From sub-solar to super-solar chemical abundances along the quasar main sequence

The 4D Eigenvector 1 sequence has proven to be a highly effective tool for organizing observational and physical properties of type 1 active galactic nuclei (AGN). In this paper, we present multiple measurements of metallicity for the broad line region gas, from new or previously published data. We demonstrate a consistent trend along the optical plane of the E1 (also known as the quasar main sequence), defined by the line width of H$\beta$ and by a parameter measuring the prominence of singly-ionized iron emission. The trend involves an increase from sub-solar metallicity in correspondence with extreme Population B (weak FeII emission, large H$\beta$ FWHM) to metallicity several tens the solar value in correspondence with extreme Population A (very strong FeII optical emission, narrower H$\beta$ profiles). The data establish the metallicity as a correlate of the 4D E1/main sequence. If the very high metallicity gas ($Z \gtrsim 10 Z_\odot$) is expelled from the sphere of influence of the central black hole, as indicated by the widespread evidence of nuclear outflows and disk wind in the case of sources radiating at high Eddington ratio, then it is possible that the outflows from quasars played a role in chemically enriching the host galaxy.


Introduction: A Main Sequence and the Eigenvector 1 for Quasars
The main sequence (MS) concept in quasar research draws parallels with stellar evolutionary sequences, but in this case, it is applied to quasar properties.A study on a sample of Palomar-Green quasars detected an interesting anti-correlation between the strength of the FeIIλ4570 emission line and the Full Width at Half Maximum (FWHM) of the broad Hβ emission line of type-1 quasars with low redshifts (z < 1, [1]).This anti-correlation suggests that, as the FeIIλ4570 feature becomes stronger, the FWHM tends to decrease, and is one of the main correlations known as the Eigenvector 1 of quasars.This finding has been established through the analysis of samples of increasing size over the years [2][3][4][5], and has proved to be fundamental to organize type-1 AGN properties in a systematic way with predictive ability.
Quasars are categorized into different spectral types along the MS [e.g., 5], and two primary populations, referred to as Population A and Population B, have been identified [3].These categories are based on specific properties such as the Eddington ratio (a measure of the accretion rate) and orientation [5][6][7].The "4DE1" classification, introduced by Sulentic et al. [3], further organizes quasar properties according to the Eddington ratio and orientation, revealing a systematic pattern of variation across different types of AGN encompassing their outflow phenomenology and their accretion mode [8].The assignation of most quasar spectral types permitted the prediction of their UV, X-ray, radio, and FIR properties with a high degree of confidence.
The MS, much like the equivalent concept in stellar evolution, is used as an evolutionary framework for understanding quasars.This sequence spans from young and rejuvenated quasars, characterized by specific spectral properties (Extreme Population A), to older and more mature quasars with distinct characteristics (Population B).Differences in factors such as black hole mass, Eddington ratio, disk winds, or outflow properties, derived from the strengths and profiles of emission lines are used to define and distinguish these evolutionary stages [9,10].
Currently, the available evidence regarding the correlation between metal content in the broad line emitting region (BLR) and the quasar main sequence (MS) remains partial and inconclusive [7,11].There is a long tradition of studies attempting to estimate the metallicity in the BLR of AGN over a broad range of redshift, from z ≈ 0, up to z ≈ 6 [e.g., [12][13][14][15][16][17][18].All these studies derive metallicity in the range from a few times solar to about 10 times solar, values that are significantly higher than the ones found even in most massive and metal-rich galaxies [19,20].When considering the past metallicity estimates along the main sequence, an intriguing trend emerges: not all quasars are accreting matter with super-solar metal content [21], and only at one end of the MS, the BLR gas may be enriched by a metal content even above ten times solar, possibly with pollution by supernova ejecta [22].
In this contribution, we will use the results of a new analysis and of several recent papers to gather a view of the global trend along the MS.Section 2 summarizes the new observations from ground-and space-based observatories.Section 3 elucidates the methodology applied to the metallicity estimates, stressing the need to isolate line components that are spectroscopically resolved, and may correspond to emitting regions in different physical conditions ( §3.1).Results of several individual sources and composite spectra representative of entire spectral types (ST) introduced in §4 clearly delineate a sequence of increasing metallicity from extreme Population B to extreme Population A ( §5).As outflows from the AGN are of special relevance for galactic evolution [23][24][25], it is important to isolate the corresponding line component, whenever possible ( §5.3).The AGN outflows may provide enrichment of the nuclear and circumnuclear region of the host galaxies ( §6), although ensuing chemical feedback is expected to be relevant only at high AGN luminosity.In § 6, we outline how the different metallicities inferred for the BLR of low-z AGN fit the evolutionary interpretation of the quasar MS.

Observations
The new spectral data employed in this study were acquired through a series of distinct observations.Specifically:

•
For Mrk 335, Mrk 478, and Fairall 9, we utilized optical spectra sourced from [26].Additionally, the UV spectra of Mrk 335 were obtained during observations conducted on the 4th and 7th of January 2013, utilizing the Cosmic Origins Spectrograph (COS) aboard the Hubble Space Telescope (HST) with the G140L grism.The UV spectra of Mrk 478 were acquired on the 5th of December 1996, utilizing the HST's Faint Object Spectrograph (FOS) and employing the G130H and G190H grisms.For Fairall 9, UV spectra were collected on the 22nd of January 1993, utilizing the HST's FOS with the G190H and G270H grisms, and subsequently on the 18th of July 2012, employing the HST's COS with the G130M and G160M grisms.

•
For NGC 1275, the data were sourced from [27].Optical spectra encompassed various observations spanning the period from 1983 to 2017.Additionally, UV spectra were acquired from the HST MAST, with FOS observations in 1993 and COS observations in 2011.

•
For PHL 1092, data were drawn from [28].Optical spectra were captured using the Goodman spectrograph at the 4.1-meter telescope of the Southern Observatory for Astrophysical Research (SOAR) on the night of 12th December 2014.The UV spectra were obtained using the HST's Space Telescope Imaging Spectrograph (STIS) on the night of 20th August 2003.
In addition, we considered: • composite spectra for radio-loud (RL) and radio-quiet (RQ) Population B sources.The data on which the composites were based were described in a previous work [21]; • median results for two samples of xA sources at intermediate redshift (z ≈ 2) [22,29] lacking the rest-frame optical spectrum.

Multicomponent analysis
A crucial technique in quasar studies is the multicomponent analysis, which involves dissecting complex emission line profiles into distinct components.This approach allows us to uncover details about different regions of emission, potentially revealing information about kinematics, velocities, and structures.Line ratios, such as those involving CIVλ1549, HeIIλ1640, AlIIIλ1860, and SiIV+OIV]λ1402, are used as diagnostic tools to infer physical conditions, such as ionization and metallicity, within the quasar environments.In the spectra of quasars, both low-ionization and high-ionization optical and UV lines are observed (LILs and HILs, respectively), and they provide valuable insights into the physical conditions and structure of the quasar BLR.These lines are emitted, at least in part, in distinct regions with different characteristics [30,31].
We distinguish a broad component (BC) that is broadened by the Doppler effect due to the rapid motion of the gas in the BLR in a velocity field dominated by Keplerian motions [32,33].Lines meeting this condition are also referred to as "virialized broad emission lines."Low-ionization lines like Hβ and MgIIλ2800 are typically emitted in the virialized BLR, which may be characterized by high densities and column densities, but relatively low ionization (ionization parameter ∼ 10 −2 .High-ionization lines like CIVλ1549, HeIIλ1640 and NVλ1240 are also emitted in the virialized BLR.However, these lines trace the regions of the BLR that are exposed to the most intense and energetic radiation from the accretion disk [30,31].They are often broader than low-ionization lines due to higher velocities in this part of the BLR, and show prominent blueshifts [34,35].
In short, we subdivide all lines into three main components that can account for the diversity of line profiles, along the quasar MS, resulting from the balance between gravitation and radiation forces [36].

Population A
Broad Component (BC) Represented by a Lorentzian function, symmetric and unshifted or slightly redshifted [4,37].

Blueshifted Component (BLUE)
It is defined as the excess of emission on the blue side of the BC.The shape can be irregular, but the profiles resemble "triangular" or "trapezoidal shapes" [38,39] that are usually well modeled by asymmetric Gaussian [38,40].
The blueshifted component can be very prominent at high Eddington ratios and high luminosity, dominating the HIL emission [41][42][43].The BLUE is increasing in prominence in the HILs along the quasar main sequence and reaches its maximum at ST A3-A4, where R FeII is also at a maximum.
The two components most likely represent coexisting regions [44], albeit in very different physical and dynamical conditions.While the BC is associated with a dense, low ionization region, capable of emitting mainly (but not exclusively) LILs and maintaining a virialized velocity field (for which a large column density is needed [45]), the outflowing gas should be of higher ionization.The assumption that BLUE and BC refer to regions with the same metallicity has been questioned by models of the AGN involving nuclear star formation [46], and in the following, we will attempt independent metallicity estimates from BLUE and BC.

Population B
Broad Component (BC) Represented by a Gaussian function, symmetric and unshifted or slightly redshifted [4].

Very Broad Component (VBC)
Represented by a Gaussian function, redshifted by about ∼ 2000km s −1 [4,47].Given the virial velocity field of the emitting regions, this component represents the innermost emission of the BLR.Several works have described the Population B Balmer profiles in terms of a BLR and a very broad line region (VBLR) [48,49].It is unclear whether the emitting gas might be so highly ionized to be optically thin to the Lyman continuum [50].The origin of the redshift is the subject of current debate, and two main alternatives have been proposed: gravitational redshift [51][52][53][54][55][56], and infall [57].The data unambiguously support the gravitational redshift hypothesis only for logM BH ≳ 8.7 [M ⊙ ], while lower M BH require very low L/L Edd for the profiles to show a significant gravitational effect [58].

Blueshifted Component (BLUE)
It is defined as the excess of emission on the blue side of the BC+VBC profile.The blueshifted component is usually not prominent at low Eddington ratios but can still affect the centroid and asymmetry index of both HILs and LILs.Due to its weakness, BLUE is always modeled by a shifted (symmetric) Gaussian.
Fig. 1 summarizes the interpretation of the line profiles of CIV and Hβ assumed as prototypical HIL and LIL, respectively, for Population B, Population A, and extreme Population A. Since BLUE is barely resolved in Pop.B, no estimates of the metallicity Z will be attempted.For the object of extreme Population A an estimate has been carried out for BC and BLUE, while for Pop.B sources, for the BC and VBC.In the latter case, we expect that the metallicity is the same, but the physical conditions would reflect a gradient in ionization (lower ionization for the BC and higher for the VBC), and hence be different on average.

Emission line ratios
Z indicator CIV/SiIV+OIV]λ1402 has been widely applied as metallicity indicator [e.g., 13,16].In photoionization equilibrium, the classical argument derived for HII regions that the electron temperature decreases with increasing metallicity [59] works for the BLR as well.The intensity of the CIV line actually decreases as the metal abundance increases.However, the reason why the ratio CIV/SiIV is a metallicity indicator resides in the "competition" of He + ions that have roughly the same creation potential of C ++ .As a result, the Strömgren sphere of C +3 decreases much more strongly with increasing Z than for Si +3 : the ionization potential of Si +2 is 2.46 Ryd, so the relatively unabsorbed continuum between 2.5 and 3.5 Ryd is available to maintain a proportionality with its abundance [60].This effect dominates over the lower electron temperature that affects the collisional excitation rates of both SiIV and CIV, expected to be higher for SiIV.

Z indicators CIV/HeIIλ1640 and SiIV/HeIIλ1640
When considering indicators like CIV/-HeIIλ1640 and SiIV/HeIIλ1640, they should show sensitivity to the abundance of Carbon and Silicon, assuming that the ratio of Helium relative to Hydrogen remains constant.However, the dependence of CIV/HeIIλ1640 on Z is not monotonic: it increases for sub-solar metallicities and then declines steadily up to 200Z ⊙ , for specific conditions with log U ∼ 0 and logn H ∼ 9cm −3 [22].For lower U values, the behavior is monotonic [29].This underscores the necessity for multiple intensity ratios that depend on Z and U.
Z indicators involving NV, NV/CIV and NV/HeIIλ1640 have also been extensively employed in previous studies [12,13,16,61,62].The strength of the NV line was unexpectedly high in a photoionization scenario, possibly due to a selective enhancement of nitrogen [e.g., 63,64], resulting from secondary production of N by massive and intermediate-mass stars, and yielding [N/H]∝ Z 2 [15,65,66].This process may be particularly significant in cases of abnormal star formation and evolution processes that are expected to occur within active nuclei.Contamination by narrow and semibroad absorption features is often significant, and even with precise modeling of high-ionization lines, it may be challenging to reconstruct the unabsorbed profile of the red wing of Lyα.In this analysis, we refrain from using ratios involving NV because they are not consistently measurable.
Ionization parameter SiIII]/SiIV, SiIIλ1814/SiIII], and SiIIλ1814/SiIV are influenced by the ionization parameter and remain insensitive to changes in Z since the lines are from different ionic states of the same element.Also, the ratio CIII]/CIV is sensitive to the ionization parameter but entails a strong dependence on the n H as well.The ratio CIV/Hβ is also a clear diagnostic, although it is also dependent on Z and, unfortunately, often made unreliable by the intrinsic variations of the quasar and by poor photometric accuracy if observations are not synoptic and dedicated.

Mixed diagnostics: FeII/Hβ
The ratio R FeII deserves a particular attention.As with any other metal to Hydrogen ratio, it entails an obvious dependence on iron abundance and hence on metallicity.Nonetheless, R FeII is dependent on density, ionization parameter and column density of the line emitting gas, in the sense that large R FeII (≳ 1) seem possible only for relatively high n H (≳ 10 11 cm −3 ), low ionization and large N c (≳ 10 23 cm −2 ) [7,67,68].
In the following, we will try to use the same ratios as much as possible for the threeline components.However, the BLUE components are often so weak to be undetectable in several LIL profiles.In this case, we consider upper or lower limits as appropriate.Table 1 provides an overview of the measured intensity ratios applied to the metallicity estimates of several of our targets.
d measured ratio with associated uncertainty.< and >: upper and lower limit to intensity ratio, respectively.-: not available.
He II UV and He II opt are HeIIλ1640 and He IIλ4686, respectively.Blue circles identify the ratios actually used for the new sources presented in this work; the use of the CIV/Hβ ratio has been considered only for Fairall 9, due to the non-contemporaneity of the rest-frame optical and UV data.Unavailable ratios involve two undetectable components.

Photoionization simulations
Understanding the physical conditions within quasar environments involves estimating parameters like metallicity (Z), density (n H ), and ionization state (ionization parameter U).The three fundamental parameters can be estimated by comparing observed line ratios with model predictions obtained through computational models, such as CLOUDY, which simulate the interactions between radiation and gas in the environments of the broad line region [69].Input parameters for photoionization computations are the photoionizing continuum spectral energy distribution (SED), the ionization parameter (or an alternative, luminosity, and distance of the emitting region), gas hydrogen density, chemical composition, and a micro-turbulence parameter.There is evidence of trends for all of these parameters along the quasar main sequence, and in the case of Population B, there is evidence of a radial stratification of the properties within the BLR [70,71] that is heuristically modeled separating a BC and VBC (BLR and VBLR).
The arrays of simulations were therefore organized as follows: 5 different SEDs, one for each of the following cases: Pop.B RL, Pop.B RQ, with a dedicated SED for NGC 1275, an SED for Pop.A sources [72] and one for extreme Pop.A [high Eddington ratio of Ref. 73].Metallicity (Z) was assumed to scale as solar (Z ⊙ ), with 12 values ranging between 0.01 and 1000 Z ⊙ for Pop.A and 14 values between 0.001 and 20 Z ⊙ for Pop.B. The micro-turbulence parameter was set to 0 km s −1 .This is relatively insignificant for resonance UV lines [22], but is expected to lead to an under-prediction of FeII emission [22,74,75].For each metallicity value, we considered an array of simulations covering the n H and U parameter plane in the range 7 ≤ logn H ≤ 14 cm −3 , −4.5 ≤ log U ≤ 1 for Pop.A (667 simulations), and 7 ≤ logn H ≤ 13 cm −3 , −3 ≤ log U ≤ 1 for Pop.B (425 simulations).For each source or composite spectra, the set of ≈ 8 − 9 diagnostic ratios were compared with a set of ≈ 8000 and 6000 simulations covering the parameter space n H , U, and Z, for Population A and B respectively.The computations were carried out independently for the three components identified in the emission lines, as they are thought to represent distinct regions in different physical conditions.The solution for the single zone model (i.e., a single point in the 3D space n H , U, Z) was identified by the minimum χ 2 computed from the difference between the observed line ratios and the predicted line ratios over the entire 3D space [see, e.g., 22,29].
Errors on measured line ratios were estimated assuming that the continuum placement was the dominant source of uncertainties and setting extreme continua as cont ± rms, where cont is the best-fit continuum and rms is the noise measured over the continuum itself and propagated according to the "triangular distribution" [76].Limits at 1 σ and 90% confidence were set by computing the ratio F = χ 2 /χ 2 min between χ 2 of different models for n ratios − 3 degrees of freedom.
There are several caveats in the method, related to both the quality of the data and the model assumption: (1) the non-simultaneity of the observations in the optical and UV.Often, optical and UV observations are separated by years in the rest frame of the quasar.In addition, photometric inter-calibration between optical and UV data is problematic: while space-based observations are precise within a few percent, optical data are affected by uncontrolled light loss.As a result, the ratio CIV/Hβ was measured but ultimately removed from the computations in the new sources analyzed in the present paper (Mark 335, Mark 478, PHL 1092) except for Fairall 9.In this case, simulations both with and without the ratio CIV/Hβ were run and gave consistent results.(2) A major assumption is that metallicity scales as solar.Albeit this is a time-honored assumption [77,78], it is not a reasonable one because major differences are expected for a disk star in a late-type spiral galaxy and the nuclear region of an active galaxy [79, and references therein].Actually, recent works discussed the evidence of pollution by supernova ejecta [29,80].(3) Photoionization computations are carried out under the assumption of single-zone emission.While this assumption seems a good one for extreme Population A, where density and ionization tend toward limiting values [7], this might not be the case for Pop.B sources.

Case studies
Basic properties of the cases considered in this paper are reported in Table 2, and include the E1 optical parameter FWHM Hβ and R FeII , as well as the accretion parameters (luminosity, black hole mass M BH , and Eddington ratio).The last columns provide the radio loudness parameter and some notes that list bibliographical sources of information or notable, recent work related to the specific object.
Composite RL -ST B1 -A RL composite spectrum was obtained from 20 RL sources, with redshift range ≈0.25 -0.65, absolute magnitudes between −23.5 and −26.5 and S/N ≈130 and ≈55 for the visual and UV ranges.The Z estimates are used as provided by Marziani et al. [21] since the method of analysis and measurement is basically the same.
NGC 1275 -NGC 1275 (Perseus A) is an elliptical galaxy, and the brightest cluster galaxy of the Perseus Cluster, one of the most massive galaxy clusters in the nearby Universe.NGC 1275 is associated with a cooling flow phenomenon, where gas in the cluster's intracluster medium is thought to cool and sink towards the central regions of the galaxy [81,82], potentially fueling its modest AGN activity.The photoionizing continuum, which is crucial for understanding the ionization state of the gas in NGC 1275, was defined ad hoc by observational constraints on the SED [see 27, for details].
The estimated photoionizing continuum had a spectral index α ν = 0.5 in the far-UV range (500 to 800 Å) and a spectral break at 800 Å.Beyond this spectral break, the spectral index α ν = 2 down to the soft X-ray band at 0.5 keV, with α ν = 1.0 in the harder X-ray band.In summary, the SED of the far-UV shown by Punsly et al. [27] indicates a broad line Seyfert-like AGN with a soft ionizing continuum, a weak hard ionizing continuum, and no Compton deflection hump.The big blue bump associated with thermal emission from optically thick, geometrically thin accretion [83] is weaker than the ones of the SED templates appropriate for Population B quasars, see Fig. 5 of Punsly et al. [27].The L/L Edd is extremely low and, interestingly, the BLR is correspondingly weak, to the point that a careful, dedicated analysis was needed to disentangle the broad line profiles from the much stronger narrow line emission.
Composite RQ -ST B1 -A composite spectrum was constructed for a group of 16 RQ sources, all falling within the spectral bin B1, within the redshift range of approximately redshift range ≈0.002-0.5.These sources span an absolute magnitude range −21 -−27, which corresponds to a bolometric luminosity range log L ∼ 45-47 [erg s −1 ].To achieve comprehensive UV coverage from 1000 Å to around 6000 Å (including the spectral region from Lyα to Hβ), which is typically demanding and necessitates space-based observations, the UV data were obtained from HST/FOS observations as discussed in Sulentic et al. [84].Additionally, the optical spectra were sourced from [26].The composite spectrum has a high signal-to-noise ratio (S/N) of approximately 90.The metallicities (Z values) for this composite and their corresponding uncertainties were adopted from [21].
Fairall 9 -ST B1 -Empirical parameters derived from the Main Sequence (MS) analysis indicate Fairall 9's spectral type as B1, a category well-populated among quasars along the MS.This classification consistently aligns with a low L/L Edd ratio.Notably, Fairall 9 exhibits radio quietness, as it eluded detection in the Sydney University Molonglo Sky Survey [85] with a detection limit of 6 mJy, implying a radio-to-optical specific flux ratio of ≲ 1.5.We use log R K ≈ −0.04 [86].
The most recent assessment of the black hole mass includes estimates using the reverberation mapping technique that have converged on values ranging from (1.5 -2.5) ×10 8 M ⊙ [87,88], contingent on the adopted virial factor, as well as a spectropolarimetric-derived M BH that allowed for a virial factor estimate, yields (1.5 ± 0.5) • 10 8 M ⊙ .A conventional estimate of Fairall 9's bolometric luminosity stands at log L bol ≈ 45.3 erg s −1 , with an Eddington ratio of log L/L Edd ≈ -2.0, placing it toward the lower end within the distribution of Population B sources.The Spectral Energy Distribution (SED) also conforms to the characteristics of Population B objects, devoid of a prominent big-blue bump.More details are given in a recent paper by Jiang et al. [89].
Mrk 335 -ST A1 -Markarian 335 is a population A Seyfert 1, spectral type A1.It is located in the nearby Universe with a redshift of 0.0256.This AGN exhibits characteristics typical of an RQ A1 AGN, with lower-than-average FeII emissions, positioning it in the lower-left corner of the MS.Emission lines in the UV and optical ranges exhibit little to no blueshifts in their profiles.The only exception to this typical behavior is the positive slope of its optical continuum, which is likely due to galactic extinction.
Mrk 478 -ST A2 -Markarian 478 is a Pop.A Seyfert 1, spectral type A2, borderline A3 from the measurements of the present analysis.It is located at a redshift of 0.077.Although classified as A2, it exhibits characteristics that suggest it may be an extreme accretor.Table 2 shows that it has an Eddington ratio ≈ 1.Additionally, it displays a strong FeII emission and a pronounced outflowing component in its emission line profiles.Hence, it could be argued that Mrk 478 could be classified as an A3-type object.
PHL 1092 -ST A4 -Palomar Haro Luyten 1092 is a population A Seyfert 1, spectral type A4, located at a redshift of 0.3965.Its spectrum is characterized by strong UV emissions, with a notably sloped SED.PHL 1092 is considered an extremely accreting quasar, as it exhibits prime characteristics of one, including a strong outflowing component in its emission line profiles, particularly noticeable in CIVλ1549 and SiIV+OIV]λ1402, which casts a shadow on the virialized component.In the optical range, the Hβ emission is overshadowed by the FeII emission, mirroring the FeII profile of I Zw 1, itself considered an extreme accretor.
Extreme Population A (A3, A4) -The intermediate redshift xA sample of Śniegowska et al. [22] and Garnica et al. [29] allowed for Z estimates from the UV spectral lines.
The sample lacks the optical data providing the important information from the Hβ spectral range, and the line width reported in Table 2 is the one of AlIII.The sources were selected based on UV criteria that were found equivalent to the criterion R FeII ≳ 1 for the identification of extreme Population A sources, at least at a high degree of confidence [90].Their luminosity is significantly higher than the luminosity of the other sources considered in this paper, although the Eddington ratio is consistent with the ones of Mark 478 and PHL 1092, ∼ O(1).The HILs in the rest-frame ultraviolet show an excess with respect to the BC+VBC extending on the blue side of the line, modeled with a skew Gaussian [91] and ascribed to emission from a spectroscopically resolved outflow that adds up to the virialized components.The magenta lines trace the full empirical model of the line profiles.The bottom panels show the (observed minus computed) residuals.

Line profile analysis
The results of the line profile analysis are shown, as an example, for the case of Fairall 9 (Fig. 2).All the lines fit with the three components introduced in Section 3.1.HeIIλ1640 and CIV are explained assuming that the components are present with consistent shifts and widths but changing their relative intensity ratios.This accounts for the flat HeIIλ1640 profile that lacks the prominent BC observed for CIV, in turn implying CIV/HeIIλ1640 ≫1 for the BC.The VBC CIV/HeIIλ1640 is much lower.A similar effect is visible for Hβ and HeIIλ4686 in the rightmost panel of Fig. 2.This has implications on the physical conditions derived for the VBLR and BLR, in turn motivating the model with two separate components, as discussed in several works [98][99][100].A second important implication is that the VBC of HeIIλ4686 and HeIIλ1640 is much better defined than the BC, allowing for a more reliable estimate of the VBC ratios involving these lines.
Similar line profile decomposition has been carried out for quasar emission lines over a broad range of redshift [41,101,102], although the heuristic technique applied in this and    7) Mass of the supermassive black hole (M BH ) of the AGN, calculated using the scaling relation from [96].(8) Ratio between the bolometric luminosity of the object and the Eddington luminosity: , where f Radio is the specific flux at a wavelength of 6 cm (5 GHz) and f B is the specific flux at 4400 Å (680 THz) in the B band. a : not detected, only a broad upper limit is estimated.b : Hβ spectral range not covered: R FeII not available.The FWHM refers to the best proximate of Hβ in the UV, the AlIII doublet.some previous papers allows to consistently fit all quasars emission lines with only three components.The approach is equivalent to measuring profile intensity ratios [29,103], with the advantage that the absorptions that are frequently found in high-redshift quasar spectra can be easily compensated.Intensity ratios for the three components are found consistent with the ones of previous work.

Estimation of Metallicity for the Virialized Emitting Region
Four of the case studies are new results on individual objects, and the metallicity values and the associated uncertainties at 1σ confidence level are reported in Table 3, along with estimates from previously published studies.Figs. 3 and 4 show the interval of confidence at 1σ level for the new cases in the planes Z vs. ionization parameter and density n H .
The case of Fairall 9 includes the best Z derived for the VLBR.The χ 2 is lower even if the number of degrees of freedom is lower, which allows for a much more restricted range in the parameter space than for the case of the BLR ratios.In this case, the parameters are rather loosely constrained, although the agreement between the minimum χ 2 derived for the BC and the one of the VBC, reinforces a metallicity estimate around Z ≈ 1 − 2Z ⊙ , as the two regions are expected to have the same Z.In the Fairall 9 case, the consideration of the ratio CIV/Hβ (an important parameter connected, in addition to Z, also to U), confirms the estimate Z ≈ 1 − 2Z ⊙ , for both the BLR and VBLR.
Fig. 4 shows the planes Z vs. ionization parameter and density n H ordered along the sequence of increasing R FeII within Pop. A. The Z values range from ≲ 0.1 Z ⊙ to 50 Z ⊙ from the B1 RL composite to PHL 1092.The Z value obtained for PHL 1092 confirms the high Z ∼ 100 Z ⊙ obtained with UV intensity ratios only [22,29].
Fig. 5 shows the location of all case studies along the E1 main sequence.The Z values along the main sequence range from 0.1Z ⊙ to ∼ 100 Z ⊙ , and the trend is one of a monotonic increase along the horizontal sequence of increasing R FeII , from solar or slightly subsolar, to highly supersolar, with Z at least a few tens the solar values.Spectral type B1 and A1 consistently show similar values around solar, with the weak but still significant FeII emission.
An important result is the realization that not all BLRs are made of gas with the same metal content.There is apparently a systematic gradient involving a range of more than a factor ∼ 100.No matter the exact values of the Z, especially at the extremes, the trend is substantiated by the trend in the most metal-sensitive ratios, (SiIV+OIV]λ1402)/CIV, and R FeII , and ratios involving AlIII.The range in Z can be compared with a recent systematic study for intermediate redshift quasars [20].Highest values around Z ∼ 20Z ⊙ are found for massive black holes (logM BH ∼ 9.7 [M ⊙ ]), and are comparable with the value we obtain at low-z for high Eddington ratio sources.
At intermediate redshift, however, the BLR Z remains always highly supersolar, ≳ 5Z ⊙ , much above the most metal-rich galaxies (Z ∼ 2Z ⊙ ) [78].Quasars in the local Universe with modest masses (≲ 10 8 [M ⊙ ]) radiating at low L/L Edd are not yet sampled in major surveys, and there is therefore no disagreement if low-Z sources are missing at intermediate redshift.

Outflows Traced by the blueshifted components
Gas outflows appear to be a phenomenon shared by the vast majority of AGNs [e.g., [105][106][107][108][109]. Emission lines like CIVλ1549 provide valuable information about massive, ionized outflows associated with the accretion disk, and in turn, contribute to understanding the dynamic processes within the BLR and its interplay with the accretion disk.
Selection of the lines most suitable for plasma diagnostics, including the metallicity, is much harder for the outflow component than for the virialized one, as the first appears spectroscopically resolved in high-and low-ionization lines only for extreme Population A or at very high luminosity.We used four diagnostic ratios for estimating the physical properties of the blue, outflowing component: CIV/HeIIλ1640, CIV/(OIV]λ1402 + SiIV), CIV/Hβ and (OIV]λ1402+SiIV)/HeIIλ1640, along with constraints from upper limits for the ratios AlIII/CIV, FeII/Hβ.The diagnostic ratios are consistent with the estimates of metallicity derived for the virialized components, for PHL 1092 (Z ∼ 50Z ⊙ , with a 1σ confidence range 20 -50 Z ⊙ ).Median values of the BLUE components are ∼ 6Z ⊙ for the xA intermediate z samples, somewhat lower and obtained only with three diagnostic ratios (Hβ was not measured).Note that this abundance estimate is obtained using ratios involving only α-elements.Other properties of the outflowing component are less wellconstrained because of the limited number of diagnostics.A better precision might be achieved with higher S/N and additional diagnostic ratios.However, it is reasonable to assume that the outflowing gas from the high Eddington ratio sources might have Z ∼ 10Z ⊙ , a conservative estimate in agreement with other works [20, and references therein].If this is the case, the mass outflow rate is Ṁ ∼ 10 3 L CIV,45 , n −1 H,9 , where L CIV,45 erg s −1 is the CIV luminosity in units of 10 45 erg s −1 and n H,9 is the Hydrogen density in units of 10 9 cm −3 [38].Solar metallicity implies that 1.46% of the mass of the gas is due to metals.For a modest luminosity of 10 44 erg s −1 , the implication is that about ∼ 15 M ⊙ /yr of metals are returned to the interstellar medium (ISM).Over a high-accretion lifetime of ∼ 10 7 yr [110,111], the metal mass expected to be returned to the ISM could be ∼ 10 8 M ⊙ .While this estimate is extremely coarse, and the actual effect will depend on how the outflow is dissipated within the host, it implies that there could be in principle a significant enrichment for a large stellar population and earlier for the ISM.
We have focused on low-luminosity quasars in the local Universe, characterized by relatively small black holes radiating near their Eddington limit.However, it's important to note that the outflow phenomena are most prominent in the highest luminosity quasars.These super-luminous quasars exhibit a high prevalence of significant blueshifts in the blueshifts in the CIVλ1549 and [OIII]λλ4959,5007 emission line profiles [39,40,112].The ionized gas mass, kinetic energy, and mechanical thrust in these cases are remarkably high, implying extensive feedback effects on the host galaxies of these exceptionally luminous quasars.These effects were particularly pronounced during cosmic epochs between 2 and 6 billion years after the Big Bang, suggesting that these quasars might have played a substantial role in enriching the chemical composition of their host galaxies.

A gradient in metal content and chemical feedback along the sequence
A cartoon depicting a global evolutionary scenario is shown in Fig. 6.At the one end of the sequence we encounter low mass, high Eddington ratio sources.They are accreting at a high rate, possibly following a merger and a burst of star formation.In the initial phases of the development of AGN and quasars, a series of events take place: wet mergers and strong interactions cause the accumulation of gas in the central regions of the galaxy.This accumulation triggers a burst of star formation (top of the inset) [113,114].Over time, mass loss from stellar winds and supernova explosions provide a source of accretion fuel for the massive black hole located at the center of the galaxy.Subsequently, radiation force and mechanical energy can clear away the dust surrounding the black hole, particularly within At a high accretion rate, a nuclear starburst can occur in the region where the disk becomes self-gravitating [115].As mentioned, the Z-values calculated for the BLR of quasars appear exceptionally high when compared to their host galaxies.For reference, the highest measured Z-value in a molecular cloud is approximately 5 times the solar metallicity [78].However, it's important to note that the nuclear and circumnuclear environments of quasars may exhibit significant deviations from a typical interstellar setting.In these regions, stars traverse the disk, giving rise to the formation of accretion-modified objects that eventually attain substantial mass and, after a brief evolutionary phase, explode as core-collapse supernovae [115,116].Stars within the nuclear vicinity can rapidly become highly massive (with masses exceeding 100 times that of the Sun), leading to core-collapse events that contribute to enriching the disk with heavy elements via the substantial metal yields produced by supernova ejecta [80].Furthermore, the compact remnants of these stars may continue accreting material, resulting in recurrent supernova occurrences [117].These accretion-modified star formation processes that enhance metallicity are projected to yield metal abundances approximately in the range of 10 to 20 times solar metallicity [44], which aligns with the values observed in the xA sources.
At the other end, we find very massive black holes, radiating at low Eddington ratios, in conditions that are proximate to "starvation" and in any case to the exhaustion of the reservoir of gas for accretion.Accretion material may come from evolved star winds that can sustain modeled accretion rates [118].Metallicity values might be in this case more conventional, around solar or subsolar, such as the values ranging from ∼0.1 to ≈ 2 times the solar metallicity found in the bulge of the Milky Way [119,120].
Pop. B sources have a slightly larger redshift than Pop.A (z ≈ 0.33 vs. 0.07) in a large, SDSS-based, low-z sample [4] that might be easily explained by selection effects.However, a ten-fold increase in black hole mass from 10 8 M ⊙ at moderate accretion rate can occur over a time ∼ 5 • 10 8 yr [10].This timescale would correspond to a change in redshift δz ≈ 0.05.Therefore, it is reasonable to assume that at least some Pop.B AGN evolved from Pop.A sources locally.

Conclusion
In conclusion, the main sequence concept in quasars reveals a gradient of metallicity across different quasar populations, shedding light on their evolutionary paths.The understanding of these populations and their associated characteristics contributes to a deeper comprehension of the processes driving quasar behavior and their influence on their host galaxies.Expectations from accretion-modified stars within the active nuclei are consistent with highly supersolar metallicity.

Figure 1 .
Figure 1.Interpretation of the line profiles of low-and high-ionization lines along the MS for isolating major spectroscopically resolved components.

Figure 2 .
Figure 2. Fits of the emission line spectrum of Fairall 9, the prototypical Population B source.From left to right, top row: 1400 Å blend; CIV + HeIIλ1640 blend; bottom row: 1900 Å blend; Hβ + FeII blend.The line profile of Hβ and the 1900 blend are accounted for by two components, BC (black lines) and VBC (red).The HILs in the rest-frame ultraviolet show an excess with respect to the BC+VBC extending on the blue side of the line, modeled with a skew Gaussian[91] and ascribed to emission from a spectroscopically resolved outflow that adds up to the virialized components.The magenta lines trace the full empirical model of the line profiles.The bottom panels show the (observed minus computed) residuals.

Figure 3 .
Figure 3. Projections of the 3D parameter space (U, n H , Z) onto the (U, Z) (left) and (n H , Z) planes (right), for the prototypical Pop.B source Fairall 9.The top and bottom panels are for χ 2 computed with and without the CIV/Hβ ratio.The blue spot identifies the model yielding minimum χ 2 for the BC; the red one is the same for the VBC.Isophotal contours are a 1σ (pale blue) and 90% confidence level (yellow); the red line is the 1σ level for the VBC.

Figure 4 .
Figure 4. Projections of the 3D parameter space (U, n H , Z) onto the (U, Z) (top) and (n H , Z) planes (bottom), for the Pop.A sources -Mark 335 (left), Mark 478 (middle), and PHL 1092 (right).The meaning of panels is the same as for the BC in Fig. 3.

Figure 5 .
Figure 5. Sketch of the quasar main sequence at low redshift, with circles of different sizes representing the metallicity estimates for different sources, composites, and samples presented in this paper and analyzed in recent literature.The abscissa is FeII prominence parameter R FeII ; ordinate is the FWHM Hβ in km s −1 .The numbers in square brackets report the prevalence of each spectral bin in an SDSS -based sample [104].The size of the circles depends on Z: smallest for Z ≲ 0.1Z ⊙ , intermediate for 0.5Z ⊙ ≲ Z ≲ 2Z ⊙ , larger for Z ∼ 10Z ⊙ , and largest for Z ∼ 50Z ⊙ .The blue circle refers to the intermediate z sample of Garnica et al. [29].

Figure 6 .
Figure 6.Sketch depicting the evolutionary interpretation of the quasar main sequence.The inset on the right shows a possible evolutionary path leading to highly accreting quasar showing evidence of high metal enrichment in the BLR.

Table 1 .
Diagnostic intensity ratios

Table 3 .
Metallicity estimates along the quasar MS