Deficiency of Broad Line AGNs in Compact Groups of Galaxies
aa r X i v : . [ a s t r o - ph ] M a r Deficiency of Broad Line AGNs in Compact Groups of Galaxies
M. A. Mart´ınez and A. Del Olmo Instituto de Astrof´ısica de Andaluc´ıa, CSIC, Apdo. 3004, 18080 Granada, Spain [email protected], [email protected]
R. Coziol Departamento de Astronom´ıa, Universidad de Guanajuato, Apdo. 144, 36000 Guanajuato,Mexico [email protected] andP. Focardi Dipartimento di Astronomia, Universit`a di Bologna, via Ranzani 1, 40127 Bologna, Italy [email protected]
ABSTRACT
Based on a new survey of AGN activity in Compact Groups of Galaxies,we report a remarkable deficiency of Broad Line AGNs as compared to NarrowLine AGNs. The cause of such deficiency could be related to the average lowluminosity of AGNs in CGs: 10 erg s − . This result may imply lower accretionrates in CG AGNs, making Broad Line Regions (BLR) undetectable, or mayindicate a genuine absence of BLRs. Both phenomena are consistent with gasstripping through tidal interaction and dry mergers. Subject headings: galaxies: active — galaxies: Seyfert — galaxies: nuclei —galaxies: evolution
1. Introduction
From the optical spectra of AGNs, one can generally distinguish two main types: thosethat show broad emission lines (BLAGNs) and those that show only narrow emission lines 2 –(NLAGNs). With an absolute magnitude M V ≥ −
23, the local BLAGNs are called Seyfert 1(Sy1), while the NLAGNs are called Seyfert 2 (Sy2). In the literature, one can also findother types of Seyfert galaxies: Sy1.2, Sy1.5, Sy1.8, and Sy1.9, all of them being some sortof Sy1 and consequently BLAGNs. NLAGNs may also come into the form of Low IonizationNuclear Emission-line Regions (LINERs).Phenomenologically, it is unclear why AGNs come in different types. Based on spectralvariation, the Narrow Line Region (NLR) is thought to be located farther out from thecentral Black Hole accretion disk than the Broad Line Region (BLR), and to be spatiallymuch more extended. Within the unification model (Antonucci 1993), which assumes allAGNs to be intrinsically the same, the BLR in NLAGNs is hidden behind an opticallythick torus of gas and dust. Consistent with this model, many observations of NLAGNshave revealed hidden BLRs through polarized spectroscopy (Antonucci 2002). However, notall NLAGNs observed with this technique have shown such structures (Tran 2001, 2003;Laor 2003; Shu et al. 2007), suggesting that in some NLAGNs the BLR might simply beabsent. This last finding is consistent with alternative models in which the accretion rateand, consequently, AGN luminosity plays a direct role in determining the presence of theBLR (Nicastro 2000; Nicastro et al. 2003; Elitzur & Shlosman 2006).One possible way how to solve this dilemma is to explore the connection of AGNs withtheir environment. According to the unification model, for example, one does not expectto find any differences in the number of AGNs in different environments. Unfortunately,such studies are usually controversial. While some authors found Sy2 to inhabit richerenvironments than Sy1 (de Robertis et al. 1998), others claimed the opposite (Schmitt 2001,and references therein). Recently, Koulouridis et al. (2006) have found no differences in theenvironment of Sy1 and Sy2 over large scales ( . =70km s − Mpc − ). These results agree with Sorrentino et al. (2006) who found two times moreSy2 than Sy1 in local ( .
100 kpc) environment.To explore further the possible conection between AGN activity and environment wehave undertaken a new survey to determine the nature and frequency of nuclear activ-ity in two different samples of Compact Groups of galaxies (CGs): the Hickson CompactGroups (Hickson 1982, HCGs) and a sample of CGs from the Updated Zwicky Catalog(Focardi & Kelm 2002, UZC-CG). Previous studies on CGs revealed a high percentage oflow luminosity AGNs in these structures, but very few luminous ones (Coziol et al. 1998,2000; Mart´ınez et al. 2006, 2007). Having in hand a statistically significant sample of CGswith complete information on the nuclear activity of their members, allows us to betterquantify the frequency of BLAGNs in these systems. 3 –
2. Data and Results
Among the HCGs, we have selected the groups with redshifts z ≤ . µ G ≤ .
4. These criteria provided us with a statistically complete sample of 283galaxies in 65 groups. We have obtained medium resolution spectroscopy for 238 of thesegalaxies. The spectra of 71 galaxies come from previous observations made by Coziol et al.(1998, 2000, 2004). The remaining 167 galaxies were observed by our group using fourdifferent telescopes: the 2.5m NOT in “El Roque de los Muchachos” (RM, Spain), the 2.2min Calar Alto (CAHA , Spain), the 2.12m in San Pedro M´artir (SPM, Mexico) and the 1.5mtelescope in Sierra Nevada Observatory (OSN, Spain).For all the galaxies, broad components search and activity classification were done onlyafter template subtraction, to correct for absorption features produced by the underlyingstellar population. Detailed of the template subtraction method used can be found inCoziol et al. (1998, 2004). Preliminary results have already been published in Mart´ınez et al.(2007). Complete description of observations, reduction and analysis method will be pub-lished elsewhere.For the UZC-CG sample, we have collected spectra from three spectroscopic archives:the Sloan Digital Sky Survey (SDSS-DR4), the Z-Machine and the FAST SpectrographArchives. We have found spectra for all the galaxy members of 215 groups (720 galaxies):210 spectra are from the SDSS, 187 from FAST and 323 from Z-Machine (Mart´ınez et al.2006). Because the Z-Machine spectra have too low S/N ratios to measure broad components,we restrict our analysis to the 397 spectra found in the SDSS and FAST databases. Spectrafrom the SDSS survey were template subtracted using Hao et al. (2005, H05) eigenspectraand their PCA method. No correction was applied to the galaxies in the FAST sample, dueto the non availability of suitable spectra to be used as templates.Emission lines were found in 153 of the 238 galaxies in the HCG sample (64%) and 274of the 397 galaxies (69%) in the UZC-CG sample. The identification of broad componentswas done by fitting a multi-components Gaussian on the emission lines, using the IRAF taskNGAUSSFIT. The FWHM of [SII] (or [OIII] when the [SII] lines were too faint or noisy)have been used to model the narrow components of the H α and [NII] lines. When a broadfourth component was necessary, it was centered on H α . A χ criterion, as described in H05, ALFOSC is owned by the IAA and operated at the Nordic Optical Telescope (NOT), under agreementbetween IAA and the NBIfAFG of the Astronomical Observatory of Copenhagen. The Centro Astron´omico Hispano Alem´an is operated jointly by the Max-Planck Institut fur Astronomieand the IAA-CSIC. ≥
500 kms − . Our analysis revealed only 1 BLAGN in the HCG sample and 8 in the UZC-CG sample.For each of these galaxies we give, in Table 1 the FWHM of the broad component andactivity classification according to Osterbrock (1989): a Sy1.9 shows a broad componentonly in H α , while a Sy1.8 shows also a weak broad component in H β . None of our BLAGNsare classified as Sy1, Sy1.2 or Sy1.5. Based on our analysis, BLAGNs represent only 1%(1/153) of emission line galaxies in the HCG sample and 3% (8/274) in the UZC-CG one.To classify NLAGNs we used the diagnostic diagram based on the four most intenseemission lines: H β , [OIII]5007˚A , H α , and [NII]6583˚A and criteria similar to those employedby Kewley et al. (2006). Galaxies located above the theoretical maximum sequence for starformation are classified as AGNs. We also distinguished between Sy2 and LINERs using theclassical upper limit log([OIII]5007˚A/H β ) < α , we classified these cases as Low Luminosity AGNs(LLAGNs) when log([NII]/H α ) > − . The fraction of BLAGNs over NLAGNs in our two CG samples is extremely low: 1%for the HCGs and 6% for the UZC-CGs. Also noticeably low are the ratios of Sy1 to Sy2:4% in the HCG and 19% in the UZC-CG. To realize how low these ratios are one has tocompare with what is usually found in other surveys.The mean H α luminosity of both NLAGNs and BLAGNs in our two samples is about10 erg s − , which is typical of the faint end of the luminosity function of AGNs. Thisvalue is comparable with the mean H α luminosity of AGNs observed in the local universeby Ho et al. (1997a, HFS97). Except for some galaxies in Virgo, all the galaxies in theHFS97 sample are located in low density environments (either loose groups or isolated). InTable 2 we compare their results (395 galaxies, excluding the galaxies in Virgo) with ours. Inthe HFS97 sample, the BLAGNs were classified as such by Ho et al. (1997b), based on the 5 –detection of a broad H α component. To be consistent with our definition, all these galaxieswere classified as Sy1. Also for comparison sake, the narrow emission lines galaxies in theHFS97 sample were reclassified using the criteria described in sect. 2.The fraction BLAGN/NLAGN in the HFS97 sample is 22% and the ratio Sy1/Sy2is 61%. There is consequently a clear deficiency of BLAGNs in CGs. This also appearsas an extremely large difference in the number of Sy1 as compared to Sy2 galaxies. Thisphenomenon is quite intriguing considering that there is no deficit of AGNs as a whole inCGs: 46% AGNs in the HCG, 51% in the UZC-CG compared to 44% in the HFS97 sample.Comparable ratios (Sy1/Sy2 ∽ ∽ <
3. Discussion3.1. Quantifying biases and detection limits
Our results suggest there is an important deficit of BLAGNs in our two CG samples ascompared to similar surveys in the field. This result confirm the tendency first encounteredby Coziol et al. (1998, 2000). To verify that the lack of BLAGNs in CGs can not be inducedby differences in observation, reduction or analysis methods we have investigated thoroughlythese possibilities.Comparison of the UZC-CG sample with the survey made by H05 is safe, becauseour SDSS data derive from the same telescope, reduction and analysis methods (includingtemplate subtraction) as theirs. The ratio BLAGN/NLAGN is 8% in our sample comparedto 43% for the sample of H05, which is already a huge difference.A possible effect due to difference in spectral resolution can also be excluded. Ho et al.(1997b) have used high resolution (2.5 ˚A) spectra, but made tests with two others lowresolution set-ups (5˚A and 10˚A) obtaining similar results. These are comparable to our ownobservations: CAHA and OSN (4˚A), NOT and SPM (8˚A), SDSS (3.5˚A) and FAST (6˚A).Both H05 and SRR06 have 3.5˚A.The S/N continuum levels of the different surveys are also comparable. On average theS/N of AGNs in our spectra is 60 with a maximum of the order of 120. This is comparable 6 –to H05 and SRR06 spectra (they both used SDSS). HFS97 did not published their values.However their BLAGNs rates are comparable to those of H05 and SRR06, suggesting this isnot an issue.There is no evidence either for a higher galaxy contamination (the amount of galaxyfalling into the aperture) in our samples. Taking into account the slit aperture and distancesof the host galaxies in each sample we find medians of 1 kpc and 1.3 kpc for the HCG andUZC-CG, respectively. Although the median for HFS97 is lower (0.5 kpc) than for H05 (7kpc) the results are similar. Obviously, template subtraction (like we also did) alleviates thedifferences. We may note also that no relation is observed, in any of these surveys (includingours), between the frequency of BLAGNs and the redshift of the galaxies where they arefound, which means that nearby galaxies are not more likely BLAGNs than remote ones.To test if our low number of Sy1 could be due to a difference in morphologies (Schmitt2001), we have divided our two samples and the HFS97 one in three morphology classes:E for early-type galaxies (E-S0), Se for early-type spirals (S0a-Sbc), and Sl for late-typespirals (Sc and later). For homogeneity sake, all the morphologies have been taken from theHyperleda database (Paturel et al. 2003). In Table 3 we give for each morphology class thefraction of galaxy and the ratios BLAGN/NLAGN and Sy1/Sy2. There are no BLAGNsin late-type spirals in any sample. In the HFS97 sample, the ratio of BLAGN/NLAGNis marginally higher in the E class while the ratio Sy1/Sy2 is significantly higher, whichindicates a definitive increase in BLAGNs in early-type galaxies. In the two CG samples wealmost see an inverse trend: the ratios of BLAGN/NLAGN and Sy1/Sy2 are both larger inthe Se class than in the E one. Moreover there is a definite rise in the number of early-typegalaxies in CGs. Following the HFS97 trend, this should have produced more BLAGNs inCGs instead of less. This eliminates a difference in morphologies as a possible explanation.We also reject the hypothesis of lower sensitivity. Comparing the median luminosity inH α of the different types of galaxies in our samples with those in the HFS97 sample, lowersensitivity would have translated into higher values in our samples. This is not observed. Inthe HFS97 sample the median H α luminosity of the NLAGNs is log(L Hα = 38 .
72 ergs s − ).Our values are comparable: 38.69 for the HCG and 38.79 for the UZC-CG.Finally we have determined the detection limits in our samples as in Ho et al. (1997b).Different simulations were performed adding to each set-up spectra a grid of synthetic spectrawith broad gaussian components of various widths and amplitudes centered on H α . We thenapplied our template and extraction analysis to deduce the following limits. For the CAHAand OSN spectra, broad components equivalent to 15% or higher of the total blended flux inH α +[NII] were recovered. Using medians of AGN blended flux and redshfit this transformsinto a detection limit in H α broad luminosity of 3 . × ergs s − . We find slightly higher 7 –fraction (20%) for the NOT and SPM spectra, equivalent to a detection limit in luminosityof 4 . × ergs s − . Only three BLAGNs in the HFS97 sample have a luminosity lowerthan these limits. Obviously, the lack of BLAGNs encountered in our samples cannot beexplain by a higher detection limits in our samples.There seem to be no obvious observational biases or differences in reduction and analysismethods capable of reproducing the lack of BLAGNs in CGs as compared to lower densityenvironments. It is consequently reasonable to conclude that this phenomenon must berelated to the environment typical of CGs. In the unification model for AGNs, a torus of matter is assumed to be responsible forhiding the BLR from the observer. To be consistent with our analysis, this mechanismshould be much more efficient in CGs. However, this assumption goes against the evidenceof tidal stripping: in CGs the infall of gas in the disk seem to be stopped, generally diminish-ing the amount of star formation (Coziol et al. 1998, 2000; Verdes-Montenegro et al. 2001;de la Rosa et al. 2007; Durbala et al. 2008). At the same time, the number of early-typegalaxies in CGs is observed to increase. Therefore, a possible reason why no BLRs appear inAGNs in CGs may be because any amount of gas that has reached the center was consumedinto stars, building larger bulges (de Carvalho & Coziol 1999). Alternatively, the bulges ofgalaxies in CGs may have grown without gas, through dry mergers (Coziol & Plauchu-Frayn2007).The fact that the average luminosity of the AGNs in CGs is low is another argumentin favor of the dissolution hypothesis for the BLR. According to recent results obtained byreverberation mapping, the size of the BLR in AGNs is correlated to the optical luminosityat 5100˚A (Kaspi et al. 2005). It is consequently possible to imagine the size of the BLRshrinking almost to zero at some low threshold luminosity (Elitzur & Shlosman 2006). Theluminosity at 5100˚A in our samples range from log( λ L λ (5100˚A)) 40.7 to 43.1 (in units of ergs − ); comparing with data of Peterson et al. (2004) we are in the lower luminosity part oftheir distribution, where few objects with broad lines have been observed. We also are atthe lower limit where no hidden BLRs have been found (Shu et al. 2007; Bian & Gu 2007).Using the relation log(L bol /L Edd ), most of our galaxies are below -1.37, which suggests thatbroad features may simply not exist in these LLAGNs.According to Nicastro (2000) and Nicastro et al. (2003) low accretion rates rather thansmaller mass black holes are responsible in explaining the absence of BLRs in Low Luminosity 8 –AGNs which is fully compatible with our observations.
4. Conclusion
REFERENCES
Antonucci, R. 1993, ARA&A, 31, 473Antonucci, R. 2002, Astrophysical Spectropolarimetry, (Cambridge, CUP), 151 (astro-ph/010348)Bian, W., & Gu, Q. 2007, ApJ, 657, 159Caon, N., Capaccioli, M., D’Onofrio, M., & Longo, G. 1994, A&A, 286, 39 9 –Coziol, R., Ribeiro, A. L. B., Capelato, H. V., & de Carvalho, R. R. 1998, ApJ, 493, 563Coziol, R., Iovino, A., & de Carvalho, R. R. 2000, AJ, 120, 47Coziol, R., Brinks, E., & Bravo-Alfaro, H. 2004, AJ, 128, 68Coziol, R., & Plauchu-Frayn 2007, AJ, 133, 2630de Carvalho, R. R., & Coziol, R. 1999, AJ, 117, 1657de la Rosa, I. G., de Carvalho, R. R., Vazdekis, A., & Barbuy, B 2007, AJ, 133, 330de Robertis, M. M., Yee, H. K. C., & Hayhoe, K. 1998, ApJ, 496, 93Durbala, A., et al. 2008, AJ, 135, 130Elitzur, M., & Shlosman, I. 2006, ApJ, 648, L101Focardi, P., & Kelm, B. 2002, A&A, 391, 35Hao, L., et al. 2005, AJ, 129, 1783 (H05)Hickson, P. 1982, ApJ, 255, 382Hickson, P., Kindl, E., & Huchra J. P. 1988, ApJ, 331, 64Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1997a, ApJS, 112, 315 (HFS97)Ho, L. C., Filippenko, A. V., Sargent, W. L. W., & Peng, C. Y. 1997b, ApJS, 112, 414Kaspi, S., Maoz, D., Netzer, H., Peterson, B. M., Vestergaard, M., & Jannuzi, B. T. 2005,ApJ, 629, 61Kewley, L. J., Groves, B., Kauffmann, G., & Heckman, T. 2006, MNRAS, 372, 961Koulouridis, E., Plionis, M., Chavushyan, V., Dultzin-Hacyan,D., Krongold, Y., & Goudis,C. 2006, ApJ, 639, 37Laor, A. 2003, ApJ, 590, 86Mart´ınez, M. A., del Olmo, A., Focardi, P., & Perea J. 2006, VII Scientific Meeting of SEA(astro-ph/0611099)Mart´ınez, M. A., del Olmo, A., Perea, J., & Coziol, R. 2007, ESO Astro. Symp., Springer-Verlag, 163 10 –Mendes de Oliveira, C., & Hickson, P. 1994, ApJ, 427, 684Nicastro, F. 2000, ApJ, 530, L65Nicastro, F., Martocchia, A., & Matt, G. 2003, ApJ, 589, L13Osterbrock, D. E. 1989, in Astrophysics of Gaseous Nebulae and Active Galactic Nuclei,University Science BooksPaturel, G., et al. 2003, A&A, 412, 45Peterson, B. M., et al. 2004, ApJ, 613, 682Peterson, B. M., et al. 2005, ApJ, 632, 799Shu, X. W., Wang, J. X., Jiang, P., Fan, L. L., & Wang, T. G. 2007, ApJ, 657, 167Schmitt, H. R. 2001, AJ, 122, 2243Sorrentino, G., Radovich, M., & Rifatto, A. 2006, A&A, 451, 809 (SRR06)Stasi´nska, G., Cid Fernandes, R., Mateus, A., Sodr´e, L., & Asari, N. V. 2006, MNRAS, 371,972Tran, H. D. 2001, ApJ, L554, 19Tran, H. D. 2003, ApJ, 583, 632Verdes-Montenegro et al. 2001,A&A, 377, 812
This preprint was prepared with the AAS L A TEX macros v5.2.
11 –Table 1. BLAGNs identificationName Source Type FWHM(H α ) FWHM(H β )(km/s) (km/s)HCG 5a CAHA 1.9 1056 · · · UZC-CG 84c SDSS 1.9 2727 · · ·
UZC-CG 89b SDSS 1.9 2159 · · ·
UZC-CG 109b SDSS 1.8 1902 1499UZC-CG 117a SDSS 1.9 2351 · · ·
UZC-CG 132b FAST 1.9 3055 · · ·
UZC-CG 139b SDSS 1.9 1941 · · ·
UZC-CG 232c SDSS 1.8 2258 1689UZC-CG 234b FAST 1.9 1328 · · ·
12 –Table 2. Nuclear classification for the AGN galaxiesSample NLAGN BLAGN
BLAGNNLAGN Sy1Sy2
Sy2 LINER LLAGN % %HCG 28 23 19 1 1 4UZC-CG 43 11 79 8 6 19HFS97 46 80 · · ·
28 22 61H05 2424 650 · · · · · · · · · · · ·
66 13 –Table 3. Activity types and morphological distributionSample E-S0 S0a-Sbc Sc-Irrf E BLAGNNLAGN Sy1Sy2 f Se BLAGNNLAGN Sy1Sy2 f Sl HCG 54% · · · · · ·
32% 3% 10% 14%UZC 34% 2% 8% 55% 9% 25% 11%HFS97 25% 25% 86% 39% 19% 56% 36% 14 –Fig. 1.— Examples of BLAGNs in our sample with multi-components Gaussian decomposi-tions centered on H αα