Bars in Starbursts and AGNs -- A Quantitative Reexamination
Lei Hao, Shardha Jogee, Fabio D. Barazza, Irina Marinova, Juntai Shen
aa r X i v : . [ a s t r o - ph . C O ] O c t Galaxy Evolution: Emerging Insights and Future ChallengesASP Conference Series, Vol. **VOLUME**, 2009Shardha Jogee, Lei Hao, Guillermo Blanc, Irina Marinova
Bars in Starbursts and AGNs – A QuantitativeReexamination
Lei Hao, Shardha Jogee, Fabio D. Barazza, Irina Marinova, andJuntai Shen Department of Astronomy, University of Texas, Austin, Texas 78712,USA. Laboratoire d’Astrophysique, ´Ecole Polytechnique F´ed´erale deLausanne (EPFL), Observatoire de Sauverny CH-1290 Versoix,Switzerland
Abstract.
Galactic bars are the most important driver of secular evolutionin galaxies. They can efficiently drive gas into the central kiloparsec of galax-ies, thus feed circumnuclear starbursts, and possibly help to fuel AGN. Theconnection between bars and AGN activities has been actively debated in thepast two decades. Previous work used fairly small samples and often lacked aproper control sample. They reported conflicting results on the correlation be-tween bars and AGN activity. Here we revisit the bar-AGN and bar-starburstconnections using the analysis of bars in a large sample of about 2000 SDSSdisk galaxies (Barazza, Jogee, & Marinova 2008). We find that AGN and star-forming galaxies have similar optical bar fractions, 47% and 50%, respectively.Both bar fractions are higher than that in inactive galaxies (29%). We discussthe implications of the study on the relationship between host galaxies and theircentral activities.
1. Introduction
Large-scale bars are very common in disk galaxies. At near infrared (NIR) wave-lengths, the optical bar fraction averaged across different Hubble types is ∼ ∼
72% from visualclassification (Eskridge et al. 2000). At optical wavelengths, quantitative meth-ods yield an average optical bar fraction of 45% to 52% (Marinova & Jogee 2007;Barazza, Jogee, & Marinova 2008; Aguerri, M´endez-Abreu, & Corsini 2009), whilevisual classification yields ∼
60% (de Vaucouleurs 1963). The optical bar frac-tion is somewhat lower than the NIR fraction due to the obscuration of bars bydust lanes and star formation along the bar.Several studies have now moved beyond considering only the bar fractionaveraged over many Hubble types. They investigated how the optical bar frac-tion varies with the Hubble types or host properties. The optical bar fractionrises in galaxies with low Bulge-to-Disk ratio or/and high luminosity (Odewahn1996; Barazza et al. 2008, 2009; Aguerri et al. 2009; Marinova et al. 2009).The non-axisymmetric stellar bar can drive gas from the outer disk to thecentral kiloparsec, where they trigger star formation. This is supported by1
Hao et al. several observations, which show that barred galaxies host high gas densitiesand circumnuclear starbursts (e.g. Jogee, Scoville, & Kenney 2005) and thatthe central gas concentration is larger in barred than unbarred galaxies (e.g.Sakamoto et al. 1999; Sheth et al. 2004). Due to their efficiency in driving gasinflows in the disks, bars are strong candidates for the triggering of nuclearactivities.There is strong evidence for a connection between large-scale bars and cir-cumnuclear starbursts. Barred galaxies show enhanced radio continuum and in-frared emissions compared with unbarred ones (e.g. Hummel 1981; Hawarden et al.1986), and starburst galaxies tend to be more barred compared with the non-starburst galaxies (e.g. Arsenault 1989; Huang et al. 1996; Ho, Filippenko, & Sargent1997; Hunt & Malkan 1999). Bars can also set up resonances, such as the innerLindblad resonances (ILRs), which can prevent the gas from going further in.Therefore, gas often builds up on a ring (a few hundred parsecs in radius) andthe circumnuclear starbursts can occur there (e.g. Jogee et al. 2005).The connection between bars and AGNs is less clear (see Jogee 2006 for areview). Over the last two decades a large number of studies were carried out toidentify if such a correlation exists. Most of them compared the bar fraction ofthe AGN sample with that of a control sample of inactive galaxies. The resultsare controversial. For example, studies like Ho et al. (1997), Hunt & Malkan(1999), Mulchaey & Regan (1997), and Martini et al. (2003) found no excess ofbars in Seyfert galaxies, while Knapen, Shlosman, & Peletier (2000), Laine et al.(2002), and Laurikainen, Salo, & Buta (2004b) reported a higher fraction of barsin Seyferts.Previous studies are limited in several aspects. Firstly, the sizes of the sam-ples are often small, including only a few tens of AGNs and control galaxies.In some cases, the control sample was not well matched to the active sample.Secondly, identifications method for bars are not always consistent across sam-ples. Thirdly, the spectral classifications of galaxies as AGNs or inactive galaxiesare significantly inconsistent. Most studies adopted the galaxy classifications inNASA/IPAC NED, which could be done by different people using different cri-teria. Our study tries to overcome some of these limitations by systematicallyinvestigating the optical bar fraction of AGNs and non-AGNs in a large numberof galaxies from the Sloan Digital Sky Survey (SDSS), using matched active andcontrol samples, and the same consistent quantitative method for identifyingbars across samples.
2. The Sample and Their Spectral Classifications
Our sample is based on the one in Barazza et al. (2008). From the 3692 galaxiesin the Sloan Digital Sky Survey (SDSS) with 18 . < M g < − . . < z < .
03, Barazza et al. (2008) selected 1961 disk galaxies viatheir blue colors. They applied ellipse fitting to find and characterize bars inthese disk galaxies. They exclude 169 disk galaxies with failed or messy ellipsefittings, or ambiguous classifications from the final analysis. 648 galaxies wereclassified as highly-inclined galaxies ( i > ◦ ) and disregarded as morphologicalanalysis is unreliable for such systems. In the final sample of 1144 moderatelyinclined galaxies, 553 were barred galaxies, and 591 were unbarred galaxies. ars in Starbursts and AGNs ′′ aperture, with a spectral resolution of 2200 covering a wavelength rangefrom 3700 ˚A to 9000 ˚A. The 3 ′′ aperture size corresponds to 606 pc to 1.78 kpcover 0 . < z < .
03, therefore, our spectra and the corresponding spectralclassifications are done to the circumnuclear region of a typical disk galaxy. Thespectral classification of these galaxies are done with various emission lines inthe SDSS spectra. We process the spectra, measure the emission lines, andclassify the galaxies following Hao et al. (2005). In particular, we measure theemission lines after removing the stellar absorption using a set of well-developedabsorption-line templates. Galaxies with weak or no emission lines, definedspecifically by having EW(H α ) < α emission lines (FWHM(H α ) > α and [OIII]/H β ratios. Galaxies with pure stellar excitation are locatedin the left branch, and those with lower metallicities have higher [OIII]/H β butlower [NII]/H α ratios. The solid line, empirically defined by Kauffmann et al.(2003), separates the two branches. These authors classify galaxies below theline as star forming galaxies, and those above it as AGNs. The dot-dashed lineis taken from Kewley et al. (2001), and demarcates the maximum position thatcan be obtained by pure photo-ionization models. Galaxies located above theline require an additional excitation mechanism, such as an AGN, or strongshocks. The general classification scheme using the two separation lines consid-ers objects above Kewley’s line as AGN dominated sources, between Kewley’sand Kauffmann’s line as composite AGN and starburst sources, and below theKauffmann’s line as star forming galaxies.Since the locations of galaxies on the BPT diagram broadly reflect the prop-erties of their nuclear activities, such as the dominance of the AGN componentor the metallicity of the pure stellar-excited galaxies, we further divide galax-ies into spectral classes of 0 to 6 based on their locations on the diagram (asshown in Figure 1). We would like to investigate whether the optical bar fractionchanges with these nuclear properties. Galaxies with spectral classes of 0 to 3are broadly considered as star-forming galaxies, and 4 to 6 as AGNs. We assigninactive galaxies, which have little emission lines as spectral class of − r -band absolutemagnitude, the stellar mass, and the sersic index. We found no significantdifferences of the host galaxy properties between inactive galaxies and star-forming galaxies or active galaxies. Hao et al.
Figure 1. The BPT diagram for optically barred (upper) and opticallyunbarred (lower) galaxies in our sample. The solid line is taken fromKauffmann et al. (2003), and the dot-dashed line is taken from Kewley et al.(2001). We divide galaxies into seven spectral classes (indicated by dashed,solid, and dot-dashed lines) by their locations on the diagram, marked withthe numbers.
3. Results
In Figure 1, we find no clear differences between the distributions of barred andunbarred galaxies on the BPT diagram. Another way to look at it is shown inthe upper panel of Figure 2, where we plot the optical bar fraction of galaxieswith different spectral classifications ( − ∼ . ars in Starbursts and AGNs Figure 2. The optical bar fractions as a function of spectral classes definedin Figure 1 are shown in the upper panel. Galaxies with spectral class of − Our result suggests that AGNs have an excess optical bar fraction comparedwith the inactive galaxies, but show no excess compared with the starburst galax-ies. Therefore accurate and consistent spectroscopic classification of both theAGN sample and the control sample is important in evaluating the excess ofbars in AGNs. Many previous studies have overlooked this issue. Among threestudies (Ho et al. 1997; Hunt & Malkan 1999; Laurikainen et al. 2004b) wherewe can clearly decide the dominant spectral classes of the comparing sample,our result agrees with two of them. The comparing sample in Ho et al. (1997)is mainly composed of star-forming galaxies and they found no excess opticalbar fraction in AGNs, which agrees with our result. Based on the classificationin NED, Laurikainen et al. (2004b) divide galaxies into Seyferts, LINERs, star-bursts, and inactive galaxies. They found a similar NIR bar fraction for Seyfertgalaxies, LINERS, and HII/starburst galaxies at 72%, compared to 55% in non-active galaxies. The pattern also agrees with our result. The absolute values of
Hao et al. the NIR bar fraction in Laurikainen et al. (2004b) are higher than our opticalbar fraction. This could be due to two factors. Firstly, the NIR bar fraction isknown to be higher than the optical one by a factor of ∼ § .
14 in Barazza et al. (2008), asthey regard galaxies with twisted position angles, but otherwise bar-like featuresto be unbarred galaxies. These galaxies could be weakly barred galaxies.Our result however, disagrees with Hunt & Malkan (1999), who found thatthe Seyfert and LINERs in the Extended 12 µ m Galaxy Sample (E12GS) havean optical bar fraction of 68% and 61%, similar to the inactive galaxies in theE12GS at 69%. Star-forming galaxies in their sample have a higher optical barfraction (85%) than both inactive galaxies and Seyferts. Our disagreement couldbe due to two factors. Firstly, Hunt & Malkan (1999) used RC3, which is thevisual classification to identify bars. Therefore, their optical bar fractions arehigher than ours where bars are identified by ellipse fitting (see § Figure 3. The bar ellipticity of optically-visible bars as a function of thespectral classes defined in Figure 1. The big asterisks are the mean ellipticityof galaxies in each spectral class.
From the ellipse fitting, Barazza et al. (2008) also obtained the ellipticity ofoptically visible bars. In Figure 3, we show the bar ellipticity of barred galaxiesin different spectral classes. The value of the bar ellipticity varies widely for ars in Starbursts and AGNs
4. Conclusions
With the classification and structural information of ∼ ′′ aperture,which corresponds to 606 pc to 1.78 kpc in the redshift range of 0 . < z < .
03. This scale matches well with the typical circumnuclear region of a galaxy,therefore, the spectra are perfect at probing the circumnuclear stubursts.We find an excess of the optical bar fraction of star-forming galaxies andAGNs compared to the inactive galaxies. This suggests that large-scale primarybars drive gas to inner kpc where they pile up near the ILRs, fueling circumnu-clear starbursts. The gas pile up is in someway also related with the increase ofthe AGN activity. But to feed the AGN directly, the gas in the inner kpc still hasto reduce its angular momentum by several orders of magnitude (see Figure 3 inJogee 2006) and a secondary mechanism is then needed to drive the gas furtherin (e.g., nuclear bars, dynamical friction). The latter may not be necessarily cou-pled to the primary bar. This agrees with our result that we do not observe anexcess of bar fraction of AGNs compared to the star-forming galaxies. Further-more, we find no evidence of bar weakening by AGNs. This agrees with previoustheoretical expectations (Shen & Sellwood 2004; Athanassoula et al. 2005).There is one caveat of our study. Our bar identification is based on theoptical data instead of NIR, therefor our bar fraction is in general lower thanthe typical NIR bar fraction. However, we do not expect our comparative re-sults to change, unless the obscuration of bars by dust and star formation havepreferential effects.
References
Aguerri, J. A. L., M´endez-Abreu, J., & Corsini, E. M. 2009, A&A, 495, 491