Boroson and Green conducted their study within the low-redshift range (z < 0.5), analyzing 87 sources from the Bright Quasar Survey (Schmidt and Green, 1983). Their primary finding from optical spectra analysis of these quasars was that the Fe ii line equivalent width consistently matched that of H β , indicating that Fe ii emission originates from the same broad-line region (BLR) clouds as H β . Additionally, they compiled optical spectral properties for each source. They integrated these with data from other spectral regions from previous studies, creating a 17-parameter correlation matrix for emission lines and continuum properties. Using principal component analysis (PCA, Francis and Wills, 1999) on this matrix to identify meaningful correlations, they focused on 13 key properties including M V (V-band magnitude), log R (radio-to-optical spectral index, Kellermann et al., 1989), α o x (optical-to-X-ray spectral index, Tananbaum et al., 1986), EW(H β ), [O iii] λ 5,007 strength, He ii λ 4,686 strength, Fe ii (4434-4684 Å) strength (see also left panel of Figure 1), [O iii] λ 5,007/H β peak height ratio, FWHM(H β ), H β profile shift, shape, asymmetry, and M [ O i i i ] . They determined that the primary eigenvector (Eigenvector 1 or EV1) derived from these properties was primarily characterized by the anti-correlation between Fe ii strength (specifically R Fe II , the ratio of EW(Fe ii) for the 4434-4684 Å blend to EW(H β ) and [O iii] λ 5,007 strength. EV1 also showed significant correlations ( > | 0.5 | ) with log R , FWHM(H β ), and H β profile asymmetry.

To interpret the PCA results, the authors identified several key parameters that could influence the observed properties: (1) the mass accretion rate, (2) the black hole (BH) mass, (3) the covering factor of the BLR clouds, (4) the degree of anisotropy in the emitted radiation from the continuum source, (5) the orientation of the source to the observer, (6) the velocity distribution of the BLR clouds, and (7) the ionization parameter.

To summarize, the paper by Boroson and Green (1992) is fundamental for two main reasons:

An advanced version of the Eigenvector 1 (EV1) schema was introduced by Sulentic et al. (2000), incorporating additional parameters beyond the initial (1) FWHM of the broad component of H β and (2) the equivalent width ratio of the optical Fe ii blend (4,434–4,684 Å) to broad H β ( R Fe II ). These new parameters include (3) the centroid shift at FWHM of the high ionization line C IV λ 1,549, c(1/2), and (4) the soft X-ray photon index ( Γ soft ) . Simplified, these measures: (1) the extent of virialized motions in a low-ionization line-emitting accretion disk or a flattened cloud distribution, acting as a virial estimator of black hole mass (e.g., Collin-Souffrin et al., 1988; Dultzin-Hacyan et al., 1999; Joly et al., 2008) (2) the ionization and size of the BLR cloud, with the Fe ii emission strength ( R Fe II ) suggesting its origin near the accretion disk; (3) indicators of winds/outflows in high ionization broad line gas; and (4) thermal emission related to the accretion disk and SMBH accretion state (e.g., Mineshige et al., 2000; Done et al., 2012). We refer the readers to a comprehensive summary in Marziani et al. (2018).

Accumulating evidence from subsequent studies, following Boroson and Green (1992), indicates that EV1 correlations involve at least two principal independent parameters: (1) the source’s bolometric luminosity ( L bol ) and its black hole mass ( M BH ), convolved with source orientation (Marziani et al., 2001; Panda et al., 2019c). These two parameters are succinctly expressed as the Eddington ratio ( L bol / L Edd ).

The onset of the new century saw the rise of large spectroscopic surveys, such as the Sloan Digital Sky Survey (SDSS, York et al., 2000; Shen et al., 2011). These surveys revitalized EV1 studies and extended their applicability to much larger samples. Shen and Ho (2014) made significant strides with their seminal paper, utilizing data from over 20,000 spectroscopically observed SDSS quasars, analyzed using an automated spectral fitting pipeline (Shen et al., 2011). This study provided spectral parameters for a wide range of emission lines, as well as estimates for black hole masses and Eddington ratios. Leveraging this comprehensive dataset, Shen and Ho redefined the main sequence of quasars and concluded that (1) the average Eddington ratio increases from left to right on the sequence, and (2) the dispersion in FWHM(H β ) at a fixed R Fe II is largely due to orientation effects (see also Sun and Shen, 2015). They proposed that quasar properties correlated with EV1 can be unified by variations in the average Eddington ratio of the accreting black hole, driven by systematic changes in the shape of the accretion disk continuum and its role in photoionizing the line-emitting regions.

More recently, the exploration of large samples has been extended to include sources from the Southern Hemisphere (Chen et al., 2018, and references therein), and the number of Type-1 AGNs, including strong Fe ii-emitting ones, has grown many folds with deeper surveys extending to fainter magnitudes (Rakshit et al., 2020; Wu and Shen, 2022; Paliya et al., 2024; Panda et al., 2024a). We are now at a stage where AGNs are frequently revisited and thus we also have a wealth of multi-epoch, multi-wavelength data for samples of AGNs. This has greatly helped to build samples of AGNs that demonstrate changes in their continuum and emission line properties - the Changing-Look AGNs (see recent compilations in Panda and Śniegowska, 2024; Guo et al., 2024; Zeltyn et al., 2024, and references therein), especially investigating the changes in the Fe ii emission in the context of the quasar main sequence (Panda and Śniegowska, 2024).

The paper is organized as follows: In Section 2, we highlight some recent advances in the last decade on the studies with the quasar main sequence - Fe ii template creation and improvements, theoretical predictions, and advancements in photoionization modeling including some direct confirmations of long-standing hypotheses. We discuss the connection of the main sequence with one of the fundamental properties demonstrated by AGNs - Variability in Section 3, especially in advancing our knowledge through techniques like reverberation mapping (RM), and the renewed interest in Changing-look/Changing-state AGNs. We then give a brief account of the present-day scenario of incorporating AGNs (and quasars) as standard(izable) candles and touch upon some relevant studies that have progressed in this direction. Finally, we conclude this mini-review with some closing remarks and perspective for the future in Section 4 with up-and-coming massive, multiplex surveys that will make things more intriguing.

Owing to its complexity and uncertainties in transition probabilities and excitation mechanisms, the most successful approach to model the Fe ii emission in AGNs consists of deriving empirical templates from observations and supplementing the missing transitions with state-of-the-art radiative transfer models, e.g., CLOUDY (Ferland et al., 2017; Chatzikos et al., 2023). The templates thus derived using this methodology are referred to as semi-empirical. The work of Kovačević et al. (2010) is seminal in this regard who provided the AGN community with an interface ¹ to create Fe ii templates by combining the Fe ii transitions from theoretical expectations and those revealed in the spectrum of the prototypical Fe ii-emitter, I Zw 1. Their methodology involves the knowledge of the temperature of the ionized cloud responsible for the Fe ii emission, information on the dynamics of the Fe ii profile in the observed spectrum, and intensities of the strongest Fe ii multiplets collected in three groups ( b 4 F, a 6 S, and a 4 G). Their semi-empirical templates have been applied to large samples of AGNs, and demonstrated to provide convincing results. All except one went back to the I Zw 1 - in the optical (e.g., Véron-Cetty et al., 2004; Marziani et al., 2021a), in the UV (e.g., Vestergaard and Wilkes, 2001; Bruhweiler and Verner, 2008; Tsuzuki et al., 2006). The exception happened more recently with the HST/STIS observations for Mrk 493 (Park et al., 2022) which has narrower lines, lower reddening, and a less extreme Eddington ratio value than I Zw 1, therefore, can be applied to a larger population of Type-1 AGNs with intrinsically lower Fe ii emission (Shen and Ho, 2014; Panda et al., 2018). These templates also allow us to infer the velocity information of the Fe ii emission, a key aspect constraining the geometry and kinematics of the Fe ii emitting region in the BLR. Studies (Hu et al., 2008; Ferland et al., 2009; Kovacevic-Dojcinovic and Popovic, 2015) have suggested that the Fe ii emission originates from a location different from, and most likely exterior to, the region that produces most of H β . These observational findings were confirmed through the analysis of emissivity profiles of AGNs using photoionization modeling (Panda et al., 2018; Sniegowska et al., 2020) and studies of Fe ii time-lags relative to the H β in samples of AGNs using reverberation mapping (Barth et al., 2013; Gaskell et al., 2022).

In recent years, there has been noteworthy development to improve the atomic datasets available for iron emission, with updated radiative and electron collisional rates, and include higher levels (up to 716) and energies as high as 26.4 eV. We refer the readers to Sarkar et al. (2021) for an overview of these datasets and their performance within the spectral synthesis code, CLOUDY.

Over the years after Boroson and Green put forward their findings from the Eigenvector 1, the expected parameters that should influence the observed correlation in the quasar main sequence have been looked at, albeit separately. Notable among them are (i) the Fe ii emission model developed in Verner et al. (1999) with 371 atomic levels producing 13,157 (permitted) emission lines with the highest energy level of ∼ 11.6 eV; (ii) study by Baldwin et al. (2004) which were among the first to suggest the importance of microturbulence ( ≳ 100 km s − 1 )- the intra-cloud pressure broadening in the broad-line emitting region (BELR) clouds, to explain both the observed shape and equivalent width of the Fe ii emission. We note that the findings in Baldwin et al. were confined to the UV regime to recover the 2,200–2,800 Å Fe ii bump feature. They also suggested the need to include higher metal abundances in the BELR clouds to recover the observed Fe ii intensities, confirming the earlier results from observations by, e.g., Hamann and Ferland (1993), (1999); Dietrich et al. (2003); (iii) the need to have higher mean densities and column densities in the BELR clouds via numerical modeling to recover the Fe ii pseudocontinuum behavior, still in the UV regime, was suggested in the paper by Bruhweiler and Verner, who re-affirmed the importance of the microturbulence in the BELR.

In more recent years, a clearer picture of the Fe ii emission, especially in the optical region, linking to the quasar main sequence has been achieved. There is a growing consensus that the main sequence of quasars, earlier thought to be primarily driven by the Eddington ratio, is in reality, dependent on a combination of parameters of the underlying accretion disk and the BELR clouds Panda et al. (2018), 2019a,c. These parameters are: (1) Eddington ratio, (2) BH mass; (3) shape of the ionizing continuum (SED); (4) BLR density; (5) BLR metallicity; (6) velocity distribution of the BLR clouds (including microturbulence); (7) source’s orientation; and (8) BLR cloud sizes (see Panda, 2021a). The 8-dimensional parameter space was first presented by Panda et al. (2019c), and extended by Panda et al. (2020b) wherein through large grids of photoionization models with CLOUDY and massive observational spectroscopic catalogs (Shen et al., 2011; Rakshit et al., 2020) the inherent trends along the main sequence have been confirmed. This almost completes the circle initiated with the hypotheses in Boroson and Green (1992) although more progress is needed, from observational and theoretical aspects. This multi-dimensional parameterization includes the viewing angle to the source (or orientation), which is constrained for a small fraction of the AGNs, especially those that show strong radio “jetted” emissions (see, e.g., Padovani et al., 2017, for an overview) or strong water masers (Neufeld et al., 1994; Greenhill et al., 2003). For the remaining sources, the viewing angle is estimated indirectly - through dynamical modeling (Pancoast et al., 2011; Li et al., 2013; Williams et al., 2018; Li et al., 2024), through polarization studies of the emission lines (Savić et al., 2018; Jiang et al., 2021; Śniegowska et al., 2023; Jose et al., 2024), or broad-band SED modeling (Yang et al., 2020; Martínez-Ramírez et al., 2024). The knowledge of the viewing angle is crucial since it can be combined with the spatial and velocity distribution of BELR clouds and their location from the central ionizing source, to estimate the black hole mass of the source. The methodology presented by Panda et al. (2020b) is powerful and acts in dual-purpose - for sources with known orientation and spectroscopically measured Fe ii emission, it can allow to constrain the BLR density and metallicity. On the other hand, through observed UV diagnostics if the BLR density and metallicity can be inferred (in addition to the Fe ii emission), one can recover the orientation angle of the source. The methodology, at present, includes the state-of-the-art broad-band SEDs presented in Panda et al. (2019c) and Ferland et al. (2020), while the BH mass, Eddington ratio, velocity distributions and line intensities are from the SDSS QSO catalogs (Shen et al., 2011; Rakshit et al., 2020) for observed AGNs, and can be refined with future multi-wavelength campaigns. Recent works by Pandey et al. (2023, 2024) have extended these results with new, and up-to-date Fe ii atomic datasets and accounting for dust within the BLR. Additionally, using these new datasets, Dias dos Santos et al. (2023, 2024) have probed into the Fe ii emission in the NIR regime with the added advantage of transitions being isolated and less in number relative to the optical and UV.

In another recent work (Floris et al., 2024), we performed a multi-component analysis on the strongest UV and optical emission lines and using ∼ 10 metal content diagnostic ratios that reveal a systematic progression in metallicity, ranging from sub-solar values to several times higher than solar values. This notable finding was a result of a series of papers (Śniegowska et al., 2021; Garnica et al., 2022; Marziani et al., 2024) wherein a robust recipe of estimating metallicity and other physical parameters in highly-accreting Type-1 AGNs were developed. These results confirm the theoretical predictions made by Panda et al. (2019c) where the increase in the metal content was noted as a key factor that proportionally led to an increase in the Fe ii emission along the main sequence.

There are multiple studies predating the aforementioned papers that have paved the way to our current understanding of the Fe ii emission and we recommend the readers to the detailed accounts by, e.g., Sulentic et al. (2000); Marziani et al. (2001); Zamfir et al. (2010); D’Onofrio et al. (2012); Shen and Ho (2014); Sulentic and Marziani (2015); Marziani et al. (2018); Gaskell et al. (2022); Panda and Marziani (2023a).

The shape of the ionizing continuum has been an integral part of the main sequence of quasars studies. Around the same time as Boroson and Green, researchers were already developing mean AGN SEDs (Vanden Berk et al., 2001; Richards et al., 2006), be it to distinguish the sources based on radio dichotomy (Laor et al., 1997) and more recently in Marziani et al. (2021a) or to reveal the prominence of the big blue bump feature in typical Type-1 AGNs (Mathews and Ferland, 1987; Korista et al., 1997). With the advent of large spectroscopic surveys spearheaded by SDSS (York et al., 2000; Shen et al., 2011) AGNs exhibiting stronger Fe ii emission alike I Zw 1 were being consistently discovered and led to the creation of a mean SED representing Narrow-line Seyfert 1 galaxies (Marziani and Sulentic, 2014). We now have broad-band mean SEDs grouped in Eddington ratios ranging from sub-to super-Eddington limits (Jin et al., 2012; 2017; Ferland et al., 2020). Although these mean SEDs have helped provide statistical inferences on the role of AGN SED in the main sequence trends, having broad-band SED for individual AGNs is a much more recent endeavor that has seen growth. With the increase in simultaneous observations across multiple spectral regimes and the development of self-consistent AGN SED models (Done et al., 2012; Kubota and Done, 2018; 2019; Hagen and Done, 2023), the number of individual sources with broad-band SEDs is growing at a rapid pace, especially for sources demonstrating the most intense Fe ii emission (Marinello et al., 2020; Jin et al., 2023). Another important extension in the area of SED building is the slim disk AGN SED models (Abramowicz et al., 1988; Wang et al., 2014; Panda and Marziani, 2023b) applicable to those sources accreting at or above the Eddington limit, that show signatures of strong outflows even in the low-ionization emitting regions (e.g., Rodríguez-Ardila et al., 2024).

Changing-look AGNs have been known for almost as long as the main sequence existed (see recent review by Komossa et al., 2024; Ricci and Trakhtenbrot, 2023). The spectral changes over multiple epochs have now been detected in numerous AGNs - be it extreme variability with the changes in the continuum and emission lines so strong that can be associated with external interference such as obscuration or tidal disruption events (LaMassa et al., 2015; Dodd et al., 2023; Trakhtenbrot et al., 2019), but could very well be associated with intrinsic effects such as disk transition/disk instabilities (Noda and Done, 2018; Ross et al., 2018; Sniegowska et al., 2020) although, the timescales of such events can be widely different (Czerny, 2006).

With the growing interest in finding new changing-look AGNs, the focus has been also to look for AGNs showing variations in their Fe ii emission (see, e.g., Gaskell et al., 2022; Petrushevska et al., 2023). The regular variable nature of AGNs has helped to gain insights into their emitting regions, with some sources where we have estimates of their Fe ii-emitting locations (see, e.g., Hu et al., 2015; Barth et al., 2013) although there are now instances of exceptional changes in the Fe ii intensities. Panda and Śniegowska (2024) made a compilation of such sources and tracked their transition along the Eigenvector 1 schema and categorized sources that either stay within the same population (A or B, see right panel of Figure 1) or make an inter-population movement as a function of spectral epoch.

In addition to Changing-look AGNs, dedicated spectro-photometric monitoring campaigns on individual sources (e.g., Mrk 6, NGC 5548, NGC 4151, NGC 4051), have allowed us to re-affirm the Pronik-Chuvaev effect, i.e., the increase, albeit with a gradual saturation, in the H β emitting luminosity with increasing AGN continuum (Pronik and Chuvaev, 1972; Wang et al., 2005; Shapovalova et al., 2008; Gaskell et al., 2021) and more recently in Panda et al. (2022); 2023a. This assists in building the R-L relation for individual epochs and gaining insights into the temporal behavior of the line-emitting BLR relative to the continuum (Zu et al., 2011; Lu et al., 2022; Feng et al., 2024) although studies to reveal the temporal behavior specifically in Fe ii are needed to complement the H β behavior in these AGNs.

Another interesting revelation has been the construction of the first Fe ii UV R-L relation in Zajaček et al. (2024b). Here, in addition to improving the existing Mg ii-based R-L with 194 sources [more recent compilation in Shen et al. (2024)], we have been able to constrain the R-L behavior in UV-emitting Fe ii for 5 AGNs. The results are motivating as the slope of the R-L is in close agreement with one expected from the standard photoionization theory (i.e., = 0.5). Although it is interesting to note that this relation appears steeper than the Mg ii R-L relations such that for low-luminosity regimes, the Fe ii emitting region is closer than the Mg ii-emitting region, whereas, at higher luminosities, both relations converge and intersect. This intriguing behaviour needs more explanation which the upcoming RM campaigns may have an answer to. We note that the optical Fe ii-based R-L relation has been around for some time (see Gaskell et al., 2022, for a recent review). A recent compilation of 17 AGNs (including multiple epoch Fe ii time lag measurements) from Prince et al. (2023) reveals an R-L for the optical Fe ii with a slope close to 0.5, and the comparison with the aforementioned UV-based R-L reveals an offset by a factor of 1.8, i.e., the optical Fe ii emitting regions are located 1.8 times further out relative to the UV Fe ii regions.

Quasars, with their extragalactic origin and persistent bright nature, have long been proposed as “standardizable candles”. With the knowledge of their luminosities (with the aid of the RM and R-L relation) and independently of their fluxes from spectroscopic monitoring, we can determine the luminosity distances of these sources. With a growing number of AGNs (now ≳ 200, Zajaček et al., 2024b; Shen et al., 2024) where we have such estimates, then allows us to prepare a Hubble diagram - stretching the redshift regime to higher ranges as compared to what we can achieve with other indicators, e.g., Cepheids, Tip of the Red Giant Branch (TRGB), and Type-1a Supernovae (SNIa). We can then scrutinize the various, existing cosmological models, in addition to gauging the performance of quasars to existing indicators, allowing us to link the cosmological measurements from the early Universe (e.g., Planck Collaboration et al., 2020) to the measurements from the late Universe (e.g., Cepheids, SNIa, and TRGBs; Riess et al., 1998; Riess et al., 2019; Freedman et al., 2019).

However, the recent detection of shorter lags in R-L linked to high-accreting sources (Du et al., 2015; Grier et al., 2017; Du et al., 2016) has put the use of R-L relation into uncertainty. In Martínez-Aldama et al. (2019), we looked into the dispersion in the R-L and upon further investigation found the extent of offset of the source’s time-lag is proportional to the Eddington ratio (or more specifically its mass accretion rate). While this helped “standardize” the R-L relation, there remained a circularity problem - the mass accretion rate needs the knowledge of luminosity apriori, and the latter can be estimated assuming a cosmological model. This defeats the purpose of using quasars for cosmology and thus, requires us to find a direct observable parameter that can replace the mass accretion rate. What can be that? In Du and Wang (2019), the authors found that the Fe ii strength (or R Fe II ) is a viable alternative to the mass accretion rate (as has been noted in earlier works of Marziani et al., 2018; Panda et al., 2019c) and can correct the dispersion in the R-L. This R Fe II -dependent R-L relation has hence been tested and confirmed in other works (Panda, 2022; Panda and Marziani, 2023a). Other empirical relations notably the L X - L UV relation, are proposed as a viable alternative to the R-L relation (Risaliti and Lusso, 2015; 2019) although there are subtle differences between the two relations and their inferences. Yet another methodology has been proposed, i.e., with the aid of the existing correlation between the luminosity and the velocity distribution of the BELR (Dultzin et al., 2020; Marziani et al., 2021b), equivalent to the original formulation of the Faber-Jackson law (Faber and Jackson, 1976). We refer the readers to Panda and Marziani (2023a) for more details.

In a parallel direction, efforts to reconcile the use of quasars along with other distance indicators, e.g., SNIa, Gamma-ray bursts, Baryon Acoustic Oscillations, and temperature anisotropy across the microwave background, have been made including the R Fe II parameter for the quasar-based R-L relations (Cao et al., 2022; Khadka et al., 2023; Dainotti et al., 2023). Other systematics, such as dust extinction can contribute to and reconcile the observed difference between the R-L and L X - L UV relation (Zajaček et al., 2024a). As we enter into the discussions around Hubble-Lemaître law and the H 0 tension, a key to resolving this is better measurements of cosmological distances. VLT/GRAVITY has opened up avenues to probe the angular sizes of the BLR in nearby AGNs using high-resolution spectroastrometry (Gravity Collaboration et al., 2018; Gravity Collaboration et al., 2020; Gravity Collaboration et al., 2021; Gravity Collaboration et al., 2024). These angular sizes can be combined with the BLR linear sizes (the latter estimated using the RM technique) to give the parallax distance to these AGNs. The technique was originally conceived in Elvis and Karovska (2002) although thanks to the recent interferometric measurements by GRAVITY coupled with their long-term RM monitoring campaign, Wang et al. (2020) have been able to estimate, for the first time, a H 0 value using this joint analysis with AGNs. Ongoing improvements with GRAVITY (see, e.g., Nowak et al., 2024) will allow the compilation of a sizable sample of AGNs extending to z ∼ two where the spectroastrometric-RM (or, SARM) technique can be applied to build the Hubble diagram for quasars. This however requires the knowledge of the BLR properties that are neatly tied to the quasar main sequence which positively affect the accuracy of the estimation of the cosmological distances to these cosmic objects.