The AGN Luminosity Fraction in Merging Galaxies
Jeremy Dietrich, Aaron S. Weiner, Matthew L.N. Ashby, Christopher C. Hayward, Juan Rafael Martínez-Galarza, Andrés F. Ramos Padilla, Lee Rosenthal, Howard A. Smith, S.P. Willner, Andreas Zezas
MMNRAS , 1–16 (2018) Preprint 3 July 2018 Compiled using MNRAS L A TEX style file v3.0
The AGN Luminosity Fraction in Merging Galaxies
Jeremy Dietrich, , (cid:63) Aaron S. Weiner, , Matthew L.N. Ashby, Christopher C. Hayward, Juan Rafael Mart´ınez-Galarza, Andr´es F. Ramos Padilla, , Lee Rosenthal, Howard A. Smith, S. P. Willner, and Andreas Zezas , , Harvard-Smithsonian Center for Astrophysics, 60 Garden St, Cambridge, MA 02138 Department of Astronomy and Steward Observatory, University of Arizona, 933 N Cherry Ave, Tucson, AZ 85719 Department of Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic Institute, 110 8th St, Troy, NY 12180 Center for Computational Astrophysics, Flatiron Institute, 162 Fifth Avenue, New York, NY 10010 Leiden Observatory, Leiden University, P.O. Box 9513, 2300RA Leiden, The Netherlands Department of Astronomy, California Institute of Technology, Pasadena, CA 91125 Physics Department & Institute of Theoretical & Computational Physics, University of Crete, 71003 Heraklion, Crete, Greece Foundation for Research and Technology-Hellas, 71110 Heraklion, Crete, Greece
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
Galaxy mergers are key events in galaxy evolution, often causing massive starburstsand fueling active galactic nuclei (AGN). In these highly dynamic systems, it is not yetprecisely known how much starbursts and AGN respectively contribute to the totalluminosity, at what interaction stages they occur, and how long they persist. Here weestimate the fraction of the bolometric infrared (IR) luminosity that can be attributedto AGN by measuring and modeling the full ultraviolet to far-infrared spectral energydistributions (SEDs) in up to 33 broad bands for 24 merging galaxies with the Codefor Investigating Galaxy Emission. In addition to a sample of 12 confirmed AGN inlate-stage mergers, found in the
Infrared Astronomical Satellite
Revised Faint SourceCatalog, our sample includes a comparison sample of 12 galaxy mergers from the
Spitzer
Interacting Galaxies Survey, mostly early stage. We also SED modeling ofmerger simulations to validate our methods, and we supplement the SEDs with mid-IR spectra of diagnostic lines obtained with
Spitzer ’s InfraRed Spectrograph. Theestimated AGN contributions to the IR luminosities vary from system to system from0% up to ∼
91% but are significantly greater in the later-stage, more luminous mergers,consistent with what is known about galaxy evolution and AGN triggering.
Key words: galaxies: interactions – galaxies: photometry – galaxies: star formation– galaxies: active galactic nuclei – infrared: galaxies
Galaxy interactions have long been known to influencegalaxy evolution, and a large majority of galaxies in the uni-verse show signs of previous interactions (e.g., Struck 1999).Mergers trigger enhanced star formation (SF) and generateor fuel active galactic nuclei (Sanders et al. (1988); Hong etal. 2015, Brassington et al. 2015 and references therein). Inaddition, mergers produce disturbed morphological features(Toomre & Toomre 1972; Lanz et al. 2013 [hereafter L13]and references therein). Tidal tails and nuclear disruptionsare the most obvious indications that two or more galaxiesare interacting or merging. (cid:63)
E-mail: [email protected]
The enhanced SF seen in galaxy mergers is, in mostcases, the main power source for (ultra) luminous infraredgalaxies ([U]LIRGs) in the local universe (Sanders & Mirabel1996). Veilleux, Kim, & Sanders (2002) have shown thatmany of these galaxies contain clear morphological indica-tors of past galaxy interactions. However, not all galaxymergers have enhanced IR emission. The star formation rate(SFR) depends on the merger stage, as SF and AGN activ-ity are enhanced in the later stages of mergers (Lackner etal. 2014).In efforts to address when and how SF and AGN activ-ity proceed, (U)LIRGs and other luminous late-stage merg-ers have become prime targets for space-based telescopessuch as the
Spitzer Space Telescope (Werner et al. 2004),the
Herschel Space Observatory (Pilbratt et al. 2010), the © a r X i v : . [ a s t r o - ph . GA ] J un J. Dietrich et al.
Galaxy Evolution Explorer (GALEX) (Martin et al. 2005),and
Swift (Gehrels et al. 2004) among others. This suiteof instruments provide highly reliable photometry by virtueof their privileged vantage point above Earth’s atmosphere,and they are sensitive to the multiple processes contributingto galaxies’ spectral energy distributions (SEDs):
Spitzer ’sinfrared cameras measure the enhanced mid-infrared (MIR)emission from both AGN and SF,
Herschel views emissionprincipally from the diffuse galactic dust, and the UV satel-lites are sensitive to the emission arising from young, hotstars.SEDs that span the range from UV to FIR wavelengthsreflect all significant energetic processes occurring in galax-ies. They are therefore indispensable for inferring galaxies’underlying physical properties, including but by no meanslimited to SFRs, masses and luminosities of the galactic dustand the effects of AGN (Hayward & Smith 2015). For galaxymergers, the MIR emission arises from dust heated by bothSF and AGN (Gruppioni et al. 2008). However, the relativeproportions each process contributes are not well understoodand vary enormously over time and from one system to an-other. Moreover, the ability to detect high-redshift galaxiesis increasing, but the SEDs of these distant sources are mucheasier to obtain than spatial morphology and are thereforeour best keys to understand the physical processes under-way. A clear understanding of SEDs in the local universe isa prerequisite for drawing reliable conclusions about distantsources.Many studies are being performed to calculate the frac-tion of luminosity that comes from the AGN in merginggalaxies. Studies utilize wide wavelength ranges (from totalIR to UV/X-ray) and span redshifts from the local universe( z < . ) out to the distant universe ( < z < ) (e.g.,Ciesla et al. 2015; Drouart et al. 2016; Fernandez-Ontiveroset al. 2016; Vaddi et al. 2016; Villforth et al. 2017). Other re-cent studies have also characterized the SEDs out to 24 µ mof radio-loud AGN (Williams et al. 2017) or specifically fo-cused on the far-IR SED shape, where dust and AGN emis-sion dominate (Safarzadeh et al. 2016; Cowley et al. 2017).Accurate constraints on the AGN contribution to the to-tal luminosity are necessary for precise estimates of otherenergetic processes such as SFR. In galaxies at cosmologicaldistances for which primarily photometric data are available,we can ascertain which physical processes are providing theluminosity for the galaxy.This work presents an analysis of 24 merging galaxiesorganized into two samples. First we re-measure in a uniformand self-consistent manner and then analyze the full SEDsof 12 late-stage merging (U)LIRGs and estimate their frac-tional AGN contributions across the entire IR range from1–1000 µ m (hereafter referred to as ‘total IR’). Our SEDscover UV to far-IR/submillimeter wavelengths, providing acomprehensive view into the processes at work in merginggalaxies. We compare the results for late-stage mergers tothose for 12 previously studied early-stage mergers.This paper is organized as follows. Section 2 presents thegalaxy samples, the observations, and the data reduction.Section 3 describes the SED analysis. Section 4 describes thesame type of SED analysis of merger simulations. Section 5discusses results, and Section 6 summarizes conclusions. We chose the late-stage merger sample to represent stronglyinteracting, infrared-luminous systems. Candidate systemswere identified by the
Infrared Astronomical Satellite ( IRAS ,Neugebauer et al. 1984) Revised Faint Source Catalogue(FSC, Wang et al. 2014). We selected interacting systemsby cross-referencing
IRAS sources with Version 2 of theGalaxy Zoo public galaxy classification program (Willett etal. 2013) to yield 453 systems. Of these, 85 have far in-frared luminosities at least in the ‘luminous infrared galaxy’(LIRG: L IR > L (cid:12) ) range, and 7 are in the ULIRG ( L IR > L (cid:12) ) range. We classified all these systems by mergerstage (Weiner et al. 2018, in preparation) finding 38 withmorphological evidence for strong interaction such as longtidal tails or heavily disturbed morphology. These systemsare designated as “Stage 4” or later by Weiner et al. (2018,in preparation). Of these 38 galaxies, only 12 had availablephotometry from all of Spitzer /IRAC,
Spitzer /MIPS 24 µ m,and Herschel /SPIRE at 250, 350, and 500 µ m. These 12constitute the late-stage merger sample listed in Table 1. Byconstruction the sample is pure for strong interactions andhigh luminosities–indeed 11 of the 12 are in the top 20 lumi-nosities of all 453 in the Galaxy Zoo sample. However, thesample is far from complete because of our requirement tohave full data sets, especially Herschel data. This may haveintroduced a bias toward ‘interesting’ systems and there-fore ones with extreme properties, but any bias is probablyrelated to obvious properties such as morphology and lumi-nosity rather than parameters that can be revealed only bydetailed analysis.For a control set to compare with the late-stage mergers,a ‘Reference Sample’ was drawn from the SIGS galaxy sam-ple (Brassington et al. 2015). SIGS consists of 103 galaxiesin 48 systems selected by a combination of galaxy proxim-ity on the sky and morphological disturbance. SIGS there-fore includes all merger stages from non-interacting systemsto early approach to late stages (Brassington et al. 2015,L13). From the SIGS sample, we selected 12 galaxies withUV–submm photometry comparable to what was availablefor the Late-Stage Merger Sample. We adopted the mergerstage classifications from L13 for these objects. Nearly allof them are Stages 2–3 implying at least mild but at mostmoderate distortions and galaxies still separated from eachother (Weiner et al. 2018, in preparation). The ReferenceSample members are listed in Table 1. The sample is delib-erately heterogeneous but contains a range of systems thatare merging but have not yet reached the final merger stage.The requirement for many-band photometry introduces abias toward well-studied systems, which are likely if anythingto be those with especially strong merger signatures, i.e., theReference Sample probably resembles the Late-Stage MergerSample more closely than the full SIGS sample would.One difference between the samples is their redshiftdistributions. The Reference Sample galaxies all lie within z ≤ . , but the Late-Stage Merger Sample galaxies arein the range . ≤ z ≤ . . This reflects the fact thatlate-stage mergers are less common than early-stage ones—the early stages last longer than late stages—and it’s nec-essary to search larger distances to find equal numbers of MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies late-stage systems. None of the galaxies requires a signifi-cant k-correction, and all inhabit the present-day universe. For most galaxies in the Reference Sample, we assem-bled photometry for single galaxies rather than the en-tire merging systems to ensure the most reliable SEDfits. This included even the merging systems M51A/B,M101/NGC 5474, NGC 3031/3077, and NGC 5394/5395.The advanced merger NGC 3690/IC 694 system is an ex-ception; it had to be observed as a single blended entity.Galaxies in the Late-Stage Merger Sample could only be im-aged as a single source. Table 1 lists the physical parametersand previously known AGN status for each galaxy. Detailsabout the SED fit for individual objects can be found inAppendix A.Photometry for all galaxies in both samples usedmatched apertures on archival broadband images of up to 33different bands (following L13). We started with the near-and far-UV bands from
GALEX
Martin et al. (2005). Atvisible wavelengths, we used ugriz imaging from the SloanDigital Sky Survey (SDSS) Data Release 12 (Alam et al.2015). For the near-IR bands, we used
JHK s imaging fromthe Two Micron All-Sky Survey (2MASS; Skrutskie et al.2006). Mid-IR comprised Spitzer
Infrared Array Camera(IRAC) 3.6, 4.5, 5.8, and 8.0 µ m imaging and MultibandImaging Photometer for Spitzer (MIPS) 24, 70, and 160 µ mimaging. We also used the IRAS (Neugebauer et al. 1984)bands at 12, 25, 60, and 100 µ m. For the late-stage merg-ers we also incorporated near- and mid-IR imaging from the Wide-Field Infrared Survey Explorer ( WISE , Wright et al.2010) at 3.4, 4.6, 12, and 22 µ m. Most of the W ISE imag-ing was taken before the W4 filter was recalibrated (Brown,Jarrett, & Cluver 2014), but because the
WISE µ m dataare outweighed in the fitting by the MIPS 24 µ m and theIRAS 25 µ m data, the difference in the SED fit betweenthe previous and new calibrations of W4 is negligible. Fi-nally, for the far-IR bands we used archival imaging fromthe Herschel
Photoconductor Array Camera and Spectrom-eter (PACS) 60–90 µ m , 90–130 µ m , and 130–210 µ m bandsas well as the Herschel
Spectral and Photometric ImagingReceiver (SPIRE) bands at 250, 350, and 500 µ m. For the Herschel imaging we used the
Herschel
Interactive Process-ing Environment (HIPE), version 14.1 (Ott 2010).Some datasets required special handling. The publiclyavailable archival IRAC mosaics for IRAS 08572+3915 andMrk 231 were not suitable for photometry. The Mrk 231mosaics (specifically, the post-basic calibrated or PBCD mo-saics) were saturated in all four IRAC bands. The 5.8 and8.0 µ m PBCD mosaics for IRAS 08573+3915 also show con-spicuous saturation. We therefore constructed our own mo-saics for these two objects using only the short exposures(0.6 s) from archived IRAC high-dynamic range observa-tions. After first verifying that the resulting short-exposuremosaics showed no saturation, we used them for our pho-tometry in place of the publicly available IRAC mosaics. Inaddition, we adopted the global IRAS photometry from the
IRAS
FSC.We assembled the non-
IRAS photometry for all 24galaxies following the procedure described by Weiner et al.(2018, in preparation). We started by subtracting the sky background using the Python package photutils and usedan elliptical aperture to capture all the flux. We took care touse the same aperture area to enclose the full galaxy emis-sion regions in all bands and to correct for any backgroundemission. Our photometric values are consistent with butmore precise than results in the open literature typicallyobtained in pipeline processing of larger datasets. The pho-tometry can be found in Appendix B.For the early-stage mergers from L13, we used the pho-tometry stated in the paper and added SDSS ugriz photom-etry, which was processed the same as by Weiner et al. (2018,in preparation). Photometric uncertainties were calculatedusing the sum in quadrature of Poisson photon noise de-termined by the photometry and calibration uncertainties,and we adopted the same calibration uncertainties as L13,following the references therein. We supplemented the SEDs for the late-stage merger sam-ple with a previously underutilized resource: spectrophoto-metric continuum measurements taken from PACS spectralscans. The PACS spectrophotometric data provide excel-lent coverage of the peak of the FIR continua. Multipleobservers obtained PACS range or spectral scans of linesof these sources, including some galaxies lacking standardPACS photometry. Some galaxies were observed many times,and for them the PACS archive has an abundance of spec-trophotometry, while others were observed in only a fewlines.The PACS observers used a variety of observing con-figurations (e.g., chopping throw, integration times, scanlengths, number of repeats), so all the PACS spectropho-tometry had to be reduced individually. We used HIPE 15and pipeline processing 14.2, which were the most recentversions of each separate program at data collection time.The sources here are adequately contained within the central × spaxels of the IFU (a practical limit being a diameter ofabout 15 (cid:48)(cid:48) ), and we used flux density values obtained withthe C129 calibration, taking the sum of the central × spaxels. The task used is extractCentralSpectrum for thechopNod Astronomer’s Observing Templates (AOTs). Weobtained the continuum level as the median of flux densityvalues away from the spectral feature. Because each scantypically has many bad values at the start and end, thosewere also excluded. As a consistency check, we performedboth automatic and manual flux density measurements, andthey were in excellent agreement.The PACS spectrophotometry required creating customsingle-pass filters. These were 0.5 µ m wide, which corre-sponds to the width of the bins generated when doing off-linespectrophotometry. These filters consisted of a delta func-tion throughput at the measured continuum wavelength. Ingeneral, we took the continuum data closest to 60, 90, and150 µ m containing the most individual observations, suchas the continuum near 63 microns or 88 microns. This al-lowed for the most consistent derived photometric valueswhile also remaining near but not on top of the PACS pho-tometry wavelengths. https://github.com/astropy/photutilsMNRAS000
IRAS photometry for all 24galaxies following the procedure described by Weiner et al.(2018, in preparation). We started by subtracting the sky background using the Python package photutils and usedan elliptical aperture to capture all the flux. We took care touse the same aperture area to enclose the full galaxy emis-sion regions in all bands and to correct for any backgroundemission. Our photometric values are consistent with butmore precise than results in the open literature typicallyobtained in pipeline processing of larger datasets. The pho-tometry can be found in Appendix B.For the early-stage mergers from L13, we used the pho-tometry stated in the paper and added SDSS ugriz photom-etry, which was processed the same as by Weiner et al. (2018,in preparation). Photometric uncertainties were calculatedusing the sum in quadrature of Poisson photon noise de-termined by the photometry and calibration uncertainties,and we adopted the same calibration uncertainties as L13,following the references therein. We supplemented the SEDs for the late-stage merger sam-ple with a previously underutilized resource: spectrophoto-metric continuum measurements taken from PACS spectralscans. The PACS spectrophotometric data provide excel-lent coverage of the peak of the FIR continua. Multipleobservers obtained PACS range or spectral scans of linesof these sources, including some galaxies lacking standardPACS photometry. Some galaxies were observed many times,and for them the PACS archive has an abundance of spec-trophotometry, while others were observed in only a fewlines.The PACS observers used a variety of observing con-figurations (e.g., chopping throw, integration times, scanlengths, number of repeats), so all the PACS spectropho-tometry had to be reduced individually. We used HIPE 15and pipeline processing 14.2, which were the most recentversions of each separate program at data collection time.The sources here are adequately contained within the central × spaxels of the IFU (a practical limit being a diameter ofabout 15 (cid:48)(cid:48) ), and we used flux density values obtained withthe C129 calibration, taking the sum of the central × spaxels. The task used is extractCentralSpectrum for thechopNod Astronomer’s Observing Templates (AOTs). Weobtained the continuum level as the median of flux densityvalues away from the spectral feature. Because each scantypically has many bad values at the start and end, thosewere also excluded. As a consistency check, we performedboth automatic and manual flux density measurements, andthey were in excellent agreement.The PACS spectrophotometry required creating customsingle-pass filters. These were 0.5 µ m wide, which corre-sponds to the width of the bins generated when doing off-linespectrophotometry. These filters consisted of a delta func-tion throughput at the measured continuum wavelength. Ingeneral, we took the continuum data closest to 60, 90, and150 µ m containing the most individual observations, suchas the continuum near 63 microns or 88 microns. This al-lowed for the most consistent derived photometric valueswhile also remaining near but not on top of the PACS pho-tometry wavelengths. https://github.com/astropy/photutilsMNRAS000 , 1–16 (2018) J. Dietrich et al.
Table 1.
Basic data for the two galaxy samplesGalaxy Name Redshift ( z) D L (Mpc) a log L I R ( M (cid:12) ) a Stage b AGN c Ref d Late-Stage Merger SampleIRAS 08572+3915 0.05835 265 12.08 4 LINER (1)IRAS 15250+3609 0.05516 249 12.04 5 LINER (1)Mrk 231 0.04217 188 12.51 6 Seyfert 1 (4)Mrk 273 0.03778 168 12.05 5 Seyfert 2 (2)Mrk 463 0.05035 227 11.73 4.5 Seyfert 2 (3)NGC 2623 0.01851 81 11.33 5 LINER (5)NGC 3758 0.02985 132 11.00 4.5 Seyfert 1 (2)NGC 6090 0.02930 129 11.49 4.5 Seyfert (2)UGC 4881 0.03930 175 11.60 4 LINER Composite (1,2)UGC 5101 0.03937 175 12.03 5 Seyfert 1 (3)VV 283 0.03748 167 11.46 5 Seyfert 2 (2)VV 705 0.04019 179 11.82 4.5 Composite (2)Reference SampleM51A 0.00155 8.58 e e f f f f g LINER Composite (1)NGC 4625 0.00212 11.75 e f f a Luminosity distance D L for the galaxies at z > . were calculated using the Hubble law with H = and scaling by (1 + z). Total IR luminosity is calculated from 5–1000 µ m, following Fritz, Franceschini, &Hatziminaoglou (2006). b Weiner et al. (2018, in preparation) c ‘Composite’ indicates strong AGN and SF activity coexist. d References for AGN classification (1) Vardoulaki et al. (2015); (2) Toba et al. (2013); (3) Baumgartneret al. (2013); (4) Ivanov et al. (2000); (5) Garc´ıa-Lorenzo et al. (2015); (6) Hern´andez-Garc´ıa et al. (2016); (7)Brassington et al. (2015); (8) Gonz´alez-Mart´ın et al. (2015); (9) V´eron-Cetty & V´eron (2010) e McQuinn et al. (2017) f Dale et al. (2017) g NGC 3690abc consists of two blended objects (a and b), with a nearby but separate third component c (IC 694).The tabulated photometry for NGC 3690 comes from the ab components only.
There are usually hundreds of datapoints in a typicalPACS spectral scan, but they are only quasi-independent.Although the formal flux density uncertainties are small be-cause of the large number of points, we adopted a value of ±
10% as more fairly accounting for the systematic uncertain-ties, similar to the PACS photometric uncertainties (Pala-dini et al. 2012). See Appendix B for the table of PACS spec-trophotometry for the 10 galaxies in the late-stage mergersample containing PACS spectra.We did not obtain PACS spectrophotometric data forthe early-stage merging galaxies from the L13 sample be-cause of their low modeled f AGN from CIGALE. The galax-ies in the late-stage sample all had large f AGN , and the AGNemission models from Fritz, Franceschini, & Hatziminaoglou(2006) peak at the PACS wavelength range, so we wantedto have the best characterization possible for the emissionin the PACS range for the late-stage merging galaxies withhigh AGN fractions. However, because this adds multiplenew data points in a small wavelength range, the risk of over-fitting the SEDs increases. For the galaxies with largeAGN fractions and a high sensitivity to changes in the PACSbands, only three spectrophotometric values were used withsimilar uncertainties to the PACS photometric data to helpcharacterize the peak without over-fitting. To avoid the issuefor the galaxies with low AGN fractions in the early-stagesample, where the emission peak is already well character-ized, we omitted PACS spectrophotometric data.
Another way to estimate the AGN luminosity fraction is byusing spectral lines that separately trace AGN and SF activ-ity. Specifically, [O iv ] at 25.89 µ m and [Ne v ] at 14.32 µ mare strong lines that signify the presence of an AGN. In con-trast, the [Ne ii ] line at 12.81 µ m is diagnostic of SF activity(Dale et al. 2009; LaMassa et al. 2012). We used spectrataken by Spitzer ’s InfraRed Spectrograph (IRS; 2004 Houcket al.), which provided spectroscopic coverage from 5–36 µ m. MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies For each galaxy, we used the Short-High (SH) and Long-High (LH) modes, which have resolving power λ / ∆ λ = .The SH mode uses a slit of 4.7 x 11.3 arcseconds, while theLH mode uses a slit of 11.1 x 22.3 arcseconds.We reduced IRS spectra using the Spectroscopic Mod-eling Analysis and Reduction Tool (SMART; Higdon et al.(2004)), version 8.2.9. SMART was specifically designed forthe IRS and provides an easy-to-use interface to reduce, an-alyze, and view the spectra. To validate our results fromSMART, we compared our final IRS spectra for a few galax-ies with those reduced using an earlier version of SMARTand placed on the Cornell Atlas of Spitzer/IRS Sources(CASSIS; Lebouteiller et al. 2011). Differences in the twoversions of the final spectra were negligible. We used thedata stored on CASSIS for spectral line analysis for 14 galax-ies in our sample because the spectra reduction was alreadycompleted and reliable. For the remaining 10 galaxies, weused the results from SMART.For these 10 galaxies, we retrieved the IRS spectra fromthe Spitzer
Heritage Archive (SHA). The observations weretaken in both Stare and Map modes. We extracted Stare ob-servations flux densities and spectra without further correc-tions, but for the Map images we used the Cube Builder forIRS Spectral Mapping (CUBISM; Smith et al. 2007) to build3D spectral cubes with 2 spatial and 1 spectral dimension.The 2 spatial dimensions had pointings in a 3x3 grid withthe centre pointing aimed at the nucleus. We confirmed thecentral spectra contained the galaxy nuclei before extract-ing the results to files for importing into SMART. Then, weused SMART to fit and extract the three target spectral linefeatures. To start, we calculated a linear baseline aroundeach spectral line and subtracted it before fitting a Gaus-sian to the line profile. For the cases where the signal wasnot strong or significant contamination caused irreversibleblending, SMART would provide a Gaussian fit with limits,so these translated into 3 σ upper limits on the integratedline flux and the line width.The [Ne ii ] line is relatively isolated with no othernearby, potentially contaminating spectral lines. The [Cl ii ]line at 14.37 µ m caused no noticeable contamination to the[Ne v ] line. However, the [O iv ] 25.89 µ m line partially over-laps with the [Fe ii ] line at 25.99 µ m, producing a slightlyblended line profile. For lines with significant potential con-tamination we used a double Gaussian profile to fit the com-posite (double-line) profiles. The [Fe ii ] contamination didend up forcing the use of an upper limit for some detectionsof the [O iv ] line as the blending caused both single anddouble Gaussian fits to fail.We calculated integrated line fluxes and widths from theGaussian fits. We compared the results for [Ne v ]/[Ne ii ]and [O iv ]/[Ne ii ], similar to the analysis of Genzel et al.(1998). Because [Ne ii ] is a strong tracer of starburst activ-ity, whereas [Ne v ] and [O iv ] are strong indicators of AGNactivity, the AGN-to-starburst tracer ratios help determinethe dominant source of luminosity for these galaxies (Ar-mus et al. 2007, Satyapal et al. 2009 and references therein).Higher ratios should indicate larger AGN activity comparedto starbursts. Ramos Padilla et al. (2018, in preparation)provide a detailed analysis of more spectral line ratios andtheir correlations with IR colours that indicate the presenceof AGN. For the SED fitting, we used the Code for InvestigatingGalaxy Emission (CIGALE; Burgarella, Buat, & Iglesias-Paramo 2005). Specifically, we used ‘pcigale’ version 0.9.0 inPython, which was released in 2016 April. In brief, CIGALEoperates by constructing a multidimensional grid of modelSEDs and identifying the SED model that best fits the datawith χ minimization. The grid dimension is set by the num-ber of user-defined parameters used to define the differentgalaxy components, e.g., intrinsic AGN and stellar emissionspectra, star formation history, dust attenuation, and neb-ular emission. After it has tested all user-specified modelsin its grid, CIGALE then outputs what it identifies as thebest-fitting model spectrum and the parameter set that bestmatches the galaxy data. CIGALE also outputs parameteruncertainties based on the range of models that are consis-tent within each galaxy’s flux density uncertainties.In this work, for simplicity we used a ‘delayed’ starformation history model (delayed with respect to the SFtimescale), assuming a single starburst with an exponentialdecay, following SFR ( t ) ∝ te − t / τ τ , (1)where τ is the e-folding time of the main stellar population,which dominates the stellar emission (Lee et al. 2010). TheSF starts at a time ‘age’ before the present day, where ‘age’is a CIGALE parameter given in the model and defined inTable 2 (Ciesla et al. 2016). We also set the separation be-tween the young and old stellar populations (the stellar sep-aration age) to 10 Myr. This means that at the time that thegalaxy is modeled in CIGALE, every star older than 10 Myris considered ‘old’ while the rest are considered ‘young’. Thecombination between τ (defined in CIGALE as τ main ), ‘age’,and stellar separation is used as a proxy to model the recentstar formation in the period of time defined by ‘stellar sep-aration’. This parametric SFH model allows for CIGALE tobe tuned to the recent SFR and can help determine the stagein some complicated cases. Tests running other SF historyoptions did not significantly alter our conclusions about theAGN fraction.For the dust attenuation, we used models jointly de-scribed by Calzetti et al. (2000) and Leitherer, Calzetti, &Martins (2002) along with the Dale et al. (2014) models forthe dust emission in the far-IR. The Calzetti law for dustextinction and attenuation is described by the following setof local piece-wise power-laws, κ ( λ ) = A ( λ ) E ( B − V ) ∗ = a + b λ + c λ + d λ , (2)where a, b, c, d are constants dependent on the wavelengthrange. The dust emission from Dale et al. (2014) follows amodified blackbody SED with a power-law distribution ofdust mass at each temperature, dM d ∝ U − α dU , (3)where M d is the dust mass heated by a radiation field atintensity U . The power-law index α was allowed to vary from1 to 3. We used the stellar emission models from Bruzual &Charlot (2003) and the standard default nebular emissionmodel included in CIGALE. MNRAS000
Heritage Archive (SHA). The observations weretaken in both Stare and Map modes. We extracted Stare ob-servations flux densities and spectra without further correc-tions, but for the Map images we used the Cube Builder forIRS Spectral Mapping (CUBISM; Smith et al. 2007) to build3D spectral cubes with 2 spatial and 1 spectral dimension.The 2 spatial dimensions had pointings in a 3x3 grid withthe centre pointing aimed at the nucleus. We confirmed thecentral spectra contained the galaxy nuclei before extract-ing the results to files for importing into SMART. Then, weused SMART to fit and extract the three target spectral linefeatures. To start, we calculated a linear baseline aroundeach spectral line and subtracted it before fitting a Gaus-sian to the line profile. For the cases where the signal wasnot strong or significant contamination caused irreversibleblending, SMART would provide a Gaussian fit with limits,so these translated into 3 σ upper limits on the integratedline flux and the line width.The [Ne ii ] line is relatively isolated with no othernearby, potentially contaminating spectral lines. The [Cl ii ]line at 14.37 µ m caused no noticeable contamination to the[Ne v ] line. However, the [O iv ] 25.89 µ m line partially over-laps with the [Fe ii ] line at 25.99 µ m, producing a slightlyblended line profile. For lines with significant potential con-tamination we used a double Gaussian profile to fit the com-posite (double-line) profiles. The [Fe ii ] contamination didend up forcing the use of an upper limit for some detectionsof the [O iv ] line as the blending caused both single anddouble Gaussian fits to fail.We calculated integrated line fluxes and widths from theGaussian fits. We compared the results for [Ne v ]/[Ne ii ]and [O iv ]/[Ne ii ], similar to the analysis of Genzel et al.(1998). Because [Ne ii ] is a strong tracer of starburst activ-ity, whereas [Ne v ] and [O iv ] are strong indicators of AGNactivity, the AGN-to-starburst tracer ratios help determinethe dominant source of luminosity for these galaxies (Ar-mus et al. 2007, Satyapal et al. 2009 and references therein).Higher ratios should indicate larger AGN activity comparedto starbursts. Ramos Padilla et al. (2018, in preparation)provide a detailed analysis of more spectral line ratios andtheir correlations with IR colours that indicate the presenceof AGN. For the SED fitting, we used the Code for InvestigatingGalaxy Emission (CIGALE; Burgarella, Buat, & Iglesias-Paramo 2005). Specifically, we used ‘pcigale’ version 0.9.0 inPython, which was released in 2016 April. In brief, CIGALEoperates by constructing a multidimensional grid of modelSEDs and identifying the SED model that best fits the datawith χ minimization. The grid dimension is set by the num-ber of user-defined parameters used to define the differentgalaxy components, e.g., intrinsic AGN and stellar emissionspectra, star formation history, dust attenuation, and neb-ular emission. After it has tested all user-specified modelsin its grid, CIGALE then outputs what it identifies as thebest-fitting model spectrum and the parameter set that bestmatches the galaxy data. CIGALE also outputs parameteruncertainties based on the range of models that are consis-tent within each galaxy’s flux density uncertainties.In this work, for simplicity we used a ‘delayed’ starformation history model (delayed with respect to the SFtimescale), assuming a single starburst with an exponentialdecay, following SFR ( t ) ∝ te − t / τ τ , (1)where τ is the e-folding time of the main stellar population,which dominates the stellar emission (Lee et al. 2010). TheSF starts at a time ‘age’ before the present day, where ‘age’is a CIGALE parameter given in the model and defined inTable 2 (Ciesla et al. 2016). We also set the separation be-tween the young and old stellar populations (the stellar sep-aration age) to 10 Myr. This means that at the time that thegalaxy is modeled in CIGALE, every star older than 10 Myris considered ‘old’ while the rest are considered ‘young’. Thecombination between τ (defined in CIGALE as τ main ), ‘age’,and stellar separation is used as a proxy to model the recentstar formation in the period of time defined by ‘stellar sep-aration’. This parametric SFH model allows for CIGALE tobe tuned to the recent SFR and can help determine the stagein some complicated cases. Tests running other SF historyoptions did not significantly alter our conclusions about theAGN fraction.For the dust attenuation, we used models jointly de-scribed by Calzetti et al. (2000) and Leitherer, Calzetti, &Martins (2002) along with the Dale et al. (2014) models forthe dust emission in the far-IR. The Calzetti law for dustextinction and attenuation is described by the following setof local piece-wise power-laws, κ ( λ ) = A ( λ ) E ( B − V ) ∗ = a + b λ + c λ + d λ , (2)where a, b, c, d are constants dependent on the wavelengthrange. The dust emission from Dale et al. (2014) follows amodified blackbody SED with a power-law distribution ofdust mass at each temperature, dM d ∝ U − α dU , (3)where M d is the dust mass heated by a radiation field atintensity U . The power-law index α was allowed to vary from1 to 3. We used the stellar emission models from Bruzual &Charlot (2003) and the standard default nebular emissionmodel included in CIGALE. MNRAS000 , 1–16 (2018)
J. Dietrich et al.
For the AGN emission we used the Fritz, Franceschini,& Hatziminaoglou (2006) AGN emission models, which as-sume isotropic emission from a central source and emis-sion from a surrounding toroidal dust structure. The as-sumed central point-like luminous source was defined witha composite power-law in λ L ( λ ) . In particular, from 0.001to 0.03 µ m, λ L ( λ ) ∝ λ . ; from 0.03 to 0.125 µ m, it is in-dependent of wavelength; and from 0.125 µ m to 20 µ m, λ L ( λ ) ∝ λ − . (Granato & Danese 1994, Nenkova, Ivezi´c,& Elitzur 2002). The rest of the IR emission comes fromthe blackbody emission due to AGN heating of the torus.The AGN emission was calculated for an intermediate-typeAGN with an axis angle of 30 . ◦ ◦ corresponds toa Seyfert 1 galaxy viewed pole-on and 90 ◦ corresponds to aSeyfert 2 galaxy viewed edge-on). The default 30 . ◦ . ◦ χ for each of 6.3 million modelSEDs. The model SED with the lowest χ was saved alongwith probability density functions (PDFs) for each param-eter and a text file containing the best models along withthe estimates and uncertainties for each parameter. Theseuncertainties were derived from the 1 σ standard deviationsof the PDFs created by CIGALE for each parameter. Thusthe ‘best-fit’ SED was not always from the ‘most probable’individual parameters as found in the PDFs, but they weregenerally within the uncertainties, in particular for the AGNfraction.For each best-fit SED identified by CIGALE, we alsocalculated the AGN luminosity fraction (denoted ‘fracAGN’by Fritz, Franceschini, & Hatziminaoglou 2006 but hereafterreferred to as f AGN ). f AGN is defined as the AGN contribu-tion to the total IR luminosity from ∼ µ m. We testedvalues of f AGN that ranged from 0 to 0.9 (90% of the totalIR luminosity) in increments of 0.1 on all merging galaxies. f AGN = accounts for the possibility that an AGN mightnot contribute to the IR luminosity. Once the full grid from0 to 0.9 was tested, CIGALE was run again on each galaxywith a finer but narrower grid for f AGN centered on the best-fit value from the previous run. Although it is theoreticallypossible to obtain an AGN fraction close to 1, the probabil-ity is extremely low even for the strongest AGN-dominatedgalaxies in our sample; empirically, we found only one casefor which the AGN fraction was significantly larger than85%. For IRAS 08572+3915, f AGN was allowed to exceed90% in the model runs with ranges from 0.7 to 0.95 by 0.05along with 0.99. CIGALE found a best-fit f AGN value of91%, which was obtained through interpolation of the pa-rameter grid points to find the best-fit solution as measuredby the reduced χ value. Table 3 contains the best-fit f AGN values and reduced χ for each galaxy along with the lineratios described in Section 3.1. Figure 1 shows an examplebest-fit model, Figure 2 shows the corresponding PDF for Figure 1.
An illustration of the data quality and CIGALE SEDmodeling. The SED shown (blue symbols in upper panel) is forIRAS 08572+3915. The best-fit CIGALE model is shown in black.Red dots indicate CIGALE-derived photometry in the modeledpassbands. The best-fit CIGALE model is the sum of contribu-tions from an AGN (green dashed line), dust-attenuated stel-lar emission (orange; the intrinsic stellar emission is indicatedin blue), nebular emission (yellow), and dust emission (red).The bottom panel shows the fractional discrepancies between themodel and photometry. The best-fit CIGALE SEDs for all 24galaxies analyzed in this work are in Figure A1. P r o b a b ili t y d e n s i t y Figure 2.
A measure of the f AG N probability forIRAS 15250+3609. CIGALE found 0 probability for valuesof f AG N less than 0.3 and greater than 0.6 f AGN , and Figure 3 shows f AGN as a function of 100 µ mluminosity for both samples. The Late-Stage Sample haslarger luminosities than most of the Reference Sample byconstruction, but nothing selected for or against AGN frac-tion in either sample. If anything, AGN of a given luminosityshould be easier to detect in low-luminosity galaxies, i.e., inthe Reference Sample.In some cases, CIGALE produced a best-fit model with f AGN = 0 having no estimated uncertainty, even for galax-
MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Table 2.
CIGALE Parameter Settings Used in This WorkParameter Definition Values Tested (range)Star Formation History—Delayed Module τ main the e-folding time of the main population (Myr) 50, 500, 1000, 2500, 5000, 7500age the age of the oldest stars (Gyr) 0.5, 1, 2, 3, 4, 5, 6sfr A multiplicative factor controlling SFR amplitude 1.0separation age separation between young and old stellar populations (Myr) 10Bruzual & Charlot (2003) Stellar Emission Moduleimf initial mass function (0 for Salpeter, 1 for Chabrier) 0metallicity initial metallicity for the stars 0.02separation age age of separation between ‘young’ and ‘old’ stellar populations in Myr 10Nebular Emission ModulelogU ionization parameter − . f esc escape fraction of Lyman continuum photons 0.0f dust absorption fraction of Lyman continuum photons 0.0lines width line width in km/s 300emission whether or not to include nebular emission TrueCalzetti et al. (2000) and Leitherer, Calzetti, & Martins (2002) Dust Attenuation ModuleE BVs young E ( B − V ) ∗ , the colour excess of the young stellar continuum light 0.1, 0.25, 0.4, 0.55, 0.7E BVs old factor reduction factor for the E ( B − V ) ∗ of old vs. young population 0.22, 0.44, 0.66, 0.88uv bump amplitude amplitude of the 220 nm bump 0.0powerlaw slope slope delta of the power law attentuation curve 0.0filters filters in which attenuation will be calculated FUVDale et al. (2014) Dust Module α slope of the dust temperature distribution in 3 1, 1.5, 2, 2.5, 3Fritz, Franceschini, & Hatziminaoglou (2006) AGN Moduler ratio the ratio between outer and inner radius of AGN torus 10, 30, 60, 100, 150 τ the optical depth at 9.7 µ m 0.6, 1, 6, 10 β the density radial exponent − , − . γ the density exponential factor 0, 2opening angle the opening angle of the torus 60, 100, 140 ψ the angle between equator and line of sight 30.1(0 is Type 2 and 89.9 is Type 1) f AG N the AGN fraction to the IR luminosity 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8,0.9 (0.95, 0.99) ies known to host an AGN from previous studies. In generalwhen f AGN < . , the uncertainties tend to be large frac-tions of the best-fit value. As a result, the AGN model isrelatively poorly constrained in low- f AGN cases. This is be-cause with just photometric data, a weak AGN cannot bedistinguished from a slightly increased SFR. CIGALE usesa slightly different, more luminous dust model in the IR toaccount for the small influx to the SED that the AGN con-tributes (Ciesla et al. 2015). However, in most of these cases,the CIGALE best-fit model underestimates the far-IR
Her-schel /SPIRE bands, up to a factor of 1.5 or 2. CIGALEmodels including non-thermal radio emission were consid-ered in an attempt to better fit the
Herschel /SPIRE datapoints, but the radio emission was ultimately discarded asnegligible because the added emission in the far-IR was 3orders of magnitude too small to make up the difference.For the spectral line analysis, with the results shown in Figure 4, we ran linear regression tests on the combina-tion of both samples for both [Ne v ]/[Ne ii ] vs. f AGN and[O iv ]/[Ne ii ] vs. f AGN , and the results were not significantat the 3 σ level. However, when running the linear regressiontest on the late-stage merger sample only, there was evidencefor a linear trend at the 2 σ level. Further analysis of spectralline ratios, including correlations with flux density ratios, isdiscussed by Ramos Padilla et al. (2018, in preparation). Numerous studies (e.g., Stern et al., 2005, Stern et al. 2012,Donley et al. 2008, Rosario et al. 2012) have demonstratedthat galaxy colours can reveal AGN. Flux ratios such as f µ m / f µ m and f µ m / f µ m as well as K–L and
L–M colours have been used to help determine the presenceof an AGN. We performed a linear regression test of f AGN
MNRAS000
MNRAS000 , 1–16 (2018)
J. Dietrich et al.
Table 3.
Derived fractional AGN contribution to the total IR luminosity, SFR, and reduced χ from the CIGALE models, and themeasured [Ne v ]/[Ne ii ] & [O iv ]/[Ne ii ] ratios from Spitzer
IRS spectra.Galaxy Name f AG N
SFR ( M (cid:12) yr − ) Reduced χ [Ne v ]/[Ne ii ] [O iv ]/[Ne ii ]Late-Stage Merger SampleIRAS 08572+3915 . ± .
05 21 . ± . < . < . IRAS 15250+3609 . ± .
04 89 . ± . < . < . Mrk 231 . ± .
02 444 ± < . < . Mrk 273 . ± .
04 57 . ± . . ± .
014 1 . ± . Mrk 463 . ± .
03 30 . ± . . ± .
136 6 . ± . NGC 2623 . ± .
05 20 . ± . . ± .
007 0 . ± . NGC 3758 . ± .
03 7 . ± . . ± .
05 35 . ± . < .
009 0 . ± . UGC 4881 . ± .
03 23 . ± . < .
005 0 . ± . UGC 5101 . ± .
04 20 . ± . ± . ± . VV 283 . ± .
04 20 . ± . ± . ± . VV 705 . ± .
08 77 . ± . < . < . Reference SampleM51A . ± .
03 2 . ± . . ± .
002 0 . ± . M51B < . < . < . < . NGC 2976 . ± .
06 0 . ± . < . < . NGC 3031 < .
01 0 . ± . < .
024 0 . ± . NGC 3077 . ± .
06 0 . ± . < . < . NGC 3190 < .
01 0 . ± . < .
081 0 . ± . NGC 3690 . ± .
05 59 . ± . < . < . NGC 4625 < .
10 0 . ± . < . < . NGC 5394 . ± .
03 3 . ± . . ± .
002 0 . ± . NGC 5395 < .
10 6 . ± . < . < . M101 < .
14 4 . ± . . ± . ...NGC 5474 < .
05 0 . ± . < . < . a Upper limits are defined at 3 σ b ‘...’ indicates the galaxy was not observed in the required wavelengths by IRS versus every flux ratio in our photometric data to determinewhether any ratio showed a significant correlation. The sig-nificant results, with Pearson correlation ratios of magnitudegreater than 0.8 and with p -values of 0.027 or less (corre-sponding to a significance level of 3 σ ), are summarized inTable C1. L − M and similar colours such as IRAC [ . ] − [ . ] and WISE W − W are the basis for the Stern et al. (2005)and Donley et al. (2008) plots showing a significant differ-ence for the AGN-dominated systems. The [ . ] − [ . ] µ mcolour is significant here at ∼ σ , as seen in Figure 5. How-ever, the corresponding WISE colours and mixing of IRACand
WISE colours are not significant due to the low samplesize of galaxies with
WISE photometry. Consistent with theSED analysis described in Sec. 3.1, the early-stage mergersconsistently populate the starburst region of the Stern et al.(2005) IRAC colour–colour plot. By contrast, even thoughour analysis shows that all the late-stage merger galaxiescontain luminous AGN and moreover that many of themare AGN-dominated, only about half of them have IRACcolours indicating that these objects host luminous AGN.The apparent discrepancy is not surprising because the SEDanalysis is based on much more information than the simplecolour–colour plot; in particular, it attempts to estimate andaccount for dust obscuration. The Stern et al. (2005) plotcould miss AGN when stars overwhelm the AGN at IRACwavelengths or when the AGN is heavily obscured by dust in the near-mid IR. However, the four largest AGN frac-tions modeled by CIGALE correspond to the four galaxieswith the reddest [3.6]–[4.5] colours are Figure 6 contains theStern colour–colour plot for the 24 galaxies in our sample.The galaxies from the Reference Sample nearly all fall in thisregion. By contrast, the late-stage merger subsample popu-lates both the AGN wedge and the star-formation region ofthe plot, albeit only the extreme red end of the latter.Sanders et al. (1988) showed that the [ ] − [ ] colouris an AGN tracer, but this colour is not correlated with ourmeasurements of f AGN as given by CIGALE. Two of thegalaxies modeled with strong AGN are among the bluest in [ ] − [ ] colour, and the reddest [ ] − [ ] measurementbelongs to a galaxy modeled with an AGN fraction of ∼ f AGN . In particular, the fluxratios of
GALEX
FUV and NUV, Sloan u, g, r, i, z , and2MASS J , H , and K s with IRAC 4.5 and 5.8 µ m and MIPS24 µ m and 70 µ m bands are significantly correlated withAGN fraction. The extremely red colors at the high end ofthe correlation show that a steep increase in the SED inthe near and mid-IR is indicative of an AGN. Photometricdata at similar wavelengths show similar correlations; UV–70 µ m flux ratios are significant for both MIPS and PACS MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Figure 3.
Estimated AGN luminosity fractions versus IR lu-minosity for both the Late-Stage Merger galaxies (red symbols)and the Reference Sample (blue symbols). The luminosity frac-tions were measured as a function of IR luminosity between 5and 1000 µ m. Symbol size indicates the percentage AGN lumi-nosity fraction: the smallest symbols are for f AG N < . , thenext largest indicate . ≤ f AG N < . , and so on. Trianglesindicate 3- σ upper limits. µ m photometry, for example. Also, negative correlationsare found between the AGN fraction and the MIPS 70 µ mand PACS/SPIRE colours, showing that the presence of anAGN makes the far-IR SED steeper than the expected cooldust power-law. We tested the reliability of CIGALE by analyzing the SEDsof simulated merging galaxies and comparing the CIGALE-derived galaxy parameters with the known galaxy parame-ters from the simulations. Performing such ‘numerical exper-iments’ using simulations is a very useful means to validatemethods of observational inference, as the ground truth isknown a priori and various uncertainties can be controlled.For previous examples and discussions of this validationprocess, see Micha(cid:32)lowski et al. (2014), Hayward & Smith(2015), Smith & Hayward (2015). As Lanz et al. (2014)have described, the simulations provide realistic SEDs (seealso 2018, in preparation Weiner et al.). The aim was to de-termine how well CIGALE recovers f AGN (i.e., the AGN’scontribution to the bolometric luminosity) of the simulated
Figure 4.
Integrated emission-line flux ratios as a function ofAGN luminosity fraction.
Upper panel: [Ne v ]/[Ne ii ]. Symbol sizeindicates the percentage AGN luminosity fraction: the smallestsymbols are for f AG N < . , the next largest indicate . ≤ f AG N < . , and so on. Triangles indicate 3- σ upper limits. Lower panel: [O iv ]/[Ne ii ]. galaxies. The simulated merger SEDs were created usinga two-step process. First, mergers were simulated using ahydrodynamical code (Springel 2005; Hayward et al. 2011;Lanz et al. 2014), and then a radiative transfer code was usedto generate the emergent light from the simulated mergersand simulate an observation (Jonsson 2006; Jonsson, Groves,& Cox 2010). The hydrodynamic simulations and radiativetransfer code used are described in detail by Lanz et al.(2014) and Weiner et al. (2018, in preparation). Here wesummarize the key aspects of our analysis. The merger simulations used the TreeSPH (Hernquist &Katz 1989) code GADGET-3 (Springel 2005), which em-ploys a hierarchical tree N-body method to compute gravi-tational interactions in an N-body cosmological simulationthat includes gravity, gas dynamics (via smoothed-particlehydrodynamics), stellar evolution, and other physical mech-anisms. GADGET-3 implements the thermodynamic trans-port of energy through gas dynamics and radiative heatingand cooling and conserves both energy and entropy. The ISMis modeled with two phases of matter in which cold, denseclouds interact with a hot, diffuse gas medium (Springel &Hernquist 2003). The hydrodynamical code models star for-
MNRAS000
MNRAS000 , 1–16 (2018) J. Dietrich et al.
Figure 5.
IRAC [ . ] − [ . ] color versus AGN luminosity frac-tion. Symbol size indicates the percentage AGN luminosity frac-tion: the smallest symbols are for f AG N < . , the next largestindicate . ≤ f AG N < . , and so on. mation according to the Kennicutt-Schmidt (‘K-S’) relation(Kennicutt 1998), an empirical relation between SFR andthe volume density of gas. When the density of gas parti-cles in the simulation surpasses a minimum threshold, gasparticles are converted into star particles according to theK-S relation. GADGET-3 uses sub-resolution models to de-scribe starforming regions because its grid is too coarse toresolve individual cold gas clouds; this limitation directlyaffects how radiative transfer is modeled.The AGN contributions to the SEDs were computedfrom the black hole accretion rate, and the correspondingAGN feedback was included using the sub-resolution modelof Springel et al. (2005). The AGN were represented inthe simulations by black hole particles which grow and ra-diate by accreting surrounding gas (Springel et al. 2005).Black hole particles accreted according to the Bondi-Hoyle-Lyttleton model, at the rate (cid:219) M BH = πα G M BH ρ ( c s + v ) / , (4)where ρ is the gas density, c s is the speed of sound in thegas, v is the black hole speed relative to the gas, and α isa system-dependent, dimensionless parameter, usually esti-mated as between 1 and 2 (Bondi 1952); we took α = . .Because the accretion occurs on spatial scales far belowGADGET-3’s resolution, the code uses a sub-resolutionmodel to interface black hole particles to surrounding gas Figure 6.
IRAC colour-colour plot following Stern et al. (2005)for all galaxies in both subsamples analyzed in this work. Sym-bol size indicates the percentage AGN luminosity fraction: thesmallest symbols are for f AG N < . , the next largest indicate . ≤ f AG N < . , and so on. The dotted line encloses the re-gion in which low-redshift galaxies reside when their IRAC coloursare dominated by luminous AGN (the AGN wedge). The straightline is an empirical boundary below which nearby star-formation-dominated galaxies typically lie. particles. GADGET-3 models accretion of gas particles asa stochastic process. Each particle near a black hole is as-signed a probability of accretion weighted by the estimatedgas density near the black hole, the location of the particlerelative to the BH, the Bondi accretion rate, and the timestep. GADGET-3 also imposes an upper limit on (cid:219) M BH atthe Eddington rate, at which the radiation pressure from anAGN overcomes the gravitational attraction of the gas.As an AGN accretes gas, its accretion disk heats up andradiates energy into the host galaxy. GADGET-3 treats thethermal energy delivered to the black hole as thermal energyradiated into the surroundings with power L r = (cid:15) r (cid:219) M BH c , (5)where (cid:15) r is the radiative efficiency, which is set to 10% inthese simulations, the consensus value for efficient black holeaccretion. As can be seen in Equation 5, the AGN luminosityis directly proportional to the accretion rate, so when theAGN is turned off, as described below, (cid:15) r = . In this way,the accreting AGN directly influence surrounding regions ofstar formation.We used the 3D polychromatic Monte Carlo dust radia- MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Figure 7.
The SUNRISE SED output for the M3–M3 majormerger. (Top) The full SED just prior to coalescence of thetwo supermassive black holes. The seven color-coded viewing an-gles in this case give nearly identical SEDs. (Bottom) The SFR(black), AGN accretion rate (blue) and black hole separation ver-sus merger time in Gyr with respect to coalescence. The diamondand square markers indicate the moment corresponding to theSED in the upper panel. tive transfer code SUNRISE (Jonsson 2006; Jonsson, Groves,& Cox 2010) to calculate spatially resolved UV–mm SEDsfor the simulated galaxies. SUNRISE performs a radiativetransfer calculation for the attenuation and re-emission fromthe dust heated by star formation and AGN activity, as wellas the stellar components, to generate ‘observed’ SEDs forthe merger. Merger steps for SED calculation were at reg-ular intervals at 10 Myr near coalescence and at 100 Myrotherwise (Lanz et al. 2014). SEDs were computed for sevendifferent viewing angles at each step to account for the im-pact of dust attenuation along different lines of sight.Five galaxy models called M4, M3, M2, M1, and M0with stellar masses respectively of 5, 4.22, 1.18, 0.38, and0.061 × M (cid:12) were used (see Table 2 of Lanz et al. 2014;Rosenthal et al. 2015; Hayward et al. 2011). One furthermodel named c6e was a massive gas-rich galaxy with a halomass of × M (cid:12) and a gas fraction of 60%, meant tomimic some submillimeter galaxy (SMG) properties. Fig-ure 7 shows the simulated SED for the M3–M3 merger case.We created output files at the specified intervals during themergers of all combinations of the six galaxy models andthen ran SUNRISE to compute the SED for each step andseven viewing angles of every merger. We did not see any evidence that CIGALE’s output reliabil-ity depended on the particulars of the merger scenario, andthe M3–M3 or M4–M4 major merger simulations represent -1 B l a c k H o l e s e p a r a t i o n ( k p c ) -4 -3 -2 -1 M fl y r − A G N c o n t r i b u t i o n t o I R l u m i n o s i t y Properties of merger simulation c6e D BH SFRBH acc
AGN fraction
Figure 8.
The SUNRISE SED output of the c6e–c6e merger. Inthis gas-rich example, the black hole separation is shown as theblue line, the star formation rate is indicated in red, and the blackhole accretion rate is in green. The AGN fractional contributionto the lumninosity is shown with the dotted black line; the L AG N was calculated from the accretion rate, and L SF R was calculatedfrom the model’s SFR. the results. As Lanz et al. (2014) have already described,those simulations give realistic SEDs (Weiner et al. 2018,in preparation). Figure 7 (lower) illustrates the black holeaccretion varying during the M3-M3 merger, peaking (forthis example) at about 0.7 M (cid:12) yr − shortly after the mo-ment of coalescence. The accretion rate hovers at a few times − M (cid:12) yr − for most of the early stages of the interaction,even during the first close pass of the two black hole nuclei,but starts to climb to its peak about 12 Myr before coales-cence, when the separation is about 150 kpc. The accretionactivity remains above the earlier baseline level for about30 Myr, during which time the increasingly dense gas in thesimulation also produces a burst of star formation, and af-terwards the AGN accretion drops to a new baseline nearly20 times smaller than the pre-merger level.The gas-rich merger simulation ‘c6e–c6e’ has an initialgas fraction of 60%, in contrast to the other simulated galax-ies which, independent of mass, used gas fractions of only15 and 38% (Cox et al. 2008; Hayward et al. 2013; Lanz etal. 2014). The c6e–c6e simulation uses a baryonic mass of × M (cid:12) , considerably more than the other simulations,but the same black hole mass of . × M (cid:12) . For this merger,Figure 8 plots the AGN fraction along with some other pa-rameters versus time. The AGN luminosity in this examplepeaks briefly as high as 55% at coalescence, and estimatesof the SFR based solely on L FI R will be correspondingly toohigh.The SED simulations for c6e–c6e made with the AGN‘turned off’, that is with (cid:15) r = , are illustrated in Figure 9.At the largest viewing angle, (edge-on), the strongest differ-ence between the two cases is a factor of two in the 5–8 µ mrange. The fact that this part of the spectral range is mostsensitive to the AGN fraction confirms what is already well-known from earlier Spitzer observations: the IRAC colour-colour diagram as manifest in the so-called Stern wedge isa useful tool to identify AGN (Stern et al. 2005). However,
MNRAS000
MNRAS000 , 1–16 (2018) J. Dietrich et al.
Figure 9.
AGN on vs. AGN off: the SUNRISE output for thec6e–c6e major merger at 0.7 Gyr. Colors correspond to differentviewing angles. (Top) The SED for the AGN turned on (solidcurves) and AGN turned off (dashed curves). (Bottom) The fluxratios for AGN on / AGN off, showing the spectral differences. the reverse is not necessarily true, as Figure 6 shows. Low-luminosity or highly obscured AGN may have blue [3.6]–[4.5]colours or red [5.8]–[8.0] colours, and by using the full SEDanalysis we can obtain a more reliable accounting of AGNemission and demographics than IRAC colours alone.
We compared CIGALE’s model results against simulationsboth with the AGN turned on (a normal (cid:15) r = ) andwith the AGN turned off ( (cid:15) r = ). The parameters of par-ticular relevance here are: the ratio of the AGN’s dust torusradii, the optical depth at 9.7 µ m, the AGN opening angle,the AGN luminosity, and the AGN fraction as estimatedboth with the Dale et al. (2014) and the Fritz, Frances-chini, & Hatziminaoglou (2006) methods. Figure 10 plots theCIGALE-modeled outputs for the SFR and AGN fractionversus the simulated output values as a function of elapsedtime with the AGN on.CIGALE does an excellent job of evaluating f AGN whenthe AGN is the dominant fraction of luminosity, but it doesa poorer job of estimating f AGN when it is below about 15%.When CIGALE errs most often it overestimates f AGN , butsometimes when f AGN < f AGN estimate is particularly egregiousin the 0.2–1.2 Gyr period, estimating a fraction of as muchas 35% when it is in fact less than 10%. This occurs becauseslight changes to the relative flux densities in the 5–30 µ mbands happen in merger phases when the SFR is low and thelonger-wavelength emission is therefore also weak. As a re-sult, these slight changes have an undue impact on CIGALEby causing large shifts in the AGN fraction upward and cor-respondingly the SFR downward.In low flux density cases, the full SED spectral outputinformation is used to supplement the calculation of AGNluminosity fraction from the IR bands, which partially cor- Figure 10.
CIGALE-derived parameters for the M3–M3 merger;the top panel shows SFR, the middle panel shows f AG N , andthe bottom panel shows τ main and age. The blue curves show themodel values output from the GADGET-3 simulations, and thered curves show values derived by CIGALE. The yellow line inthe bottom panel is the derived age, and the green is the derived e -folding time for the delayed star formation history. rects for the f AGN estimated by CIGALE. Similarly, whenCIGALE underestimates the observed flux density at wave-lengths below 10 µ m but overestimates it longward of 10and 100 µ m (as we can see in the SED fit to UGC 5101, forexample, in Figure A1), additional correction can be appliedby shifting upward the allowed range for the f AGN parame-ter accordingly and/or using a cooler dust model. Not least,spectroscopic information (as per Section 2.3) can also beused to calibrate a CIGALE output when the AGN fractionis low. In all of the AGN-dominated systems considered here,the AGN fraction is high enough that the CIGALE resultsare credible.
As shown in Figure 3, the luminosity is strongly correlatedwith merger stage, which is expected. The AGN luminosityfraction is also correlated with both, as all of the stage 3.5 orless merging galaxies except for NGC 5394 have AGN lumi-nosity fractions below the values of all of the stage 4 or highermerging galaxies. The lowest f AGN values of 0% were foundin early-stage merging galaxies, and the late-stage merginggalaxy with the lowest AGN fraction was Mrk 231, whichis classified as a stage 6 at post-coalescence. This galaxyhas already completed the merging process, and the AGNluminosity is dropping while the SFR is still relatively high.Nearly half of the late-stage merging galaxies in oursample of 11 host an AGN that is at least as luminous asthe rest of the galaxy in the IR. These behemoths can evenexceed the IR luminosities of their hosts by an order of mag-nitude, as in IRAS 08572+3915, which at an estimated 91%
MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies surpasses the original limit we had imposed on the AGNfraction at 90%. This is consistent with Efstathiou et al.(2014), who have also reported an AGN fraction of ∼ v ]/[Ne ii ] or [O iv ]/[Ne ii ] ratio and theAGN fraction in the combined sample. This could be due tolimitations in the measurements, as there were many upperlimits for the line ratios in the early-stage merging galaxies inthe Reference Sample at low AGN fraction. Some line ratiosin our sample are similar to the line ratios for galaxies shownby Dale et al. (2009), but their sample is low-luminosity andunlikely to contain strong AGN. The AGN fraction in our Late-Stage Sample is systemat-ically and significantly higher than that measured in ourReference Sample. For the Late-Stage Sample, f AGN rangesfrom 17 to 91%. The 91% estimate is for IRAS 08572+3915, alate-stage ULIRG. The 17% estimate is for Mrk 231, which isclassified as a stage 6 post-coalescence merger having a highSFR. In contrast, in the Reference Sample, f AGN is below15% for all but three galaxies. The difference is probably be-cause the Late-Stage Sample galaxies are advanced in theirinteraction level, with material flowing to their centers andfeeding the AGN that reside there, with the exception ofMRK 231 which is consistent with being in the post-mergerstage.CIGALE SED modeling of late-stage snapshots of a setof SPH merger simulations yields AGN luminosity fractionsthat are in good agreement in general with the simulationvalues and also are consistent with values measured in theLate-Stage Merger Sample. However, CIGALE incorrectlyinferred AGN fractions up to 30% in earlier stages of the sim-ulated mergers when the true value was close to 0. Galaxiesin the Reference Sample with little to no empirical evidencein the SED for AGN activity were not modeled with large f AGN , so the SED fits for these galaxies are reliable.We also measured spectral line ratios for [Ne v ]/[Ne ii ] and [O iv ]/[Ne ii ] to provide another tool to estimatethe strength of the AGN. We found no overall correlationabove the 2 σ level in our samples; some strong AGN havecomparatively weak line ratios, similar to those of the weakerAGN. The effects of extinction in these mid-IR lines likelyplays a significant role. We do, however, find that the late- stage merging galaxies alone do show a possible linear trendbetween AGN fraction and line ratios. ACKNOWLEDGEMENTS
We thank the National Science Foundation, the SmithsonianAstrophysical Observatory, and Jonathan McDowell for pro-viding JD the ability to complete this research through theNSF Research Experience for Undergraduates Program heldat the SAO. We would also like to thank Aliza Beveragefor her assistance and feedback with the research and writ-ing process, and D. Burgarella and the CIGALE team fortheir advice. The SAO REU program is funded in part bythe National Science Foundation REU and Department ofDefense ASSURE programs under NSF Grant no. 1262851,and by the Smithsonian Institution. The Flatiron Instituteis supported by the Simons Foundation. HAS, ASW, andJRM-G acknowledge partial support from NASA GrantsNNX14AJ61G and NNX15AE56G. This research has madeuse of the SIMBAD database, operated at CDS, Strasbourg,France. This research has made use of the NASA/IPAC Ex-tragalactic Database (NED), operated by the Jet Propul-sion Laboratory, California Institute of Technology, undercontract with the National Aeronautics and Space Admin-istration.
REFERENCES
Alam, S.; Albareti, F.D.; Allende Prieto, C.; et al. 2015, ApJ, 219,12Armus, L.; Charmandaris, V.; Bernard-Salas, J.; et al. 2007, ApJ,656, 148Baumgartner, W.H.; Tueller, J.; Markwardt, C.B.; et al. 2013,ApJS, 207, 19Bondi, H. 1952, MNRAS, 112, 195Brassington, N.; Zezas, A.; Ashby, M.L.N.; et al. 2015, ApJS, 218,6Brown, M.J.I.; Jarrett, T.H.; & Cluver, M.E. 2014, PASA, 31, 49Bruzual, G. & Charlot, S. 2003, MNRAS, 344, 1000Burgarella, D.; Buat, V.; & Iglesias-Paramo, J. 2005, MNRAS,360, 1413Calzetti, D.; Armus, L.; Bohlin, R.C.; et al. 2000, ApJ, 533, 682Ciesla, L.; Charmandaris, V.; Georgakakis, A.; et al. 2015, A&A,576, 10Ciesla, L.; Boselli, A.; Elbaz, D.; et al. 2016, A&A, 585, 43Corwin, H.G, Buta, R.J. & de Vaucouleurs, G. 1994, AJ, 108, 212Cowley, W.I.; B´ethermin, M.; Lagos, C.d.P; et al. 2017, MNRAS,467, 1231Cox, T.J.; Jonsson, P.; Somerville, R.S.; et al. 2008, MNRAS, 384,386Dale, D.A.; Smith, J.D.T.; Schlawin, E.A.; et al. 2009, ApJ, 693,1821Dale, D.A.; Helou, G.; Magdis, G.E.; et al. 2014, ApJ, 784, 83Dale, D.A.; Cook, D.O.; Roussel, H.; et al. 2017, ApJ, 837, 90Donley, J.L.; Rieke, G.H.; P´erez-Gonz´alez, P.G.; et al. 2008, ApJ,687, 111Drouart, G.; Rocca-Volmerange, B.; De Brueck, C.; et al. 2016,A&A, 593, 109Efstathiou, A.; Pearson, C.; Farrah, D.; et al. 2014, MNRAS, 437,16Fernandez-Ontiveros, J.A.; Spinoglio, L.; Pereira-Santaella, M.;et al. 2016, ApJS, 226, 19Franceschini, A.; Braito, V.; Persic, M.; et al. 2003, MNRAS, 343,1181MNRAS000
Alam, S.; Albareti, F.D.; Allende Prieto, C.; et al. 2015, ApJ, 219,12Armus, L.; Charmandaris, V.; Bernard-Salas, J.; et al. 2007, ApJ,656, 148Baumgartner, W.H.; Tueller, J.; Markwardt, C.B.; et al. 2013,ApJS, 207, 19Bondi, H. 1952, MNRAS, 112, 195Brassington, N.; Zezas, A.; Ashby, M.L.N.; et al. 2015, ApJS, 218,6Brown, M.J.I.; Jarrett, T.H.; & Cluver, M.E. 2014, PASA, 31, 49Bruzual, G. & Charlot, S. 2003, MNRAS, 344, 1000Burgarella, D.; Buat, V.; & Iglesias-Paramo, J. 2005, MNRAS,360, 1413Calzetti, D.; Armus, L.; Bohlin, R.C.; et al. 2000, ApJ, 533, 682Ciesla, L.; Charmandaris, V.; Georgakakis, A.; et al. 2015, A&A,576, 10Ciesla, L.; Boselli, A.; Elbaz, D.; et al. 2016, A&A, 585, 43Corwin, H.G, Buta, R.J. & de Vaucouleurs, G. 1994, AJ, 108, 212Cowley, W.I.; B´ethermin, M.; Lagos, C.d.P; et al. 2017, MNRAS,467, 1231Cox, T.J.; Jonsson, P.; Somerville, R.S.; et al. 2008, MNRAS, 384,386Dale, D.A.; Smith, J.D.T.; Schlawin, E.A.; et al. 2009, ApJ, 693,1821Dale, D.A.; Helou, G.; Magdis, G.E.; et al. 2014, ApJ, 784, 83Dale, D.A.; Cook, D.O.; Roussel, H.; et al. 2017, ApJ, 837, 90Donley, J.L.; Rieke, G.H.; P´erez-Gonz´alez, P.G.; et al. 2008, ApJ,687, 111Drouart, G.; Rocca-Volmerange, B.; De Brueck, C.; et al. 2016,A&A, 593, 109Efstathiou, A.; Pearson, C.; Farrah, D.; et al. 2014, MNRAS, 437,16Fernandez-Ontiveros, J.A.; Spinoglio, L.; Pereira-Santaella, M.;et al. 2016, ApJS, 226, 19Franceschini, A.; Braito, V.; Persic, M.; et al. 2003, MNRAS, 343,1181MNRAS000 , 1–16 (2018) J. Dietrich et al.
Fritz, J.; Franceschini, A.; & Hatziminaoglou, E. 2006, MNRAS,366, 767Garc´ıa-Lorenzo, B.; M´arquez, I.; Buerra-Ballesteros, J.K.; et al.2015, A&A, 573, 59Gehrels, N.; Chincarini, G.; Giommi, P.; et al. 2004, ApJ, 611,1005Genzel, R.; Lutz, D.; Sturm, E.; et al. 1998, ApJ, 498, 579Gingold, R.A. & Monaghan, J.J. 1977, MNRAS, 181, 375Gonz´alez-Mart´ın, O.; Masegosa, J.; M´arquez, I.; et al. 2015, A&A,578, 74Granato, G.L. & Danese, L. 1994, MNRAS, 268, 253Groves, B.; Dopita, M.A.; Sutherland, R.S.; et al. 2008, ApJS,176, 438Gruppioni, C.; Pozzi, F.; Polleta, M.; et al. 2008, ApJ, 701, 850Hayward, C.C.; Duˆsan, K.; Jonnson, P.; et al. 2011, ApJ, 743,159Hayward, C.C.; Narayanan, D.; Duˆsan, K.; et al. 2013, MNRAS,428, 2529Hayward, C.C.; Lanz, L.; Ashby, M.L.N.; et al. 2014, MNRAS,445, 1598Hayward, C.C.; Torrey, P.; Springel, V.; et al. 2014, MNRAS, 442,1992Hayward, C.C.; & Smith, D.J.B. 2015, MNRAS, 446, 1512Hern´andez-Garc´ıa, L.; Masegosa, J.; Gonz´alez-Mart´ın, O.; et al.2016, ApJ, 824, 7Hernquist, L & Katz, N. 1989, ApJS, 70, 419Higdon, S.J.U.; Devost, D.; Higdon, J.L.; et al. 2004, PASP, 116,975Hong, J.; Im, M.; Kim, M.; et al. 2015, ApJ, 804, 34Hopkins, P.F.; Richards, G.T.; & Hernquist, L. 2007, ApJ, 654,731Houck, J.R.; Roellig, T.L.; van Cleve, J.; et al. 2004, ApJS, 154,18Ivanov, V.D.; Rieke, G.H.; Groppi, C.E.; et al. 2000, ApJ, 545,190Jonsson, P. 2006, 372, 2Jonsson, P.; Groves, B.; and Cox, T.J. 2010, MNRAS, 403, 17Katz, N.; Weinberg, D.H.; & Hernquist, L. 1996, ApJS, 105, 19Keel, W.C.; Kennicutt, R.C.; Jr.; Hummel, E.; et al. 1985, AJ,90, 708Kennicutt, R.C., Jr. 1998, ApJ, 498, 541Knapen, J.H.; Erroz-Ferrer, S.; Roa, J.; et al. 2014, A&A, 569, 91Lackner, C.N.; Silverman, J.D.; Salvato, M.; et al. 2014, AJ, 148,137LaMassa, S.M.; Heckman, T.M.; Ptak, A.; et al. 2012, ApJ, 758,1Lanz, L.; Zezas, A.; Brassington, N.; et al. 2013, ApJ, 768, 90(L13)Lanz, L.; Hayward, C.C.; Zezas, A.; et al. 2014, ApJ, 785, 39Lebouteiller, V.; Barry, D.J.; Spoon, H.W.W.; et al. 2011, ApJS,196, 8Lee, S.-K.; Ferguson, H.C.; Somerville, R.S.; et al. 2010, ApJ, 725,1644Leitherer, C.; Schaerer, D.; Goldader, J.D.; et al. 1999, ApJS, 123,3Leitherer, C.; Calzetti, D.; & Martins, L.P. 2002, ApJ, 574, 114Lucy, L.B. 1977, AJ, 82, 1013Martin, C.D.; Fanson, J.; Schiminovich, D.; et al. 2005, ApJ, 619,L1McQuinn, K.B.W.; Skillman, E.D.; Dolphin, A.E.; et al. 2017,AJ, 154, 51Micha(cid:32)lowski, M.J.; Hayward, C.C.; Dunlop, J.S.; et al. 2014,A&A, 571, 75Moshir, M.; Kopman, G.; & Conrow, T.A.O. 1992, Pasadena:IPAC, CaltechNenkova, M.; Ivezi´c, ˇZ.; & Elitzur, M. 2002, ApJ, 570, 9Neugebauer, G.; Habing, H.J.; van Duinen, R.; et al. 1984, ApJ,278, 1 Ott, S. 2010, ASP Conf. Ser., 434, 139Paladini, R.; Linz, H.; Altieri, B.; & Ali, B. 2012, AssessmentAnalysis of the Extended Emission Calibration for the PACSRed Channel, Document: PICC-NHSC-TR-034 (Pasadena,CA: NHSC)Pilbratt, G.L.; Riedinger, J.R.; Passvogel, T.; et al. 2010, A&A,518, 1Ramos Padilla, A.F.,; et al. 2018Rosario, D.J.; Santini, P.; Lutz, D.; et al. 2012, A&A, 545, 45Rosenberg, M.J.F.; Meijerink, R.; Israel, F.P.; et al. 2014, A&A,568, 90Rosenberg, M.J.F.; van der Werf, P.P.; Aalto, S.; et al. 2015, ApJ,801, 72Rosenthal, L.; Hayward, C.C.; Smith, H.; et al. 2015, AAS, 225,143.47Safarzadeh, M.; Hayward, C.C.; Ferguson, H.; & Somerville, R.S.2016, ApJ, 818, 62Sanders, D.B. and Mirabel, I.F.; ARA&A, 34, 749Sanders, D.B.; Soifer, B.T.; Elias, J.H.; et al. 1988, ApJ, 325, 74Sanders, D.B.; Mazzarella, J.M.; Kim, D.-C.; et al. 2003, AJ, 126,1607Satyapal, S.; B¨oker, T.; McAlpine, W.; et al. 2009, ApJ, 704, 439Skrutskie, M.F.; Cutri, R.M.; Steining, R.; et al. 2006, AJ, 131,1163Smith, J.D.T.; Armus, L.; Dale, D.A.; et al. 2007, PASP, 119,1133Smith, H.A.; Li, A.; Li, M.P.; et al. 2010, ApJ, 716, 490Smith, D.J.B.; & Hayward, C.C. 2015, MNRAS, 453, 1497Snyder, G.F.; Hayward, C.C.; Sajina, A.; et al. 2013, ApJ, 768,168Springel, V.; & Hernquist, L. 2003, IAUS, 208, 273Springel, V. 2005, MNRAS, 364, 1105Springel, V.; White, S.D.M.; Jenkins, A.; et al. 2005, Nature, 435,629Springel, V. 2010, ARA&A, 48, 391Stern, D.; Eisenhardt, P.; Gorjian, V.; et al. 2005, ApJ, 631, 163Stern, D.; Assef, R.J.; Benford, D.J.; et al. 2012, ApJ, 753, 30Toba, Y.; Oyabu, S.; Matsuhara, H.; et al. 2013, PASJ, 65, 113Toomre, A.; and Toomre, J. 1972, ApJ, 178, 623de Vaucouleurs, G.; de Vaucouleurs, A.; Corwin, H.G.; et al. 1991,New York: Springer-VerlagVaddi, S.; O’Dea, C.P.; Baum, S.A.; et al. 2016, ApJ, 818, 182Vardoulaki, E.; Charmandaris, V.; Murphy, E.J.; et al. 2015,A&A, 574, 4V´eron-Cetty, M.-P. and V´eron, P. 2010, A&A, 518, 10Veilleux, S.; Kim, D.C.; & Sanders, D.B. 2002, ApJ, 143, 315Villforth, C.; Hamilton, T.; Pawlik, M.M.; et al. 2017, MNRAS,466, 812Wang, L.; Rowan-Robinson, M.; Norberg, P.; et al. 2014, MNRAS,442, 2739Weiner, A.S.; Smith, H.A.; Ashby, M.L.N.; et al. 2018Wenger, M.; Ochsenbein, F.; Egret, B.; et al. 2000, A&AS, 143, 9Werner, M.W.; Roellig, T.L.; Low, F.J.; et al. 2004, ApJ, 154, 1Willett, K.W.; Lintott, C.J.; Bamford, S.P.; et al. 2013, MNRAS,435, 2835Williams, W.L.; Calistro Rivera, G.; Best, P.N.; et al. 2017, arXiv,1711.10504Wright, E.L.; Eisenhardt, P. R. M.; Mainzer, A. K.; et al. 2010,AJ, 140, 1868
APPENDIX A: NOTES ON INDIVIDUALGALAXIES
Group I: The Late-Stage Merger Sample
IRAS 08572+3915 : IRAS 08572+3915 is a ULIRG.
MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Its very steep spectrum from 2 to 20 µ m implies an ex-tremely powerful AGN, which approaches 91% of the totalIR luminosity coming from the galaxy. This is consistentwith Efstathiou et al. (2014), who also found an AGN lu-minosity fraction around 0.9, and Dale et al. (2014), whoestimated an AGN contribution of 85%. IRAS 15250+3609 : IRAS 15250+3609 is a ULIRG.The 2–10 µ m slope is steep enough to imply an AGN contri-bution to the SED, which CIGALE estimates at ∼
47% of theIR luminosity. Franceschini et al. (2003) defines this galaxyas not AGN-dominated but still containing a LINER-typenucleus.
Mrk 231 : Mrk 231 is a ULIRG that is the most IR-luminous system in the sample. We measure the AGN con-tribution at ∼
17% of the IR luminosity, the lowest value inthe late-stage sample, lower than the value found by Rosen-berg et al. (2015) by almost a factor of 5, and lower thanthat from Fritz, Franceschini, & Hatziminaoglou (2006) byalmost a factor of 2. This low fraction is because the dataat wavelengths greater than 60 µ m are well fit by the dustmodel with no AGN contribution needed. Mrk 273 : Mrk 273 is a ULIRG. The steep 3–24 µ mspectrum implies a large AGN contribution that CIGALEestimates at ∼
66% of the IR luminosity. This is higher thanthe value given by Rosenberg et al. (2015) by around a factorof 2.
Mrk 463 : Mrk 463 is a LIRG. The 3–24 µ m SED is wellfit by an AGN model, and the derived AGN contributionis ∼
68% of the IR luminosity. The spectral line ratios for[Ne v ]/[Ne ii ] and [O iv ]/[Ne ii ] are a factor of 5 largerthan others in this sample due to relatively low [Ne ii ] flux,indicating very little star formation is occurring. That theline ratio is so particularly strong is surprising, because anAGN strong enough to ionize neon and oxygen that heavilywould also be expected to have powerful UV emission, whichis not seen in Mrk 463. That no enhanced UV emission isseen is presumably due to a high internal extinction. Mrk 463is also a luminous X-ray source, another indication of strongAGN activity. NGC 2623 : NGC 2623 is a LIRG with the 3.6–24 µ mdata well fit by an AGN model. We measure the AGN con-tribution to be ∼ χ value is relatively large. NGC 3758 : NGC 3758 is a LIRG and the least lumi-nous galaxy in the late-stage merger sample. The 3.6–24 µ mSED is well fit by an AGN model, but the estimated AGNcontribution is ∼ NGC 6090 : NGC 6090 is a LIRG. The steep 4.5–8.0 µ m SED implies an AGN is present, but the estimatedAGN luminosity fraction is ∼ UGC 4881 : UGC 4881 is a LIRG. We measure theAGN contribution to be ∼ µ m colourssuggest an AGN is present UGC 5101 : UGC 5101 is a ULIRG. We measure theAGN contribution to be ∼ µ m SED implies an AGN is present. VV 283 : VV 283 is a LIRG. We measure an AGN con- tribution of ∼ µ m SED implies anAGN is present. VV 705 : VV 705 is a LIRG. We measure an AGN con-tribution of ∼ µ m SED implies an AGN ispresent.Group II: The Reference Sample M51A : M51A, also known as the Whirlpool Galaxy,is a well-known spiral galaxy with an elliptical companion.CIGALE modeled M51A with an AGN at 9% of the IRluminosity. Hern´andez-Garc´ıa et al. (2016) modeled it as anobscured AGN, while L13 calculated a best-fit model thatdid not have an AGN. Nothing in the SED demands thepresence of an AGN.
M51B : M51B is the companion to M51A, and it wasmodeled with an AGN at < % of the IR luminosity.Hern´andez-Garc´ıa et al. (2016) classified it as a LINERgalaxy, but L13 did not calculate an AGN luminosity frac-tion for it, and nothing in the SED requires an AGN to bepresent. NGC 2976 : NGC 2976 is a spiral galaxy in the M81group. CIGALE modeled NGC 2976 with f AGN = 28%, butprevious results from Gonz´alez-Mart´ın et al. (2015) and L13produced models without AGN. Nothing in the SED requiresan AGN.
NGC 3031 : NGC 3031, also known as M81, is a nearbyspiral galaxy. CIGALE modeled the galaxy with f AGN ≤ . , although L13 reported a total IR AGN luminosity frac-tion of 4% and a maximum of 16% in the 8–35 µ m range.The galaxy nucleus has a unique dust spectrum (Smith etal. 2010), and modeling based on standard templates is un-reliable. However, nothing in the total-galaxy SED requiresan AGN. NGC 3077 : NGC 3077 is a low-luminosity irregulargalaxy. We measure the AGN contribution at ∼ (cid:48) away (Knapen et al. 2014), indicatingit is in the earliest stages of merging. It does show signs ofprevious galaxy interaction, but nothing in the SED requiresan AGN. NGC 3190 : NGC 3190 is an edge-on spiral galaxy withprominent dust lanes. CIGALE places a 3 σ upper limit forthe AGN contribution at 1%, although it has been shown tohave a LINER-type nucleus (Gonz´alez-Mart´ın et al. 2015).L13 also described a best-fit model with no AGN contribu-tion, and nothing in the SED suggests an AGN. NGC 3690 : NGC 3690 is the most IR-luminous galaxyin the Reference Sample and its only LIRG (Rosenberg etal. 2014). This galaxy is nearing final pass; although not atcoalescence, it is still classified as a late-stage merger. Wemeasure the AGN contribution at ∼ µ m SED suggests an AGN is present. NGC 4625 : NGC 4625 is a peculiar spiral with ablue SED. CIGALE fits f AGN ≤ . V´eron-Cetty & V´eron(2010) classified it as a Seyfert galaxy (type unknown), but MNRAS000
NGC 3031 : NGC 3031, also known as M81, is a nearbyspiral galaxy. CIGALE modeled the galaxy with f AGN ≤ . , although L13 reported a total IR AGN luminosity frac-tion of 4% and a maximum of 16% in the 8–35 µ m range.The galaxy nucleus has a unique dust spectrum (Smith etal. 2010), and modeling based on standard templates is un-reliable. However, nothing in the total-galaxy SED requiresan AGN. NGC 3077 : NGC 3077 is a low-luminosity irregulargalaxy. We measure the AGN contribution at ∼ (cid:48) away (Knapen et al. 2014), indicatingit is in the earliest stages of merging. It does show signs ofprevious galaxy interaction, but nothing in the SED requiresan AGN. NGC 3190 : NGC 3190 is an edge-on spiral galaxy withprominent dust lanes. CIGALE places a 3 σ upper limit forthe AGN contribution at 1%, although it has been shown tohave a LINER-type nucleus (Gonz´alez-Mart´ın et al. 2015).L13 also described a best-fit model with no AGN contribu-tion, and nothing in the SED suggests an AGN. NGC 3690 : NGC 3690 is the most IR-luminous galaxyin the Reference Sample and its only LIRG (Rosenberg etal. 2014). This galaxy is nearing final pass; although not atcoalescence, it is still classified as a late-stage merger. Wemeasure the AGN contribution at ∼ µ m SED suggests an AGN is present. NGC 4625 : NGC 4625 is a peculiar spiral with ablue SED. CIGALE fits f AGN ≤ . V´eron-Cetty & V´eron(2010) classified it as a Seyfert galaxy (type unknown), but MNRAS000 , 1–16 (2018) J. Dietrich et al.
L13 did not define an AGN contribution for NGC 4625.Nothing in the SED requires an AGN.
NGC 5394 : NGC 5394 is a companion of NGC 5395in the middle stages of merging at a projected separation of28 kpc. We measure the AGN contribution at ∼ µ mSED suggests an AGN is present. NGC 5395 : NGC 5395 is the larger spiral companion ofNGC 5394. CIGALE fits f AGN <
10% to NGC 5395, consis-tent with both V´eron-Cetty & V´eron (2010) and L13 callingit a LINER and attributing to it 3–12% of the bolometricand mid-IR luminosities. Nothing in the SED requires anAGN.
M101 : M101, also known as the Pinwheel Galaxy, isa nearby spiral galaxy showing some tidal disruptions inits outer arms with multiple small companions, includingNGC 5474. CIGALE posits f AGN < NGC 5474 : NGC 5474 is a smaller companion toM101 at a projected separation of 87 kpc. CIGALE mod-els NGC 5474 with f AGN ≤ . Brassington et al. (2015)and L13 did not fit a model containing an AGN to the data,and nothing in the SED requires an AGN. APPENDIX B: PHOTOMETRY AND PACSSPECTROPHOTOMETRIC DATAAPPENDIX C: AGN OBSERVABLES LINEARFIT ANALYSIS
This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies MNRAS000
GN Fraction in Merging Galaxies MNRAS000 , 1–16 (2018) J. Dietrich et al.
MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Figure A1.
Best-fit SED models for the 24 galaxies in the sample containing the nebular emission (gold dotted lines), both attenuatedstellar emission (orange) and non-attenuated stellar emission (blue dot-dashed), dust emission (red solid), and AGN emission (greendashed). The red dots are the best model flux densities and the blue squares mark the observed flux densities with 1 σ error bars.MNRAS000
Best-fit SED models for the 24 galaxies in the sample containing the nebular emission (gold dotted lines), both attenuatedstellar emission (orange) and non-attenuated stellar emission (blue dot-dashed), dust emission (red solid), and AGN emission (greendashed). The red dots are the best model flux densities and the blue squares mark the observed flux densities with 1 σ error bars.MNRAS000 , 1–16 (2018) J. Dietrich et al.
Table B1.
GALEX and
Swift
UV Photometry
GALEX Swift
Galaxy Name FUV NUV UVOT UVW2 UVOT UVM2 UVOT UVW1(mJy) (mJy) (mJy) (mJy) (mJy)Late-Stage Merger SampleIRAS 08572+3915 . ± .
012 0 . ± . ... ... ...IRAS 15250+3609 ... . ± . ... ... ...Mrk 231 ... ... . ± .
050 0 . ± .
056 1 . ± . Mrk 273 . ± .
041 0 . ± . ... ... ...Mrk 463 ... ... . ± .
066 0 . ± .
069 2 . ± . NGC 2623 . ± .
06 0 . ± . ... ... ...NGC 3758 . ± .
040 0 . ± . ... ... ...NGC 6090 . ± .
071 1 . ± . ... ... ...UGC 4881 . ± .
036 0 . ± . ... ... ...UGC 5101 . ± .
021 0 . ± . ... ... ...VV 283 . ± .
019 0 . ± . ... ... ...VV 705 . ± .
062 0 . ± . ... ... ...Reference SampleM51A . ± .
09 162 . ± . ... . ±
63 1650 . ± M51B . ± .
20 4 . ± .
45 89 . ± .
23 282 . ± . . ± . NGC 2976 . ± .
24 19 . ± .
92 85 . ± .
90 200 . ± . . ± . NGC 3031 . ± . . ± . ... . ±
83 6100 . ± . NGC 3077 ... ... . ± .
70 243 . ± . . ± . NGC 3190 . ± .
05 1 . ± .
15 16 . ± .
63 60 . ± .
83 126 . ± . NGC 3690 . ± .
88 13 . ± . ... ... ...NGC 4625 . ± .
33 4 . ± .
49 14 . ± .
88 28 . ± .
09 41 . ± . NGC 5394 . ± .
05 1 . ± .
12 5 . ± .
16 14 . ± .
85 22 . ± . NGC 5395 . ± .
24 4 . ± .
41 23 . ± .
06 61 . ± .
50 103 . ± . M101 . ± . . ± . ... . ±
175 2610 . ± NGC 5474 . ± .
75 22 . ± . ... . ± . . ± . MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Table B2.
SDSS Photometry SDSSGalaxy Name u g r i z(mJy) (mJy) (mJy) (mJy) (mJy)Late-Stage Merger SampleIRAS 08572+3915 . ± .
075 0 . ± .
028 1 . ± .
047 1 . ± .
068 1 . ± . IRAS 15250+3609 . ± .
084 1 . ± .
055 3 . ± . Mrk 231 . ± .
23 14 . ± .
43 17 . ± .
48 20 . ± .
59 39 . ± . Mrk 273 . ± .
19 5 . ± .
17 9 . ± .
28 13 . ± .
41 15 . ± . Mrk 463 . ± .
12 6 . ± .
13 8 . ± .
19 12 . ± .
26 11 . ± . NGC 2623 . ± .
15 9 . ± .
20 16 . ± .
34 21 . ± .
44 26 . ± . NGC 3758 . ± .
13 7 . ± .
16 13 . ± .
29 19 . ± .
41 24 . ± . NGC 6090 . ± .
08 7 . ± .
15 11 . ± .
24 15 . ± .
32 18 . ± . UGC 4881 . ± .
16 7 . ± .
21 12 . ± .
36 17 . ± .
52 19 . ± . UGC 5101 . ± .
051 4 . ± .
15 8 . ± .
26 11 . ± .
38 10 . ± . VV 283 . ± .
17 3 . ± .
11 6 . ± .
19 8 . ± .
27 10 . ± . VV 705 . ± .
14 5 . ± .
16 8 . ± .
26 11 . ± .
35 12 . ± . Reference SampleM51A . ± . ±
23 1920 ±
38 2520 ±
50 2980 ± M51B . ± .
88 218 . ± . . ± . . ± . . ± . NGC 2976 . ± .
57 213 . ± . . ± . . ± . . ± . NGC 3031 ±
24 3850 ±
77 7980 ±
160 11800 ±
240 15900 ± NGC 3077 . ± . . ± . . ± . . ± . . ± . NGC 3190 . ± .
44 94 . ± . . ± . . ± . . ± . NGC 3690 . ± .
59 58 . ± . . ± . . ± . . ± . NGC 4625 . ± .
26 29 . ± .
59 47 . ± .
94 60 . ± .
21 71 . ± . NGC 5394 . ± .
13 14 . ± .
29 25 . ± .
51 34 . ± .
69 39 . ± . NGC 5395 . ± .
33 51 . ± .
02 93 . ± .
86 130 . ± . . ± . M101 . ± . ±
38 2860 ±
57 3650 ±
73 4160 ± NGC 5474 . ± .
93 106 . ± . . ± . . ± . . ± . MNRAS000
93 106 . ± . . ± . . ± . . ± . MNRAS000 , 1–16 (2018) J. Dietrich et al.
Table B3.
IRAS
Photometry2MASS
IRAS
Galaxy Name
J H K s µ m 25 µ m 60 µ m 100 µ m(mJy) (mJy) (mJy) (mJy) (mJy) (mJy) (mJy)Late-Stage Merger SampleIRAS 08572+3915 . ± .
51 3 . ± .
84 3 . ± .
82 318 ±
35 1700 ±
90 7430 ±
370 4770 ± Mrk 231 . ± . . ± . . ± . ±
90 8660 ±
430 32000 ± ± Mrk 273 . ± . . ± . . ± . ±
27 2280 ±
130 21700 ±
870 22500 ± Mrk 463 . ± . . ± . . ± . ±
40 1580 ±
90 2180 ±
110 1920 ± NGC 2623 . ± . . ± . . ± . ±
20 1810 ±
40 23700 ±
930 25900 ± NGC 3758 . ± . . ± . . ± . ±
30 309 ±
43 1260 ±
130 2410 ± NGC 6090 . ± . . ± . . ± . ±
20 1110 ±
40 6660 ±
270 9400 ± UGC 4881 . ± . . ± . . ± . ±
31 599 ±
48 5960 ±
360 10300 ± UGC 5101 . ± . . ± . . ± . ±
40 1030 ±
60 11500 ±
810 19900 ± VV 283 . ± . . ± . . ± . ±
33 386 ±
66 5070 ±
460 7950 ± VV 705 . ± . . ± . . ± . ±
20 1390 ±
70 9210 ±
370 10000 ± Reference SampleM51A ±
79 4690 ±
94 3810 ±
77 7210 ±
75 9560 ±
77 97400 ±
190 221000 ± M51B ±
28 1670 ±
34 1400 ±
28 721 . ± . ±
51 15200 ±
800 31300 ± NGC 2976 . ± . . ± . . ± . . ± . ±
20 13100 ±
30 33400 ± NGC 3031 ±
446 25700 ±
515 21300 ±
427 5860 ±
879 5420 ±
813 44700 ± ± NGC 3077 . ± . . ± . . ± . . ± . ± . ±
390 26500 ± NGC 3190 . ± . . ± . . ± . . ± . . ± . ±
35 10100 ± NGC 3690 . ± . . ± . . ± . ±
400 24100 ± ± ± NGC 4625 . ± .
97 104 . ± . . ± . . ± . . ± . ±
132 3580 ± NGC 5394 . ± .
36 74 . ± .
76 65 . ± .
75 520 . ± . ±
110 5620 ± ...NGC 5395 . ± . . ± . . ± . . ± . . ± . ± ± M101 ±
92 5270 ±
107 4570 ±
94 6200 ±
930 11800 ± ± ± NGC 5474 . ± . . ± . . ± . ... ... ± ... Table B4.
WISE
Photometry for the Late-Stage Merger Sample; it was not used for the Reference Sample (see Section 2)272 Galaxy Name 3.4 µ m 4.6 µ m 12 µ m 22 µ m(mJy) (mJy) (mJy) (mJy)Late-Stage Merger SampleIRAS 08572+3915 . ± .
44 113 . ± .
78 308 . ± . ± Mrk 231 . ± . . ± . . ± . ± Mrk 273 . ± .
58 36 . ± .
45 212 . ± . ± Mrk 463 . ± .
86 206 . ± . . ± . ± NGC 2623 . ± .
64 25 . ± .
52 181 . ± . ± NGC 3758 . ± .
46 43 . ± .
61 127 . ± .
63 227 . ± . NGC 6090 ... ... ... ...UGC 4881 . ± .
76 15 . ± .
35 114 . ± .
69 352 . ± . UGC 5101 . ± .
75 79 . ± .
71 158 . ± . . ± . VV 283 . ± .
11 11 . ± .
51 89 . ± .
61 252 . ± . VV 705 . ± .
65 17 . ± .
47 169 . ± . . ± . MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Table B5.
Spitzer /IRAC and MIPS Photometry IRAC MIPSGalaxy Name 3.6 µ m 4.5 µ m 5.8 µ m 8.0 µ m 24 µ m 70 µ m 160 µ m(mJy) (mJy) (mJy) (mJy) (mJy) (mJy) (mJy)Late-Stage Merger SampleIRAS 08572+3915 . ± . . ± . . ± . ±
10 1390 ±
56 6160 ±
250 1850 ± IRAS 15250+3609 . ± .
22 9 . ± .
29 38 . ± . . ± . ±
40 8410 ±
330 2750 ± Mrk 231 . ± . . ± . . ± . ± ... ... ...Mrk 273 . ± . . ± . . ± . . ± . ±
70 27400 ± ± Mrk 463 . ± . . ± . . ± . . ± . ±
60 3310 ±
140 1190 ± NGC 2623 . ± .
88 27 . ± .
82 62 . ± . . ± . ±
60 27300 ± ± NGC 3758 . ± . . ± . . ± . . ± . ± ... ...NGC 6090 . ± . . ± . . ± . . ± . ± . ± ...UGC 4881 . ± . . ± .
76 37 . ± . . ± . ±
24 9740 ±
960 7600 ± UGC 5101 . ± . . ± . . ± . . ± . ±
43 15400 ± ± VV 283 . ± .
64 13 . ± .
55 34 . ± . . ± . ±
17 6390 ±
730 7190 ± VV 705 . ± .
85 17 . ± .
75 43 . ± . . ± . ±
64 7480 ±
650 5690 ± Reference SampleM51A ±
71 1550 ±
47 3690 ±
111 9810 ±
294 12000 ±
480 135000 ± ...M51B . ± . . ± . . ± . ... ±
62 18000 ± ...NGC 2976 . ± . . ± . . ± . . ± . ±
55 20000 ±
800 50300 ± NGC 3031 ±
330 6930 ±
208 5700 ±
171 7060 ±
212 5410 ±
216 82400 ± ± NGC 3077 . ± . . ± . . ± . . ± . ± ... ...NGC 3190 . ± . . ± . . ± . . ± . . ± . ±
224 15400 ± NGC 3690 . ± . . ± . ±
31 2370 ±
71 17400 ± ... ...NGC 4625 . ± . . ± .
85 54 . ± . . ± . . ± . ... ...NGC 5394 . ± .
26 29 . ± .
89 82 . ± .
47 222 . ± . . ± . ... ...NGC 5395 . ± . . ± .
58 164 . ± . . ± . . ± . ... ...M101 ±
80 1770 ±
53 3110 ±
93 7470 ±
224 10500 ±
420 117000 ± ± NGC 5474 . ± . . ± . . ± . . ± . . ± . ... ± MNRAS000
420 117000 ± ± NGC 5474 . ± . . ± . . ± . . ± . . ± . ... ± MNRAS000 , 1–16 (2018) J. Dietrich et al.
Table B6.
Herschel /PACS and SPIRE PhotometryPACS SPIREGalaxy Name 75 µ m 110 µ m 170 µ m 250 µ m 350 µ m 500 µ m(mJy) (mJy) (mJy) (mJy) (mJy) (mJy)Late-Stage Merger SampleIRAS 08572+3915 ±
620 4120 ±
410 1830 ±
180 446 ±
33 131 ±
16 2 . ± . Mrk 231 ± ± ± ±
500 1670 ±
180 456 ± Mrk 273 ± ± ± ±
460 1200 ±
130 334 ± Mrk 463 ... ... ... ±
45 199 ±
19 55 . ± . NGC 2623 ± ± ± ± ±
200 473 ± NGC 3758 ± ... ±
280 1100 ±
90 416 ±
42 85 . ± . NGC 6090 ±
600 7020 ±
700 4860 ±
490 2910 ±
200 1080 ±
80 280 ± UGC 4881 ± ± ±
930 3090 ±
320 1150 ±
140 344 ± UGC 5101 ± ... ± ±
550 2140 ±
220 610 ± VV 283 ±
880 7840 ± ±
840 2320 ±
240 891 ±
97 244 ± VV 705 ± ± ±
820 1950 ±
200 667 ±
78 185 ± Reference SampleM51A ± ... ± ± ± ± M51B ± ... ± ±
790 4360 ±
310 1380 ± NGC 2976 ± ± ± ± ±
800 4220 ± NGC 3031 ± ... ± ± ± ± NGC 3077 ± ± ± ±
626 3290 ±
241 1030 ± NGC 3190 ±
614 11800 ± ± ±
558 3420 ±
243 1130 ± NGC 3690 ± ± ± ± ±
503 2070 ± NGC 4625 ±
187 3810 ±
389 4450 ±
455 2240 ±
160 1050 ±
79 360 . ± . NGC 5394 ±
787 10000 ± ±
766 2920 ±
205 1120 ±
81 339 . ± . NGC 5395 ±
528 11200 ± ± ±
574 3570 ±
251 1230 ± M101 ± ± ± ± ± ± NGC 5474 ±
403 3320 ±
582 6120 ±
648 4210 ±
304 2330 ±
176 985 . ± . Table B7.
The PACS Spectrophotometric (SP) Data Points, with wavelength, number of observations, and average flux; ... signifiesunknown number of observations were used.PACS ‘Blue’ SP PACS ‘Green’ SP PACS ‘Red’ SPGalaxy Name λ ( µ m) λ ( µ m) λ ( µ m) ±
89 497 ±
151 60 ± Mrk 231 66 701 ±
90 714 ±
160 490 ± Mrk 273 63 419 ±
88 307 ±
160 154 ± Mrk 463 63 ... ±
88 ... ±
157 ... ± NGC 2623 64 71 ±
90 61 ±
160 49 ± NGC 6090 65 209 ±
81 234 ±
163 57 ± UGC 4881 66 154 ±
92 153 ±
164 69 ± UGC 5101 63 284 ±
92 48 ±
151 122 ± VV 705 65 206 ±
82 128 ±
164 65 ± MNRAS , 1–16 (2018)
GN Fraction in Merging Galaxies Table C1.
The linear regression test results of f AG N vs. colours showing the strongest correlations (Pearson r > σ significance).Flux Ratio Number of systems Slope r p wo GALEX
FUV–MIPS 24 µ m 20 . ± .
08 0 .
82 8 . × − GALEX
FUV–
IRAS µ m 18 . ± .
03 0 .
83 1 . × − GALEX
FUV–
IRAS µ m 19 . ± .
99 0 .
83 1 . × − GALEX
FUV–PACS 70 µ m 20 . ± .
05 0 .
81 1 . × − GALEX
FUV–MIPS 70 µ m 13 . ± .
14 0 .
82 0 . × − GALEX
NUV–
IRAS µ m 20 . ± .
65 0 .
82 7 . × − GALEX
NUV–MIPS 24 µ m 21 . ± .
88 0 .
86 7 . × − GALEX
NUV–
IRAS µ m 19 . ± .
84 0 .
87 1 . × − GALEX
NUV–
IRAS µ m 20 . ± .
85 0 .
85 1 . × − GALEX
NUV–PACS 70 µ m 21 . ± .
88 0 .
83 2 . × − GALEX
NUV–MIPS 70 µ m 14 . ± .
95 0 .
85 1 . × − Sloan u –MIPS 24 µ m 23 . ± .
96 0 .
82 1 . × − SloaNGC n u –MIPS 70 µ m 15 . ± .
94 0 .
83 148 . × − Sloan g –MIPS 24 µ m 23 . ± .
01 0 .
82 1 . × − Sloan g –PACS 70 µ m 23 . ± .
97 0 .
81 3 . × − Sloan g –MIPS 70 µ m 15 . ± .
02 0 .
83 0 . × − Sloan r –MIPS 24 µ m 23 . ± .
03 0 .
82 2 . × − Sloan r –MIPS 70 µ m 15 . ± .
05 0 .
83 145366 . × − Sloan i –MIPS 24 µ m 23 . ± .
05 0 .
81 2 . × − Sloan i –MIPS 70 µ m 15 . ± .
07 0 .
83 0 . × − Sloan z –MIPS 24 µ m 23 . ± .
04 0 .
83 1 . × − Sloan z –PACS 70 µ m 23 . ± .
99 0 .
81 2 . × − Sloan z –MIPS 70 µ m 15 . ± .
04 0 .
85 6 . × − Sloan z –PACS 100 µ m 18 . ± .
93 0 .
81 4 . × − µ m 23 . ± .
01 0 .
83 8 . × − µ m 23 . ± .
97 0 .
82 1 . × − µ m 15 . ± .
07 0 .
85 7 . × − IRAS µ m 12 . ± .
81 0 .
83 77 . × − µ m 18 . ± .
93 0 . . × − µ m 24 . ± .
61 0 . . × − µ m 24 . ± . . . × − µ m 23 . ± .
04 0 .
82 1 . × − µ m 23 . ± .
94 0 .
83 1 . × − µ m 15 . ± .
14 0 .
82 16 . × − IRAS µ m 12 . ± . .
85 431 . × − µ m 18 . ± .
87 0 .
83 2 . × − µ m 23 . ± .
87 0 .
82 1 . × − IRAS µ m 12 . ± .
67 0 .
83 7 . × − µ m 18 . ± .
76 0 .
83 1 . × − IRAC 3.6 µ m–IRAC 4.5 µ m 24 . ± .
19 0 .
84 2 . × − IRAC 3.6 µ m– IRAS µ m 12 . ± .
36 0 .
85 0 . × − IRAC 4.5 µ m– IRAS µ m 12 . ± .
21 0 .
83 89 . × − IRAC 5.8 µ m– IRAS µ m 12 . ± .
79 0 .
83 87 . × − IRAS µ m– IRAS µ m 12 − . ± . − .
89 1069 . × − IRAS µ m–MIPS 160 µ m 14 − . ± . − .
85 139 . × − PACS 70 µ m–MIPS 160 µ m 14 − . ± . − .
85 1 . × − MIPS 70 µ m– IRAS µ m 7 − . ± . − .
93 222578 . × − MIPS 70 µ m–MIPS 160 µ m 13 − . ± . − .
87 114097 . × − MIPS 70 µ m–PACS 160 µ m 14 − . ± . − .
85 12987 . × − MIPS 70 µ m–SPIRE 250 µ m 15 − . ± . − .
87 2 . × − MIPS 70 µ m–SPIRE 500 µ m 15 − . ± . − .
87 3 . × − MIPS 70 µ m–SPIRE 350 µ m 15 − . ± . − .
86 3 . × − IRAS µ m–PACS 160 µ m 12 − . ± . − .
94 5 . × − IRAS µ m–SPIRE 500 µ m 12 − . ± . − .
87 248 . × − IRAS µ m–SPIRE 350 µ m 12 − . ± . − .
89 9 . × − IRAS µ m–SPIRE 250 µ m 12 − . ± . − .
91 3 . × − PACS 100 µ m–PACS 160 µ m 18 − . ± . − . . × − MNRAS000