the bar, whilethe radiallycooler populations isverticallythin. Debattistaetal. 2006; Martinez-Valpuestaetal. 2006; Sahaetal. In this scenario, the declining metallicity profile towards higher 2012).Iftrue,ourobservationsaresettinganimportantconstraint latitudesandverticallyrisingabundanceratiointheGalacticbulge onthetimethatprocessoccurred.Themeanagevaluesandespe- are interpreted as a result of this separation of initially nearly ciallythepositive[Mg/Fe]gradientssuggestthisprocesstookplace co-spatialpopulations.Thisisaninterestingevolutionaryscenario, veryearlyoninthelifeofgalaxies.Theseresultsgoalongthelines but how the original superposition of differentiated discs could ofrecentworksintheliteraturepointingattheearlyformationof be formedisstill unknown. Whilethisscenario isalso consistent bars(seee.g.Seideletal.2015,2016)andthoseanalysingtheevo- with our observations of more distant galaxies, we are not yet in (cid:13)c 0000RAS,MNRAS000,000–000 10 Molaeinezhadet al. position to robustly disentangle the different chemo-kinematic workingonthispaper.A.M.alsoacknowledges theIsaacNewton componentsofourgalaxies,asdoneintheMW.Infact,whilewe GroupofTelescopes(ING),theInstitutodeAstrof´ısicadeCanarias agree on the final set of observed properties in the galaxies we (IAC) and the Kapteyn Astronomical Institute for hospitality and see today, we provide a different explanation to the way we get support while this paper was in progress. The authors acknowl- to those properties. Current attempts to do so are limited to very edgesupportfromtheSpanishMinistryofEconomyandCompet- dramatic cases of large-scale counter-rotating disc components itiveness (MINECO) through grants AYA2009-11137, AYA2016- (e.g. Coccatoetal. 2013; Johnstonetal. 2013). Meanwhile, more 77237-C3-1-P and AYA2014-58308-P. Funding for SDSS-III has chemo-dynamical simulations are also necessary to confirm this been provided by the Alfred P. Sloan Foundation, the Participat- issueandexplainpossiblesegregationprocesses,inducedbybars, ingInstitutions,theNationalScienceFoundation,andtheU.S.De- toproduce the verticalgradients instellarpopulation of secularly partment of Energy Office of Science. The SDSS-III web site is formedbulges. http://www.sdss3.org/. 5 CONCLUSIONS Thispaper investigates theimprintsof barson the stellarpopula- tionproperties of bulgesand morespecifically, thevertical gradi- entsofstellarpopulations.Forthispurpose,wehavecarriedouta detailed analysis of thestellar age, metallicityand [Mg/Fe] of 28 highly-inclined (i > 65o) disc galaxies (i > 65o) , from S0 to S(B)c,observedwiththeSAURONintegralfieldspectrograph.The choiceofhigh-inclinationgalaxiesensuresminimalcontamination bythestellardisc.Followingtheapproach,appliedinPaperI,the sampleisdividedintotwocleansamplesofbarred(n=21)andun- barredgalaxies(n=7).Comparingthestellarpopulationproperties and vertical gradients of bulges inthese twoclasses, wefind that while: • The distribution of the mean stellar age, metallicity and [Mg/Fe]inthebulgesofbarredandunbarredgalaxiesarenotsta- tisticallydistinct. • Galaxiesinoursamplespanawiderangeofverticalgradients inSSP-equivalentstellaragesandthereisnosignificantdifference betweenbarredandunbarredgalaxies. • Bothclassesofbulgespresentnegativemetallicityandthegra- dientsarenotstatisticallydistinctforbarredandunbarredgalaxies. AsimilarpatternhasbeenreportedfortheBPbulgesofMW. Weseeastrongdifferenceintheverticalgradientin[Mg/Fe]in barredandunbarredgalaxies.Theverticalgradientsin[Mg/Fe]for barredgalaxiesaremostlypositive,whileforunbarredgalaxiesthe profileisalmostflat.Inotherwords,bulgesofbarredgalaxiesare ingeneralmore[Mg/Fe]–enhancedathigherlatitudesthanregions, close to disc plane and therefore demonstrate a positive [Mg/Fe] gradientsalongtheminoraxis. Suchhighlevelof[Mg/Fe]ratioinbarredbulgestogetherwith the relatively old population of bar material, as old as unbarred bulgesinoursample,suggeststhatbarsarelong-lastingstructures madeofoldmaterial.Theoriginofsuch[Mg/Fe]–enhancedstarsis notyetclear,butasimpleinterpretationwouldbethatthebuckling phaseleadingtotheformationofthebartookplacelongtimeago (>10Gyr).Moredetailedchemo-dynamicalstudiesarestillneeded toshedmorelightonthisissue.Whatitseemsclearisthattheposi- tiveverticalgradientin[Mg/Fe]isacharacteristicfeatureofbarred bulgesandthereforedifferentevolutionmechanismsarerequiredto interpretthestellarpopulationdifferencesobservedbetweenbulges ofbarredandunbarredgalaxies. ACKNOWLEDGMENTS A.M. wishes to thank the School of Astronomy, IPM and the Iranian National Observatory (INO) for providing support while (cid:13)c 0000RAS,MNRAS000,000–000