The VIRUS-P Exploration of Nearby Galaxies (VENGA): The stellar populations and assembly of NGC 2903's bulge, bar, and outer disc
Andreia Carrillo, Shardha Jogee, Niv Drory, Kyle F. Kaplan, Guillermo Blanc, Tim Weinzirl, Mimi Song, Rongxin Luo
MMNRAS , 1–14 (2018) Preprint 17 February 2020 Compiled using MNRAS L A TEX style file v3.0
The VIRUS-P Exploration of Nearby Galaxies (VENGA):The stellar populations and assembly of NGC 2903’s bulge,bar, and outer disc
Andreia Carrillo , , (cid:63) Shardha Jogee , Niv Drory , Kyle F. Kaplan , Guillermo Blanc , ,Tim Weinzirl , Mimi Song , Rongxin Luo Department of Astronomy, University of Texas at Austin, 2515 Speedway, Stop C1400, Austin, TX 78712-1205, USA Large Synoptic Survey Telescope Corporation Data Science Fellow McDonald Observatory, The University of Texas at Austin, 1 University Station, Austin, TX 78712, U.S.A. SOFIA-USRA, NASA Ames Research Center, MS N232-12, Moffett Field, CA 94035-1000, USA The Observatories of the Carnegie Institution for Science, 813 Santa Barbara St., Pasadena, CA 91101, USA Departamento de Astronomia, Universidad de Chile, Castilla 36-D, Santiago, Chile School of Physics and Astronomy, The University of Nottingham, University Park, Nottingham, NG7 2RD, UK Astrophysics Science Division, Goddard Space Flight Center, Code 665, Greenbelt, MD 20771, USA Shanghai Astronomical Observatory, 80 Nandan Road, Shanghai 200030, China
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
We study the stellar populations and assembly of the nearby spiral galaxy NGC 2903’sbulge, bar, and outer disc using the VIRUS-P Exploration of Nearby Galaxies IFSsurvey. We observe NGC 2903 with a spatial resolution of 185 pc using the MitchellSpectrograph’s 4.25 arcsec fibres at the 2.7 Harlan J. Smith telescope. Bulge-bar-discdecomposition on the 2MASS K s -band image of NGC 2903 shows that it has ∼ ∼ . , suggestive of a disky bulge. We performstellar population synthesis and find that the outer disc has 46% of its mass in stars > <
10% in younger stars. Its stellarbar has 65% of its mass in ages 1-5 Gyr and has metallicities similar to the outerdisc, suggestive of the evolutionary picture where the bar forms from disc material.Its bulge is mainly composed of old high-metallicity stars though it also has a smallfraction of young stars. We find enhanced metallicity in the spiral arms and centralregion, tracing areas of high star formation as seen in the H α map. These results areconsistent with the idea that galaxies of low bulge-to-total mass ratio and low bulgeS´ersic index like NGC 2903 has not had a recent major merger event, but has insteadgrown mostly through minor mergers and secular processes. Key words: galaxies: formation – galaxies: fundamental parameters – galaxies: spiral
Observational constraints on the assembly history of dif-ferent components of a galaxy are essential to understandhow quickly and through what mechanisms galaxies grow.The outer disc, primary bar, and different types of centralbulges (classical, disky and boxy/peanut) are fundamentalstellar components of spiral galaxies that define the Hubblesequence (Kormendy 1993; Athanassoula 2003; Sheth et al.2005). These components are thought to be built through (cid:63)
E-mail: [email protected] different evolutionary pathways and their morphology, kine-matics, star formation rate (SFR), stellar ages, stellar metal-licities, and star formation history (SFH), provide importantconstraints on the assembly history of these components andtheir host galaxy.Simulations have suggested that the outer disc growsfrom minor mergers or gas accretion through cold anddense filaments (Somerville & Dav´e 2015). After which,bar-formation could be spontaneously or tidally induced(Romano-D´ıaz et al. 2008). The bar then efficiently redis-tributes mass and angular momentum in the galaxy (Shlos-man & Noguchi 1993; Knapen et al. 1995; Athanassoula © a r X i v : . [ a s t r o - ph . GA ] F e b A. Carrillo et al. σ ) rather than ordered motion(characterized by circular velocity, v), therefore having lowv/ σ . A classical bulge is thought to be built generally byviolent relaxation in dry or modestly gas-rich galaxy majormergers. Disky pseudobulges have high v/ σ and are domi-nated by ordered motion, reminiscent of galactic discs. Theyare believed to have been built via gas that is funneled tothe central region of a galaxy (Kormendy 1993; Kormendy& Kennicutt 2004a; Jogee et al. 2005; Athanassoula 2005;Weinzirl et al. 2009). Boxy (or peanut) pseudobulges formfrom stellar bar material scattered to larger scale heightsvia instabilities or vertical resonances (Combes et al. 1990;Athanassoula 2005). It is important to note that a galaxydoes not necessarily have just one type of bulge. For exam-ple, a disky pseudobulge can coexist with a boxy pseudob-ulge (e.g., Barentine & Kormendy 2012) or a classical bulge(e.g., Erwin et al. 2015).High resolution images have allowed for morphologi-cal and structural studies that place some constraints ona galaxy’s assembly history (Kormendy 1993; Kormendy& Kennicutt 2004a; Weinzirl et al. 2009, 2014; Fisher &Drory 2008; Jogee et al. 2004; Marinova & Jogee 2007; Jo-gee et al. 2009; Weinzirl et al. 2011). Integral Field Spec-troscopy (IFS or IFU) surveys at high spatial resolution canstrongly complement morphological studies by constrainingthe stellar populations and ionized gas properties of differentgalaxy components. One such survey is the VIRUS-P Ex-ploration of Nearby Galaxies (VENGA) IFU survey (Blancet al. 2013) that includes observations of 30 nearby spiralsfrom their central regions to their outskirts. VENGA com-plements many recent IFU surveys (MaNGA, Bundy et al.2015; CALIFA, S´anchez et al. 2012; SAMI, Croom et al.2012), that target large samples of galaxies out to largedistances, have spatial resolution of typically several kpc,and have primarily provided important statistical insights onglobal properties of galaxies. With VENGA, we can specif-ically explore the assembly of the outer disc, primary bar,and different types of central bulges.We are doing a pilot study on the stellar populationsand mass build-up of the bulge, the bar, and the outer discfor one of the galaxies in the VENGA survey, NGC 2903. Itis an isolated spiral galaxy (Irwin et al. 2009) with circum-nuclear star formation, a bar, and grand-design spiral arms.With this pilot study, we aim to advance the field of stellarpopulations and aid our understanding of the evolution inthe different parts of a galaxy because of the following advan-tages: (1) NGC 2903’s low distance (8.9 Mpc) leads to a su-perb spatial resolution (185 pc), and the large coverage fromthe bulge to the outer disc allows for the study of spatially-resolved galaxy properties; (2) its structural properties areinteresting: it is a SAB(rs)bc spiral (de Vaucouleurs et al.1995) with a strong bar in the near-infrared (NIR elliptic-ity of 0.88 and a bar-to-total (Bar/T) light ratio = 0.06;see Table 1), a low bulge-to-total (B/T) light ratio of 0.06,and low S´ersic index, n ∼ < ∼
2, and have builttheir outer discs as well as part of their bulge at later timesthrough gas accretion, minor mergers, and secular processes.This is supported by more recent studies on bulge formationwith hydrodynamical simulations (e.g. Gargiulo et al. 2019;Tacchella et al. 2019) that show the importance of secularevolution and significance of disc formation in building lowS´ersic index and low B/T bulges; (3) it is one of the barredgalaxies in Kaplan et al. (2016) that also uses VENGA tolook at gas-phase metallicity with 7 different indicators pro-viding us with emission line analysis results to compare to;(4) it is a well-studied galaxy and has a wealth of ancil-lary data (e.g., 2MASS near-infrared among others) whichwe can take advantage of to do a morphological study incombination with the spectral analysis.This paper is organized as follows: In Section 2 we de-scribe observation, data reduction and data cube building,we outline in Section 3 the bulge-bar-disc decomposition us-ing GALFIT, in Section 4 we explain the stellar populationfitting procedure, in Section 5 we present mass-build up his-tories for the bulge, bar, and outer disc, in Section 6 we dis-cuss the implications of our derived stellar population mapsand star formation histories, and lastly, we summarize ourresults in Section 7.
VENGA (Blanc et al. 2013) includes 30 nearby spirals ob-served over ∼
150 nights with the Mitchell Spectrograph onthe 2.7 meter Harlan J. Smith telescope (Hill et al. 2008).The survey benefits from high spatial sampling of typicallya few 100 pc resolution (fibre size of 4.25 (cid:48)(cid:48) ) with 80% ofthe galaxies less than 20 Mpc away, large coverage from thebulge to the outer disc, broad wavelength range (3600-6800˚A), and medium spectral resolution (120 km/s at 5000 ˚A orR ∼ (cid:48)(cid:48) × (cid:48)(cid:48) per pointing as seen in Figure 1. Observationswere done with a blue setup (3600-5800 ˚A) and a red setup(4600-6800 ˚A) to increase the wavelength coverage of thesurvey without compromising the medium spectral resolu-tion. These were then later combined during the reductionprocedure.These observations result in data with high signal-to-noise ratio (SNR) of up to ∼
300 in the central region. Wehave a spatial resolution of 185 pc at the distance of NGC2903 (8.9 Mpc), enough to resolve the bulge area with mul-tiple spaxels.
MNRAS000
MNRAS000 , 1–14 (2018)
GC 2903’s bulge, bar, and outer disc Figure 1.
SDSS image of NGC 2903 (inverted colormap) with3 Mitchell Spectrograph pointings, each with a field of view of110 (cid:48)(cid:48) × (cid:48)(cid:48) . The VENGA pointings encompass the bulge, bar,and part of the outer disc of NGC 2903. We refer the reader to Blanc et al. (2013) for details on theVENGA data reduction. To summarize, we used the pipelineVACCINE (Adams et al. 2011) which does bias and darksubtraction, flat fielding, wavelength calibration, cosmic rayrejection, and sky subtraction after which, the astrometrywas determined and the data was flux-calibrated. This wasa two-step process: first by using a standard star for rela-tive flux calibration and next by using SDSS g and r bandimages, convolved to the fiber size of Mitchell spectrograph,for absolute flux-calibration and astrometry. Lastly all sci-ence frames were combined into a datacube (two versionsfor each galaxy with linearly-sampled or logarithmically-sampled wavelengths) that contains the wavelength, flux,flux error, spatial position, and instrumental resolution.Additionally, we Voronoi-binned (Cappellari & Copin2003) the data to increase the SNR to at least 50 per bin.This had a net effect of combining spaxels on the edges ofthe field of view. In the end, we had a total of 6,812 spaxelsavailable for full-spectrum fitting for NGC 2903. We follow the methods Weinzirl et al. (2009) in decomposingthe galaxy light into the bulge, bar, and disc components.This method specifically uses GALFIT (Peng et al. 2010) tofit different S´ersic profiles to the galaxy light. We use the2MASS K s -band image from 2MASS Large Galaxy Atlas(Jarrett et al. 2003) for our decomposition because it besttraces stellar mass and the stellar mass-to-light ratio in the K s -band remains roughly constant with stellar populationage. We also determined a PSF image from stars in the field,to be fed into GALFIT.We do the following steps based on Weinzirl et al.(2009): we first fit with a single S´ersic component which de-termines the total luminosity, the center of the galaxy, andthe inclination of the outer disc which are then held fixed for the later stages. Next we fit two components, a S´ersiccomponent representative of the bulge, and an exponentialdisc. We then proceed to a three component fit with an ex-ponential disc and two S´ersic components (for the bulge andfor the bar) using the best fit model from the two compo-nent (bulge+disc) fit as initial guesses on the parameters.The models and the residuals divided by the data are shownin Figure 2. We decide whether the 2-component (bulge-disc) or 3-component (bulge-bar-disc) decomposition is bet-ter based on the prescription from Section 3.3 in Weinzirlet al. (2009). An indication that the 3-component fit is mostappropriate for NGC 2903 can be seen in the residual di-vided by the data map (see bottom row of Figure 2), wherelighter regions indicate lower residual over data and there-fore a better fit. Since the galaxy has a bar as seen in Figure1, the residuals in the one component and two-componentfits contain prominent and m = symmetric light distribu-tion pattern due to the unsubtracted bar light. However,this residual pattern is significantly reduced with the threecomponent (bulge+bar+disc) fit. Other light patterns in theresidual map remain, though these are due to the spiral armswhich are unaccounted for in our light decomposition. Table1 shows the component-to-total ratios and statistical resultsfrom this GALFIT analysis of NGC 2903.We then fed the 3-component best-fit model back toGALFIT to get the image blocks for each light-component(see Figure 3). For a given position, we now know how muchlight is contributed by the bulge, bar, and disc components.We assign the NGC 2903 VENGA spaxels by matching theirposition to the GALFIT decomposition and determiningwhich component contributes >
50% of the light for thatarea. This breakdown of spaxel assignment is illustrated inFigure 4 where nine bulge spaxels are shown in orange, 640bar spaxels are in blue, 6,062 outer disc spaxels are in purple,and 101 unassigned spaxels are in gray. Though we know howmuch each component contributes to the light, we do notknow the breakdown of the stellar populations that shouldbe ascribed to each component at a given spaxel. We there-fore chose to assign a spaxel to one component as long asthat component dominates more than half of the light inthat region. Going forward, we use these spaxel member-ships to assign a given spaxel to the bulge or bar or outerdisc components, to determine their associated properties.The unassigned spaxels (1.5%) were excluded from the indi-vidual component mass build-up histories. We do not use amore stringent spaxel assignment (i.e. > Fitting physically-motivated combinations of stellar popula-tions to IFU data is challenging, and so parameterization ofthe SFH is usually implemented (see discussions in Conroy2013). However, numerous codes do exist for fitting stellarpopulations to IFU data (STARLIGHT: Cid Fernandes et al.2005, PIPE3D: S´anchez et al. 2016, FIREFLY: Wilkinsonet al. 2017) that have nonparametric SFH. The advantageof a nonparametric solution is that it does not assume a de-clining, rising, or constant SFH, and is therefore a better
MNRAS , 1–14 (2018)
A. Carrillo et al.
Figure 2.
GALFIT decomposition of NGC 2903 using 2MASS K S -band image. The top row shows the best fit GALFIT model, and thebottom row shows the residual (data-model) divided by the data. The grayscale applies to the bottom row where lighter shade indicatessmaller residual fraction and therefore a better fit. The columns are arranged as one component fit ( χ /DOF = 1.55), 2 component fitwith the bulge and outer disc decomposition ( χ /DOF = 1.17), and three component fit with the bulge, bar, and outer disc decomposition( χ /DOF = 1.11) following Weinzirl et al. (2009). The residuals for the 3-component decomposition shows that it best fits the lightprofile of NGC 2903. Table 1.
Results of GALFIT decomposition on 2MASS K s -band imageFIT χ χ /DOF R e or h ( (cid:48)(cid:48) , kpc) S´ersic index n b/a PA X/T(1) (2) (3) (4) (5) (6) (7) (8) (9) S´ersic 294,523 1.55 173.51, 7.55 1.790 ± ± Bulge 221,773 1.17 13.94, 0.61 0.390 ± ± ± ± Bulge 202,364 1.11 13.25, 0.58 0.266 ± ± ± ± ± ± ± χ foreach step (4) reduced χ (5) Effective radius or scale length in case of the disc measured in arcseconds (6) S´ersic index n (S´ersic1963) (7) semi major axis to semi minor axis ratio i.e. 1 - ellipticity, e (8) Position angle measured in degrees (9) component tototal light ratio. approach in getting the SFH of a spiral galaxy, which mayhave a bursty or stochastic star formation history.For fitting the stellar populations in this study, we usedthe Penalized Pixel-Fitting (pPXF) method (Cappellari &Emsellem 2004; Cappellari 2017). Though usually employedfor getting gas and stellar kinematics, it has also been usedin a number of stellar populations studies (e.g., Onoderaet al. 2012; McDermid et al. 2015). In general, pPXF derivesthe mass-weighted, best-fit template as a linear combinationof stellar populations with different ages and metallicities.Regularization is suggested for using pPXF in stellar popu-lation studies as deriving the SFH from a grid of templatesis an ill-conditioned inverse problem (see discussion in Sec-tion 3.5 from Cappellari 2017) and requires further assump-tions. This is especially important for low SNR data like other IFU surveys of nearby galaxies where over-fitting tonoisy data could be an issue. A regularized solution gives thesmoothest star formation history consistent with the data.However, with our high SNR data, the non-regularized bestfit model from pPXF performed well in our cross-validationtests and produced similar SFH trends to the regularizedcase (tested on 120 spaxels), as we collapse our age resolu-tion into four bins in the end. The differences in the averageages and metallicities between the non-regularized and regu-larized fits are also quite small i.e. . ± . Gyr for age and . ± . dex for metallicity. From these tests, we decided toproceed on using pPXF without regularization which ben-efits from being computationally less expensive while stillproducing physically-motivated results.Throughout the full-spectrum fitting process, we mask MNRAS000
Results of GALFIT decomposition on 2MASS K s -band imageFIT χ χ /DOF R e or h ( (cid:48)(cid:48) , kpc) S´ersic index n b/a PA X/T(1) (2) (3) (4) (5) (6) (7) (8) (9) S´ersic 294,523 1.55 173.51, 7.55 1.790 ± ± Bulge 221,773 1.17 13.94, 0.61 0.390 ± ± ± ± Bulge 202,364 1.11 13.25, 0.58 0.266 ± ± ± ± ± ± ± χ foreach step (4) reduced χ (5) Effective radius or scale length in case of the disc measured in arcseconds (6) S´ersic index n (S´ersic1963) (7) semi major axis to semi minor axis ratio i.e. 1 - ellipticity, e (8) Position angle measured in degrees (9) component tototal light ratio. approach in getting the SFH of a spiral galaxy, which mayhave a bursty or stochastic star formation history.For fitting the stellar populations in this study, we usedthe Penalized Pixel-Fitting (pPXF) method (Cappellari &Emsellem 2004; Cappellari 2017). Though usually employedfor getting gas and stellar kinematics, it has also been usedin a number of stellar populations studies (e.g., Onoderaet al. 2012; McDermid et al. 2015). In general, pPXF derivesthe mass-weighted, best-fit template as a linear combinationof stellar populations with different ages and metallicities.Regularization is suggested for using pPXF in stellar popu-lation studies as deriving the SFH from a grid of templatesis an ill-conditioned inverse problem (see discussion in Sec-tion 3.5 from Cappellari 2017) and requires further assump-tions. This is especially important for low SNR data like other IFU surveys of nearby galaxies where over-fitting tonoisy data could be an issue. A regularized solution gives thesmoothest star formation history consistent with the data.However, with our high SNR data, the non-regularized bestfit model from pPXF performed well in our cross-validationtests and produced similar SFH trends to the regularizedcase (tested on 120 spaxels), as we collapse our age resolu-tion into four bins in the end. The differences in the averageages and metallicities between the non-regularized and regu-larized fits are also quite small i.e. . ± . Gyr for age and . ± . dex for metallicity. From these tests, we decided toproceed on using pPXF without regularization which ben-efits from being computationally less expensive while stillproducing physically-motivated results.Throughout the full-spectrum fitting process, we mask MNRAS000 , 1–14 (2018)
GC 2903’s bulge, bar, and outer disc Figure 3.
Light contributions from the bulge(left), bar(center),and disc(right) from the 3-component GALFIT decomposition.The colorbar indicates the brightness counts. out wavelength ranges that have poor sky subtraction. Wealso mask higher order Balmer lines (up to H θ ) to improvethe fit, especially for areas with high star formation activity.In determining the best fit combination of stellar popula-tions, we normalize the continuum with a Legendre polyno-mial (LP) Later on however, we do derive the reddening foreach spaxel from the continuum (see Section 4.2) in order toget the intrinsic luminosity and therefore the intrinsic massof the galaxy. We first used stellar population templates from the MILEStemplate library from Vazdekis et al. (2010) that werealready provided within the pPXF package distribution.Specifically, these spectra are based on the MILES empiricalstellar library with a Salpeter (1955) initial mass function(IMF), and Padova isochrone tracks (Girardi et al. 2000).The spectra have FWHM of 2.51 ˚A that cover 6 metallici-ties from [M/H] = -2.3 to [M/H] = +0.22 and a full age range(from 63.1 Myr to 15.85 Gyr). We convolved the templateswith the wavelength-dependent instrumental dispersion ofour data.However, using these templates for NGC 2903, a star-forming galaxy, forced the best fit spectra to contain a siz-able portion of the mass to come from the youngest templateat the lowest metallicity. This is unphysical unless pristine,un-enriched gas is funneled into the galaxy. The more likelyscenario is that young stars are present and pPXF erro-neously tries to fit the bluest template, which in MILEScorrespond to the youngest, most metal-poor templates.To alleviate this problem, we used the more exten-sive library Base GM (Rosa Gonzalez-Delgado priv. comm.)adopted from the code STARLIGHT. This template libraryalso includes the Vazdekis et al. (2010) single stellar popu-lations (SSP) with the addition of younger templates fromGonz´alez Delgado et al. (2005). These younger models con-sist of 15 ages from 1 to 63 Myr at four metallicities from[M/H] = -0.71 to [M/H] = + 0.22, and uses Geneva isochrone
Figure 4.
We assign spaxels to the bulge, bar, and outer disc ifthese components contribute to >
50 % of the total light at eachlocation. The spaxels are shown as orange for the bulge, bluefor the bar, and purple for the outer disc. The grey areas havespaxels with no dominating (i.e. > tracks (Schaller et al. 1992; Schaerer et al. 1993; Charbonnelet al. 1993) as well as a Salpeter IMF.Base GM templates has a FWHM of 6 ˚A, which matchesthe CALIFA resolution (S´anchez et al. 2016) and is higheror lower than our data’s spectral resolution (4.6˚A to 6.1˚A),depending on the wavelength range under consideration.pPXF is ran with the prescription that the template libraryhas similar or better resolution than the observed data. Wetherefore had to degrade our data to the same resolutionas the templates and re-measure the final FWHM from theconvolved arc lamps. Not doing this properly could resultinto inaccurate velocity dispersion determination.With the convolved and updated set of templates, the 63Myr lowest metallicity template was no longer being fit andinstead, as expected, there are contributions from templates < α -enhancement.Alpha elements (O, Mg, Si, S, Ca, Ti) are dispersed intothe interstellar medium by core-collapse supernovae as aresult of massive star death, but are then diluted by su-pernova Type Ia (SNe Ia) events that do not contribute asmuch α elements. An α -abundant stellar population has anisochrone track shifted to cooler temperatures (or reddercolors). Early-type galaxies and galaxy regions that haveundergone rapid star formation have been found to have en-hanced α -abundances, having very little successive star for- MNRAS , 1–14 (2018)
A. Carrillo et al. mation events to dilute them. We therefore investigated thecontribution of α -enhanced stellar populations in the differ-ent components of NGC 2903.The alpha-enhanced SSPs (Vazdekis et al. 2015) usethe same MILES empirical stellar library, Salpeter IMF,and BaSTI isochrones (Pietrinferni et al. 2004, 2006) with α abundance, [ α /Fe] = +0.4. This is different from thescaled solar model in that for the scaled solar model,[Fe/H] = [M/H] while for the alpha-enhanced model,[Fe/H] = [M/H] - 0.75[ α /Fe]. The templates obtained alsohave FWHM = 6 ˚A to match the other templates. See Figure5 for a pPXF fit to a sample spaxel, employing the procedureoutline in this section. In the previous section, we fit away the continuum in order toget the stellar populations based mostly on the presence andshapes of lines. However, in trying to get a mass for a spaxeland more importantly, a total mass of contributing spaxelsto each galaxy component, we need to get the intrinsic, dust-corrected continuum light from the galaxy.We derived E ( B − V ) from pPXF by fitting the best-fit spectra derived from the LP fit with a Calzetti curve(Calzetti et al. 1994). pPXF does this by determininghow much the continuum shape has changed from an un-attenuated spectra i.e. the default for the template spectra.We then deredden the observed spectra, which we multiplyby the M/L in the SDSS g -band and corresponding weightderived from the LP fit, to get the g -band mass for thatspaxel for each contributing stellar population. We discussthis procedure in greater detail in Section 5.To check for consistency, we also derived the E ( B − V ) from the Balmer decrement method. We apply Equation 1(adopted from Dom´ınguez et al. 2013) using H α and H β fluxes measured from pPXF to get the E ( B − V ) for eachspaxel, assuming H α /H β = 2.87 (Case B recombination at , K and − cm − , Osterbrock & Bochkarev 1989).Here, k λ is defined by the Calzetti curve. E ( B − V ) = − . lo g (cid:18) [ H α / H β ] obs . (cid:19) k ( λ H α ) − k ( λ H β ) (1)However, one must be wary that the reddening derivedfrom the gas is not necessarily the same as the reddeningfrom the continuum. One complication to the Balmer decre-ment method is that H α /H β is a function of temperatureand density and we assume that the emission is coming froma star-forming region. However, diffuse ionized gas (DIG)which has non-local origins, also contribute H α (see Kaplanet al. 2016 for how they dealt with DIG emission in a sampleof VENGA galaxies). Figure 6 shows the extinction mapsderived from both methods as well as the non extinction-corrected H α map of NGC 2903. Both extinction maps showareas with higher values that correspond to where currentstar formation is happening as seen in the H α map and wherethe dust lanes are in the optical (Figure 1). We use the red-dening derived from the continuum to get the intrinsic lightfrom the galaxy to be consistent with the stellar populationsanalysis. The traditional way of determining mass is by using an in-frared image, where the dominant light source are K giants.This justifies our light decomposition in Section 3 where weuse the 2MASS K s -band data in decomposing the galaxyinto the bulge, bar, and disc. But to be more consistentwith how we derived the stellar population properties fromthe template fitting, we compute the mass from the M/Lratio for each stellar population (of different metallicities)at different ages, by knowing how much they contribute tothe total light.We use the M/L in the SDSS g band since its wholerange is within the observed wavelength of VENGA andwithin the wavelength range of our SSP templates. TheMILES stellar populations (Vazdekis et al. 2010) has magni-tudes, colors, M/L, and masses tabulated in their website for all their single stellar populations in different filters, in-cluding SDSS. As for the GRANADA-based templates, wederived the M/L ourselves using the fact that the spectraare in units of L (cid:12) / M (cid:12) / ˚A. We convolve the spectra with theSDSS g -band transmission curve and integrate. We also takeinto account that the contributing mass to the spectra is acombination of the stellar remnants and stars, and is smallerthan the initial mass of the cloud (at 1 M (cid:12) ). Therefore theper M (cid:12) is actually interpreted as per M (cid:12) × remaining massfraction left in remnants and stars. For example, this frac-tion is 1 i.e. no mass lost yet for a 10 Myr population and ∼ g band. They are at most, different by only 3%, withour values being higher. For self-consistency, we use the de-rived M/L instead of the tabulated M/L for the MILES tem-plates and warn the reader that the mass we compute maybe overestimated by 3%.Once we have the M/L in the g band for all the tem-plates and the dust-corrected light from the galaxy as out-lined in Section 4.2, we can then multiply them by each otherand get the mass of the spaxel if 100% of its light came fromthat stellar population. This has to be multiplied by theweights determined from the stellar population fitting andthen added up to get the total mass of the spaxel in g-band.Using the spaxel membership from Section 3, we com-bine the masses in four different age bins for spaxels thatbelong to the same component, color-coded by the contri-bution of different metallicities. Figure 7 shows these massfraction build up for the whole galaxy, the bulge, the bar, andthe outer disc. We group the ages accordingly: 1-63 Myr forstellar populations coming from the GRANADA templatesand whose light is dominated by current star formation, 63Myr - 1 Gyr for the younger MILES templates that are alsodominated by light from O and B stars, 1 - 5 Gyr for thestellar populations dominated by A type stars, and > http://miles.iac.es/ MNRAS000
A. Carrillo et al. mation events to dilute them. We therefore investigated thecontribution of α -enhanced stellar populations in the differ-ent components of NGC 2903.The alpha-enhanced SSPs (Vazdekis et al. 2015) usethe same MILES empirical stellar library, Salpeter IMF,and BaSTI isochrones (Pietrinferni et al. 2004, 2006) with α abundance, [ α /Fe] = +0.4. This is different from thescaled solar model in that for the scaled solar model,[Fe/H] = [M/H] while for the alpha-enhanced model,[Fe/H] = [M/H] - 0.75[ α /Fe]. The templates obtained alsohave FWHM = 6 ˚A to match the other templates. See Figure5 for a pPXF fit to a sample spaxel, employing the procedureoutline in this section. In the previous section, we fit away the continuum in order toget the stellar populations based mostly on the presence andshapes of lines. However, in trying to get a mass for a spaxeland more importantly, a total mass of contributing spaxelsto each galaxy component, we need to get the intrinsic, dust-corrected continuum light from the galaxy.We derived E ( B − V ) from pPXF by fitting the best-fit spectra derived from the LP fit with a Calzetti curve(Calzetti et al. 1994). pPXF does this by determininghow much the continuum shape has changed from an un-attenuated spectra i.e. the default for the template spectra.We then deredden the observed spectra, which we multiplyby the M/L in the SDSS g -band and corresponding weightderived from the LP fit, to get the g -band mass for thatspaxel for each contributing stellar population. We discussthis procedure in greater detail in Section 5.To check for consistency, we also derived the E ( B − V ) from the Balmer decrement method. We apply Equation 1(adopted from Dom´ınguez et al. 2013) using H α and H β fluxes measured from pPXF to get the E ( B − V ) for eachspaxel, assuming H α /H β = 2.87 (Case B recombination at , K and − cm − , Osterbrock & Bochkarev 1989).Here, k λ is defined by the Calzetti curve. E ( B − V ) = − . lo g (cid:18) [ H α / H β ] obs . (cid:19) k ( λ H α ) − k ( λ H β ) (1)However, one must be wary that the reddening derivedfrom the gas is not necessarily the same as the reddeningfrom the continuum. One complication to the Balmer decre-ment method is that H α /H β is a function of temperatureand density and we assume that the emission is coming froma star-forming region. However, diffuse ionized gas (DIG)which has non-local origins, also contribute H α (see Kaplanet al. 2016 for how they dealt with DIG emission in a sampleof VENGA galaxies). Figure 6 shows the extinction mapsderived from both methods as well as the non extinction-corrected H α map of NGC 2903. Both extinction maps showareas with higher values that correspond to where currentstar formation is happening as seen in the H α map and wherethe dust lanes are in the optical (Figure 1). We use the red-dening derived from the continuum to get the intrinsic lightfrom the galaxy to be consistent with the stellar populationsanalysis. The traditional way of determining mass is by using an in-frared image, where the dominant light source are K giants.This justifies our light decomposition in Section 3 where weuse the 2MASS K s -band data in decomposing the galaxyinto the bulge, bar, and disc. But to be more consistentwith how we derived the stellar population properties fromthe template fitting, we compute the mass from the M/Lratio for each stellar population (of different metallicities)at different ages, by knowing how much they contribute tothe total light.We use the M/L in the SDSS g band since its wholerange is within the observed wavelength of VENGA andwithin the wavelength range of our SSP templates. TheMILES stellar populations (Vazdekis et al. 2010) has magni-tudes, colors, M/L, and masses tabulated in their website for all their single stellar populations in different filters, in-cluding SDSS. As for the GRANADA-based templates, wederived the M/L ourselves using the fact that the spectraare in units of L (cid:12) / M (cid:12) / ˚A. We convolve the spectra with theSDSS g -band transmission curve and integrate. We also takeinto account that the contributing mass to the spectra is acombination of the stellar remnants and stars, and is smallerthan the initial mass of the cloud (at 1 M (cid:12) ). Therefore theper M (cid:12) is actually interpreted as per M (cid:12) × remaining massfraction left in remnants and stars. For example, this frac-tion is 1 i.e. no mass lost yet for a 10 Myr population and ∼ g band. They are at most, different by only 3%, withour values being higher. For self-consistency, we use the de-rived M/L instead of the tabulated M/L for the MILES tem-plates and warn the reader that the mass we compute maybe overestimated by 3%.Once we have the M/L in the g band for all the tem-plates and the dust-corrected light from the galaxy as out-lined in Section 4.2, we can then multiply them by each otherand get the mass of the spaxel if 100% of its light came fromthat stellar population. This has to be multiplied by theweights determined from the stellar population fitting andthen added up to get the total mass of the spaxel in g-band.Using the spaxel membership from Section 3, we com-bine the masses in four different age bins for spaxels thatbelong to the same component, color-coded by the contri-bution of different metallicities. Figure 7 shows these massfraction build up for the whole galaxy, the bulge, the bar, andthe outer disc. We group the ages accordingly: 1-63 Myr forstellar populations coming from the GRANADA templatesand whose light is dominated by current star formation, 63Myr - 1 Gyr for the younger MILES templates that are alsodominated by light from O and B stars, 1 - 5 Gyr for thestellar populations dominated by A type stars, and > http://miles.iac.es/ MNRAS000 , 1–14 (2018) GC 2903’s bulge, bar, and outer disc Figure 5. pPXF output for spaxel 3124 in NGC 2903. Top panel: observed VENGA data (black) with the model for the continuum(red), model for the gas flux (pink), combined gas and continuum models (orange), the residuals(green), and wavelength ranges notincluded in the fit(blue). We zoom into a smaller portion of the spectra (bottom panel) to highlight how well the model spectra (orange)reproduces the observed spectra (black). Middle 3 panels: contributions from different stellar populations in a grid of [M/H] (dex) vs logAge (yr) for the MILES alpha-enhanced templates (second panel, from the top), MILES solar-abundance templates (third panel), andyoung GRANADA templates (fourth panel). Take note that each sets of templates have different heat scales for their contribution tothe bestfit template. This figure highlights pPXF’s ability to reproduce the observed spectra from a combination of templates that spana wide range in ages and metallicities.MNRAS , 1–14 (2018)
A. Carrillo et al.
Figure 6. E ( B − V ) determined from Balmer decrement (left) using emission line flux from the gas, E ( B − V ) determined from fitting thestellar continuum shape in pPXF (center), and non extinction-corrected H α flux ( erg s − cm − ˚A − ) map derived from the simultaneousgas-fitting with the stellar population-fitting in pPXF (right). We note that the areas with high extinction in both E ( B − V ) maps tracesimilar regions i.e. areas with higher star formation as seen in the H α map. based on the lifetime of A-type stars whose age we can tellfrom the Balmer lines and Balmer Break.We derived masses for NGC 2903, listed in Table 2,from light-weighted and mass-weighted stellar populationfitting. The light-weighted masses are lower which is ex-pected, as younger stellar populations are weighed more inthis case which have lower mass-to-light ratios. Because thetotal light-weighted mass of NGC 2903 is smaller than massmeasurements in the literature (i.e. 21% of mass from Leeet al. 2009) and the mass-weighted stellar population fittingfrom this study (22%), we do not discuss these results inthe context of the mass growth of each galaxy component(Section 6.1). One major advantage of our spatially resolved IFU-basedstudy is that we can dissect the galaxy into its outer disc,bar, and bulge, and analyze the stellar populations of eachcomponent. Figure 7 shows the age, metallicity, and mass ofthe stellar populations present in the whole galaxy, as wellas in its outer disc, bar, and central bulge. The y-axis showsthe stellar mass fraction as determined in Sections 4 and5. The x-axis shows the four age bins used for the fits, asdiscussed in Section 5.The hashed marks show the contribution from alpha-enhanced stellar populations, and the color bar below thefour panels show the metallicity ([M/H]) for the solar abun-dance templates and the alpha-enhanced ([Mg/Fe]=0.4)templates. One can see that at smaller lookback times (i.e.,for younger stars), the contribution from alpha-enhanced stellar populations decreases. This is consistent with chemi-cal evolution models where the α elements produced in core-collapse supernovae are diluted by SNe Ia events.We next discuss the properties of the stars in the outerdisc, bar, and bulge of NGC 2903 and the implications forthe assembly history of NGC 2903. The bottom right panel of Figure 7 showsthe stellar populations of the outer disc of NGC 2903. Theouter disc contributes to 72% of the total stellar mass of thewhole galaxy, 16% lower than the Disk/T from the GALFITdecomposition. This is because of the absolute nature of ourspaxel assignment to only one component, when in fact therecould be superposition of different components in a givenregion. Specifically, as seen in Figure 3, the outer disc lightdistribution peaks in the central region but these spaxels areassigned to either the bulge or the bar in Figure 4. The outerdisc has 46% of its stellar mass in the form of stars of age > MNRAS000
Figure 6. E ( B − V ) determined from Balmer decrement (left) using emission line flux from the gas, E ( B − V ) determined from fitting thestellar continuum shape in pPXF (center), and non extinction-corrected H α flux ( erg s − cm − ˚A − ) map derived from the simultaneousgas-fitting with the stellar population-fitting in pPXF (right). We note that the areas with high extinction in both E ( B − V ) maps tracesimilar regions i.e. areas with higher star formation as seen in the H α map. based on the lifetime of A-type stars whose age we can tellfrom the Balmer lines and Balmer Break.We derived masses for NGC 2903, listed in Table 2,from light-weighted and mass-weighted stellar populationfitting. The light-weighted masses are lower which is ex-pected, as younger stellar populations are weighed more inthis case which have lower mass-to-light ratios. Because thetotal light-weighted mass of NGC 2903 is smaller than massmeasurements in the literature (i.e. 21% of mass from Leeet al. 2009) and the mass-weighted stellar population fittingfrom this study (22%), we do not discuss these results inthe context of the mass growth of each galaxy component(Section 6.1). One major advantage of our spatially resolved IFU-basedstudy is that we can dissect the galaxy into its outer disc,bar, and bulge, and analyze the stellar populations of eachcomponent. Figure 7 shows the age, metallicity, and mass ofthe stellar populations present in the whole galaxy, as wellas in its outer disc, bar, and central bulge. The y-axis showsthe stellar mass fraction as determined in Sections 4 and5. The x-axis shows the four age bins used for the fits, asdiscussed in Section 5.The hashed marks show the contribution from alpha-enhanced stellar populations, and the color bar below thefour panels show the metallicity ([M/H]) for the solar abun-dance templates and the alpha-enhanced ([Mg/Fe]=0.4)templates. One can see that at smaller lookback times (i.e.,for younger stars), the contribution from alpha-enhanced stellar populations decreases. This is consistent with chemi-cal evolution models where the α elements produced in core-collapse supernovae are diluted by SNe Ia events.We next discuss the properties of the stars in the outerdisc, bar, and bulge of NGC 2903 and the implications forthe assembly history of NGC 2903. The bottom right panel of Figure 7 showsthe stellar populations of the outer disc of NGC 2903. Theouter disc contributes to 72% of the total stellar mass of thewhole galaxy, 16% lower than the Disk/T from the GALFITdecomposition. This is because of the absolute nature of ourspaxel assignment to only one component, when in fact therecould be superposition of different components in a givenregion. Specifically, as seen in Figure 3, the outer disc lightdistribution peaks in the central region but these spaxels areassigned to either the bulge or the bar in Figure 4. The outerdisc has 46% of its stellar mass in the form of stars of age > MNRAS000 , 1–14 (2018)
GC 2903’s bulge, bar, and outer disc Table 2.
Total masses for each galactic component of NGC 2903 from mass-weighted and light-weighted stellar population fittinggalaxy component mass-weighted ( M (cid:12) ) % of total MW mass light-weighed ( M (cid:12) ) % of total LW masswhole galaxy . × . × . × . × . × . × . × . × Figure 7.
Clockwise from top left: stellar mass fraction for the whole galaxy, the bulge, the outer disc, and the bar, of NGC 2903 in 4age bins, color-coded by the contribution of different metallicities. The vertical axis shows the mass fraction of the mass of the galaxy orgalaxy component, indicated on the top left of each panel. The contribution of the alpha-enhanced templates are shown as the hashedpart of the bar graphs. The color bar below the panels shows the corresponding metallicity ([M/H]) for the solar abundance templates(top) and the alpha-enhanced ([ α /Fe]=0.4) templates. recently or over an extended period of time (Kereˇs et al.2005; Dekel et al. 2009); and b) the remaining substantialfraction of the outer disc may have been built via older( > > The outer disc and stellar bar:
As shown in thebottom left panel of Figure 7), the stellar bar contributes to20% of the total stellar mass of the galaxy, and it is about 3 times less massive than the outer disc. We measure 14%more than the Bar/T (0.06) in the GALFIT decomposition.This could be due to our relatively lenient criterion onspaxel assignment (i.e. component dominance by > %) asdiscussed in Section 3. The outer disc and bar have 46%and 35% of their respective stellar masses made of starsof age > MNRAS , 1–14 (2018) A. Carrillo et al. from disc material (Miller & Prendergast 1968; Combes &Sanders 1981).We perform a more detailed exploration of what onemight expect by looking at cosmological simulations of barformation by Romano-D´ıaz et al. (2008). These simulationsshow that at high redshifts (e.g., z ∼ ∼ The bulge and outer disc:
As shown in the topright panel of Figure 7, the central bulge has 4% of thetotal stellar mass of the galaxy.This is comparable to whatGALFIT gives (5.7%). Most (75%) of the stars in thebulge are older than 5 Gyr, and it has a sharply decliningfraction of younger stars. This star formation history isreminiscent of that of early type galaxies. The presence ofa small fraction (below 10%) of young stars (63 Myr to 1Gyr old) is consistent with gas inflow driven by the bar andother mechanisms building a younger stellar population inthe central part of the galaxy. The presence of old, highmetallicity stars is intriguing and will be discussed in § > < Gyr. The chemical evolution of thebulge is also different from the bar and the disc, with alarge fraction of low metallicity, high- α stars at > < . , such as NGC 2903 (which has B/T ∼ ∼ ∼
10 Gyr).Instead, such spirals have likely built their outer discsas well as part of their bulge at later times through gasaccretion, minor mergers, and secular processes.More recently, there have been studies with hydrody-namical simulations of galaxies that focus on disc and bulgeformation. Gargiulo et al. (2019) used the Auriga simula-tions (Grand et al. 2017) to look at galaxies with S´ersicindex n < . < B/T < . . Comparisons with theirgalaxies that have ∼
20% of the bulge particles at ages be-tween 1-5 Gyr (as we see in Figure 7 for the bulge) showlow B/T (i.e. < . ) and growth through minor mergers andsecular processes. Tacchella et al. (2019) used IllustrisTNGhydrodynamical simulations (Pillepich et al. 2018) to studythe spheroidal and disc components of galaxies. They findthat similar-mass galaxies to NGC 2903 have bulges thatgrow very little between the epoch of disc formation (z ∼ ∼ ∼ From the stellar population synthesis and best-fit combina-tion of stellar populations for each spaxel, we derived mass-weighted and light-weighted age and metallicity maps asshown in Figure 8. Both the age and metallicity are higherfor the mass-weighted maps than the light-weighted maps.This physically makes sense as older stars at a given metal-licity and higher-metallicity stars at a given age have highermass-to-light ratios and would therefore dominate the mass-weighted maps. On the other hand, the light-weighted mapsshow lower values for the average age and metallicity for aspaxel. Younger stars dominate the light for a given stel-lar population, therefore the light-weighted age is biased to-wards younger ages. Also quite noticeable is the enhancedmetallicity, both in the mass-weighted and light-weightedmaps, for the spiral arms and central region, areas withhigher star formation.To look into these patterns further, we made light-weighted averaged age maps for stellar populations with ages < Myr. We also compare it to the H α map derived fromthe simultaneous gas-fitting of pPXF together with the stel-lar population fitting. These are shown in the left and cen-ter panels of Figure 9. For the map of younger stars ( < Myr), the fact that the bulge and the spiral arms have rel-atively older ages means that there is a continuous produc-tion of stars in those regions for the last 10 Myr, as seenand traced by the H α map. We compare these two maps tointermediate-age stars (1-5 Gyr) as seen on the right panelof Figure 9. This map, on the other hand, does not sharethe same features as the H α map.It is also interesting that we see two populations inthe central region, though we note that only the inner-most spaxels were classified as part of the bulge in makingthe mass build up histories. From Figure 8, we can see a MNRAS000
20% of the bulge particles at ages be-tween 1-5 Gyr (as we see in Figure 7 for the bulge) showlow B/T (i.e. < . ) and growth through minor mergers andsecular processes. Tacchella et al. (2019) used IllustrisTNGhydrodynamical simulations (Pillepich et al. 2018) to studythe spheroidal and disc components of galaxies. They findthat similar-mass galaxies to NGC 2903 have bulges thatgrow very little between the epoch of disc formation (z ∼ ∼ ∼ From the stellar population synthesis and best-fit combina-tion of stellar populations for each spaxel, we derived mass-weighted and light-weighted age and metallicity maps asshown in Figure 8. Both the age and metallicity are higherfor the mass-weighted maps than the light-weighted maps.This physically makes sense as older stars at a given metal-licity and higher-metallicity stars at a given age have highermass-to-light ratios and would therefore dominate the mass-weighted maps. On the other hand, the light-weighted mapsshow lower values for the average age and metallicity for aspaxel. Younger stars dominate the light for a given stel-lar population, therefore the light-weighted age is biased to-wards younger ages. Also quite noticeable is the enhancedmetallicity, both in the mass-weighted and light-weightedmaps, for the spiral arms and central region, areas withhigher star formation.To look into these patterns further, we made light-weighted averaged age maps for stellar populations with ages < Myr. We also compare it to the H α map derived fromthe simultaneous gas-fitting of pPXF together with the stel-lar population fitting. These are shown in the left and cen-ter panels of Figure 9. For the map of younger stars ( < Myr), the fact that the bulge and the spiral arms have rel-atively older ages means that there is a continuous produc-tion of stars in those regions for the last 10 Myr, as seenand traced by the H α map. We compare these two maps tointermediate-age stars (1-5 Gyr) as seen on the right panelof Figure 9. This map, on the other hand, does not sharethe same features as the H α map.It is also interesting that we see two populations inthe central region, though we note that only the inner-most spaxels were classified as part of the bulge in makingthe mass build up histories. From Figure 8, we can see a MNRAS000 , 1–14 (2018)
GC 2903’s bulge, bar, and outer disc younger and higher metallicity stellar population surround-ing an older, lower metallicity stellar population in both thelight-weighted and mass-weighted maps. This is evidence ofthe co-existence of an older and lower metallicity bulge anda younger and higher metallicity bulge. Such composite sys-tems have been found in other galaxies through light profileand kinematics studies (e.g., Fisher & Drory 2010; Erwinet al. 2015), preferentially in spirals with bars. In this study,we distinguish these composite bulges in metallicity and age,complementary to previous work. The presence of youngerstars in the bulge as well as our structural fit that gives ita S´ersric index of n < < Gyr compared to 2.5 - 4 Gyr forthe central part of the bulge. The mass-weighted age map inFigure 8 shows an even bigger difference in the average age,with the younger population being < . Gyr and the olderpopulation > Gyr.
Figure 7 shows the stellar mass fraction of the whole galaxy,bulge, bar, and disc, divided into bins of different stellarages. Most (75%) of the stars in the bulge are older than 5Gyr and largely have high metallicity, followed by low metal-licity and alpha-enhanced compositions. In this section, wetry to understand the metallicity pattern better and explorewhether it is plausible to have such high metallicities in bulgestars with age > > ∼ In this study, we have used the VENGA IFU survey to dis-sect the age, metallicity, and mass of stars in the bulge,bar and outer disc of NGC 2903, and how their ensembleformation histories paint a comprehensive picture of galaxyevolution. We performed bulge-bar-disc decomposition withGALFIT to better assign a spectra to a galaxy componentand performed full-spectrum fitting with pPXF to derive thestellar populations from which we got star formation histo-ries. Our main results are:(1)
The outer disc : The outer disc has 46% of its stel-lar mass in the form of stars with ages > <
10% in younger stars.There is a larger fraction of solar and sub-solar metallicitystellar populations in the outer part of the disc comparedto the inner parts of the galaxy. These observed propertiessupport a scenario where the outer disc may have mainlyformed via in-situ star formation recently (1-5 Gyr ago) andat earlier times ( > > The stellar bar : The stellar bar is about threetimes less massive than the outer disc. It has a broadly sim-ilar stellar metallicity make-up and distribution of stellarages as the outer disc, but has a higher fraction of youngerstars of ages between 1 and 5 Gyr (65% in the bar vs 48%in the outer disc). These results support the general picturewhere stellar bars form, spontaneously or via tidal triggers,from disc material, and where gas inflows along a weak ormoderately strong bar can lead to star formation along thebar. 3)
The bulge : The central bulge contributes less than5% of the total stellar mass of the galaxy and bulge-bar-disc decomposition of the 2MASS Ks-band image shows ithas a low S´ersic index of ∼ > MNRAS , 1–14 (2018) A. Carrillo et al.
Figure 8.
Top: Mass-weighted average age (left) and average metallicity (right). The color bars represent log of age in years and [M/H],respectively. Bottom: Same as above but light-weighted. The light-weighted age and metallicity are younger and lower respectively, asthese stellar populations have lower mass-to-light ratio compared to older and higher metallicity stellar populations. Both age maps showthat the disc generally has no features but that the bulge has a young population and an old population. On the other hand, both themetallicity maps trace areas of high star formation (in the spiral arms and central regions), showing enhanced metallicities.MNRAS000
Top: Mass-weighted average age (left) and average metallicity (right). The color bars represent log of age in years and [M/H],respectively. Bottom: Same as above but light-weighted. The light-weighted age and metallicity are younger and lower respectively, asthese stellar populations have lower mass-to-light ratio compared to older and higher metallicity stellar populations. Both age maps showthat the disc generally has no features but that the bulge has a young population and an old population. On the other hand, both themetallicity maps trace areas of high star formation (in the spiral arms and central regions), showing enhanced metallicities.MNRAS000 , 1–14 (2018)
GC 2903’s bulge, bar, and outer disc Figure 9.
Light-weighted average age for stars that are less than 10 Myr old (left), extinction-corrected H α map (center), and light-weighted average age for stars with ages between 1 and 5 Gyr (right), to compare areas with recent star formation based on the continuumand gas vs an older population. The color bars for the first and third panels represent log of age in years while the color bar for the secondpanel is log of the extinction-corrected H α flux in erg s − cm − ˚A − . H α recombination lines trace stars formed <
10 Myr. As expected,the young stellar population map agrees with the features being traced in the H α map. The stellar populations with ages 1-5 Gyr do nothave the same features as the H α map and instead shows older ages inside corotation i.e. location of the bar. > α map.(5) Our results in (1) to (3) are consistent with the studyby Weinzirl et al. (2009), which suggests that most spiralgalaxies in the local Universe with low B/T have not had amajor merger since z ∼ , and have grown their outer discmostly through gas accretion, minor mergers, and secularprocesses, based on their comparison of structural propertiesof spirals (such as the distributions of S´ersic index and B / T of their bulges) to semi-analytic models of galaxy evolutionfrom Hopkins et al. (2009), Khochfar & Burkert (2005) andKhochfar & Silk (2006). This is further supported by recenthydrodynamical simulations (e.g., Gargiulo et al. 2019; Tac-chella et al. 2019) that show that galaxies with B/T, S´ersic index, and mass similar to NGC 2903 have not experienceda major merger in the recent past.In this work, we have emphasized the importance ofthe comprehensive study of individual and synergistic stel-lar populations and masses of the different parts of a galaxyto fully understand its formation and evolution. With moreexpansive spectral templates, and telescopes like JWST andthe Giant Magellan Telescope (GMT ) that let us peerdeeper into space at higher resolutions, we will be able fillin our knowledge about how galaxies and their componentschange over time. Doing it in the nearby universe as we didin this study is but the first step. ACKNOWLEDGEMENTS
This paper includes data taken at The McDonald Obser-vatory of The University of Texas at Austin. A.C. wouldlike to thank Keith Hawkins, Milos Milosavljevic, JasonJaacks, and Rodrigo Luger for insightful discussions andRosa Gonzalez-Delgado for providing the Base GM tem-plates. A.C. and S.J. acknowledge support from NationalScience Foundation (NSF) grant AST-1413652, NSF AST-1614798, NSF AST-1757983 and the McDonald Observatoryand the Department of Astronomy Board of Visitors. A.C.also thanks the Large Synoptic Survey Telescope Corpora-tion (LSSTC) Data Science Fellowship Program, which is , 1–14 (2018) A. Carrillo et al. funded by LSSTC, NSF Cybertraining Grant 1829740, theBrinson Foundation, and the Moore Foundation.
REFERENCES
Adams J. J., et al., 2011, ApJS, 192, 5Athanassoula E., 2003, MNRAS, 341, 1179Athanassoula E., 2005, MNRAS, 358, 1477Barentine J. C., Kormendy J., 2012, ApJ, 754, 140Bensby T., et al., 2017, A&A, 605, A89Bezanson R., van Dokkum P. G., Tal T., Marchesini D., KriekM., Franx M., Coppi P., 2009, ApJ, 697, 1290Blanc G. A., et al., 2013, ApJ, 764, 117Bundy K., et al., 2015, ApJ, 798, 7Calzetti D., Kinney A. L., Storchi-Bergmann T., 1994, ApJ, 429,582Cappellari M., 2017, MNRAS, 466, 798Cappellari M., Copin Y., 2003, MNRAS, 342, 345Cappellari M., Emsellem E., 2004, PASP, 116, 138Charbonnel C., Meynet G., Maeder A., Schaller G., Schaerer D.,1993, A&AS, 101, 415Cid Fernandes R., Mateus A., Sodr´e L., Stasi´nska G., GomesJ. M., 2005, MNRAS, 358, 363Combes F., Sanders R. H., 1981, A&A, 96, 164Combes F., Debbasch F., Friedli D., Pfenniger D., 1990, A&A,233, 82Conroy C., 2013, ARA&A, 51, 393Croom S. M., et al., 2012, MNRAS, 421, 872Dekel A., et al., 2009, Nature, 457, 451Dom´ınguez A., et al., 2013, ApJ, 763, 145Erwin P., et al., 2015, MNRAS, 446, 4039Fisher D. B., Drory N., 2008, AJ, 136, 773Fisher D. B., Drory N., 2010, ApJ, 716, 942Gargiulo I. D., et al., 2019, arXiv e-prints, p. arXiv:1907.02082Girardi L., Bressan A., Bertelli G., Chiosi C., 2000, A&AS, 141,371Gonz´alez Delgado R. M., Cervi˜no M., Martins L. P., Leitherer C.,Hauschildt P. H., 2005, MNRAS, 357, 945Gonz´alez Delgado R. M., et al., 2015, A&A, 581, A103Grand R. J. J., et al., 2017, MNRAS, 467, 179Hill G. J., et al., 2008, Design, construction, and performance ofVIRUS-P: the prototype of a highly replicated integral-fieldspectrograph for HET. p. 701470, doi:10.1117/12.790235Hopkins P. F., et al., 2009, MNRAS, 397, 802Irwin J. A., et al., 2009, ApJ, 692, 1447Jarrett T. H., Chester T., Cutri R., Schneider S. E., Huchra J. P.,2003, AJ, 125, 525Jogee S., et al., 2004, ApJ, 615, L105Jogee S., Scoville N., Kenney J. D. P., 2005, ApJ, 630, 837Jogee S., et al., 2009, ApJ, 697, 1971Jones T., Ellis R., Jullo E., Richard J., 2010, ApJ, 725, L176Jones T., Ellis R. S., Richard J., Jullo E., 2013, ApJ, 765, 48Kaplan K. F., et al., 2016, MNRAS, 462, 1642Kereˇs D., Katz N., Weinberg D. H., Dav´e R., 2005, MNRAS, 363,2Khochfar S., Burkert A., 2005, MNRAS, 359, 1379Khochfar S., Silk J., 2006, MNRAS, 370, 902Khochfar S., Silk J., 2009, MNRAS, 397, 506Knapen J. H., Beckman J. E., Heller C. H., Shlosman I., de JongR. S., 1995, ApJ, 454, 623Kormendy J., 1993, in Dejonghe H., Habing H. J., eds, IAU Sym-posium Vol. 153, Galactic Bulges. p. 209Kormendy J., Kennicutt Jr. R. C., 2004a, ARA&A, 42, 603Kormendy J., Kennicutt Robert C. J., 2004b, ARA&A, 42, 603Lee J. C., et al., 2009, ApJ, 706, 599Marinova I., Jogee S., 2007, ApJ, 659, 1176McDermid R. M., et al., 2015, MNRAS, 448, 3484 Miller R. H., Prendergast K. H., 1968, ApJ, 151, 699Molina J., Ibar E., Swinbank A. M., Sobral D., Best P. N., SmailI., Escala A., Cirasuolo M., 2017, MNRAS, 466, 892Naab T., Trujillo I., 2006, MNRAS, 369, 625Ness M., Freeman K., 2016, Publ. Astron. Soc. Australia, 33, e022Ness M., et al., 2013, MNRAS, 432, 2092Onodera M., et al., 2012, ApJ, 755, 26Osterbrock D. E., Bochkarev N. G., 1989, Soviet Ast., 33, 694Peng C. Y., Ho L. C., Impey C. D., Rix H.-W., 2010, AJ, 139,2097Pietrinferni A., Cassisi S., Salaris M., Castelli F., 2004, ApJ, 612,168Pietrinferni A., Cassisi S., Salaris M., Castelli F., 2006, ApJ, 642,797Pillepich A., et al., 2018, MNRAS, 473, 4077Romano-D´ıaz E., Shlosman I., Heller C., Hoffman Y., 2008, ApJ,687, L13Salpeter E. E., 1955, ApJ, 121, 161S´anchez S. F., et al., 2012, A&A, 538, A8S´anchez S. F., et al., 2016, Rev. Mex. Astron. Astrofis., 52, 21Schaerer D., Meynet G., Maeder A., Schaller G., 1993, A&AS,98, 523Schaller G., Schaerer D., Meynet G., Maeder A., 1992, A&AS,96, 269Scott N., et al., 2017, MNRAS, 472, 2833S´ersic J. L., 1963, Boletin de la Asociacion Argentina de Astrono-mia La Plata Argentina, 6, 41Sheth K., Vogel S. N., Regan M. W., Thornley M. D., TeubenP. J., 2005, ApJ, 632, 217Shlosman I., Noguchi M., 1993, ApJ, 414, 474Somerville R. S., Dav´e R., 2015, ARA&A, 53, 51Swinbank A. M., Sobral D., Smail I., Geach J. E., Best P. N.,McCarthy I. G., Crain R. A., Theuns T., 2012, MNRAS, 426,935Tacchella S., et al., 2019, MNRAS, p. 1573Vazdekis A., S´anchez-Bl´azquez P., Falc´on-Barroso J., CenarroA. J., Beasley M. A., Cardiel N., Gorgas J., Peletier R. F.,2010, MNRAS, 404, 1639Vazdekis A., et al., 2015, MNRAS, 449, 1177Wang X., et al., 2017, ApJ, 837, 89Weinzirl T., Jogee S., Khochfar S., Burkert A., Kormendy J.,2009, ApJ, 696, 411Weinzirl T., et al., 2011, ApJ, 743, 87Weinzirl T., et al., 2014, MNRAS, 441, 3083Wilkinson D. M., Maraston C., Goddard D., Thomas D., ParikhT., 2017, MNRAS, 472, 4297Yuan T.-T., Kewley L. J., Swinbank A. M., Richard J., LivermoreR. C., 2011, ApJ, 732, L14de Vaucouleurs G., de Vaucouleurs A., Corwin H. G., Buta R. J.,Paturel G., Fouque P., 1995, VizieR Online Data Catalog, p.VII/155This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000