Elevation or Suppression? The Resolved Star Formation Main Sequence of Galaxies with Two Different Assembly Modes
Qing Liu, Enci Wang, Zesen Lin, Yulong Gao, Haiyang Liu, Berzaf Berhane Teklu, Xu Kong
DDraft version March 30, 2018
Typeset using L A TEX twocolumn style in AASTeX62
Elevation or Suppression? The Resolved Star Formation Main Sequence of Galaxies withTwo Different Assembly Modes
Qing Liu,
1, 2
Enci Wang,
1, 2
Zesen Lin,
1, 2
Yulong Gao,
1, 2
Haiyang Liu,
1, 2
Berzaf Berhane Teklu,
1, 2 andXu Kong
1, 2 CAS Key Laboratory for Research in Galaxies and Cosmology, Department of Astronomy, University of Science and Technology ofChina, Hefei 230026, China School of Astronomy and Space Science, University of Science and Technology of China, Hefei 230026, China
ABSTRACTWe investigate the spatially resolved star formation main sequence in star-forming galaxies usingIntegral Field Spectroscopic observations from the Mapping Nearby Galaxies at the Apache PointObservatory (MaNGA) survey. We demonstrate that the correlation between the stellar mass surfacedensity (Σ ∗ ) and star formation rate surface density (Σ SFR ) holds down to the sub-galactic scale,leading to the sub-galactic main sequence (SGMS). By dividing galaxies into two populations based ontheir recent mass assembly modes, we find the resolved main sequence in galaxies with the ‘outside-in’mode is steeper than that in galaxies with the ‘inside-out’ mode. This is also confirmed on a galaxy-by-galaxy level, where we find the distributions of SGMS slopes for individual galaxies are clearlyseparated for the two populations. When normalizing and stacking the SGMS of individual galaxieson one panel for the two populations, we find that the inner regions of galaxies with the ‘inside-out’mode statistically exhibit a suppression in star formation, with a less significant trend in the outerregions of galaxies with ‘outside-in’ mode. In contrast, the inner regions of galaxies with ‘outside-in’ mode and the outer regions of galaxies with ‘inside-out’ mode follow a slightly sublinear scalingrelation with a slope ∼ Keywords: galaxies: evolution — galaxies: star formation — galaxies: fundamental parameters INTRODUCTIONOne of the well-established relationships in galaxy for-mation and evolution is the correlation between the starformation rate (SFR) and the stellar mass (M ∗ ) for star-forming galaxies (SFGs), which is usually referred to asthe star-forming main sequence (SFMS). Observationshave demonstrated that the SFMS holds from the localuniverse (e.g. Brinchmann et al. 2004; Salim et al. 2007)to the high-redshift one (e.g. Daddi et al. 2007; Elbazet al. 2007; Noeske et al. 2007; Barro et al. 2017). Inaddition, the SFMS exhibits an evolution with redshift(summarized in Speagle et al. 2014). [email protected]@[email protected] Recently, a tight correlation between surface stellarmass density (Σ ∗ ) and surface SFR density (Σ SFR ) haveattracted attention (e.g. S´anchez et al. 2013; Wuyts etal. 2013; Cano-D´ıaz et al. 2016, hereafter C16; Magdiset al. 2016; Abdurro’uf & Akiyama 2017; Hsieh et al.2017, hereafter H17; Maragkoudakis et al. 2017), whichis referred to as the resolved star formation main se-quence, or sub-galactic main sequence (SGMS). Thisindicates a more fundamental relation between local on-going star-forming activities and the underlying stellarpopulations. C16 have found an SGMS on an ∼ ∼ . a r X i v : . [ a s t r o - ph . GA ] M a r Q.Liu et al.
Thanks to IFS surveys, galaxies can be divided intotwo populations according to their assembly modes(P´erez et al. 2013; Pan et al. 2015; Ibarra-Medel et al.2016; Goddard et al. 2017). In a parallel paper (Wanget al. 2017a), we found that galaxies that are in the re-cent ‘outside-in’ mode have smaller sizes, higher concen-trations, and higher global gas-phase metallicities withrespect to galaxies in the recent ‘inside-out’ mode, sug-gesting that they are likely in the transitional phase fromnormal SFG to quiescent populations with rapid centralstellar mass growth. Thus, it is essential to understandthe star formation regulation in these two populations,especially their behaviors on SGMS.In this work, we present our results based on ∼ = 71 km s − Mpc − , Ω M = 0.27,and Ω Λ = 0.73. DATA2.1.
Sample and Classification
In this work, we use the newly released MaNGA datafrom SDSS DR14 (Abolfathi et al. 2017), including morethan 2700 galaxies. The M ∗ , SFR, effective radius (R e ),and the other parameters we used are drawn from theMaNGA Pipe3D Value Added Catalog (S´anchez et al.2017). The sample is first reduced to have an inclina-tion < ◦ to avoid high inclination effects, with ∼ < ∼
800 galaxies selected.We further remove galaxies with a field of view (FoV) < e to ensure that most of the galaxies in the remainingsample have enough sampling in both inner and outerparts, with 647 galaxies left. It is worth noting that weare not sampling the entire disk for our sample, whichwould be a more suitable task for the CALIFA survey(S´anchez et al. 2012) or the MUSE survey (Bacon et al.2017).Then we divide our galaxy sample into two popula-tions according to their recent mass assembly modes.Briefly, the classification is based on information fromspatially resolved information of the 4000 ˚A break log M * [M ] l o gS F R [ M y r ] Inside-outOutside-in
Figure 1.
Distribution of sample and subsamples in ourstudy. Red circles stand for ‘inside-out’ galaxies and blue cir-cles for ‘outside-in’ galaxies. Distributions of M ∗ and SFR forour subsamples are shown in small panels by red (‘inside-out’galaxies) and blue (‘outside-in’ galaxies) histograms. Col-ored lines in red and blue show their linear fittings. All ∼ (D4000) as follows: Galaxies in the ‘outside-in’ massassembly mode are defined to have D4000 . e (D4000at 1.5 R e ) > D4000 cen (D4000 within 0.25R e ). Con-versely, those with D4000 . e < D4000 cen are classi-fied as being in the ‘inside-out’ mass assembly mode.D4000 is defined as the ratio between the average fluxin red-end and blue-end continua at 4000 ˚A wavelength(Bruzual A. 1983), which has been widely used to tracethe stellar populations with mean stellar ages youngerthan 1-2 Gyr (e.g., Balogh et al. 1999; Kauffmann etal. 2003; Brinchmann et al. 2004; Gallazzi et al. 2005;Wang et al. 2017b). More details and discussions aboutthe classification based on D4000 can be found in Wanget al. (2017a).Through visual inspection, we remove major mergersand galaxies with very few available spaxels or dubi-ous R e measurements for ∼ e smaller than the PSF (2.5 (cid:48)(cid:48) ) of the MaNGAsurvey, which account for 5% of the sample. Finally,we obtain 406 galaxies classified as being in the ‘inside-out’ mass assembly mode and 155 classified as beingin the ‘outside-in’ mass assembly mode. For simplicity,we hereafter refer to them as ‘inside-out’ galaxies and‘outside-in’ galaxies. The distribution of our sample on GMS of Two Modes ∗ –SFR plane is shown in Figure 1. In agreementwith Wang et al. (2017a), ‘outside-in’ galaxies > M (cid:12) are typically preferred to be located above ‘inside-out’ galaxies. We use the same criteria in S´anchez etal. (2017) that their central 2.5 arcsec/aperture regionshave (a) emission line ratios above the Kewley limit(Kewley et al. 2001) and (b) EW(H α ) > ∗ andrange of local densities with redshift in the MaNGA sam-ple, we perform tests by restricting galaxies to redshift < < Spaxels
The fitting results we use are from the Pipe3D(S´anchez et al. 2016a; S´anchez et al. 2016b) dataprod-ucts released with SDSS DR14. The Initial Mass Func-tion (IMF) adopted in Pipe3D is Salpeter (1955). As-suming the intrinsic flux ratio of (H α/ H β ) = 2.86 anda Calzetti et al. (2000) attenuation law, we then use theBalmer Decrement to correct the emission line fluxes,with which we obtain the SFR, adopting the Kennicutt(1998) conversion (in Salpeter IMF).The spaxels are then selected with regard to the fol-lowing criteria: (1) EW(H α ) (equivalent width of H α ) > iii ] / H β and[N ii ] / H α lying below the Kewley limit (Kewley et al.2001) on the BPT diagram (Baldwin et al. 1981); (3)signal-to-noise ratio (S/N) of H α greater than 3 andS/N of other involved emission lines greater than 1. Thefirst two criteria are to select pure star-forming regions,which have been used in many studies (e.g. Cano-D´ıaz et al. 2016; S´anchez-Menguiano et al. 2016; S´anchez etal. 2017). In Pipe3D dataproducts, a spatial binningis performed in order to reach a S/N of 50 measuredin the range of 5590–5680 ˚A across the FoV (S´anchezet al. 2016a). However, as the binning method requiresflux homogeneity in the same bin, some bins would havelower S/Ns than expected. Therefore, we recalculatethe continuum S/N for each bin following Stoehr et al.(2008) and clip those with S/N smaller than 3 ( ∼
4% ofdata). RESULTS3.1.
Resolved Main Sequence for Two Assembly Modes
In panel (a) of Figure 2, we show the SGMS for all141,114 spaxel bins in all the sample galaxies. The blacksolid line represents the best optimized least square(OLS) linear fitting using all spaxel bins in the formof log (Σ SFR ) = α log (Σ ∗ ) + β . The fitted slopes ( α )and zero points ( β ) are shown in each panel. For com-parison, results from C16 and H17 are also drawn. Witha slope of 0.75, our result lies between C16 (0.72) andH17 (1.0).To investigate whether the two subsamples show dif-ferent patterns on their SGMS, we display the SGMSwith regard to the two types of mass assembly modes inpanel (b) and (c) of Fig. 2, respectively. It turns outthat the two populations of galaxies occupy different lo-cations on the Σ ∗ –Σ SFR plane, with ‘outside-in’ galaxieson average lying above ‘inside-out’ galaxies at the high-Σ ∗ end (log (Σ ∗ ) > Galaxy-by-Galaxy SGMS
To explore statistical trends and variations amonggalaxies, it is also worth inspecting the SGMS within in-dividual galaxies. Therefore, we conduct a similar anal-ysis in Maragkoudakis et al. (2017) by linearly fittingstar-forming bins from each galaxy one-by-one. Thishelps to mitigate the potential problem in resolved stud-ies that the fitting might depend on the sampling sincethe numbers of bins out of each galaxy are different.For each galaxy, the fitting is iterated for three times toremove outliers beyond 3 σ regions. We present resultsfor individual galaxies in Figure 3 and 4, excluding the Q.Liu et al. log * [M kpc ] l o g S F R [ M y r k p c ] (a)= 0.75= 8.19C16H17Total log * [M kpc ] (b)= 0.68= 7.57C16H17Inside-out log * [M kpc ] (c)= 0.94= 9.74C16H17Outside-in Figure 2.
SGMS for our sample and subsamples. White squares show medians of Σ
SFR for Σ ∗ equally binned between 3% and97% quantiles. The best OLS fittings are shown by colored solid lines in each panel in black (total), red (‘inside-out’) and blue(‘outside-in’). α and β are slopes and zero points for the OLS fittings. The OLS fitting for 80% data of C16 and orthogonalfitting from H17 are shown by the green dashed line and the magma dotted line, respectively. In each panel, contours showlevels of 30% / 1 σ (67%) / 2 σ (95%) / 3 σ (99%) of data. S l o p e o f G - b y - G S G M S (a) log M * [M ] D: 0.27p: 5.0e-07 (b) log sSFR[yr ] D: 0.08p: 4.6e-01 (c)
Age LW D: 0.69p: 1.2e-42 1.250.7 D: 0.67p: 1.3e-40
Figure 3.
Slopes of G-by-G SGMS with respect to (from left to right) M ∗ , sSFR, and α (cid:104) Age LW (cid:105) . In each panel, ‘outside-in’and ‘inside-out’ galaxies are represented by blue and red dots. Colored solid lines show trends of five medians of Σ SFR for Σ ∗ equally binned between 3% and 97% quantiles for the two subsamples. Marginalized distributions of parameters are shown byred (‘inside-out’ galaxies) and blue (‘outside-in’ galaxies) histograms in small panels with their results of K–S tests (D-statisticsand p-values). Colored dashed lines show their peaks and the black dashed line shows where the SGMS slope equals 1.04. bad fittings with the Pearson correlation coefficient < ∼
18% of the sample).We perform Kolmogorov–Smirnov (K–S) tests for eachmarginalized distribution of our two subsamples to testwhether they differ, with the D-statistics and p-valuesshown in Fig. 3 and 4. If the D-statistic is small or thep-value is high (at a certain significance level), then wecannot reject the hypothesis that their distributions arethe same; otherwise, it can be concluded that they arefrom different distributions.In each panel of Fig. 3, slopes fitted from G-by-GSGMS are plotted as a function of M ∗ , specific star for- mation rate (sSFR) and the luminosity-weighted (LW)age gradient of the host galaxy. The distributions ofslopes for the two populations are clearly separated. Inpanel (a) and (b), trends of medians shown in coloredsolid lines suggest a slight increase for slopes of ‘inside-out’ galaxies with regard to sSFR and weak or no depen-dence with M ∗ . The dependences are less significant for‘outside-in’ galaxies as the slopes have a larger scatter.The age gradient is drawn from the Pipe3D catalog inS´anchez et al. (2017), measured within 0.5–2 R e . Thewell-separated distributions of age gradient in panel (c) GMS of Two Modes S l o p e o f G - b y - G S G M S (a) Sersic n
D: 0.21p: 3.3e-04 (b) (B/T) g D: 0.26p: 2.9e-06 (c) R (1.5 R e ) D: 0.29p: 1.1e-07 1.250.7 D: 0.67p: 1.6e-40
Figure 4.
Same plot as Fig. 3 for slopes of G-by-G SGMS with respect to (from left to right) Sersic n, (B/T) g , and λ R (1 . λ R (1 . are due to the close correlation between D4000 and stel-lar age (see the sample selection in Section 2.1).In each panel of Fig. 4, we explore the relation ofG-by-G SGMS with bulge indicators including the Ser-sic n index (from the MaNGA DRP catalog), the bulgeratio in G band (B/T) g (from Simard et al. 2011) andthe specific angular momentum ( λ R ) of the host galaxy.The λ R parameter is drawn from the Pipe3D catalog(S´anchez et al. 2017), measured within 1.5 R e . This pa-rameter is defined as λ R = (cid:104) R | V |(cid:105) / (cid:104) R √ V + σ (cid:105) (Em-sellem et al. 2007), where V and σ are locally measuredmaximum radial velocity and stellar velocity dispersion,with R representing projected galactocentric distanceand brackets symbolizing flux weighting. Panel (a) andPanel (b) show that for ‘inside-out’ galaxies the slopeshave weak or no correlations with Sersic n or (B/T) g .On the other hand, ‘outside-in’ galaxies appear to havelarger slopes with higher Sersic n or bulge ratios. How-ever, the Sersic n and B/T parameter may not be di-rectly translated into the morphology because they arenot one-to-one tightly correlated (see, e.g. S´anchez etal. 2017). In panel (c), the slopes of ‘inside-out’ galaxiespresent a slight anti-correlation with λ R , i.e. galaxieswith higher λ R (1 . λ R (1 . Stacking of G-by-G SGMS
The fitting of G-by-G SGMS also shows differencesin the zero points among galaxies, which might be at-tributed to differences in their physical properties orconditions. For this reason, we shift the SGMS of each individual galaxy to have the median values of Σ ∗ andΣ SFR equal to zero, and stack them on one panel for‘inside-out’ and ‘outside-in’ galaxies to see their statis-tical patterns. This is equivalent to select the charac-teristic values of Σ ∗ and Σ SFR for each galaxy and ap-ply a normalization on them. The results of stackingSGMS for the two subsamples are shown in Figure 5.The shifted Σ ∗ and Σ SFR are referred to as (cid:101) Σ ∗ and (cid:101) Σ SFR .Firstly, the small scatter ( ∼ ∗ end for ‘inside-out’ galaxies and a less sig-nificant but perceptible steepening at the low-Σ ∗ end for‘outside-in’ galaxies. Such patterns are hard to observein Fig. 2 since differences in zero points mix the SGMStogether. To further illustrate, we apply two-componentlinear fittings on modes of Σ SFR for Σ ∗ equally binnedbetween 1% and 99% quantiles, with a restriction thatthe turning points lie within the central 40% of data. Across-validation technique (Ivezi´c et al. 2014) is appliedto demonstrate that the piecewise fittings have less vari-ance, which also have smaller χ dof when uncertaintiesof (cid:101) Σ SFR are considered. However, we emphasize thatthe main purpose of fitting is to strengthen the visualinspection.The slopes of SGMS in the literature are dependenton the sample selection, analyzing method, and fittingrecipe; however, they all range from 0.66 to 1.0 (e.g.S´anchez et al. 2013; Wuyts et al. 2013; Cano-D´ıaz et al.2016; Magdis et al. 2016; Abdurro’uf & Akiyama 2017;Hsieh et al. 2017; Maragkoudakis et al. 2017). In Fig.5, the fitting slopes for outer parts of ‘inside-out’ galax-ies and inner parts of ‘outside-in’ galaxies are close to0.9, which is situated within the range given in previous
Q.Liu et al. * (dex) S F R ( d e x ) k :0.94 k :0.38 Inside-out * (dex) S F R ( d e x ) k :1.61 k :0.92 Outside-in EW ( H )[ Å ] 0 31 EW ( H )[ Å ] Figure 5.
Stacking for G-by-G SGMS of two assembly modes. Colored dots (green for ‘outside-in’ and gold for ‘inside-out’)show modes of (cid:101) Σ SFR for (cid:101) Σ ∗ in equal bins between 1% and 99% quantiles. Two-component linear fittings for the modes are showby solid lines, with dashed lines representing their extrapolations. The inner parts of ‘inside-out’ and outer parts of ‘outside-in’galaxies appear to be in good alignment with slopes close to 0.9, which is consistent with results of Magdis et al. (2016) andMaragkoudakis et al. (2017). Contours show levels of 30% / 1 σ (67%) / 2 σ (95%) / 3 σ (99%) of data. SDSS gri images, EW(H α )maps and G-by-G SGMS of two typical galaxies (MaNGA ID: 9041-6102 and 8611-3702) showing a clear piecewise pattern ona G-by-G level are shown in small panels for each subsample. In the bottom small panels, outliers > σ are shown by emptycircles. Colored stars show the characteristic values of SGMS, which are shifted to zero. studies and is well consistent with results of Magdis et al.(2016) and Maragkoudakis et al. (2017). Thus, we treatthem as the standard pattern of star formation. Theremaining parts show a deviation from them, especiallyin ‘inside-out’ galaxies. This suggests that the expectedin-situ star formation in the outer regions of ‘inside-out’galaxies and the inner regions of ‘outside-in’ galaxies aregoverned by a common set of regulations without theinfluence of additional physical processes, and thus weinfer that the inner parts of ‘inside-out’ galaxies clearlyshow an evidence of suppression in star formation. For‘outside-in’ galaxies, they also show an indication of sup-pression in star formation for their outer regions, as theyappear to have insufficient SFR at certain surface den-sity (which leads to a steeper slope). We have stackedthe two panels of Fig. 5 into one and find that theouter regions of ‘outside-in’ galaxies indeed have differ-ent distributions from ‘inside-out’ galaxies. We will havemore discussions on this in the next section. We furtherperform a test with a lower continuum S/N criteria toconfirm that this pattern is not due to an S/N cut inlower surface density regions. DISCUSSIONAs confirmed by many previous studies, the SGMSholds well down to kiloparsec scales. In this work, ourresults further reveal that, the recent mass assemblymode of the galaxy would have impacts on the shapeof their SGMS. It is worth noting that our subsample construction for ‘inside-out’ and ‘outside-in’ galaxies isdifferent from the fossil record method used in manyprevious studies (e.g. P´erez et al. 2013; Ibarra-Medelet al. 2016; Garc´ıa-Benito et al. 2017). Given the shorttimescales and stochasticity for star-forming activities,the mass assembly mode of the galaxy is possibly notfixed. For the same reason, it is reasonable to infer thatthe SGMS is more related to the recent mass assemblymode, which can be well traced by the D4000 diagnostic.Furthermore, it is model-independent with fewer biases,since different fitting methods might lead to different re-sults (Goddard et al. 2017). However, we notice here thepotential caveat of using D4000 as the indicator is that,although primarily correlated with the age of the stellarpopulation, D4000 also depends on the metallicity (Pog-gianti & Barbaro 1997), which is related to the age-colordegeneracy. To check the mass assembly histories of oursample, we have also stacked and compared the massgrowth curves in inner and outer regions. We find that >
40% of galaxies have complicated overlapping massgrowth curves hard to define their assembly histories,which supports our previous speculation and is consis-tent with episodic transitions proposed in Ibarra-Medelet al. (2016).4.1. ‘Inside-out’ SGMS: Bulge Effects or CentralSuppression?
The smaller slope of ‘inside-out’ SGMS mainly comesfrom the suppression in their central high surface density
GMS of Two Modes g and systematically larger λ R , indicating their relative significance in presence ofthe bulge from both photometric and kinematic perspec-tives. Meanwhile, they do not present a significant bendin their centers. Therefore, we infer that, at least in oursample, the presence of the bulge only is not a predom-inant factor contributing to the bend of SGMS.In Figure 6, we plot profiles of EW(H α ) for ‘inside-out’ galaxies and ‘outside-in’ galaxies in red and blue,respectively, where it can be seen that ‘inside-out’ galax-ies show a strong decrease with their star formationpeaking in disks. Recently, Belfiore et al. (2018) havedemonstrated that EW(H α ) and sSFR profiles of SFGswith higher M ∗ , especially for those with M ∗ > . ,generally present a stronger decrease at their centers.Gonz´alez Delgado et al. (2016) have observed a simi-lar trend with the morphology, since morphologies ofgalaxies correlate with M ∗ . Our result is consistent withBelfiore et al. (2018), where we check that more massivegalaxies have lower EW(H α ) in their centers. It couldbe inferred that the difference of profiles observed inFig. 6 is due to the fact that ‘inside-out’ galaxies aresystematically more massive than ‘outside-in’ galaxies.However, by constructing control samples for ‘inside-out’ and ‘outside-in’ galaxies in M ∗ / λ R with no signifi-cant changes observed, i.e. their overall trends remain,we confirm that the difference of profiles of our subsam-ples in inner regions is not led by their mass/morphologydistinctions.4.2. ‘Outside-in’ SGMS: Standard or PeripheralSuppression? On the other hand, the larger slope of the ‘outside-in’ SGMS approaching (in the pixel-by-pixel case) or
R [R e ] l o g E W ( H )[ Å ] Inside-outOutside-in
Figure 6.
EW(H α ) profiles of ‘inside-out’ galaxies (red) and‘outside-in’ galaxies (blue) equally sampled along the radius.The solid lines stand for the median profiles of EW(H α ) andthe translucent bands show the 30%–70% distributions. Starformation in ‘inside-out’ galaxies peaks in disks, wile starformation in ‘outside-in’ galaxies peaks in central regions anddecreases with the radius. even exceeding (in G-by-G cases) unity is worth in-vestigation. When compared with ‘inside-out’ galaxies,one would naturally infer that the ‘outside-in’ galaxiespresent a ‘standard’ star-forming pattern in both innerregions and outskirts. This point might be supported bythe similarity in its slope to the high-redshift integratedSFMS (e.g. Speagle et al. 2014, Barro et al. 2017), sincegalaxies in that epoch contain a higher fraction of star-forming regions. However, many ‘outside-in’ galaxies ac-tually present a ‘truncated’ star-forming pattern, whichcan be revealed in Fig. 6 that ‘outside-in’ galaxies typi-cally present a decreasing trend in star formation alongradius (and decreases faster after reaching 1.25 R e ) in-stead of a flat profile. Therefore, we consider that theyare not forming stars in a uniform manner throughouttheir optical extensions.Our result for ‘outside-in’ galaxies is in good agree-ment with the compaction model, which predicts thecentral enhanced star formation (Dekel & Burkert 2014;Tacchella et al. 2016a; Tacchella et al. 2016b) accompa-nied by the suppression or consistency (Tacchella et al.2016b) in outer regions. This is what we have observedin the case of stacking SGMS for ‘outside-in’ galaxies(Fig. 5) and in some of their G-by-G SGMS (but notall). These galaxies have higher surface densities andare more concentrated in optical morphologies (Wanget al. 2017a). We also have checked that star forma-tion in ‘outside-in’ galaxies tend to be more compactand centralized, by measuring the concentration param-eter (Conselice 2014) of their EW(H α ) maps, which are Q.Liu et al. roughly equivalent to sSFR maps. The preferred loca-tions above the main sequence for ‘outside-in’ galaxieswith M ∗ > M (cid:12) are also consistent with Tacchellaet al. (2016a) and correspond to the central enhancedSFGs observed in Ellison et al. (2018). The compactionprocess may be attributed to a series of physical mech-anisms, such as disk instability, bar-induced inflow, andinteraction with the environment. This process is ex-pected to happen more frequently in high-redshift cos-mic epochs (z > CONCLUSIONWe investigate the spatially resolved SFMS in ∼ ∗ –Σ SFR plane, with ‘outside-in’ galaxies appearingto be steeper, and more elevated than ‘inside-out’ galax-ies at higher surface densities. We further inspect theSGMS in individual galaxies and find their distributionsof slopes to be clearly separated. ‘Inside-out’ galaxiesshow an increase in their slopes with increasing sSFR and an anti-correlation with the specific angular momen-tum, while ‘outside-in’ galaxies have a larger scatter intheir distribution of slopes. By normalizing and stackingthe SGMS for each galaxy, we find a clear suppressionin inner regions of ‘inside-out’ galaxies and a less signifi-cant suppression in outer regions of ‘outside-in’ galaxies,leaving the remaining parts following a slightly sublinearscaling relation with a slope of ∼ . GMS of Two Modes A. RESOLUTION EFFECT WITH MANGA FWHM PSFThe reconstructed PSF of the MaNGA datacube is 2.5 (cid:48)(cid:48) in FWHM. Therefore, according to the Nyquist samplingtheorem, those galaxies in the MaNGA sample with R e < (cid:48)(cid:48) (1 PSF) are not resolved, and those with 2.5 (cid:48)(cid:48) < R e < (cid:48)(cid:48) (1 ∼ R e [arcsec] N u m b e r Figure A.1. R e histogram for our 561 MaNGA galaxies used in the main analysis. The red dashed line stands for the MaNGAFWHM PSF (2.5 (cid:48)(cid:48) ), and the blue dashed line stands for two times of PSF (5 (cid:48)(cid:48) ). Fractions of galaxies lying between the PSFlimits are indicated in numbers at the bottom. In Figure A.1, we show the R e distribution of our whole sample, including both ‘inside-out’ and ‘outside-in’ galaxies.The red dashed line represents the FWHM PSF of MaNGA and the blue dashed line represents two times of the PSFof MaNGA. For our sample, about 5% of the whole population have R e smaller than the FWHM, which have alreadybeen excluded in our main analysis because these galaxies might induce severe bias. About 39% of our sample haveR e lying between 1 ∼ e > e > e , and thus they are more susceptible toresolution effects. However, the overall distributions and trends remain the same; therefore, we conclude that ourresults are not significantly influenced by resolution effects and our main conclusions still hold.0 Q.Liu et al. * (dex) S F R ( d e x ) Inside-out FoV>2.5"Inside-out FoV>5" * (dex) S F R ( d e x ) Outside-in FoV>2.5"Outside-in FoV>5"
Figure A.2.
Contours showing 30% / 1 σ (67%) / 2 σ (95%) / 3 σ (99%) distributions of the stacking of G-by-G SGMS beforeand after excluding galaxies with PSF < R e < e < PSFwhereas green lines represent those with R e > e < PSF whereas orange lines represent those with R e > R [R e ] l o g E W ( H )[ Å ] Inside-outOutside-in
Figure A.3.
Reproduction of Fig. 6 using galaxies with R e > α ) in red (‘inside-out’ galaxies) and blue (‘outside-in’ galaxies), whereas the bands represent their 30%–70% distributions.For comparison, profiles using galaxies with R e > PSF (95% of data) are shown in dashed lines (i.e. solid lines in Fig. 6).
REFERENCES
Abadi, M. G., Moore, B., & Bower, R. G. 1999, MNRAS,308, 947Abdurro’uf, & Akiyama, M. 2017, MNRAS, 469, 2806Abolfathi, B., Aguado, D. S., Aguilar, G., et al. 2017,arXiv:1707.09322Bacon, R., Conseil, S., Mary, D., et al. 2017, A&A, 608, A1Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981,PASP, 93, 5Balogh, M. L., Morris, S. L., Yee, H. K. C., Carlberg,R. G., & Ellingson, E. 1999, ApJ, 527, 54Barro, G., Faber, S. M., Koo, D. C., et al. 2017, ApJ, 840,47Belfiore, F., Maiolino, R., Bundy, K., et al. 2018, MNRAS,Boselli, A., & Gavazzi, G. 2014, A&A Rv, 22, 74 Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004,MNRAS, 351, 1151Bruzual A., G. 1983, ApJ, 273, 105Bundy, K., Bershady, M. A., Law, D. R., et al. 2015, ApJ,798, 7Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, ApJ,533, 682Cano-D´ıaz, M., S´anchez, S. F., Zibetti, S., et al. 2016,ApJL, 821, L26Cappellari, M. 2016, ARA&A, 54, 597Conselice, C. J. 2014, ARA&A, 52, 291Daddi, E., Dickinson, M., Morrison, G., et al. 2007, ApJ,670, 156Dekel, A., & Burkert, A. 2014, MNRAS, 438, 1870
GMS of Two Modes11