EDP Sciences
Free Access
Issue
A&A
Volume 543, July 2012
Article Number A143
Number of page(s) 8
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201118477
Published online 12 July 2012

© ESO, 2012

1. Introduction

The chemical composition of galaxies is a powerful tool to distinguish various evolutionary scenarios because it is the result of past star-formation activity and is also affected by gas inflows and outflows. The chemical properties of active galactic nuclei (AGN) are particularly interesting, because the huge luminosities of AGN enable us to accurately measure spectroscopic features of quasars even at high redshifts and thus to explore the chemical properties in the early universe. Another interesting aspect of assessing the chemical composition of AGNs is that it provides clues on the co-evolution between galaxies and their supermassive black holes (SMBHs), which has been inferred by the tight correlation between the mass of SMBHs (MBH) and their host spheroidal observed in the local Universe (e.g., Marconi & Hunt 2003; Ferrarese & Merritt 2000; Gebhardt et al. 2000). Matsuoka et al. (2011) reported a tight relationship between the metallicity in broad-line regions (BLRs) and MBH at z ~ 2 − 3, suggesting that the evolution of SMBHs is associated with the cumulative star formation in the host galaxies (see also Warner et al. 2004; Nagao et al. 2006b).

A widely adopted method of estimating the BLR metallicity (ZBLR) exploits the flux of nitrogen emission lines such as N vλ1240. Nitrogen is a secondary element and thus its relative abundance is proportional to the metallicity, [N/O]  ∝  [O/H] or, equivalently,  [N/H]  ∝  [O/H] 2 (see, e.g., Hamann & Ferland 1992). Based on this method, it has been found that the BLR in most quasars show super-solar metallicities (e.g., Nagao et al. 2006b), even for quasars at z ≳ 6 (Jiang et al. 2007; Juarez et al. 2009; Mortlock et al. 2011). The BLR metallicity is up to ZBLR ~ 10   Z in the most extreme cases although these high metallicities are hard to understand with ordinary galaxy chemical evolution models (e.g., Hamann & Ferland 1993). More interestingly, some quasars show extremely strong emission in N v and in other nitrogen lines (especially N iv]λ1486 and N iii]λ1750 semi-forbidden lines). The estimated ZBLR reaches  ~15 Z, or higher, and these quasars are classified as “N-loud quasars” (e.g., Baldwin et al. 2003; Bentz & Osmer 2004; Bentz et al. 2004). Here the definition of a N-loud quasar is that of Jiang et al. (2008); i.e., type 1 quasars with strong nitrogen emission of EWrest(N iv]λ1486)  >  3 Å or EWrest(N iii]λ1750)  >  3 Å (see Sect. 2). Note that the measurement of the N vλ1240 flux is sometimes uncertain because of the heavy blending with the Lyα emission (see, e.g., Nagao et al. 2006b); however, the N-loudness of these quasars is convincing because the well-isolated N iv]λ1486 and N iii]λ1750 are used for defining the N-loud quasar population. Standard chemical evolution models cannot predict these extremely high metallicities at any epoch (e.g., Ballero et al. 2008), therefore these N-loud quasars is a great challenge for galaxy evolutionary models.

Jiang et al. (2008) pointed out that N-loud quasars may not have such high ZBLR, but simply have an unusually high relative abundance of nitrogen in the BLR, mainly because the emission-line spectrum of N-loud quasars is not significantly different from that of typical quasars except for the nitrogen lines. This idea questions the use of the N v emission for ZBLR measurements in quasars. It is therefore very important to verify observationally if N-loud quasars have either high ZBLR or just high nitrogen relative abundances. More recently, Matsuoka et al. (2011) reported that the strength of nitrogen lines is correlated with the quasar Eddington ratio (Lbol/LEdd) and discussed the possibility that quasars with strong nitrogen emission are in a special phase of the evolutionary history of their host galaxies. These studies outline the importance of understanding the nature of N-loud quasars, especially in terms of their chemical evolution. Therefore it is important to assess the metallicity of BLRs and host galaxies of N-loud quasars with a method that does not rely on rest-frame UV broad nitrogen lines.

In this paper, we focus on the emission of the narrow-line region (NLR) in N-loud quasars. Since strong nitrogen lines from their BLRs suggest a high ZBLR, we examined the properties of the NLR to independently investigate the metallicity in the N-loud quasar population. While the BLR is located in very compact, subpc-scale regions around SMBHs, NLR gas clouds are distributed on kpc scales, i.e. on the scales of the host galaxies, which enabled us to study the possible relationships between the properties of N-loud quasars and their host galaxies. Throughout this paper, we adopt a cosmology with H0 = 70 km s-1 Mpc-1, Ωm = 0.3, and ΩΛ = 0.7.

2. Sample selection, observation, and the data reduction

Table 1

Observational properties of SDSS J1707+6443.

To investigate NLR properties of N-loud quasars in detail, we focused on SDSS J1707+6443 at z = 3.163 from the N-loud quasar catalog of Jiang et al. (2008). This catalog contains 293 N-loud quasars selected from the Fifth Data Release quasar catalog (Schneider et al. 2007) of the Sloan Digital Sky Survey (SDSS; York et al. 2000). The N-loud quasars in the catalog of Jiang et al. (2008) were selected according to the following criteria: (i) i′ < 20.1; (ii) 1.7 < z < 4.0; and (iii) EWrest(N iv]λ1486)  >  3 Å or EWrest(N iii]λ1750)  >  3 Å. Among the N-loud quasars listed in the catalog of Jiang et al. (2008), we focused on SDSS J1707+6443 because it is relatively bright and its redshift is adequate for detecting redshifted NLR lines without suffering from the deep atmospheric absorption (Fig. 1). In Fig. 2, we show the frequency distributions of EWrest(N iv]λ1486) and EWrest(N iii]λ1750), and also where SDSS J1707+6443 is located in these frequency distributions (see also Table 1). The averages and standard deviations of EWrest(N iv]λ1486) distributions are 4.6 Å  ±  1.7 Å, with minimum and maximum values in the parent sample of 1.2 Å and 8.9 Å. The averages and standard deviations of EWrest(N iii]λ1750) are 5.6 Å  ±  2.6 Å, with minimum and maximum values of 2.5 Å and 27.1 Å. Figure 2 does not suggest that SDSS J1707+6443 is located at the lowest end in the distribution of EWrest(N iii]λ1750), because some of N-loud quasars do not always show detectable emission lines of N iv]λ1486 and N iii]λ1750. More specifically, only 50 N-loud quasars out of 293 show detectable N iv]λ1486 emission while 280 out of 293 show detectable N iii]λ1750 emission (Jiang et al. 2008). Although SDSS J1707+6443 has a high N iv]λ1486/N iii]λ1750 ratio with respect to typical N-loud quasars, we do not discuss this question in this paper since it is beyond the scope of this work. The SDSS spectrum and the basic observational properties of SDSS J1707+6443 are shown in Fig. 3 and Table 1.

thumbnail Fig. 1

Absolute i′-band magnitudes of 293 N-loud quasars as a function of redshift. Filled circle and crosses denote SDSS J1707+6443 and the other N-loud quasars, respectively. Data are taken from Jiang et al. (2008).

Open with DEXTER

thumbnail Fig. 2

Histograms of EWrest of N iv]λ1486 (upper panel) and N iii]λ1750 (lower panel). The arrows denote the EW   rest values of SDSS J1707+6443. Data are taken from Jiang et al. (2008).

Open with DEXTER

To assess the NLR properties of type 1 quasars, it is necessary to investigate rest-frame optical spectra and consequently, near-infrared spectroscopic data are required for this target. Therefore we observed SDSS J1707+6443 with MOIRCS (Ichikawa et al. 2006; Suzuki et al. 2008), the near-infrared spectrograph at the Subaru Telescope, on 30 May 2009. We used the HK grism, which yielded a wavelength coverage of 1.4   μm ≲ λobs ≲ 2.3   μm. By using the 0.6′′-wide slit, the resulting spectral resolution was R ~ 570, which was measured through the widths of the observed OH airglow emission. The total integration time was 1800 s, consisting of six 300 s independent exposures. We also observed HIP 86687 (A2 star with HVega = 8.9 and an assumed effective temperature of 8810 K) for the flux calibration and the telluric absorption correction. During the observation, the typical seeing size was 0.8′′ in the optical.

Standard data processing was performed using the available IRAF1 routines. Frames were flat-fielded using domeflat images, and sky emission subtraction was performed by subtracting pairs of subsequent frames with the target at different positions along the slit. Cosmic-ray events were then removed by using the lineclean task. We extracted the one-dimensional spectrum of the target using the apall task, with an aperture size of  ± 5 pixels (1.17′′) from the emission center and, in this process, sky residuals were removed. Wavelength calibration was performed using OH sky lines. Finally, flux calibration and telluric absorption correction were carried out using the observed spectra of HIP 86687. The final processed spectrum was obtained by combining all single exposure frames.

thumbnail Fig. 3

Optical spectrum of SDSS J1707+6443 obtained from the SDSS database. Line IDs are given for strong emission lines. The inset shows spectral features around the C iv emission, where N iv]λ1486 and N iii]λ1750 lines are seen.

Open with DEXTER

To check the consistency of the flux calibration between the SDSS data and our MOIRCS data, we extrapolated the UV continuum emission toward the longer wavelength and compared the flux density at λrest = 5100 Å between the extrapolated spectrum and our MOIRCS spectrum. The adopted spectral index is derived by fitting the SDSS spectrum at the wavelength regions where strong emission-line features are not present (λrest ~ 1350   Å,1450   Å,1670   Å, and 1970 Å). Here we assume a constant spectral index at λrest < 4000   Å (see Vanden Berk et al. 2001). The discrepancy in the continuum flux is  ~ 10%, and therefore we conclude that the flux calibration is consistent between the SDSS spectrum and our MOIRCS spectrum.

thumbnail Fig. 4

Near-infrared spectrum of SDSS J1707+6443 obtained in our MOIRCS run (middle panel), with the Mauna Kea atmospheric transmission (upper panel) and the typical sky spectrum obtained during our MOIRCS run (lower panel). Important emission lines are labeled in the middle panel.

Open with DEXTER

3. Result

The final processed spectrum of SDSS J1707+6443 is shown in Fig. 4, with the Mauna Kea atmospheric transmission2 and a typical sky spectrum obtained during our MOIRCS observing run. A prominent broad Hβ emission is detected at λobs ~ 2.0 μm, a typical spectral feature from quasar BLRs. In addition to the broad Hβ emission, some narrow emission lines from the NLR are also detected, like the [O iii] doublet in the K-band, a feature common in several high-z quasars (e.g., McIntosh et al. 1999; Netzer et al. 2004; Marziani et al. 2009; Greene et al. 2010). Interestingly, we also detected the quite rare [Ne iii]λ3869 emission line but not the [Ne v]λ3426 and [O ii]λ3727 lines.

Table 2

Measured spectral features.

To measure the emission-line fluxes and velocity widths, we fit the spectral features by using the specfit routine (Kriss 1994). Here we adopted a single Gaussian profile for forbidden narrow lines and a double Gaussian profile for Hβ line. The measured emission-line properties are summarized in Table 2, where the presented quantities are based on the fitting models, not on the actual data themselves. In Fig. 5, we show fitting results and residuals. The measured velocity widths (ΔvFWHM ~ 600 − 850 km s-1 for the NLR lines and ΔvFWHM ~ 5500 km s-1 for the BLR line) are consistent with typical values for NLR and BLR emission lines seen in type 1 AGNs (426 ± 251 km s-1 and 4420 ± 3210 km s-1 for the NLR and BLR lines; Ho & Kim 2009). Note that the systematic errors given for the emission-line widths were estimated by applying some fitting functions for emission lines (such as Gaussian, Lorentzian, and so on) and examining the standard deviation of the width. Here the statistical errors in the velocity width are smaller than the systematic errors, given the achieved signal-to-noise ratio.

thumbnail Fig. 5

Fitting result (upper panel) and the residual (lower panel) for the processed MOIRCS spectrum. See the main text for details of the fitting procedure.

Open with DEXTER

4. Discussion

4.1. The chemical properties of the narrow line region

One of the aims of this study is to estimate the metallicity of the host galaxy of the N-loud quasar SDSS J1707+6443. More specifically, we wish to test whether this quasar has extremely metal-rich gas clouds, as expected from its strong broad nitrogen emission lines. Although strong broad nitrogen lines are usually interpreted as indications of high ZBLR (for both permitted lines and semi-forbidden lines; see, e.g., Shields 1976; Hamann & Ferland 1992), Jiang et al. (2008) pointed out that N-loud quasars might simply have unusually high nitrogen relative abundances. This is because the emission-line spectrum of the N-loud quasars is similar to that of common type 1 quasars. In particular, many N-loud quasars show no anomalous behavior in other broad emission-line flux ratios, not involving nitrogen, but are sensitive to ZBLR, such as (Si iv+O iv])/C iv (Nagao et al. 2006b; Juarez et al. 2009; Simon & Hamann 2010). However, the nuclear BLR involves only a very small fraction of the gas content in the galaxy and, therefore, may well not be representative of the metallicity in the host galaxy. As a consequence, it may be more instructive to investigate the metallicity in the NLR (ZNLR) through the rest-frame optical spectrum to investigate whether the high metallicities inferred for the BLR are confirmed on the larger scales traced by the NLR.

We focus first on the metallicity dependence of the line emissivity on the physical conditions typical of the NLR. The emissivity of collisionally excited emission lines strongly depends on the gas temperature. The equilibrium temperature of ionized gas clouds depends on gas metallicity, because metal emission lines are the main coolants of these clouds. Therefore, collisionally excited emission lines from the NLR become fainter for increasing ZNLR (e.g., Nagao et al. 2006a; Matsuoka et al. 2009). To show this effect more explicitly, we computed simple photoionization model calculations with Cloudy, version 08.00 (Ferland et al. 1998; Ferland 2006). We assumed ionization-bounded plane-parallel gas clouds with a constant density, photoionized by the typical spectral energy distribution for quasars (Mathews & Ferland 1987; Ferland 2006). We examined the metallicity dependence of the equivalent width (EW) of the [O iii] emission with the relative elemental abundance ratios fixed to the solar values except for nitrogen. We assumed that the nitrogen relative abundance (N/H) scales with the square of the metallicity. In Fig. 6, we show photoionization model results for gas densities of nH = 101 cm-3 and 104 cm-3, and ionization parameters of U = 10-3.5 and 10-1.5. The emissivity curves in this figure are normalized by their peak to highlight their dependence on the gas metallicity. Clearly, the EW([O iii]) decreases rapidly as the metallicity increases, and the resulting EW([O iii]) is quite small at ZNLR > 5   Z. Note that we cannot derive accurate values of ZNLR based only on [O iii], since detailed metallicity studies require multiple emission lines (e.g., Storchi-Bergmann et al. 1998; Nagao et al. 2002; Iwamuro et al. 2003; Groves et al. 2006; Nagao et al. 2006a; Matsuoka et al. 2009). However, the aim of this paper is only to assess whether the NLR in SDSS J1707+6443 has a high metallicity, not to derive its accurate value. These results suggest that we can distinguish whether or not N-loud quasars have a relatively high ZNLR with respect to other type 1 quasars by comparing the EW([O iii]) distributions of the two populations.

In Fig. 7, we show the frequency distribution of EWrest([O iii]) of type 1 quasars at 1 < z < 4, which are compiled from the works of McIntosh et al. (1999), Netzer et al. (2004), Marziani et al. (2009), and Greene et al. (2010) (shown with a solid histogram in Fig. 7). In these four papers, there are 86 [O iii]-detected quasars, whose average and median EW([O iii])rest values are 16.5 Å and 13.0 Å with the standard deviation of 14.6 Å (the minimum and maximum values are 0.4 Å and 73.0 Å, respectively). Since the EWrest([O iii]) of SDSS J1707+6443 is 17.4   Å (Table 2), the [O iii] emission of SDSS J1707+6443 is not weaker than that of typical type 1 quasars. In Fig. 7 we also show the frequency distribution of EWrest([O iii]) of SDSS quasars at 0 < z < 1 (shown with filled circles with error bars in Fig. 7). This is the distribution by Risaliti et al. (2011), who used the spectral measurements of SDSS DR5 quasars (Schneider et al. 2007) performed by Shen et al. (2008). The frequency distribution of EWrest([O iii]) of the 0 < z < 1 SDSS sample (Risaliti et al. 2011; filled circles in Fig. 7) is not significantly different from that of the higher-z (1 < z < 4) sample (solid histogram in Fig. 7). The EWrest([O iii]) value of SDSS J1707+6443 is higher than the peak value of these distributions. Therefore we conclude that the EWrest([O iii]) of SDSS J1707+6443 is not significantly smaller than that of typical type 1 quasars in both low-z and high-z samples. Note that the actual relative [O iii]λ5007 strength of SDSS J1707+6443 with respect to the whole parent sample of type 1 quasars could be higher, because only [O iii]-detected quasars (i.e., relatively strong [O iii] emitters) are selectively shown in Fig. 7.

thumbnail Fig. 6

Predicted equivalent width of [O iii]λ5007, as a function of ZNLR. Solid, dashed, dot-dashed, and dotted lines denote the models with (log nH, log U) = (1, –1.5), (1, –3.5), (4, –1.5), and (4, –3.5), respectively. The EW predictions are normalized by their peak values.

Open with DEXTER

The relatively large EW([O iii])rest of SDSS J1707+6443 with respect to the global population of AGNs (Fig. 7) suggests that the NLR in SDSS J1707+6443 is not characterized by a very high ZNLR, when the photoionization model results shown in Fig. 6 are taken into account. If a positive correlation between ZNLR and ZBLR is assumed, a lack of very high metallicity clouds in the NLR seems to be inconsistent with the strong broad nitrogen lines seen in the rest-frame UV spectrum of this object. Obviously these are two possible scenarios to explain this inconsistency: (1) the strong UV broad nitrogen lines are caused by a very high relative abundance of nitrogen in the BLR (with respect to non-N-loud quasars) and not by a BLR metallicity extremely higher than non-N-loud quasars, or (2) the BLR metallicity is significantly higher than the NLR metallicity, which better represents the metallicity of the host galaxy. The relation between ZBLR and ZNLR is not yet well understood observationally, although Fu & Stockton (2007, 2008, 2009) reported that these two quantities are probably related with each other, at least in low-z (z < 0.5) quasars. Whatever case applies, (1) or (2), the strong broad nitrogen lines of SDSS J1707+6443 are not consistent with the possibility that the host galaxy of this quasar is characterized by a very high metallicity. This implies that broad UV nitrogen lines of quasars are not (at least in some cases) a good tool to explore the chemical evolution of quasar host galaxies. Here we note that the spatial scale of NLRs is far larger than that of BLRs but does not necessarily coincide with the spatial scale of their host galaxies; we show for SDSS J1707+6443, however, that the NLR spatial scale corresponds to the host-galaxy scale (~kpc scale; see Sect. 4.3). Since quasar spectra are frequently used to investigate the chemical evolution at high redshifts, it will be crucial to examine whether the broad UV lines of quasars are good (or bad) tracers of the metallicity through more detailed studies for larger samples of AGNs (see, e.g., Matsuoka et al. 2011).

thumbnail Fig. 7

Solid-line histogram denotes the compiled data of high-z type 1 quasars from the literature (McIntosh et al. 1999; Netzer et al. 2004; Marziani et al. 2009; Greene et al. 2010). Filled circles with error bars denote the data of SDSS type 1 quasars at 0 < z < 1 (Risaliti et al. 2011). The frequency distributions of the two datasets are normalized by the number of objects in the bin at the peak of their frequency distributions. The arrow denotes the EWrest([O iii]) value of SDSS J1707+6443 measured in our work.

Open with DEXTER

4.2. Black hole mass and the Eddington ratio

The results in the previous section suggest that the strong nitrogen emission in the BLR of SDSS J1707+6443 (or, possibly, N-loud quasars in general) is not indicative of high metallicity in the NLR and hence in the host galaxy. It is important, however, to verify whether the properties of the broad nitrogen lines of SDSS J1707+6443 do follow the trends of the global population of high-z type 1 quasars. Recently Matsuoka et al. (2011) found that a high relative nitrogen abundance is seen in BLRs of high-Lbol/LEdd quasars. To examine whether SDSS J1707+6443 is consistent with the result of Matsuoka et al. (2011), we estimated MBH and Lbol/LEdd based on the velocity widths of C iv in the SDSS spectrum and Hβ in our MOIRCS spectrum. Note that it is debatable whether C iv-based or Hβ-based estimations are more accurate (see, e.g., Netzer et al. 2004; Peterson et al. 2004; Denney et al. 2009). Therefore we used both emission lines to minimize possible systematic errors in the estimation of those parameters.

We derived MBH for SDSS J1707+6443 by adopting the calibrations given by Vestergaard & Peterson (2006), (1)for C iv, and (2)for Hβ, where FWMHCIV and FWHMHβ are the velocity width of the C iv and Hβ emission in full-width at half maximum, respectively. For estimating the Eddington ratio (Lbol/LEdd), we adopted a bolometric correction of Lbol = 9.26 λLλ5100 (Shen et al. 2008). The derived values are log (MBH/M) = 8.98 and log (Lbol/LEdd) =  −0.53 when using C iv, and log (MBH/M) = 9.73 and log (Lbol/LEdd) =  −0.21 when using Hβ, respectively. Accordingly we adopted their means, log (MBH/M) = 9.50 and log (Lbol/LEdd) =  −0.34. The derived Eddington ratio is relatively high (regardless of the adopted emission line for deriving MBH), with respect to the frequency distribution of the Eddington ratio of SDSS type 1 quasars at similar redshifts (−0.48 ± 0.41 for 3144 quasars at 3.0 < z < 3.3; Shen et al. 2011).

Our results is consistent with the finding of Matsuoka et al. (2011) that quasars with a high accretion rate are not necessarily characterized by high metallicities, but are characterized by high nitrogen abundances (resulting in stronger nitrogen lines). In the scenario proposed by Matsuoka et al. (2011), the high black hole accretion is delayed by a few 100 Myr, relative to the main episode of star formation, when intermediate-mass stars have evolved and have enriched the ISM with nitrogen.

4.3. Emission-line diagnostics in the narrow line region

By combining the emission-line fluxes of [O iii]λ5007 and [Ne iii]λ3869 and the upper limit on the flux of [O ii]λ3727, it is possible to investigate the properties of NLR gas clouds. This analysis enables us to study the physical properties of gas clouds in the host galaxy, because the NLR is extended on galactic scales in contrast to the BLR. Moreover, since the redshift of SDSS J1707+6443 (z ~ 3) corresponds to the peak of the global quasar activity (e.g., Richards et al. 2006), the gas properties of the quasar host galaxy are interesting to explore the interplay between AGNs and their host galaxies (i.e., the galaxy-SMBH coevolution). In this context, host galaxies of the N-loud quasars are particularly interesting since they may be in a special evolutionary stage, as mentioned in Sect. 4.2. Note that NLR emission lines other than [O iii]λ5007 have only rarely been observed in luminous high-z type 1 quasars, which prevents a detailed investigation of NLR gas properties similar to our MOIRCS spectrum.

thumbnail Fig. 8

Emission-line flux ratios of [O ii]λ3727/[O iii]λ5007 versus [Ne iii]λ3869/[O iii]λ5007. The filled circle with an arrow denotes the data of SDSS J1707+6443 (where the 3σ upper limit is adopted for the [O ii] flux), and cross symbols denote the data of the SDSS low-z (z ~ 0.7) type 1 quasars. The arrow at the upper-left corner in the panel is the reddening vector for observed data in the case of AV = 1.0 mag, adopting the extinction curve of Cardelli et al. (1989). The grids are predicted emission-line flux ratios from Cloudy model runs, adopting ZNLR = 3   Z and varying the hydrogen density and the ionization parameter.

Open with DEXTER

In Fig. 8, we place SDSS J1707+6443 on the diagram comparing the distributions of the emission-line flux ratios of [O ii]λ3727/[O iii]λ5007 and [Ne iii]λ3869/[O iii]λ5007. For comparison, we also plot the data of the SDSS DR7 type 1 quasar sample (Shen et al. 2011). We checked their SDSS archival spectra and selected quasars whose [O ii]λ3727, [Ne iii]λ3869, and [O iii]λ5007 are significantly detected (i.e., S/N > 10). As a consequence we plot the emission-line flux ratio of 25 SDSS DR7 type 1 quasars, whose redshift range is 0.37 < z < 0.80 with the median redshift of 0.65. Note that the average luminosity λLλ(5100) for the 25 SDSS type 1 quasars is 1045.77 erg s-1, that is  ~1 dex lower than the corresponding value for SDSS J1707+6443 (1046.66 erg s-1). As shown in Fig. 8, these 25 SDSS type 1 quasars show completely different flux ratios from SDSS J1707+6443; i.e., SDSS J1707+6443 shows a higher [Ne iii]λ3869/[O iii]λ5007 ratio and a lower [O ii]λ3727/[O iii]λ5007 ratio than SDSS type 1 quasars. These differences can be interpreted qualitatively if the NLR in SDSS J1707+6443 is characterized by a much higher average density than the NLRs in SDSS type 1 quasars at z < 1. This is because the [O ii]λ3727 emission and partly [O iii]λ5007 emission are suppressed by the collisional de-excitation process in high-density clouds with nH ~ 106 cm-3 or higher, by taking the critical densities of these emission lines (ncr([O ii]) = (3.3 − 14) × 103 cm-3, ncr([O iii]) = 7.0 × 105 cm-3, and ncr([Ne iii]) = 9.7 × 106 cm-3) into account.

To investigate the differences in the NLR emission-line flux ratios between SDSS J1707+6443 and lower-z type 1 quasars more quantitatively, we have performed Cloudy model runs for the parameter ranges of nH = 102 − 106 cm-3 and U = 10-4.0 − 10-1.5. Here we adopted a metallicity of ZNLR = 3  Z, which is the typical value for NLR clouds (e.g., Groves et al. 2006), because we concluded in Sect. 4.1 that SDSS J1707+6443 is not characterized by very high ZNLR. The other adopted parameters such as the input SED are the same as those described in Sect. 4.1. The results of the model runs are overlaid in Fig. 8. The NLRs in the SDSS type 1 quasars are characterized by nH ~ 104.5 cm-3 and U ~ 10-3.5 − 10-2.5, consistent with the parameter ranges inferred by previous studies (e.g., Nagao et al. 2001b). On the other hand, the NLR in SDSS J1707+6443 is characterized by a higher density, nH ~ 106 cm-3 or more, although the ionization parameter is similar to that of type 1 SDSS quasars. These results are not sensitive to the adopted metallicity.

There are two possible scenarios to explain the distinct physical properties of SDSS J1707+6443. One possibility is considering a strong contribution from the inner wall of the dusty torus. Because the clouds located at the inner side of dusty tori are characterized by a high density and photoionized by the strong central continuum radiation, emission lines with high critical densities are radiated from these clouds (e.g., Murayama & Taniguchi 1998b; Nagao et al. 2000, 2001a). If these high-density clouds at the inner dusty torus show strong NLR emission, its emission dominates the whole NLR emission and then low-ncr emission lines become relatively weak, explaining the observed emission-line flux ratios of SDSS J1707+6443 (see also Murayama & Taniguchi 1998a; Nagao et al. 2001c). Another possibility is the existence of high-density clouds distributed on galactic scales (i.e.,  ~kpc). Such a situation has been proposed for high-z and/or high-luminosity quasars (Netzer et al. 2004; see also Ho 2005), possibly related to violent star-formation activity in the host galaxy. Note that the latter scenario is not ruled out by the non-detection of the [O ii]λ3727 emission, which is used as an indicator of the star formation rate (e.g., Gallagher et al. 1989; Kennicutt 1998; Ideue et al. 2009). This is because the [O ii]λ3727 emission is significantly suppressed in high-density H ii regions when the density is higher than 104 cm-3, even when the star formation activity is very vigorous.

To distinguish between the above two scenarios, we estimated the typical distance of the ionized clouds from the nucleus in SDSS J1707+6443 by combining the estimates of nH and U obtained in the previous section. Since we already estimated Lbol of SDSS J1707+6443 in Sect. 4.1, the number of the ionizing photon can be estimated for a given spectral energy distribution (SED). Here we assumed the same SED as in the above photoionization model calculations (Mathews & Ferland 1987) and adopted the parameters of nH = 106 cm-3 and U = 10-3.5. Accordingly, we obtained the result that the typical distance of the clouds is  ~1.5 kpc, which is consistent with the latter scenario that considers high-density clouds in the host galaxy. Note that a smaller radius would have been derived by assuming a higher density, e.g.,  ~15 pc for nH ~ 108 cm-3. However, this is not plausible, because [O iii]λ5007 emission would be significantly suppressed at such high-densities due to the collisional de-excitation. We thus conclude that there are high-density clouds at the kpc scale in the host galaxy of SDSS J1707+6443, which dominate the emission of the detected NLR lines. The torus scenario is disfavored also in terms of the observed NLR velocity width. By assuming that the SMBH is the dominant source of the gravitational potential field in the nucleus of SDSS J1707+6443, the expected velocity width of [O iii]λ5007 is  ~1300 km s-1 if the most of this emission arises at  ~15 pc from the SMBH with MBH = 109.78   M (see Sect. 4.2). This expected velocity width is apparently larger than the observed width (617 km s-1; Table 2), suggesting that most of the [O iii]λ5007 emission in this object arises at larger spatial scales, which probably correspond to the scale of its host galaxy.

Here we briefly discuss the implication of the extended (~kpc) dense (~106 cm-3) gas clouds in SDSS J1707+6443 inferred from emission-line diagnostics. It is well known that these dense clouds in galaxies are closely related to the star-forming activity. Plume et al. (1997) reported their observations of the carbon monosulphide (CS) molecule for  >100 Galactic H ii regions and showed that their typical density is n ~ 106 cm-3 (but less than 2 × 107 cm-3). Such a high density for clouds in H ii regions has been inferred also from some other molecular-line radio observations (e.g., Hofner et al. 1996; Bergin et al. 1996; Lada et al. 1997). Although such a high density is not the “typical” density for H ii region clouds, the star formation efficiency of the denser clouds is actually much higher than less dense clouds. This is suggested by, e.g., a clear positive correlation between the LHCN/LCO ratio (i.e., the dense-gas fraction) and the LIR/LCO ratio (i.e., the star-formation efficiency) that is seen in H ii region clouds (e.g., Gao & Solomon 2004). Therefore high-density clouds generally dominate the star formation, when they exist (see also, e.g., Wang et al. 2011). A similar situation is also seen in high-z galaxies (e.g., Gao et al. 2007; Riechers et al. 2007, 2010, 2011) but with a significant difference compared to low-z galaxies; that is, the spatial extension of vigorous star-forming regions. The spatial scale of star-forming regions in high-z actively star-forming galaxies (such as sub-millimeter galaxies) extends up to  ~ kpc scales or more (e.g., Chapman et al. 2004; Walter et al. 2009), which is different from low-z actively star-forming galaxies such as ultra-luminous infrared galaxies (e.g., Downes & Solomon 1998; see also, e.g., Iono et al. 2009). These pictures well match with our results in the sense that dense gas clouds (~106 cm-3) are distributed at the  ~kpc scale in a galaxy at z ~ 3.2. Therefore, given these considerations, we speculate that there is ongoing vigorous star-forming activity in SDSS J1707+6443.

5. Summary

To assess the physical and chemical properties of the host galaxies of N-loud quasars, we analyzed the MOIRC near-infrared spectrum of SDSS J1707+6443, at z = 3.16, obtaining the following results:

  • this N-loud quasar shows strong [O iii]λ5007emission. Since photoionization models predict weak[O iii]λ5007 emission when the NLR metallicity isvery high (>5   Z), the detected strong [O iii]λ5007emission suggests that the NLR in this object is not characterizedby very high metallicities;

  • the UV nitrogen lines from BLRs are not ideal tools to discuss the chemical evolution of quasar host galaxies, because whatever the origin of strong broad nitrogen emission (high ZBLR, or high relative abundance of nitrogen with an ordinary metallicity), ZBLR are likely unrelated with the metallicities of the host galaxies;

  • the Eddington ratio of SDSS J1707+6443 is moderately high (Lbol/LEdd = 10-0.26), as derived from single-epoch MBH estimates using both C iv and Hβ emission lines. This is consistent with the discovery by Matsuoka et al. (2011) that a high relative nitrogen abundance is associated to quasars with high Eddington ratios;

  • the flux ratio of [O ii]λ3727/[O iii]λ5007 in SDSS J1707+6443 is significantly lower than that in lower-z type 1 quasars, indicating that NLR clouds in SDSS J1707+6443 are characterized by much higher densities than those in other type 1 quasars;

  • photoionization models suggest that those high-density clouds (with nH ~ 106 cm-3) are located at the kpc scales in the host galaxy of SDSS J1707+6443, and we speculate that this might be related to the vigorous star-formation activity in the host galaxy.

These results possibly reveal a strong connection between the high AGN activity (characterized by a high Eddington ratio) and the vigorous star formation activity suggested by the kpc-scale distribution of dense gas cloud in a N-loud quasar, SDSS J1707+6443. Currently it is not clear whether such a connection is seen also in other N-loud quasars generally or not, given the paucity of detailed near-infrared spectroscopic studies for N-loud quasars. Therefore it is highly interesting to examine whether such a situation is common in other N-loud quasars through more near-infrared spectroscopic observations for a large sample of N-loud quasars.


1

IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation.

2

Data obtained from the UKIRT web site.

Acknowledgments

We thank the Subaru Telescope staff for supporting our MOIRCS observation. We also thank the anonymous referee, Takayuki Saitoh, and Bunyo Hatsukade, for their useful comments. Funding for the creation and distribution of the SDSS Archive has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Aeronautics and Space Administration, the National Science Foundation, the US Department of Energy, the Japanese Monbukagakusyo, and the Max Planck Society. The SDSS web site is http://www.sdss.org/. We thank Gary Ferland for providing his photoionization code Cloudy to the public. N.A. is supported in part by a grant from the Hayakawa Satio Fund awarded by the Astronomical Society of Japan. T.N. is financially supported by JSPS (grant no. 23654068), the Kurata Memorial Hitachi Science and Technology Foundation, the Itoh Science Foundation, Ehime University (the Research Promotion Award), and Kyoto University (the Hakubi Project grant). K.M. acknowledges financial support from JSPS through JSPS Research Fellowships for Young Scientists. Y.T. is financially supported by JSPS (grant no. 23244031).

References

All Tables

Table 1

Observational properties of SDSS J1707+6443.

Table 2

Measured spectral features.

All Figures

thumbnail Fig. 1

Absolute i′-band magnitudes of 293 N-loud quasars as a function of redshift. Filled circle and crosses denote SDSS J1707+6443 and the other N-loud quasars, respectively. Data are taken from Jiang et al. (2008).

Open with DEXTER
In the text
thumbnail Fig. 2

Histograms of EWrest of N iv]λ1486 (upper panel) and N iii]λ1750 (lower panel). The arrows denote the EW   rest values of SDSS J1707+6443. Data are taken from Jiang et al. (2008).

Open with DEXTER
In the text
thumbnail Fig. 3

Optical spectrum of SDSS J1707+6443 obtained from the SDSS database. Line IDs are given for strong emission lines. The inset shows spectral features around the C iv emission, where N iv]λ1486 and N iii]λ1750 lines are seen.

Open with DEXTER
In the text
thumbnail Fig. 4

Near-infrared spectrum of SDSS J1707+6443 obtained in our MOIRCS run (middle panel), with the Mauna Kea atmospheric transmission (upper panel) and the typical sky spectrum obtained during our MOIRCS run (lower panel). Important emission lines are labeled in the middle panel.

Open with DEXTER
In the text
thumbnail Fig. 5

Fitting result (upper panel) and the residual (lower panel) for the processed MOIRCS spectrum. See the main text for details of the fitting procedure.

Open with DEXTER
In the text
thumbnail Fig. 6

Predicted equivalent width of [O iii]λ5007, as a function of ZNLR. Solid, dashed, dot-dashed, and dotted lines denote the models with (log nH, log U) = (1, –1.5), (1, –3.5), (4, –1.5), and (4, –3.5), respectively. The EW predictions are normalized by their peak values.

Open with DEXTER
In the text
thumbnail Fig. 7

Solid-line histogram denotes the compiled data of high-z type 1 quasars from the literature (McIntosh et al. 1999; Netzer et al. 2004; Marziani et al. 2009; Greene et al. 2010). Filled circles with error bars denote the data of SDSS type 1 quasars at 0 < z < 1 (Risaliti et al. 2011). The frequency distributions of the two datasets are normalized by the number of objects in the bin at the peak of their frequency distributions. The arrow denotes the EWrest([O iii]) value of SDSS J1707+6443 measured in our work.

Open with DEXTER
In the text
thumbnail Fig. 8

Emission-line flux ratios of [O ii]λ3727/[O iii]λ5007 versus [Ne iii]λ3869/[O iii]λ5007. The filled circle with an arrow denotes the data of SDSS J1707+6443 (where the 3σ upper limit is adopted for the [O ii] flux), and cross symbols denote the data of the SDSS low-z (z ~ 0.7) type 1 quasars. The arrow at the upper-left corner in the panel is the reddening vector for observed data in the case of AV = 1.0 mag, adopting the extinction curve of Cardelli et al. (1989). The grids are predicted emission-line flux ratios from Cloudy model runs, adopting ZNLR = 3   Z and varying the hydrogen density and the ionization parameter.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.