SDSS-IV MaNGA: The Radial Profile of Enhanced Star Formation in Close Galaxy Pairs
Joshua L. Steffen, Hai Fu, J. M. Comerford, Y. Sophia Dai, Shuai Feng, Arran C. Gross, Rui Xue
DD RAFT VERSION F EBRUARY
9, 2021
Preprint typeset using L A TEX style emulateapj v. 01/23/15
SDSS-IV MANGA: THE RADIAL PROFILE OF ENHANCED STAR FORMATION IN CLOSE GALAXY PAIRS J OSHUA
L. S
TEFFEN , H AI F U , J. M. C OMERFORD , Y. S OPHIA D AI ( 戴 昱 ) , S HUAI F ENG , A
RRAN
C. G
ROSS , AND R UI X UE Draft version February 9, 2021
ABSTRACTWe compare the radial profiles of the specific star formation rate (sSFR) in a sample of 169 star-forminggalaxies in close pairs with those of mass-matched control galaxies in the SDSS-IV MaNGA survey. We findthat the sSFR is centrally enhanced (within one effective radius) in interacting galaxies by ∼ ∼ M / M (cid:12) ) = 9.0 − Subject headings: galaxies: star formation — galaxies: nuclei — galaxies: interactions — galaxies: massevolution INTRODUCTIONIn the Λ CDM model, galaxy evolution is a hierarchical pro-cess. In this model, massive galaxies are the product of sev-eral past merger events of smaller galaxies. In fact, cosmo-logical hydrodynamical simulations have shown that repeatedmerger events may be responsible for as much as ∼
60% ofstellar mass in massive galaxies like M87 (e.g., Rodriguez-Gomez et al. 2016; Pillepich et al. 2018). As the galaxiesundergo the merging process, the gas within the galaxies aresubjected to gravitational torques which perturb the morphol-ogy of the galaxies.The internal dynamics of these interacting galaxies werefirst modeled in the seminal work, Toomre & Toomre (1972).Since then, hydrodynamical simulations have expanded uponthe N-body simulations of Toomre & Toomre (1972) by mod-eling gas-dynamics within the galaxies. These simulationsshow the process by which barred structures develop withinthe disks of the interacting galaxies due to the tidal torquesbetween them (Barnes & Hernquist 1991). As the bars form,the gases within the galaxy’s disk lose angular momentum andget funneled into the centers of the galaxies.When the gas-inflows impact upon the gases in the nu-cleus of the galaxy, a burst of new star formation is triggered(Barnes & Hernquist 1996; Mihos & Hernquist 1996). These Department of Physics & Astronomy, University of Iowa, Iowa City,IA 52242 Insititute for Astronomy, University of Hawaii, Honolulu, HI 96822 Department of Astrophysical and Planetary Sciences, University ofColorado, Boulder, CO 80309 Chinese Academy of Sciences South America Center for Astronomy(CASSACA)/National Astronomical Observatories of China (NAOC),20A Datun Road, Beijing 100012, China Department of Physics, Hebei Normal University, 20 South ErhuanRoad, Shijiazhuang, 050024, China Key Laboratory for Research in Galaxies and Cosmology, Shang-hai Astronomical Observatory, Chinese Academy of Sciences, 80 NandanRoad, Shanghai 200030, China University of the Chinese Academy of Sciences, No.19A YuquanRoad, Beijing 100049, China gas inflows will also bring metal-poor gases from the disk intothe center of the galaxy which can dilute the central metallic-ity (Rupke et al. 2010; Perez et al. 2011; Scudder et al. 2012).The gas-inflows may also be able to reach into the very cen-ter of the galaxy and trigger an episode of supermassive blackhole (SMBH) accretion (Capelo et al. 2017).Interaction induced star formation was first seen observa-tionally in the bluer colors of peculiar galaxies in Larson& Tinsley (1978). Similar observations have has also beenshown in more recent works using the single-fiber spectro-scopic survey, SDSS (Sloan Digital Sky Survey) (Ellison et al.2008; Li et al. 2008; Scudder et al. 2012; Patton et al. 2013;Bustamante et al. 2020). From these previous works it hasbeen shown that the strength of the star formation enhance-ment in the centers of paired galaxies is dependent on the stel-lar mass of the pairs (Li et al. 2008), the projected separationbetween the pairs (Ellison et al. 2008; Li et al. 2008; Scud-der et al. 2012), and the mass ratio between the pairs (Ellisonet al. 2008).The previously mentioned works using SDSS were re-stricted to studying the centers of the paired galaxies through1-1.5 (cid:48)(cid:48) -radius optical fibers. With the recent large integralfield spectroscopic (IFS) surveys, interacting galaxies cannow be studied with unprecedented spatial detail. These sur-veys allow us to study the centers of merging galaxies morerigorously since apertures can be set to the physical scale ofthe galaxies instead of being bound by a fixed sky aperture.These IFS surveys will also allow us to see the extent of thecentrally induced star formation and to see how the star for-mation in the disks of the galaxies are affected.Indeed, Barrera-Ballesteros et al. (2015) used the CALIFA(Calar Alto Legacy Integral Field Area) survey to study a sam-ple of 103 paired galaxies by varying the size of the aperturethrough which the EW(H α ) is extracted. In this study, theyfound a moderate enhancement to the sSFR in the centers ofpaired galaxies and a moderate suppression to the sSFR inoutskirts of the paired galaxies.Pan et al. (2019) used the SDSS-IV MaNGA survey to a r X i v : . [ a s t r o - ph . GA ] F e b study radial profiles of a sample of 205 paired galaxies. Theenhancement to the sSFR was shown to be the strongest inthe centers of the paired galaxies. This central enhancementlinearly fell with increasing galactocentric radii; however, amoderate enhancement to the sSFR remains in the outskirts ofthe galaxies. Pan et al. (2019) further studied the paired galax-ies as a function of merger stage, from well separated pairsto post-merger galaxies. Across the different merger stages,the sSFR enhancement was greatest in close pairs with tidalfeatures and in post-merger galaxies. This was in agreementwith previous hydrodynamical simulations which showed thata burst of star formation is triggered after the first pericen-ter and as the two galaxies begin to coalesce (Scudder et al.2012).The radial profile of sSFR in post-merger galaxies has alsobeen studied with the MaNGA survey by Thorp et al. (2019).The post-merger galaxies were shown to have a strong en-hancement to the sSFR in their centers as well as a moderateenhancement in their outskirts.Where previous studies on the radial profile of the sSFR off-sets in paired galaxies have focused on studying the profiles asa function of interaction stage, we will focus on the radial pro-file as a function of the stellar mass, projected separation, andmass ratio. As mentioned previously, these parameters havebeen covered by studies restricted to the nuclear region of thepaired galaxies. With the MaNGA survey, we will be able toexpand upon these previous studies in greater spatial detail.We will be able to analyze how these three parameters affectboth the level of the sSFR offsets in the centers of the pairedgalaxies and the offsets in the outskirts of the galaxies. Wewill also be able to study whether any of the three parametersinfluence the gradient of the sSFR enhancement profiles or ifthe gradient is preserved between different configurations.In our previous work using the MaNGA data included in the14th Public Data Release (DR14; Abolfathi et al. 2018), webuilt a sample of close galaxy pairs where both componentsof the pair were contained within the field of view of a singleintegral field unit (Fu et al. 2018, hereafter Paper I). We foundthat approximately 5.7% of the MaNGA galaxies have a com-panion galaxy contained within the field-of-view of a singleIFU. In this work, we update this sample and supplement itwith a sample of companion galaxies identified outside thefield-of-view of the MaNGA IFUs.This paper is organized as follows; in § 2 we will discussthe properties of the MaNGA survey along with the construc-tion of our pair and control samples, in § 3 we will discusshow we measure star formation rates and how we build radialprofiles of star formation, in § 4 we study the radial profiles asa function of stellar mass, projected separation, and the massratio, in § 5 we compare our work against previous works, andin § 6 we summarize the findings of the work. Throughout weadopt the Λ CDM cosmology with Ω m = 0 . Ω Λ = 0 .
7, and h = 0 . DATA AND SAMPLESMaNGA is an IFS survey at the Apache Point Observa-tory (APO) which uses the SDSS (Sloan Digital Sky Sur-vey) 2.5-meter telescope along with two dual-channel BOSSspectrographs (Drory et al. 2015). MaNGA captures spectrathrough 17 integral field units (IFUs) with variable numbersof fibers: 19, 37, 61, 91, and 127 fibers covering 12.5 (cid:48)(cid:48) , 17.5 (cid:48)(cid:48) ,22.5 (cid:48)(cid:48) , 27.5 (cid:48)(cid:48) , and 32.5 (cid:48)(cid:48) on the sky respectively (Law et al.2015). MaNGA is an optical survey with a spectral coverage of 3600 − ∼ (cid:48)(cid:48) FWHM (Bundy et al. 2015).The MaNGA survey targets galaxies from a subset of41,154 galaxies in the NASA-Sloan Atlas (NSA v1_0_1; ) with a redshift range of0.01 < z < < M i < -24.0, where M i is the rest frame i -band magnitude within thesurvey’s elliptical Petrosian apertures. MaNGA plans to cover10,000 galaxies with a flat stellar mass distribution at two spa-tial coverages, 1.5 R eff and 2.5 R eff (where R eff is the radiuswhich contains 50% of the galaxy’s total light). In this workwe use the data from the 8th MaNGA Product Launch (MPL-8), which covers 6142 unique galaxies observed by July 3,2018. 2.1. Spectral Fitting
We use
SPFIT to model the MaNGA datacubes. TheIDL package is publicly available and was first used inour previous study of close galaxy pairs in MaNGA (Pa-per I). While data products from the MaNGA Data Analy-sis Pipeline (DAP; Belfiore et al. 2019) and PIPE D (Sánchezet al. 2016a,b) are available, SPFIT allows us to combine spec-tra inside a given aperture before fitting the combined spec-trum. The feature alone made it better suited for this project.Additional features of
SPFIT include: (1) simultaneously fit-ting emission lines and stellar continuum, (2) coadding spax-els with either a Voronoi tessellation or arbitrary polygonswhile accounting for covariances, (3) modeling asymmet-ric emission lines, broad AGN emission lines, AGN contin-uum, and various dust-extinction laws, and (4) utilizing multi-threading when processing a large number of datacubes.Here we provide a brief description of the fitting procedureof
SPFIT . Except for broad-line AGN,
SPFIT models the ob-served spectrum as a superposition of emission lines and sim-ple stellar populations (SSPs). The SSP library of
MIUSCAT (Vazdekis et al. 2012) is matched to the MaNGA spectral res-olution and is convolved with the line-of-sight velocity dis-tribution (LOSVD). Both the LOSVD and the profile of theemission lines are parameterized as separate Gauss-Hermiteseries (van der Marel & Franx 1993) to the fourth order.For model optimization, we prefer a fast algorithm likethe Levenberg-Marquardt nonlinear least-squares minimiza-tion algorithm implemented in
MPFIT (Markwardt 2009). Butfor complex spectral models like ours, the success of the fit-ting routine relies on a good initial “guess” solution, whichcan be provided by the penalized pixel-fitting method (pPXF;Cappellari 2017). The pPXF method is robust because itsolves the weights of the templates with a linear algorithm(Lawson & Hanson 1974) independently from solving theGauss-Hermite LOSVD with a nonlinear optimizer (
MPFIT ).The
SPFIT package implements a three-stage fitting proce-dure: First, it masks out spectral regions around emissionlines from the input spectrum and use pPXF on the maskedspectrum with SSP-only templates to obtain the initial “best-fit” parameters of the stellar continuum; Then, it subtractsthe best-fit stellar continuum model from the spectrum anduse pPXF to fit the residual emission-line-only spectrum withGaussian emission-line templates to obtain the initial “best-fit” parameters of the emission lines; Finally, it uses
MPFIT onthe full input spectrum with the two sets of “best-fit” parame-ters from pPXF to simultaneously fit all of the parameters de-scribing the emission lines and the stellar continuum. Saved https://github.com/fuhaiastro/spfit M i N U V - r Red SequenceGreenValley Blue Cloud 1.0 0.5 0.0 0.5 log([N II]/H ) l o g ([ O III ] / H ) SF Comp.AGN
RetiredStarformingCompositeAGN log([N II]/H ) l o g ( E W ( H ) SF sAGNwAGNRG0.0 0.2 0.4 0.6 0.8 1.0
Normalized Number Density F IG . 1.— (Left) Color-magnitude diagram for MaNGA galaxies. (Center) BPT diagram for the MaNGA galaxies. (Right) The WHAN diagram for the MaNGAgalaxies. The grey circles represent the whole MaNGA sample and the color scale reflects the local density around each data point in this color-magnitude plane,as indicated by the color bar on the top. The purple diamonds represent the paired galaxies which are classified as retired (by the EW(H α ) cut), the blue diamondsrepresent the star forming paired galaxies (by the BPT diagram), the green diamonds represent the composite star forming-AGN paired galaxies (by the BPTdiagram), and the Red diamonds represent the paired galaxies containing an AGN (by the BPT diagram). in the final FITS file are the best-fit parameters and their un-certainties, including emission-line fluxes, equivalent widths,and kinematics and luminosity-weighted stellar masses, ages,metallicities, and kinematics.2.2.
Selecting Star Forming Galaxies
We classify galaxies in this survey as star forming galaxiesby selecting galaxies in the blue cloud on the color-magnitudediagram (CMD). We show the CMD for the MaNGA surveyand our pair sample in Figure 1 along with demarcation lineswhich separate the blue cloud, red sequence, and green val-ley. We established the demarcation lines by collapsing theCMD to a color histogram for each of the three regions. Wethen varied the slopes between the regions until we found theslopes which best fit the data. These demarcation lines are;
NUV − r = 3 . − . M i +
18) (1)
NUV − r = 4 . − . M i + , (2)Where NUV − r is the color from SDSS’s k -corrected absolutemagnitude and M i is the i -band magnitude from the NSAcatalog.We use the BPT diagnostic (Baldwin et al. 1981), shownin the center of Figure 1, to remove galaxies with possibleAGN in their centers. Specifically, we use the emission lineratios, log([O III ]/H β ) and log([N II ]/H α ), extracted from a1.3 kpc aperture along with the maximum starburst line ofKewley et al. (2001) as the demarcation between the star-forming branch and the AGN branch.Based on the WHAN diagram, shown in the right panel ofFigure 1, we apply an EW(H α ) ≥ R eff aperture to ensure that all retiredgalaxies are removed from the sample (Cid Fernandes et al.2011).On top of the star formation cuts, we require that all galax-ies are in either the Primary or Secondary MaNGA subsam-ples (Wake et al. 2017), and that the stellar mass range of TABLE 1C
LOSE G ALAXY P AIRS AND M ULTIPLES IN M A NGA IFU S Plate-IFU Index R.A. (J2000) Decl. (J2000) Redshift(deg) (deg)7443-12703 0 229.5255758 +42.7458538 0.04043 ··· ··· ··· ··· ··· ···
OTE . — This table lists a total of 404 plate-IFUs that contain closegalaxy pairs and multiples in the eighth MaNGA Product Launch (MPL-8).The second column lists the component index, where “0” indicates the pri-mary target of the MaNGA observation. For each IFU, all components within ± − of the primary are listed, sorted in ascending angular distancefrom the primary.(This table is available in its entirety in a machine-readable form in the onlinejournal. A portion is shown here for guidance regarding its form and content.) the galaxy sample is between log( M / M (cid:12) ) = 9.0 − Inside-IFU Sample
To identify potential paired galaxies covered by individ-ual IFUs, we start by overlaying SDSS photometric objectsover each MaNGA fields of view. We manually inspecteach MaNGA field, removing photometric objects which areover-deblended galaxy fragments and, very rarely, adding anyobjects missed in the SDSS photometric catalog. At thisstage the object catalog includes foreground stars and fore-ground/background galaxies along with the potential pairedgalaxies.To select paired galaxies out of our object catalog, we in-spect the spectra of each object. The spectra of the identi-fied objects is extracted through a 1 (cid:48)(cid:48) circular aperture andfitted assuming the MaNGA target’s redshift and then manu-ally sorted into the following categories: “good" galaxy spec-tra, broad-line AGN, foreground star, foreground/backgroundgalaxies, or poor S/N objects. The “good" galaxy spectra arethe objects whose spectra are well modeled by
SPFIT at thetarget galaxy’s redshift, whether it is the target galaxy itselfor a nearby companion galaxy. This means that the compan-ion galaxy can be within approximately ± − of theMaNGA target. We found 6573 “good" objects, 57 broad-line AGN, 836 foreground stars, 319 foreground/backgroundgalaxies, and 1546 objects with poor S/N.Broad-line AGN comprise ∼ ∼ ∼ ∼ ∆ v <
500 km s − . Given the redshiftrange of the MaNGA sample and the size of MaNGA’s IFUs,the maximum projected separation for a companion galaxyin the IFU is ∼
40 kpc. Again, the galaxy sample is also re-stricted to galaxies with a stellar mass range of log( M / M (cid:12) ) =9.0 − Outside-IFU Sample
We supplement the inside-IFU sample with a set of pairsidentified outside of the field of view of the MaNGA IFU.We select these outside-IFU pairs from the NSA catalog. Wesearch for these external pairs by selecting objects with a pro-jected separation from the MaNGA targets of r p <
50 kpcusing the MaNGA target’s redshift. We further use a relativevelocity cut of ∆ v <
500 km s − to remove projected galaxiesfrom the selection.From the NSA catalog’s 641,409 galaxies, we find 492galaxies which are paired to MaNGA targets. After restrict-ing the sample to SFGs, we have another 115 MaNGA tar-gets with paired galaxies outside of the IFU. MaNGA targetswhich have both an inside-IFU and an outside-IFU pair areleft to the inside-IFU sample.2.5. Control Sample
To show how the population of galaxy pairs differs fromisolated galaxies, we will compare our pair sample to a sam-ple of control galaxies in the MaNGA survey. We build thiscontrol sample from the MaNGA target galaxies which haveno spectroscopic companions within r p <
50 kpc and ∆ v <
500 km s − either inside or outside of the IFU. This gives us acontrol sample of 1830 star forming control galaxies.It has been shown that the SFR in galaxies is dependenton both the stellar mass and the redshift of the galaxies (e.g.,Noeske et al. 2007). We will compare our interacting galax-ies with control galaxies of a similar stellar mass, redshift,and radial size, to account for these other dependencies. Todo this, we will use two different methods of pair - controlcomparison.In the first method, we simply control for the stellar massbetween the pairs and controls. We split both the pair andcontrol samples into five evenly spaced stellar mass bins overthe range, log( M / M (cid:12) ) = 9.0 − −
60 control galaxies. Since we want each paired galaxy tobe treated in a similar manner, we randomly down-select thetotal number of acquired control galaxies to 20. A given con-trol galaxy may be selected for multiple pairs by the pipeline;most of the control galaxies are used at least once and the av-erage number of times a given control is reused is between2 − R eff by 5% until at least 20 control galaxies are found. 39pairs required 1 extra iteration, 15 pairs required 2 extra iter-ations, 5 pairs required 3 extra iterations, and 2 pairs required4 extra iterations. ANALYSIS METHODS3.1.
Specific Star Formation Rate
We calculate the star formation rates in the pair and con-trol galaxies with the emission lines extracted with our spec-tra fitting code,
SPFIT (described in Section 2.1). We correctthe emission lines for reddening using the extinction curvefrom Cardelli et al. (1989) with updated coefficients fromO’Donnell (1994). The extinction is parameterized as R V ≡ A V / E ( B − V ) = 3.1, where we estimate the value of the V -bandextinction, A V , by comparing the H α /H β ratio to the expectedvalue of 2.85 for Case-B recombination.We measure the SFR from the extinction corrected H α lu-minosity, L H α . We use the SFR formula, Equation 3, fromMurphy et al. (2011) which uses a Kroupa IMF, Solar metal-licity, a constant SFR at an age of 100 Myr, and Case-B re-combination: SFR M (cid:12) yr − = L H α . × erg s − . (3)Since the stellar mass of a galaxy is not uniformly dis-tributed within the galaxy, we normalize the SFR by the stellarmass in the same spaxel, M , giving us the specific star forma-tion rate (sSFR): sSFR = SFR M . (4)The local stellar masses used here is derived from SPFIT ’sbest-fit stellar continuum. Utilizing sSFR, instead of SFR,allows us to compare regions of high mass surface density toregions of low mass surface density.We check our measurement of the specific star formationrate with the equivalent width of the H α line, EW(H α ), sinceit is a known proxy for the sSFR. This is useful as the sSFR isdependent on SPFIT ’s measurement of the stellar mass whileEW(H α ) is an observable. As we will show in Section 4, theradial profiles of EW(H α ) is consistent with radial profiles ofthe sSFR.The signal to noise ratio of the data will decrease with widegalactocentric radii. MaNGA is designed to cover 1.5 R eff forgalaxies in the Primary sample and 2.5 R eff in the secondarysample. These are not hard limits, data will exist beyond theseradii, especially along the semi-minor axis of galaxy; how-ever, the data beyond these limits may be unreliable due tolow signal to noise. To control the quality of the used data,we only use spaxels with S/N ≥ α line. We alsorestrict the used spaxels to those with EW(H α ) ≥ − as the lower limit tothe sSFR in our galaxies.3.2. Radial Profiles
In order to spatially characterize the star formation in thepaired galaxies we build radial profiles of sSFR. First, the ge-ometry of the galaxies needs to be defined. Specifically wewill need the position angle, the inclination angle, and theeffective radius of each of the MaNGA targets. We use the r -band elliptical Petrosian apertures from the NSA catalog andthe r -band Sérsic apertures from Simard et al. (2011) to definethe geometries of the galaxies.The NSA catalog has complete coverage over the MaNGAsample (since MaNGA selects its targets from this cata-log); however, it tends to fail to properly fit paired galaxieswith close on-sky separations. We found that the aperturesfrom Simard et al. (2011) work better for these close pairedgalaxies; however, the catalog does not completely cover theMaNGA sample. We use the NSA catalog for the outside-IFUpair sample because they are well separated on the sky. TheSimard et al. (2011) catalog is used for the inside-IFU sample.If the paired galaxy is not covered by Simard et al. (2011), weuse the ellipse from the NSA catalog.We fit the geometry with apertures from both catalogs whenavailable. When using the first method of pair-control com-parison, where pairs and controls are grouped into stellarmass bins, we fit the control galaxies with the NSA apertures.When making the comparison with the second method, pairedgalaxies fitted with the NSA apertures are compared againstcontrols fitted with the NSA apertures and paired galaxiesfitted with the Simard et al. (2011) apertures are comparedagainst controls fitted with the Simard et al. (2011) apertures.Further, when defining a galaxy’s geometry with the Simardet al. (2011) apertures we also use the masses given in the cat-alog for the total stellar mass of the galaxy. Note that althoughwe carefully treat the two catalogs separately whenever possi- ble, we find excellent agreement in the geometry parametersof control galaxies in both catalogs.We calculate the inclination angle, i , of the galaxies usingthe major-to-minor axis ratios from the elliptical apertures;cos ( i ) = ( b / a ) − q − q , (5)Where b / a is the major-to-minor axis ratio and q is the intrin-sic oblateness. We use the empirically determined oblatenessof q = 0 .
13, from Giovanelli et al. (1994).The inclination angle, along with the galaxy’s position an-gle, is used to deproject the geometries of the galaxies. Weuse the 50% half light radius ( i.e , the effective radius, R eff )to scale the sizes of the galaxies. Doing this will allow us tocompare galaxies of different sizes against each other.Once the geometries of the galaxies are set, we can buildazimuthally averaged radial profiles. The spaxels are binnedinto radius increments of 0.2 R eff from 0.0 − R eff . Withineach radius bin we take the median of the specific star forma-tion rate. We build an azimuthally averaged radial profile foreach MaNGA star forming galaxy and later stack and differ-entiate these profiles in the subsequent analysis.The MaNGA sample does not have a uniform spatial cov-erage, 63% of the sample is designed to cover 1.5 R eff (thePrimary+ subsample) and 37% of the sample is designed tocover 2.5 R eff (the Secondary subsample). This means thatthere will be fewer selected spaxels beyond 1.5 R eff and theS/N of the selected spaxels will be lower than those within1.5 R eff . We decide to still extend our radial profiles out to 2.5 R eff to make full use of the MaNGA data; however, we empha-size that the difference in the sampling of the data within 1.5 R eff and the data beyond 1.5 R eff may create artificial slopes inthe data. RESULTS4.1.
Star Formation Enhancement
With the individual log(sSFR) profiles built for eachMaNGA galaxy, we now use two different methods to com-pare the profiles of paired galaxies against the profiles of con-trol galaxies. In the first method, in § 4.1.1, paired galaxiesare compared to control galaxies within evenly spaced stellarmass-bins. In the second method, in § 4.1.2, paired galax-ies are compared to a subset of 20 control galaxies which arematched in stellar mass and redshift.4.1.1.
Mass-Binned Difference Profiles
In the first method, paired and control galaxies are groupedinto five evenly spaced stellar mass bins between log( M / M (cid:12) )= 9.0 − R eff -scaled galactocentricdistance for both paired and control galaxies. The control pro-files are constructed using all available control galaxies withineach mass bin. The error associated with the “stacked” profileis the standard error of mean of the data at each radius bin.Here we define the standard error of the mean, σ x , as, σ x = σ √ n , (6)where σ is the standard deviation and n is the sample size.We show these stacked profiles for control galaxies in theleft hand panel of Figure 2 and the stacked profiles for pairedgalaxies in the middle panel of Figure 2. Star formation in R/R eff l o g ( s S F R , y r ) Controls log(M/M ) = 9.0 9.5 (398)log(M/M ) = 9.5 10.0 (566)log(M/M ) = 10.0 10.5 (454)log(M/M ) = 10.5 11.0 (322)log(M/M ) = 11.0 11.5 (93)
R/R eff l o g ( s S F R , y r ) Pairs log(M/M ) = 9.0 9.5 (33)log(M/M ) = 9.5 10.0 (40)log(M/M ) = 10.0 10.5 (46)log(M/M ) = 10.5 11.0 (34)log(M/M ) = 11.0 11.5 (16)
R/R eff l o g ( s S F R , d e x ) Pairs - Controls F IG . 2.— The log(sSFR) as a function of galactocentric radius for control galaxies (Left) and galaxy pairs (Middle). The difference between the profiles of thepaired galaxies and the control galaxies are shown in the Right panel. The dashed black profile represents the mean of the difference profiles. The colors of theprofiles represent the mass range of the selected galaxies which is given in the legend along with the number of galaxies in that mass bin in parantheses. Thehighlighted region around the profiles represent the standard error of the mean of the data at the given radius interval. The vertical dash-dot line marks 1.5 R eff ,beyond which the radial sampling is expected to fall. R/R eff l o g ( E W ( H )) Controls log(M/M ) = 9.0 9.5 (398)log(M/M ) = 9.5 10.0 (566)log(M/M ) = 10.0 10.5 (454)log(M/M ) = 10.5 11.0 (322)log(M/M ) = 11.0 11.5 (93)
R/R eff l o g ( E W ( H )) Pairs log(M/M ) = 9.0 9.5 (33)log(M/M ) = 9.5 10.0 (40)log(M/M ) = 10.0 10.5 (46)log(M/M ) = 10.5 11.0 (34)log(M/M ) = 11.0 11.5 (16)
R/R eff l o g ( E W ( H )) Pairs - Controls F IG . 3.— Same as Figure 2 but for EW(H α ). the control galaxies are quenched in their centers with re-spect to their disks and the difference between the sSFR intheir centers and their disks increases with stellar mass, con-sistent with previously published results using the MaNGAsurvey (Belfiore et al. 2018). In contrast, the paired galaxiesshow flatter sSFR profiles where the level of the sSFR remainsroughly consistent across their disks except for paired galax-ies with stellar masses above log( M / M (cid:12) ) = 10.5, which stillshow some central quenching.We then take the difference between the stacked profilesof the paired galaxies and the control galaxies, pair - con-trol, in log space (this means that the difference profiles re-ally represent a ratio between the pairs and controls in lin-ear space). This gives us the difference profile, ∆ log(sSFR),which shows us where the sSFR is enhanced or suppressed(shown in the right hand panel of Figure 2). Across all stellarmass bins, the sSFR of paired galaxies are centrally enhancedby ∼ ± ∼ R eff . In the outskirts of the pairs beyond 1.5 R eff , the pairsfeature lightly suppressed sSFR of 0.0 − M / M (cid:12) ) =11.0 − M / M (cid:12) ) = 11.0 − R eff is significantly higher than the me-dian profile (reaching 0.5 − ∼ × (0.3 dex) greater rates in galaxieswith close companions than in more isolated galaxies, acrossa wide mass range between 9.0 ≤ log( M / M (cid:12) ) ≤ α ) in Figure 3. Wefind that the EW(H α ) profiles are in close agreement withthe sSFR profiles, which shows that we can be confident of SPFIT ’s measurement of the sSFR. We do see that the pair-control offsets are lightly suppressed using the EW(H α ) asthe ∆ log(EW(H α )) is only enhanced by 0.25 dex where the ∆ log(sSFR) was enhanced by 0.30 dex.This method of stacking sSFR profiles by a single param-eter (stellar mass) has the advantage of simplicity and largestatistical samples. It has revealed the first order result ofmass-independent, centrally enhanced star formation in closegalaxy pairs. To proceed with exploring the dependency of ∆ sSFR on merger parameters such as separation and mass ra-tio, we will utilize a more sophisticated method of selectingcontrol samples for individual paired galaxies based on stellarmass and redshift in the following subsections.4.1.2. Tailored-Controls Difference Profiles
In the second method, we match each paired galaxy to aset of 20 control galaxies of similar stellar masses and red-shifts, as described in § 2.5. We then take the median of theazimuthally averaged profiles of the tailored control sample.Finally, we obtain the ∆ logsSFR profile for each of the 169paired galaxies by calculating the difference between its sSFRprofile and the median profile of its control sample.Before delving into merger parameters, we decided to stackthe profiles by stellar mass to see how this method comparesto the previous mass-binning method. We split the individualdifference profiles into five evenly spaced stellar mass binsbetween log( M / M (cid:12) ) = 9.0 − R eff . Thecentral sSFR is calculated for each galaxy by taking spaxelswithin 0.5 R eff and whose H α flux has S/N ≥ ∆ log(sSFR) using the same methodas the ∆ log(sSFR) radial profiles are made. The central sSFRof a paired galaxy is compared against the sSFR of a set of 20similar control galaxies, the same set of controls which wereused for the ∆ log(sSFR) profiles, by taking the difference be-tween the central sSFR of the paired galaxy and the mediansSFR of the 20 selected control galaxies. We depict the cen-tral ∆ log(sSFR) as a function of stellar mass in the right handpanel of Figure 4.The difference profiles are shown to be centrally enhancedby ∼ − ± M / M (cid:12) ) =11.0 − R eff and the outskirts of the galaxies are lightly suppressedby 0.0 − z = 0.15 and a 127 spaxel IFU is about40 kpc.The highest mass galaxies in the sample are at closer pro-jected separations in comparison to the rest of the sample with70% of the highest mass galaxies being within 20 kpc. Thecause of this may be due to clustering effects where high massgalaxies tend to be at the centers of galaxy clusters (Cooray& Sheth 2002; Zehavi et al. 2002). As we will show in § 4.3,the central sSFR enhancement is strongest at small projectedseparation which means that higher sSFR in the centers of themassive galaxies may be driven by their close projected sepa-rations.The results we obtain from the tailored-control method islargely consistent with the mass-binning method. Betweenthe two methods we see that the ∆ log(sSFR) profile is inde-pendent of the total stellar mass of the galaxies. In the nexttwo sections, we will use the tailored-control method to studyhow the difference profiles behave as a function of the massratio (§ 4.2) and projected separation (§ 4.3).4.2. Dependency on Mass Ratio
While the centrally enhanced star formation in close galaxypairs seems mostly mass-independent, the mass ratio of thegalaxy pair, like the projected separation, may be an importantparameter that controls the level of enhancement (Ellison et al.2008).The mass ratio here is defined as, ∆ log( M ) = log( M target ) − log( M comp ) , (7)where M target is the stellar mass of the MaNGA target galaxyand M comp is the stellar mass of its identified companiongalaxy. In the inside-IFU sample, we have stellar masses forthe MaNGA target galaxy but not the other identified pairs.Because of this, we leave the inside-IFU companions out ofthis analysis.In the top left of Figure 6, we split the ∆ log(sSFR) profilesinto four mass ratio bins from | ∆ log(M) | = 0.0 − | ∆ log(M) | ≤ | ∆ log(M) | ≥ ∼ − ∼ − R eff in the top right panel. Herethe ∆ log(sSFR) falls linearly with wider mass ratios, reachingzero enhancement to the sSFR at | ∆ log(M) | = 1.5 − R/R eff l o g ( s S F R , d e x ) log(M/M ) = 9.0 9.5 (33)log(M/M ) = 9.5 10.0 (40)log(M/M ) = 10.0 10.5 (46)log(M/M ) = 10.5 11.0 (34)log(M/M ) = 11.0 11.5 (16) log(M/M ) l o g ( s S F R , d e x ) Outside-IFU PairsInside-IFU Pairs F IG . 4.— The Left panel shows the ∆ log(sSFR) profiles where the difference profiles are constructed from the difference between the paired galaxy profilesand a set of 20 control galaxies. The profiles are split into five different stellar mass bins and the highlighted region about the profiles represent the standard errorof mean of the profile. The black dashed line represents the mean profile between the four difference mass ranges. The number of paired galaxies in each massrange is given in the legend in parentheses. The Right panel shows the nuclear ∆ log(sSFR) extracted from a 0.5 R eff aperture. The black squares are the meanvalues within a stellar mass bin (where the size of the bins are shown the the horizontal error bars). The vertical error bars on the black squares represent thestandard deviation within the bin. The horizontal, dashed black line represents the median central enhancement of the pair sample. Galaxies in the outside-IFU(Blue) and inside-IFU (Red) samples are separately depicted. The vertical dash-dot line marks 1.5 R eff , beyond which the radial sampling is expected to fall. Separation [kpc] F r a c t i o n Full Samplelog(M/M ) = 11.0 - 11.5 F IG . 5.— The projected separation distribution of the whole sample (blue)against the projected separation distribution for the highest mass galaxies,log( M / M (cid:12) ) = 11.0 − In bottom left panel of Figure 6, we split the profiles bymass ratio in four bins. The primary companions are the pairswith ∆ log(M) ≥ ∆ log(M) ≤ ∆ log(sSFR) in comparison to the more massive companionof a pair. In major mergers, the sSFR enhancement is ∼ ∼ ∆ log( M ) = 0.0). These galaxies feature a central enhance-ment of ∼ ∆ log( M ) = -1.0, the sSFR isenhanced by 0.2 dex while at ∆ log( M ) = 1.0 the sSFR en-hancement is zero.We also see that the ∆ log(sSFR) in the outskirts ( R > R eff ) of the pairs have a dependence on the mass ratio in thebottom right of Figure 6. Primary companions show a pos-itive enhancement of 0.0 − − R eff ; however, a similar result has been ob-served in the hydrodynamical simulations of Moreno et al.(2015) and Moreno et al. (2020).From the mass ratio, we see two different effects. First,we see that the central enhancement to the sSFR is strongestfor pairs with 1:1 mass ratios. Second, we see that secondarycompanions show steeper enhancement profiles with higherlevels of sSFR enhancement in their centers and stronger lev-els of sSFR suppression in their disks with respect to primarycompanions. Finally, we see that primary companions featuresSFR enhancement at wide radii while secondary companionsfeature sSFR suppression at wide radii.4.3. Dependency on Projected Separation
A number of previous studies have shown that the sSFR en-hancement increases as projected separation decreases (e.g.,Li et al. 2008; Ellison et al. 2008; Scudder et al. 2012; Pat-ton et al. 2013), in agreement with simulations (Scudder et al.2012).Figure 7 shows the profile and the central level of sSFR en-hancement as a function of the projected separation ( r p ) fromour MaNGA data. The ∆ log(sSFR) profiles show only a weakdependency on the projected separation. The pairs below a R/R eff l o g ( s S F R , d e x ) | log(M)| = 0.0 0.25 (21)| log(M)| = 0.25 0.5 (23)| log(M)| = 0.5 0.75 (20)| log(M)| = 0.75 1.0 (25) | log(M)| l o g ( s S F R , d e x ) Outside-IFU Pairs
R/R eff l o g ( s S F R , d e x ) log(M) = -1.0 -0.5 (21)log(M) = -0.5 0.0 (16)log(M) = 0.0 0.5 (28)log(M) = 0.5 1.0 (24) log(M) l o g ( s S F R , d e x ) Outside-IFU Pairs F IG . 6.— Same as Figure 4 except the difference profiles are split by the mass ratio of the pair. The top row is the absolute value of the mass ratio andthe bottom row is the mass ratio without taking the absolute value. Taking the mass ratio without the absolute value allows us to separately study the moremassive companions of pairs from the less massive companions from pairs. The inside-IFU sample is left out of this analysis because we do not yet have reliablemeasurements of their total stellar mass ratios. R/R eff l o g ( s S F R , d e x ) Separation [kpc] = 0.0 12.5 (53)Separation [kpc] = 12.5 25.0 (32)Separation [kpc] = 25.0 37.5 (34)Separation [kpc] = 37.5 50.0 (50)
Separation [kpc] l o g ( s S F R , d e x ) Outside-IFU PairsInside-IFU Pairs F IG . 7.— Same as Figure 4 except the difference profiles are split by projected separation. ∆ log(sSFR) profiles which lie ∼ ∼ R eff . The level of the enhancement gradually increases withcloser separation from 50 kpc to 10 kpc. While ∆ log(sSFR)falls at higher separations, there is still a substantial level ofenhancement between 40 and 50 kpc, ∼ ∆ log(sSFR) enhancement jumps to ∼ r p = 50 kpc, which is the limit of our pair selection. DISCUSSION5.1.
Radial Profiles of Enhancement
In Figure 8 we compare the ∆ log(sSFR) profile and ∆ log(SFR) as a function of galactocentric radius between ourwork and previous works using the MaNGA data. Thorp et al.(2019) studies a sample of 36 post-mergers from the MaNGAsample. The centers of the post-merger galaxies feature agreater level of sSFR enhancement in their centers comparedto our galaxies pairs of ∆ log(SFR) = 0.40 − − R eff thepost-merger galaxies and our pairs are in closer agreement,being 0.05 − R eff the post-merger galaxies have a higher ∆ log(SFR) byabout 0.1 − ∆ log(sSFR) profiles of the galaxies pairs in Pan et al. (2019)are ∼ − R eff while the galaxy pairs in Pan et al. (2019) showan enhancement of ∼ ∆ log(SFR)when compared to pairs without morphological distortions.Further, Thorp et al. (2019) showed that post-merger galaxieshave ∆ log(SFR) profiles which are elevated over our sample’sprofiles. From this, infer that the lower ∆ log(sSFR) seen inour sample is due to our sample’s lack of post-merger galax-ies. 5.2. Central Enhancement vs. Merger Parameters
While in this work we find that ∆ log(sSFR) has essentiallyno dependence on the stellar mass of the galaxy, Li et al.(2008) found that lower mass galaxies experience greater lev-els of enhancement than higher mass galaxies. Galaxies in themass range log( M / M (cid:12) ) ≥ ∼ M / M (cid:12) ) ≥ R/R eff l o g ( s S F R , d e x ) This WorkPan+19
R/R eff l o g ( S F R , d e x ) This WorkThorp+19 F IG . 8.— The mean radial profile of ∆ log(sSFR) (Top) and ∆ log(SFR)(Bottom) between pairs and controls of this work with the tailored controlmethod ( Black ) compared against those of Pan et al. (2019) (
Red ), a pairsample, and Thorp et al. (2019) (
Blue ), a post-merger sample. hancement function as the average sSFR of the paired galaxiesat a given separation, weighted by the number of companions,subtracted by the average sSFR of the whole sample. Whilewe do not find the same dependency between the sSFR en-hancement and stellar mass, the sSFR enhancement as a func-tion of projected separation between our studies is in goodagreement (see Figure 9).The dependency of the SFR enhancement on the mass ratiowas studied in Ellison et al. (2008). They found that pairs withmass ratios of 2:1 have higher SFR enhancements of ∼ ∆ log( M ) = 0.25 (1.8:1) are 0.2 dex over pairswith mass ratios of ∆ log( M ) = 1.0 (10:1). Ellison et al. (2008)also saw tentative evidence for SFR enhancement in the sec-ondary companions of minor pairs. We not only confirm that1 l o g ( S F R , d e x ) This WorkEllison+08Scudder+12Patton+13Bustamante+20
Separation [kpc] l o g ( s S F R , d e x ) This WorkLi+08Patton+20 F IG . 9.— (Top) ∆ log(SFR) extracted from the inner 0.5 R eff over projectedseparation from this study ( Black ), Ellison et al. (2008) (
Red ), Scudder et al.(2012) (
Blue ), Patton et al. (2013) (
Green ), and Bustamante et al. (2020)(
Purple ). (Bottom) ∆ log(sSFR) extracted from the inner 0.5 R eff over pro-jected separation from this study ( Black ), Li et al. (2008) (
Red ), and Pattonet al. (2020) (
Blue ). secondary companions of minor pairs feature SFR enhance-ment, but also that the enhancement is higher than the primarycomponent of pairs at the same mass ratios.The observations are also consistent with simulations.Moreno et al. (2015) used G ADGET -3, a smoothed particlehydrodynamics code, to study the spatial extent of the starformation enhancement. The merger simulations showed thatthe star formation enhancement was largely concentrated inthe centers (R < < R <
10 kpc) showed a suppression tothe star formation. Moreno et al. (2015) also found that lowermass secondary galaxies (in mergers with stellar mass ratiosof 2.5:1) have higher levels of sSFR enhancement in theircenters and have stronger levels of sSFR suppression in theirdisks. This is in good agreement with our work, we foundthat secondary companions show moderately higher levels ofsSFR enhancement in their centers in comparison to primarycompanions at the same mass ratio. Further, the outskirts ofour primary companions feature positive enhancement to theirsSFR while the secondary companions show a strong suppres-sion to the sSFR in their outskirts.Figure 9 compares the enhancement in both SFR and sSFR as a function of projected separation. For our data points, weuse the mean ∆ log SFR and ∆ log sSFR measured within adeprojected radius of 0.5 R eff (i.e., the black data points inthe right side panel of Figure 7). The literature data is mea-sured from single 1-1.5 (cid:48)(cid:48) -radius SDSS fibers (Ellison et al.2008; Scudder et al. 2012; Patton et al. 2013; Bustamanteet al. 2020). We find that our central SFR enhancements forclosely separated pairs are higher than many of the previousstudies, our sSFR enhancement is ∼ ∼ ∼ − ∆ log(sSFR)enhancements across 2D separation (projected separation)from the TNG300-1 simulation. Patton et al. (2020) found asSFR enhancement which is 1.8 × that of the isolated controlswithin a separation of 15 kpc and is statistically significant outto a 2D separation of ∼
250 kpc. This enhancement is ∼ R eff aperture while Pat-ton et al. (2020) extracted the sSFR from a 50% half-massradius. This means that our extraction radius is effectivelytwice a small and is more restricted to the central burst of starformation.Our sample only covers projected separations within 50kpc, while previous surveys cover out to 100 −
200 kpc. Whileour sample covers a smaller separation range, the ∆ log(SFR)of our sample at 50 kpc is roughly the same level as Ellisonet al. (2008) and Bustamante et al. (2020) at the same pro-jected separation. From this we see that our sSFR enhance-ment as a function of projected separation is consistent withwith what has been found in previous works.5.3. Comparison with Simulations
Our results are consistent with the idea that galaxy mergerevents trigger a burst of star formation in the centers of thepaired galaxies. In this model, when two galaxy pairs un-dergo their first pericenter, the tidal torques exerted on thedisks of interacting galaxies form barred structures. Thesebarred structures funnel gases into the centers of the galaxieswhich triggers a burst of new star formation. Eventually, asthe pairs separate from each other after the first pericenter, theburst of star formation begins to subside. This is shown in ourwork and many previous works in Figure 9 where the sSFRenhancement is greatest for close separations and falls withwider separations.We also found that the ∆ log(sSFR) is independent of thestellar mass. This means that the merger event will producethe same amount of new stellar mass for low mass galaxies asit produces for high mass galaxies. This also means that thelow mass galaxies will experience a greater change in totalstellar mass before and after the merger event so the mergerevent will have a greater impact on the mass evolution of lowmass galaxies as opposed to high mass galaxies.We find that the strength of the central burst of merger in-duced star formation is dependent on the relative masses be-tween the two galaxies. Equal mass galaxies see the strongestbursts of star formation while wide mass ratios see weakerbursts of star formation. We further find that the strength2of the central burst of star formation differs between the pri-mary and secondary companion of a pair where the less mas-sive secondary companion features a higher level of sSFR en-hancement in comparison to its higher mass primary compan-ion.Moreno et al. (2020) uses galaxy merger simulations tostudy the origin of the enhanced star formation in galaxypairs, whether its an increase in the star formation efficiency,defined as the ratio between the SFR and the mass of cold-dense gas, or its an increase in the availability of cold densegas. They find that the star formation in secondary galaxiesis evenly split between being efficiency driven or fuel drivensystems while primary galaxies are more likely to be fueldriven systems (71%). Moreno et al. (2020) also finds thatthat secondary galaxies feature higher levels of star formationenhancement in comparison to primary galaxies, which is inagreement with what we find in this work. This indicates thatthe reason why primary galaxies behave differently from sec-ondary galaxies may be due to a difference in the physicalmechanism which drives the enhanced star formation.We also find that there is a difference in the ∆ log(sSFR)offsets at wide radii between primary and secondary com-panions. The primary companion features an enhancement totheir sSFR at wide radii while the secondary companion fea-tures a suppression to their sSFR at wide radii. The reason forthis is unclear; however, this means that the primary compan-ions will experience enhanced stellar mass growth across theirdisks while the secondary companions will see only see sub-stantial stellar mass growth in their centers which will result inmore bulge dominated galaxies. Differences in the bulge-to-total ratio, (B/T), has been observed in previous works likeBluck et al. (2019) who found that satellites tend to haveslighter higher (B/T) ratios in comparison to central galaxiesat the same stellar mass. SUMMARY AND CONCLUSIONIn this paper, we have demonstrated the power of a mas-sive integral-field spectroscopic survey in comparison studiesof galaxy properties. The nearby galaxy populations showconsistent behaviors despite of their large diversity in starformation properties. Stacking was able to detect the signalburied in the noise by averaging, in the time domain, the likelystochastic star forming histories of galaxies (which drives thescatter in the sample). We focused on comparing the az-imuthally averaged radial profiles of sSFR between galaxiesin close pairs and a control sample of isolated galaxies. Inagreement with previous studies, we found that, on average,star formation is elevated in close galaxy pairs. The propertiesof these purported merger-induced differences in sSFR can besummarized as follows:1. Star formation is enhanced within the inner 1.0 R eff andit peaks at a level of 0.20 − ± ∼ × faster star formation). On the other hand, the outskirtsof the paired galaxies ( R eff = 1.0 − ∼ | ∆ log( M ) | = 0.0 − ∼ − | ∆ log( M ) | = 1.0.5. The merger-induced changes in sSFR also seem to dif-fer between the more massive and the less massivemember of a galaxy pair. At the same mass ratio, theless massive member in a galaxy pair shows a highersSFR enhancement (by ∼ − R eff > ∼ (cid:48)(cid:48) spatial sampling up to 2-3 R eff