Eigenvector 1 (EV1) in Quasar Spectroscopy

• Eigenvector 1 (EV1) is the principal axis derived from PCA of quasar spectra, characterizing the anticorrelation between optical Fe II and [O III] emissions.
• It is quantified using observables like FWHM(Hβ) and the Fe II to Hβ ratio, forming a surrogate H–R diagram that organizes quasar diversity.
• EV1 underpins the classification into Population A and B, linking spectral line properties with accretion rates and outflow dynamics in AGN.

Eigenvector 1 (EV1) is the principal axis of variance in quasar and Type 1 AGN spectra identified through principal component analysis, and it is used to organize the spectroscopic diversity of unobscured active galaxies into a quasar “main sequence” (Panda et al., 2017). In its optical form, EV1 is dominated by the anticorrelation between Fe II optical emission and the [O III] line, and EV1 alone contained 30% of the total variance in the original Boroson & Green analysis as summarized in later work (Panda et al., 2017). In current usage, EV1 commonly denotes both that principal component and the observational parameter space built around it, especially the plane defined by FWHM(H$\beta$) and the Fe II strength parameter $R_{\mathrm{FeII}}$, with the broader 4D Eigenvector 1 framework extending the scheme into the UV and X-ray domains (Sulentic et al., 2015).

1. Origin and formal definition

EV1 emerged from principal component analysis of quasar spectra as the leading component behind significant correlations among measured spectral parameters (Panda et al., 2017). Its defining optical signature is the anticorrelation between optical Fe II emission and [O III] $\lambda 5007$, a relation that remains the canonical empirical core of the EV1 concept (Panda et al., 2017).

A standard proxy for EV1 in the optical is the Fe II to H$\beta$ ratio

$$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$

or, equivalently in flux form, the ratio of the optical Fe II blend to broad H$\beta$ (Panda et al., 2017). The EV1 optical plane is commonly displayed as FWHM(H$\beta$) versus $R_{\mathrm{FeII}}$, and several papers describe this diagram as analogous to the Hertzsprung–Russell diagram for stars or as a “surrogate H-R diagram for quasars” (Sulentic et al., 2015).

This usage combines two closely related ideas. First, EV1 is a statistical component derived from PCA. Second, EV1 is an empirical organizing sequence in which quasars occupy characteristic regions according to line width, Fe II prominence, [O III] behavior, high-ionization line shifts, and X-ray spectral slope (Marziani et al., 2014). This suggests that EV1 is not simply a mathematical artifact of PCA but a phenomenological framework for connecting observed diversity to accretion and line-emitting structure.

2. The 4DE1 parameter space

The 4D Eigenvector 1 framework extends the original optical EV1 by combining four observational parameters from optical, UV, and X-ray spectroscopy (Marziani et al., 2014). In this formulation, the optical plane remains central, but it is interpreted together with high-ionization kinematics and soft X-ray properties (Marziani et al., 2012).

Parameter	Definition	Diagnostic role
FWHM(H$\beta$)	Full width at half maximum of broad H$\beta$	Virial broadening estimator; BLR velocity field
$R_{\mathrm{FeII}}$0	Fe II $R_{\mathrm{FeII}}$1/H$R_{\mathrm{FeII}}$2	Fe II emission prominence; EV1 position
$R_{\mathrm{FeII}}$3(C IV $R_{\mathrm{FeII}}$4)	Centroid shift at half maximum of C IV	Wind/outflow diagnostic
$R_{\mathrm{FeII}}$5	Soft X-ray photon index	Accretion-state indicator

Within this space, the optical EV1 plane is the most widely used projection. Binning schemes such as A1–A4 and B1–B1++ are used to classify source occupation across the FWHM(H$R_{\mathrm{FeII}}$6)–$R_{\mathrm{FeII}}$7 diagram (Sulentic et al., 2015). The same framework supports low-redshift optical work and high-redshift UV work, where parameters such as C IV centroid shift and UV line ratios are used when H$R_{\mathrm{FeII}}$8 and optical Fe II are inaccessible (Marziani et al., 2014).

The significance of 4DE1 is methodological as well as descriptive. It provides a common coordinate system in which line widths, low-ionization and high-ionization emission, outflow signatures, and X-ray continuum slope can be compared without reducing quasar diversity to a single observable. This makes EV1 useful both for classification and for tests of physical drivers.

3. Population A, Population B, and the quasar main sequence

A central application of EV1 is the Population A/Population B division. Population A sources are defined by FWHM(H$R_{\mathrm{FeII}}$9) $\lambda 5007$0 km s$\lambda 5007$1, while Population B sources have FWHM(H$\lambda 5007$2) $\lambda 5007$3 km s$\lambda 5007$4 (Marziani et al., 2012). In the reviewed literature, Population A is associated with stronger Fe II emission, weaker [O III], steeper soft X-ray spectra, and more prominent outflow signatures, whereas Population B is associated with weaker Fe II, stronger [O III], and much less prominent high-ionization outflows (Sulentic et al., 2015).

This division is not only morphological. In 4DE1, Population A is generally described as the higher-$\lambda 5007$5 end of the sequence, and Population B as the lower-$\lambda 5007$6 end (Marziani et al., 2014). A critical Eddington ratio of $\lambda 5007$7–$\lambda 5007$8 has been discussed as the boundary between the two populations, possibly marking a structural transition in the accretion flow or BLR configuration (Sulentic et al., 2015). In the specific case of NGC 5548, the structural boundary was indicatively placed at $\lambda 5007$9 (Bon et al., 2018).

The outflow phenomenology is strongly organized by this population split. C IV $\beta$0 blueshift increases toward the extreme Population A bins, and [O III] blue outliers are predominantly associated with extreme Population A and NLSy1 sources (Marziani et al., 2012). By contrast, Population B is characterized by broader and more symmetric low-ionization profiles, little or no high-ionization outflow signature, and in some formulations a very broad line region component (Sulentic et al., 2015). This supports the use of EV1 as an empirical sequence linking accretion state, line formation, and outflow prominence.

4. Physical drivers and competing interpretations

The dominant physical driver of EV1 remains debated. A broad consensus in the cited literature is that EV1 is mainly related to the Eddington ratio $\beta$1, with black hole mass, orientation, composition, and related parameters contributing secondary effects (Bon et al., 2018). In the 4DE1 reviews, EV1 is described as a source main sequence driven by Eddington ratio convolved with line-of-sight orientation (Sulentic et al., 2015).

At the same time, several theoretical studies proposed that the maximum accretion disk temperature, $\beta$2, reflected in the shape of the spectral energy distribution, may be the more direct driver of the optical EV1 sequence (Panda et al., 2017). In these CLOUDY calculations, the BLR was represented as a constant-density cloud in plane-parallel geometry, and one model sequence found a clear increase of $\beta$3 with $\beta$4 over most of the considered temperature range, supporting $\beta$5 as the direct determinant of the optical Fe II/H$\beta$6 ratio (Panda et al., 2017). A related study also concluded that the maximum accretion disk temperature is the primary physical driver of EV1, while hydrogen density and microturbulence modulate the relation (Panda et al., 2017).

However, another CLOUDY-based study reported the opposite monotonic trend: increasing $\beta$7 caused the Fe II/H$\beta$8 ratio to decrease, leading the authors to conclude that either the hypothesis is incorrect or the approximations used for the description of the line emissivity are inadequate (Panda et al., 2017). This establishes a genuine modeling tension rather than a resolved result.

Observational studies of extreme EV1 sources further complicate a one-parameter interpretation. A detailed re-analysis of 27 SDSS quasars initially selected with $\beta$9 confirmed only six such sources after individualized spectral modeling, and these six did not form a homogeneous population: three had FWHM(H$$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$0) below 2100 km s$$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$1, while three had broad lines above 4500 km s$$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$2, with considerable diversity in black hole mass and Eddington ratio (Sniegowska et al., 2017). All six showed strong [O III] emission, contrary to the simple Fe II–[O III] anticorrelation expected from the classic EV1 picture (Sniegowska et al., 2017). The interpretation of EV1 therefore remains an open issue in the literature.

5. Variability, single-object evolution, and state changes

EV1 has also been used as a time-domain diagnostic. The long-term monitoring of NGC 5548 provides a single-object test of whether motion in the EV1 plane traces structural changes or only state changes. Over a 43-year spectroscopic baseline, NGC 5548 showed large-amplitude continuum fluctuations, flaring behavior, and even changes in Seyfert type from 1 to 1.8 and back, yet it remained confined mostly within Population B on the EV1 diagram (Bon et al., 2018). In low activity states it moved deeper within Population B, while bright states moved it toward B1 or only marginally into Pop A1 territory, never solidly into Population A (Bon et al., 2018).

For NGC 5548, the analysis assumed inclination and black hole mass to be constant during the observational time, so movement in EV1 was interpreted primarily in terms of accretion rate changes (Bon et al., 2018). The reported results included an anticorrelation between $$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$3 and $$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$4, a weak anticorrelation between FWHM(H$$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$5) and continuum luminosity, and no crossing of the structural boundary near $$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$6 (Bon et al., 2018). The weak FWHM response was linked to “BLR breathing,” in which BLR size increases with luminosity and broad lines become slightly narrower as the source brightens (Bon et al., 2018). A plausible implication is that EV1 motion within a single object need not reproduce ensemble correlations measured across many quasars.

A larger variability study reinforced this difference between ensemble EV1 trends and time evolution in individual sources. Using 13,438 SDSS quasars, the authors found that later brightness changes were clearly related to position on the EV1 plane (Nagoshi et al., 2022). A Changing-Look quasar sample showed that bright and dim states occupy opposite sides of the typical quasar distribution, and a multi-epoch sample of 2,839 quasars showed that brightening and dimming sources move on similar paths in opposite directions across the EV1 ridge (Nagoshi et al., 2022). The same study explicitly noted that, for individual quasars over time, the trend is opposite to the empirical rule that $$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$7 is positively correlated with Eddington ratio in large quasar samples (Nagoshi et al., 2022). This suggests that EV1 contains both a population sequence and a state variable component.

6. Applications, extreme accretors, and current directions

EV1 has practical applications in black hole mass estimation, source selection, and the study of high-accretion AGN. In the ultraviolet, the continuum-subtracted peak flux ratio of the Si IV + O IV] blend near $$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$8 to C IV $$R_{\mathrm{FeII}} = \frac{\mathrm{EW}_{\mathrm{FeII}(4434{-}4684\,\text{\AA})}}{\mathrm{EW}_{\mathrm{H}\beta}},$$9 serves as an EV1 indicator, denoted peak $\beta$0/C IV (Brotherton et al., 2015). Using a sample of 85 quasars with quasi-simultaneous optical-ultraviolet spectrophotometry, one study showed that EV1 biases in reverberation-mapped calibration samples systematically bias C IV-based black hole mass scaling relationships, leading to black hole masses nearly 50% too high for the average quasar (Brotherton et al., 2015). This made EV1 a calibration variable rather than only a classification variable.

The 4DE1 context has also been used to define highly accreting quasars. At low redshift, $\beta$1 was proposed as a sufficient condition for identifying xA sources, while at high redshift the UV criteria Al III $\beta$2/Si III] $\beta$3 and Si III] $\beta$4/C III] $\beta$5 were proposed as analogs of the optical criterion (Marziani et al., 2014). These sources were argued to isolate objects with a defined physical structure, namely a geometrically thick, optically thick advection-dominated accretion disk or “slim” disk, and their Eddington ratio was expected to saturate toward values of order unity, making them possible cosmological probes (Marziani et al., 2014).

Recent survey work has extended the extreme EV1 end with much larger samples. A DESI DR1 study mapped 18,749 NLSy1 galaxies onto the EV1 plane and found that the DESI population is shifted toward the extreme end of EV1, with stronger Fe II emission than the SDSS DR17 NLSy1 catalog and a larger fraction of sources above the Eddington limit (Domínguez et al., 9 Jun 2026). The reported medians were $\beta$6 for the DESI sample and $\beta$7 for the SDSS sample, with 43.8%–47.7% of the DESI sources showing $\beta$8 Eddington ratio $\beta$9, compared with 20.6%–37.4% in SDSS (Domínguez et al., 9 Jun 2026). In this interpretation, DESI sensitivity revealed a large population of low-mass, super-Eddington accretors that were underrepresented in previous surveys (Domínguez et al., 9 Jun 2026).

Across these applications, EV1 functions as both a taxonomy and a physical hypothesis. It organizes quasar spectra into a main sequence; it separates Population A and B phenomenology; it supports tests of accretion-rate, orientation, and SED-based interpretations; it exposes sample-selection biases in virial mass work; and it provides an empirical framework for tracking both ensemble trends and long-term state changes. The literature consistently treats EV1 as indispensable for quasar phenomenology, but it does not yet support a unique physical reduction of EV1 to a single parameter (Sniegowska et al., 2017).