X-Ray Properties of Narrow-Line Seyfert 1 Galaxies with Very Small Broad-Line Widths
aa r X i v : . [ a s t r o - ph . H E ] N ov To be appear in
The Astrophysical Journal
X-RAY PROPERTIES OF NARROW-LINE SEYFERT 1 GALAXIESWITH VERY SMALL BROAD-LINE WIDTHS
Y. L. Ai , , W. Yuan , , , H. Y. Zhou , T. G. Wang , S. H. Zhang [email protected], [email protected] ABSTRACT
Narrow-line Seyfert 1 galaxies (NLS1s) with very small broad-line widths (say,FWHM(H β ) . - ) represent the extreme type of Seyfert 1 galaxies that havesmall black hole masses ( M BH ) and/or high Eddington ratios ( L / L Edd ). Here we studythe X-ray properties of a homogeneously and optically selected sample of 13 suchobjects, termed as very narrow line Seyfert 1 galaxies (VNLS1s), using archival
XMM-Newton data. It is found that the Fe K α emission line is at most weak in these objects.A soft X-ray excess is ubiquitous, with the thermal temperatures falling within a strictrange of 0.1–0.2 keV. Our result highlights the puzzling independence of the thermaltemperature by extending the relations to even smaller FWHM(H β ), i.e., smaller M BH ( ∼ M ⊙ ) and/or higher L / L Edd . The excess emission can be modeled by a rangeof viable models, though the disk reflection and Comptonization models generallygive somewhat better fits over the smeared absorption and the p -free models. At theEddington ratios around unity and above, the X-ray spectral slopes in the 2–10 keVband are systematically flatter than the Risaliti et al.’s predictions of the relationshipwith L / L Edd suggested previously. Short timescale (1–2 hours) X-ray variability iscommon, which, together with the variability amplitude computed for some of theobjects, are supportive of the scenario that NLS1s are indeed AGN with relativelysmall M BH . National Astronomical Observatories/Yunnan Observatory, Chinese Academy of Sciences, Kunming, Yunnan,P.O. BOX 110, China, [email protected] Key Laboratory for the Structure and Evolution of Celestial Objects, Chinese Academy of Sciences, Kunming,China National Astronomical Observatories, Chinese Academy of Sciences, Beijing, 100012, China, [email protected]([email protected]) Key Laboratory for Research in Galaxies and Cosmology, Center for Astrophysics, University of Science andTechnology of China, Hefei, Anhui, China
Subject headings: galaxies: active — galaxies: Seyfert — X-rays: galaxies
1. Introduction
Type 1 active galactic nuclei (AGNs) are characterized by prominent broad emission lines intheir optical/UV spectra. The lower end of the line widths is mostly populated by the so-callednarrow-line Seyfert 1 galaxies (NLS1s), defined as having the broad hydrogen Balmer lines nar-rower than ∼ - in full width at half maximum (FWHM) and the relatively weak [O III ]lines (Osterbrock & Pogge 1985; Goodrich 1989). NLS1s show some extreme properties amongAGNs (see Komossa 2008, for a recent review). Previous studies have revealed a set of corre-lations among AGN optical emission line and X-ray properties—the so-called eigenvector 1 (EV1) correlations (Boroson & Green 1992), which is believed to be driven by the Eddington ratio( L / L Edd ). A narrow H β line is generally associated with strong optical Fe II and weak [O III ]emission (Goodrich 1989; Véron-Cetty et al. 2001), a steep soft X-ray spectral slope (Boller et al.1996; Wang et al. 1996), and fast and large amplitude X-ray variability (Leighly 1999; Grupe2004). However, these correlations were found based on AGNs with FWHM(H β ) mostly broaderthan ∼ - , below which only a small number of objects were known by then. Onewould expect naively, based on the EV1 correlations, that NLS1s with very small width (say,FWHM . - ) would show even extreme properties in X-ray, i.e., even steeper soft X-rayslopes and even faster and larger-amplitude variability. We refer to such AGNs as very narrow-lineSeyfert 1 galaxies (VNLS1s) hereafter in this paper.It has been recently found that the gas motion in the broad-line region (BLR) is virialized(Peterson & Wandel 2000a; Onken & Peterson 2002) and that the BLR size scales with opticalluminosity with an index of roughly 0 . β ) with black hole mass M BH and L / L Edd in a way asFWHM ∼ M BH ( L / L Edd ) - (McHardy et al. 2006). Therefore, a narrower FWHM(H β ) indicatesa larger ( L / L Edd )/ M BH ratio, provided that the inclination is not a dominating effect. This is whatNLS1s are commonly thought to be, as argued extensively in the literatures (e.g., Mineshige et al.2000; Peterson et al. 2000b; Sulentic et al. 2000), and should be even more extreme for VNLS1sas expected.As such, the X-ray properties of these extreme black hole accreting systems are of particularinterest, in light of the following considerations. Firstly, VNLS1s are well suited for investigatingthe soft X-ray excess emission commonly detected in Seyfert 1 galaxies and quasars, whose ori-gins remain controversial. One motivation is to attempt to link the soft X-ray excess with the blackbody emission from accretion disks, whose maximum temperature would be the highest amongAGNs currently known (since T max ∝ [( L / L Edd )/ M BH ] / ∝ ( FW HM ) - ), and might be detectable 3 –with the current X-ray satellites. Interestingly, this attempt was successful in at least one AGN,RX J1633+4718, that is also a VNLS1 (FWHM(H β ) ∼
900 km s - ) but radio-loud, as recently dis-covered by Yuan et al. (2010), though similar cases are extremely rare. Alternatively, the observedsoft X-ray excess can be mimicked by relativistically blurred line emission of the reflection compo-nent from a highly ionized inner disk, which may be dominant in high L / L Edd systems (Fabian et al.2002). Secondly, there were suggestions that NLS1s resemble the fastest accreting states (‘high’and ‘very high’ states) of X-ray binaries, in both the X-ray spectra (e.g., Pounds et al. 1995;Middleton & Done 2007) and X-ray quasi-periodic oscillations (QPOs) (Gierli´nski et al. 2008),and thus VNLS1s are more suitable for studying such an analogy. Thirdly, the X-ray spectral andtemporal properties of VNLS1s can be compared to AGNs with genuine small mass black holes,say, M BH < M ⊙ (e.g., Greene & Ho 2007a; Dewangan et al. 2008; Miniutti et al. 2009). Thismay provide possible insight in the black hole masses of VNLS1s and help distinguish differentmodels of NLS1s (Osterbrock & Pogge 1985; Mineshige et al. 2000; Sulentic et al. 2000).Although some VNLS1s have been studied in X-rays individually in the literature, systematicstudies of their ensemble X-ray properties are rare, however, given the lack of homogeneouslyselected samples in the past. Recently, the X-ray properties of small samples of AGNs with M BH . M ⊙ with XMM-Newton observations have been presented (e.g., Dewangan et al. 2008;Miniutti et al. 2009), among which several objects are in fact VNLS1 considering their opticalspectral properties and the high L / L Edd . It was found that these VNLS1s are characterized bystrong and rapid X-ray variability and soft X-ray excess emission. However, more observations fora larger, homogeneously selected sample are needed to confirm these results.Using a large NLS1 sample selected from the SDSS, Zhou et al. (2006) found that, to one’ssurprise, the previously known FWHM(H β )– Γ s (soft X-ray photon index) anti-correlation becomesflattened at FWHM ∼ - . Though in the small FWHM regime AGNs having flat Γ s have previously been noted to exist , the lack of expected steep soft X-ray slopes is intriguing.However, enhanced X-ray absorption in VNLS1s may explain such a trend, which, though seemsto be unlikely, cannot be ruled, since in Zhou et al. (2006) the Γ s (estimated from the ROSAT hardness ratios) are subject to large uncertainties. Detailed spectral modeling of the X-ray spectrawith higher spectral resolution and S/N for these objects is needed to confirm this interesting trend.Motivated by the above considerations, here we present a study on the X-ray properties of asample of NLS1 with extreme linewidth, FWHM ≤ - , using data from archival XMM-Newton observations. The sample and data reduction are described in Section 2. The modeling ofthe
XMM-Newton spectra are presented in Section 3, with a focus on the soft X-ray excess. The For such AGNs, their deviation from the above relation is explained as due to their low luminosity (low L / L Edd );see Laor (2000). H = 70 km s - Mpc - , Ω Λ = 0 .
73, and Ω M = 0 .
27. Allquoted errors correspond to the 90% confidence level unless specified otherwise.
2. Sample and X-ray data2.1. X-ray VNLS1 sample
We select VNLS1s from a large, homogeneous sample of ∼ β ) . - forVNLS1s, in consideration of the fact that below roughly this value the Γ s –FWHM anti-correlationbecomes flattened (see figure 17 in Zhou et al. 2006). There are 384 NLS1s meeting this criterion.We match these VNLS1s with the 2XMM source catalogue (Watson et al. 2007) using a matchingradius of 5 ′′ and then select those detected in X-rays with at least 200 net source counts. Weconsider radio-quiet objects only, since X-rays from radio-loud NLS1s may be contaminated byemission from relativistic jets (Zhou et al. 2007; Yuan et al. 2008; Abdo et al. 2009). The aboveselection results in 13 objects with reasonable signal-to-noise (S/N) ratios, which form our workingsample of VNLS1s in this study. The sample objects are listed in Table 1, and the logs of the XMM-Newton observations are summarized in Table 2. Among the sample, the
XMM-Newton data of seven objects are presented here for the first time; while the
XMM-Newton spectra of sixobjects have been presented previously in various details in the literatures for different aims. Forthe purpose of sample study using homogeneously derived results, we also re-analysis the XMM-Newton spectra of these objects, in the same way as for the other objects whose
XMM-Newton dataare presented for the first time here.The optical spectral and continuum parameters of the sample objects are taken from Zhou et al.(2006) and given in Table 1 . The black hole masses are estimated from the broad componentof the H α line using the M BH –linewidth–luminosity relation in Greene & Ho (2007b). We alsoestimate the Eddington ratio L / L Edd assuming the bolometric luminosity as 9 λ L (Elvis et al.1994), where λ L is the monochromatic luminosity at 5100 Å. Figure 1 shows the distributions Having radio-loudness less than 10, defined as the rest frame flux ratio between the radio 1.4 GHz and the optical g -band (see Zhou et al. 2006) They are J0107+1408, J1140+0307, J1357+6525 (Dewangan et al. 2008; Miniutti et al. 2009), J1246+022(Porquet et al. 2004), J2219+1207 (Gallo et al. 2006b) , and J1415-0030 (Foschini et al. 2004). M BH , L / L Edd and the optical Fe II emissionmultiplets strength R , in comparison with those of the NLS1 sample of Zhou’06 as well asthe overall FWHM < - (VNLS1) sub-sample. A few remarks can be made concerningthe bulk properties of the sample. Firstly, both of these two VNLS1 sub-samples have similar R distributions to that of the parent sample, confirming their typical NLS1 nature. The generalstrong Fe II emission is also demonstrated in the composite SDSS spectrum of our XMM-Newton
VNLS1 sample (Figure 2). Secondly, our VNLS1s have lower M BH and higher L / L Edd distributionsin general than the overall NLS1 sample, as expected from their narrower linewidths. Thirdly, our
XMM-Newton sample is roughly consistent with the overall VNLS1s in these distributions. Thusour X-ray VNLS1 sample is not biased from, but rather representative of, optically selected NLS1swith the smallest linewidth. This should be kept in mind when comparing our results with thoseobtained in previous studies, especially for X-ray selected NLS1 samples.
The observational data with
XMM-Newton were retrieved from the
XMM-Newton sciencearchival center. For all but one object the observations were operated in the full window mode. Forthe only exception, J2219+1207, the MOS cameras were operated in the small window mode, inwhich a considerable fraction of the source counts in the wing of the point spread function (PSF)were lost, and thus only the PN data are used. The PN observation of J1246+0222 experienced apile-up, which is corrected by excising the core of the PSF with a radius of 10 ′′ . Some of the dataof the individual cameras as listed in Table 2 cannot be utilized due to the sources being either atthe edges or in the gaps of the CCDs, out of the field of view, or on a bad CCD column.For XMM-Newton data reduction we use the standard Science Analysis System (SAS, v8.1.0.).The Observation Data Files (ODF) are processed to create calibrated events files with ‘bad’ (e.g.,‘hot’, ‘dead’, ‘flickering’) pixels removed. The time intervals of high flaring backgrounds con-tamination are identified and subsequently removed following the standard SAS procedures andthresholds. Source counts are extracted from a circle with a radius ranging from, depending on thesource position on the detector, 30 ′′ to 65 ′′ at the source position, and the background counts froma source-free region with a usually larger radius. To extract X-ray spectra only X-ray events withthe pattern ≤ ≤
12 for MOS are used. Background subtracted light curves are alsoextracted from cleaned events files and are subsequently corrected for instrumental effects (such Defined as the Fe II ( λλ - β flux ratio, where Fe II ( λλ - II multiplets integrated over the wavelength range of 4434–4684 Å, and H β the flux of the broad component of H β ; seeZhou et al. (2006).
3. X-ray spectra analysis3.1. Continuum shape and Fe K α emission line In order to compare the X-ray continuum slopes with results from other previous AGN studies,we first characterize the X-ray spectral shape with an absorbed power-law model in both the soft(0.2–2.4 keV) and "hard" (2–10 keV) bands , respectively. The results are listed in Table 3. In thesoft X-ray band, the model with a neutral absorption column density ( N H ) fixed at the Galacticvalue ( N GalH ) yields acceptable fit for about half of the sample objects. In the remaining objectsthe fit can be improved by adding an extra neutral absorber in the objects’ rest frame. The fittedexcess absorption N H are small, however, comparable to N GalH . In only one object, J1415-0030,ionized absorption is required to yield acceptable fit, with an edge-like feature around 0.6 keV (seeSection 3.2). We conclude that intrinsic absorption is not significant in these objects. We thussuggest that the observed flattening of the FWHM(H β )– Γ s relation below FWHM ∼ - as found in Zhou et al. (2006), is not caused by X-ray absorption, but most likely an intrinsicproperty. For those having more than one measurement, there seems to be little or no changes in thesoft X-ray spectral shape, and the mean Γ s are calculated. The fitted Γ s values range from 2.03 to3.72. We quantify the intrinsic distribution of Γ s (assumed to be Gaussian) that is disentangled frommeasurement errors using the maximum-likelihood method as first applied by Maccacaro et al. For three objects only the 2–7 keV band is used since the spectra above 7 keV are dominated by backgrounds; seeTable 2. h Γ s i = 2 . + . - . , and a standard deviation σ = 0 . + . - . (90% confidence),whose confidence contours are shown in Figure 3.In the hard X-ray band there are nine objects having high enough spectral S/N ratios formeasuring photon indices Γ h . The absorption N H is fixed at the Galactic value. A power-lawcan well reproduce the observed hard X-ray spectrum in five objects, whereas the remaining fourobjects (J0922+5120, J1140+0307, J1246+0222, and J2219+1207) show a possible broad excessemission feature in the residuals around 5 keV. The fitted Γ h values are in the range of 1.95–2.39.The maximum-likelihood intrinsic distribution of Γ h has a mean of h Γ h i = 2 . + . - . , which is flat fortypical NLS1s, and a small intrinsic scatter, σ = 0 . + . - . (see Figure 3 for their confidence contours).No Fe K α emission line feature appears to be present in all the objects except J1357+6525,which shows a marginal feature of a narrow-line at around 6.4 keV. For this object adding a Gaus-sian (to an absorbed power-law model in the 2–10 keV band) improved slightly the fit, though withonly ∆ χ =5 for 3 degrees of freedom (dof), i.e. the addition of an extra Gaussian component isnot statistically significant. Thus only upper limits can be derived on the equivalent width (EW)of any potentially narrow Fe K α line at 6.4 keV (assuming σ = 10 eV). The derived upper limits atthe 90% significance are given in Table 3. We compared our objects with the Fe K α line EW andX-ray luminosity relation for AGNs as given in Page et al. (2004), and found that the derived lineEW limits are well consistent with the relation. A comparison of the soft and hard X-ray spectral indices obtained above indicates an overallspectral steepening toward low energies in all of the objects, suggesting the presence of the softX-ray excess. As demonstration, we show in Figure 4 the X-ray spectra of the 4 VNLS1s which arepresented for the first time, and the extrapolation down to 0.2 keV of the power-law model fittedin the hard X-ray band. Significant excess emission in the soft X-ray band is prominent, similar tothat reported in the other objects of the sample (e.g., Miniutti et al. 2009), which is also confirmedhere. We thus conclude that the apparent soft X-ray excess emission is ubiquitous in our VNLS1sample.Given the small M BH and high accretion rate ( L / L Edd ) in VNLS1s, the expected blackbodyemission from accretion disks is shifted toward higher energies compared to classical AGNs withmore massive black holes, and the high energy turnover may start to emerge in the soft X-rays(e.g., kT max ∼
72 eV for J0940+0324 assuming a Schwarzschild black hole, e.g. Peterson 1997).Thus the blackbody emission directly from the disks might be detected. We first model the softX-ray excess with a blackbody. The model yields acceptable or marginally acceptable fits for all 8 –except for two objects. For J0107+1408 an additional neutral absorber is required to improve thefit. For J1415-0030 a moderately ionized absorber is needed; adding an absorption edge improvesthe fit significantly with ∆ χ = 12 for 3 additional free parameters. The fitted edge energy is0.67 ± ± - M ⊙ (Gierli´nski & Done 2004; Porquet et al.2004; Crummy et al. 2006; Bianchi et al. 2009). Hence our results confirm the extension of thecanonical 100–200 eV temperatures down to AGNs with M BH as low as around ∼ M ⊙ (Miniutti et al.2009) by adding more objects in this M BH range. The result is clearly demonstrated in Figure 5, inwhich our results are compared with those of AGNs and quasars with more massive black holes(Piconcelli et al. 2005; Crummy et al. 2006). These values are still systematically higher than themaximum temperatures predicted for standard accretion disks. The independence of the thermaltemperature on M BH over such a wide M BH range argues against the direct blackbody emissionfrom accretion disks as the origin of the observed soft X-ray excess for the vast majority of AGNs,except for RX J1633+4718 (Yuan et al. 2010).Recent studies suggest that, similar to the blackbody temperature, the relative strength of thesoft X-ray excess also falls within a relatively small range for PG quasars (Piconcelli et al. 2005)and Seyfert 1 AGNs (Middleton & Done 2007) and AGNs with small masses (Miniutti et al. 2009).Here we examine this quantity for the VNLS1s of our sample. We estimate the relative strength asthe luminosity ratio of the excess component, modeled as a blackbody, to the total luminosity in the0.5–2 keV band. The values are in the range of 8%–38% with a mean of 21% (Table 4). Apparently,when combined with previous results, as shown in Figure 6, there seems no strong dependence onH β linewidth over a large range, FWHM(H β ) = 600–10,000 km s - (a Spearman correlation testprobability of 0.79). The mean relative strength of our VNLS1s is somewhat smaller than that(31%) of the PG quasars in (Piconcelli et al. 2005) derived from their fitting results, though furtherconfirmation is needed given the relatively small size of our sample.There are currently several viable models to account for the soft X-ray excess. Photon trap-ping in high accreting system where advection is important (Abramowicz et al. 1988), or Comp-tonization of ultraviolet photon from the accretion disc by electrons as hotter skin above the disc(Czerny & Elvis 1987; Wandel & Petrosian 1988; Shimura & Takahara 1993; Czerny et al. 2003),can explain the required higher temperature. On the other hand, absorption or emission linesarising from atomic processes, when blurred due to relativistic motion, can mimic the soft X-rayexcess. For example, strong relativistically-blurred emission/absorption lines between ∼ N H fixed at the Galactic value is always included. For objects whosedata have been analyzed previously, we compare our results with previous results individually inAppendix A. We use the Comptonization model ( comptt in XSPEC, Titarchuk 1994) and fix the input seedphoton energy at the innermost temperature of the standard accretion disk based on the estimationof the black hole masses and accretion rates ( L / L Edd ). In this case the emergent spectral shapedepends on only two parameters, the temperature and the optical depth of scattering electrons.This model gives significantly improved fits over, or at least as good as, the above blackbody fitsfor all of the objects (see Table 4, Figure 7). The inferred electron temperatures are found in arelatively small range, kT plasma ∼ τ ∼ We use the latest ionized disk reflection model from Ross & Fabian (2005) ( reflion in XSPEC)and in the fits the photon indices of the ionizing continuum and the observed continuum are tied to-gether, and the solar abundance is assumed. For relativistic blurring the laor kernel model ( kdblur in XSPEC, Laor 1991) is used with an outer radius fixed at 400 r g , an emissivity index of the diskfixed to the standard value of 3, and the inner disk radius allowed to vary. Although this modelcan reproduce acceptable fitting results for most of the objects, in three objects the residuals inthe soft X-rays indicate possible contribution from another component. Following Miniutti et al.(2009), we then include an additional blackbody component in the fits to account for possible con-tribution from the accretion disk. This improves the fits for all of the three, namely J0107+1408( ∆ χ / do f = 14/3), J0922+5120 ( ∆ χ / do f = 204/3), and J2219+1207 ( ∆ χ / do f = 24/2), withthe addition of the blackbody component is statistically significant (a probability level < ± ± + . - . keV,respectively, broadly consistent with the maximum temperatures at near the inner disk predictedfrom the estimated black hole masses and accretion rates.The disk reflection model, either with or without additional blackbody emission, providesacceptable fits for all and the best fits for some of the objects (see Figure 7 and Table 4). The best-fit disk inner radius is less than 6r g , suggesting a highly spinning Kerr black hole for most of our 10 –objects. The inferred disk inclination varies from 0 ◦ to 50 ◦ . The ionization parameters are log ξ ∼ Finally we fit the spectra with the relativistically smeared absorption model ( swind1 in XSPEC).This model provides acceptable fits for nearly half of the objects but not for the remaining objects(Table 4, Figure 7). The inferred column densities are in the range of N H ∼ × cm - , theionization parameters log ξ ∼ In this work we also try the p-free disk model ( diskpbb in XSPEC) to account for the softexcess. For some of the objects, this model gives as good fits as the Comptonization model, withthe inferred temperatures at the inner disk radius of 0.15–0.35 keV, and the index of the temperatureprofile p = 0.32–0.64. However, the innermost disk radii derived from the fitted normalization aresignificantly less than the estimated gravitational radii, even after the correction for the spectralhardening factor ( ∼ . p -free model to be physically unrealistic and hence no fittedparameters are listed here. In general, the soft X-ray excess of these VNLS1s can be reproduced by more than one model,which often cannot be distinguished based on fitting statistics. However, disk reflection and/orComptonization are much more preferred than the other two ones, i.e., smeared absorption modelwith marginally improved fitting and p-free model which is physically unrealistic. In the mod-eling the X-ray spectra, no additional intrinsic absorption are required for all the objects exceptJ0107+1408 and J1415-0030, of which additional neural and warm absorption needed, respec-tively. 11 –For SDSS J0107+1408, neutral absorption with N H ∼ × cm - is required. ForJ1415-0030, in the above spectral fittings for this object with various soft X-ray excess models, weadd ionized absorption ( zxipcf in XSPEC) applicable to all the emission components. The overallspectrum and the edge feature can be well reproduced and all the fits are improved significantly,with a decrease of ∆ χ = 10 in general. The covering fraction of the absorber is close to unity, theabsorbing column density is in the range of 9.3–15.8 × cm - and the ionization parameter of10 . - . , depending on the exact model for the soft X-ray excess. Γ s –FWHM relation We have showed above that there is little or no significant intrinsic absorption in the X-rayspectra of most of these VNLS1s. Therefore the photon indices derived in Zhou et al. (2006)from the
ROSAT hardness ratios are mostly reliable, and hence the spectral flattening towards thelower FWHM end (Zhou et al. 2006) is likely real, rather than being caused by X-ray absorption.Recently, Grupe et al. (2010) studied the spectral indices of a sample of soft X-ray selected AGNsmeasured with the
Swift
XRT, some of which also have linewidths similar to ours. We comparethe soft X-ray photon indices of our VNLS1s with those of the Grupe et al. (2010) sample, asshown in Figure 8. It can be seen that in the lowest linewidth regime our Γ s values are statisticallycompatible with the result of Grupe et al. (2010) . It also appears that there is a lack of AGNshaving both narrow FWHM ( . - ) and very steep soft X-ray spectra ( Γ s & . Γ s –FWHM relationat FWHM . - using various methods; the result is inconclusive in the statistical sense,however. This might be partly due to the relative small sample size and/or the heterogeneity ofthe combined samples. A larger and homogeneously selected sample is needed to test the possibledeviation of this well known Γ s –FWHM relation at the low-FWHM end in the future. Note that the Γ s values in Grupe et al. (2010) are measured in the 0.2-2 keV band, slightly different from 0.2-2.4 keV adopted here; however, we find from spectral fits that the differences in Γ s thus caused are negligible for ourobjects.
12 –
4. X-ray variability
Figure 9 shows, as examples, the 0.2-10 keV lightcurves for the five objects in our sample,which are presented for the first time . In fact, we find that, for most of the sample objects, theX-ray flux varied by more than a factor of 2 on timescales of 1–2 hours within the observationalintervals. We conclude that short timescale variability is common to VNLS1s. Remarkable fluxvariations in short-timescales are evident. We also investigate possible spectral variability withinthe observational intervals, which are divided into time bins, using the hardness ratios; however, noconclusive remarks can be made mainly due to the insufficient S/N of the data for such a purpose.The X-ray variability amplitude can be characterized by the excess variance , which hasbeen found to be strongly correlated with the black hole masses (Lu & Yu 2001; Papadakis 2004;O’Neill et al. 2005). Using the XMM-Newton data, some are included in the current sample,Miniutti et al. (2009) extended this relation to AGNs with black hole masses < 10 M ⊙ (three in-cluded in our sample) and demonstrated that the relationship is relatively tight. This result indicatesthat black hole mass is a primary parameter that drives the relative X-ray variability in AGNs. In thesame way as in Miniutti et al. (2009), we calculate the excess variance for the four objects not pre-sented previously, that makes use of lightcurve segments of equal duration (20 ks) and equal timebin size (500 s) in the 2–10 keV band. However, for only one object, J0922+5120, the lightcurvehas S/N high enough to yield reliably determined excess variance, log σ = - . ± .
68. Welocate this object ( M BH = 10 . M ⊙ ) on the excess variance vs. black hole mass relation presentedin Figure 8 of Miniutti et al. (2009), and find that it does follow closely the relation. It should benoted that, J1140+0307, one of the three objects in Miniutti et al. (2009) that are in our sampleis also typical NLS1. As is generally believed that M BH is the underlying physical parameter thatdrives the dependence of the X-ray variability, the fact that these NLS1s follow the same excessvariance– M BH relation as for normal broad-line Seyfert 1 galaxies (BLS1s) and quasars tends tovalidate their M BH estimation. That is to say, the black hole masses of the VNLS1s in our sampleare indeed relatively small, i.e. their narrow Balmer linewidth is caused primarily by relativelysmall M BH , rather than by a face-on flattened BLR as claimed in some papers in the literatures. For J0107+1408, J1140-0307, and J1357+6525 the X-ray lightcurves have been presented in Miniutti et al. (2009)and Dewangan et al. (2008); for J1415-0030 it was in Foschini et al. (2004). Defined as variance of the lightcurve in a specific time series after subtracting the contribution expected frommeasurement errors (e.g., Vaughan et al. 1999).
13 –
5. Optical/UV to X-ray spectral index
The broad-band spectral energy distribution of AGN is commonly parameterized by the α ox parameter, α ox = - . ν (2500 Å) / L ν (2keV)] (Tananbaum et al. 1979), which is claimedto be correlated with the optical-UV luminosity for Seyferts and quasars (Vignali et al. 2003;Yuan et al. 1998; Strateva et al. 2005). Gallo (2006a) reported that the objects in their NLS1 sam-ple, which have systematically broader linewidths than ours, also follow the same relation forBLS1s. Here we investigate this relation for our VNLS1s. The flux density at 2500 Å is calculatedfrom the SDSS u band (effective observed-frame wavelength of 3543 Å) PSF-magnitude adopt-ing the spectral slope of the composite SDSS quasar spectrum ( α ν = - .
44; Vanden Berk et al.2001). The α ox values of our objects are found to lie in the range from -1.41 to -1.18. We showin Figure 10 the α ox vs. 2500 Å monochromatic luminosity relation for our objects, along withthe normal NLS1s from the Gallo (2006a) sample, as well as the regression relation for normalSeyferts and quasars (Strateva et al. 2005). It shows that the VNLS1s do follow closely the α ox – L uv relation defined by BLS1s, and consistent with the result for NLS1s with broader linewidth(FWHM & - ).
6. Discussion6.1. Soft X-ray excess in VNLS1s
The ubiquity of the soft X-ray excess in our VNLS1 sample is interesting, which may havesomething to do with the generally high L / L Edd in our sample. The reflection model has beenfound to be a good description of the complexity of the X-ray spectra and spectral variabilityfor several NLS1s (Miniutti & Fabian 2004; Crummy et al. 2006; Ballo et al. 2008; Zoghbi et al.2008). In some cases the reflection component is found to dominate the observed X-ray band,which can be explained either in terms of a corrugated disk or strong gravitational light bendingeffects (Fabian et al. 2002). Miniutti et al. (2009) found that this model reproduce well the
XMM-Newton spectra of several AGNs with M BH ≤ M ⊙ , three of which are included in this paper. Thespectral fitting for the other objects in our sample also supports this result. The small inner radiiof the accretion disk derived argue for fast rotating black holes in most of these VNLS1s, whichmay be a consequence of their fast accretion process. In a few cases the disk thermal emission isrequired, which is not unexpected given the relatively high disk temperatures for small M BH ; thisindicates that the disk reflection model is self-consistent.The Comptonization model successful explains the soft excess in normal Seyferts (Gierli´nski & Done2004) and the spectral variability in RE J1034+396 (Middleton et al. 2009). However, as pointedout by Gierli´nski & Done (2004), the derived electron temperatures and the optical depth are both 14 –found in a small range ( kT e ∼ . - . τ ∼ - Our result is supportive of the lack of objects with very steep soft X-ray slopes and FWHM(H β ) . - in the Γ s –FWHM diagram; those objects may be expected simply based on this well-known EV1 correlation. Here we try to link this relationship with L / L Edd which is believed tobe the underlying driver of the EV1 correlations. The soft X-ray spectral slope is primarily deter-mined by two factors, the slope of the underlying (hard X-ray) continuum and the effect of the softX-ray excess. We examine the effect of the former by comparing the soft and hard band spectralindices of our sample objects, and find that they are strongly correlated with each other (a Spear-man correlation test probability P < - ). Thus we conclude that soft X-ray slope Γ s is largelydetermined by, or tracing the underlying continuum slope Γ h .Compared to the Γ –FWHM relation, a more significant, and perhaps fundamental, correlationis the Γ h – L / L Edd correlation that has been detected in the range of L / L Edd < Γ h – L / L Edd relation for our VNLS1s whose Γ h are available. Also plotted is, for a comparison, the regression of Risaliti et al. (2009) for thestrongest Γ h – L / L Edd relation ( M BH estimated from H β , the same as in our paper) and its extrapola-tion to the high L / L Edd regime where our sample objects are located. Interestingly, the Γ h values ofall our VNLS1s fall systematically below the extrapolation of the Risaliti et al.’s relation to high- L / L Edd values. To check whether this flattening might be caused by the enhanced Compton reflec-tion hump—known to exist in Seyfert galaxies and to make the hard X-ray spectrum flatten—inhigh- L / L Edd objects, we over-plot the ‘underlying’ X-ray continuum slope inferred from the abovedisk reflection model fitting. As can be seen, the ‘underlying’ hard X-ray continua (open circles)are indeed slightly steeper than the ‘observed’ one, but are still systematically flatter than the pre-diction of the Risaliti et al.’s relation. We thus suggest that, at L / L Edd ∼ Γ h – L / L Edd relation may naturally explain the above observed lack of objects with very steepsoft X-ray spectra (i.e. Γ s & .
5) at the lowest FWHM end.
Three objects in our sample were studied by Miniutti et al. (2009) as AGNs with intermedi-ate mass black holes (IMBHs), a term sometimes used to refer to black holes with M BH < M ⊙ in the literature (e.g. Greene & Ho 2004). Since M BH ∝ FWHM , most IMBH AGNs must havebroad line widths falling within the conventional criterion of NLS1s ( . - ), but may notnecessarily possess the characteristics of typical NLS1s, i.e., strong FeII emission, high Edding-ton ratios, and significant soft X-ray excess (e.g. Greene & Ho 2004). The X-ray properties ofIMBH AGN samples have been studied by several authors (Greene & Ho 2007a; Dewangan et al.2008; Desroches et al. 2009; Miniutti et al. 2009), and a large diversity has been found. For in-stance, the soft X-ray (0.5–2 keV) photon indices are found to fall into a large range ( Γ s = 1–2.7,Desroches et al. 2009). The flat X-ray spectral slopes, as well as some other properties, are verysimilar to those of typical Seyfert galaxies with M BH = 10 - M ⊙ . Some, especially those withlow L / L Edd , do not show soft X-ray excess (Iwasawa et al. 2000; Dewangan et al. 2008), as inNGC 4395, the prototype of this kind.We suggest that IMBH AGNs, albeit their small linewidths as for NLS1s, have diverse ob-served properties, depending on the Eddington ratio. Those accreting at high L / L Edd values areprobably more NLS1-like, e.g. the presence of a significant soft X-ray excess, strong FeII, steep Γ s , such as the three IMBHs in Miniutti et al. (2009) and also included in our sample as VNLS1.On the other hand, there exists a population accreting at low L / L Edd , which exhibit properties re-sembling closely those of classical Seyfert galaxies with more massive M BH , in both optical andX-ray (week FeII, relatively flat Γ s , non-ubiquity of the soft X-ray excess). The observed spectralproperties of Seyfert galaxies depend much more strongly on mass accretion rate than on blackhole mass. In this regards, the conventional definition of NLS1s may have to be revised. On theother hand, NLS1s and IMBH AGNs show similar timing property. This is not surprising given thepostulation that the X-ray variability of AGNs is believed to be largely determined by black holemass, rather than accretion rate.
7. Summary
NLS1s with very small broad-line widths represent the extreme of Seyfert 1 AGNs, whichhave the largest L / L Edd / M BH ratio among all AGNs known so far. Here we investigated the X-ray 16 –properties of a homogeneously selected sample of NLS1s with FWHM(H β ) . - , usingthe archival XMM-Newton data. We note that our sample is not complete in the sense that onlythose observed with
XMM-Newton with good spectral S/N ratios are included, which might bebiased toward relatively bright objects in X-rays. This should be kept in mind when comparing ourresults with the others.No significant Fe K α emission line is detected, which should be at most weak in such objects.It is found that the soft X-ray excess is ubiquitous in the objects which have the 0.2–10 keV spectraavailable. The temperatures of this component, when fitted with a blackbody (or disk blackbody)model, all fall within 0.1–0.2 keV, significantly higher than the prediction of the standard diskmodel. Our result highlights the puzzling independence of the thermal temperature on M BH byextending it to NLS1s with narrower FWHM(H β ), i.e., smaller M BH and/or higher L / L Edd . Thefailure to ascribe the soft X-ray excess to the Wien tail of the disk blackbody emission in theseVNLS1s (with similar M BH and L / L Edd values to RX J1633+4718, though) highlights the questionas to why RX J1633+4718 is so unique (see Yuan et al. 2010, for a brief discussion). A range of vi-able models, including Comptonization, disk reflection, smeared absorption, and the p -free modelwere used to fit the soft X-ray excess. In general, the disk reflection and Comptonization modelstend to give the best fits. The relative strength of the soft X-ray excess appears to be independentof FWHM(H β ) over a large range, indicating that the excess component is not particularly strongin these VNLS1s compared to PG quasars with much broader linewidth.The soft X-ray spectra in 0.2–2.4 keV have a mean photon index of Γ s =2 . + . - . with a largeintrinsic scatter ( σ = 0 . + . - . ), while the 2–10 keV spectra have a mean Γ h of 2 . + . - . with asmall intrinsic scatter consistent with zero. Thus VNLS1s have the spectral slopes in both bands nosteeper than "normal" NLS1s with broader linewidth. There is little or no intrinsic X-ray absorptionin most of these VNLS1s, indicating that the flattening of the Γ s –FWHM anti-correlation belowFWHM ∼ - , as suggested in Zhou et al. (2006, using a much larger sample but with Γ s estimated from hardness ratios), is not caused by absorption but most likely intrinsic. Althoughthis trend is not statistically significant when combining our current sample with the Swift sampleof Grupe et al. (2010), both with Γ s derived from spectral fitting, there appears a lack of AGNs withboth very narrow FWHM(H β ) ( . - ) and very steep soft X-ray spectra (i.e., Γ s & . Γ h of our objects fall systematicallybelow the extrapolation of the suggested Γ h – L / L Edd correlation (Risaliti et al. 2009) to high L / L Edd values. We argue that these two trends, if confirmed, might in fact be driven by the same underlyingphysical process. Similar to other "normal" NLS1s, these VNLS1s also follow the same α ox – L opt relation as for normal Seyferts and quasars.All of the sample objects show rapid variability in X-rays, with two-fold timescales of 1–2hours. The short variability timescales and the conformance with the variance excess–black hole 17 –mass relation for normal Seyferts tends to suggest that the black hole masses in these VNLS1s arelikely truly small, as commonly thought, and the present M BH estimators based on the linewidth–luminosity scaling relation is applicable to NLS1s.We thank the referees for helpful comments and suggestions, which helped to improve thepaper significantly. Y. Ai is grateful to Stefanie Komossa for her tutoring the X-ray data analysisand discussion on this work, and her kind hosptality during the visit at MPE. Y. Ai is grateful tothe support by the MPG–CAS Joint Doctoral Promotion Programme and the hospitality of MPE,based on which part of the research was carried out. This work is supported by NSFC grants10533050, 11033007, the National Basic Research Program of (973 Program) 2009CB824800,2007CB815405. This research is based on observations obtained with XMM-Newton
REFERENCES
Abramowicz, M. A., Czerny, B., Lasota, J. P., & Szuszkiewicz, E. 1988, ApJ, 332, 646Abdo, A. A. et al. 2009, ApJ, 699, 976Arnaud, K.A., 1996, in ASP Conf. Ser. vol. 101, Astronomical Data Analysis Software and Sys-tems V, eds. Jacoby G. & Barnes J., 17Ballo, L., Giustini, M., Schartel, N., Cappi, M., Jiménez-Bailón, E., Piconcelli, E., Santos-Lleó,M., & Vignali, C. 2008, A&A, 483, 137Bentz, M. C., Peterson, B. M.. Pogge, R. W., Vestergaard, M., & Onken, C. A. 2006, ApJ, 644,133Bianchi, S., Guainazzi, M., Matt, G., Fonseca Bonilla, N., & Ponti, G. 2009, A&A, 495, 421Boller, T., Brandt, W. N., & Fink, H. 1996, A&A, 305, 53Boroson, T. A., & Green, R. F. 1992, ApJS, 80, 109Crummy, J., Fabian, A.C., Gallo, L., & Ross, R. R. 2006, MNRAS, 365, 1067 18 –Czerny, B., Elvis, M. 1987, ApJ, 321, 305Czerny B., Nikołajuk, M., Ró˙za´nska, Agata, Dumont, A.-M., Loska, Z., & ˙Zycki, P. T. 2003, A&A,412, 317Desroches, L.-B., Greene, J. E., & Ho, L. C. 2009, ApJ, 698, 1515Dewangan, G. C., Mathur, S., Griffiths, R. E., & Rao, A. R. 2008, ApJ, 689, 762Elvis, M. et al. 1994, ApJS, 95, 1Fabian, A. C., Ballantyne, D. R., Merloni, A., Vaughan, S., Iwasawa, K., & Boller, Th. 2002,MNRAS, 331, 35Foschini L. et al. 2004, A&A, 428, 51Gallo, L. C., 2006a, MNRAS, 368, 479Gallo, L. C., Lehmann, I., Pietsch, W., Boller, Th., Brinkmann, W., Friedrich, P., & Grupe, D.2006b, MNRAS, 365, 688Gierli´nski, M., & Done, C. 2004, MNRAS, 349, 7Gierli´nski, M., Middleton, M., Ward, M., & Done, C. 2008, Natur. 445, 369Goodrich R. W., 1989, ApJ, 342, 224Greene, J. E., & Ho, L. C. 2004, ApJ, 610, 722Greene, J. E., & Ho, L. C. 2007a, ApJ, 656, 84Greene, J. E., & Ho, L. C. 2007b, ApJ, 670, 92Grupe, D. 2004, AJ, 127, 1799Grupe, D., Komossa, S., Leighly, K. M., Page, K. L. 2010, ApJS, 187, 64GGuainazzi, M. 2009, XMM-EPIC status of calibration and data analysis, XMM-SOC-CAL-TN-0018Iwasawa, K., Fabian, A. C., Almaini, O., Lira, P., Lawrence, A., Hayashida, K., & Inoue, H. 2000,MNRAS, 318, 879Kaspi, Shai, Maoz, D., Netzer, H., Peterson, B. M., Vestergaard, M., & Jannuzi, B. T. 2005, ApJ,629, 61 19 –Kawaguchi, T. 2003, ApJ, 593, 69Komossa, S. 2008, RevMexAALaor, A. 1991, ApJ, 376, 90Laor, A. 2000, NewAR, 44, 503Leighly, K. M. 1999, ApJS, 125, 297Lu, Y.-J., & Yu, Q.-J., 2001, ApJ, 561, 660Maccacaro, T., Gioia, I. M., Wolter, A., Zamorani, G., & Stocke, J. T., 1988, ApJ, 326, 680McHardy, I. M., Koerding, E., Knigge, C., Uttley, P., Fender, R. P. 2006, Natur, 444, 730Middleton, M., Done, C., & Gierli´nski, M., 2007, MNRAS, 381, 1426Middleton, M., Done, C., Ward, M., Gierli´nski, M., & Schurch, N. 2009, MNRAS, 394, 250Mineshige, S., Kawaguchi, T., Takeuchi, M., & Hayashida, K., 2000, PASJ, 52, 499Miniutti G., & Fabian A. C. 2004, MNRAS, 349, 1435Miniutti, G., Ponti, G., Greene, J. E., Ho, L. C., Fabian, A. C., & Iwasawa, K., 2009, MNRAS,394, 443O’Neill, P. M., Nandra, K., Papadakis, I, E., Turner, T. J. 2005, MNRAS, 358, 1405Onken, C. A., & Peterson, B. M. 2002, ApJ, 572, 746Osterbrock, D. E., & Pogge, R. W. 1985, ApJ, 297, 166Page, K. L., O’Brien, P. T., Reeves, J. N., & Turner, M. J. L. 2004, MNRAS, 347, 316Papadakis, I. E. 2004, MNRAS, 348, 207Peterson, B. M., 1997, An Introduction to Active Galactic Nuclei, Cambridge Univ. Press, Cam-bridgePeterson, B. M., & Wandel, A. 2000a, ApJ, 540, 13Peterson, B. M., et al. 2000b, ApJ, 542, 161Piconcelli, E., Jimenez-Bailón, E., Guainazzi, M., Schartel, N., Rodríguez-Pascual, P. M. &Santos-Lleó, M., 2005, A&A, 432, 15 20 –Porquet, D., Reeves, J. N., O’Brien, P., & Brinkmann, W., 2004, A&A, 422, 85Pounds, K. A., Done, C., & Osborne, J. P. 1995, MNRAS, 277, 5Risaliti, G., Young, M., & Elvis, M. 2009, ApJ, 700, 6Ross, R. R., & Fabian, A. C., 1993, MNRAS, 261, 74Ross, R. R., & Fabian, A. C., 2005, MNRAS, 358, 211Shemmer, O., Brandt, W. N., Netzer, H., Maiolino, R., & Kaspi, S. 2006, ApJ, 646, 29Shimura, T., Takahara, F. 1993, ApJ, 419, 78Strateva, I. V., Brandt, W. N., Schneider, D. P., Vanden Berk, D. G., & Vignali, C. 2005, AJ, 130,387Sulentic, J. W., Zwitter, T., Marziani, P., & Dultzin-Hacyan, D. 2000, ApJ, 536, L5Tananbaum, H., et al. 1979, ApJ, 234, L9Titarchuk, L., 1994, ApJ, 434, 570Vaughan, S., Reeves, J., Warwick, R., & Edelson, R., 1999, MNRAS, 309, 113Véron-Cetty, M.-P., Véron, P., & Gonçalves, A. C. 2001, A&A, 372, 730Vignali, C., Brandt, W. N., & Schneider, D. P. 2003, AJ, 125, 433Wandel, A., & Petrosian, V. 1988, ApJ, 329, 11Wang, T., Brinkmann, W., & Bergeron, J. 1996, A& A, 309, 81Watson et al. 2007, "The
XMM-Newton serendipitous survey. V. The Second
XMM-Newton
Serendipitous Source Catalogue," , in preparation.Yuan, W., Brinkmann, W., Siebert, J., Voges, W. 1998, A&A, 330, 108Yuan, W., Zhou, H. Y., Komossa, S., Dong, X. B., Wang, T. G., Lu, H. L., & Bai, J. M. 2008, ApJ,685, 801Yuan, W., Liu, B. F., Zhou, H., Wang, T. G., 2010, ApJ, 723, 508Zhou, H.-Y., Wang, T.-G., Yuan, W., Lu, H.-L., Dong, X.-B., Wang, J.-X., & Lu, Y.-J. 2006, ApJS,166, 128 21 –Zhou, H.-Y., et al. 2007, ApJ, 658, 13Zoghbi, A., Fabian, A. C., & Gallo, L. C. 2008, MNRAS, 391, 2003
This preprint was prepared with the AAS L A TEX macros v5.2.
22 –Table 1. Basic parameters of the sample objects
No. Name SDSS Name z log λ L λ FWHM(H β ) FWHM(H α ) F(H α bc ) R logM BH logL bol /L Edd (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)1 J0107+1408 J010712.0+140845 0.076 42.96 787 ±
31 709 ±
12 1500 ±
12 0.35 ± ±
28 1090 ±
17 2121 ±
24 0.92 ± ±
27 1002 ±
13 3058 ±
23 1.37 ± ±
95 810 ±
34 2249 ±
36 0.92 ± ±
75 1216 ±
35 1056 ±
17 0.28 ± ±
60 866 ±
21 1158 ±
14 0.65 ± ±
41 571 ±
18 1337 ±
17 0.97 ± ±
23 1200 ±
332 1143 ±
255 0.57 ± ±
27 709 ±
15 11669 ±
118 0.83 ± ±
42 1044 ±
16 1929 ±
15 0.35 ± ±
41 694 ±
16 1471 ±
16 0.45 ± ±
27 954 ±
11 1735 ±
13 0.76 ± ±
38 886 ±
15 4369 ±
38 1.11 ± - ); Col.(6): H β linewidth (km s - ); Col.(7): H α linewidth (km s - ); Col.(8): H α broad component flux (10 - ergs s - cm - ); Col.(9): the optical Fe II strength relative to the H β broad component; Col.(10): black hole mass (M ⊙ ); Col.(11): Eddington ratio
23 –Table 2. Log of
XMM-Newton observations
Name Date Off-Axis Exposure time NotePN MOS1 MOS2(1) (2) (3) (4) (5) (6) (7)J0107+1408 2005-07-22 0.0 15.1 27.1 27.1 cJ0740+3118 2001-04-19 12.7 - 2.2 - aJ0922+5120 2005-10-08 0.0 5.2 19.2 - cJ0940+0324 2005-10-30 11.8 21.9 26.1 26.2 bJ1000+5536 2001-04-13 12.8 - - 8.2 a2003-10-14 13.5 14.2 - - aJ1114+5258 2003-04-25 11.3 3.9 6.6 7.0 aJ1140+0307 2005-12-03 0.0 30.8 38.5 39.2 cJ1231+1051 2003-07-13 12.0 - 45.3 - a2005-12-13 12.0 - - 68.2 a2005-12-17 12.0 - 91.6 91.6 aJ1246+0222 2001-06-17 0.0 3.1 - - cJ1331-0152 2001-07-29 12.0 - 32.3 32.3 bJ1357+6525 2005-04-04 0.0 14.5 21.1 20.6 cJ1415-0030 2003-02-08 10.0 9.5 13.7 14.2 bJ2219+1207 2001-06-07 0.1 7.2 - - cNote. — Col.(2): observation date; Col.(3): off axis angle in arcmin; Col.(4)–(6):cleaned exposure time of the three EPIC cameras in kilo-second; Col.(7): energyrange in which source spectrum is extracted, (a) 0.2–2.4 keV, (b) 0.2–7 keV, (c)0.2–10 keV.
24 –Table 3:
XMM-Newton spectral fits with an absorbed power-law model
Name N GalH N inH Γ soft log L soft χ ν /dof Γ hard log L hard χ ν /dof EW(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)J0107+1408 3.37 4 . ± . . ± .
08 43.07 1.0/230 2.26 ± < . ± .
30 44.72 c-stat - - - -J0922+5120 1.20 2.7 ± ± ± < . ± .
03 43.00 1.0/244 2.02 + . - . < ± . ± .
25 43.14 1.5/18 - - - -J1114+5258 0.99 - 2 . ± .
07 43.16 0.8/60 - - - -J1140+0307 1.88 1.3 ± . ± .
03 43.44 1.3/375 2.06 ± < . ± .
18 44.10 0.9/26 - - - -- - 2.80 ± . ± .
07 43.93 1.0/69 - - - -J1246+0222 1.74 1.83 ± . ± .
06 43.65 1.2/198 2.25 + . - . < . ± .
05 43.65 1.0/58 1.95 + . - . < . ± . . ± .
06 43.43 1.1/265 2.29 + . - . < . ± .
06 43.49 1.0/104 2.21 + . - . < . ± . . ± .
07 44.19 1.0/264 2.39 ± < Note. — Col.(2): Galactic column density in 10 cm - ; Col.(3): column density of intrinsic neutral absorptionin the object’s rest frame in 10 cm - ; Col.(4): fitted power-law photon index in the soft X-ray band (0.2–2.4 keV);Col.(5): absorption corrected luminosity in 0.2-2.4 keV in ergs s - ; Col.(6): reduced χ ; Col.(7): fitted power-lawphoton index in the hard X-ray band (mostly in 2–10 keV; see Table 2); Col.(8): absorption corrected luminosity in2–10 keV in ergs s - ; Col.(9): reduced χ ; Col.(10): rest frame equivalent width of the Fe K α line in units of eV
25 –Table 4:: Results of spectral fits in 0.2-10 keV with differentmodels for the soft excessBlackbodyName Γ kT (keV) BB/total ( a ) χ /dofJ0107+1408 2.39 + . - . ± .
02 0.166 277/264J0922+5120 3.18 ± ± ± + . - . ± ± ± ± ± ± ± ± + . - . ± ± ± Γ kT plasma (keV) τ χ /dofJ0107+1408 2.32 ± .
06 0.25 ± .
07 10.9 ± . ± ± + . - . ± .
19 0.30 ± .
10 13.5 ± . ± .
15 0.22 ± .
03 18.7 ± . ± .
23 0.21 ± .
02 21.0 ± . ± .
21 0.17 ± .
07 17.1 ± . ± .
04 0.19 ± .
02 23.5 ± . ± .
20 0.20 ± .
06 20.0 ± . ± .
15 0.21 ± .
04 18.1 ± . Γ r in / r g log ξ kT (keV) ( b ) Flux Frac. ( c ) χ /dofJ0107+1408 2.24 ± + . - ± ± ± + . - + . - . ± ± + . - +- . - 0.9 273/271J1140+0307 2.38 ± + . - + . - . - 0.9 524/470J1246+0222 2.43 ± + . - +- . - 1.0 212/210J1331-0152 2.29 ± + . - +- . - 0.3 69/66J1357+6525 2.19 ± + . - . + . - . - 0.5 320/301J1415-0030 2.24 ± + . - + . - . - 0.7 104/109J2219+1207 2.43 ± + . - +- . + . - . Γ log ξ N H (10 cm - ) σ (v/c) χ /dofJ0107+1408 2.47 ± ± +- +- . ± ± +- +- . ± ± +- +- . ± ± +- +- . ± ± +- +- . ± ± + - +- . ± ± +- +- . ± ± .
50 9 +- +- . ± ± +- +- . Notes. ——For SDSS J0107+1408 with free absorption in the fitting. For SDSS J1415-0030an additional ionized absorption is applied (see text). The blank parameter errors denote that theupper or lower limits are outside of the tabulated parameters range, which are considered to be notphysically meaningful. ( a ) Luminosity ratio of the blackbody component to the total componentin 0.5–2 keV. ( b ) Inferred temperatures of an additional blackbody component. ( c ) Flux ratio of thereflected component to the total component in 0.2–10 keV. 27 –Fig. 1.— Distributions of the Fe II to H β flux ratio (R ), black hole mass and the Eddingtonratio for our sample (red), the total Zhou’06 NLS1s sample (black) and the whole subsample withFWHM(H β ) ≤ - (blue).Fig. 2.— Composite optical spectrum of our sample objects derived from their individual SDSSspectra. The strong Fe II multiplet emission is characteristic of typical NLS1 spectra. 28 –Fig. 3.— Confidence contours (at the 68%, 90% and 99% confidence levels) of the mean and stan-dard deviation of the intrinsic distributions (assumed to be Gaussian) of the soft ( Γ s ) and hard ( Γ h )X-ray photon indices for our VNLS1, which are derived using the Maximum-likelihood method(see text). Pluses indicate the best-estimated values. 29 –Fig. 4.— XMM-Newton spectra of the four objects among our VNLS1 sample (PN spectra exceptfor J1331-0152 of which MOS spectra are used). The power-law model fitted to 2–10 keV spectraand its extrapolation to the soft X-ray band is also shown. Soft X-ray excess emission is clearlypresent. 30 –Fig. 5.— The observed temperature of the soft excess is plotted versus the black hole mass. Filledcircles are our results, stars for the radio-quiet PG quasars (Piconcelli et al. 2005) and diamondsfor type 1 AGNs (Crummy et al. 2006). The dotted-dashed lines are the maximum temperatureexpected from the accretion disc. 31 –Fig. 6.— Soft X-ray excess strength, parameterized as the ratio of the blackbody to the total lumi-nosity in the 0.5–2 keV range, versus the linewidth for the VNLS1s in our sample (open circles)and the radio-quiet PG quasars (diamonds) in Piconcelli et al. (2005). 32 –Fig. 7.— Residuals of spectral fits to the
XMM-Newton spectra with various models to account forthe soft X-ray excess, which are, from top to bottom for each panel, blackbody, Comptonization,disk reflection, and smeared absorption model. An additional blackbody component is added inthe disk reflection model for J0107+1408, J0922+5120 and J2219+1207. 33 –Fig. 8.— Soft X-ray photon index Γ s versus H β linewidth relation for our VNLS1s (filled circles).Stars are the results from Grupe et al. (2010). 34 –Fig. 9.— XMM-Newton
X-ray lightcurves in the 0.2–10 keV band for five of the VNLS1s in oursample. The time binsize is 200 s, except for SDSS J133141.0-015213 (500 s). 35 –Fig. 10.— Dependence of α ox on the 2500 Å monochromatic luminosity for the VNLS1s in oursample (filled dots) and the ordinary NLS1s (open squares) in the sample of Gallo (2006a). Thesolid line represents the relation for radio-quiet type 1 AGNs given in Strateva et al. (2005). 36 –Fig. 11.— Relationship between the hard X-ray (2–10 keV) photon index and the Eddington ratiofor our sample objects. Filled circles represent the results from the fits with a simple power-law model and open circles from a power-law plus disk reflection model. The solid line is theextrapolation of the relation suggested by Risaliti et al. (2009) ( M BH estimated from H β , the sameas in our paper) and the dash-dotted line represents the dispersion. The vertical dotted line marksthe higher end of the L / L Edd range in the Risaliti’s sample. 37 –
A. COMPARISONS WITH PREVIOUS RESULTS FOR INDIVIDUAL OBJECTS
For six objects in our sample, the
XMM-Newton data have been presented previously. Herewe compare our results with those in the literatures.
J0107+1408, J140+0307, J1357+6525 — These three objects were studied as IMBH AGNsby Miniutti et al. (2009) and Dewangan et al. (2008). For the model fits with blackbody, ionizeddisk reflection, and smeared absorption, our results are consistent with the previous ones except thatthe disk ionization parameters of ours are somewhat higher than those in Miniutti et al. (2009). Inaddition, we fit the spectra with the Comptonization and ‘p-free’ model, which was not consideredin the previous papers.
J1246+0222 — The fitting results for the soft X-ray excess component are consistent withthose presented in the literatures (Porquet et al. 2004; Crummy et al. 2006; Middleton & Done2007). However, in the simple power-law fitting we find a soft X-ray photon index of 2.97 ± ± J1415-0030 — The