A Combined Optical and X-ray Study of Unobscured Type 1 AGN. I. Optical Spectra and SED Modeling
aa r X i v : . [ a s t r o - ph . H E ] S e p Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 11 July 2018 (MN L A TEX style file v2.2)
A Combined Optical and X-ray Study of Unobscured Type1 AGN. I. Optical Spectra and SED Modeling
Chichuan Jin ⋆ , Martin Ward , Chris Done , Jonathan Gelbord , Department of Physics, University of Durham, South Road, Durham, DH1 3LE, UK Department of Astronomy & Astrophysics, Penn State University, PA 16801, USA
Accepted by MNRAS
ABSTRACT
We present modeling and interpretation of the continuum and emission lines for asample of 51 unobscured Type 1 active galactic nuclei (AGN). All of these AGNs havehigh quality spectra from both XMM-Newton and Sloan Digital Sky Survey (SDSS).We extend the wavelength coverage where possible by adding simultaneous UV datafrom the OM onboard XMM-Newton. Our sample is selected based on low reddeningin the optical and low gas columns implied by their X-ray spectra, except for one case,the BAL-quasar PG 1004+130. They also lack clear signatures for the presence of awarm absorber. Therefore the observed characteristics of this sample are likely to bedirectly related to the intrinsic properties of the central engine.To determine the intrinsic optical continuum we subtract the Balmer continuumand all major emission lines (including FeII). We also consider possible effects ofcontamination from the host galaxy. The resulting continuum is then used to derivethe properties of the underlying accretion disc. We constrain the black hole massesfrom spectral fits of the Balmer emission lines and determine the best fit value fromthe modeling of broadband spectral energy distributions (SED). In addition to thedisc component, many of these SEDs also exhibit a strong soft X-ray excess, plus apower law extending to higher X-ray energies. We fit these SEDs by applying a newbroadband SED model which comprises the accretion disc emission, low temperatureoptically thick Comptonisation and a hard X-ray tail by introducing the concept ofa corona radius (Done et al. 2011). We find that in order to fit the data, the modeloften requires an additional long wavelength optical continuum component, whoseorigin is discussed in this paper. We also find that the Photo-recombination edge ofBalmer continuum shifts and broadens beyond the standard limit of 3646˚A, implyingan electron number density which is far higher than that in the broad line regionclouds.Our results indicate that the Narrow Line Seyfert 1s in this sample tend to havelower black hole masses, higher Eddington ratios, softer 2-10 keV band spectra, lower2-10 keV luminosities and higher α ox , compared with typical broad line Seyfert 1s(BLS1), although their bolometric luminosities are similar. We illustrate these differ-ences in properties by forming an average SED for three subsamples, based on theFWHM velocity width of the H β emission line. Key words: accretion, broadband SED modeling, active-galaxies: nuclei
The spectral energy distribution (SED) of AGN has beenmodeled for several decades. Initial studies focused on theinfrared, optical and ultraviolet continuum (e.g. Wills et al.1985; Canalizo & Stockton 2001; Lacy et al. 2007). With ⋆ E-mail: [email protected] the inclusion of X-ray data, it was possible to define thecontinuum on both sides of the ultraviolet/X-ray gap (im-posed by galactic photoelectric absorption), and so constrainthe properties of the accretion disc (e.g. Ward et al. 1987;Elvis et al. 1994). Refinements to modeling the optical/UVcontinuum include subtraction of the complex blended fea-tures arising from permitted iron emission, the so-calledsmall blue-bump from the Balmer continuum, and contami- c (cid:13) C. Jin, M. Ward, C. Done and J. M. Gelbord nation across the entire spectrum from a stellar component(Maoz et al. 1993; Boisson et al. 2000)The observed spectral differences between various typesof AGN are not only due to selective absorption andorientation effects, as implied by the simplest version ofAGN unification model (Antonucci 1993), but also resultfrom a wide range in basic physical parameters, such asblack hole mass and accretion rate (e.g. Boroson & Green1992; Boller, Brandt & Fink 1996; Done & Gierli´nski 2005;Zhou et al. 2006). To better understand the accretion pro-cesses occurring close to the super massive black hole(SMBH), we construct broadband SEDs. Galactic dust red-dening, together with the intrinsic reddening of the AGN it-self, attenuates the optical/UV band emission. Furthermore,Photoelectric absorption from gas modifies the lower energyX-ray continuum. But these factors can be quantified andcorrected. Thereby we can recover the intrinsic SED, exceptfor the unobservable far-UV region. If we have reliable dataon both sides of the energy gap between the UV and softX-ray, we can apply a multi-component model which spansacross it.
Many multi-wavelength studies have been carried out previ-ously. Puchnarewicz et al. (1992) studied the optical prop-erties of 53 AGNs in C´ordova et al. (1992)’s sample with ul-tra soft X-ray excesses, and found that they tend to havenarrower permitted lines than optically selected samples.Supporting this finding, Boller, Brandt & Fink (1996) stud-ied ROSAT selected AGN with extremely soft X-ray spec-tra, and found that they tend to be Narrow-Line Seyfert 1s(NLS1s). Correspondingly they found that optically selectedNLS1s often have large soft X-ray excesses. Walter & Fink(1993) combined soft X-ray and optical data for 58 Seyfert1s, and showed that their broadband SED have a bumpfrom UV to soft X-rays, which is now refered to as thebig blue bump (BBB). Grupe et al. (1998) and Grupe et al.(1999) used a sample of 76 bright soft X-ray selected Seyfertswith infrared data, optical spectra and soft X-ray spectra.Their results reinforced the connection between the opti-cal and soft X-ray spectra, and confirmed the existence ofstrong BBB emission in these objects. Elvis et al. (1994)studied 47 quasars in a UV-soft X-ray sample, and derivedthe mean SEDs for radio-loud and radio-quiet sources. Re-cently, more detailed spectral models have been applied tobroadband SEDs including simultaneous optical/UV andX-ray observations which avoid potential problems causedby variability. Vasudevan & Fabian (2007) (hereafter VF07)combined a disc and broken powerlaw model to fit opti-cal, far UV and X-ray data for 54 AGN. They found awell-defined relationship between the hard X-ray bolomet-ric correction and the Eddington ratio. Brocksopp et al.(2006) analysed the data from XMM-Newton’s simultane-ous EPIC (X-ray) and OM (optical/UV) observations for22 Palomar Green (PG) quasars. Another sample consist-ing of 21 NLS1s and 13 broad line AGNs was also definedusing simultaneous data from XMM-Newton’s EPIC andOM monitor (Crummy et al. 2006). The SEDs of this sam-ple were then fitted using various broadband SED modelssuch as disc plus powerlaw model, disc reflection model anddisc wind absorption model (Middleton, Done & Gierli´nski 2007). Vasudevan & Fabian (2009) derived SEDs usingXMM-Newton’s simultaneous X-ray and optical/UV obser-vations for 29 AGNs selected from Peterson et al. (2004)’sreverberation mapped sample. The well constrained blackhole masses available for this sample enabled them to fita better constrained accretion disc model, combined witha powerlaw, to the source’s broadband SEDs. Hence theyderived more reliable Eddington ratios.
In this paper we define an X-ray/optically selected sampleof 51 AGN, all of which have low reddening (so excludingSeyfert 2s and 1.9/1.8s), to construct SEDs ranging fromabout 0.9 microns to 10 keV. We also apply corrections forthe permitted iron features, the Balmer continuum and stel-lar contribution, in order to model the non-stellar continuumfree from emission line effects. Included in this sample are anumber of NLS1s, a subclass of AGN whose permitted linewidths are comparable to those of forbidden lines. Their[OIII] λ β ratio is also lower than the typical value ofbroad line Seyfert 1s (BLS1s) (Shuder & Osterbrock 1981;Osterbrock & Pogge 1985). For consistency with previouswork, we classify AGNs in our sample as NLS1s if they haveratios of [OIII] λ β < Hβ < ∼
12 NLS1s in our sample .All objects in our sample have high quality optical spec-tra taken from the Sloan Digital Sky Survey (SDSS) DR7,X-ray spectra from the XMM-Newton EPIC cameras, andin some cases simultaneous optical/UV photometric datapoints from the XMM-Newton OM monitor. Combiningthese data reduces the impact of intrinsic variability andprovides a good estimate of the spectral shape in the opti-cal, near UV and X-ray regions. In addition, by analyzingthe SDSS spectra, we can derive the parameters of the prin-cipal optical emission lines and underlying continuum. Animportant result from reverberation mapping study is thecorrelation between black hole mass, monochromatic lumi-nosity at 5100 ˚A and H β FWHM (e.g. Kaspi et al. 2000;Woo & Urry 2002; Peterson et al. 2004). We measure thesequantities from the SDSS spectra, and then estimate blackhole masses using this correlation.Compared with previous work, a significant improve-ment of our study is that we employ a new broadband SEDmodel which combines disc emission, Comptonisation and ahigh energy powerlaw component in the context of an en-ergetically self-consistent model for the accretion disc emis-sion (Done et al. 2011, also see Section 5.2). By fitting thismodel to our data, we can reproduce the whole broadbandSED from the optical to X-ray. From this detailed SED fit-ting, we derive a number of interesting AGN properties suchas: the bolometric luminosity, Eddington ratio, hard X-rayslope, and the hard X-ray bolometric correction. Combin-ing all the broadband SED parameters with the optical pa-rameters, we can provide further evidence for many previ- Although 2XMM J112328.0+052823 and 1E 1346+26.7 haveH β FWHMs of 2000 km s − , 2050 km s − respectively, they bothhave H α FWHM of 1700 km s − , and also share other NLS1’sspectral characteristics. Thus they could both potentially be clas-sified as NLS1s, making a total of 12.c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Table 1.
The Seyfert 1 Galaxy Sample Set 2XMMi Catalog XMM-Newton SDSS DR7 SDSS EPICID Common Name a Redshift IAU Name (2XMM b ) Obs Date MJD-Plate-Fibre Obs Date Counts c ∗ ∗ ∗ a for some targets without well-known names, we simply use ‘2XMMi/DR7’; b the full name should be ‘2XMM J...’, but for those targets with * symbol, their full names should be ‘2XMMi J...’; c the total counts in all three EPIC monitors, namely pn, MOS1 and MOS2, and there are at least 2000 counts in at least oneof these three monitors;c (cid:13)000
The Seyfert 1 Galaxy Sample Set 2XMMi Catalog XMM-Newton SDSS DR7 SDSS EPICID Common Name a Redshift IAU Name (2XMM b ) Obs Date MJD-Plate-Fibre Obs Date Counts c ∗ ∗ ∗ a for some targets without well-known names, we simply use ‘2XMMi/DR7’; b the full name should be ‘2XMM J...’, but for those targets with * symbol, their full names should be ‘2XMMi J...’; c the total counts in all three EPIC monitors, namely pn, MOS1 and MOS2, and there are at least 2000 counts in at least oneof these three monitors;c (cid:13)000 , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord ously suggested correlations, including all the correlationsbetween optical and X-ray claimed in previous work, plusmany others such as the H β FWHM versus X-ray slope,black hole mass versus Eddington ratio, FeII luminosity ver-sus [OIII] λ λ λ H = 72 km s − Mpc − , Ω M = 0 .
27 and Ω Λ = 0 .
73 is adopted. In anotherpaper, we will present our analysis of correlations betweenselected optical/UV emission features and the SED compo-nents, and discuss their physical implications (Jin et al. inprep., hereafter Paper II).
To identify a sample of Type 1 AGNs having both highquality X-ray and optical spectra, we performed a cross-correlation between catalog and
SDSS DR7 catalog.We filtered the resulting large sample as described below.Our final sample consists of 51 Type 1 AGNs including 12NLS1s, all with high quality optical and X-ray spectra andlow reddening/absorption, and with H β line widths rangingfrom 600 kms − up to 13000 kms − . All the sources arelisted in Table 1. The first step was to cross-correlate between and
SDSS DR7 catalogs. The catalog contains 4117 XMM-Newton EPIC camera observations obtained between 03-02-2000 and 28-03-2008, and covering a sky area of ∼
420 deg .The SDSS DR7 is the seventh data release of the Sloan skysurvey. The SDSS spectroscopic data has sky coverage of ∼ , with spectra from 3800 ˚A to 9200 ˚A, and spectralresolution between 1800 and 2200.Our cross-correlation consisted of three steps:1. We first searched for all XMM/SDSS position pairs thatlay within 20 ′′ of each other, resulting in 5341 such cases.2. For these 5341 unique X-ray sources, we imposed two fur-ther selection criteria: that source positions be separated byless than 3 ′′ , or that sources be separated by no more than3 × the XMM-Newton position uncertainty and no morethan 7 ′′ . This filtering resulted in 3491 unique X-ray sources.The 3 ′′ separation is chosen because we want to include allpossible XMM/SDSS pairs during these early filtering steps.From the 2XMMi and SDSS DR7 cross-correlation, there are114 XMM/SDSS pairs whose separations are less than 3 ′′ ,but are still nevertheless greater than 3 × the XMM position uncertainty. We included all of these pairs. The 7 ′′ separa-tion upper limit mitigates spurious matches, especially forfainter objects and/or those located far off-axis.3. We selected only objects classified as extragalactic, givinga total of 3342 for further analysis. Within these 3342 unique X-ray sources which satisfied allthe above criteria, we applied further filtering to select onlyType 1 AGNs having both high quality optical and X-rayspectra. The five steps in the filtering were as follows:1. In order to obtain black hole mass estimates and also asreddening indicators, we require Hβ and Hα emission linesto be measurable. So we only selected sources with Hβ inemission (as indicated by the SDSS Hβ line models with atleast 3 σ significance and EW >
0) and redshift z < . Hβ line showed strongreddening or low S/N, which would distort the Hβ line pro-file. We also excluded one object, RBS 0992, because itsSDSS spectrum did not show an Hβ line, due to a bad datagap. We ensured that the remaining 73 objects all had good Hβ line profiles.4. As a simple method to assess the spectral quality of the X-ray data, we used wabs*powerlaw model in xspec11.3.2 tofit the rest-frame 2-10 keV X-ray spectra of all 73 objects.The error command was used to estimate the 90% confi-dence region for the photon index parameter. Based on theresults, 16 objects with photon index uncertainties greaterthan 0.5 were thereby excluded, leaving 57 Type 1 AGNswith relatively well constrained 2-10 keV spectra.5. By examining the 0.2-10 keV X-ray spectra, we ex-cluded another 6 objects (i.e. IRAS F09159+2129, IRASF12397+3333, PG 1114+445, PG 1307+085, PG 1309+355and PG 1425+267) whose spectral shapes all showedclear evidence of an absorption edge at ∼ The sample selection procedure described above ensures thatevery source in our AGN sample has both high quality op-tical and X-ray spectra. In addition, a large fraction of thesample have simultaneous optical/UV photometric points c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Figure 1.
The aperture effect correction results for 17 extended sources in the sample. The point like source RBS 0769 (the last figuremarked by **) is also shown for comparison. We over-plot OM data points on to the SDSS spectrum. Red OM points are data obtaineddirectly from the OM PPS files. Blue OM points are the corresponding data after applying a smaller 6 ′′ aperture to all OM filters, andapplying appropriate OM corrections to the flux eg. deadtime correction, coincidence loss correction and OM time sensitivity degradationcorrection.c (cid:13)000
The aperture effect correction results for 17 extended sources in the sample. The point like source RBS 0769 (the last figuremarked by **) is also shown for comparison. We over-plot OM data points on to the SDSS spectrum. Red OM points are data obtaineddirectly from the OM PPS files. Blue OM points are the corresponding data after applying a smaller 6 ′′ aperture to all OM filters, andapplying appropriate OM corrections to the flux eg. deadtime correction, coincidence loss correction and OM time sensitivity degradationcorrection.c (cid:13)000 , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord from the OM monitor. Such high quality data enables ac-curate spectral fitting. In the optical band our sample isselected to have low reddening, since if present this wouldsignificantly modify the intrinsic continuum as well as theoptical emission lines. This requirement reduces the com-plexity and uncertainty in our modeling of the intrinsic con-tinuum, and also increases the overall quality of H β andH α line profiles useful for estimating the black hole masses.Furthermore, low reddening is essential in the UV band. Theinclusion of OM-UV photometric data observed simultane-ously with the X-ray spectra provides a reliable link betweenthese bands. This helps to reduce fitting uncertainty of theSED resulting from optical and X-ray variability. Besides,all sources are well constrained in the 2-10 keV band, whichis directly associated with the compact emitting region ofthe AGN. Our exclusion of objects with evidence of a warmabsorber means that the 2-10 keV spectral index is likely tobe intrinsic rather than hardened by absorption in the softX-ray region.In summary, compared with previous AGN samplesused for broadband SED modelling, the spectrally ‘cleaner’nature of our sample should make the reconstructed broad-band SEDs more reliable. Consequently, the parameters de-rived from the broadband spectral fitting should be moreaccurate. This may reveal new and potentially importantbroadband correlations, which we will discuss in detail inpaper II. The 51 Type 1 AGNs all have SDSS survey-quality spectra(flagged as “sciencePrimary” in SDSS catalog), including 3objects that have multiple SDSS spectra (i.e. NVSS J030639,1RXS J111007 and Mrk1018). In such cases we adopt theSDSS spectrum which connects most smoothly with the OMdata.For each object, we used all available EPIC X-ray spec-tra (i.e. pn, MOS1 and MOS2) for the broadband SED mod-eling, unless the spectrum had few counts and low S/N. Wealso searched through the
XMM-OM SUSS catalog for all datain the OM bands (i.e. V, B, U, UVW2, UVM2 and UVW1),which are observed simultaneously with the correspondingEPIC spectrum. Of our 51 sources, we have 14 sources withSDSS optical spectra and XMM EPIC X-ray spectra, and37 sources which in addition to this also have XMM-OMphotometry.
In the procedure of combining the SDSS spectra and OMdata points, we identified that in some objects there is aclear discrepancy between these two data sets. The OMpoints often appear higher on the spectral plots (brigher)than is consistent from a smooth extrapolation of the SDSSspectral shape. In fewer cases this discrepancy appears inthe opposite sense, with the OM points apparently too low(fainter), see Figure 1 for some examples). This discrepancymay arise for several reasons, including a simple apertureeffect. Compared to 3 ′′ diameter for the SDSS spectroscopyfibres, the OM monitor has a much larger aperture, i.e. 12 ′′ and 35 ′′ diameter for the OM optical and OM UV filters respectively (Antonio Talavera.OMCal Team 2009). If thehost galaxy is sufficiently extended, e.g. in the case of REJ1034+396, the larger aperture of the OM would includemore host galaxy emission than that in the SDSS spectrum(see also section 5.3.1 for other possible reasons to accountfor this discrepancy). To investigate the aperture issue inmore detail, we performed the following tests:(1) We examined the combined SDSS and OM data plots,searching for those objects with excess OM flux comparedwith that expected from the extrapolated SDSS spectrum.We identified 27 such cases out of the 51 sources;(2) Within this sample of 27 sources, we checked the cata-log flag for an extended source in each OM filter. We notedthose flagged as an extended source in at least one OM fil-ter. This yielded 13 sources out of the 27.(3) We also extracted the SDSS CCD images for all 51 ob-jects and visually checked whether they appeared extended.As a result, we included another 4 objects for which theirSDSS CCD images show that their host galaxy is extendedbeyond the 3 ′′ diameter of the SDSS aperture. Either theywere not flagged as extended sources in any OM filter, orthey did not have any OM optical data. For these 17 objects,an aperture effect could at least be partially responsible foran excess flux in the OM data.(4) For these 17 objects we downloaded all available OMimage files. In each OM image, we applied a 6 ′′ diameteraperture from which to extract the flux. We used the samesized aperture placed on a blank region of sky close to the ob-ject, to estimate the background. The quoted PSF FWHMof the OM for the different filters are: V(1.35 ′′ ), B(1.39 ′′ ) ,U(1.55 ′′ ), UVW1(2.0 ′′ ), UVM2(1.8 ′′ ), UVW2(1.98 ′′ ). Thusin all cases 6 ′′ is at least 3 × PSF FWHM. So this apertureincludes effectively all optical flux for a point source, andmore than 90% that from a UV point source detected bythe OM.Before subtracting the background flux from thesource+background flux, we performed three count rate cal-ibrations, according the method described in the OM in-strument document. The first is the deadtime correction,required because for a small fraction of the exposure timethe CCD is in readout mode, and so cannot record events.The second calibration is for coincidence losses, which oc-cur when more than one photon arrives on the CCD at thesame location and within the same frame time, so resultsin under counting. The third calibration is for the OM timesensitivity degradation correction. We performed these cal-ibrations, according to the algorithms set out in the OMinstrument document, separately for the background andsource+background count rates. We then subtracted thebackground count rate from the source+background countrate to obtain the corrected source count rate.Figure 1 shows the OM data points before and after cor-rection for aperture effects for the 17 objects. The reducedOM aperture does improve the alignment between the OMpoints and SDSS spectrum. This correction not only lowersthe OM flux, but also changes the continuum shape definedby the OM points. Although choice of an aperture smaller URL: http://xmm2.esac.esa.int/docs/documents/CAL-TN-0019.ps.gz;Also see the XMM-Newton User Handbook:http://xmm.esac.esa.int/external/xmm user support/document-ation/uhb/index.html.c (cid:13) , 000–000
Spectral Study of Unobscured Type 1 AGN - I Fig-2a: An Example of SDSS Spectrum Fitting Fig-2b: Balmer Line Fitting
Figure 2.
An example of results from SDSS spectrum fitting. The left panel shows a good fit for PG 2233+234. The black line is theobserved spectrum, the red line is the total model spectrum. The green line represents the observed underlying continuum. The Balmercontinuum (blue), FeII emission (light blue) and other strong emission lines (orange) are shown underneath. The right panel shows anexample of detailed line profile fitting to the FeII subtracted region around the H β (upper) and H α lines (lower) including H α , H β , [OIII] λ λ λ λ λ β and H α , two components for [OIII] λ than 6 ′′ will lower the OM fluxes by a larger factor, it willalso introduce uncertainties and systematics caused by thePSF. Therefore we compromise by adopting a 6 ′′ diameteraperture. In our subsequent SED modeling we use the aper-ture corrected OM data. Our optical spectral modeling employs linked Hα and Hβ profile fitting and the complete optical spectral fitting. Wewrote the code in IDL (Interactive Data Language) v6.2, toperform all the optical spectral fitting. The ‘MPFITEXPR’program from the Markwardt IDL Library is incorporatedwithin our code to perform the Levenberg-Marquardt least-squares algorithm used to obtain the best-fit parameters.The SDSS spectra (stored in SDSS spSpec files) were ex-tracted directly from the SDSS DR7 data archive and ana-lyzed in IDL using our code. A detailed description of ourspectral modeling procedures is presented in the followingsubsections. α , H β and [OIII] λ Based on current AGN emission line models, there arethought to be stratified regions emitting different lines.These regions are divided somewhat arbitrarily into a narrow line region (NLR), a broad line region (BLR)and possibly an intermediate line region (ILR, e.g.Grupe et al. 1999; Hu et al. 2008; Mei, Yuan & Dong 2009;Zhu, Zhang & Tang 2009). Following previous studies, weuse several separate Gaussian profiles representing each ofthese emitting regions to model the Balmer line profiles.The H α and H β line profiles each pose distinct difficul-ties for the spectral analysis. In the case of the Hβ line, thepermitted FeII emission features (which are often strong inNLS1s) and broad HeII 4686 line blended with the Hβ line,which can affect the determination of the underlying contin-uum and hence the Hβ line profile. For the Hα line, there isthe problem of blending with the [NII] λ Hα ’s intrinsicprofile. Our approach, therefore, is to fit Hα and Hβ simul-taneously using the same multi Gaussian components. Theassumed similarity between the intrinsic profiles of these twoBalmer lines assists in deblending from other nearby emis-sion lines, and should yield a more robust deconvolution forthe separate components of their profile. We use the theoretical FeII model templates of Verner et al.(2009). These include 830 energy levels and 344,035 tran-sitions between 2000˚A and 12000 ˚A, totaling 1059 emis-sion lines. The predicted FeII emission depends on physicalconditions such as microturbulence velocity and hardnessof the radiation field, but we use the template which best c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord matches the observed spectrum of I ZW 1 (Boroson & Green1992, V´eron-Cetty et al. 2004) i.e. the one with n H =10 cm − , v turb = 30 kms − , F ionizing = 20 . cm − s − .Detailed modelling of high signal-to-noise spectra showsthat the FeII emission is often complex, with four ma-jor line systems in the case of 1 Zw 1, (one broad linesystem, two narrow high-excitation systems and one low-excitation system V´eron-Cetty et al. 2004; Zhou et al. 2006;Mei, Yuan & Dong 2009). However, for simplicity we will as-sume only one velocity structure and convolve this templatewith a single Lorentzian profile.We fit this to the actual FeII emission line features be-tween 5100 ˚A and 5600 ˚A (no other strong emission lines liein this wavelength range) of the de-redshifted SDSS spectra,leaving the FWHM of the Lorentzian and the normalizationof the FeII as free parameters. The resulting best-fit FeIImodel to this restricted wavelength range, was then extrap-olated and subtracted from the entire SDSS spectrum. Amajor benefit from subtracting the FeII features is that theprofiles of the [OIII] λ λ After fitting the [OIII] λ α andH β line profiles simultaneously. Following previous studieswe consider a simplified picture in which the Balmer lineshave three principal components, namely a narrow compo-nent (from the NLR), an intermediate component (from atransition region ILR between the NLR and BLR or fromthe inner edge of dusty torus (Zhu, Zhang & Tang 2009)),and a broad component (from the BLR). The intermediateand broad components are both represented by a Gaussianprofile, whereas the narrow component is assumed to be sim-ilar to that of [OIII] λ λ / λ λ λ λ λ λ Hα and Hβ lines are left as free parame-ters;2. Only the central narrow component of the [OIII] λ λ Hα and Hβ lines arefree parameters; 3. The shape of the narrow component is held the sameas the entire [OIII] λ Hβ line narrow component is set to be 10% of[OIII] λ λ λ Hβ line profile, and then measuredthe FWHM from this model. The rationale for using thismethod, instead of directly measuring the FWHM of the Hβ line from the data, is because for low signal/noise lineprofiles direct measurement of FWHM can lead to large un-certainties, whereas our profile models are not prone to lo-calized noise in the data. The H β FWHM measurements foreach of the 51 sources, after de-convolving using the instru-mental resolution of 69 kms − , are listed in Table 3. In order to obtain the underlying continuum, we must modelthe entire SDSS spectrum so that we can remove all theemission lines as well as the Balmer continuum and hostgalaxy contribution. As we are now concerned with thebroad continuum shape, we choose to refit the FeII spec-trum across the entire SDSS range, rather than restrictingthe fit to the H α and H β line regions as discussed in theprevious section.Figure 2 shows an illustrative example of our opticalspectral fitting, and the results for each of the 51 sourcesare presented in Figure A1. In the following subsections wegive further details of the components that make up thesemodeled spectra. We use the models for [OIII], Hα and Hβ as derived above.We add to this a series of higher order Balmer lines: from5 → Hγ ) to 15 →
2. We fix the line profile of these to thatof Hβ up to 9 →
2, then simply use a single Lorentzian profilefor the rest weak higher order Balmer lines. We fix the lineratios for each Balmer line using the values in Osterbrock(1989), Table 4.2, with T e between 10,000 K and 20,000 K.We similarly use a single Lorentizan to model the series ofHelium lines (HeI 3187, HeI 3889, HeI 4471, HeI 5876, HeII3204, HeII 4686) and some other emission lines (MgII 2798, c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I [NeIII] λ λ λ λ λ Another potentially significant contribution at shorter wave-lengths is from the Balmer continuum. Canfield & Puetter(1981) and Kwan & Krolik (1981) predicted the opticaldepth at the Balmer continuum edge to be less than 1,we use Equation 1 to model the Balmer continuum underthe assumptions of the optically thin case and a single-temperature electron population (also see Grandi 1982;Wills et al. 1985). F BCv = F BEv e − h ( v − v BE ) / ( kT e ) ( v > v BE ) (1)where F BEv is the flux at Balmer edge, v BE correspondsto the Balmer edge frequency at 3646˚A. T e is the electrontemperature. h is the Planck’s constant, k is the Boltzmann’sconstant. This Balmer continuum equation is then convolvedwith a Gaussian profile to represent the real Balmer bumpin SDSS spectra.There are several parameters that may slightly mod-ify or significantly change the shape of the Balmer contin-uum. It is already seen that the electron temperature T e appearing in Equation 1 and the optical depth can bothchange the Balmer continuum shape, but there are addi-tional important factors. Any intrinsic velocity dispersionwill Doppler broaden all the Hydrogen emission features.Therefore a better description of the Balmer continuum canbe obtained by convolving Equation 1 with a Gaussian pro-file, whose FWHM is determined by the line width of Hβ (or other broad lines), as shown by Equation 2, where G(x)represents a Gaussian profile with a specific FWHM. F BCλ = F BEλ e hc/ ( λ BE kT e ) Z + ∞ e − hc/ ( λkT e ) G ( λ − λ ) dλ (2)Figure 3 shows how the Balmer continuum’s shape de-pends on the electron temperature and velocity broaden-ing in Equation 2. The electron temperature modifies thedecrease in the Balmer continuum towards shorter wave-lengths, but has little effect on the broadening of (BalmerPhoto-recombination) BPR edge. On the contrary, velocitybroadening mainly affects the shape of the BPR edge, butthe emission longward of 3646˚A is still very weak comparedto the emission blueward of the BPR edge, i.e. the BPR edgeis still sharp.We initially applied Equation 2 to fit the Balmer contin-uum bump below 4000˚A in the SDSS spectra. We assumedthe velocity profile for the convolution was a Gaussian withits FWHM determined from the Hβ line profile, and thewavelength of the position of the BPR edge was taken asthe laboratory wavelength of 3646˚A. However, this modeldid not provide an acceptable fit, for example see the modelshown by the blue line in Figure 4. It appears that the ob-served spectrum requires a model with either a more ex-tended wing redward of the BPR edge, or a BPR edge thatshifts to longer wavelength than 3646˚A. However, additional Figure 3.
The Balmer continuum models of Grandi (1982). Theupper panel shows the dependence of the model on the electrontemperature. The lower panel shows the dependence of the modelon the FWHM of the convolved Lorentzian profile. velocity broadening should affect both the Balmer contin-uum and Balmer emission lines equally, as they are producedfrom the same material, although the multiple componentspresent in the line make this difficult to constrain.One way the wavelength of the edge may be shiftedwithout affecting the lines is via density (collisional, orStark) broadening (e.g. Pigarov et al. 1998). Multiple colli-sions disturb the outer energy levels, leading to an effective n max for the highest bound level ≪ ∞ , i.e. lowering theeffective ionization potential. We set the edge position andthe FWHM as free parameters, and let the observed spectralshape determine their best fit values. The red line shownin Figure 4 represents a good fit, obtained with FWHMof 6000 kms − and the BPR edge wavelength of 3746 ˚A,which implies n max ∼
12. The theoretical n max can be de-termined by the plasma density N e and temperature T e as n max = 2 × ( T e /N e ) / (Mihalas 1978), so for a typicaltemperature of 10 − K , the required density is 7 × − cm − . Such high density is not generally associated with theBLR clouds, and may give support to models where the lowionization BLR is from the illuminated accretion disc (e.g.Collin-Souffrin & Dumont 1990). However, any reliable es-timation of the density would require more accurate sub-traction of other optical components such as the FeII lineblends and many other non-hydrogen emission lines, whichis not the focus of this paper. Nonetheless, this remains aninteresting problem which is worthy of further study.Yet another issue in modeling the Balmer continuumis how to quantify the the total intensity of this continuumcomponent, especially when there is limited spectral cov-erage bellow 4000˚A, which makes it difficult to define theoverall shape. The theoretical flux ratio between the Balmercontinuum and the Hβ line under case B conditions can beexpressed by Equation 3 (Wills et al. 1985), I ( Bac ) /I ( Hβ ) = 3 . T . (3) c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Figure 4.
An expanded view of the region around the BPR edgein PG 1427+480. The blue and dashed lines represent the Balmercontinuum model superposed on the underlying disc continuum(green solid line) using standard parameters (blue dash), and alsoa set of best fit parameters (red dash line). The red and blue solidlines are models of the total optical spectrum, including the cor-responding Balmer continuum components and plus other com-ponents described in the text. The observed spectrum is shownin black. but other theoretical calculations of photonionization mod-els show that by varying the Balmer optical depth, elec-tron temperature and electron number density, this canresult in very different values of I(Bac)/I(H β ). For exam-ple, Canfield & Puetter (1981)’s calculation resulted in aI(Bac)/H α range of 0.05 ∼
10, Kwan & Krolik (1981) sug-gested I(Bac)/I(H β )=1.6 ∼
15, and other theoretical workalso confirmed a large range in flux ratios (Puetter & Levan1982; Kwan 1984; Hubbard & Puetter 1985). The observedranges in I(Bac)/I(H β ) are also large. Canfield & Puetter(1981) showed an observed range of 0.5 ∼ α ).Wills et al. (1985) observed 9 intermediate redshift QSOswhose I(Bac)/I(H β ) ranges from 4.65 ∼ Our basic assumption is that the residual optical spectrum,after subtraction of the Balmer continuum, FeII emissionand other emission lines mentioned previously, arises mainlyfrom the accretion disc emission. As a reasonable approxi-mation over a limited wavelength range we use a powerlawof the following form to fit the underlying continuum, F ( λ ) = C · ( λ/ A ) − C (4)The powerlaw approximation for the optical underlyingdisc continuum is also widely adopted in previous andrecent AGN optical spectral studies. (e.g. Grandi 1982;Tsuzuki et al. 2006; Zhou et al. 2006; Landt et al. 2011).We model the dust reddening using the Seaton (1979)’s1100˚A to 10000˚A reddening curve, and we apply this tothe overall model, i.e. emission lines, Balmer continuum andthe disc continuum. There are also other reddening curvesavailable such as Fitzpatrick (1986) for the Large Magel-lanic Cloud, Pr´evot et al. (1984) and Bouchet et al. (1985)for the Small Magellanic Cloud and Calzetti et al. (2000) forstarburst galaxies, but over the wavelength range of 2500˚Ato 10000˚A, the difference between these reddening curves issmall, except for Calzetti et al. (2000)’s curve which is ap-propriate for starburst galaxies, and is thus not applicablefor our AGN sample. Many previous studies on AGN’s optical/infrared spectrahave adopted a powerlaw as a reasonable approximationfor the accretion disc continuum blueward of 1 µ m (e.g.Mei, Yuan & Dong 2009; Bian & Huang 2010), but thesestudies also needed to include additional contributions fromthe host galaxy and emission from the dusty torus to accountfor the extra continuum emission at long wavelengths of theoptical spectrum (e.g. Kinney et al. 1996; Mannucci et al.2001; Landt et al. 2011). In our work we have also identifiedan inconsistency between the 3000˚A ∼ ′′ diameter fibreused to obtain the SDSS spectra also helps to reduce the con-tribution of stellar emission from a host galaxy, particularlyfor nearby sources in our sample such as KUG 1031+398.These evidences argues against the possibility that the redoptical continuum is primarily dominated by host galaxyemission. In fact, it is possible that the observed additionalcomponent arises due to emission from the outer regions ofa standard accretion disc (e.g. Soria & Puchnarewicz 2002; c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Collin & Kawaguchi 2004; Hao et al. 2010). The existence ofsuch an additional red optical continuum component reducesthe consistency of a powerlaw fit to the optical spectra.
Our optical spectral fitting is performed only for data blue-ward of 7000˚A. The choice to truncate the model at 7000˚Ais made for several reasons. We wish to include H α line inthe spectral fitting range, and the broad wing of H α pro-file sometimes extend to ∼ ∼ ∼ For each object we extracted the original data files (ODFs)and the pipeline products (PPS) from XMM-Newton Sci-ence Archive (XSA) . In the following data reduction pro-cess, tasks from XMM-Newton Science Analysis System http://xmm.esac.esa.int/external/xmm data acc/xsa/index.shtml (SAS) v7.1.0 were used. First, EPCHAIN/EMCHAIN tasks wereused to extract events unless the events files had alreadybeen extracted for each exposure by PPS. Then
ESPFILT task was used to define background Good Time Intervals(GTIs) that are free from flares. In each available EPIC im-age, a 45 ′′ radius circle was used to extract the source region,and an annulus centered on the source with inner and outerradii of 60 ′′ and 120 ′′ was used to define the background re-gion. For other sources listed in the region files of PPS thatare included in these regions, these were subtracted using thedefault radii generated by PPS, which scaled with the sourcebrightness. Then the GIT filter, source and background re-gion filters were applied to the corresponding events filesto produce a set of source and background events files. Weonly accepted photons with quality flag =0 and pattern 0 ∼ EPATPLOT task was then used to check for pile-up ef-fects. When pile-up was detected, an annulus with inner andouter radii of 12 ′′ and 45 ′′ was used instead of the previous45 ′′ radius circle to define the source region. Then sourceevents files were reproduced using the new source regionfilter. Source and background spectra were extracted fromthese events files for each available EPIC exposure. Tasks RMFGEN/ARFGEN were used to produce response matrices andauxiliary files for the source spectra. These final spectra weregrouped with a minimum of 25 counts per bin using the
GRPPHA v3.0.1 tool for spectral fitting in
Xspec v11.3.2 .To prepare the OM data, the om filter default.pi file andall response files for the V,B,U, UVW1, UVM2, UVW2 fil-ters were downloaded from the OM response file directoryin HEASARC Archive . We then checked the OM sourcelist file for each object to see if there were any available OMcount rates. Each count rate and its associated error wereentered into the om filter default.pi file and then combinedwith the response file of the corresponding OM filter, againby using the GRPPHA tool to produce OM data that could beused in
Xspec .Finally, the XMM-Newton EPIC spectra are combinedwith the aperture corrected OM photometric points, and theoptical continuum points produced from the optical under-lying continuum (obtained from the full optical spectrumfitting) using
FLX2XSP tool. From these data we constructeda broadband nuclear SED of each AGN. There is a ubiqui-tous data gap in the far UV region which is due to photo-electric absorption by Galactic gas. Unfortunately, in mostcases of low-redshift AGN, their intrinsic SED also peaks inthis very UV region, and so this unobservable energy bandoften conceals a large portion of the bolometric luminosity.In order to account for this, and to estimate the bolomet-ric luminosity, we fit the X-ray and UV/optical continua alltogether using a new broadband SED model (Done et al.2011,
Xspec model: optxagn ). We then calculate the bolo-metric luminosity by summing up the integrated emissionusing the best-fit parameters obtained for each continuumcomponent.
A standard interpretation of the broadband SED is thatthe emission is dominated by a multi-temperature accretion http://heasarc.gsfc.nasa.gov/FTP/xmm/data/responses/om/c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Table 2.
Broadband SED Fitting Parameters, and Model Outputs (L bol , f d , f c , f p ). ID: object number, the same as Table 1;N H,gal and N
H,int : the fixed galactic and free intrinsic neutral hydrogen column densities in 10 cm − ; Γ pow : the powerlawcomponent’s slope in the SED fitting, (*) denotes the objects whose powerlaw slopes hit the uplimit of 2.2 and were fixedthere; Fpl: the fraction of powerlaw component in the total reprocessed disc emission; R cor : corona (truncation) radius in unitof Gravitational radii (r g ) within which all disc emission is reprocessed into the Comptonisation and powerlaw components;T e : temperature of the Compton up-scattering electron population; Tau: optical depth of the Comptonisation component;log(M BH ): the best-fit black hole mass; log( ˙ M ): total mass accretion rate; L bol : bolometric luminosity integrated from 0.001 keVto 100 keV; f d , f c , f p : luminosity fractions of disc emission, soft Comptonisation and hard X-ray Compotonisation componentsin the bolometric luminosity; χ : the reduced χ of the broadband SED fitting.ID N H,gal N H,int Γ pow Fpl R cor T e Tau log (M BH ) log ( ˙ M ) L bol f d f c f p χ × × r g keV M ⊙ g s − reduced ∗ ∗ ∗ ∗ ∗ ∗ ∗ (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Table 3.
Broadband SED Key Parameters. ID: object number, the same as Table 1; Γ − keV : the slope of the singlepowerlaw fitted to 2-10 keV spectrum. L − keV : 2-10 keV luminosity (in 10 erg s − ); κ − keV : the 2-10keV bolometriccorrection coefficient; λ L ˚ A : the monochromatic luminosity at 2500˚A (in 10 erg s − ); ν L keV : the monochromaticluminosity at 2keV (in 10 erg s − ); α ox : the optical X-ray spectral index; λ L : the monochromatic luminosity at5100˚A (in 10 erg s − ); κ : the 5100˚A bolometric correction coefficient; FWHM Hβ : the narrow component subtractedH β FWHM; L bol /L Edd : the Eddington Ratio.ID Γ − keV L − keV κ − keV λ L ˚ A ν L keV α ox λ L κ FWHM Hβ L bol /L Edd × × × × km s − ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord disc component which peaks in the UV (e.g. Gierli´nski et al.1999,
Xspec model: diskpn ). This produces the seed photonsfor Compton up-scattering by a hot, optically thin elec-tron population within a corona situated above the disc,resulting in a power law component above 2 keV (e.g.Haardt & Maraschi 1991; Zdziarski, Poutanen & Johnson2000,
Xspec model: bknpl ). However, the X-ray dataclearly show that there is yet another component whichrises below 1 keV in almost all high mass accretionrate AGNs. The ubiquity of this component can beseen, for example, in the compilation of AGN SEDspresented in Middleton, Done & Gierli´nski (2007), andone of the strongest cases is the NLS1 RE J1034+396(Casebeer, Leighly & Baron 2006; Middleton et al. 2009)and RX J0136.9-3510 (Jin et al. 2009). The origin ofthis so-called soft X-ray excess is still unclear (e.g.Gierli´nski & Done 2004; Crummy et al. 2006; Turner et al.2007; Miller, Turner & Reeves 2008), and so some previousbroadband SED modeling studies have explicitly excludeddata below 1 keV. An obvious consequence is that in suchstudies a soft excess component cannot influence the mod-els, so making it possible to fit the data using just a disc and(broken) power law continuum (VF07; Vasudevan & Fabian2009). However, in our current study we include all of thedata, and so we require a self-consistent model which incor-porates this soft component.Whatever the true origin of the soft X-ray excess, thesimplest model which can phenomenologically fit its shapeis the optically thick, low temperature thermal Comptonisa-tion model ( compTT ). But the observed data are used to con-strain the three separate components, discpn + compTT +bknpl , which is generally problematic given the gap in spec-tral coverage between the UV and soft X-ray regions causedby interstellar absorption. So instead, we combine thesethree components together using a local model in xspec ,assuming that they are all ultimately powered by gravita-tional energy released in accretion. A complete descriptionof this model can be found in the
Xspec website and is alsogiven in Done et al. (2011). It is in essence a faster versionof the models recently applied to black hole binary spec-tra observed close to their Eddington limit (Done & Kubota2006) and to the (possibly super Eddington) Ultra-Luminous X-ray sources (Gladstone, Roberts & Done 2009;Middleton & Done 2010), thus this model is more appropri-ate for fitting a medium sized sample of objects. A com-prehensive comparison with the model of Done & Kubota(2006) is given in Done et al. (2011). To make this paperself contained we give a brief synopsis of the model. We as-sume that the gravitational energy released in the disc ateach radius is emitted as a blackbody only down to a givenradius, R corona . Below this radius, we further assume thatthe energy can no longer be completely thermalised, and isdistributed between the soft excess component and the highenergy tail. Thus the model includes all three componentswhich are known to contribute to AGN SED in a self con-sistent way. As such it represents an improvement on thefits in VF07 in several respects, by including the soft excessand by requiring energy conservation, and it improves onDone & Kubota (2006) by including the power law tail.In our SED fitting, the optical/UV data constrains themass accretion rate through the outer disc, provided we havean estimate of the black hole mass. We constrain this by our analysis of the Hβ emission line profile. The main differencefrom previous studies based on non-reverberation samples isthat we do not directly use the FWHM of the H β profile toderive the black hole mass. Rather, we use the FWHM of theintermediate and broad line component determined from theemission line fitting results presented in Section 3.1. Theseare then used in Equation 5 (Woo & Urry 2002 and refer-ences therein) to derive the black hole mass limits requiredfor the SED fitting: M BH = 4 . × [ λL λ (5100˚ A )10 ergs − ] . F W HM (5)where L λ (5100˚A) is measured directly from the SDSS spec-tra. The rms difference between the black hole masses fromthis equation and from the reverberation mapping study is ∼ β pro-file. Section 6.5 discusses the differences between the best-fitblack hole masses and those estimated using other methods.Once the black hole mass is constrained, the opticaldata then sets the mass accretion rate ˙ M , and hence thetotal energy available is determined by the accretion effi-ciency. We assume a stress-free (Novikov-Thorne) emissiv-ity for a Schwarzschild black hole, i.e. an overall efficiencyof 0.057 for R in = 6 R g . Thus the total luminosity of thesoft excess and power law is 0 .
057 ˙
M c (1 − R in /R corona ).This constrains the model in the unobservable EUV region,with the input free parameter R corona setting the modeloutput of the luminosity ratio between the standard discemission and Comptonisation components. The upper limitof R corona is set to be 100 R g , which corresponds to 81% re-leased accretion disc energy. This upper limit is based on therequirement that the seed photons should be up-scattered(Done et al. 2011). We assume that both the Comptonisa-tion components scatter seed photons from the accretiondisc with temperature corresponding to R corona . The othermodel input parameters are; the temperature ( kT e ) and op-tical depth ( τ ) of the soft Comptonisation component whichare determined by the shape of the soft X-ray excess, thespectral index (Γ) of the hard X-ray Comptonisation thatproduces the 2-10 keV power law, with electron tempera-ture fixed at 100 keV. The model output f pl represents thefraction of the non-thermalised accretion energy (i.e. givenby the luminosity originating from the region of R corona to R in ), which is emitted in the hard X-ray Comptonisation.We also included two sets of corrections for attenuation( reddening - wabs ), to account for the line of sight Galacticabsorption and for the absorption intrinsic to each source,the latter is redshifted ( zred and zwabs in Xspec ). TheGalactic HI column density is fixed at the value taken fromKalberla et al. (2005), but the intrinsic HI column densityis left as a free parameter. The standard dust to gas con-version formula of E(B-V)=1 . × − N H (Bessell 1991)is used for both Galactic and intrinsic reddening. We setthe initial value of the powerlaw photon index to be thatof the photon index in 2-10 keV energy band, but it canvary during the fitting process. However, we set an upperlimit of 2.2 for the powerlaw photon index, not only becausethe photon index is < c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I (Middleton, Done & Gierli´nski 2007), but also because oth-erwise the much higher signal-to-noise in the soft excess insome observed spectra can artificially steepen the hard X-ray powerlaw and result in unphyiscal best-fit models.All free parameters used in the broadband SED fittingare listed in Table 2. For completeness, we also explicitlycalculate the fraction of the total luminosity carried by eachcomponent of the model (i.e. disc: f d ; soft Comptonization: f c ; hard X-ray Comptonization: f p ) from the model fit pa-rameters R cor and F pl (see Table 2) . Table 3 lists theimportant characteristic parameters. The main uncertaintyin these parameters, especially the black hole mass, is dom-inated by other systematic uncertainties introduced by theobservational data, model assumptions (e.g. the assumptionof a non-spinning black hole and the inclination dependenceof the disc emission) and the analysis methods involved.Therefore the parameter fitting uncertainties which are oftenless than 10%, are not significant in comparison, and thusare not listed. The statistical properties of these parametersare discussed in section 6. We further discuss two problems we encountered during thefitting procedure in the following subsections. The first prob-lem is the discrepancy between the OM and SDSS contin-uum points (mentioned in Section 2.5). The second problemis that of the observed flat optical continuum, whose shapecannot be accounted for in our SED model (mentioned inSection 4.4).
There remains a significant discrepancy between many ofthe OM and SDSS continuum points, even after applying theaperture correction discussed in Section 2.2 (see Figure A1).The OM points often appear above (brighter) the extrapo-lation of the SDSS continuum to the OM wavelengths. Weidentify three possible reasons for this discrepancy:(1) Remaining aperture effects: There is an aperturedifference between the SDSS fibres (3 ′′ diameter) and theOM apertures we used (6 ′′ diameter). Clearly the OM pointswill still include more host galaxy starlight than the SDSSpoints, and so will appear above the SDSS spectrum.(2) Contamination from emission lines: The wavelengthranges for each OM filter (over which the effective trans-mission is greater than 10% of the peak effective trans-mission) are as follows: UVW2 1805-2454˚A, UVM2 1970-2675˚A, UVW1 2410-3565˚A, U 3030-3890˚A, B 3815-4910˚A,V 5020-5870˚A. We exclude the contribution from strong op-tical emission lines within the OM U, B, V bandpass (andalso the Balmer continuum contribution in U band) by us-ing the best-fit optical underlying continuum which excludessuch features from the SDSS spectral fitting. In fact, this a full description of the model parameters can be found on the Xspec web page:http://heasarc.nasa.gov/xanadu/xspec/models/optxagn.html was an important initial motivation of the study, i.e. to ob-tain more accurate estimates of the true underlying contin-uum rather than simply to use the SDSS ‘ugriz’ photometricdata. Inclusion of strong emission lines within these photo-metric data would result in over-estimation of the opticalcontinuum, and so compromise our aim to study the shapeof the optical underlying continuum. This is an importantspectral characteristic used to constrain the accretion disccomponent in the SED fitting (see also the discussion insection 5.1.2). There are some strong emission lines withinthe UV bandpasses such as Ly α , CIV 1549, CIII 1909 andMgII 2798, whose fluxes are not available from SDSS spec-trum. Accurate subtraction of these line fluxes for each ob-ject would require new UV spectroscopy. We conclude thatinclusion of emission line flux within the OM photometricpoints may account for some of the observed discrepancy.(3) Intrinsic source variability: AGN are well known tobe variable across their SEDs. In general there is a signif-icant time difference between acquistion of the SDSS andOM-UV data, so intrinsic variation may contribute to anyobserved discrepancy. Mrk 110 is the most extreme exam-ple of this phenomena in our sample, as its SDSS spectrumhas a very large discrepancy compared with the OM data.The recent paper by Landt et al. (2011) gives another set ofoptical spectra for Mrk 110, which is more consistent withour best-fit model. It shows that the inclusion of OM datais useful to help identify cases such as this. As an additionaltest for variability, we assembled all available GALEX datafor our sample. We find that 43 objects in our sample haveGALEX data. Using a GALEX aperture of 12 ′′ , which islimited by the PSF and which is also similar to the UV OMapertures, we compare these values with the SED model.The ratio of the GALEX data and our SED model withinthe same bandpass differ by less than a factor of 2 for themajority of our sample, and significantly the flux ratio dis-tribution is almost symmetric and is centered close to unity.This suggests that the non-simultaneous OM and SDSS datais not likely to be a major impediment to our modeling.In effect, these three factors will merge together to pro-duce the observed discrepancy between the SDSS and OMdata. Since the combined effects of Point (1) and (2) whichwill add flux and generally be greater than that caused byoptical/UV variability as shown by previous long term rever-beration mapping studies (Giveon et al. 1999; Kaspi et al.2000), we should treat the OM points included in our SEDmodeling as upper limits when interpreting the results ofour modeling. Indeed, the 90% confidence uncertainties inthe BH masses derived directly from the Xspec fitting arealmost certainly small compared with the systematic errorsintroduced by the above uncertainties.
A related problem in our fitting is about the SDSS contin-uum shape. For some AGNs, their SDSS continuum datapoints exhibit a very different spectral slope from that ofthe SED model. This cannot be reconciled by adjusting theparameters of the accretion disc model, and thus impliesthe presence of an additional component at longer opticalwavelengths, which flattens compared with that predictedby the accretion disc models. One obvious explanation forthis flux excess is the contribution from the host galaxy. In c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Fig-a1: 2XMM J112328.0+052823 Optical Spectrum Fig-a2: SED Fitting Before and After Host Galaxy SubtractionFig-b1: PG1415+451 Optical Spectrum Fig-b2: SED Fitting Before and After Host Galaxy Subtraction
Figure 5.
A comparison between the results of two subtractions of host galaxy contribution. 2XMM J112328+052823 (Fig-a1 and Fig-a2)shows an underlying continuum that more closely resembles a disc continuum (solid green line in Fig-a1) after modelling and subtractingthe host galaxy contribution (light blue spectrum in Fig-a1). The left panel of Fig-a2 shows the original broadband SED fitting withoutsubtracting the host galaxy contribution. The dashed green line shows the modelled accretion disc emission in the best-fit SED. Theinserted panel shows a magnification of the fit in the optical/UV region, where a big discrepancy exists between the SDSS data andbest-fit SED model. The right panel of Fig-a2 is the new SED fit using the new underlying disc continuum (shown as solid green line inFig-a1) after subtracting the host galaxy contribution. The new fit is improved in the optical region compared with the previous resultsin the left panel of Fig-a2. In contrast to the above example, PG 1415+451 (Fig-b1 and Fig-b2) has little host galaxy contribution inthe SDSS optical spectrum (see the light blue component in Fig-b1), and its broadband SED fitting in the optical region remains poorregardless of the amount of host galaxy subtraction applied (see the two panels in Fig-b2). The spectral template for Elliptical galaxiesin Kinney et al. (1996) was used in both cases since their host galaxies both have elliptical morphologies in SDSS image. late type host galaxies such as elliptical and S0 galaxies,emission from their old stellar populations peaks at nearinfrared wavelengths. Kinney et al. (1996) combined spec-tra of quiescent galaxies and constructed an average spec-tral template for each morphological type, including bulge,elliptical, S0, Sa, Sb, Sc and starburst galaxies. For someobjects in our sample with high S/N SDSS spectra whichshow at least marginal stellar absorption features, we haveadded the corresponding type of host galaxy spectral tem-plate taken from Kinney et al. (1996), into the overall SDSSspectral fitting. This revised underlying continuum in theoptical, and was then used in the broadband SED fitting.We are then able to compare it with the original fit, to seehow the subtraction of a stellar population template effectsthe overall SED fitting.Figure 5 shows two examples. The first is 2XMM J112328.0+052823, in which after subtracting the hostgalaxy component, the observed optical continuum is closerto the slope of the SED model. However, the results forPG1415+451 in Figure 5 lower panel imply that its hostgalaxy cannot be the origin of the flat optical spectrum.The reason is that its optical spectrum does not show anystrong stellar absorption features. This means that the maxi-mum amount of host galaxy contribution is small, and so andthere remains a substantial inconsistency in the slope versusthe SED model. In addition to 2XMM J112328.0+052823above, only Mrk1018 and 2XMM J125553.0+272405 showclear stellar absorption features. Also the 3 ′′ diameter fibreexcludes much of the host galaxy component at these red-shifts. Therefore, on these general grounds we conclude thathost galaxy contamination is small for most sources in oursample, and consequently cannot fully account for the ob- c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I served flat optical continuum. Additional support for thisview comes from good correlations between the X-ray com-ponents and the red optical continuum, suggesting that thisextra optical flux is likely related to the intrinsic activity(e.g. Soria & Puchnarewicz 2002; Collin & Kawaguchi 2004;Hao et al. 2010; Landt et al. 2011). Histograms of data on our sample are shown in Figure 6,Figure 8 and Figure 9, including redshift, HI column density,optical and X-ray modeling parameters etc. The red regionin the histograms show the distributions for the 12 NLS1sin our sample. It is clear that NLS1s are distinct among thewhole sample in several respects.
Figure 6 shows some basic properties of our sample whichare not model dependent:(1). Redshift: the sample’s redshift ranges from 0.031 (Mrk493) to 0.377 (HS 0810+5157). The NLS1s are found mainlyat lower redshifts, with < z > n = 0 .
12 compared to the < z > n = 0 .
19 for the BLS1s. For comparison we see thatthe sample of VF07 has a similar redshift range, but it hasa lower average redshift of 0.10.(2). The Galactic nH: the average Galactic nH is 2 . × .(3). The photon indexes obtained from simple power lawfits to the restricted energy range of 2-10 keV. The NLS1scluster on the higher photon index side, with an aver-age of 2 . ± .
20, which differs from the sample average of1 . ± .
25 and the BLS1s’ average of 1 . ± .
18. This meansthat NLS1s tend to have softer X-ray spectra, which is fur-ther confirmed in the following section on the mean SEDs.(4). The X-ray continuum and 2-10 keV luminosity: this dis-tribution shows that NLS1s have lower 2-10 keV luminositiesin spite of their steeper slopes. We note that the VF07 sam-ple has a similar distribution, except for their inclusion ofthree extremely low X-ray luminosity AGN (i.e. NGC4395,NGC3227 and NGC6814), these objects were not includedin our sample due to our selection criteria and/or a lack ofSDSS spectra.(5). The optical continuum luminosity at 5100 ˚A. On aver-age the NLS1 have lower optical luminosities than BLS1.(6-8). The [OIII] λ α and H β emission line luminosi-ties. Again the NLS1s have on average lower luminositiesthan BLS1s.(9). The Balmer decrement. The average value for the wholesample is 3.14 ± ± Figure 7 shows properties derived from the SED fits:(1). The bolometric luminosity: the distribution range is be-tween 1 . × ergs s − (Mrk 464) and 1 . × ergs s − (PG 2233+134). There is no clear difference in the distribu-tion of the complete sample and the sub-set of NLS1s. The average luminosity is Log(L bol )=45.49 ± ± M ⊙ and10 M ⊙ . Equation 5 suggests that the black hole mass shoulddepend on both H β FWHM and L , and the results fromour SED fitting suggest that NLS1s with smaller Balmerline FWHM do indeed harbour lower mass black holes. KUG1034+396 has the lowest black hole mass in our sample. Thevalue of 1 . × M ⊙ is consistent with the estimate based onthe first firmly detected AGN QPO (quasi periodic oscilla-tion) found in this source (Gierli´nski et al. 2008). Again wecan compare our results with those of VF07 sample. We findthat their average black hole mass is 7.89 ± , FWHM Hβ ) relation. Adopting this samemethod for our sample, we find a very similar average of7.99 ± ± ± ± ± ± ∼ α ox index, is defined between restframe continuumpoints at 2500 ˚A and 2 keV (see Lusso et al. 2010 and ref-erences therein). The distribution for NLS1 is peaked atmarginally higher values than for BLS1.(5). The κ − bolometric correction, is defined as L bol /L − (see VF07 and references therein). We find thatNLS1s have a significantly higher fraction of their bolometricluminosity emitted as hard X-rays than the BLS1s. Com-pared with the VF07 sample, both distributions peak at κ − =10 ∼
30, but our sample shows a smoother distribu-tion decreasing as κ − increases after ∼
30, and so resultsin a slightly higher average value of κ − .(6). The intrinsic nH: This distribution shows that the in-trinsic equivalent neutral hydrogen column densities are lowfor our sample, which is a natural consequence of our initialsample selection criteria. The NLS1s have slightly higher in-trinsic absorption than BLS1s, which may imply a slightlyhigher dust reddening. However the distribution of Balmerdecrements shows no significant difference between these twotypes of AGNs.(7). The temperature of the Comptonisation componentused to describe the soft X-ray excess. This is close to0.2 keV in all objects, confirming the trend seen in previ-ous studies for this component to exhibit a narrow range ofpeak energy (Czerny et al. 2003; Gierli´nski & Done 2004).The distribution peak at this energy is more marked for theBLS1 than for NLS1, although the small number statisticsmeans that this difference cannot be considered as definitivefor our sample.(8). The optical depth of the soft excess Comptonised com-ponent. It is clear that this component is always optically c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Figure 6.
Distributions of our sample for different properties. In each panel the blue areas show the distribution for the whole sample,while the red areas show the distribution for the 12 NLS1s in our sample. We note that the H α , H β and [OIII] λ thick, with most objects having τ ∼ −
30. There is no sig-nificant difference in temperature or optical depth betweenthe broad and narrow line objects.(9). There is a difference in the coronal radii distributionbetween the BLS1s and NLS1s. Corona radius controls therelative amount of power emerging from the accretion discand the soft X-ray excess/hard tail. There are two peaksin the distribution for the broad line objects, one between10 and 20 R g (where R g = GM/c ), and the other at 100R g (which is set as the upper limit of this parameter in ourbroadband SED model). By contrast these radii in NLS1are consistent with just the first peak. At first sight this issurprising, since NLS1 are expected to be those with thestrongest soft X-ray excess. However, their similar soft ex-cess temperatures around 0.2 keV suggests that atomic pro-cesses may be significant (reflection and/or absorption frompartially ionized material), and this may influence our fits.The average coronal radii are 32 ±
26 R g for NLS1, 59 ±
37 R g for BLS1 and 53 ±
36 R g for the whole sample. This supportsthe conclusion of VF07 that high Eddington ratio AGN have lower coronal fractions compared to those with low Edding-ton ratios. Figure 8 shows further details of the modeled profiles of Hα (first row), and Hβ (second row).(1). The FWHM of the broad emission profile. This is cal-culated from co-adding the two best fit Gaussian profiles forthe broad and intermediate line components, and then usingthe resultant profile to determine the FWHM. This is equiv-alent to subtracting the narrow line core from the observedprofile and measuring the resultant FWHM.Note, the NLS1s by definition have H α < s − .(2-3). The equivalent widths and line luminosities are againmeasured using the total broad emission line profile asabove. The NLS1s have both lower equivalent widths andline luminosities.(4). By contrast, there is no pronounced difference betweenNLS1s and BLS1s in their Balmer narrow line component.This suggests that the narrow line region is less influenced c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Figure 7.
The distribution of model dependent parameters using the same colour coding as in Figure 6. Comments on each distributionare given in Section 6.2. by whatever difference in properties is responsible for thedefining difference between NLS1s and BLS1s in the broadline region.
The fraction of the total luminosity contained in each com-ponent of the SED model is shown in Figure 9. The upperleft panels show these fractions as a function of the bolomet-ric luminosity. It seems that as the bolometric luminosity in-creases, the disc component slightly increases in importance.However, the total numbers of objects at high luminositiesis small, as seen in the upper right panels, where the frac-tion is multiplied by the number of objects in the bin, so weshould be cautious about this finding.The lower panel shows this fraction for each of the ob-jects ranking from the smallest to biggest Hβ FWHM. Thuslow rank objects have the narrowest Hβ (and hence areby definition NLS1s). These also have the lowest black holemasses and highest Eddington ratios. They are more likelyto have a smaller fraction of their total luminosity emittedin the soft X-ray excess component, than the BLS1s. This Table 4.
The average black hole masses, as shown in Figure 10.NLS1 BLS1 ALL < M BH,IC > ± ± ± < M BH,BC > ± ± ± < M BH,IC + BC > ± ± ± < M BH,σ > ± ± ± < M BH,F IT > ± ± ± < M BH,RP > ± ± ± relates to the issue of the coronal radii, see Point (9) ofSection 6.2. There are also some BLS1s which have an ap-parently high fraction of power in their soft X-ray excesses,but they may also have alternative spectral fits includingreflection and/or absorption.We note that in all these plots the lower limit to thedisc fraction of 0 .
19 results from setting an upper limit of100 R g for the coronal radius parameter, as mentioned inSection 5.2 c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Figure 8.
The Balmer line parameter distributions. The first row is for H α and the second is for H β . We combine the intermediate andbroad components in each Balmer line profile to form the total broad line properties, giving values of the FWHM, EW and luminosity.The final panel shows the luminosity distribution of the narrow component for comparison. The distributions for the 12 NLS1s areindicated by the red regions, as in Figure 6. Figure 9.
The bolometric luminosity distribution for the differ-ent continuum components of the SED, i.e. accretion disc (green),Comptonisation (orange) and hard X-ray Comptonisation (blue).The upper left panel shows the percentage within each luminos-ity bin for each of these three SED components. The Upper rightpanel shows the luminosity distribution of the whole sample, witheach bin divided into three regions according to the fractional con-tribution from the different components in that luminosity bin.The lower panel shows how the contribution from each compo-nent changes as a function of rank order in H β FWHM, after thenarrow line component has been removed.
The black hole mass is one of the key parameters used in ourSED fitting, and it largely determines the continuum shapein the optical/UV region. The masses derived from rever-beration mapping are considered to be the most accurate,but the total number of objects which have been studied us-ing this technique is still relatively small (e.g. Peterson et al.2004; Denney et al. 2010; Bentz et al. 2010). In the absenceof reverberation mapping, the empirical relation betweenM BH and H β linewidth and L is often used as a proxyto estimate the black hole mass (Peterson et al. 2004). Aserious limitation of this method is that it is still not clearwhich specific measure of the H β profile provides the clos-est association with the velocity dispersion of the gas in thebroad line region.There are various alternative measures of the velocitywidth used for determining the black hole mass, includingthe FWHMs of the intermediate component (IC) and thebroad component (BC) (e.g. Zhu, Zhang & Tang 2009). Onecould also use the model independent second momentum(e.g. Peterson et al. 2004; Bian et al. 2008), or more sim-ply the FWHM of the H β line after subtracting the narrowcomponent (NC) (e.g. Peterson et al. 2004). The NC sub-tracted FWHM and the second momentum estimates oftenlie within the range of values covered by the IC and BCFWHMs, except for some peculiar objects such as those withbroad double-peaked profiles, for example UM 269. Given allthese uncertainties we decided to adopt the the best-fit blackhole mass obtained from the SED model, rather than sim-ply fixing it at a value determined from a specific linewidthmeasurement. Moreover, it is now suggested that radiationpressure may be important in modifying the black hole massderived using the relation between M BH and L and H β c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Figure 10.
A comparison of various methods used to derive blackhole mass. The total distributions are shown with the 12 NLS1sshow by the red regions. The purple dashed line indicate the av-erage black hole mass for the whole sample. The orange and cyandotted lines indicate the average masses of NLS1s and BLS1s,respectively. The average values are listed in Table 4. Values forindividual objects are listed in Table C1.
FWHM, especially for objects with high Eddington ratiossuch as most NLS1s (e.g. Marconi et al. 2008).In order to compare our results with those from otherstudies, we have made various estimates of black hole massesfor every source in our sample as follows:(1) M
BH,IC , M
BH,BC and M
BH,IC + BC are derived usingEquation 5 with different H β FWHMs obtained from ourBalmer line fitting procedure.(2) M
BH,σ is the black hole mass calculated from the secondmomentum of the total H β line profile (see Peterson et al.(2004) for details of the definition of ‘second momentum’),by using R BLR ∝ L . and a geometry factor of f =3 .
85. These assumptions are considered to be appropriatewhen using second momentum as a measure of the veloc-ity dispersion in BLR (Bentz et al. 2006; Collin et al. 2006;Bian et al. 2008).(3) M
BH,RP is the black hole mass corrected for radiationpressure, using equation (9) in Marconi et al. (2008) with f = 3 . log ( g ) = 7 .
6. We compare the black hole mass distributions obtained fromthese different methods in Figure 10. The mean values arelisted in Table 4.The M
BH,IC and M
BH,BC represent the two extremeestimates of black hole masses. The M
BH,IC could still beinfluenced by contamination from a NLR component, es-pecially for NLS1s where deconvolution of the narrow andbroad components is very difficult. If there is a residual nar-row line component, it will introduce a bias that under-estimates black hole masses. Conversely, using M
BH,BC ismore likely to bias towards higher black hole masses, due tothe presence of low contrast very broad wings often seenin H β profiles. We found FWHM IC + BC / σ Hβ =1.30 ± ± BH,σ than M
BH,IC + BC , but these two methods bothgive black hole masses between M BH,IC and M
BH,BC , withM
BH,IC + BC spanning a broader mass range.Our best-fit SED black hole masses (M BH,F IT ) are alsodistributed between M
BH,IC and M
BH,BC , with similar av-erage masses as M
BH,IC + BC (a comparison is shown in Fig-ure 11 Panel-A). Note that M BH,F IT is a free parameterin the SED fitting unless it hits the lower or upper lim-its set by M
BH,IC and M
BH,BC , which occasionally hap-pened (see Tabel C1). It is clearly shown in Figure 11 thatthe black hole masses from the SED fitting are not consis-tent with estimates based on either extremely narrow or ex-tremely broad lines. So for NLS1s, the mean M
BH,F IT is 0.36dex higher than M
BH,IC + BC ; while for BLS1s, the meanM BH,F IT is 0.22 dex lower than M
BH,IC + BC . Interestingly,this also implies that the M BH,F IT of NLS1s may have lessdeviation from the established M-Sigma relation than thatusing the M(L , FWHM Hβ ) relation as shown in severalprevious studies (e.g. Wang & Lu 2001; Bian & Zhao 2004;Zhou et al. 2006).The situation may be further complicated asMarconi et al. (2008) showed that NLS1s could beconsistent with the M- σ ∗ relation if a correction for ra-diation pressure is applied to black hole masses derivedfrom M(L , FWHM Hβ ). In our sample, correction forradiation pressure adds to the average M BH,IC + BC by0.67 dex for NLS1, 0.07 dex for BLS1 and 0.21 dex forthe whole sample. We also found a very similar massdistribution between M BH,RP and M
BH,F IT , except foran average of 0.36 dex higher in M
BH,RP . The differencesbetween the average mass of NLS1s and BLS1s are 0.78dex and 0.72 dex in M
BH,RP and M
BH,F IT , separately (seeFigure 11 Panel-B). Therefore, if M
BH,RP can provide agood match to the M-Sigma relation even down to low massNLS1s as proposed by Marconi et al. (2008), then our SEDdetermined M
BH,F IT may also give similar results. Thisimplies that the suggested deviation from the M- σ ∗ relationfor NLS1s may not be an intrinsic property, but rather aconsequence of using black hole estimates based on M(L ,FWHM Hβ ) relation, which may not be appropriate forNLS1s (e.g. Grupe & Mathur 2004; Komossa 2008). Elvis et al. (1994) constructed SED templates for bothradio-loud and radio-quiet AGN, based on a sample of 47quasars between redshift 0.025 and 0.94. VF07 modeled c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Figure 11.
Correlations of best-fit black hole mass (‘M BH -Fitting’ or ‘M BH,F IT ’) vs. H β FWHM determined black hole mass (‘M BH -H β FWHM’ or ‘M
BH,IC + BC ’) and vs. radiation pressure corrected black hole mass (‘M BH -Radiation Pressure’ or ‘M BH,RP ’). Redpoints represent the 12 NLS1s. The inserted panel in panel-A shows the distribution of the mass difference between M
BH,IC + BC andM BH,F IT , while the inserted panel in panel-B shows the distribution of the mass difference between M
BH,RP and M
BH,F IT . Red regionshighlight the distribution of NLS1s. optical-to-X-ray SED for a sample of 54 AGNs with red-shifts between 0.001 and 0.371, and showed that the SEDwas related to Eddington ratio. They also suggested that κ − keV is well correlated with Eddington ratio. In a laterstudy (Vasudevan & Fabian 2009) based on SED model-ing of 29 local AGNs from Peterson et al. (2004), the SEDdependence on Eddington ratio was reinforced. Recently,Lusso et al. (2010) studied 545 X-ray selected type 1 AGNover the redshift range of 0.04 to 4.25. They computed SEDsat different redshifts, and investigated α ox correlations withother parameters such as redshift, κ − keV , λ Edd etc.We present a mean SED for our sample which is sub-divided according to their Hβ FWHM. This gave three sub-samples, those with the narrowest lines, those with mod-erately broad lines, and those with very broad lines. Allobjects were de-redshifted to their local frame. First, eachof the best-fit SEDs was divided into 450 energy bins be-tween 1 eV and 100 keV. For each energy bin we calculatedthe monochromatic luminosity for the sub-sample with 12NLS1s, using their individual SED models. Then an aver-age value and standard deviation in each energy bin werecalculated in logarithm space. Thus a mean SED for the12 NLS1s was constructed. Using the same method for the12 moderate and 12 broadest line objects, their mean SEDswere produced. The total SED energy range is 1 eV to 100keV, but we note that only spectral ranges from 1.5-6 eV and0.3-10 keV are actually covered by the observational data,and all other ranges are based on model extrapolations.Obviously, limitations of our mean SEDs include therelatively small sample sizes composing the SEDs, and theredshift restriction z < β FWHM. As the line width increases,so the big blue bump (BBB) in the UV region becomesweaker relative to the hard X-rays, and its peak shifts to-wards lower energy. Also the spectral slope at high energiesbecomes harder.This evolution in spectral shape is similar to that foundby VF07 and Vasudevan & Fabian (2009), in which twomean SEDs of different mean Eddington ratio were com-pared. This relation might be expected since the FWHMand Eddington ratio are also strongly (anti)correlated in oursample. VF07 interpreted the spectral diversity as a scaledup version of the different accretion states of Galactic blackhole binaries. The low Eddington ratio AGN could be analo-gous to the low/hard state in black hole binaries in having aweak disc giving a strong high energy tail, and the high Ed- c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I dington ratio sources are analogous to the high/soft state, inwhich the disc emission dominates. Our SED templates donot extend down to such low Eddington ratios as in VF07,but we still see a similar behaviour. In this paper we presented a spectral study of 51 unob-scured Type 1 AGNs, including 12 NLS1s. We assembledX-ray data from the EPIC monitor on board the XMM-Newton satellite, and optical data from the SDSS DR7. Inaddition we added optical/UV data from the XMM-NewtonOM monitor when available. Our results confirm some pre-viously known correlations. For example, NLS1s often havesofter powerlaw fits from 2-10 keV, and have lower 2-10 keVluminosities. Their H α , H β and [OIII] λ Hα and Hβ line pro-files, with multi-components to deblend the narrow, inter-mediate and broad components by means of simultaneousmodeling of the FeII continuum and other blended lines. Wethen use results from the Hβ line fitting to constrain theblack hole mass. The FWHM of the intermediate and broadcomponents give a lower and upper limit for the mass, re-spectively. This supports previous studies which find thatNLS1s tend to have lower black hole masses and higher Ed-dington ratios, although their bolometric luminosities arenot significantly different from those of BLS1s.We include the Balmer continuum and permitted ironemission, and extend the modeling across the entire SDSSspectrum in order to isolate the intrinsic optical underlyingcontinuum. However, this pure optical continuum is often(in 32/51 objects) flatter than is predicted by the standardaccretion disc model. This could indicate some contamina-tion from the host galaxy, but the lack of stellar absorptionfeatures in most of the SDSS spectra suggests that this can-not be a general explanation. Instead it seems more likelythat there is an additional component in the optical regionrelated to the AGN, which is as yet not well understood.We also show that the Balmer continuum is not wellmodeled if the edge wavelength is fixed at its laboratoryvalue of 3646˚A. It is shifted redwards, and smoothed bymore than predicted by the FWHM of the Balmer emis-sion lines. These effects could both be produced by densitybroadening. Potentially more detailed models of the opti-cal emission could employ this as a new diagnostic tool forstudying the physical conditions e.g. electron density andtemperature, in the innermost Balmer emitting regions.The optical, UV and X-ray data were fitted using anew broadband SED model, which assumes that the gravi-tational potential energy is emitted as optically thick black-body emission at each radius down to some specific coronalradius. Below this radius the remaining energy down to thelast stable orbit is divided between a soft X-ray excess com-ponent and a hard X-ray tail. This energetically constrainsthe model fits in the unobservable EUV region. We constructthe resulting SEDs for each of the sources.A multi-component decomposition of the broadbandSED shows that relative contributions to the bolometricluminosity from the accretion disc, Compotonisation andpowerlaw components vary among sources with different lu- minosity and H β linewidth. We find a slight increase in con-tribution from the accretion disc as the luminosity increases,but a larger sample with more sources at both low and highluminosities is needed to confirm this.Our study also supports the distinctiveness of theNLS1s among the whole sample. We find that NLS1s tendto have a softer 2-10 keV spectrum, lower 2-10 keV lumi-nosity, lower black hole mass, higher Eddington ratio andhigher α ox index. However NLS1s do not stand out fromthe whole sample in terms of their bolometric luminositydistribution. We estimate the corona radii for every AGN inour sample from the SED fitting. This shows that on aver-age NLS1s have smaller corona radii, and correspondingly asmaller coronal component contribution.We compare the best-fit black hole masses with thosecorrected for radiation pressure, and other estimates of blackhole mass based on the R BLR -L relation, including nu-merous options for measuring the velocity width of the H β emission line. These results show that the black holes massesderived from SED fitting have a similar distribution to thatderived from profiles corrected for radiation pressure effects,except for an offset of 0.3 dex lower in both the NLS1 andBLS1 subsamples. The black hole mass difference betweenNLS1s and BLS1s from these two methods (i.e. SED fittingand radiation pressure corrected profiles) are both smallerthan inferred from other mass measurements. This impliesthat compared with black hole mass estimates based onlyon the H β FWHM, NLS1s may lie closer to the establishedM- σ ∗ relation at the low mass end, when their black holemasses are corrected for radiation pressure, and when weuse masses derived from our SED fitting.Finally, we form three broadband SED templates by co-adding SEDs in three subsamples (consist of 12 objects ineach) to examine how the broadband SED depends on Hβ FWHM velocity width, and by extension the Eddington ra-tio. The results show that there is a change in the SED shapeas the FWHM increases, with NLS1s having the largest bigblue bump in the extreme UV region. Other important pa-rameters such as Γ − keV , κ − keV and α ox , also change asthe H β FWHM increases. The implications of correlationsamong these parameters will be discussed in our next paper.
ACKNOWLEDGEMENTS
C. Jin acknowledges financial support through the award ofDurham Doctoral Fellowship. This work is partially basedon the data from SDSS, whose funding is provided by theAlfred P. Sloan Foundation, the Participating Institutions,the National Science Foundation, the U.S. Department ofEnergy, the National Aeronautics and Space Administra-tion, the Japanese Monbukagakusho, the Max Planck So-ciety, and the Higher Education Funding Council for Eng-land. This work is also partially based on observations ob-tained with XMM-Newton, an ESA science mission with in-struments and contributions directly funded by ESA Mem-ber States and the USA (NASA). We also used GALEXdata, which is based on observations made with the NASAGalaxy Evolution Explorer. GALEX is operated for NASAby the California Institute of Technology under NASA con-tract NAS5-98034. c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Figure 12.
The average SED of our sample. The panel on the left shows the averaged SED for the 12 NLS1s (including two marginalNLS1s, 2XMM 112328.0+052823 and 1E 1346+26.7). The average H β FWHM is 1400 ±
500 km s − . The red area indicates a onestandard deviation region on either side of the average spectrum. The central panel is for 12 objects with moderate line width. Theaverage FWHM is 3700 ±
600 km s − . The green region indicates one standard deviation. The panel on the right is the mean SED forthe 12 broadest line objects in our sample, including the one double-peak source. The average FWHM is 9800 ± − . We alsoshow the average value of the 2-10 keV powerlaw photon index, the 2-10 keV bolometric correction, and the α ox value with a one sigmaerror. D L on the Y-axis title is the luminosity distance. The unit of Y-axis is ‘keV (ergs s − keV − )’ in logarithm. The same arbitraryconstant of 1.31 × − is used for rescaling each plot. REFERENCES
Antonio Talavera .OMCal Team, 2009, Ap&SS, 320, 177Antonucci R., 1993, ARA&A, 31, 473Bentz M. C. et al., 2006, ApJ, 651, 775Bentz M. C. et al., 2010, ApJ, 716, 993Bessell M. S., 1991, A&A, 242, L17Bian W., Huang K., 2010, MNRAS, 401, 507Bian W., Zhao Y., 2004, MNRAS, 347, 607Bian W., Hu C., Gu Q., Wang J., 2008, MNRAS, 390, 752Boisson C., Joly M., Moultaka J., Pelat D., Serote RoosM., 2000, A&A, 357, 850Boller T., Brandt W. N., Fink H., 1996, A&A, 305, 53Boroson T. A., Green R. F., 1992, ApJS, 80, 109Bouchet P., Lequeux J., Maurice E., Prevot L., Prevot-Burnichon M. L., 1985, A&A, 149, 330Brocksopp C., Starling R. L. C., Schady P., Mason K. O.,Romero-Colmenero E., Puchnarewicz E. M., 2006, MN-RAS, 366, 953Calzetti D., Armus L., Bohlin R. C., Kinney A. L., Koorn-neef J., Storchi-Bergmann T., 2000, ApJ, 533, 682Canalizo G., Stockton A., 2001, ApJ, 555, 719Canfield R. C., Puetter R. C., 1981, ApJ, 243, 390Casebeer D. A., Leighly K. M., Baron E.,2006, ApJ, 637,157Collin-Souffrin S., Dumont A. M., 1990, A&A, 229, 292Collin S., Kawaguchi T., 2004, A&A, 426, 797Collin S., Kawaguchi T., Peterson B. M., Vestergaard M.,A&A, 456, 75C´ordova F. A., Kartje, J. F., Thompson R. J. Jr., Mason K.O., Puchnarewicz E. M., Harnden F. R. Jr., 1992, ApJS,81, 661Crenshaw D. M., Kraemer S. B., George I. M., 2003,Ann.Rev.A&A, 41, 117Crummy J., Fabian A. C., Gallo L., Ross R. R., 2006, MN-RAS, 365, 1067Czerny B., Elvis M., 1987, ApJ, 321, 305Czerny B., Niko lajuk M., R´o˙za´nska A, Dumont A.-M., Loska Z., Zycki P. T., 2003, A&A, 412, 317Denney K. D. et al., 2010, ApJ, 721, 715Done C., Gierli´nski M., 2005, Ap&&SS, 300, 167Done C., Kubota A., 2006, MNRAS, 371, 1216Done C., Nayakshin S., 2007, MNRAS, 377, L59Done C., Davis S., Jin C., Blaes O., Ward M., MNRAS,2011, accepted, arXiv:1107.5429v1Elvis M. et al., 1994, ApJS, 95, 1Fitzpatrick E. L., 1986, AJ, 92, 1068Giveon Uriel, Maoz D., Kaspi S., Netzer H., Smith P. S.,1999, MNRAS, 306 637Gierli´nski M., Done C., 2004, MNRAS, 349, L7Gierli´nski M., Middleton M., Ward M., Done C., 2008, Na-ture, 455, 369Gierli´nski M., Zdziarski A. A., Poutanen J., Coppi P. S.,Ebisawa K., Johnson W. N., 1999, MNRAS, 309, 496Gladstone J., Roberts T., Done C., 2009, MNRAS, 397,1836Goodrich R. W., 1989, ApJ, 342, 224Grandi S. A., 1982, ApJ, 255, 25Grupe D., Mathur S., ApJ, 606, L41Grupe D., Beuermann K., Thomas H. C., Mannheim K.,Fink H. H., 1998, A&A, 330, 25Grupe D., Beuermann K., Mannheim K., Thomas H. C.,1999, A&A, 350, 805Hao H. et al., 2010, ApJ, 724, L59Haardt F., Maraschi L., 1991, ApJ, 380, L51Hu C., Wang J., Ho L. C., Chen Y., Bian W., Xue S., 2008,ApJ, 683, L115Hubbard E. N., Puetter R. C., 1985, ApJ, 290, 394Jin C., Done C., Ward M., Gierli´nski M., Mullaney J., 2009,MNRAS, 398, L16Kalberla P. M. W., Burton W. B., Hartmann Dap, Arnal E.M., Bajaja E., Morras R., P¨oppel W. G. L., 2005, A&A,440, 775Kaspi S., Smith P. S., Netzer H., Maoz D., Jannuzi B.,Giveon U., 2000, ApJ, 533, 631Kinney A. L., Calzetti D., Bohlin R. C., McQuade K., c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Storchi-Bergmann T., Schmitt H. R., 1996, ApJ, 467, 38Komossa S., 2008, RMxAC, 32, 86Kwan J., 1984, ApJ, 283, 70Kwan J., Krolik J. H., 1981, ApJ, 250, 478Lacy M., Sajina A., Petric A. O., Seymour N., Canalizo G.,Ridgway S. E., Armus L., Storrie-Lombardi L. J., 2007,ApJ, 669, L61Landt H., Elvis M., Ward M. J., Bentz M. C., Korista K.T., Karovska M., 2011, MNRAS, 414, 218Lee J. C., Ogle P. M., Canizares C. R., Marshall H. L.,Schulz N. S., Morales R., Fabian A. C., Iwasawa K., 2001,ApJ, 554, L13Leighly K. M., 1999, ApJS, 125, 317Lusso E. et al., 2010, A&A, 512, 34Mannucci F., Basile F., Poggianti B. M., Cimatti A., DaddiE., Pozzetti L., Vanzi L., 2001, MNRAS, 326, 745Maoz D. et al., 1993, ApJ, 404, 576Marconi A., Axon D. J., Maiolino R., Nagao T., PastoriniG., Pietrini P., Robinson A., Torricelli G., 2008, ApJ, 678,693Mei L., Yuan W., Dong X., 2009, RAA, 9, 269Middleton M., Done C., 2010, MNRAS, 403, 9Middleton M., Done C., Gierli´nski M., 2007, MNRAS, 381,1426Middleton M., Done C., Ward M., Gierli´nski M., SchurchN., 2009, MNRAS, 394,250Mihalas D., 1978, Stellar Atmospheres, (2nd ed.; San Fran-cisco, CA: Freeman), 650Miller L., Turner T. J., Reeves J. N., 2008, A&A, 483, 437Mullaney J. R., Ward M. J., Done C., Ferland G. J.,Schurch N., 2009, MNRAS, 394, L16Nandra K., Pounds K. A., 1994, MNRAS, 268, 405Osterbrock D. E., 1989, Astrophysics of Gaseous Nebu-lae and Active Galacitc Nuclei (University Science Books,Mill Valley, California)Osterbrock D. E., Pogge R. W., 1985, ApJ, 297, 166Peterson B. M. et al., 2004, ApJ, 613, 682Pigarov A. Y., Terry J. L., Lipschultz B., 1998, PlasmaPhysics and Controlled Fusion, 40, 12Prevot M. L., Lequeux J., Prevot L., Maurice E., Rocca-Volmerange B., 1984, A&A, 132, 389Puchnarewicz E. M. et al., 1992, MNRAS, 256, 589Puetter R. C., Levan P. D., 1982, ApJ, 260, 44Schurch N. J., Done C., 2006, MNRAS, 371, 81Seaton M. J., 1979, MNRAS, 187, 73Shuder J. M., Osterbrock D. E., 1981, ApJ, 250, 55Sim S. A., Long K. S., Miller L., Turner T. J., 2008, MN-RAS, 388, 611Soria R., Puchnarewicz E. M., 2002, MNRAS, 329, 456Stephens S. A., 1989, AJ, 97, 10Sulentic J. W., Zwitter T., Marziani P., Dultzin-HacyanD., 2000, ApJ, 536, L5Tsuzuki Y., Kawara K., Yoshii Y., Oyabu S., Tanab´e T.,Matsuoka Y., 2006, ApJ, 650, 57Turner A. K., Fabian A. C., Lee J. C., Vaughan S., 2004,MNRAS, 353, 319Turner A. K., Miller L., Reeves J. N., Kraemer S. B., 2004,A&A, 475, 121Vasudevan R. V., Fabian A. C., 2007, MNRAS, 381, 1235Vasudevan R. V., Fabian A. C., 2009, MNRAS, 392, 1124Verner E., Bruhweiler F., Johansson S., Peterson B., 2009,Physica Scripta, 2009, T134 V´eron-Cetty M. P., Joly M., V´eron P., 2004, A&A, 417,515Walter R., Fink H. H., 1993, A&A, 274, 105Wang T., Lu Y., 2001, A&A, 377, 52Ward M., Elvis M., Fabbiano G., Carleton N. P., WillnerS. P., Lawrence A., 1987, ApJ, 315, 74Wills B. J., Netzer H., Wills D., 1985, ApJ, 288, 94Woo J., Urry C. M., 2002, ApJ, 579, 530Zdziarski A. A., Poutanen J., Johnson W. N., 2000, ApJ,542, 703Zhou H., Wang T., Yuan W., Lu H., Dong X., Wang J., LuJ., 2006, ApJS, 166, 128Zhu L., Zhang S., Tang S., 2009, ApJ, 700, 1173 c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord −3 − − − − . . Energy (keV) nH_gal = 1.79 E+20nH_int = 0.00 E+20 E F E (No:01 - a) (No:01 - b) UM 269 (No:01 - c) −3 − − − . . Energy (keV) nH_gal = 2.43 E+20nH_int = 1.05 E+20 E F E (No:02 - a) (No:02 - b) MRK 1018 (No:02 - c) −3 − − − . . Energy (keV) nH_gal = 6.31 E+20nH_int = 9.88 E+20 E F E (No:03 - a) (No:03 - b) NVSS J030639 (No:03 - c) Figure A1.
The spectral fitting results. Object order follows all other tables in this paper as increasing RA and DEC. 1. BroadbandSED fitting plot (panel-a): X-ray data has been rebinned for each object. Green solid line is the pure accretion disc component peakingat optical/UV region, orange line is Comptonisation component producing soft X-ray excess below 2 keV, blue line is the hard X-rayComptonisation component dominating 2-10 keV spectrum, and red is the total broadband SED model. 2. SDSS spectrum fitting plot(panel-b): only the fitted spectrum below 7000˚A is plotted. Green solid line is the best-fit underlying continuum from accretion disc.Orange line shows all best-fit emission lines, including the results from detailed Balmer line fitting in panel-c. FeII emission is plotted aslight blue, while Balmer continuum being dark blue. The total best-fit model with reddening is drawn in red solid line. 3. Balmer emissionline fitting plot(panel-c): spectral ranges containing H α and H β profiles are plotted separately, with blue lines showing individual linecomponents and red line showing the whole best-fit model. These are also the corresponding zoom-in plots of nearby regions of H α andH β in panel-b. The given black hole mass is the broadband SED best-fit value, see Section 5 for detailed descriptions. APPENDIX A: THE SPECTRAL MODELLING RESULTS c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I −3 − − − − . . Energy (keV) nH_gal = 3.49 E+20nH_int = 2.81 E+20 E F E (No:04 - a) (No:04 - b) 2XMM J074601.2+280732 (No:04 - c) −3 − − − − . . Energy (keV) nH_gal = 3.53 E+20nH_int = 4.03 E+20 E F E (No:05 - a) (No:05 - b) 2XMM J080608.0+244421 (No:05 - c) −3 − − − − . . Energy (keV) nH_gal = 4.24 E+20nH_int = 0.00 E+20 E F E (No:06 - a) (No:06 - b) HS 0810+5157 (No:06 - c) −3 − − − . . Energy (keV) nH_gal = 1.33 E+20nH_int = 3.74 E+20 E F E (No:07 - a) (No:07 - b) RBS 0769 (No:07 - c) Figure A1. continued c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord −3 − − . . Energy (keV) nH_gal = 3.12 E+20nH_int = 7.34 E+20 E F E (No:08 - a) (No:08 - b) RBS 0770 (No:08 - c) −3 − − . . Energy (keV) nH_gal = 1.30 E+20nH_int = 1.35 E+20 E F E (No:09 - a) (No:09 - b) MRK 0110 (No:09 - c) −3 − − − . . Energy (keV) nH_gal = 1.74 E+20nH_int = 0.00 E+20 E F E (No:10 - a) (No:10 - b) PG 0947+396 (No:10 - c) −3 − − − − . . Energy (keV) nH_gal = 1.72 E+20nH_int = 1.99 E+20 E F E (No:11 - a) (No:11 - b) 2XMM J100025.2+015852 (No:11 - c) Figure A1. continued c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I −3 − − − − . . Energy (keV) nH_gal = 1.20 E+20nH_int = 1.08 E+20 E F E (No:12 - a) (No:12 - b) 2XMM J100523.9+410746 (No:12 - c) −3 − − − − . . Energy (keV) nH_gal = 3.56 E+20nH_int = 0.00 E+20 E F E (No:13 - a) (No:13 - b) PG 1004+130 (No:13 - c) −3 − − − . . Energy (keV) nH_gal = 1.76 E+20nH_int = 0.00 E+20 E F E (No:14 - a) (No:14 - b) RBS 0875 (No:14 - c) −3 − − − . . Energy (keV) nH_gal = 1.31 E+20nH_int = 2.42 E+20 E F E (No:15 - a) (No:15 - b) KUG 1031+398 (No:15 - c) Figure A1. continued c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord −3 − − − . . Energy (keV) nH_gal = 1.70 E+20nH_int = 0.65 E+20 E F E (No:16 - a) (No:16 - b) PG 1048+342 (No:16 - c) −3 − − − − . . Energy (keV) nH_gal = 0.64 E+20nH_int = 0.84 E+20 E F E (No:17 - a) (No:17 - b) 1RXS J111007 (No:17 - c) −3 − − − . . Energy (keV) nH_gal = 1.45 E+20nH_int = 0.19 E+20 E F E (No:18 - a) (No:18 - b) PG 1115+407 (No:18 - c) −3 − − − − . . Energy (keV) nH_gal = 3.70 E+20nH_int = 1.40 E+20 E F E (No:19 - a) (No:19 - b) 2XMM J112328.0+052823 (No:19 - c) Figure A1. continued c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I −3 − − − . . Energy (keV) nH_gal = 1.91 E+20nH_int = 4.76 E+20 E F E (No:20 - a) (No:20 - b) RX J1140.1+0307 (No:20 - c) −3 − − − . . Energy (keV) nH_gal = 1.77 E+20nH_int = 0.00 E+20 E F E (No:21 - a) (No:21 - b) PG 1202+281 (No:21 - c) −3 − − − − . . Energy (keV) nH_gal = 2.75 E+20nH_int = 8.83 E+20 E F E (No:22 - a) (No:22 - b) 1AXG J121359+1404 (No:22 - c) −3 − − − − . . Energy (keV) nH_gal = 1.59 E+20nH_int = 0.00 E+20 E F E (No:23 - a) (No:23 - b) 2E 1216+0700 (No:23 - c) Figure A1. continued c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord −3 − − − − . . Energy (keV) nH_gal = 1.63 E+20nH_int = 7.32 E+20 E F E (No:24 - a) (No:24 - b) 1RXS J122019 (No:24 - c) −3 − − − . . Energy (keV) nH_gal = 2.34 E+20nH_int = 0.00 E+20 E F E (No:25 - a) (No:25 - b) LBQS 1228+1116 (No:25 - c) −3 − − − − . . Energy (keV) nH_gal = 2.31 E+20nH_int = 7.24 E+20 E F E (No:26 - a) (No:26 - b) 2XMM J123126.4+105111 (No:26 - c) −3 − − − . . Energy (keV) nH_gal = 2.75 E+20nH_int = 0.00 E+20 E F E (No:27 - a) (No:27 - b) MRK 0771 (No:27 - c) Figure A1. continued c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I −3 − − − − . . Energy (keV) nH_gal = 1.45 E+20nH_int = 6.20 E+20 E F E (No28 - a) (No:28 - b) RX J1233.9+0747 (No:28 - c) −3 − − − − . . Energy (keV) nH_gal = 1.18 E+20nH_int = 1.36 E+20 E F E (No:29 - a) (No:29 - b) RX J1236.0+2641 (No:29 - c) −3 − − . . Energy (keV) nH_gal = 1.87 E+20nH_int = 2.64 E+20 E F E (No:30 - a) (No:30 - b) PG 1244+026 (No:30 - c) −3 − − − − . . Energy (keV) nH_gal = 0.83 E+20nH_int = 0.00 E+20 E F E (No:31 - a) (No:31 - b) 2XMM J125553.0+272405 (No:31 - c) Figure A1. continued c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord −3 − − − . . Energy (keV) nH_gal = 0.90 E+20nH_int = 0.14 E+20 E F E (No:32 - a) (No:32 - b) RBS 1201 (No:32 - c) −3 − − − − . . Energy (keV) nH_gal = 1.07 E+20nH_int = 0.81 E+20 E F E (No:33 - a) (No:33 - b) 2XMM J132101.4+340658 (No:33 - c) −3 − − − − . . Energy (keV) nH_gal = 1.83 E+20nH_int = 0.93 E+20 E F E (No:34 - a) (No:34 - b) 1RXS J132447 (No:34 - c) −3 − − − . . Energy (keV) nH_gal = 1.76 E+20nH_int = 0.89 E+20 E F E (No:35 - a) (No:35 - b) UM 602 (No:35 - c) Figure A1. continued c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I −3 − − − . . Energy (keV) nH_gal = 1.18 E+20nH_int = 3.94 E+20 E F E (No:36 - a) (No:36 - b) 1E 1346+26.7 (No:36 - c) −3 − − − . . Energy (keV) nH_gal = 1.82 E+20nH_int = 0.00 E+20 E F E (No:37 - a) (No:37 - b) PG 1352+183 (No:37 - c) −3 − − − − . . Energy (keV) nH_gal = 1.42 E+20nH_int = 0.37 E+20 E F E (No:38 - a) (No:38 - b) MRK 0464 (No:38 - c) −3 − − − − . . Energy (keV) nH_gal = 1.36 E+20nH_int = 4.77 E+20 E F E (No:39 - a) (No:39 - b) 1RXS J135724 (No:39 - c) Figure A1. continued c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord −3 − − − . . Energy (keV) nH_gal = 0.77 E+20nH_int = 5.20 E+20 E F E (No:40 - a) (No:40 - b) PG 1415+451 (No:40 - c) −3 − − − . . Energy (keV) nH_gal = 1.81 E+20nH_int = 0.00 E+20 E F E (No:41 - a) (No:41 - b) PG 1427+480 (No:41 - c) −3 − − − . . Energy (keV) nH_gal = 2.86 E+20nH_int = 3.28 E+20 E F E (No:42 - a) (No:42 - b) NGC 5683 (No:42 - c) −3 − − − . . Energy (keV) nH_gal = 2.69 E+20nH_int = 0.00 E+20 E F E (No:43 - a) (No:43 - b) RBS 1423 (No:43 - c) Figure A1. continued c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I −3 − − . . Energy (keV) nH_gal = 2.78 E+20nH_int = 5.89 E+20 E F E (No:44 - a) (No:44 - b) PG 1448+273 (No:44 - c) −3 − − − . . Energy (keV) nH_gal = 1.46 E+20nH_int = 0.00 E+20 E F E (No:45 - a) (No:45 - b) PG 1512+370 (No:45 - c) −3 − − − − . . Energy (keV) nH_gal = 4.02 E+20nH_int = 0.54 E+20 E F E (No:46 - a) (No:46 - b) Q 1529+050 (No:46 - c) −3 − − − . . Energy (keV) nH_gal = 3.78 E+20nH_int = 16.6 E+20 E F E (No:47 - a) (No:47 - b) 1E 1556+27.4 (No:47 - c) Figure A1. continued c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord −3 − − − . . Energy (keV) nH_gal = 2.11 E+20nH_int = 0.86 E+20 E F E (No:48 - a) (No:48 - b) MRK 0493 (No:48 - c) −3 − − − . . Energy (keV) nH_gal = 4.90 E+20nH_int = 0.36 E+20 E F E (No:49 - a) (No:49 - b) II Zw 177 (No:49 - c) −3 − − − . . Energy (keV) nH_gal = 4.51 E+20nH_int = 0.00 E+20 E F E (No:50 - a) (No:50 - b) PG 2233+134 (No:50 - c) −3 − − − . . Energy (keV) nH_gal = 2.91 E+20nH_int = 1.31 E+20 E F E (No:51 - a) (No:51 - b) MRK 0926 (No:51 - c) Figure A1. continued c (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I APPENDIX B: XMM-NEWTON AND SDSS DR7 SOURCE POSITION AND SEPERATION OF OURSAMPLE c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Table B1.
XMM-Newton and SDSS DR7 source position and separation of our sample. ID: object number, the same asTable 1; XMM Ra and XMM Dec: source’s right ascension and declination in the corresponding XMM-Newton observa-tion; XMM PosErr: X-ray position uncertainty from XMM-Newton; SDSS Ra and SDSS Dec: source’s right ascension anddeclination measured by SDSS; Separation: the angular separation between source’s XMM-Newton and SDSS coordinates;Sep./XMM PosErr: the ratio between coordinates separation and X-ray position uncertainty, showing the significance ofcoordinate separation.ID XMM Ra XMM Dec XMM PosErr SDSS Ra SDSS Dec Separation Sep./XMM PosErr degree degree arcsec degree degree arcsec (cid:13) , 000–000
Spectral Study of Unobscured Type 1 AGN - I APPENDIX C: BLACK HOLE MASSES FROM DIFFERENT METHODS c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Table C1.
Black hole masses from different methods. M
BH,IC : black hole mass calculated from the FWHM of H β in-termediate component in logarithm and solar mass; M BH,BC : black hole mass calculated from the FWHM of H β broadcomponent; M BH,IC + BC : black hole masses calculated from the FWHM of superposing H β intermediate component (IC)and broad component (BC) (i.e. narrow component subtracted), using Equation 5; M BH,σ : black hole mass calculated fromthe second momentum of the whole H β line profile, see subsection 6.5 for details; M BH,F it : the best-fit black hole massesin logarithm, which is constrained by M
BH,IC and M
BH,BC , but values within 0.5 lower than log(M
BH,IC ) were alsoadopted in the fitting, see subsection 6.5; log(M
BH,RP ): the radiation pressure corrected black hole mass using Equation 9in Marconi et al. (2008) with f =3.1 and log g =7.6; (*): note that M BH,IC + BC is always within the range of M BH,IC andM
BH,BC , except for UM269 whose H β shows double-peak profile.ID Common Name M BH,IC M BH,BC M BH,IC + BC M BH, σ M BH,F it M BH,RP log , M ⊙ log , M ⊙ log , M ⊙ log , M ⊙ log , M ⊙ log , M ⊙ ∗ (cid:13) , 000–000 Spectral Study of Unobscured Type 1 AGN - I Table D1.
Emission line parameters for the whole sample. Narrow component (NC), intermediate component (IC), broad component(BC) and intermediate plus broad component (I+B) are shown separately for H α and H β . IC and BC are both Gaussian, while NCmay have the same profile as the whole [OIII] λ λ λ λ λ λ λ kms − .The velocity of ’NC’ is small and may come from the redshift uncertainty in Sloan’s final redshift measurement, and thus should notbe taken seriously. FWHMs of ’NC’, ’IC’ and ’BC’ are directly from the Gaussian profile parameters. FWHM for ’I+B’ is measureddirectly from the superposed model profile. The numbers are all in kms − . ’lum’ and ’ew’ means luminosity in Log ( ergss − ) andequivalent width in ˚A.ID H α H β [OIII] 5007 HeII FeVII FeX vel fwhm lum ew vel fwhm lum ew vel fwhm lum ew lum lum lum d NC — 456 42.09 21 — 457 41.46 3.1 — — — — — 42.3 —(1) f IC -3700 6080 43.18 270 -3700 6080 42.67 50 47 203 41.49 3.5 — — — BC I+B — 13000 43.51 570 — 13000 43.00 110 — 462 42.27 21 — — —2 NC — 405 40.77 5.0 — 401 40.13 0.94 — — — — 41.1 41.0 —(1) f IC BC I+B — 4330 42.21 140 — 6220 41.82 45 — 401 41.23 12 — — —3 NC — 470 41.84 42 — 469 41.17 6.4 — — — — 41.3 41.3 41.2(1) f IC
120 2040 42.36 140 120 2040 41.86 32 52 396 41.59 17 — — — BC I+B — 2190 42.53 210 — 2310 42.14 59 — 468 41.85 32 — — —4 ∗ NC — 251 40.89 3.5 — 249 40.35 0.81 -34 239 40.82 2.5 40.8 — —(1) f IC -400 5260 42.23 77 -400 5260 41.31 7.4 140 192 41.23 6.2 — — — BC -450 13700 42.29 86 -450 13700 42.20 57 -110 1370 40.87 2.7 — — — I+B — 6500 42.56 160 — 10800 42.25 64 — 248 41.49 11 — — —5 NC — 326 41.76 13 — 325 41.01 1.6 — — — — 41.9 41.9 —(2) f IC
190 2360 43.15 320 190 2360 42.56 58 -23 325 41.71 8.4 — — — BC
900 9030 42.96 200 900 9030 42.52 53 -150 794 42.01 17 — — —
I+B — 2610 43.37 520 — 2720 42.84 110 — 474 42.19 25 — — —6 NC — 461 42.66 51 — 457 41.88 4.7 — — — — 42.4 — —(1) f IC
400 3980 43.40 280 400 3980 42.86 45 -76 387 42.25 12 — — — BC I+B — 4930 43.77 650 — 5430 43.34 140 — 462 42.52 22 — — —7 NC — 580 41.88 38 — 575 41.38 8.0 — — — — 41.5 — —(2) f IC
160 1570 42.10 62 160 1570 41.60 13 -160 580 40.96 3.2 — — — BC -120 4300 41.99 48 -120 4300 41.70 17 -610 1200 41.27 6.5 — — — I+B — 1820 42.35 110 — 1980 41.95 30 — 1030 41.45 9.7 — — —8 ∗ NC — 442 41.75 78 — 445 41.08 12 -27 300 41.10 13 41.6 41.0 41.0(1) f IC
73 2360 42.28 260 73 2360 41.68 48 -160 663 41.26 19 — — — BC
920 8080 41.97 130 920 8080 41.72 53 -170 1390 41.03 11 — — —
I+B — 2580 42.45 390 — 2840 42.00 100 — 444 41.62 43 — — —9 ∗ NC — 317 41.46 150 — 319 40.81 34 -10. 297 41.66 240 40.6 40.4 —(1) f IC
200 2360 42.02 530 200 2360 40.99 50 150 1450 40.44 14 — — — BC
840 7230 41.48 150 840 7230 41.14 72 -41 509 41.10 64 — — —
I+B — 2500 42.13 680 — 3030 41.37 120 — 316 41.79 320 — — —
APPENDIX D: EMISSION LINE FITTING PARAMETERS OF THE WHOLE SAMPLE c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Table D1. continuedID H α H β [OIII] 5007 HeII FeVII FeX vel fwhm lum ew vel fwhm lum ew vel fwhm lum ew lum lum lum NC — 405 42.06 18 — 401 41.13 1.3 — — — — 41.9 42.0 41.2(1) f IC
180 4060 43.45 430 180 4060 42.87 70 28 330 41.96 8.9 — — — BC I+B — 4440 43.62 630 — 4810 43.15 130 — 407 42.27 18 — — —11 NC — 279 42.03 47 — 281 41.38 5.8 — — — — 41.7 — —(1) f IC
230 3440 42.89 340 230 3440 42.11 31 -43 249 42.06 29 — — — BC
730 14100 43.10 550 730 14100 42.66 110 -82 634 41.77 15 — — —
I+B — 4360 43.31 900 — 5640 42.77 140 — 279 42.24 44 — — —12 NC — 396 41.55 36 — 394 40.60 2.8 — — — — 41.6 — —(1) f IC -76 3850 42.53 350 -76 3850 41.90 57 36 1590 40.92 6.2 — — — BC -140 12600 42.11 130 -140 12600 41.77 43 -23 372 41.33 16 — — — I+B — 4130 42.67 480 — 4390 42.14 100 — 395 41.47 22 — — —13 NC — 396 42.50 10. — 395 41.94 1.6 — — — — 42.8 42.3 —(1) f IC
580 6160 43.53 110 580 6160 42.85 13 -44 304 42.16 2.8 — — — BC
700 17200 43.62 140 700 17200 43.46 52 -250 1190 42.37 4.5 — — —
I+B — 7730 43.88 240 — 10800 43.55 65 — 395 42.58 7.3 — — —14 NC — 298 41.97 23 — 300 41.48 5.8 — — — — 42.0 41.9 —(1) f IC
300 5580 43.31 520 300 5580 42.59 74 6.0 893 42.10 25 — — — BC I+B — 5940 43.43 680 — 7060 42.95 170 — 297 42.52 65 — — —15 NC — 342 41.03 30 — 345 40.43 7.4 — — — — 40.7 40.0 40.4(2) f IC
58 918 41.22 46 58 918 40.73 15 53 299 40.84 19 — — — BC -20 4400 40.97 26 -20 4400 40.48 8.3 -320 1000 40.74 15 — — — I+B — 994 41.42 73 — 987 40.92 23 — 340 41.09 34 — — —16 NC — 279 41.62 9.9 — 281 40.98 1.5 — — — — 42.1 41.4 —(2) f IC
37 2810 43.14 330 37 2810 42.47 46 -55 968 41.53 5.4 — — — BC I+B — 3080 43.32 500 — 3560 42.84 110 — 297 42.13 22 — — —17 ∗ NC — 608 42.00 37 — 607 41.08 2.9 -28 617 41.84 17 41.7 41.8 41.7(3) f IC
160 1930 42.91 300 160 1930 42.35 53 140 235 41.14 3.4 — — — BC
430 7450 42.67 170 430 7450 42.34 51 -530 1600 41.57 9.2 — — —
I+B — 2120 43.10 470 — 2250 42.64 100 — 607 42.08 30 — — —18 NC — 400 42.23 32 — 401 41.53 3.6 — — — — 42.1 — 41.9(1) f IC
220 1810 43.02 200 220 1810 42.44 29 -60 260 41.28 2.1 — — — BC
360 7220 42.93 160 360 7220 42.66 48 -210 701 41.61 4.6 — — —
I+B — 2060 43.28 360 — 2310 42.86 76 — 401 41.78 6.7 — — —19 NC — 188 40.90 13 — 191 40.25 2.3 — — — — 41.0 40.4 40.5(2) f IC
46 1500 41.77 96 46 1500 41.16 19 75 186 40.93 11 — — — BC -54 5010 41.66 74 -54 5010 41.40 33 -100 445 40.80 8.4 — — — I+B — 1730 42.02 170 — 2000 41.60 51 — 223 41.17 20 — — —20 NC — 232 40.64 14 — 230 40.03 3.1 — — — — 41.2 — 39.8(2) f IC
29 578 41.12 40 29 578 40.51 9.5 43 234 40.29 5.8 — — — BC
120 1970 41.12 41 120 1970 40.76 17 -230 460 39.80 1.9 — — —
I+B — 686 41.42 82 — 774 40.95 27 — 254 40.41 7.6 — — —21 NC — 354 42.05 49 — 357 41.53 11 — — — — — 41.5 —(2) f IC -250 3990 43.03 470 -250 3990 42.14 44 110 357 42.29 64 — — — BC
730 13900 42.93 380 730 13900 42.58 120 -200 1160 42.27 63 — — —
I+B — 4570 43.29 840 — 6090 42.71 170 — 419 42.58 130 — — —22 NC — 317 40.86 6.5 — 319 40.08 0.86 — — — — 41.5 — —(2) f IC
750 4700 42.28 170 750 4700 41.59 28 -19 316 40.92 6.0 — — — BC
59 12600 42.18 130 59 12600 41.98 68 -150 764 40.90 5.8 — — —
I+B — 5510 42.53 300 — 7050 42.13 96 — 389 41.21 12 — — —c (cid:13) , 000–000
Spectral Study of Unobscured Type 1 AGN - I Table D1. continuedID H α H β [OIII] 5007 HeII FeVII FeX vel fwhm lum ew vel fwhm lum ew vel fwhm lum ew lum lum lum NC — 340 41.56 39 — 344 40.79 5.0 — — — — 41.0 — 41.0(1) f IC
130 1700 42.15 150 130 1700 41.66 37 31 256 40.59 3.2 — — — BC
340 5970 41.92 89 340 5970 41.60 33 -79 930 40.81 5.4 — — —
I+B — 1890 42.35 240 — 1980 41.93 70 — 340 41.01 8.7 — — —24 NC — 576 41.37 3.9 — 575 41.04 1.6 — — — — 42.3 42.2 —(1) f IC
150 4700 42.74 91 150 4700 41.59 5.5 44 327 41.00 1.4 — — — BC I+B — 7280 43.32 350 — 13900 42.93 120 — 577 41.64 6.2 — — —25 NC — 400 42.49 47 — 395 41.79 5.6 — — — — 42.1 — —(1) f IC -74 3820 43.36 350 -74 3820 42.72 47 29 365 42.51 31 — — — BC -87 12700 43.22 250 -87 12700 42.92 75 -99 943 42.13 13 — — — I+B — 4350 43.59 590 — 4980 43.13 120 — 395 42.66 43 — — —26 NC — 543 42.16 88 — 544 41.63 16 — — — — 41.5 — —(2) f IC
44 1540 42.53 210 44 1540 42.05 43 -40 545 41.71 21 — — — BC
490 6770 42.42 160 490 6770 41.92 32 -330 1630 41.21 6.6 — — —
I+B — 1730 42.78 370 — 1720 42.29 76 — 577 41.83 27 — — —27 ∗ NC — 255 40.98 7.6 — 256 40.48 1.8 49 254 41.53 20 41.5 40.9 40.9(2) f IC -150 2500 42.40 200 -150 2500 41.64 26 -130 504 41.15 8.6 — — — BC
780 7780 42.43 220 780 7780 42.17 86 -350 1240 41.00 6.1 — — —
I+B — 3030 42.72 420 — 4310 42.28 110 — 291 41.76 35 — — —28 NC — 368 42.31 55 — 369 41.74 11 — — — — 42.7 — —(1) f IC BC -690 10900 42.75 150 -690 10900 42.52 68 -21 1530 41.94 18 — — — I+B — 3610 43.19 420 — 4240 42.74 110 — 365 42.71 110 — — —29 NC — 497 42.03 34 — 494 41.40 5.6 — — — — 41.8 41.5 41.0(1) f IC
12 2810 42.87 240 12 2810 42.30 44 39 385 41.69 11 — — — BC
810 9770 42.71 170 810 9770 42.46 64 -160 789 41.75 13 — — —
I+B — 3170 43.10 400 — 3560 42.69 110 — 498 42.02 24 — — —30 NC — 386 41.31 39 — 388 40.66 5.1 — — — — 41.1 39.8 40.3(1) f IC
110 808 41.43 52 110 808 41.09 14 -19 336 40.97 11 — — — BC
110 3040 41.41 49 110 3040 41.10 14 -210 703 40.74 6.5 — — —
I+B — 943 41.72 100 — 953 41.39 28 — 389 41.17 17 — — —31 NC — 382 41.61 8.0 — 382 40.54 0.51 — — — — 41.2 — —(3) f IC
610 5730 43.14 270 610 5730 42.59 57 20 333 41.29 3.0 — — — BC
850 21500 42.75 110 850 21500 42.62 61 -460 814 41.17 2.3 — — —
I+B — 6140 43.29 380 — 6810 42.90 120 — 377 41.54 5.2 — — —32 NC — 425 41.24 23 — 424 40.61 5.3 — — — — 41.1 41.0 40.3(1) f IC
140 2550 42.17 200 140 2550 41.48 39 -58 388 41.51 42 — — — BC I+B — 2710 42.30 260 — 3100 41.80 82 — 419 41.65 58 — — —33 NC — 423 41.68 16 — 426 41.03 2.7 — — — — 41.6 — —(1) f IC
17 4700 42.92 290 17 4700 42.37 59 100 412 41.99 25 — — — BC
23 16700 42.66 160 23 16700 42.44 68 -610 1370 41.42 6.8 — — —
I+B — 5170 43.11 450 — 5690 42.71 130 — 425 42.09 32 — — —34 NC — 838 42.57 44 — 836 41.70 4.0 — — — — 41.8 — —(3) f IC
260 2470 43.06 130 260 2470 42.52 26 -150 1440 42.48 25 — — — BC I+B — 2910 43.32 250 — 3310 42.93 67 — 829 42.70 41 — — —35 NC — 572 42.67 130 — 576 42.01 22 — — — — 42.0 41.8 42.1(1) f IC
270 2380 43.31 560 270 2380 42.65 96 -120 1250 42.28 43 — — — BC I+B — 2640 43.49 840 — 2790 42.90 170 — 571 42.74 120 — — —c (cid:13) , 000–000 C. Jin, M. Ward, C. Done and J. M. Gelbord
Table D1. continuedID H α H β [OIII] 5007 HeII FeVII FeX vel fwhm lum ew vel fwhm lum ew vel fwhm lum ew lum lum lum ∗ NC — 274 41.15 36 — 271 40.32 5.1 54 175 40.54 8.6 41.1 40.2 39.6(1) f IC
210 1540 41.29 49 210 1540 40.69 12 -62 448 40.62 10. — — — BC -160 4450 40.98 24 -160 4450 40.75 14 -290 1150 40.42 6.5 — — — I+B — 1690 41.47 73 — 1890 41.02 26 — 248 41.01 25 — — —37 NC — 567 41.92 13 — 566 41.17 2.2 — — — — 42.5 41.7 41.0(1) f IC
220 3440 43.17 230 220 3440 42.68 72 72 294 41.37 3.7 — — — BC
870 10000 42.88 120 870 10000 42.55 53 -170 824 41.79 9.8 — — —
I+B — 3790 43.35 350 — 3960 42.92 130 — 558 41.93 13 — — —38 ∗ NC — 326 41.61 140 — 325 41.05 38 38 635 41.22 56 41.1 — —(1) f IC -100 4830 42.23 580 -100 4830 41.25 60 60 286 41.51 110 — — — BC -390 12200 41.83 230 -390 12200 41.51 110 -140 1280 40.77 20 — — — I+B — 5250 42.38 800 — 6630 41.70 170 — 328 41.74 190 — — —39 NC — 212 41.00 32 — 217 40.44 6.9 — — — — 40.8 40.0 40.1(1) f IC
29 829 41.47 93 29 829 40.87 19 33 193 41.06 29 — — — BC
50 3000 41.25 57 50 3000 40.90 20 -20 427 40.73 14 — — —
I+B — 925 41.67 150 — 990 41.19 38 — 216 41.22 43 — — —40 ∗ NC — 446 41.34 8.0 — 451 40.45 0.88 -130 465 40.73 1.7 — — 41.0(3) f IC
210 1930 42.64 160 210 1930 41.84 22 130 123 40.59 1.2 — — — BC
76 6350 42.47 110 76 6350 42.21 51 -160 1590 41.27 6.0 — — —
I+B — 2180 42.86 270 — 2790 42.37 73 — 450 41.45 8.9 — — —41 NC — 400 42.10 23 — 401 41.47 3.5 — — — — 42.3 — —(2) f IC
170 2300 43.38 440 170 2300 42.85 84 -52 401 42.32 25 — — — BC
890 8300 43.09 220 890 8300 42.73 62 -260 908 42.25 22 — — —
I+B — 2510 43.56 660 — 2610 43.10 150 — 486 42.59 47 — — —42 NC — 307 40.74 14 — 306 39.90 1.6 — — — — 41.1 40.6 39.9(1) f IC
120 3760 41.89 190 120 3760 41.27 38 26 721 40.43 5.6 — — — BC -300 9930 41.66 110 -300 9930 41.44 56 -3.0 287 41.02 22 — — — I+B — 4240 42.09 300 — 4920 41.67 94 — 303 41.12 27 — — —43 NC — 414 42.10 29 — 419 41.60 6.5 — — — — 42.0 — 41.0(1) f IC
12 3550 43.23 380 12 3550 42.51 53 79 324 42.07 20 — — — BC I+B — 3920 43.41 590 — 4550 42.90 130 — 419 42.45 47 — — —44 NC — 265 41.86 26 — 262 41.34 5.1 — — — — 41.8 40.3 40.8(1) f IC
73 852 42.41 91 73 852 41.80 15 110 220 41.67 11 — — — BC
97 4190 42.32 74 97 4190 42.04 26 -98 852 41.72 13 — — —
I+B — 952 42.67 170 — 1070 42.24 40 — 260 41.99 24 — — —45 NC — 345 42.82 33 — 344 42.22 5.2 — — — — — — —(1) f IC -870 6210 43.81 330 -870 6210 43.01 32 15 267 42.79 20 — — — BC I+B — 7740 44.16 720 — 10900 43.66 140 — 346 43.17 48 — — —46 ∗ NC — 514 42.22 51 — 516 41.66 13 -11 432 42.32 63 41.4 41.9 —(3) f IC BC -1400 8410 42.93 260 -1400 8410 42.09 36 -150 1010 42.30 60 — — — I+B — 9970 43.08 370 — 9930 42.34 64 — 534 42.66 140 — — —47 NC — 391 41.58 27 — 391 40.93 5.7 — — — — 41.7 — —(1) f IC
160 3550 42.70 360 160 3550 41.94 58 50 1080 41.19 11 — — — BC
670 7400 42.02 74 670 7400 41.78 40 -24 355 41.70 34 — — —
I+B — 3730 42.78 430 — 4100 42.17 98 — 383 41.82 45 — — —48 NC — 302 41.27 51 — 306 40.67 8.7 — — — — 40.4 — 40.2(2) f IC
110 1040 41.53 92 110 1040 41.03 20 28 306 40.37 4.6 — — — BC
33 4300 41.38 66 33 4300 41.00 19 -130 719 40.30 3.9 — — —
I+B — 1160 41.77 160 — 1190 41.32 39 — 365 40.64 8.4 — — —c (cid:13) , 000–000
Spectral Study of Unobscured Type 1 AGN - I Table D1. continuedID H α H β [OIII] 5007 HeII FeVII FeX vel fwhm lum ew vel fwhm lum ew vel fwhm lum ew lum lum lum NC — 284 41.27 19 — 281 40.65 3.8 — — — — 41.5 — 40.5(1) f IC
16 1010 41.65 45 16 1010 41.05 9.4 44 268 40.98 8.1 — — — BC
160 4040 41.49 31 160 4040 41.33 18 -260 578 40.37 2.0 — — —
I+B — 1120 41.88 76 — 1340 41.51 27 — 285 41.07 10. — — —50 ∗ NC — 274 42.84 36 — 275 42.23 4.5 -73 402 41.87 2.1 42.3 42.2 41.9(1) f IC
120 1880 43.60 210 120 1880 43.06 30 130 224 42.34 6.2 — — — BC
91 8960 43.50 170 91 8960 43.12 35 -470 1770 42.42 7.4 — — —
I+B — 2090 43.86 370 — 2200 43.39 65 — 266 42.74 16 — — —51 ∗ NC — 451 41.73 65 — 445 41.20 16 11 364 42.03 110 — 41.2 —(1) f IC -1300 6490 42.44 330 -1300 6490 41.59 39 130 1150 41.57 39 — — — BC I+B — 8170 42.79 750 — 11100 42.21 170 — 450 42.21 170 — — — f : The final fitting method used for each object. see subsection 3.1 for detailed description of each fitting methods. ∗ : Three gaussian profiles are used for these objects, in this case (I+B) means the total of all three components. d : UM 269, the only object in our sample showing double-peak feature in Balmer lines. Two gaussian profiles are used for fitting thetwo peaks, thus the velocity shift of each component related to the line centre is huge.c (cid:13)000