The Connection between Star-Forming Galaxies, AGN Host Galaxies and Early-Type Galaxies in the SDSS
Joon Hyeop Lee, Myung Gyoon Lee, Taehyun Kim, Ho Seong Hwang, Changbom Park, Yun-Young Choi
aa r X i v : . [ a s t r o - ph ] J un Draft version January 14, 2019
Preprint typeset using L A TEX style emulateapj v. 11/26/03
THE CONNECTION BETWEEN STAR-FORMING GALAXIES, AGN HOST GALAXIES AND EARLY-TYPEGALAXIES IN THE SDSS
Joon Hyeop Lee , Myung Gyoon Lee , Taehyun Kim , Ho Seong Hwang , Changbom Park , and Yun-YoungChoi Astronomy Program, Department of Physics and Astronomy, Seoul National University, Seoul 151-742, Korea Korea Institute for Advanced Study, Dongdaemun-gu, Seoul 103-722, Koreaemail: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]
Draft version January 14, 2019
ABSTRACTWe present a study of the connection between star-forming galaxies, AGN host galaxies, and normalearly-type galaxies in the Sloan Digital Sky Survey (SDSS). Using the SDSS DR5 and DR4plusdata, we select our early-type galaxy sample in the color versus color-gradient space, and we classifythe spectral types of the selected early-type galaxies into normal, star-forming, Seyfert, and LINERclasses, using several spectral line flux ratios. We investigate the slope in the fundamental spacefor each class of early-type galaxies and find that there are obvious differences in the slopes of thefundamental planes (FPs) among the different classes of early-type galaxies, in the sense that theslopes for Seyferts and star-forming galaxies are flatter than those for normal galaxies and LINERs.This may be the first identification of the systematic variation of the FP slope among the subclasses ofearly-type galaxies. The difference in the FP slope might be caused by the difference in the degree ofnonhomology among different classes or by the difference of gas contents in their merging progenitors.One possible scenario is that the AGN host galaxies and star-forming galaxies are formed by gas-richmerging and that they may evolve into normal early-type galaxies after finishing their star formationor AGN activities.
Subject headings: galaxies: active — galaxies: star-forming — galaxies: elliptical and lenticular, cD— galaxies: evolution — galaxies: formation — galaxies: fundamental parameters INTRODUCTION
One of the interesting issues of modern observationalcosmology is the evolutionary connection between var-ious classes of galaxies: early-type galaxies, late-typegalaxies, dwarf galaxies, mergers, star-forming galaxies,AGN host galaxies, and so on. Such connections, iffound, may be important constraints on the galaxy for-mation and evolution scenario. Early-type galaxies arean ideal target to investigate this connection problem.Early-type galaxies were considered for a long time tobelong to a single family. However, recent studies foundthat there are variations of earl-type galaxies. For ex-ample, some early-type galaxies have young stellar pop-ulations unlike typical normal early-type galaxies (Abra-ham et al. 1999; Menanteau et al. 2001), and some early-type galaxies have active nuclei (AGNs) in their cen-ters (S´anchez et al. 2004; Capetti & Balmaverde 2006).These abnormal classes of early-type galaxies are pos-sibly the links between normal early-type galaxies andother classes of galaxies, and the relationship betweennormal early-type galaxies and other classes of galax-ies may be important for constraining the formationmodel of early-type galaxies. Recently, Lee et al. (2006)suggested that there are possible evolutionary connec-tions between star-forming galaxies, AGN host galaxies,and normal early-type galaxies, from the study of “blueearly-type galaxies” in the GOODS (Great ObservatoriesOrigins Deep Survey; Giavalisco et al. 2004)
HST /ACSfields.The fundamental plane (FP; Dressler et al. 1987; Djor-govski & Davis 1987) of galaxies provides important cluesabout the relationship among various classes of galaxies. Tacconi et al. (2002) studied the properties of 18 ultralu-minous infrared galaxies (ULIRGs) and concluded thatthose ULIRGs may evolve into normal elliptical galaxiesbased on their position in the fundamental space. It isalso known that AGN host galaxies reside at the sameplane as the normal early-type galaxies in the fundamen-tal space (Snellen et al. 2003; Woo et al. 2004), whichimplies that those active galaxies and normal early-typegalaxies have a very close relationship in the evolutionarysequence. However, it was difficult to secure a large sam-ple of abnormal classes of early-type galaxies (e.g., AGNhost early-type galaxies or post–star-forming early-typegalaxies) in the previous studies.In this Letter, we present a study of the connection be-tween the various sub-classes of early-type galaxies usingthe large sample in the Sloan Digital Sky Survey (SDSS;York et al. 2000). The outline of this Letter is as follows.Section 2 describes the data set we used and the meth-ods to select early-type galaxies and to classify early-type galaxies into several sub-classes. We present theFP analysis of each class of early-type galaxies in § § h = 0 .
7, Ω Λ = 0 .
7, and Ω M = 0 . DATA AND GALAXY CLASSIFICATION
We used the SDSS Data Release 5 (DR5; Adelman-McCarthy et al. 2007) in this study. The DR5 imagingdata cover about 8000 deg in the ugriz bands, and theDR5 spectroscopic data cover 5600 deg . The pho-tometric and spectroscopic observations were conducted Lee et al.
Fig. 1.—
Upper panel : Segregation between early-type galaxies(dark gray points) and late-type galaxies (light gray points) in thecolor vs. color-gradient space. The color gradient ∆( g − i ) is definedas the difference in ( g − i ) color between the region at R < . R petand the region at 0 . R pet < R < R pet [negative ∆( g − i ) forblue outside]. The lines represent different segregation guidelinesfor different magnitude ranges (solid line for 14 . < r pet < . . < r pet < .
5, long-dashed line for16 . < r pet < .
0, and dotted line for 17 . < r pet < . Lowerleft panel : AGN selection in the BPT diagram (Baldwin et al.1981). The short-dashed line is the theoretical upper limit of star-forming galaxies (Kewley et al. 2001), and the long-dashed line isthe empirical criterion of Kauffmann et al. (2003). In this study, weused our empirical criterion ( solid line ) to distinguish AGNs fromstar-forming galaxies.
Lower right panel : Seyfert-LINER segrega-tion in the [O III]/[H β ] vs. [O I]/[H α ] diagram. The solid lineis our criterion, and the long-dashed line is that of Groves et al.(2006). with the 2.5 m SDSS telescope at the Apache Point Ob-servatory in New Mexico between 1999 March and 2005June.We used only the galaxy catalog with spectroscopicinformation, which includes 573,113 available objects.From this catalog, we selected early-type galaxies withthe galaxy classification method using the color versuscolor-gradient space (Park & Choi 2005) as shown inFig. 1. In this classification method, colors and colorgradients are the main criteria for classification, andthe inverse concentration (C − ≡ R / R ) cut isalso adopted differentially for different magnitude ranges:C − < .
43 for 14 . < r P et < .
0, C − < .
45 for16 . < r P et < .
5, C − < .
47 for 16 . < r P et < . − < .
48 for 17 . < r P et < . Park &Choi (2005) estimated the completeness and reliabilityof this classification method to be as large as ∼ r P et < .
5. The color-gradient estimation was con-ducted for the 389,789 objects in the SDSS DR4plus(Choi et al. 2007) sample, which is one of the productsof the New York University Value-Added Galaxy Cata-log (Blanton et al. 2005). Finally, we selected 139,183early-type galaxies.We selected AGN host galaxies in our early-type galaxy R n % is the n % Petrosian radius, and r Pet is the Petrosianmagnitude in the r band. sample, using the line flux ratio diagram of [O III]/[H β ]versus [N II]/[H α ] (BPT diagram; Baldwin et al. 1981)as shown in Fig. 1. Kewley et al. (2001) suggested atheoretical upper limit of star-forming galaxies in theBPT diagram, and Kauffmann et al. (2003) adopted theirempirical criterion to select AGNs as drawn in Fig. 1.However, since our sample shows small offsets from theprevious samples in the BPT diagram, we used our ownempirical criterion to segregate AGN host galaxies fromstar-forming galaxies: [O III] / [H β ] = 0 . / ([N II] / [H α ] − .
25) + 1 .
25, which is intermediate between the criteriaof Kewley et al. (2001) and the criteria of Kauffmannet al. (2003). Among the selected AGN host early-typegalaxies, we distinguished Seyferts from LINERs in the[O III]/[H β ] versus [O I]/[H α ] diagram. Since there isalso a small offset between the sample of Groves et al.(2006) and our sample in this diagram as shown in Fig. 1,we used our own empirical guideline: [O III]/[H β ] = 1 . α ] +1 .
7. This process returns 1913 star-forminggalaxies, 1129 Seyfert galaxies, and 320 LINERs with asignal-to-noise ratio (S/N) of > >
10 for their velocity dispersion, which returns130,147 normal early-type galaxies, 1608 star-forminggalaxies, 1050 Seyfert galaxies, and 293 LINER galaxies. THE FUNDAMENTAL PLANE
We estimated the slope in the fundamental space foreach class of early-type galaxies, using the ordinary least-square bisector method (Isobe et al. 1990). The normalearly-type galaxies in the SDSS show a good FP relation,as already shown by Bernardi et al. (2003b). In addition,AGN host galaxies and a large fraction of star-forminggalaxies also reside in the FP of normal early-type galax-ies. The slopes of normal early-type galaxies and LINERearly-type galaxies are consistent within 1 σ slope error(1 . ± .
002 for normal and 1 . ± .
052 for LINER).However, the slopes of Seyfert galaxies and star-forminggalaxies are significantly ( > σ slope ) smaller than thoseof normal and LINER galaxies (1 . ± .
023 for Seyfertsand 0 . ± .
019 for star-forming galaxies).These differences in the FP slope, however, are possi-bly artifacts caused by the different parameter distribu-tions of different classes in the fundamental space. Sincethe slope of the FP varies in the different ranges of fun-damental parameters (Bernardi et al. 2003b), it is fair We used the emission line fluxes measured by the SDSS pipeline(Stoughton et al. 2002), while Kauffmann et al. (2003) and Groveset al. (2006) used those measured by the pipeline of Tremonti etal. (2004). We define normal early-type galaxies as early-type galaxieswithout evidence of obvious line emission. onnection between Star-forming, AGN, and Early-Type Galaxies 3
Fig. 2.—
FP slope estimations with the same σ v distributionin given σ v ranges, for 0 . < z < . Upper panels : Normaland star-forming early-type galaxies, with 100 km s − < σ v < − . Middle panels : Normal and Seyfert early-type galaxies,with 100 km s − < σ v <
220 km s − . Lower panels : Normal andLINER early-type galaxies, with 100 km s − < σ v <
240 km s − .The lines are OLS bisector fittings, and the resulting FP slope and1 σ slope error are shown in the lower right-hand corner of eachpanel. Each velocity dispersion boundary was selected within thecommon range of the distributions of the two classes. to compare the FP slope with the galaxies in the samedomain of fundamental space. Among the three fun-damental parameters, we set boundaries with velocitydispersion ( σ v ), because this parameter may be less sen-sitive to galaxy evolution effects than effective radius orsurface brightness. The approximate range of σ v in eachclass is as follows: 20 −
420 km s − for normal galaxies,20 −
170 km s − for star-forming galaxies, 20 −
220 kms − for Seyfert galaxies, and 100 −
240 km s − for LINERgalaxies. Because the estimation error of σ v is very largefor σ v <
100 km s − (Choi et al. 2007), we use galax-ies with σ v >
100 km s − only. In addition, it is safeto compare the galaxies at 0 . < z < .
1, to minimizethe selection effects caused by the fixed fiber-size andthe LINER incompleteness (Kewley et al. 2006). Fig. 2shows the FP slope estimates with the limits of σ v andredshift. The normal early-type galaxies were randomlyresampled to have the same σ v distribution as the com-pared abnormal early-type galaxies. In this comparison,the difference between star-forming galaxies and normalgalaxies is 2 . σ slope , and the difference between Seyfertgalaxies and normal galaxies is 2 . σ slope , while LINERgalaxies and normal galaxies have consistent FP slopes.Interestingly, Figure 4 of Snellen et al. (2003) and Fig-ure 3 of Woo et al. (2004) show that AGN host galaxiesreside in the FP of normal early-type galaxies but thatthey are slightly tilted from the FP of normal early-typegalaxies, for the same direction as our Seyfert galaxiesare.However, it is necessary to estimate the sampling errorfor each abnormal class of galaxies, because the samplesizes of abnormal galaxies are much smaller than that ofnormal galaxies. We conducted random sampling testsusing the normal galaxy sample, with the same sample Table 1. Random Sampling Tests (Ntrial=10000) with the Same Sam-ple Size and σ v Distribution σ v Range FP Slope of Class 1 FP Slope(km s − ) For Normal Galaxies of Class 2 Error100 −
170 1.315 1.187 a −
220 1.302 1.220 b −
240 1.302 1.327 c a For star-forming galaxies. b For Seyfert galaxies. c For LINER galaxies. size and σ v distribution as each abnormal class. The ran-dom sampling was repeated 10,000 times for each com-parison, and the results are summarized in Table 1. Theerror values in Table 1 include both estimation errors andsampling errors. Considering these errors, star-forminggalaxies and Seyfert galaxies have different FP slopesfrom normal early-type galaxies with probabilities morethan 98% (2 . σ ) in the same σ v and redshift ranges asthe Fig. 2, and LINER galaxies and normal galaxies haveconsistent FP slopes.One observational bias may affect the FP properties ofthe AGN host galaxies. Since spectroscopy was targetedto the central parts of galaxies, the σ v estimates of AGNhost galaxies are possibly distorted by the existence ofan AGN. However, it is difficult not only to correct suchan effect but also to guess how much that changes theFP slope of AGN host galaxies, using current data. DISCUSSION AND CONCLUSION
The FP of early-type galaxies is considered to beclosely related to the virial theorem. However, the virialplane (VP) does not match the FP exactly, and severalstudies have been conducted to investigate the origin ofthe tilt between the two planes. Busarello et al. (1997)found from the analysis of 40 nearby elliptical galaxiesthat the primary origin of the tilt between the VP andthe FP is the non-homology of elliptical galaxies; that is,the velocity dispersion profiles of more massive galaxiesare steeper than those of less massive galaxies. In addi-tion to the non-homology effect, they also suggested thatthe effect of stellar populations is about 30% responsiblefor the FP tilt. This result was supported by Trujilloet al. (2004), who investigated the FP of 911 early-typegalaxies in the combined catalog of the SDSS and 2MASS(Two Micron All Sky Survey; Bell et al. 2003).Adopting the nonhomology effect to the FP slope dif-ference among different classes, it is inferred that theSeyfert and star-forming early-type galaxies may havestronger non-homology than the normal early-type galax-ies. It is not easy to tell what causes such strong non-homology in Seyfert and star-forming early-type galax-ies, but it is possible to assume that some events causingtheir star formation or AGN activity might also affect (orbe affected by) the dynamical structures of those classes.Recently, several numerical experiments were con-ducted to explain the physical origin of the FP tilt.Some self-consistent hydrodynamic simulations showedthat the major merging with gas dissipation can repro-duce the tilt between the VP and the FP (O˜norbe et al.2005; Dekel & Cox 2006). Robertson et al. (2006) studied Lee et al.the variation of the slope of the FP in diverse mergingcases. They found that the dissipational merging pro-duces the FP tilt, whereas the FP of the dissipationlessmerging is parallel to the VP. Furthermore, they showedthat the merging of disk galaxies with a larger fractionof gas makes a larger tilt between the VP and the FP.These previous results are very interesting, since theyprovide a hint to understanding the origin of the dif-ference in the FP slopes of different classes of early-type galaxies. According to those simulations, the differ-ent classes of our early-type galaxies might evolve frommergers with different gas contents. Normal and LINERearly-type galaxies might be formed by the merging ofgas-poor disk galaxies, while Seyfert and star-formingearly-type galaxies might be formed by the merging ofgas-rich disk galaxies. This interpretation is also consis-tent with the conclusion of Kewley et al. (2006) that themajor difference between Seyferts and LINERs may bethe gas accretion rate. However, it is expected that thestar formation or AGN activities in abnormal early-typegalaxies will not continue forever. Since the structuralparameters and the loci in the fundamental space of theabnormal classes are very similar to those of the normalearly-type class, it is not too absurd to infer that theAGN host or star-forming early-type galaxies may evolveinto normal early-type galaxies after finishing their starformation or AGN activities.The observational analysis of Kewley & Dopita (2003)led to an evolutionary scenario: starbursts initially trig-gered by tidal interactions → AGN activated by gas fun-neled toward the merger nucleus → circumnuclear starformation at late stages of the merger. When combiningthis scenario with the hierarchical merging scenario ofearly-type galaxies (Toomre 1977; Searle & Zinn 1978),it is possible to extend the scenario of Kewley & Dopita(2003) as follows: mergers → galaxies with star forma-tion or AGN → galaxies finishing their star formation orAGN activity → REFERENCESAbraham, R. G., Ellis, R. S., Fabian, A. C., Tanvir, N. R., &Glazebrook, K. 1999, MNRAS, 303, 641Adelman-McCarthy, J., et al. 2007, ApJS, in pressBaldwin, J, Phillips, M., & Terlevich, R. 1981, PASP, 93, 5Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003,ApJS, 149, 289Bernardi, M., et al. 2003a, AJ, 125, 1817Bernardi, M., et al. 2003b, AJ, 125, 1866Blanton, M. R., et al. 2005, AJ, 129, 2562Busarello, G., Capaccioli, M., Capozziello, S., Longo, G., & Puddu,E. 1997, A&A, 320, 415Capetti, A., & Balmaverde, B. 2006, A&A, 453, 27 Choi, Y.-Y., Park, C., & Vogeley, M. S. 2007, ApJ, 658, 884Dekel, A., & Cox, T. J. 2006, MNRAS, 370, 1445Djorgovski, S., & Davis, M. 1987, ApJ, 313, 59Dressler, A., Lynden-Bell, D., Burtein, D., Davies, R. L., Faber, S.M., Terlevich, R. J., & Wengner, G. 1987, ApJ, 313, 42Giavalisco, M., et al. 2004, ApJ, 600, L93Groves, B., Kewley, L., Kauffmann, G., & Heckman, T. 2006, NewAstronomy Reviews, 50, 743Isobe, T., Feigelson, D., Akritas, M. G., & Babu, G. J. 1990, ApJ,364, 104Jørgensen, I., Franx, M., & Kjærgaard, P. 1995, MNRAS, 276, 1341Kauffmann, G., et al. 2003, MNRAS, 346, 1055 onnection between Star-forming, AGN, and Early-Type Galaxies 5onnection between Star-forming, AGN, and Early-Type Galaxies 5