High metal content of highly accreting quasars
Marzena Śniegowska, Paola Marziani, Bożena Czerny, Swayamtrupta Panda, Mary Loli Martínez-Aldama, Ascensión del Olmo, Mauro D'Onofrio
DDraft version September 30, 2020
Typeset using L A TEX twocolumn style in AASTeX63
High metal content of highly accreting quasars
Marzena ´Sniegowska,
1, 2
Paola Marziani, Bo˙zena Czerny, Swayamtrupta Panda,
2, 1
Mary Loli Mart´ınez-Aldama, Ascensi´on del Olmo, and Mauro D’Onofrio Nicolaus Copernicus Astronomical Center, Polish Academy of Sciences, Bartycka 18, 00-716 Warsaw, Poland Center for Theoretical Physics, Polish Academy of Sciences, Al. Lotnik´ow 32/46, 02-668 Warsaw, Poland Istituto Nazionale di Astrofisica (INAF), Osservatorio Astronomico di Padova, 35122 Padova, Italy Instituto de Astrof´ısica de Andaluc´ıa (IAA- CSIC), Glorieta de Astronom´ıa, E-18080 Granada, Spain Dipartmento di Fisica & Astronomia, Universit`a di Padova, Padova, Italy (Received; Revised; Accepted)
Submitted toABSTRACTWe present an analysis of UV spectra of 13 quasars believed to belong to extreme Population A(xA) quasars, aimed at the estimation of the chemical abundances of the broad line emitting gas.Metallicity estimates for the broad line emitting gas of quasars are subject to a number of caveats,although present data suggest the possibility of an increase along the quasar main sequence along withprominence of optical Fe ii emission. Extreme Population A sources with the strongest Fe ii emissionoffer several advantages with respect to the quasar general population, as their optical and UV emissionlines can be interpreted as the sum of a low-ionization component roughly at quasar rest frame (fromvirialized gas), plus a blueshifted excess (a disk wind), in different physical conditions. Specifically,in terms of ionization parameter, cloud density, metallicity and column density. Capitalizing on theseresults, we analyze the component at rest frame and the blueshifted one, exploiting the dependenceof several intensity line ratios on metallicity Z . We find that the validity of intensity line ratios asmetallicity indicators depends on the physical conditions. We apply the measured diagnostic ratiosto estimate the physical properties of sources such as density, ionization, and metallicity of the gas.Our results confirm that the two regions (the low-ionization component and the blue-shifted excess)of different dynamical conditions also show different physical conditions and suggest metallicity valuesthat are high, and probably the highest along the quasar main sequence, with Z (cid:38) Z (cid:12) . We foundsome evidence of an overabundance of Aluminium with respect to Carbon, possibly due to selectiveenrichment of the broad line emitting gas by supernova ejecta. Keywords: quasars: emission lines — quasars: supermassive black holes — Line: profiles — quasars:NLSy1 — quasars: super Eddington INTRODUCTIONThanks to large public databases as the Sloan DigitalSky Survey (SDSS) catalogs, we have unrestricted accessto a large wealth of astronomical data (for example, sev-eral editions of quasar catalogues, Schneider et al. 2010;Pˆaris et al. 2017, and of value-added measurements byShen et al. 2011). SDSS spectra of high redshift quasars
Corresponding author: Marzena ´[email protected] ( z (cid:38)
2) cover the rest frame UV spectral range. It isknown since the 1970s that measurements of UV emis-sion lines can be used to explore the physical and chem-ical properties of active galactic nuclei (AGN). Land-mark papers provided the basic understanding of lineformation processes due to photoionization (e.g., Wills& Netzer 1979; Davidson & Netzer 1979; Baldwin et al.2003).The chemical composition of the line emitting gas isan especially intriguing problem from the point of viewof the evolution of cosmic structures, but also from thetechnical side. Nagao et al. (2006b) investigated BLR a r X i v : . [ a s t r o - ph . H E ] S e p ´Sniegowska et al. metallicities using various emission-line flux ratios andclaimed that the typical metallicity of the gas in that re-gion is at least super-solar, with typical Z ∼ Z (cid:12) . More-over, studies of metallicity-redshift dependence (Nagaoet al. 2006b; Juarez et al. 2009) show a lack of metallic-ity evolution up to z ≈
5. Similar results are obtainedfor (Nagao et al. 2006a). The highest-redshift quasars( z (cid:38) Perhaps surprisingly, these sources aresuspected to have high metal content in their line emit-ting gas, due to the consistent values of several diag-nostic ratios measured in quasars with similar spectralproperties at low and high z (Mart´ınez-Aldama et al.2018), and indicating highly super-solar metal content.Several techniques are applied to estimate the chemi-cal composition in Galactic nebulae (see e.g., Feibelman& Aller 1987 for planetary nebulae). Classical tech-niques used for H ii and other nebulæ (including theNarrow Line Regions, NLRs) are unfortunately not ap-plicable to the broad line regions of quasars. Permittedand inter-combination lines are too broad to resolve finestructure components of doublets; line profiles are com-posites and may originate in regions that are spatiallyunresolved, and unresolved or only partially resolved inradial velocity as well.However, quasar emission line profiles still offer im-portant clues in the radial velocity domain. The shapeof the profile is strongly dependent on the ionization po-tential of the ionic species from which the line is emitted:it is expedient to subdivide the broad lines in low- andhigh ionization lines (LILs and HILs). The LIL groupin the spectral range under analysis (1200 ˚A – 2000˚A) includes the following lines: Si ii λ ii λ ii λ iii λ iii] λ Ciii] λ Niv] λ iv] λ Civ λ iv λ Oiii] λ ii λ iii , Si iii] , and Ciii] lines sometimes referred to as “intermediate ionizationlines:” even if they are mainly produced within the fullyionized region of the emitting gas clouds (Negrete et al.2012), the ionization potential of their ionic species iscloser to the ones of the LILs, and typically (cid:46)
20 eV. The effect is most likely due to a bias: for a flux limited sample,the highest radiators at a given black hole mass are the ones thatremain detectable at highest z (Sulentic et al. 2014). The two groups of lines (HILs and LILs) do notonly show different kinematic properties (Sulentic et al.1995), but their emission is also likely to occur in fun-damentally different physical conditions (Marziani et al.2010). The HILs are characterized also by the evi-dence of strong blueshifted emission, very evident in
Civ (e.g., Sulentic et al. 2007; Richards et al. 2011;Coatman et al. 2016). Therefore, a careful line compar-ison/decomposition is necessary, lest inferences may beassociated with a non-existent region with inexplicableproperties.The interpretation of two line components involves avirialized region, of relatively low ionization (hereafterreferred to the virialized, low-ionization BLR associatedwith a symmetric broad component, BC), possibly in-cluding emission from the accretion disk, and a regionof higher ionization, associated with a disk wind or aclumpy outflow, a scenario first proposed by Collin-Souffrin et al. (1988, and further developed by Elvis2000), and observationally supported by reverberationmapping (e.g., Peterson & Wandel 1999) and the ap-parent lack of correlation between HILs and LILs in lu-minous quasars (e.g., Mej´ıa-Restrepo et al. 2016; Su-lentic et al. 2017). Even if all lines were emitted by awind (Murray et al. 1995; Murray & Chiang 1997; Proga2007a), the conditions at the base of the textcolorwindmay strongly differ from the ones downstream in theoutflow.While each UV metal line contains information re-lated to composition (Hamann & Ferland 1992), not allof the lines listed above can be used in practice. For in-stance, the N v and Si ii λ α ; other lines such as Si ii λ Niv] λ α , only the strongest broad features will be consideredas potential metallicity estimators in this work (Sect.3). The ratio (Si iv +O iv] )/ Civ has been widely used inpast studies (Hamann & Ferland 1999, and referencestherein); this ratio is relatively easy to measure andseems to be the most stable ratio against distributionof gas densities and ionization parameter in the BLR(Nagao et al. 2006b). The ratios involving NV λ v / Civ , are apparently more sensitive to ioniza-tion parameter and sensitive to the nitrogen abundance(e.g Dietrich et al. 2003; Wang et al. 2012a). We willrediscuss the use of these ratios in the context of the xAquasar spectral properties (Sect. 5.7). etallicity in highly accreting quasars ii emission is often unde-tectable above noise (e.g., Hamann et al. 2002; Punslyet al. 2018). At the other extreme, where Fe ii is mostprominent, estimates suggest Z (cid:38) Z (cid:12) (Panda et al.2018, 2019). Baldwin et al. (2003) derived Z ≈ Z (cid:12) ,although in the particular case of a “nitrogen-loud”quasars. Apart from the extremes, it is not obviouswhether there is a continuous systematic trend alongthe sequence. Previous estimates consistently suggestsuper-solar metallicity up to Z (cid:46) Z (cid:12) (Warner et al.2004). Other landmark studies consistently found super-solar metallicity: Hamann & Ferland (1992) derived Z up to (cid:46) Z (cid:12) ; Nagao et al. (2006b) found typical val-ues Z ≈ Z (cid:12) , with Z ∼ Z (cid:12) for the most luminousquasars from the (Si iv +O iv] )/ Civ ratio. Sulentic et al.(2014) inferred a large dispersion with the largest valuein excess of 10 Z (cid:12) . Similar results were reached by Shinet al. (2013) whose Si iv +O iv] / Civ ratio measurementssuggested Z (cid:38) Z (cid:12) .Most interesting along the quasar main sequence arethe high accretors. They are selected according to em-pirical criteria (e.g., Wang et al. 2013; Marziani & Sulen-tic 2014; Wang et al. 2014; Du et al. 2016a), and definedby having R FeII >
1, that is with the Fe ii λ H β (as defined by Boroson & Green1992) flux exceeding the flux of H β . In the optical dia-gram of the quasar main sequence (Sulentic et al. 2000b;Shen & Ho 2014) they are at the extreme tip in termsof Fe ii prominence, and identified as extreme Popula-tion A (hereafter xA), following Sulentic et al. (2002).Depending on redshift, we look for high accretors usingdifferent criteria. In case of z (cid:38)
1, it is expedient to usea criterion based on two UV line intensity ratios: • Al iii /Si iii] > • Ciii] /Si iii] < z theyare characterized by relatively low black hole masses fortheir luminosities and high Eddington ratios (Mathur2000; Sulentic et al. 2000a). There is evidence that xAsources tend to have high-metallicity (Shemmer et al.2004; Mart´ınez-Aldama et al. 2018). Similar proper-ties have been identified as characteristic of narrow-line Seyfert 1 galaxies (NLSy1s) with strong Fe ii emission.NLS1s also have unusually high metallicities for their lu-minosities. Shemmer & Netzer (2002) have shown thatNLSy1s deviate significantly from the nominal relation-ship between metallicity and luminosity in AGN. As sev-eral studies distinguish between NLSy1s and “broader-lined” AGN, we remark here that all Fe ii strong NLSy1smeeting the selection criterion R FeII > The aim of this work is to investigate the metallicity-sensitive diagnostic ratios of the UV spectral range forextreme Population A quasars i.e., for highly accretingquasars. Section 2 defines the selection of our sam-ple, and provides some basic information on the sam-ple quasars. In Sect. 3 we define the diagnostic ratios,and describe the basic observational results. In Section4 we compare measured diagnostic ratios and we com-pare them with the ones obtained from photoionizationsimulations. In Sect. 5 we discuss our results in termsof method caveats, metal enrichment, accretion param-eters and their implications on the nature of xA sources.We show the UV spectra in Appendix A along with themulticomponent fit analysis of the emission blends, andin the Appendix B we show the trend of Z -sensitive ra-tios as a function of ionization parameter, density, andmetallicity. SAMPLE2.1.
Sample definition
Qualitatively, extreme Pop. A objects show promi-nent Al iii and weak or absent
Ciii] emission lines. Ingeneral, they show low emission line equivalent widths( ≈ of them meet the W ( Civ ) (cid:46)
10 ˚A, and qualifyas weak-lined quasars following Diamond-Stanic et al.2009), and a spectrum that is easily recognizable evenby a visual inspection, also because of the “trapezoidal”shape of the Civ profile and the intensity of the λ Civ (Martinez-Aldamaet al. 2018).xA sources were selected according to the criteriagiven in Sect. 1, using line measurements automatically NLSy1s are identified by the line width of the H β broad compo-nent being FWHM( H β BC ) ≤ − (Osterbrock & Pogge1985), Pop. A sources are identified FWHM( H β BC ) ≤ − (Sulentic et al. 2000a). Imposing a fixed limit on lineFWHM, although very convenient observationally, has no directphysical meaning, and its interpretation might be sample depen-dent. See Marziani et al. (2018) for a discussion of the issue. Weak-lined quasars are mostly xA sources, judging from theirlocation along the MS (Marziani et al. 2016a), and that the limitat W ≈
10 ˚A separates the low- W side of a continuous distribu-tion of the xA Civ equivalent width peaked right at around 10 ˚A(Mart´ınez-Aldama et al. 2018). ´Sniegowska et al. obtained by the splot task with a cursor script withinthe IRAF data reduction package. We focus on the spec-tral range from ≈ α + N v blend is usually too heavilycompromised by absorptions which make it impossibleto reconstruct the emission components especially forLy α . We will make some consideration on the meanstrength of the N v with respect to Civ and H e ii λ v as a diagnostics.We selected SDSS DR12 spectra in the redshift range2.15 < z < r <
19) to ensuremoderate-to-high S/N in the continua (in all cases S/N (cid:38) (cid:38) δ <
10. The redshift rangewas chosen to allow for the possibility of H β coveragein the H band by eventual near-IR spectroscopic obser-vations. The DR12 sample selected with these criteriais ≈
500 sources strong. xA sources were selected outof this sample with an automated procedure, inspectedto avoid broad absorption lines, and further vetted forobtained a small pilot sample of ∼
10 sources. A largersample of xA sources will be considered in a subsequentwork (Garnica et al., in preparation). The final selec-tion includes 13 sources. With the adopted selectioncriteria in flux and redshift, we expect a small disper-sion in the accretion parameters (especially luminosity;Sect. 5.2). Indeed, the selected sources are rather ho-mogeneous in terms of spectral appearance, with a fewsources included in our sample that however show bor-derline criteria. They will be considered is Sect. 4.1.1 interms of their individual U , n H .2.2. Sample properties
Table 1 provides basic information for the 13 sourcesof our sample: SDSS name, redshift from the SDSS, thedifference between our redshift estimation using Al iii (described in 3.1) and the SDSS redshift δz = z − z SDSS ,the g -band magnitude provided by Adelman-McCarthyet al. (2008a), the g − r color index, the rest-frame-specific continuum flux at 1700 ˚A and 1350 ˚A mea-sured on the rest frame, the S/N at 1450 ˚A. All othersources were covered by the FIRST (Becker et al. 1995),but undetected. Considering that the typical rms scat-ter of FIRST radio maps is ≈ g band, we have upper limits (cid:46) provided by Sulentic et al. (2006, their Eq. B.5), andΛCDM cosmology (Ω Λ = 0 . , Ω M = 0 . , H = 70 km s − Mpc − ). The bolometric luminosity is around ∼ erg s − , assuming a bolometric correction B.C. = 3.5(Richards et al. 2006). The sample rms is just ≈ . ≈ L/L
Edd , consideringthe large uncertainty and serious biases associated withthe estimation of M BH from UV high-ionization lines.Accretion parameters will be discussed in Sect. 5.2. METHODS3.1.
Redshift determination
The estimate of the quasar systemic redshift in theUV is not trivial, as there are no low-ionization narrowlines available in the spectral range (Vanden Berk et al.2001). In practice, one can resort to the broad LIL. Ne-grete et al. (2014) and Mart´ınez-Aldama et al. (2018)consider the Si ii λ i λ iii doublet which is found, in almostall cases, to have a consistent redshift. To determine theAl iii shift those authors used multicomponent fits withall the lines in the region of the blend λ iii is clearly visible in the spectra of oursample, since in high accretors emission of Al iii is strongwith respect to the other lines in the blend at λ splot task of the Al iii doubletand/or of the Si iii] line, depending on which feature issharper. The obtained values are usually ≥ z SDSS (Ta-ble 1). This is not a surprise as z SDSS is based on linesthat are mainly blueshifted in xA sources, and hence isa systematic underestimation of the unbiased redshift.3.2.
Diagnostic ratios sensitive to U , density, Z Line ratios are sensitive to different parameters. Inthe UV range, three groups of diagnostic ratios are de-fined in the literature (e.g. Negrete et al. 2012; Martinez-Aldama et al. 2018). • Civ /Si iv +O iv] , Civ /He ii have been widely ap-plied as metallicity indicators (e.g., Shin et al.2013). In principle, Civ /He ii and Si iv /He ii should be sensitive to C and Si abundance be-cause the He abundance can be considered con-stant. The ionization potentials of C and He + are similar. The main difference is that the He ii line is a recombination line, equivalent to H i H α , etallicity in highly accreting quasars Table 1.
Source identification and basic properties
SDSS NAME z SDSS δz g g − r f λ (1700 ˚A) f λ (1350 ˚A) S/N(1) (2) (3) (4) (5) (6) (7) (8)J010657.94-085500.1 2.355 0.006 18.18 0.095 662 951 20J082936.30+080140.6 2.189 0.008 18.366 0.302 672 939 11J084525.84+072222.3 2.269 0.017 18.204 0.331 668 989 13J084719.12+094323.4 2.295 0.004 18.940 0.234 368 511 17J085856.00+015219.4 2.160 0.002 17.916 0.255 709 1204 21J092641.41+013506.6 2.181 0.004 18.591 0.337 377 670 21J094637.83-012411.5 2.212 0.002 18.561 0.178 385 595 18J102421.32+024520.2 2.319 0.008 18.49 0.177 478 694 23J102606.67+011459.0 2.253 0.003 18.982 0.206 428 525 13J114557.84+080029.0 2.338 0.009 18.545 0.369 243 360 5J150959.16+074450.1 2.255 0.008 18.938 0.278 223 346 9J151929.45+072328.7 2.394 0.008 18.662 0.171 405 507 19J211651.48+044123.7 2.352 0.000 18.825 0.220 404 573 32 Note —Columns are as follows: (1) SDSS coordinate name; (2) SDSS redshift; (3) correction toredshift estimated in the present work ( δz = z − z SDSS ); (4) g -band magnitude from Adelman-McCarthy et al. (2008b); (5) color index g − r ; (6) continuum flux measured at 1700 ˚A in unitsof 10 − erg s − cm − ˚A − ; (7) continuum flux measured at 1350 ˚A in the same units; (8)S/N measured at continuum level at 1450 ˚A. and the regions where they are formed are not co-incident (see Fig. 4). • Ratios involving N v , N v / Civ and N v /He ii havebeen also widely used in past work, after it wasnoted that the N v line was stronger than ex-pected in a photoionization scenario (e.g., Osmer& Smith 1976). A selective enhancement of nitro-gen (Shields 1976) is expected due to secondaryproduction of N by massive and intermediate massstars, yielding [N/H] ∝ Z (Vila-Costas & Ed-munds 1993; Izotov & Thuan 1999). This processmight be especially important at the high metal-licities inferred for the quasar BLR. Therefore esti-mates based on N v may differ in a systematic wayfrom estimates based on other metal lines (e.g.,Matsuoka et al. 2011). In the present sample ofquasars, contamination by narrow and semi-broadabsorption is severe, and even if we model preciselythe high ionization lines, it might be impossible toreconstruct the unabsorbed profile of the red wingof Ly α . In addition, S/N is not sufficient to al-low for a careful measurement of Niv] λ Niii] λ v measures in ahigh- Z scenario (Sect. 5.3). • The ratios Al iii /Si iii] and Si iii] / Ciii] are sensitiveto density, as the ratios involve intercombinationlines with a well defined critical density ( n c ∼ cm − for Ciii] (Hamann et al. 2002) and n c ∼ cm − for Al iii and Si iii] (Negrete et al. 2010)). • Si iii] /Si iv , Si ii λ iii] , and Si ii λ iv are sensitive to the ionization parameters and in-sensitive to Z , as they are different ionic speciesof the same element.Other intensity ratios entail a dependence on metal-licity Z , but also on ionization parameter U and density n H . 3.3. Line interpretation and diagnostic ratios
The comparison between LILs and HILs has providedinsightful information over a broad range of redshift andluminosity (Corbin & Boroson 1996; Marziani et al.1996, 2010; Sulentic et al. 2017; Bisogni et al. 2017; Shen2016; Vietri et al. 2018). A LIL-BLR appears to remainbasically virialized (Marziani et al. 2009; Sulentic et al.2017), as the H β profile remains (almost) symmetricand unshifted with respect to rest frame even if Civ blueshifts can reach several thousands of km s − . InPopulation A, the lines have been decomposed into twocomponents: • The broad component (BC), also known as theintermediate component, the core component orthe central broad component following various au-thors (e.g., Brotherton et al. 1994; Popovi´c et al.2002; Kovaˇcevi´c-Dojˇcinovi´c & Popovi´c 2015; Ad-hikari et al. 2016). The BC is modeled by a sym-metric and unshifted profile (Lorentzian for Pop. ´Sniegowska et al.
A; V´eron-Cetty et al. 2001; Sulentic et al. 2002;Zhou et al. 2006), and is believed to be associatedwith a virialized BLR subsystem. • The blue shifted component (BLUE). A strongblue excess in Pop. A
Civ profiles is obvious, asin some
Civ profiles – like the one of the xA pro-totype I Zw1 or high luminosity quasars – BLUEdominates the total emission line flux (Marzianiet al. 1996; Leighly & Moore 2004; Sulentic et al.2017). For BLUE, there is no evidence of a reg-ular profile, and the fit attempts to empiricallyreproduce the observed excess emission. BLUEis detected in a LIL such as H β at a very lowlevel, and is not strongly affecting FWHM mea-surements (Negrete et al. 2018).3.3.1. Broad component
Diagnostic ratios are not equally well measurable forthe BC and the BLUE. For the BC, the following con-straints and caveats apply:
Civ /, Si iv /, Al iii / over He ii — He ii is weak but mea-surable in most of the objects. Ratios such as Civ / H e ii λ iv / H e ii λ iii / H e ii λ U -dependent) offer Z indicators, although, as mentioned,we have to consider that the metal line formation zonesare not always coincident with the one of H e ii λ Z estimates presented in this paper. Si iv / Civ — There are problems in estimating the Si iv line intensity: an overestimation might be possible be-cause of difficult continuum placement (see, for example,the case of SDSSJ085856.00+015219.4 in Appendix A).The relative contribution of Si iv to the blend at λ iv] is unlikely, as this line has a critical density n c ∼ cm − (Zheng 1988, see also the isophotal con-tour of Si iv /O iv] in Appendix B). Our measurementsare nonetheless compared to Si iv + total O iv CLOUDYprediction. Al iii /Si iii] — This ratio is sensitive to density in the low-ionization BLR domain (Negrete et al. 2012). ValuesAl iii /Si iii] > cm − , the critical density of Si iii] . We will not usethis parameter as a metallicity estimator, although, inprinciple, for fixed physical conditions (setting n H and U ) the Al iii /Si iii] and Si iii] / Ciii] ratios may becomedependent mainly on electron temperature and so onmetallicity (Sect. 3.5). The ratio of the total emission in the 1900 blend Al iii +Si iii] + Ciii] over
Civ has beenused as a metallicity estimator (Sulentic et al. 2014).Considering the uncertain contribution of Fe iii emissionand especially of the Fe iii λ λ Civ /Al iii — Biases might be associated with the estimateof the
Civ λ BC especially when BLUE is so promi-nent that Civ λ BC contributes to a minority frac-tion. 3.3.2. BLUE component
Civ / H e ii λ — The H e ii λ Civ . The ratio
Civ / H e ii λ ii emis-sion. This ratio is in principle sensitive to metallicity.However, the increase is not monotonic at high U (Fig.2). The resulting effect is that the Civ / H e ii λ Z unconstrainedbetween 0.1 and 100 solar. Civ /(O iv] + Si iv ) — The blueshifted excess at λ iv + Si iv emission. A significant con-tribution can be associated with O iv] and several tran-sitions of O iv that are computed by CLOUDY (seee.g., Keenan et al. 2002) are especially relevant at high U values and moderately low n H ( ∼ cm − ). Theblue side of the line is relatively straightforward to mea-sure for computing Civ / λ λ Civ is possible if,assuming log U (cid:38)
0, log n H (cid:38) − ], the metallicityvalue is very high 20 (cid:46) Z (cid:46) Z (cid:12) , (Sect. 3.6). (O iv] +Si iv )/ H e ii λ — By the same token, the H e ii λ iv] +Si iv )/ H e ii λ Analysis via multicomponent fits
We analyze 13 objects using the specfit task fromIRAF (Kriss 1994). The use of the χ minimiza-tion is aimed to provide a heuristic separation be-tween the broad component (BC) and the blue com-ponent (BLUE) of the emission lines. After red-shift correction following the method described in Sect.3.1, for each source of our sample we perform adetailed modeling using various components as de-scribed below, including computation of asymmetricerrors (Sect. 3.4.1). As mentioned in Sect. 3.2,in our analysis we consider five diagnostic ratios for etallicity in highly accreting quasars Civ / λ Civ / H e ii λ iii / H e ii λ λ H e ii λ λ iii , and three for the BLUE: Civ / λ Civ / H e ii λ λ H e ii λ Civ / H e ii λ H e ii λ χ i.e., with min-imized difference between the observed and the modelspectrum. Following the data analysis by Negrete et al.(2012), we use the following components: The continuum — was modeled as a power-law, and weuse the line-free windows around 1300 and 1700 ˚A (twosmall ranges where there are no strong emission lines) toscale it. If needed, we divide the continuum into threeparts (corresponding to the three regions mentioned be-low). Assumed continua are shown in the Figures ofAppendix A. Fe ii emission — usually does not contribute significantlyin the studied spectral ranges. We consider the Fe ii template which is based on CLOUDY simulations ofBr¨uhweiler & Verner (2008) when necessary. In practicethe contamination by the blended Fe ii emission yieldinga pseudo-continuum is negligible. Some Fe ii emissionlines were detectable in only a few objects and around ≈ Fe iii emission — affects more the 1900 ˚A region andseems to be strong when AlIII λ Region 1300 - 1450˚A — is dominated by the Si iv + O iv] high ionization blend with strong blueshifted compo-nent. The fainter lines as Si ii λ i λ ii λ Civ and H e ii λ Region 1450 - 1700 ˚A — is dominated by Civ emissionline which we model as a fixed in the rest-frame wave-length Lorentzian profile representing the BC and twoblueshifted asymmetric Gaussian profiles vary freely.The same model is used for H e ii λ Region 1700 - 2200 ˚A — is dominated by Al iii , Si iii] andFe iii intermediate - ionization lines. We model Al iii and Si iii] using Lorentzian profiles, following Negreteet al. (2012). Ciii] emission is also included in the fit,although the dominant contribution around λ iii (Mart´ınez-Aldama et al. 2018, andreferences therein). We use the template of Vestergaard& Wilkes (2001) to model Fe iii emission. No BLUE isascribed to these intermediate ionization lines. Absorption lines — are modeled by Gaussians, and in-cluded whenever necessary to obtain a good fit.The fits to the observed spectral ranges are shown inthe Figures of Appendix A.3.4.1. Error estimation on line fluxes
The choice of the continuum placement is the mainsource of uncertainty in the measurement of the emissionline intensities. The fits in Appendix A show that, in themajority of cases, the FWHM of the Al iii and Si iii] lines(assumed equal) satisfy the condition FWHM(Al iii ) ∼ FWHM(
Civ BC ∼ FWHM(Si iv BC ). Figure 1 shows thebest fit, maximum and minimum placement of the con-tinuum, which we choose empirically. With this ap-proach the continua of Figure 1 should provide the con-tinuum uncertainty at a ± σ confidence.The continuum placement strongly affects the mea-surement of an extended feature such as the Fe iii blendsand the H e ii λ σ ( X ) = ∆ x + + ∆ x − + ∆ x + + ∆ x −
18 (1) ´Sniegowska et al. where ∆ x + and ∆ x − are differences between measure-ment with maximum and best continuum and with bestand minimum continuum respectively. To analyze er-ror of diagnostic ratios we propagate uncertainties usingstandard formulas of error propagation. N o r m a li z ed F λ Rest−frame wavelength (Å)J211651.48+044123.7
Figure 1.
Continuum estimation for J211651.48+044123.7from our sample. Range 1300 - 1450 ˚A is shown in blue, 1450- 1700 ˚A in red and 1750 - 2200 ˚A in green. The continuumlines in each range represent from the top: the maximum,the best and the minimum continuum placement.
Photoionization modeling
To interpret our fitting results we compare the lineintensity ration for BC and BLUE with the ones pre-dicted by CLOUDY simulations (Ferland et al. 2013). An array of simulations is used as reference for compar-ison with the observed line intensity ratios. It was com-puted under the assumption that (1) column density is N c = 10 cm − ; (2) the continuum is represented by themodel continuum of Mathews & Ferland (1987) which isbelieved to be appropriate for Population A quasars, and(3) microturbulence is negligible. The simulation arrayscover the hydrogen density range 7.00 ≤ log( n H ) ≤ − . ≤ log( U ) ≤ Z (cid:12) . Extremely high metallicity Z (cid:38) Z (cid:12) is considered physically unrealistic ( Z ≈ Z (cid:12) impliesthat more than half of the gas mass is made up by met-als!), unless the enrichment is provided in situ within The arrays were computed over several years with CLOUDY13.05, in large part before CLOUDY 17 became available. the disk (Cantiello et al. 2020). In several cases, sim-ulations suffered convergence problems if Z (cid:38) Z (cid:12) .The behavior of diagnostic line ratios as a function of U and n H for selected values of Z is shown in Fig. 18 ofAppendix B. 3.5.1. Basic Interpretation
The line emissivity (cid:15) coll (ergs cm − s − ) of a collision-ally excited line emitted from an element X in its i − thionization stage has a strong temperature dependence.In the high density limit (cid:15) X i , coll = n X i , l βA X i , ul hν g X i , l g X i , u exp (cid:18) − hν kT e (cid:19) (2) ∝ n X i , l exp (cid:18) − hν kT e (cid:19) the line is said to be “thermalized,” as its strength de-pends only on the atomic level population and not onthe transition strength (Hamann & Ferland 1999). β isthe photon escape probability and A X i , ul is the sponta-neous decay coefficient. At low densities we have, (cid:15) X i , coll = n X i , l n e q X i , lu hν ∝ n i T − / exp (cid:18) − hν kT e (cid:19) (3)The recombination lines considered in our analysis are H β and He ii λ α on electron temperature, Osterbrock & Ferland2006) becomes: (cid:15) Y j , rec = n Y j n e α hν ∝ n j T − (4)and n Y j is the number density of the parent ion.Under these simplifying, illustrative assumption wecan write: (cid:15) X i , coll (cid:15) Y j , rec ∝ (cid:18) n X i n Y j (cid:19) T e exp (cid:18) − hν kT e (cid:19) (5)for the low-density case, and (cid:15) X i , coll (cid:15) Y j , rec ∝ n X i n j T e exp (cid:18) − hν kT e (cid:19) (6)for the high density case.Similarly, for the ratio of two collisionally excited linesat frequencies ν and ν , (cid:15) X i , coll (cid:15) Y j , coll ∝ (cid:18) n X i n Y j (cid:19) κ exp (cid:18) − h ( ν − ν ) kT e (cid:19) (7)where κ = 1 , etallicity in highly accreting quasars T e . Inother words, electron temperature is the main parame-ter connected to metallicity. This is especially true forfixed physical condition ( U , n H , N c = 10 , SED given).This is most likely the case of xA sources: the spectralsimilarity implies that the scatter in physical propertiesodest. We further investigate this issue in Sect. 4.3.3.6. Explorative analysis of photoionization trends atfixed ionization parameter and density
One of the main results of previous investigations isthe systematic differences in ionization between BLUEand BC (Marziani et al. 2010; Negrete et al. 2012; Su-lentic et al. 2017). Previous inferences suggest verylow ionization ( U ∼ − . ), also because of the rela-tively low Civ / H β ratio for the BC emitting part of theBLR, and high density. A robust lower limit to density n H ∼ . cm − has been obtained from the analysis ofthe CaII triplet emission (Matsuoka et al. 2007; Pandaet al. 2020a). Less constrained are the physical condi-tions for BLUE emission. Apart from Civ / H β (cid:29) α / H β and Civ / Ciii] also (cid:29)
1, little constraints ex-ist on density and column density. This result hardlycomes as a surprise considering the difference in dynam-ical status associated with the two components. Whileit is expected that the BC is emitted in a region of highcolumn density log N c (cid:38)
23 [cm − ], not last because radi-ation forces are proportional to the inverse of N c (Netzer& Marziani 2010). Following Netzer & Marziani (2010)we wrote the equation of motions for a gas cloud underthe combined effect of gravitation and radiation forces,and showed that the acceleration term due to radiationis inversely proportional to N c (see also Ferland et al.2009), this high N c region is expected to be relativelystable (at rest frame, with no sign of systematic, largeshifts in Population A) and presumably devoid of low-density gas (considering the weakness of Ciii] , Negreteet al. 2012). The same cannot be assumed for BLUE.BLUE is associated with a high radial velocity outflow,probably with the outflowing streams creating BAL fea-tures when intercepted by the line-of-sight (e.g., Elvis2000). Here we consider log U = − .
5, log n H =12 (-2.5,12),and log U = 0 , log n H =9 (0,9) as representative of thelow and high- ionization emitting gas. Fig. 2 illus-trates the behavior of the Civ / H β , H e ii λ H β and Civ / H e ii λ Civ intensity with re-spect to H β has a steep drop around Z (cid:38) Z (cid:12) , aftera steady increase for sub-solar Z . The H e ii λ H β ratio decreases steadily, with a steepening at round so-lar value. Physically, this behavior is due to the highvalue of the ionization parameter (assumed constant),while the electron temperature decreases with metallic-ity, implying a much lower collisional excitation rate for Civ production. The dominant effect for the H e ii λ Civ and H e ii λ Civ / H e ii λ Civ / H β is followed by a saturation to a maximum Civ / H β . The H e ii λ H β ratio is constant up to so-lar, and steadily decreases above solar, where the ioniza-tion competition with triply ionized carbon sets on. Theresult is a smooth, steady increase in the Civ / H e ii λ iv +O iv] / Civ and Si iv +O iv] / H e ii λ Z (cid:12) . Only around Z ∼ Z (cid:12) val-ues Civ /Si iv +O iv] (cid:46) Z (cid:38) Z (cid:12) , with the unpleasant consequence that a ratio Civ /Si iv +O iv] ≈ . Z (cid:12) as well as1000 Z (cid:12) . The ratios usable for the BC also show regu-lar behavior. The Civ /Al iii ratio remains almost con-stant up Z ∼ . Z (cid:12) , and the starts a regular decreasewith increasing Z , due to the decrease of T e with Z ( Civ is affected more strongly than Al iii ). Interest-ingly, Al iii / H e ii λ H e ii λ Z . Especially of interest is however the behavior of ratioAl iii / H e ii λ iv +O iv] )/ Civ (cid:38) Z ≈ Z (cid:12) ) compli-cates the interpretation of the observed emission lineratios.The ionization structure within the slab remains selfsimilar over a wide metallicity range, with the same sys-tematic differences between the high and low-ionization0 ´Sniegowska et al. Figure 2.
Computed intensity ratios involving
Civ and H e ii λ U and n H fixed: (log U , log n H ) = (-1,9) (top) and (log U , log n H ) = (-2.5,12) (bottom). Columns from left to right show Civ / H β , H e ii λ H β , Civ / H e ii λ case (Fig. 4), consistent with the assumption of a con-stant ionization parameter. As expected, the electrontemperature decreases with metallicity, and the transi-tion between the fully and partially ionized zone (FIZand PIZ) occurs at smaller depth. In addition, close tothe illuminated side of the cloud the electron tempera-ture remains almost constant; the gas starts becomingcolder before the transition from FIZ to PIZ. The depthat which T e starts decreasing is well-defined, and itsvalue becomes lower with increasing Z (Fig. 4). The ef-fect is present for both the low- and high- ionization case,although it is more pronounced for the high-ionization.Fig. 5 shows that an increase in metallicity is affectingthe T e in the line emitting cloud. Fig. 5 reports the be-havior of T e at the illuminated face of the cloud ( τ ∼ τ (corresponding to N c = 10 cm − ,the side facing the observer) for the high-ionization andlow-ionization case. The T e monotonically decreases asa function of metallicity. The difference between thetwo cloud faces is almost constant for the low ionizationcase, with δ log T e ≈ . δ log T e ≈ .
75 dex at thehighet Z value considered, 10 Z (cid:12) . RESULTS4.1.
Immediate Results
The observational results of our analysis involve themeasurements of the intensity of the line BC and BLUEcomponent separately. The rest-frame spectra with thecontinuum placements, and the fits to the blends of thespectra are shown in Appendix A. Table 2 reports themeasurement for the λ − ) and equivalent width and flux of Al iii (the sumof the doublet lines, in units of ˚A and 10 − erg s − cm − , respectively), FWHM and flux of Ciii] , and fluxof Si iii] (its FWHM is assumed equal to the one of thesingle Al iii lines.) with Similarly, Table 3 reports theparameter of the
Civ blend: equivalent width, FWHMand flux of the
Civ
BC, the flux of the
Civ blueshiftedcomponent, as well as the fluxes of the BC and BLUE of H e ii λ (cid:38) − should be considered as highlyuncertain. There is the concrete possibility of an ad- etallicity in highly accreting quasars Figure 3.
Behavior of the intensity ratios employed in this work (with the exception of
Civ / H e ii λ U and n H fixed: (log U , log n H ) = (-1,9) and (log U , log n H ) =(-2.5,12). Top panels, from left to right: Civ /Si iv +O iv] ,(Si iv +O iv] )/ H e ii λ Civ /Al iii . Bottom panes, from left to right:Al iii / H e ii λ Civ /(Si iv +O iv] ), (Si iv +O iv] )/ H e ii λ ditional broadening ( ∼
10 % of the observed FWHM)associated with non-virial motions for the Al iii line (delOlmo et al., in preparation). The fluxes of the BC andof BLUE of Si iv and O iv] are reported in Table 4. In-tensity ratios with uncertainties are reported in Table 5.The last row lists the median values of the ratios withtheir semi-interquartile ranges (SIQR).4.1.1. Identification of xA sources and of “intruders”
Figure 6 shows that the majority of sources meetboth UV selection criteria, and should be considered xAquasars. The median value of the Al iii /Si iii] (last row ofTable 5) implies that the Al iii is strong relative to Si iii] .Also Si iii] is stronger than
Ciii] . Both selection crite-ria are satisfied by the median ratios. Only one source(SDSS J084525.84+072222.3) shows
Ciii] /Si iii] signifi-cantly larger than 1. This quasar is however confirmedas an xA by the very large Al iii /Si iii] , by the blueshiftof
Civ , and by the prominent λ Civ emission. The lines in the spectrum of SDSSJ084525.84+072222.3 are broad, and any
Ciii] emis- sion is heavily blended with Fe iii emission. The
Ciii] value should be considered an upper limit. Three out-lying/borderline data points (in orange) in Fig. 6 haveratio
Ciii] /Si iii] ∼
1, and Al iii /Si iii] consistent with theselection criteria within the uncertainties, but other cri-teria support their classification as xA. The borderlinesources will be further discussed in Section 4.3. In con-clusion, all the 13 sources of the present sample save oneshould be considered bona-fide xA sources.It is intriguing that the intensity ratios
Ciii] /Si iii] andAl iii /Si iii] are apparently anti-correlated in Figure 6,if we exclude the two outlying points. Excluding thetwo outlying data point the Spearman rank correlationcoefficient is ρ ≈ .
8, which implies a 4 σ significance fora correlation, but the correlation coefficient between thetwo ratios for the full sample is much lower. Given thesmall number of sources, a larger sample is needed toconfirm the trend.4.1.2. BC intensity ratios ´Sniegowska et al. Figure 4.
Computed ionization fraction and electron temperature (thick grey line) as a function of depth within the emittinggas slab, for physical parameters U and n H fixed: (log U , log n H ) = (-1,9) (representative of BLUE and high ionization case,left) and (log U , log n H ) = (-2.5,12) (representative of the low-ionization BLR, right), in order of increasing metallicity fromtop to bottom. etallicity in highly accreting quasars Figure 5.
Electron temperature T e as a function of metal-licity Z , for physical parameters U and n H fixed: (log U , log n H ) = (-1,9) (representative of BLUE and high ionizationcase, blue and cyan) and (log U , log n H ) = (-2.5,12) (repre-sentative of the low-ionization BLR, red and orange). Blueand red refer to the first zone of the CLOUDY computationi.e., to the illuminated surface of the clouds; cyan and or-ange, to the side of the cloud farther from the continuumsources i.e., facing the observer. Fig. 7 shows the distributon of diagnostic in-tensity ratios
Civ / H e ii λ iv / H e ii λ iii / H e ii λ µ ( Civ / H e ii λ ≈ . µ (Al iii / H e ii λ ≈ µ (Si iv / H e ii λ ≈ iv / H e ii λ Civ / H e ii λ µ ( Civ /Si iv ) ≈ Civ /Al iii ratio is also out of scale:the minimum
Civ /Al iii predicted by the simulations atlow ionization is ≈ .
6, reached at extremely high Z ( ∼ Z (cid:12) ).However, the distribution of the data points is rel-atively well-behaved, with individual ratios showingsmall scatter around their median values. In the his-togram, we see a tail made by a 4-5 objects sug-gesting systematically higher values. In particular, atleast two objects (SDSS J102606.67+011459.0 and SDSSJ085856.00+015219.4) show systematically higher ra-tios, with Civ / H e ii λ ≈
10, and Al iii / H e ii λ ≈ Figure 6.
Relation between intensity ratiosAlIII λ λ λ λ Both of them show extreme
Civ blueshifts and SDSSJ102606.67+011459.0 shows the highest Al iii /Si iii] ra-tio in the sample.Since the three ratios are, for fixed physical condi-tions, proportional to metallicity, we expect an overallconsistency in their behavior i.e., if one ratio is higherthan the median for one object, also the other inten-sity ratios should be also higher. The lower diagramsare helpful to identify sources, for which only one in-tensity ratio deviates significantly from the rest of thesample. A case in point is SDSS J082936.30+080140.6whose ratio Al iii / H e ii λ ≈ Civ / H e ii λ iv / H e ii λ iii emis-sion. The Civ and λ Civ and Si iv BCis very difficult, as it accounts for a small fraction of theline emission. The H e ii λ BLUE intensity ratios
Similar considerations apply to the blue intensity ra-tios. We see systematic trends in Figure 8 that implyconsistency of the ratios for most sources, although theuncertainties are larger, especially for
Civ / H e ii λ Civ /(Si iv + O iv] ) values are systematicallyhigher than for the BC, while the Civ / H e ii λ ´Sniegowska et al.
0 5 10 15 20 25
CIV/HeII
2 4 6 8 10 12 14 16 18
SiIV/HeII
0 2 4 6 8 10 12 14 J010657.94−085500.1 J094637.83−012411.5 J085856.00+015219.4 J102606.67+011459.0 J211651.48+044123.7 J092641.41+013506.6 J102421.32+024520.2 J082936.30+080140.6 J151929.45+072328.7 J084525.84+072222.3 J084719.12+094323.4 J114557.84+080029.0 J150959.16+074450.1
AlIII/HeII
AlIII/HeII
SiIV/HeII
CIV/HeII
Figure 7.
Distribution of diagnostic intensity ratios based on the BC (top) for
Civ / H e ii λ iv / H e ii λ iii / H e ii λ slightly higher (median BLUE 5.8 vs. median BC 4.38).The ratio (Si iv + O iv] )/ H e ii λ iv from O iv] , and by the fre-quent occurrence of absorptions affecting the blue sideof the blend. Both factors may conspire to depressBLUE. The lower diagrams of Fig. 8 are again help-ful to identify sources for which intensity ratios devi-ate significantly from the rest of the sample. SDSS J102606.67+011459.0 shows a strong enhancement of Civ / H e ii λ iv +O iv] , confirming the trendseen in its BC.4.1.4. Correlation between diagnostic ratios
Fig. 9 shows a matrix of correlation coefficientsfor all diagnostic ratios which we considered in thiswork. The 2 σ confidence level of significance for theSpearman’s rank correlation coefficient for 13 objectsis achieved for ρ ≈ .
54. The highest degree of cor- etallicity in highly accreting quasars Table 2.
Measurements in the 1900˚A blend region.
SDSS JCODE Al iii Al iii Al iii Ciii] Ciii] Si iii] W FWHM Flux FWHM Flux Flux(1) (2) (3) (4) (5) (6) (7)J010657.94-085500.1 7.9 5560 5.41 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note —Columns are as follows: (1) SDSS name, (2) and (3) report the FWHM ofAl iii and
Ciii] in km s − ; (4), (5) and (6) list the fluxes in units of 10 − ergs − cm − for Al iii , Ciii] , and Si iii] . Table 3.
Measurements in the
Civ spectral region
SDSS JCODE
Civ Civ BC Civ BC Civ
BLUE He ii BC He ii BLUE W FWHM Flux Flux Flux Flux(1) (2) (3) (4) (5) (6) (7)J010657.94-085500.1 18.6 5530 6.35 ± ± ± ± ±
670 1.83 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
650 6.92 ± ± ± ± ±
700 6.84 ± ± ± ± ±
690 4.19 ± ± ± ± ±
590 5.47 ± ± ± ± ± ± ± ± Note.
Columns are as follows: (1) SDSS name, (2) rest-frame equivalent width ofthe total
Civ emission i.e.,
Civ
BLUE+BC, in ˚A; (3) the FWHM of the
Civ linein km s − ; Cols. (4) and (5) list fluxes of the Civ
BC and BLUE line; Cols. (6)and (7) report fluxes of the BC and BLUE components for the H e ii λ − erg s − cm − . relation is found between the ratios Civ / H e ii λ Civ /Si iv (0.87), and between Civ / H e ii λ iv / H e ii λ iii / H e ii λ Civ / H e ii λ iv / H e ii λ iv and Civ are likely affected in a related way by asingle parameter. The main parameter is expected to be T e , and hence Z (Sect. 3.5.1). The Al iii (normalized tothe H e ii λ iv and Civ are basicallythe same line, the Al iii formation may not be exclusivelycollisional, as shown by the results of the
CLOUDY sim-ulations. Therefore there could be a different responseto individual U , n H and optical depth variations. Theprominence of Ciii] with respect to Si iii] decreases withSi iv / H e ii λ Civ / H e ii λ iv / H e ii λ Civ /Si iv +O iv] . Apparentlythe Ciii] /Si iii] ratio is strongly affected by an increase6 ´Sniegowska et al.
Table 4.
Measurements in the λ SDSS JCODE Si iv +O iv] BC Si iv +O iv] BC Si iv +O iv] BLUEFWHM Flux Flux(1) (2) (3) (4)J010657.94-085500.1 5070 9.23 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note.
Columns are as follows: (1) SDSS name, (2) the FWHMof the Si iv line in km s − . (3) and (4) list fluxes of the broadcomponents and the blue component line in units of 10 − ergs − cm − . Table 5.
Intensity ratios for the BC and BLUE line components
SDSS JCODE Al iii /Si iii] Ciii] /Si iii] Civ /Si iv Civ / H e ii λ iv / H e ii λ Civ /Al iii Al iii / H e ii λ Civ / H e ii λ Civ /Si iv + O iv] Si iv + O iv] / H e ii λ ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± in metallicity and more in general by ratios that are in-dicative of “extremeness” in our sample. For BLUE,the two main independent Z estimators are correlated( ρ ≈ . Analysis of Z distributions: global inferences onsample
Fixed ( U , n H ) We propagated the diagnostic intensity ratios mea-sured on the BC and BLUE components with their lowerand upper uncertainties following the relation betweenratios and Z in the Figures 2, for the fixed physicalconditions assumed in the low- and high-ionization re-gion. The results are reported in Table 6 and Table 7for the BC and for the blueshifted component, respec- tively. The last row reports the median values of theindividual sources Z estimates with the sample SIQR.The distributions are shown in Figs. 10 and 11, alongwith a graphical presentation of each source and its as-sociated uncertainties.Table 6 and Table 7 permit to quantify the systematicdifferences that are apparent in Figs. 10 and 11. Theagreement between the various estimators is in good onaverage (the medians scatter around log Z ≈ Z obtained from the various diagnostic ra-tios. Si iv and Al iii over H e ii λ Z by a factor 2 with respect to Civ / H e ii λ Civ /Si iv and Civ /Al iii maysuggest that metallicity scaling according to solar pro- etallicity in highly accreting quasars
0 1 2 3 4 5
CIV/SiIV+OIV
0 2 4 6 8 10 12 14 16
CIV/HeII
0 2 4 6 8 10 J010657.94−085500.1 J094637.83−012411.5 J085856.00+015219.4 J102606.67+011459.0 J211651.48+044123.7 J092641.41+013506.6 J102421.32+024520.2 J082936.30+080140.6 J151929.45+072328.7 J084525.84+072222.3 J084719.12+094323.4 J114557.84+080029.0 J150959.16+074450.1
SiIV+OIV/HeII
SiIV+OIV/HeII
CIV/HeII
CIV/SiIV+OIV
Figure 8.
Distribution of diagnostic intensity ratios based on the BLUE components (top) for
Civ /Si iv +O iv] (blue), Civ / H e ii λ iv +O iv] / H e ii λ ´Sniegowska et al. A l III / S i III C III i / S i III C I V / S i I V C I V / H e II S i I V / H e II C I V / A l III A l III / H e II C I V / H e II B L U E C I V / S i I V + O I V B L U E S i I V / H e II B L U E AlIII/SiIIICIIIi/SiIIICIV/SiIVCIV/HeIISiIV/HeIICIV/AlIIIAlIII/HeIICIV/HeII BLUECIV/SiIV+OIV BLUESiIV/HeII BLUE -0.36-0.33 0.17-0.074 0.069 0.870.32 0.038 0.47 0.81-0.59 0.6 0.65 0.54 0.320.74 -0.55 -0.075 0.23 0.44 -0.510.53 -0.39 -0.12 0.0022 0.14 -0.5 0.63-0.2 0.68 0.21 -0.053 -0.16 0.4 -0.33 -0.10.13 -0.46 0.036 0.2 0.14 -0.36 0.52 0.68 -0.57
Figure 9.
The correlation matrix between diagnostic ra-tios in BC and BLUE. The numbers in each square show theSpearman rank correlation coefficient. Red colors indicate apositive correlation, blue colors indicate a negative correla-tion.. portion may not be strictly correct (Sect. 5.6). In thecase of BLUE, several estimates from
Civ / H e ii λ Z and Civ / H e ii λ Z ≈ Z (cid:12) (cid:46) Z (cid:46) Z (cid:12) . There is appar-ently a systematic difference between BC and BLUE,in the sense that Z derived from the BC is systemati-cally higher than Z from blue. The difference is smallin the case of Civ / H e ii λ iv + O iv] / H e ii λ Z from BLUEare a factor of 10 lower. We have stressed earlier thatthere are often absorptions affecting the BLUE of Si iv +O iv] / H e ii λ ii λ iv BC lines make it difficult to prop-erly define the continuum underlying the λ iv +O iv] BLUE intensity estimate is more of a lower limit.Another explanation might be related to the assump-tion of a constant density and U for all sources. Whilethere are observational constraints supporting this con-dition for the BC (Panda et al. 2018, 2019, 2020b), thereare no strong clues to the BLUE properties, save a highionization degree.4.3. Z for individual sources for fixed U , n H Table 8 reports the Z estimates for the BC, BLUE,and a combination of BC and BLUE for each individualobject. The values reported are the median values of theindividual objects’ estimates from the different ratios.For BC, the three ratios of Table 6 were always used.The last column of Table 8 reports the number of ratiosused from the BLUE component. Here the Z value foreach object is computed by vetting the ratios accordingto concordance. If the discordance is not due on phys-ical ground, but rather to instrumental problems (forexample, contamination by absorption lines, non lineardependence on Z of some ratios), a proper strategy is touse estimators such as the median that eliminate discor-dant values even for small sample sizes ( n ≥ Z (cid:46) iv +O iv] / H e ii λ Civ / H e ii λ Z ; apartfrom J211651.48+044123.7, the upper uncertainty of thenegative estimates is so large that Z is actually uncon-strained. The difference between BLUE and BC is evenmore evident: the median (last row) indicates a factor ≈ Z ≈ Z (cid:12) , while the BLUE Z ≈ Z (cid:12) .The assumption that the wind and disk component havethe same Z in each object is a reasonable one, with thecaveats mentioned in Sect. 5.6. Therefore the two esti-mates, for BLUE and BC could be considered two inde-pendent estimators of Z . If the two estimates are com-bined for each individual object, 10 Z (cid:12) (cid:46) Z (cid:46) Z (cid:12) ,with a median value of Z ≈ Z (cid:12) .4.4. Estimates of Z relaxing the constraints on U and n H The Z estimates for the BC are mainly based on thethree ratios involving H e ii λ Z dependence on thephysical parameters, we assigned a score from 0 to 3and considered the domain of the parameter space U , n H , Z for which there is agreement with all the threediagnostic ratios. The left panels of Fig. 12 and of Fig.13 shows the 3D space U , n H , Z where each point inspace correspond to an element of the grid of CLOUDY the parameter space compatible with all three observedratios within the uncertainties. The case shown in Fig.12 and in Fig. 13 is the one with the median values ofthe sample objects.Similar considerations can be made if we consider the χ behavior. We compute the χ in the following form,to identify the value of the metallicity for median valuesof the diagnostic ratios and for the diagnostic ratios of etallicity in highly accreting quasars −0.5 0 0.5 1 1.5 2 2.5 log10(Z(CIV/HeII))
1 1.2 1.4 1.6 1.8 2 2.2 2.4 log10(Z(SiIV/HeII))
1 1.2 1.4 1.6 1.8 2 2.2 2.4 log10(Z(AlIII/HeII)) −0.5 0 0.5 1 1.5 2 2.5 J010657.94−085500.1 J094637.83−012411.5 J085856.00+015219.4 J102606.67+011459.0 J211651.48+044123.7 J092641.41+013506.6 J102421.32+024520.2 J082936.30+080140.6 J151929.45+072328.7 J084525.84+072222.3 J084719.12+094323.4 J114557.84+080029.0 J150959.16+074450.1 log10(Z)
Figure 10. (Upper panel) Distribution of metallicity measurements for broad component obtained from ratios
Civ / H e ii λ iv / H e ii λ iii / H e ii λ Table 6.
Metallicity (log Z ) of the BC assuming fixed U , n H SDSS JCODE
Civ / H e ii λ iv / H e ii λ iii / H e ii λ +0 . − . +0 . − . +0 . − . J085856.00+015219.4 2.19 +0 . − . +0 . − . +0 . − . J082936.30+080140.6 0.63 +0 . − . +0 . − . +0 . − . J084525.84+072222.3 1.38 +0 . − . +0 . − . +0 . − . J084719.12+094323.4 1.47 +0 . − . +0 . − . +0 . − . J092641.41+013506.6 1.25 +0 . − . +0 . − . +0 . − . J094637.83-012411.5 1.78 +0 . − . +0 . − . +0 . − . J102421.32+024520.2 1.14 +0 . − . +0 . − . +0 . − . J102606.67+011459.0 1.79 +0 . − . +0 . − . +0 . − . J114557.84+080029.0 1.10 +0 . − . +0 . − . +0 . − . J150959.16+074450.1 0.18 +0 . − . +0 . − . +0 . − . J151929.45+072328.7 1.15 +0 . − . +0 . − . +0 . − . J211651.48+044123.7 2.09 +0 . − . +0 . − . +0 . − . Median 1.37 ± ± ± Note —Columns: (1) SDSS identification, (2), (3) and (4) metallicity val-ues for
Civ / H e ii λ iv +O iv] / H e ii λ iii / H e ii λ Table 7.
Metallicity (log Z ) of BLUE assuming fixed U , n H SDSS JCODE Si iv +O iv] / H e ii λ Civ /Si iv +O iv] Civ / H e ii λ +0 . − . +0 . − . +0 . − . J082936.30+080140.6 1.11 +0 . − . +0 . − . +0 . − . J084525.84+072222.3 0.63 +0 . − . +0 . − . +0 . − . J084719.12+094323.4 0.73 +0 . − . +0 . − . +0 . − . J085856.00+015219.4 1.29 +0 . − . +0 . − . +0 . − . J092641.41+013506.6 0.94 +0 . − . +0 . − . -0.56 +1 . − . J094637.83-012411.5 0.75 +0 . − . +0 . − . -1.01 +2 . − . J102421.32+024520.2 0.75 +0 . − . +0 . − . -0.51 +2 . − . J102606.67+011459.0 1.04 +0 . − . +0 . − . +0 . − . J114557.84+080029.0 0.9 +0 . − . +0 . − . +0 . − . J150959.16+074450.1 1.09 +0 . − . +0 . − . +0 . − . J151929.45+072328.7 0.81 +0 . − . +0 . − . +0 . − . J211651.48+044123.7 -0.34 +0 . − . +0 . − . -1.47 +0 . − . Medians 1.160 ± ± ± Note —Columns: (1) SDSS identification, (2), (3) and (4) metallicity valuesfor Si iv +O iv] / H e ii λ Civ /Si iv +O iv] and Civ / H e ii λ ´Sniegowska et al. log10(Z(CIV/SiIV+OIV)) −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 log10(Z(CIV/HeII)) −1 −0.5 0 0.5 1 1.5 2 J010657.94−085500.1 J094637.83−012411.5 J085856.00+015219.4 J102606.67+011459.0 J211651.48+044123.7 J092641.41+013506.6 J102421.32+024520.2 J082936.30+080140.6 J151929.45+072328.7 J084525.84+072222.3 J084719.12+094323.4 J114557.84+080029.0 J150959.16+074450.1 log10(Z(SiIV+OIV/HeII)) −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 log10(Z) Figure 11. (Upper panel) Distribution of metallicity measurements for blue component obtained from ratios:
Civ /Si iv +O iv] (blue), Civ / H e ii λ iv +O iv] / H e ii λ Table 8.
Metallicity (log Z ) of individual quasars assumingfixed U , n H SDSS JCODE BC BLUE Combined ± ± ± ± ± ± ± ± ± ± ± ± ± ± . . . 2.09 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note —Columns: (1) SDSS identification, (2), (3) and (4) metallicitymedians for BC, BLUE and the combination of the two, with un-certainties. Column (5) yields the number of ratios used for theBLUE estimates. No uncertainty is reported for BLUE of SDSSJ211651.48+044123.7 since only one ratio was used. individual objects relaxing the assumption of fixed den- sity and ionization parameters. For each object k , andfor each component c , we can write: χ ( n H , U, Z ) = (cid:88) i w ci (cid:18) R kci − R kci , mod ( n H , U, Z ) δR kci (cid:19) (8)where the summation is done over the available diag-nostic ratios, and the χ is computed with respect tothe results of the CLOUDY simulations as a functionof U , n H , and Z (subscript ‘mod’). Weights w ci = 1were assigned to Civ / H e ii λ iv / H e ii λ iii / H e ii λ w ci = 0 or 0.5 were assigned to Civ /Al iii and and
Civ /Si iv . For BLUE, the three di-agnostic ratios were all assigned w ci = 1.The distribution of the data point for the median ± SIQR values in Fig. 12 is constrained in a relativelynarrow range of U , n H , Z , at very high density, lowionization, and high metallicity. Within the limit in U , n H , the distribution of Z is flat and thin, around Z ∼ Z (cid:12) . This implies that, for a change of the U and n H within the limits allowed by the data, the estimate of Z is stable and independent on U and n H . Table 8 reportsthe individual Z estimates and the SIQR for the sourcesin the sample (the last row is the median). etallicity in highly accreting quasars χ distributions. The spread in ionization and density isvery large, although the concentration of data points ishigher in the case of low n H (log n H ∼ − ]) andhigh ionization (log U ∼ iii / H e ii λ Z are however asstable as for the BC, even if the dispersion is large, andsuggest values in the range 10 (cid:46) Z (cid:46) Z (cid:12) .Summing up, all meaningful estimators converge to-ward high Z values, Z (cid:38) Z (cid:12) . Ratios Civ /Si iv signifi-cantly less than < iv / H e ii λ Z . Also the high Al iii / Civ requires extremelyhigh values of Z . A conclusion has to be tentative, con-sidering the possible systematic errors affecting the es-timates of the Civ and Si iv intensities: for Civ , theBC in the most extreme cases is often buried under anoverwhelming BLUE; a fit is not providing a reliableestimate of the BC (by far the fainter component) butprovides a reliable BLUE intensity; for Si iv we may over-estimate the intensity due to “cancellation” of the BLUEby absorptions. This said, the present data are consis-tent with the possibility of a selective enhancement of Aland Si, as already considered by Negrete et al. (2012).The issue will be briefly discussed in Sect. 5.At any rate, the absence of correlation between BLUEand BC parameters (Fig. 9), the difference in the diag-nostic ratios and differences in inferred Z , as well as theresults for individual sources described below justify theapproach followed in the paper to maintain a separa-tion between BLUE and BC. The meaning of possiblesystematic differences between the BC and BLUE arefurther discussed in § Individual sources
The best n H , U , and Z for each object have been ob-tained by minimizing the χ as defined in Eq. 8, andthey are reported in Table 9. The last two rows list theminimum χ values for the median (with the SIQR ofthe sample) and for the median on the medians reportedfor individual sources. In other words, the choice of thebest physical conditions was obtained by minimizing thesum of the deviations between the model predictionsand the observer diagnostic ratios. The obtained value of Z cover the range 20 (cid:46) Z (cid:46) (cid:46) Z (cid:46) Z (cid:12) , and medians of inten-sity ratios yielding Z ∼ Z (cid:12) . There is some spreadin the ionization parameter values, − . (cid:46) U (cid:46) − . n H ≈ .
75, and in several cases n H reaches 10 cm − . The median values are Z = 100 Z (cid:12) , log U = − . n H =13, therefore validating the original assumptionof log U = − .
5, log n H =12 for fixed physical condi-tion. The result for individual sources confirm the sce-nario of Fig. 12 for the wide majority of the samplesources. The higher n H values are consistent with re-cent inferences for the low-ionization BLR derived fromTemple et al. (2020), based on the FeIII UV emissionwhich is especially prominent in the UV spectra of xAquasars (Mart´ınez-Aldama et al. 2018). It is interest-ing to note that borderline objects (Al iii /Si iii] ≈ . Ciii] /Si iii] ≈
1) show higher values of the ionization pa-rameter (log U ≈ − . Z for the BC ( ≈ Z (cid:12) ).The inferences are less clear from BLUE (Table 10.In some cases, the permitted volume in the 3D pa-rameter space for individual sources covers a broadrange in U and n H as for the median. In some oth-ers the volume is more limited but the n H can bevery high log n H ∼
14. Most sources show high degreeof ionization, − (cid:46) log U (cid:46)
0, and only in one caseSDSS J102606.67+011459.0 there is apparently a low-ionization solution with U comparable to that of thelow-ionization BLR. The median suggests log n H ∼ ∼ -0.5, close to the values that we assumed for thefixed ( U , n H ) approach. The results on metallicity sug-gest in most cases Z (cid:38) Z (cid:12) , even if Z for the medianis Z = 10 Z (cid:12) . However, within 3 σ from the minimum χ , Z values up to 100 are also possible.In summary, the low-ionization BLR of xA sourcesseems to be consistently characterized by extremely lowionization, high density and very high metallicity, underthe assumption that Z scales with the solar chemicalcomposition. Diagnostics on BLUE is less constraining,and measurements are more difficult. The 0-order re-sults are however consistent again with high metallicity Z (cid:38) Z (cid:12) . Inferences on Z are in agreement with thecase for fixed physical conditions, as the Z determina-tions are weakly dependent on U , n H . DISCUSSION5.1.
A method to estimate Z The determination of the metal content of the broadline emitting region of xA quasars was made possible bythe following procedure:2 ´Sniegowska et al.
Figure 12.
The parameter space n H , U , Z . Left: data points in 3D space are elements in the grid of the parameter space thatare in agreement with the three main diagnostic ratios used for the BC, within the SIQR of the median estimate from Table 5.The individual contour was smoothed with a Gaussian kernel. Right: data points in the parameter space selected for not beingdifferent from χ within a confidence level of 3 σ . Figure 13.
The parameter space n H , U , Z . Left: data points in 3D space are elements in the grid of the parameter space thatare in agreement with the three main diagnostic ratios used for BLUE, within the SIQR of the median estimate from Table 5.The individual contour was smoothed with a Gaussian kernel. Right: data points in the parameter space selected for not beingdifferent from χ within a confidence level of 3 σ .
1. the estimation of an accurate redshift. Even if alllines are affected by significant blueshifts whichreduce the values of measured redshift, in the ab-sence of information from the H β spectral rangethe Al iii and the λ λ Civ andthe λ iii doublet can be used as atemplate BC. The component BLUE is defined asthe excess emission on the blue side of the BC. 3. a first estimate of metallicity can be obtained fromthe assumption that the low-ionization BLR asso-ciated with the BC and wind/outflow componentassociated with BLUE can be described by simi-lar physical conditions in different objects. Severaldiagnostic ratios can be associated with the inten-sity ratios predicted by an array of photoionizationsimulations, namely • for the BC: Al iii / H e ii λ Civ / H e ii λ iv / H e ii λ U , log n H ) =(-2.5,12) or (log U , log n H ) = (-2.5,13) • for the BLUE: Civ / H e ii λ iv +O iv] / H e ii λ Civ /Si iv +O iv] as-suming (log U , log n H ) = (0,9). etallicity in highly accreting quasars Table 9. Z , U , n H of individual sources andmedian derived from the BC SDSS JCODE Z [ Z (cid:12) ] log U log n H (1) (2) (3) (4)J010657.94-085500.1 200 -2.25 14J094637.83-012411.5 20 -1.25 13J085856.00+015219.4 100 -2.25 12J102606.67+011459.0 500 -2.00 13.75J211651.48+044123.7 50 -1.75 11.75J092641.41+013506.6 50 -1.75 13.75J102421.32+024520.2 100 -2.50 13.5J082936.30+080140.6 200 -3.75 13.75J151929.45+072328.7 500 -3.75 14J084525.84+072222.3 200 -2.50 13.75J084719.12+094323.4 100 -2.5 13J114557.84+080029.0 100 -2.00 14J150959.16+074450.1 100 -2.25 14Median 100 -2.5 13 µ (Medians) 100 ±
50 -2.25 ± ± Note —Columns: (1) SDSS identification, (2), (3) and(4) Z in units of Z (cid:12) , log U and log n H in cm − in thesame order. Table 10. Z , U , n H of individual sources andmedian derived from BLUE SDSS JCODE Z [ Z (cid:12) ] log U log n H (1) (2) (3) (4)J010657.94 − µ (Medians) 50 ±
15 -0.75 ± ± Note —Columns: (1) SDSS identification, (2), (3) and(4) Z in units of Z (cid:12) , log U and log n H in cm − in thesame order.
4. Estimates can be refined for individual sources re-laxing the constant (log U , log n H ) assumptions.Tight constraints can be obtained for the BC. TheBLUE is more problematic, because of both obser-vational difficulties and the absence of unambigu-ous diagnostics.Our method relies on ratios involving H e ii λ N c =10 ) as fixed. The role of turbulence is further discussedin Sect. 5.5, and is found to be not relevant unlike in thecase of Fe ii emission in the optical spectral range, whereeffects of self- and Ly α -fluorescence are important, (e.g.,Verner et al. 1999; Panda et al. 2018), while the N c effectis most likely negligible.Extension of the method to the full Population A is alikely possibility, since we do not expect a very strongeffect of the SED on the metallicity estimate, as long asthe SED has a prominent big blue bump, as it seemsto be case for Population A. The role of SED is likelyimportant if the method has to be extended to sourcesof Pop. B along the main sequence. Intensity ratiosinvolving H e ii λ L/L
Edd , witha much flatter SED at low
L/L
Edd . The extension toPop. B would therefore require a new dedicated arrayof simulations.5.2.
Accretion parameters of sample sources
The bolometric luminosity has been computed assum-ing a flat ΛCDM cosmological model with Ω Λ = 0.7, Ω m = 0.3, and H = 70 km s − Mpc − . Following Marziani& Sulentic (2014) we decided to use Al iii as virial broad-ening estimator for computing the M BH . Our estimatesadopt two different scaling laws: (1) the scaling laws ofVestergaard & Peterson (2006) for Civ and a second, un-published one based on Al iii (del Olmo et al., in prepa-ration). Eddington ratios have been obtained using theEddington luminosity L Edd ≈ . × ( M BH /M (cid:12) ) ergs − . The luminosity range of the sample is very lim-ited, less than a factor 3, 46 . (cid:46) log L (cid:46) .
3, in linewith the requirement of similar redshift and high fluxvalues. Correspondingly, the M BH and the Eddingtonratio are constrained in the range 8 . (cid:46) log M BH (cid:46) . − . (cid:46) log L/L
Edd (cid:46) .
18, respectively. The M BH sample dispersion is relatively small, with log M BH ∼ ± . M (cid:12) ]. The scatter in M BH and L/L
Edd is reducedto ≈ M BH and highest L/L
Edd . Applying a small correction(10%) to the FWHM to account for an excess broadeningin Al iii due to non-virial motions will decrease the M BH by 0.1 dex (as found by Negrete et al. 2018 for H β ), andincrease L/L
Edd correspondingly. If this correction isapplied the median
L/L
Edd is ≈ Civ
BCFWHM as a virial broadening estimator, also decreasing M BH median estimate by 0.1 dex, The accretion param-eters are consistent with extreme quasars of PopulationA at high mass and luminosity; they are mainly at the4 ´Sniegowska et al. Z BC AlIII/SiIIICIIIi/SiIIICIV/SiIVCIV/HeIISiIV/HeIICIV/AlIIIAlIII/HeII
Log ( L BOL ) Log ( M BH ) Log ( L / L Edd ) 0.57-0.220.270.650.81-0.0240.76-0.37-0.24-0.42 0.250.000.250.500.75 Z BLUE
AlIII/SiIIICIIIi/SiIIICIV/HeII BLUECIV/SiIV+OIV BLUESiIV/HeII BLUE
Log ( L BOL ) Log ( M BH ) Log ( L / L Edd ) -0.2-0.180.59-0.650.740.650.630.53 0.500.250.000.250.50
Figure 14.
Left panel: the correlation matrix between Z computed for the BC, and BC diagnostic ratios along withlog of bolometric luminosity, log of black hole mass and logof Eddington ratio. Right panel: Same, but for the BLUEcomponent. The numbers in each square show the Spearmanrank correlation coefficient. The color hue is proportional tothe correlation, from dark blue (strong negative correlation)to red (strong positive correlation). low L/L
Edd of Sample 3 (based on M BH estimates fromAl iii of Marziani & Sulentic (2014)). The small disper-sion in physical properties of the present sample (0 . Correlation between diagnostic ratios and AGNphysical properties
Considering the small dispersion in M BH , L/L
Edd andbolometric luminosity, it is hardly surprising that noneof the ratios utilized in this paper is significantly corre-lated with the accretion parameter. The highest degreeof correlation is seen between
L/L
Edd and
Civ /Al iii ,but still below the minimum ρ needed for a statisticallysignificant correlation.In Figure 14 we present the correlation between metal-licity and diagnostic ratios along with log of bolomet-ric luminosity, log of black hole mass and log of Ed-dington ratio for BC and BLUE. The strongest cor-relation between Z BC and intensity ratios are withSi iv / H e ii λ iii / H e ii λ BLUE , Si iv / H e ii λ BLUE correlates with physical parame-ters, whereas Z BC rather anti-correlates with them butnot at a statistically significant level. Considering thelimited range in luminosity and M BH , and small samplesize, these trends should be confirmed.The metallicity values we derive are very high amongquasars analyzed with similar techniques (e.g., Nagaoet al. 2006b; Shin et al. 2013; Sulentic et al. 2014): as mentioned, typical values for high- z quasars are around5 Z (cid:12) . This value could be taken as a reference over abroad range of redshift, and also for the sample con-sidered in the present paper, as there is no evidence ofmetallicity evolution in the BLR up to z ≈ . z , and a luminousxA object at redshift z ≈ .
23. Even if these authorsdid not derive Z from their data, the I Zw 1 intensityratios reported in their paper indicate very high metal-licity consistent with the values derived for the presentsample.More than inferences on the global enhancement of Z in the host galaxies, the absence of evolution pointstoward a circumnuclear source of metal enrichment, ulti-mately associated with a Starburst (e.g., Collin & Zahn1999a; Xu et al. 2012).A detailed comparison with previous work on the de-pendence of Z on accretion parameters is hampered bytwo difficulties. (1) Before comparing the intensity ra-tios of this paper, we should consider that other authorsdo not distinguish between BLUE and BC when com-puting the ratios. This has the unfortunate implica-tions that in some cases such as Al iii / Civ , the ratio istaken between lines emitted predominantly in differentregions (virialized and wind), presumably in very dif-ferent physical conditions. Not distinguishing betweenBC and BLUE yields
Civ /Al iii ∼ (cid:29)
1. (2) Methodsof M BH estimate differ. For example Matsuoka et al.(2011) use the Vestergaard & Peterson (2006) scalinglaws without any correction to the line width of Civ .This might easily imply overestimates of the M BH byfactor 5 – 10 (Sulentic et al. 2007). More properly, theanalysis by Shemmer et al. (2004) used H β from opticaland IR observations to compute M BH and to examinethe dependence of metallicity on accretion parameters.These authors found the strongest dependence on Ed-dington ratio (with respect to luminosity and mass) over6 orders of magnitude in luminosity, suggesting that lu-minosity and black hole mass are by far less relevant (asalso found, for example, by Shin et al. 2013).5.3. A posteriori analysis of N v strength As it was stressed in several works (e.g. Wang et al.2012a; Sulentic et al. 2014, the intensity of the N v lineis difficult to estimate due to blending with Ly α andstrongly affected by absorption. We model Ly α and N v using the same criteria as in Si iv modelling. However,in this work, we give only a qualitative judgement of N vetallicity in highly accreting quasars v broad component intensity is slightly higher or com-parable to Ly α broad component. Blue componentsdominate both lines. We notice also significantly higherintensity of blue component in comparison to broad onein Si iv and Civ blends. An example of source of thistype is shown in the upper half Fig. 15. On the contrary,sources with the lowest metallicities obtained from BCintensity ratios show the Ly α BC intensity higher thanin N v and the BC is stronger than BLUE. The samebehaviour of strong broad component we see in the Si iv and Civ ranges. An example of sources of this type isshown in the lower panels of Fig. 15. Shin et al. (2013)compared Si iv +O iv] and N v fluxes and found strong,significant correlation between them ( ρ = 0.75). The N v over H e ii λ H β should be a strong tracer of Z , asit is sensitive to secondary Z production and hence pro-portional to Z (Hamann & Ferland 1999). Therefore,we conclude that the N v emission is extremely strong,and consistent with very high metal content. A muchmore thorough investigation of the quasar absorption /emission system is needed to include N v as a Z estima-tor. This is deferred to further work.5.4. Role of column density
The column density assumed in the present paper islog N c =23 [cm − ]. With this value the emitting cloudsin the low-ionization conditions remain optically thick tothe Lyman continuum for most of the geometrical depthof the cloud. Even if the value log N c =23 may appear asa lower limit for the low-ionization BLR, as higher val-ues are required to explain low-ionization emission suchas Ca ii and Fe ii (Panda 2020; Panda et al. 2020a), theemission of the intermediate and high-ionized region isconfined within the fully ionized part of the line emit-ting gas whose extension is already much less than thegeometrical depth of the gas slab for log N c =23. There-fore, we expect no or negligible effect from an increasein the column density for the low ionization part of theBLR.For BLUE, the situation is radically different, andwe have no actual strong constraints on column den-sity. Most emission may come from a clumpy outflow(Matthews 2016, and references therein), and thereforeassuming a constant N c may not be appropriate. Con-sidering the poor constrain that we are able to obtain,we leave the issue to an eventual investigation.5.5. Role of turbulence
The results presented in this work refer to the case inwhich there is no significant micro turbulence included in the
CLOUDY computations. Fig. 16 shows that atlow ionization the effect is relatively modest, and thatin the high-ionization case appropriate for BLUE theeffect is very modest. Less obvious is the behavior atlow-ionization for R FeII : it shows an increase for t = 10km s − , but then it has a surprising drop at larger valueof the micro-turbulence. While the increase can be ex-plained by an increase of the transitions for which flu-orescence is possible, the decrease is not of obvious in-terpretation. It has been however confirmed by the in-dependent set of simulation of Panda et al. (2018, 2019)who used the more recent version of CLOUDY C17.01 (Ferland et al. 2017).5.6.
Metal segregation?
Metals are expected to be preferentially acceleratedby resonance scattering (e.g., Proga 2007b; Risaliti &Elvis 2010). In principle, for a sufficiently large photonflux, the acceleration of metals by radiation pressuremight become larger than the Coulomb friction, there-fore causing a decoupling of the metals with respect totheir parent plasma (Baskin & Laor 2012). This pos-sibility has been explored in the context of the BALs,and broad absorption and emission components are ex-pected to be related (Elvis 2000; Xu et al. 2020). Theionization parameter values are however several orders ofmagnitudes higher than the ones derived for the BLUEemission component. In addition our Z estimates forthe BLUE suggest, if anything, values lower or equalthan for the BC, whose Z might be related more to theoriginal chemical composition of the gas in the accretionmaterial. However, we ascribe the systematic differencesbetween BC and BLUE as uncertainties in the methodand measurement, so that Z from BLUE and BC shouldbe considered intrinsically equal.Considering that the most metal rich stars, galaxies,and molecular clouds in the Universe do not exceed Z ≈ Z (cid:12) (Maiolino & Mannucci 2019), circumnuclear starformation is needed for the chemical enrichment of theBLR gas (e.g., Collin & Zahn 1999a,b; Wang et al. 2011,2012b). Star formation may occur in the self gravitating,outer part of the disk. An alternative possibility is that amassive star could be formed inside the disk by accretionof disk gas (Cantiello et al. 2020).5.6.1. Abundance pollution?
An implication of the scenarios involving circumnu-clear or even nuclear star formation is that there couldbe an alteration of the relative abundance of elementswith respect to the standard solar composition (Anders& Grevesse 1989; Grevesse & Sauval 1998). Some sup-port is provided by the extreme
Civ /Si iv and Civ /Al iii ´Sniegowska et al. ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1450 1500 1550 1600 1650 1700 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 0.5 1 1.5 2 2.5 3 3.5 4 1200 1300 1400 1500 1600 1700Ly α +NV SiII SiII+OI CII SiIV CIV HeII OIII]+AlII N o r m a li z ed F λ Rest−frame wavelength (Å)J085856.00+015219.4 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1450 1500 1550 1600 1650 1700 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 0.5 1 1.5 2 2.5 3 1200 1300 1400 1500 1600 1700Ly α +NV SiII SiII+OI CII SiIV CIV HeII OIII]+AlII N o r m a li z ed F λ Rest−frame wavelength (Å)J211651.48+044123.7
Figure 15.
Analysis of sources showing the Ly α + N v blend. Top: calibrated rest-frame spectrum of SDSSJ085856.00+015219.4. before continuum subtraction. Global or local continuum are specified by a continuous coloured line,while the black line mark rest-framed data. Dot-dashed vertical lines correspond to the rest-frame wavelength of each emissionline. Bottom: multicomponent fits after continuum subtraction for the L αλ λ λ αλ λ λ λ λ λ λ λ λ λ λ − and in ˚A. Bottom: Same as in the previouspanels, for SDSS J211651.48+044123.7. etallicity in highly accreting quasars Figure 16.
Effects of turbulence on diagnostic line ratios,for
Civ / H e ii λ iv / H e ii λ Z (cid:12) (squares). The top panelassumes the low ionization conditions appropriate for BCemission, the bottom one for BLUE. In the top panel thegreen lines trace the same trends for the FeII blend at λ β ratio i.e., R FeII . that may hint to a selective enhancement of Al with re-spect to C. As suggested by Negrete et al. (2012), core-collapse supernovæ with very massive progenitors couldbe at the origin of a selective enhancement. Supernovaewith progenitors of masses between 15 and 40 M (cid:12) haveselective enhancement in their yields of Al and Si byfactors of ≈
100 and 10 relative to hydrogen with re-spect to solar (Chieffi & Limongi 2013). Since Carbonis also increased by a factor ∼
10 with respect to solar,the [Al/C] is expected to be a factor ∼
10 the solar valuein supernova ejecta. The case for Silicon is less clear, asthe enhancement is of the same order of magnitude ofthe one expected for Carbon. Pollution of gas by super-novæ may therefore lead to an estimate of the Z higherthan the actual one, if solar relative abundances are as-sumed. This possibility will be explored in an eventualwork (Garnica et al., in preparation).5.7. Implications for quasar structure evolution
Metallicity and the outflow prominence of quasarswere found to be highly correlated (Wang et al. 2012a;Shin et al. 2017). The implication of these results is thatxA sources, which show the highest blueshifts (Sulenticet al. 2017; Vietri et al. 2018; Martinez-Aldama et al.2018; Mart´ınez-Aldama et al. 2018), should also be themost metal rich. The xA sources should be at the topof the Z outflow parameter correlation of Wang et al.(2012a), if Z (cid:38) Z (cid:12) .There is evidence of a metallicity correlation betweenBLR and NLR (Du et al. 2014), as expected if the out-flows on spatial scales of kpc are originating in a diskwind. Zamanov et al. (2002) derived very small spa-tial scales at low luminosity. This provides additionalsupport to the idea that xA sources – which at low- z phenomenologically appear as Fe ii -strong NLSy1s, arerelatively young sources. Their low [Oiii] λ z ≈ L/L
Edd although there are no exam-ples of the extremes of xA sources showing blueshiftedemission in Al iii as prominent as the one of
Civ (e.g.,Mart´ınez-Aldama et al. 2017). There is no evidenceof heavy obscuration. They are certainly out of theobscured early evolution stage in which the accretingblack hole is enveloped by gas and dust (see the sketchin D’Onofrio & Marziani 2018). The W Civ distri-bution covers the upper half of the one of Mart´ınez-Aldama et al. (2018). There are no weak-lined quasarsfollowing Diamond-Stanic et al. (2009). The xA sourcesof the present sample may have reached a sort of sta-ble equilibrium between gravitation and radiation forces8 ´Sniegowska et al. made perhaps possible by the development by an opti-cally thick, geometrically thick accretion disk, and by itsanisotropic radiation properties (e.g., Abramowicz et al.1988; Szuszkiewicz et al. 1996; S¸adowski et al. 2014).The median value of the peak displacement of theBLUE component is around ≈ − , and thecentroid at half maximum is shifted by 5000 km s − .The extreme blueshifts in the metal lines imply out-flows that may not remain bound to the potential well ofthe black hole and of the inner bulge of the host galax-ies (e.g., Marziani et al. 2016b, and references therein).The high metal content of the outflows, estimated bythe present work to be in the range 10 − Z (cid:12) , im-plies that these sources are likely to be a major sourceof metal enrichment of the interstellar gas of the hostgalaxy and of the intergalactic medium. Using a stan-dard estimate for the mass outflow rate ˙ M (Marzianiet al. 2016b), ˙ M ≈ L CIV , v r − n − M (cid:12) yr − , weobtain an outflow rate of ˙ M ≈ M (cid:12) yr − , assumingmedian values for the sources of our sample: median out-flow velocity from the peak of BLUE ≈ − − , amedian luminosity of the Civ
BLUE (corrected becauseof Galactic extinction) of 4.2 · erg s − , a medianradius 5.9 · cm from the Kaspi et al. (2007) radius-luminosity correlation for Civ , and n = 1. For a dutycycle of ∼ yr, the expelled mass of heavily enriched-gas could be ∼ M (cid:12) . CONCLUSIONThe sources at the extreme end of Population A alongthe main sequence are defined by the prominence of theirFe ii emission and, precisely, by the selection criterion R FeII (cid:38) ii -strong sources since Lipari et al. (1993); Grahamet al. (1996), their relevance to galactic and large scalestructure evolution is being reconsidered anew with thehelp of the quasar main sequence. This paper adds toother aspects that were considered by previous inves-tigations (for example, the very powerful outflows, thedisjoint low- and high-ionization emitting regions, firstsuggested by Collin-Souffrin et al. 1988), a quantitiveanalysis of the chemical composition of xA sources. Themain aspects of the present investigations can be sum-marized as follows: • We distinguish between two emission line com-ponents most likely origination from emitting inwidely different physical conditions: a virializedlow-ionization BLR, and a high-ionization regionassociated to a very strong blueshifted excess in the
Civ emission line. This is the conditio sinequa non for meaningful Z estimates. • The physical conditions in the low and high regionswere confirmed to be very different, with the lowionization ( U , n H ) ≈ (-2.75, 13) and the high ion-ization ( U , n H ) ≈ (-0.5,8. – 9). The high ionizationregion parameters are however poorly constrained. • Using intensity ratios between the strongest metallines and H e ii λ λ (cid:46) Z (cid:46) Z (cid:12) ,with most likely values around several tens of thetime solar metallicity. • We find evidence of overabundance of Al with re-spect to C. This result points toward possible pol-lution of the broad line emitting gas chemical com-position by supernova ejecta.xA quasars are perhaps the only quasars whose ejec-tion are able to overcome the potential well of the blackhole and of the host galaxy. Applying the method tolarge samples of quasars would permit to constrain themetal enrichment processes on a galactic scale.The present analysis relied heavily on the H e ii λ Z of the emit-ting regions). Therefore, a more precise analysis wouldrequire spectra of moderate dispersion but of higher S/Nratio. A large part of the scatter and/or systematic dif-ference for various Z estimators is related to difficultyto isolate faint and broad emission in relatively noisyspectra. etallicity in highly accreting quasars Abramowicz, M. A., Czerny, B., Lasota, J. P., &Szuszkiewicz, E. 1988, ApJ, 332, 646Adelman-McCarthy, J. K., Ag¨ueros, M. A., Allam, S. S.,et al. 2008a, ApJS, 175, 297—. 2008b, ApJS, 175, 297Adhikari, T. P., R´o˙za´nska, A., Czerny, B., Hryniewicz, K.,& Ferland, G. J. 2016, ApJ, 831, 68Anders, E., & Grevesse, N. 1989, GeoCoA, 53, 197Ba˜nados, E., Venemans, B. P., Decarli, R., et al. 2016,ApJS, 227, 11Baldwin, J. A., Hamann, F., Korista, K. T., et al. 2003,ApJ, 583, 649Baskin, A., & Laor, A. 2012, MNRAS, 426, 1144Becker, R. H., White, R. L., & Helfand, D. J. 1995, ApJ,450, 559Bisogni, S., di Serego Alighieri, S., Goldoni, P., et al. 2017,A&A, 603, A1Boroson, T. A., & Green, R. F. 1992, ApJS, 80, 109Brotherton, M. S., Wills, B. J., Francis, P. J., & Steidel,C. C. 1994, ApJ, 430, 495Br¨uhweiler, F., & Verner, E. 2008, ApJ, 675, 83Cantiello, M., Jermyn, A. S., & Lin, D. N. C. 2020, arXive-prints, arXiv:2009.03936Chieffi, A., & Limongi, M. 2013, ApJ, 764, 21 Coatman, L., Hewett, P. C., Banerji, M., & Richards, G. T.2016, MNRAS, 461, 647Collin, S., & Zahn, J.-P. 1999a, A&Ap, 344, 433—. 1999b, Ap&SS, 265, 501Collin-Souffrin, S., Dyson, J. E., McDowell, J. C., & Perry,J. J. 1988, MNRAS, 232, 539Collin-Souffrin, S., & Lasota, J.-P. 1988, PASP, 100, 1041Corbin, M. R., & Boroson, T. A. 1996, ApJS, 107, 69D’ Agostini, G. 2003, Bayesian reasoning in data analysis -A critical introduction (Singapore: World Scientific Co.)Davidson, K., & Netzer, H. 1979, Reviews of ModernPhysics, 51, 715Diamond-Stanic, A. M., Fan, X., Brandt, W. N., et al.2009, ApJ, 699, 782Dietrich, M., Hamann, F., Shields, J. C., et al. 2003, ApJ,589, 722D’Onofrio, M., & Marziani, P. 2018, Frontiers inAstronomy and Space Sciences, 5, 31Du, P., Wang, J.-M., Hu, C., et al. 2016a, ApJL, 818, L14—. 2014, MNRAS, 438, 2828Du, P., Lu, K.-X., Hu, C., et al. 2016b, ApJ, 820, 27Elvis, M. 2000, ApJ, 545, 63Feibelman, W. A., & Aller, L. H. 1987, ApJ, 319, 407 ´Sniegowska et al. Ferland, G. J., Done, C., Jin, C., Landt, H., & Ward, M. J.2020, MNRAS, 494, 5917Ferland, G. J., Hu, C., Wang, J., et al. 2009, ApJL, 707,L82Ferland, G. J., Porter, R. L., van Hoof, P. A. M., et al.2013, RevMexA&Ap, 49, 137Ferland, G. J., Chatzikos, M., Guzm´an, F., et al. 2017,RMxAA, 53, 385Gaskell, C. M. 2000, NewAR, 44, 563Graham, M. J., Clowes, R. G., & Campusano, L. E. 1996,MNRAS, 279, 1349Grevesse, N., & Sauval, A. J. 1998, SSRv, 85, 161Hamann, F., & Ferland, G. 1992, ApJL, 391, L53—. 1999, ARA&A, 37, 487Hamann, F., Korista, K. T., Ferland, G. J., Warner, C., &Baldwin, J. 2002, ApJ, 564, 592Hartig, G. F., & Baldwin, J. A. 1986, ApJ, 302, 64Izotov, Y. I., & Thuan, T. X. 1999, ApJ, 511, 639Juarez, Y., Maiolino, R., Mujica, R., et al. 2009, A&Ap,494, L25Kaspi, S., Brandt, W. N., Maoz, D., et al. 2007, ApJ, 659,997Keenan, F. P., Ahmed, S., Brage, T., et al. 2002, MNRAS,337, 901Kovaˇcevi´c-Dojˇcinovi´c, J., & Popovi´c, L. ˇC. 2015, ApJS,221, 35Kriss, G. 1994, Astronomical Data Analysis Software andSystems III, A.S.P. Conference Series, 61, 437Kuraszkiewicz, J., Wilkes, B. J., Schmidt, G., et al. 2009,ApJ, 692, 1180Leighly, K. M., & Moore, J. R. 2004, ApJ, 611, 107Lipari, S., Terlevich, R., & Macchetto, F. 1993, ApJ, 406,451Maiolino, R., & Mannucci, F. 2019, A&A Rv, 27, 3Mart´ınez-Aldama, M. L., Del Olmo, A., Marziani, P., et al.2017, Frontiers in Astronomy and Space Sciences, 4, 29Mart´ınez-Aldama, M. L., del Olmo, A., Marziani, P., et al.2018, A&A, 618, A179Martinez-Aldama, M. L., Del Olmo, A., Marziani, P., et al.2018, Frontiers in Astronomy and Space Sciences, 4, 65Marziani, P., Mart´ınez Carballo, M. A., Sulentic, J. W.,et al. 2016a, Ap&SS, 361, 29Marziani, P., & Sulentic, J. W. 2014, MNRAS, 442, 1211Marziani, P., Sulentic, J. W., Dultzin-Hacyan, D., Calvani,M., & Moles, M. 1996, ApJS, 104, 37Marziani, P., Sulentic, J. W., Negrete, C. A., et al. 2010,MNRAS, 409, 1033Marziani, P., Sulentic, J. W., Stirpe, G. M., et al. 2016b,Ap&SS, 361, 3 Marziani, P., Sulentic, J. W., Stirpe, G. M., Zamfir, S., &Calvani, M. 2009, A&Ap, 495, 83Marziani, P., Dultzin, D., Sulentic, J. W., et al. 2018,Frontiers in Astronomy and Space Sciences, 5, 6Mathews, W. G., & Ferland, G. J. 1987, ApJ, 323, 456Mathur, S. 2000, MNRAS, 314, L17Matsuoka, K., Nagao, T., Marconi, A., Maiolino, R., &Taniguchi, Y. 2011, A&A, 527, A100Matsuoka, Y., Oyabu, S., Tsuzuki, Y., & Kawara, K. 2007,ApJ, 663, 781Matthews, J. H. 2016, PhD thesis, University ofSouthamptonMej´ıa-Restrepo, J. E., Trakhtenbrot, B., Lira, P., Netzer,H., & Capellupo, D. M. 2016, MNRAS, 460,arXiv:1603.03437Murray, N., & Chiang, J. 1997, ApJ, 474, 91Murray, N., Chiang, J., Grossman, S. A., & Voit, G. M.1995, ApJ, 451, 498Nagao, T., Maiolino, R., & Marconi, A. 2006a, A&A, 447,863Nagao, T., Marconi, A., & Maiolino, R. 2006b, A&Ap, 447,157Nardini, E., Lusso, E., Risaliti, G., et al. 2019, A&A, 632,A109Negrete, A., Dultzin, D., Marziani, P., & Sulentic, J. 2012,ApJ, 757, 62Negrete, C. A., Dultzin, D., Marziani, P., & Sulentic, J.2010, ArXiv e-prints, arXiv:1011.4248Negrete, C. A., Dultzin, D., Marziani, P., & Sulentic, J. W.2014, ApJ, 794, 95Negrete, C. A., Dultzin, D., Marziani, P., et al. 2018, A&A,620, A118Netzer, H., & Marziani, P. 2010, ApJ, 724, 318Onoue, M., Ba˜nados, E., Mazzucchelli, C., et al. 2020, ApJ,898, 105Osmer, P. S., & Smith, M. G. 1976, ApJ, 210, 267Osterbrock, D. E., & Ferland, G. J. 2006, Astrophysics ofgaseous nebulae and active galactic nuclei (Mill Valley,CA: University Science Books)Osterbrock, D. E., & Pogge, R. W. 1985, ApJ, 297, 166Panda, S. 2020, arXiv e-prints, arXiv:2004.13113Panda, S., Czerny, B., Adhikari, T. P., et al. 2018, ApJ,866, 115Panda, S., Mart´ınez-Aldama, M. L., Marinello, M., et al.2020a, arXiv e-prints, arXiv:2004.05201Panda, S., Marziani, P., & Czerny, B. 2019, ApJ, 882, 79—. 2020b, Contributions of the Astronomical ObservatorySkalnate Pleso, 50, 293Pˆaris, I., Petitjean, P., Ross, N. P., et al. 2017, A&A, 597,A79 etallicity in highly accreting quasars Peterson, B. M., & Wandel, A. 1999, ApJL, 521, L95Plotkin, R. M., Shemmer, O., Trakhtenbrot, B., et al. 2015,ApJ, 805, 123Popovi´c, L. ˇC., Mediavilla, E. G., Kubiˇcela, A., &Jovanovi´c, P. 2002, A&A, 390, 473Proga, D. 2007a, ApJ, 661, 693Proga, D. 2007b, in Astronomical Society of the PacificConference Series, Vol. 373, The Central Engine ofActive Galactic Nuclei, ed. L. C. Ho & J.-W. Wang, 267Punsly, B., Marziani, P., Bennert, V. N., Nagai, H., &Gurwell, M. A. 2018, ApJ, 869, 143Richards, G. T., Lacy, M., Storrie-Lombardi, L. J., et al.2006, ApJS, 166, 470Richards, G. T., Kruczek, N. E., Gallagher, S. C., et al.2011, AJ, 141, 167Risaliti, G., & Elvis, M. 2010, A&A, 516, A89Risaliti, G., Salvati, M., & Marconi, A. 2011, MNRAS, 411,2223S¸adowski, A., Narayan, R., McKinney, J. C., &Tchekhovskoy, A. 2014, MNRAS, 439, 503Schneider, D. P., Richards, G. T., Hall, P. B., et al. 2010,AJ, 139, 2360Shemmer, O., & Netzer, H. 2002, ApJL, 567, L19Shemmer, O., Netzer, H., Maiolino, R., et al. 2004, ApJ,614, 547Shen, Y. 2016, ApJ, 817, 55Shen, Y., & Ho, L. C. 2014, Nature, 513, 210Shen, Y., Richards, G. T., Strauss, M. A., et al. 2011,ApJS, 194, 45Shields, G. A. 1976, ApJ, 204, 330Shin, J., Nagao, T., & Woo, J.-H. 2017, ApJ, 835, 24Shin, J., Woo, J.-H., Nagao, T., & Kim, S. C. 2013, ApJ,763, 58Sulentic, J. W., Bachev, R., Marziani, P., Negrete, C. A., &Dultzin, D. 2007, ApJ, 666, 757Sulentic, J. W., Marziani, P., del Olmo, A., et al. 2014,A&A, 570, A96Sulentic, J. W., Marziani, P., & Dultzin-Hacyan, D. 2000a,ARA&A, 38, 521Sulentic, J. W., Marziani, P., Dultzin-Hacyan, D., Calvani,M., & Moles, M. 1995, ApJL, 445, L85Sulentic, J. W., Marziani, P., Zamanov, R., et al. 2002,ApJL, 566, L71 Sulentic, J. W., Repetto, P., Stirpe, G. M., et al. 2006,A&Ap, 456, 929Sulentic, J. W., Zwitter, T., Marziani, P., &Dultzin-Hacyan, D. 2000b, ApJL, 536, L5Sulentic, J. W., del Olmo, A., Marziani, P., et al. 2017,A&A, 608, A122Szuszkiewicz, E., Malkan, M. A., & Abramowicz, M. A.1996, ApJ, 458, 474Temple, M. J., Ferland, G. J., Rankine, A. L., et al. 2020,MNRAS, 496, 2565Vanden Berk, D. E., Richards, G. T., Bauer, A., et al. 2001,AJ, 122, 549Verner, E. M., Verner, D. A., Korista, K. T., et al. 1999,ApJS, 120, 101V´eron-Cetty, M.-P., V´eron, P., & Gon¸calves, A. C. 2001,AAp, 372, 730Vestergaard, M., & Peterson, B. M. 2006, ApJ, 641, 689Vestergaard, M., & Wilkes, B. J. 2001, ApJS, 134, 1Vietri, G., Piconcelli, E., Bischetti, M., et al. 2018, A&A,617, A81Vila-Costas, M. B., & Edmunds, M. G. 1993, MNRAS, 265,199Wang, H., Zhou, H., Yuan, W., & Wang, T. 2012a, ApJL,751, L23Wang, J.-M., Du, P., Baldwin, J. A., et al. 2012b, ApJ, 746,137Wang, J.-M., Du, P., Li, Y.-R., et al. 2014, ApJL, 792, L13Wang, J.-M., Du, P., Valls-Gabaud, D., Hu, C., & Netzer,H. 2013, Physical Review Letters, 110, 081301Wang, J.-M., Ge, J.-Q., Hu, C., et al. 2011, ApJ, 739, 3Warner, C., Hamann, F., & Dietrich, M. 2004, ApJ, 608,136Wildy, C., Czerny, B., & Panda, S. 2019, A&A, 632, A41Wills, D., & Netzer, H. 1979, ApJ, 233, 1Xu, D., Komossa, S., Zhou, H., et al. 2012, AJ, 143, 83Xu, X., Zakamska, N. L., Arav, N., Miller, T., & Benn, C.2020, MNRAS, 495, 305Zamanov, R., Marziani, P., Sulentic, J. W., et al. 2002,ApJL, 576, L9Zheng, W. 1988, Astrophysical Letters andCommunications, 27, 275Zhou, H., Wang, T., Yuan, W., et al. 2006, ApJS, 166, 128 ´Sniegowska et al. APPENDIX A. REST-FRAME SPECTRA AND FITSThe spectral analysis of the 13 objects of our sample is shown in the figures below. ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J010657.94−085500.1
Figure 17.
Top panels: calibrated rest-frame spectrum of SDSS J010657.94-085500.1 before continuum substraction. Globalor local continuum are specified by a continuous coloured line, while the black line mark rest-framed data. Dot-dashed verticallines correspond to the rest-frame wavelength of each emission line. Bottom: multicomponent fits after continuum subtractionfor the SiIV λ λ λ λ λ λ λ λ λ λ λ λ λ λ λ λ − and in ˚A.B. DIAGNOSTIC INTENSITY RATIOS IN THE PLANE ( U , n H ) AS A FUNCTION OF METALLICITYThe results of the arrays of simulations as a function of n H , U , and Z are shown below, for N c = 10 . The SEDshape is the same for all simulations (table agn) which corresponds to the SED of Mathews & Ferland (1987). Noturbulence was assumed. etallicity in highly accreting quasars ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J082936.30+080140.6 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1300 1320 1340 1360 1380 1400 1420 1440 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1450 1500 1550 1600 1650 1700 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.2 0.4 0.6 0.8 1 1.2 1.4 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J084525.84+072222.3 0 0.1 0.2 0.3 0.4 0.5 1300 1320 1340 1360 1380 1400 1420 1440 0 0.1 0.2 0.3 0.4 0.5 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.1 0.2 0.3 0.4 0.5 1450 1500 1550 1600 1650 1700 0 0.1 0.2 0.3 0.4 0.5 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J084719.12+094323.4
Figure 17.
Same of the previous panel, for SDSS J082936.30+080140.6 and SDSS J084525.84+072222.3, and SDSSJ084719.12+094323.4. ´Sniegowska et al. ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.4 0.6 0.8 1 1.2 1.4 1.6 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J085856.00+015219.4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1300 1320 1340 1360 1380 1400 1420 1440 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1450 1500 1550 1600 1650 1700 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 30000 45000 0.4 0.6 0.8 1 1.2 1.4 1.6 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J092641.41+013506.6 0 0.1 0.2 0.3 0.4 0.5 0.6 1300 1320 1340 1360 1380 1400 1420 1440 0 0.1 0.2 0.3 0.4 0.5 0.6 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.1 0.2 0.3 0.4 0.5 0.6 1450 1500 1550 1600 1650 1700 0 0.1 0.2 0.3 0.4 0.5 0.6 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J094637.83−012411.5
Figure 17.
Same of the previous panel, for SDSS J085856.00+015219.4, SDSS J092641.41+013506.6, and SDSS J094637.83-012411.5. etallicity in highly accreting quasars ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J102421.32+024520.2 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1300 1320 1340 1360 1380 1400 1420 1440 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1450 1500 1550 1600 1650 1700 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J102606.67+011459.0 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1300 1320 1340 1360 1380 1400 1420 1440 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1450 1500 1550 1600 1650 1700 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J114557.84+080029.0
Figure 17.
Same of the previous panel, for SDSS J102421.32+024520.2 SDSS J102606.67+011459.0 SDSS J114557.84+080029.0. ´Sniegowska et al. ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.4 0.6 0.8 1 1.2 1.4 1.6 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J150959.16+074450.1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1300 1320 1340 1360 1380 1400 1420 1440 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1450 1500 1550 1600 1650 1700 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 1450 1500 1550 1600 1650 1700 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0 0.5 1 1.5 2 2.5 3 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J151929.45+072328.7 0 0.05 0.1 0.15 0.2 0.25 0.3 1300 1320 1340 1360 1380 1400 1420 1440 0 0.05 0.1 0.15 0.2 0.25 0.3 1300 1320 1340 1360 1380 1400 1420 1440 −0.15−0.1−0.05 0 0.05 0.1−20000 −10000 0 10000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 −0.15−0.1−0.05 0 0.05 0.1 −15000 0 15000 ∆ v r (km s −1 ) 0 0.05 0.1 0.15 0.2 0.25 1750 1800 1850 1900 1950 2000 2050 −0.15−0.1−0.05 0 0.05 0.1−15000 0 15000 30000 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1300 1400 1500 1600 1700 1800 1900 2000SiII+OICII SiIV CIV HeII OIII]+AlII SiII AlIII SiIIICIII]+FeIII N o r m a li z ed F λ Rest−frame wavelength (Å)J211651.48+044123.7
Figure 17.
Same of the previous panel, for SDSS J150959.16+074450.1, SDSS J151929.45+072328.7, SDSSJ211651.48+044123.7. etallicity in highly accreting quasars F i g u r e . I s o ph o t a l c o n t o u r i n t h e l og U – l og Z f o r d i ag n o s t i c li n e i n t e n s i t y r a t i o s f o r n H = c m − (t o p ) a nd n H = c m − ( b o tt o m ) , f o r c o l u m nd e n s i t y N c = c m − . ´Sniegowska et al. F i g u r e . I s o ph o t a l c o n t o u r i n t h e l og U – l og n H f o r li n e i n t e n s i t y r a t i o s ( f r o m l e f tt o r i g h t) a s a f un c t i o n o f m e t a lli c i t y ( f r o m t o p - t o - b o tt o m : Z = , , , Z (cid:12) ) , f o r c o l u m nd e n s i t y N c = c m −2