Mon. Not. R. Astron. Soc.

000, 000000 (0000)

Printed 23 May 2006

A (MN L TEX style le v2.2)

Stellar and gaseous velocity dispersions in type II AGNs at 0.3 < z < 0.83 from the Sloan Digital Sky Survey
W.Bian1,2, Q. Gu3 , Y. Zhao3, L. Chao1 and Q. Cui1 1
2 Key

Department of Physics and Institute of Theoretical Physics, Nanjing Normal University, Nanjing 210097, China Laboratory for Particle Astrophysics, Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100039, China 3 Department of Astronomy, Nanjing University, Nanjing 210093, China

23 May 2006

ABSTRACT

We apply the stellar population synthesis code by Cid Fernandes et al. to model the stellar contribution for a sample of 209 type II AGNs at redshifts 0.3 < z < 0.83 from the Sloan Digital Sky Survey. The reliable stellar velocity dispersions ( ) are obtained for 33 type II AGNs with signicant stellar absorption features. According to the L[OIII] criterion of 3 108 L , 20 of which can be classied as type II quasars. We use the formula of Greene & Ho to obtain the corrected stellar velocity dispersions c c ( ). We also calculate the supermassive black holes masses from in these higherredshift type II AGNs. The [O III] luminosity is correlated with the black hole mass, and no correlation is found between the [O III] luminosity and the Eddington ratio. Three sets of two-component proles are used to t multiple emission transitions ([O III]4959, 5007 and [O II]3727, 3729) in these 33 stellar-light subtracted spectra. We also measure the gas velocity dispersion (g ) from these multiple transitions, and c nd that g can trace (although with considerable scatter), which conrms that the gaseous kinematics of narrow line regions in these type II quasars are primarily c dominated by the gravitational potential of the bulge. The distribution of < g / > is 1.24 0.76 for the core [O III] line and 1.20 0.96 for the [O II] line, which suggests c that g of the core [O III] and [O II] lines can trace within about 0.1 dex in the c c logarithm of . For the secondary driver, we nd that the deviation of g from is correlated with the Eddington ratio. Key words: galaxies:active galaxies:nuclei quasars: emission lines

INTRODUCTION

Recent advances on the study of normal galaxies and active galactic nuclei (AGNs) are that we found more evidence for the existence of central supermassive black holes (SMBHs) and the relationship between SMBHs and bulge properties of host galaxies (Gebhardt et al. 2000; Ferrarese & Merritt 2000; Tremaine et al. 2002; Begelman 2003; Shen et al. 2005). We can use the stellar and/or gaseous dynamics to derive the SMBHs masses in nearby inactive galaxies. However, it is much more dicult for the case of AGNs. With the broad emission lines from broad-line regions (BLRs)(e.g. H, Mg II, CIV; H), the reverberation mapping method and the empirical size-luminosity relation can be used to derive the virial SMBHs masses in AGNs (Kaspi et al. 2000; Vestergaard 2002; McLure & Jarvis 2002; Wu et al. 2004; Greene & Ho 2006a). It has been found that nearby galaxies and AGNs follow the same strong correlation between the central SMBHs masses (MBH ) and stellar bulge velocity dispersion ( ) (the MBH relation) (Nelson et al. 2001;
c 0000 RAS

Tremaine et al. 2002; Greene & Ho 2006a, 2006b), which also implied that the mass from reverberation mapping method is reliable. According to the unication model of active galactic nuclei (e.g. Antonucci 1993; Urry & Padovani 1995), AGNs can be classied into two classes depending on whether the central engine and BLRs are viewed directly (type I AGNs) or are obscured by circumnuclear medium (type II AGNs). In type I AGNs, by using the broad emission lines from BLRs (the reverberation mapping method or the empirical sizeluminosity relation), we can derive virial SMBHs masses. It is not easy to study their host galaxies because their optical spectra are dominated by the non-stellar emission. This is especially true for luminous AGNs, where the continuum radiation from central source outshines the stellar light from the host galaxy. In type II AGNs, the obscuration of BLRs makes both the reverberation mapping method and the empirical sizeluminosity relation impossible to derive SMBHs masses.

W. Bian, Q. Gu, Y. Zhao, L. Chao, Q. Cui

per resolution element. The bers in the SDSS spectroscopic survey have a diameter of 3 on the sky, for our Type II AGNs sample at redshifts 0.3 < z < 0.83, the projected bre aperture diameter typically contains about 90% of the total host galaxy light (Kaumann & Heckman 2005), and thus makes it feasible to observe signicant stellar absorption features, which is the key point to accurately measure the stellar velocity dispersion ( ) in this paper. We rst model the stellar contribution in the SDSS spectra of type II AGNs through the modied version of the stellar population synthesis code, STARLIGHT (version 2.0, Cid Fernandes et al. 2001; Cid Fernandes et al. 2004a; Cid Fernandes et al. 2004b; Garcia-Rissman et al. 2005), which adopted the new stellar library from Bruzual & Charlot (2003). The code does a search for the linear combination of Simple Stellar Populations (SSP) to match a given observed spectrum O . The model spectrum M is:
N

However, we can use the well-known MBH relation to derive SMBHs masses if we can accurately measure the stellar bulge velocity dispersion ( ). There are mainly two methods to measure , one is in Fourier space (e.g. Tonry & Davis 1979), the other is in pixel space (e.g. Greene & Ho 2006b and reference therein). These years it is popular to use the combination of galaxy template spectra broadened by a Gaussian kernel (e.g. Kaumann et al. 2003; Cid Fernandes et al. 2004a; Greene & Ho 2006b). Though it is still not an easy task to measure , it has been shown successfully to derive through tting stellar absorption features: such as Ca II 8498, 8542, 8662 triplet, Mg Ib 5167, 5173, 5184 triplet, and Ca H+K 3969, 3934, etc. On the other hand, Nelson & Whittle (1996) nd that the gaseous velocity dispersion (g ) of [O III]5007 from the narrow-line regions (NLRs) is nearly the same as for a sample of 66 Seyfert galaxies, and suggest that the gaseous kinematics of NLRs be primarily governed by the bulge gravitational potential. Nelson (2000) nd a tight relation between MBH and [OIII] (the [O III]5007 velocity dispersion) for AGNs, very similar to the relation of MBH , although with more scatter, which strongly suggests that g can be used as a proxy for . For lower-redshift type II AGNs with 0.02 < z < 0.3, Kaumann et al. (2003) have investigated the properties of their hosts from the Sloan Digital Sky Survey (SDSS) Data Release One (DR1), measured and estimated the SBMHs masses from (Brinchmann et al. 2004). By using this sample, Greene & Ho (2005) measured the gaseous velocity dispersion (g ) from multiple transitions ([O II] 3727, [O III] 5007, and [S II] 6716, 6731) and compared and g . They found that g from these multiple transitions trace very well, although for some emission features showing considerable scatter. Type II quasars are the luminous analogs of lowluminosity type II (Seyfert 2) galaxies. The obscuration of BLRs makes quasars appear to be type II quasars (obscured quasars). Some methods have been used to discover type II quasars, but only a handful have been found. Recently, Zakamsa et al. (2003) present a sample of 291 type II AGNs at redshifts 0.3 < Z < 0.83 from the SDSS spectroscopic data. About half are type II quasars if we use the [O III] 5007 line luminosity to represent the strength of the nuclear activity. How about the g relation for type II quasars? And how about their SMBHs masses and the Eddington ratios Lbol /LEdd (i.e. the bolometric luminosity as a fraction of the Eddington luminosity)? Here we used the sample of Zakamsa et al. (2003) to study these questions in type II quasars. In section 2, we introduce the data and the analysis. Our results and discussion are given in Sec. 3. All of the cosmological calculations in this paper assume H0 = 70 km s1 Mpc1 , M = 0.3, = 0.7.

M (x, M0 , AV , v , ) = M0
j=1

xj bj, r G(v , )(1)

DATA AND ANALYSIS

With SDSS, Zakamsa et al. (2003) present a sample of 291 type II AGNs at redshifts 0.3 < z < 0.83. We download these spectra from SDSS Data Release Four (DR4) and the spectra for 202 type II AGNs at redshifts 0.3 < z < 0.83 are obtained. SDSS spectra cover 3800-9200 , with resoluA tion (/) of 1800 < R < 2100 and sampling of 2.4 pixels

where bj, LSSP (tj , Zj )/LSSP (tj , Zj ) is the spectrum of 0 the j th SSP normalized at 0 , r 100.4(A A0 ) is the reddening term, x is the population vector, M0 is the synthetic ux at the normalization wavelength, G(v , ) is the line-of-sight stellar velocity distribution, modeled as a Gaussian centered at velocity v and broadened by . The match between model and observed spectra is calculated by N 2 (x, M0 , AV , v , ) = =1 [(O M ) w ]2 , where the weight spectrum w is dened as the inverse of the noise in O . For more detail, please refer to Cid Fernandes et al. (2005). Prior to the synthesis, the Galactic extinction is corrected by using the extinction law of Cardelli, Clayton & Mathis (1989) and the AV value from Schlegel, Finkbeiner & Davis (1998) as listed in the NASA/IPAC Extragalactic Database (NED). The spectra are transformed into the rest frame dened by the redshift given in their FITS header. The spectrum is normalized at 4020 and the signal-to-noise raA tio is measured in the S/N window between 4730 and 4780 . Masks of 20 30 around obvious emission lines are A A constructed for each object individually. Because the redshift coverage of this type II AGNs sample, we focus on the strongest stellar absorption features of Ca II K and the Gband, which are less aected by nearby emission lines. An f 1.5 power-law component is used to account for the contribution from the AGN continuum emission (Watanabe et al. 2003). Finally we check visually our spectral tting results one by one. For our sample, the S/N in the S/N window varies between 0.3 and 21.5. The tting results for high S/N objects are usually better than those for low S/N ones. After inspecting the tting results, we nd that the tting goodness depends not only on the S/N (> 5), but also on the absorption lines equivalent widths (EWs > 1.5). At last we A select 33 type II AGNs, which are shown signicant stellar absorption features and are well tted to derive reliable measurement of stellar velocity dispersion . The [O III]5007 luminosity (L[OIII] ) is subject to extinction by interstellar dust in the host galaxy and in our Galaxy, which is usually corrected by using the Balmer decrement. However, for most objects in the sample of Zac 0000 RAS, MNRAS 000, 000000

3
kamsa et al. (2003), H cannot be showed in SDSS spectra. Therefore the extinction is not corrected. In order to check that these sub-sample is representative of total sample of Zakamsa et al. (2003) respect to [O III]5007 luminosity (L[OIII] ), we plotted the histograms of the L[OIII] distribution for sub-sample and total-sample (see gure 1). And then we used the T-test and found that at the 0.95 level, the difference of these two population is not signicantly dierent. L[OIII] is directly adopted from table 1 in Zakamsa et al. (2003). Using the L[OIII] criterion of 3108 L (the logarithm is 8.477), which corresponds to the intrinsic absolute magnitudes MB < 23 (Zakamsa et al. 2003), 20 objects can be classied as type II quasars. Fig. 1 shows a sample tting for SDSS J150117.96+545518.2 with S/N=20.5. The nal results are presented in table 1. All the ttings for 33 type II AGNs are appended in the appendix. After subtracting the synthetic stellar components and the AGNs continuum, we obtain the clean pure emissionline spectra as shown in the top panel of gure 2, where we can analysis the pure emission-line proles in detail by using the multi-component spectral tting task SPECFIT (Kriss 1994) in the IRAF-STSDAS package 1 . Because of the asymmetry of proles of [O III]4959, 5007 lines, two sets of two-gaussian proles are used in order to remove properly the asymmetric blue/red wings of [O III] line. We take the same linewidth for each component, and x the ux ratio of [O III]4959 to [O III]5007 to be 1:3. For three objects, we cant t the [O III] lines for their irregular [O III] lines (See table 1). For the [O II] 3727, 3729 lines, we use two-gaussian proles and x their wavelength separation to the laboratory value, the ratio of the line intensities is allowed to vary during the tting. The decomposition for [O II] lines is more dicult because of relatively low S/N and that the expected line widths are comparable to the pair separation (2.4). We do not t [S II] 6716, 6731 for our A larger redshifts of our sample. For more details, please refer to Bian, Yuan & Zhao (2005, 2006). Our sample tting for SDSS J150117.96+545518.2 is showed in gure 3. is consistent very well with that of the MPA/JHU group (Kaumann et al. 2003). The median of the dierence between the two estimates is just 9 km s1 . We have carefully checked our synthesis tting result one by one and picked out 33 type II AGNs that are well tted and the stellar velocity dispersion ( ) are reliably derived. The spectral S/N for these objects are in the range of 5 to 21.5, most of which are larger than 10, thus the typical uncertainty in should be around 20 km s1 . Recent, Greene & Ho (2006a, 2006b) used the directtting method (Barth et al. 2002) to study the the systematic biases of form dierent regions around Ca II triplet, Mg Ib triplet, and Ca H+K stellar absorption features, which are introduced by both template mismatch and contamination from AGNs. They argue that the Ca II triplet provides the most reliable measurements of and there is a systematic oset between CaK and derived from other spectral regions. For our higher-redhsift sample and the SDSS wavelength coverage 3800-9200, it is impossiA ble to measure from Ca II triplet. Therefore, for higherredhsift type II AGNs, new observation around Ca II triplet is necessary in the future. Here we used the following forc mula to obtain the corrected velocity dispersion (Greene & Ho 2006b),
c = (1.40 0.04) (71 5).

(2)

3 3.1

RESULTS AND DISCUSSION The uncertainties of stellar velocity dispersion ( ) and gaseous velocity dispersion (g )

The derived stellar velocity dispersion was corrected by the instrumental resolutions of both the SDSS spectra and the STELIB library. Cid Fernandes et al. (2004a) have used their stellar population synthesis method to study a sample of 79 nearby galaxies observable from the southern hemisphere, of which 65 are Seyfert 2 galaxies. The S/N in the S/N window varies between 10 and 67. They compared their with values from the literature and found the agreement is good. And they estimated that the uncertainty in is typically about 20 km s1 . Recently, Cid Fernandes et al. (2005) apply their synthesis method to a larger sample of 50362 galaxies from the SDSS Data Release 2 (DR2). Their derived
1

For three objects, is near the instrumental resolution and c the corrected is unliable. These objects are omitted in our next analysis, which are denoted as in table 1. The gaseous velocity dispersion (g ) is obtained from full width half maximum (FWHM) of emisg sion lines by assuming the Gaussian prole: [OII] = g FWHM[OII]/2.35 and [OIII] = FWHM[OIII]/2.35. Considering the SDSS spectrum resolution, the intrinsic g derived from FWHM([OIII]) may be instrumentally broadened. The 2 intrinsic g value can be approximated by g = (obs 2 1/2 (inst /(1 + z)) ) , where z is the redshift. For the spectra from SDSS, the mean values of inst are 74 km s1 (the logarithm is 1.87 dex) for [O II], and 60 km s1 (the logarithm is 1.78 dex) for [O III] (Greene & Ho 2005), respectively. The results of g are listed in table 1 (Columns 7 and Columns 8). After removing the eect of instrumental broadening, some objects become unresolved (see table 1). Measurements of g below the resolution limit (74 and 60 km s1 for [O II] and [O III], respectively) are not reliable. The error of g is derived from the error of the linewidth. For the linewidth, the typical error is about 10 per cent. However, the systematic eects are neglected, e.g. the uncertainties of the continuum subtraction, and component decomposition (Bian, Yuan & Zhao 2005).

3.2

The SMBHs masses and Eddington ratios in type II quasars

IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. c 0000 RAS, MNRAS 000, 000000

Using the reverberation mapping method or the empirical size-luminosity relation, it is impossible to estimate the SMBHs masses in type II quasars for the lack of emission line from BLRs. Here we use the formulae to derive the SMBHs masses in type II quasars from stellar velocity dispersion (Tremaine et al. 2002), which is MBH ( M ) = c 108.13 ( /(200 kms1 ))4.02 .

W. Bian, Q. Gu, Y. Zhao, L. Chao, Q. Cui

the higher-redshift narrow-line AGNs follow the same trend as the lower-redshift ones, where the gas kinematics in NLRs is dominated by gravitational potential of the bulge and the velocity dispersion in the gas and stars are comparable. In order to qualify the comparison between g and c c , we calculate the distribution of < g / >. The value is 1.24 0.76 for the the core component of [O III] line, 1.20 0.96 for the [O II] line, which suggest that g of the c the core component of [O III] and [O II] lines can trace c within 0.09 and 0.08 dex in the logarithm of , respectively, c and that the high-ionization [O III] line traces as well as the low-ionization [O II] line. If we use the line width of [O III] core component to estimate the SMBHs masses from M[OIII] = 108.13 ([OIII] /(200kms1 ))4.02 M , we would overestimate SMBHs masses by 0.38 dex. For comparison, Greene & Ho (2005) derived the distribution of < g / > of lower-redshift type II AGNs with 0.02 < z < 0.3, which are 1.34 0.66 for the [O III] line, 1.000.35 for the [O III] core line, and 1.130.38 for the [O c II] line. We also calculate the the distribution of < g / > for the sample of Seyfert galaxies (Nelson & Whittle 1996), the value is 1.15 0.68, suggesting that g of the [O III] line c c can trace within 0.06 dex in the logarithm of . Our results are thus consistent with theirs. To rst order, the line widths of the core component of [O III] and [O II] for both low-redshift and high-redshift Type II AGNs is primarily controlled by the gravitational potential of the bulges of host galaxies, and can proximately trace the stellar velocity dispersion. As we know, the errors in virial SMBHs masses derived from galaxy dynamics or size-luminosity relation is about 0.5 dex (e.g. Magorrian et al. 1998; Kaspi et al. 2000; Bian & Zhao 2004a, 2004b). We also can use the line-width of the core component of [O III] or [O II] to estimate the black hole mass. In order to nd the secondary eect of the line broadening in gas lines from NLRs for our higher-redshift narrowline AGNs, we study the relation between the deviation of c c g from ( = logg log ) and the Eddington ratio (Lbol /LEdd ) (Greene & Ho 2005). Using the least-squares regression, we nd a median strong correlation between these two dimensionless parameters. For the [O III] line, the relation is: = (0.09 0.04) + (0.27 0.04)log(Lbol /LEdd ). The Spearman rank correlation coecient R is 0.77, with a probability of P < 104 for rejecting the null hypothesis of no correlation. For the [O II] line, the relation is = (0.11 0.04) + (0.22 0.05)log(Lbol /LEdd ), R=0.68, Pnull = 0.00181 (see gure 7). We also nd the comparable correlation between and Mbh . These results conrm that the nuclei accretion process and/or nuclei SMBHs would effect the line width of gas lines from NLRs, although the primary driver is from the gravitational potential of the bulge.

We also calculate the Eddington ratio, Lbol /LEdd . We use [O III] luminosity as a surrogate for the AGN luminosity (Heckman et al., 2004; Greene & Ho 2005), Lbol = 3500L[OIII] , to calculate the bolometric luminosity Lbol , and LEdd = 1.261038 MBH / M ergs s1 . The results of SMBHs masses and Eddington ratios in type II AGNs are also presented in table 1 (Columns 12 and 13). We also calculated the SMBHs masses and Eddington ratios for the lowerredshifts type II AGNs at 0.02 < z < 0.3 presented by Kaumann et al. (2003). For the typical uncertainties of 20km s1 for = 200 km s1 , the errors of log would be about 0.05 dex, corresponding 0.17 dex for logMBH , and almost the same for Lbol /LEdd (Bian, Yuan, & Zhao 2005). Here we didnt consider the cosmology evolution of MBH relation (e.g. Woo et al. 2006), which is a question open to debate. In gure 4, we compared the distribution of SMBHs masses and Eddington ratios in the lower-redshifts sample and the higher-redshifts sample. It is found that the type II AGNs at higher-redshifts have higher SMBHs masses and higher Eddington ratios. It is well known that the Eddington ratio is an important parameter to describe the accretion process in AGNs. The [O III] luminosity is usually used as a surrogate for the AGN luminosity (Heckman et al. 2004 and the reference therein). In gure 5, we plot the [O III] luminosity versus the SMBHs masses. Using a least-squares regression, we derive the correlation between MBH and L[OIII] to be: log(L[OIII] /L ) = (6.830.65)+(0.220.08)log(MBH /M ). The correlation coecient R is 0.45, with a probability of P = 0.012 for rejecting the null hypothesis of no correlation. However, no correlation is found between the [O III] luminosity and the accretion ratio. From the peptonization model, the strength of [O III] is controlled by the NLRs covering factor, its density, and ionization parameter (e.g. Baskin & Laro 2005). The relation between the [O III] luminosity and the AGN bolometric luminosity is still a question to debate. 3.3 The g relation

The existence of a good correlation between stellar velocity dispersion ( ) and ionized gas velocity dispersion (g ) (e.g. Nelson & Whittle 1996) suggests that the gaseous kinematics of NLRs in Seyfert galaxies be primarily dominated by the bulge gravitational potential, which is further conrmed by Nelson & Whittle (1996). Most recently, Greene & Ho (2005) have investigated a large and homogeneous sample of lower-redshift (0.02 < z < 0.3) Type II AGNs from the SDSS, and found that g traces . Though the gas kinematics of NLRs are governed by the gravitational potential of the bulges, they also nd that the accretion rate plays an important secondary role. c Following Greene & Ho (2005), we study the g relation for 33 Type II AGNs at redshifts 0.3 < z < 0.83 after measuring the gas velocity dispersion (g ) from the narrow emission lines from NLRs, and the stellar velocity c dispersion ( ) from the Ca H+K, G-band absorption feature, which is shown in gure 2. It is obvious that, the line widths of the core component of [O III] and [O II] can approximately trace the stellar velocity dispersion, although with a considerable scatter (see gure 6). We conrm that

CONCLUSION

The stellar population synthesis code is used to model the stellar contribution for a sample of 209 type II AGNs at redshifts 0.3 < z < 0.83 from SDSS. According to the L[OIII] criterion of 3 108 L , 20 of which can be classied as type II quasars. The main conclusions can be summarized as follows.
c 0000 RAS, MNRAS 000, 000000

5
The reliable are measured for 33 type II AGNs with signicant stellar absorption features. We used the formula of Greene & Ho to obtain the corrected stellar velocity disc persions ( ). And SMBHs masses are calculated from the c MBH relation. A median strong relation between the [O III] luminosity and the SMBH mass is found, no correlation between the [O III] luminosity and the Eddington ratio. The gas velocity dispersion (g ) in NLRs is measured using three sets of two-gaussian proles to t [O III]4959, 5007 and [O II]3727, 3729) in these 33 stellar-light subc tracted spectra. We nd that g can trace with considerable scatter, which conrms that the gaseous kinematics of NLRs in these type II quasars are primarily dominated by the gravitational potential of the bulge. c The distribution of < g / > is 1.24 0.76 for the core [O III] line and 1.20 0.96 for the [O II] line, which c suggests that g can trace within about 0.1 dex in the c c logarithm of . The deviation of g from is correlated with the Eddington ratio.
Greene J. E., Ho L. C., 2006a, ApJL, in press, astro-ph/0512461 Greene J. E., Ho L. C., 2006b, ApJ, in press, astro-ph/0512462 Gu Q., Melnick J., Cid Fernandes R., Kunth D., Terlevich, Terlevich R., et al. 2006, MNRAS, 366, 480 Heckman T. M., et al., 2004, ApJ, 613, 109 Kaspi S., Smith P.S., Netzer H., Maoz D., Jannuzi B.T., Giveon U., 2000, ApJ, 533, 631 Kaumann G., et al. 2003, MNRAS, 346, 1055 Kaumann, G., & Heckman, T. M. 2005, Philos. Trans. R. Soc. London, A363, 621 Kriss G. A. 1994, in ASP Conf. Ser. 61, Astronomical Data Analysis Software and Systems III, ed. Crabtree D. R., Hanisch R. J., Barnes J. (San Francisco: ASP), 437 Maggorian J., et al., 1998, AJ, 115, 2285 McLure R. J., Jarvis M. J., 2002, MNRAS, 337, 109 Nelson C. H., 2001, ApJ, 544, L91 Nelson C. H., Whittle M., 1996, ApJ, 465, 96 Schlegel D. J., Finkbeiner D. P., Davis M., 1998, ApJ, 500, 525 Shen Z. Q., et al., Nature, 2005, 438, 62 Tonry J., & Davis M., 1979, ApJ, 84, 1511 Tremaine S., et al., 2002, Ap J, 574, 740 Urry C. M., Padovani P., 1995, PASP, 107, 803 Vestergaard M., 2002, ApJ, 571, 733 Watanabe M., et al., 2003, ApJ, 591, 714 Wu X. B., et al., 2004, A& A, 424, 793 Zakamsa N. L., et al., 2003, AJ, 126, 2125 Woo J. H., et al., 2006, ApJ, in press, astro-ph/0603648

ACKNOWLEDGMENTS This work has been supported by the NSFC (No. 10403005; No. 10473005; No. 10273007) and NSF from Jiangsu Provincial Education Department (No. 03KJB160060). QGU would like to acknowledge the nancial supports from China Scholarship Council (CSC) and the National Natural Science Foundation of China under grants 10103001 and 10221001. Funding for the creation and distribution of the SDSS Archive has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, NASA, the National Science Foundation, the US Department of Energy, the Japanese Monbukagakusho, and the Max Planck Society. The SDSS Web site is http:// www.sdss.org/. This research has made use of the NASA/IPAC Extragalactic Database, which is operated by the Jet Propulsion Laboratory at Caltech, under contract with NASA.

REFERENCES
Antonucci R., 1993, ARA&A, 31, 4 Barth A. J., Ho L. C., & Sargent W. L. W., 2002, AJ, 124, 2607 Begelman M. C., 2003, Science, 300, 1898 Baskin A., Laro A., 2005, MNRAS, 358, 1043 Bian W., Zhao Y., 2004a, MNRAS, 347, 607 Bian W., Zhao Y., 2004b, MNRAS, 352, 823 Bian W., Yuan Q., Zhao Y., 2005, MNRAS, 364, 187 Bian W., Yuan Q., Zhao Y., 2006, MNRAS, 367, 860 Brinchmann J., et al., 2004, ApJ, 613, 109 Bruzual G., Charlot S., 2003, MNRAS, 344, 1000 Cardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345, 245 Cid Fernandes R., Sodre L., Schmitt H. R., Leao J. R. S., 2001, MNRAS, 325, 60 Cid Fernandes R., Gu Q., Melnick J., et al., 2004a, MNRAS, 355, 273 Cid Fernandes R., Gonzalez Delgado R., Storchi-Bergmann, T., Pires Martins, L., Schmitt H., 2004b, MNRAS, 356, 270 Cid Fernandes R., Mateus A., Sodre L., Stasinska G., Gomes J., 2005, MNRAS, 358, 363 Ferrarese, L., Merritt D. 2000, ApJ, 539, L9 Gebhardt K., et al. 2000, ApJ, 539, L13 Greene J. E., Ho L. C., 2005, ApJ, 627, 721 c 0000 RAS, MNRAS 000, 000000