Abstract
We present measurements of the spectral properties for a total of 526,265 quasars, out of which 63% have a continuum signal-to-noise ratio > 3 pixel−1, selected from the fourteenth data release of the Sloan Digital Sky Survey (SDSS-DR14) quasar catalog. We performed a careful and homogeneous analysis of the SDSS spectra of these sources to estimate the continuum and line properties of several emission lines such as Hα, Hβ, Hγ, Mg ii, C iii], C iv, and Lyα. From the derived emission line parameters, we estimated single-epoch virial black hole masses (MBH) for the sample using Hβ, Mg ii, and C iv emission lines. The sample covers a wide range in bolometric luminosity ( erg s−1) between 44.4 and 47.3 and between 7.1 and 9.9 M⊙. Using the ratio of Lbol to the Eddington luminosity as a measure of the accretion rate, the logarithm of the accretion rate is found to be in the range between −2.06 and 0.43. We performed several correlation analyses between different emission line parameters and found them to match the correlation known earlier using smaller samples. We note that strong Fe ii sources with a large Balmer line width and highly accreting sources with large MBH are rare in our sample. We make an extended and complete catalog available online that contains various spectral properties of 526,265 quasars derived in this work along with other properties culled from the SDSS-DR14 quasar catalog.
Export citation and abstract BibTeX RIS
1. Introduction
Quasars are a class of active galactic nuclei (AGN). They are powered by accretion of matter onto a supermassive black hole surrounded by an accretion disk (e.g., Antonucci 1993). The availability of a large number of quasars with measured line and continuum properties is of paramount importance in a wide variety of astrophysical research such as galaxy evolution, black hole growth, etc. For example, the mass of the black holes (MBH) in AGN is found to be strongly correlated with the velocity dispersion of the host galaxies, suggesting that the black hole and host galaxy evolve together (e.g., Kormendy & Ho 2013). Thus, measuring MBH for a large sample of quasars is required for studying the growth and evolution of black holes across cosmic time. A direct method for measuring MBH in quasars over a large range of redshifts is the technique of reverberation mapping (Blandford & McKee 1982; Peterson 1993). Studies that used this method show a strong correlation between the quasar monochromatic luminosity (L) at 5100 Å and the size (R) of the broadline region (BLR; e.g., Kaspi et al. 2000; Bentz et al. 2009, 2013). Because reverberation mapping requires a long-term monitoring campaign, which is difficult for high-redshift and high-luminosity objects, the size–luminosity (R − L) relation has been used to estimate MBH from a single-epoch spectrum for which monochromatic luminosity and emission line width measurements are available (e.g., Woo & Urry 2002; Shen et al. 2011). The values of MBH estimated from a single-epoch spectrum are mostly consistent with the reverberation mapping MBH estimates within a factor of few (e.g., Wandel et al. 1999; McLure & Jarvis 2002; Vestergaard 2002; Grier et al. 2017, but also see Vestergaard & Peterson 2006; Shen 2013 and Peterson 2014 for merits and caveats of single-epoch MBH).
Statistical studies of quasars will also help to understand quasar properties better (Kellermann et al. 1989; Urry & Padovani 1995), such as the quasar luminosity function (Richards et al. 2006b); the black hole mass function, which shows a peak at z ∼ 2 (Vestergaard & Osmer 2009; Kelly et al. 2010); and the Eddington ratio distribution, which peaks at Lbol/LEdd ∼ 0.05 (Kelly et al. 2010), where Lbol is the bolometric luminosity and LEdd = 1.3 × 1038 (MBH/M⊙) erg s−1 is the Eddington luminosity. Several correlations between continuum and emission line properties in quasars are available, e.g., the anticorrelation between the line equivalent width (EW) and continuum luminosity (Baldwin 1977), which is especially strong in the C iv and Mg ii lines (Shen et al. 2011), and the correlations between continuum luminosity, line widths, and luminosities, etc. (e.g., Boroson & Green 1992; Greene & Ho 2005; Shen et al. 2011; Calderone et al. 2017; Rakshit et al. 2017). Studies of the emission lines from AGN will also help our understanding of the physical conditions of the gas close to the central regions of AGN (Osterbrock 1989).
All of the above studies require large samples of quasars. Since the discovery of quasars about more than half a century ago (Schmidt 1963), the number of quasars that are known has increased gradually. A significant increase in the number of quasars occurred in the last two decades, with the bulk of the contribution coming from the Sloan Digital Sky Survey (SDSS; York et al. 2000). In addition to the SDSS, other surveys have also contributed to the increase in the number of quasars, such as the 2dF quasar redshift survey (2QZ; Croom et al. 2004), the bright quasar survey (Schmidt & Green 1983), and the large bright quasar survey (LBQS; Hewett et al. 1995). The number of quasars is also expected to increase manifold in the future with the next-generation large optical imaging survey using the Large Synoptic Survey Telescope (LSST, now known as the Vera C. Rubin Observatory; Ivezić et al. 2019, 2014).
Among the many available quasar surveys, SDSS has provided us with the largest homogeneous sample of quasars with optical spectra. Each SDSS quasar survey had different science goals. For example, the SDSS DR7 quasar catalog (Schneider et al. 2010) consists of 105,783 spectroscopically confirmed quasars from the SDSS-I/II survey (York et al. 2000), whose aim was studying the quasar luminosity function (e.g., Richards et al. 2006b) and clustering properties (e.g., Hennawi et al. 2006; Shen et al. 2007). The survey also led to the discovery of many high-redshift z ∼ 6 quasars (e.g., Fan et al. 2006; Jiang et al. 2008) and broad absorption line quasars (e.g., Reichard et al. 2003; Trump et al. 2006; Gibson et al. 2008). The SDSS-III/BOSS survey (Eisenstein et al. 2011; Dawson et al. 2013) was intended to discover a large sample of quasars within the Lyα forest, which fall in the redshift range of 2.15–3.5, to constrain the baryon acoustic oscillation (BAO) scale. This survey led to the discovery of 270,000 quasars, mostly at z > 2, which helped to provide strong cosmological constraints at z ∼ 2.5 through the autocorrelation of the Lyα forest (e.g., Bautista et al. 2017) and by cross-correlation of quasars and Lyα forest (e.g., du Mas des Bourboux et al. 2017).
The SDSS-IV has multiple goals. SDSS-IV/eBOSS (see Dawson et al. 2016) is dedicated to measure the percent-level angular diameter distance dA(z) and Hubble parameter H(z) using 250,000 new spectroscopically confirmed luminous red galaxies, 195,000 new emission line galaxies, 500,000 spectroscopically confirmed quasars, and 60,000 new Lyα forest quasar measurements at redshifts z > 2.1. The time-domain Spectroscopic Survey (TDSS) of SDSS-IV was designed to study the spectroscopic variability of quasars, and the Spectroscopic Identification of eROSITA Sources (SPIDERS) program was designed to investigate X-ray sources in SDSS-IV. Pâris et al. (2018) recently compiled a quasar catalog from SDSS-IV including all previously spectroscopically selected quasars from the SDSS I, II, and III surveys. This catalog consists of 526,356 quasars over a region of the sky of 9376 degree2 from SDSS, with 144,046 newly discovered quasars from SDSS-IV. The catalog of Pâris et al. (2018) therefore is a unique and the largest list of spectroscopically confirmed quasars selected homogeneously and covering a large part of the northern sky. When all the spectral properties of the quasars in Pâris et al. (2018) are available, the catalog can serve as the largest quasar database useful to address a wide variety of astrophysical problems and/or revisit the already known correlations between various quasar properties.
The spectral properties of SDSS DR7 quasars have been studied by Shen et al. (2011, hereafter S11). DR7 consists of about 100,000 quasars. Calderone et al. (2017, hereafter C17) studied spectral properties of about 70,000 quasars at z < 2 from SDSS-DR10 (Ahn et al. 2014), which contains the first data release from SDSS-III. The latest SDSS-DR14 quasar catalog of Pâris et al. (2018) not only increases the number of quasars by a factor of 5 compared to SDSS DR7, it also covers about 1.5 mag fainter sources (i-band absolute magnitude ) than SDSS DR7 (). As the DR14 quasar catalog includes much fainter quasars, this opens up the possibility of exploring the properties of quasars over a large range in luminosity. Although the catalog contains the X-ray, UV, optical, IR, and radio imaging properties of the quasars wherever available, it lacks spectral information of the sources. About 332,000 (∼63%) sources in the DR14 catalog have a continuum signal-to-noise ratio (S/N) > 3 pixel−1. This is a factor of 3 larger than the entire sample of the S11 catalog. Thus, the DR14 catalog with its spectral information will be useful for the astronomical community not only for statistical studies of quasars, but also to discover and investigate peculiar objects.
We therefore carried out detailed spectral modeling of all the quasars cataloged in Pâris et al. (2018) and provide a new catalog of continuum and emission line properties of 526,265 quasars along with MBH and the Eddington ratio. This paper is structured as follows. Our data and spectral analysis procedures are described in Section 2. We compare our measurements with the previous works in Section 3. In Section 4 we discuss the impact of the S/N of the spectra on the derived spectral quantities. We discuss some applications of the catalog in Section 5 with a summary in Section 6. In Appendix A we define and describe the quality of our spectral measurements, and in Appendix B we present the spectral catalog. A cosmology with , Ωm = 0.3, and Ωλ =0.7 is assumed throughout.
2. Data and Spectral Analysis
We started with the SDSS DR14 quasar catalog (version "DR14Q_v4_4") by Pâris et al. (2018, hereafter DR14Q), which includes all the spectroscopically confirmed quasars observed during any SDSS data release, consisting of 526,356 quasars based on i-band absolute magnitude and having at least one emission line with a full width at half maximum (FWHM) larger than 500 km s−1 or having interesting or complex absorption features. It was constructed from SDSS-DR14 (Abolfathi et al. 2018), and a great part of the newly discovered quasars in DR14Q are from the extended Baryon Oscillation Spectroscopic Survey (eBOSS) of SDSS IV (Myers et al. 2015). A detailed description of DR14Q can be found in Pâris et al. (2018).
To measure the spectral information of the quasars in DR14Q, we first downloaded all the processed and calibrated4 spectra from the SDSS database.5 We then analyzed each spectrum using the publicly available multicomponent spectral fitting code PyQSOFit6 developed by Guo et al. (2018). A detailed description of the code and its applications can be found in Guo et al. (2019) and Shen et al. (2019). First, we corrected each spectrum for Galactic extinction using the Schlegel et al. (1998) map and the Milky Way extinction law of Fitzpatrick (1999) with RV = 3.1. We then transformed the observed spectrum to the rest-frame wavelength7 using the redshift (z) value provided in DR14Q. Finally, we performed multicomponent spectral fitting for each spectrum.
2.1. Continuum Components
The light from stars in the host galaxy of a quasar can contribute to the observed quasar's spectrum. This is particularly significant for low-z quasars (). Thus, to extract intrinsic AGN properties, the contribution of the host galaxy to each spectrum must be removed. We therefore decomposed the host galaxy from spectra for z < 0.8 quasars based on the principal component analysis (PCA; Yip et al. 2004a, 2004b) that is implemented in PyQSOFit code. The PCA method has been used in several previous studies (Vanden Berk et al. 2006; Shen et al. 2008a, 2015) to decompose host galaxy and quasar contribution. It assumes that the observed composite spectrum is a combination of two independent sets of eigenspectra taken from pure galaxy (Yip et al. 2004a) and pure quasar (Yip et al. 2004b) samples. The first three galaxy eigenspectra contain 98% of the galaxy sample information, while the first 10 quasar eigenspectra contain 92% of the quasar sample information. Vanden Berk et al. (2006) performed a PCA on 11,000 SDSS quasars. They also studied the reliability of the spectral decomposition with the S/N of the spectrum, host-galaxy fraction, galaxy class, etc. The host-galaxy decomposition is considered to be successful if the host-galaxy fraction in the wavelength range of 4160–4210 Å is larger than 10%. This method has also been applied by Shen et al. (2008a, 2015) to decompose the host galaxy from the SDSS spectra. Here we also applied the PCA to decompose the host-galaxy contribution using 5 PCA components for galaxies that can reproduce about 98% of the galaxy sample and 20 PCA components for quasars that can reproduce about 96% of the quasar sample and the global model (independent of redshift and luminosity). We then subtracted the host contribution, if present, from each spectrum.
Using the host-galaxy subtracted spectrum, we modeled the entire continuum, masking the prominent emission lines as
where the power-law continuum (fpl) is
with a reference wavelength λ0 = 3000 Å. The parameters α and β are the power-law slope and normalization parameter, respectively.
The Fe ii model () is
where the parameters b0, b1, and b2 are the normalization, the Gaussian FWHM used to convolve the Fe ii template, and the wavelength shift applied to the Fe ii template, respectively, to fit the data. Both the UV and optical Fe ii emission were modeled. In PyQSOFit, the UV Fe ii template is a modified template built by Shen et al. (2019) with a constant velocity dispersion of 103.6 km s−1 from the templates of Vestergaard & Wilkes (2001), Tsuzuki et al. (2006), and Salviander et al. (2007). For the wavelength range of 1000–2200 Å, the template is from Vestergaard & Wilkes (2001), for 2200–3090 Å, the template is from Salviander et al. (2007), who extrapolated the Fe ii flux below the Mg ii line, and for 3090–3500 Å, the template is from Tsuzuki et al. (2006). The optical Fe ii template (3686−7484 Å) is based on Boroson & Green (1992). To model the Fe ii emission for each of our spectra, we first convolved the template with the parameter b1, which is constrained in the range of 1200-10,000 km s−1 with an initial guess of 3000 km s−1. A small wavelength shift (b2), constrained to be within 1% of the template wavelength, was also applied to fit the data. Then the parameters b0, b1, and b2 were varied within the range mentioned above until the best-fit model that represents the data was found. In Figure 1 we show examples of template broadening for different values of the Gaussian FWHM (b1).
The Balmer continuum (fBC; Grandi 1982; Dietrich et al. 2002) is defined as
where FBE is the normalized flux density, τλ is the optical depth at the Balmer edge of the wavelength λBE = 3646Å, and Bλ(Te) is the Planck function at the electron temperature Te. As the spectral coverage of many low-z objects is not high enough to fit a Balmer component, it is fitted whenever the continuum window has at least 100 pixels below λBE. We used FBE as a free parameter and kept Te = 15,000 K and τλ = 1 fixed to avoid degeneracy between the parameters following Dietrich et al. (2002) and C17. Moreover, for sources with z > 1.1, the Balmer component resembles a simple power law. Thus, following previous studies (e.g., C17), we further allowed FBE to vary between 0 and 0.1 × F3675. Here F3675 is the flux density at λ = 3675Å, where the Fe ii contribution is insignificant. The upper limit is also justified from the distribution of the flux ratio of the Balmer to power-law continuum at 3000 Å (see Figure 2) for z < 1.1 sources, which has a median of ∼0.1.
Download figure:
Standard image High-resolution imageWe note that in a few low-z (z < 0.3) objects, the blue part of the spectrum between 3000–4000 Å is much steeper and the entire continuum cannot be well fit with a power-law and Balmer component. For those objects,8 we limited the spectral fitting range to above 4000 Å, thereby excluding the steep rise toward the UV.
Many high-z spectra were affected by broad and narrow absorption lines that could bias the line-fitting results, hence, we used the "rej_abs = True" option in PyQSOFit to reduce this bias. The code first performed continuum modeling ("tmp_cont") of the spectrum and removed the 3σ (where σ is the flux uncertainty) outliers below the continuum (i.e., pixels where flux <tmp_cont −3× flux uncertainty) for wavelengths <3500Å and then performed a second iteration of the continuum model fit to the 10-pixel box-car smoothed spectrum excluding the outliers. This method is found to be useful to reduce the impact of absorption features, as noted in Shen et al. (2011, 2019).
2.2. Emission Line Components
The best-fit continuum model was subtracted from each spectrum, which left the line spectrum alone. Each individual line complex was fit separately, while all the emission lines within a line complex were fit together. The full list of emission lines and the number of Gaussian components used for the individual lines are given in Table 1. Broad emission line profiles in many objects can be very complex (e.g., double-peaked, with a flat top, or asymmetric) and cannot be well represented by a single Gaussian. Moreover, the line width estimated by a single-Gaussian model is systematically larger by 0.1 dex than the multiple-Gaussian model (Shen et al. 2008b, 2011). Thus, following previous studies, we used multiple Gaussians to model the broad emission line profiles (e.g., Greene & Ho 2005; Shen et al. 2011). During the fitting, the velocity and width of all the narrow components in the Hβ and Hα complex were tied together with an added constraint that the maximum allowed FWHM of narrow components is 900 km s−1, while the broad components have FWHM >900 km s−1. The FWHM criterion was adopted to separate Type 1 AGN from Type 2 AGN following previous studies (e.g., Wang et al. 2009; Calderone et al. 2017; Coffey et al. 2019; Wang et al. 2019). The velocity offsets of the broad and narrow components were restricted to ± 3000 km s−1 and ± 1000 km s−1, respectively. Furthermore, the flux ratios of the [O iii] and [N ii] doublets were fixed to their theoretical values, i.e., F(5007)/F(4959) = 3 and F(6585)/F(6549) = 3. Note that we did not use any narrow component to model the C iii] and C iv emission lines, and the line FWHM and flux were determined from the whole line because of ambiguity in the presence of narrow components in these lines (also see Shen et al. 2011, 2019).
Table 1. Emission Line Fitting Parameters
Complex Name | Wavelength Range | Emission Line Name | Number of Gaussians |
---|---|---|---|
(Å) | |||
(1) | (2) | (3) | (4) |
Hα | 6400–6800 | Hα broad | 3 |
Hα narrow | 1 | ||
[N ii]6549 | 1 | ||
[N ii]6585 | 1 | ||
[S ii]6718 | 1 | ||
[S ii]6732 | 1 | ||
Hβ | 4640–5100 | Hβ broad | 3 |
Hβ narrow | 1 | ||
[O iii]4959 core | 1 | ||
[O iii]4959 wing | 1 | ||
[O iii]5007 core | 1 | ||
[O iii]5007 wing | 1 | ||
Hγ | 4250–4440 | Hγ broad | 1 |
Hγ narrow | 1 | ||
[O iii]4364 | 1 | ||
Mg ii | 2700–2900 | Mg ii broad | 2 |
Mg ii narrow | 1 | ||
C iii] | 1850–1970 | C iii] | 2 |
C iv | 1500–1600 | C iv | 3 |
Lyα | 1150–1290 | Lyα | 3 |
N v 1240 | 1 |
Note. The columns are as follows: (1) name of the line complex, (2) wavelength range (in Å) of the line-fitting window, (3) name of the emission line, and (4) number of Gaussians used in the fitting.
Download table as: ASCIITypeset image
A few examples of the spectral decomposition are shown in Figures 3 and 4 for spectra of different qualities. The median continuum S/N (estimated from the rest-frame spectrum in the regions around 5100 Å, 4210 Å, 3000 Å, 2245 Å, and 1350 Å depending on the spectral coverage) is also noted in the figure. Note that due to the large number of quasars, visual inspection of all the spectral fittings was not possible. Thus, only random checks of a few thousand spectra in various redshift and S/N bins were made. All the spectral fitting plots and individual model components are made publicly available for the users. We also provide various quality flags on the spectral quantities to access the reliability of the measurements. Good-quality measurements are given the quality flag = 0. Any measurements with quality flag >0 may not be reliable either due to poor S/N or poor spectral decomposition. Therefore, sources with flag >0 should be used with caution. The criteria for fulfillment of each quality flag are defined and described in Appendix A, including detailed statistics on each quality flag.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution image2.3. Spectral Quantities
We measured the continuum (slope, luminosity) and emission line (line peak, FWHM, EW, luminosity, etc.) properties from the best-fit model.9 Various studies (e.g., Collin et al. 2006; Rafiee & Hall 2011) have suggested that the line dispersion (σline), i.e., the second moment of the line (see Peterson et al. 2004), is a better measure of the emission line width than the FWHM. However, the FWHM is less affected by the noise in the line wings and by the treatments for line blending (e.g., Hβ blended with Fe ii, He IIλ4686, and [O iii]) than the σline. On the other hand, σline is less sensitive to the treatments of narrow-component removal and peculiar line profiles. Instead of line dispersion, the FWHM is preferred because it is easy to measure and repeat, especially in poor-quality spectra where the line wings are difficult to constrain and σline cannot be measured reliably. Despite that, following the prescription of Wang et al. (2019), we also measured σline for all the broad emission lines and included them in the catalog.
The uncertainty in each of the spectral quantities was estimated using a Monte Carlo approach (e.g., Shen et al. 2011, 2019; Rakshit & Woo 2018). We created a mock spectrum by adding to the original spectrum at each pixel a Gaussian random deviate with zero mean and σ given by the flux uncertainty at that pixel. We then performed the same spectral fitting on the mock spectrum as was done for the original spectrum and estimated all the spectral quantities from the mock spectrum. We created 50 such mock spectra for each object, allowing us to obtain the distribution of each spectral quantity. Finally, for each spectral quantity, the semiamplitude of the range enclosing the 16th and 84th percentiles of the distribution was taken as the uncertainty of that quantity. Therefore, all the uncertainties of the spectral quantities reported in this work were calculated using the Monte Carlo approach.
We calculated Lbol from the monochromatic luminosity using the bolometric correction factor given in S11 as adapted from the analysis in Richards et al. (2006a),
Note that the above correction factors are derived from the mean spectral energy distribution of AGN, and using a single value could lead to a 50% uncertainty in Lbol measurements (Richards et al. 2006a). Estimating the bolometric luminosity for an individual source requires multiband data from radio to X-ray to build the spectral energy distribution, which is not available for most of the quasars. However, the bolometric correction factor allows us to estimate the bolometric luminosity from their monochromatic luminosity, but with a large uncertainty.
In Figure 5 we plot Lbol against redshift for DR14 quasars. The DR7 quasars from S11 are also shown. As mentioned in Pâris et al. (2018), the peak of the redshift distribution at z ∼ 2.5 is due to the quasars that were observed by SDSS-III to access the Lyα forest, while the peaks at z ∼ 0.8 and 1.6 are due to the known degeneracy in color-redshift relation of the quasar target selection (see also Ross et al. 2012). For example, in a large number of quasars at z ∼ 0.8, the Mg ii line lies at the same wavelength as Lyα at z ∼ 3.1, providing the same broadband color. Similarly, in quasars at z ∼ 1.6, the C IV line lies at the same wavelength as Lyα at z ∼ 2.3. The bolometric luminosity has a median of with a range of 44.43–47.32 erg s−1 (3σ around the median). The errors given in the median bolometric luminosity do not include the uncertainties in the bolometric correction factor. The fraction of low-luminosity quasars in DR14 is larger than that in DR7. For example, the fraction of quasars with in DR14 is about 54%, compared to 27% in DR7.
Download figure:
Standard image High-resolution imageWe included the commonly used BALnicity index (BI; Weymann et al. 1991) and its uncertainty from the SDSS DR14 quasar catalog of Pâris et al. (2018) to flag the broad absorption line quasars (BAL-QSOs) in this work. Due to a large number of quasars, Pâris et al. (2018) performed a fully automated detection of BAL for all z > 1.57 quasars by focusing on C iv absorption troughs. A total of 21,876 quasars with C iv absorption troughs wider than 2000 km s−1 are present in this work. We also included the BAL flag of SDSS DR7 quasars from S11, who culled the BAL flag from the Gibson et al. (2009) SDSS DR5 BALQSO catalog and visually inspected post-DR5 BALQSO with redshift z > 1.45.
All the parameters and their uncertainties derived in this work are compiled into a catalog ("dr14q_spec_prop.fits") that is described in Section B and Table B4, containing 274 columns. We also provide an extended catalog ("dr14q_spec_prop_ext.fits") where we appended all other information from Pâris et al. (2018), which includes multiband imaging properties, thereby leading to 380 columns in our extended catalog. Both catalogs are available on Zenodo (doi:10.5281/zenodo.3878152); the catalogs and additional supplementary materials (e.g., best-fit model components and spectral decomposition plots for all objects) are also available online.10
3. Comparison with Previous Studies
We compared our measurements with the S11 and C17 catalogs. The first catalog is based on all DR7 quasars up to z = 5 with 105,783 entries, and the second catalog is based on DR10 quasars up to z = 2 with 71,261 entries (catalog version "qsfit_1.2.4"). Although different methods were used by the two authors than the PyQSOFit code used in our analysis, a comparison can be made. We refer to Calderone et al. (2017) for a discussion of S11 and the spectral analysis method of C17. Here we summarize the main differences.
- 1.
- 2.S11 modeled the local AGN continuum (using a power law) including the Fe ii emission, then fit the emission lines of the continuum + Fe ii-subtracted spectrum. C17 fit the continuum (power-law + Balmer continuum) of the whole spectrum, and at the final fitting step, they fit all components (continuum + galaxy + iron + emission lines) simultaneously. We first removed the host-galaxy contribution, if present, and then fit the AGN continuum (power law + Balmer continuum) and Fe ii template of the whole host-subtracted spectrum. Finally, we fit the emission lines of the continuum-subtracted spectrum.
- 3.
- 4.S11 fit broad lines with up to three Gaussians, while C17 started their modeling of broad lines with a single Gaussian ("known" line) and added more Gaussian if "unknown" emission lines were present close to the known lines. We fit most of the broad emission lines using multiple Gaussians (see Table 1).
We cross-matched our catalog with those of S11 and C17 using TOPCAT11 (Taylor 2005) and took only the common entries (71,163) for comparison. However, different catalogs may include spectra of different quality for the same target because repeated observations have been performed by SDSS. Quasars also show spectral variability, which can affect the measurements included in different catalogs. Therefore, we cross-matched sources with the same SDSS plate-mjd-fiber in all three catalogs and found 65,170 matches. Furthermore, we only included measurements with a quality flag of 0 in C17 and our work.
In Figure 6 we compare our continuum luminosity measurements with those of S11 and C17, where our measurements are plotted along the x-axis in the left panels. We also plot the measurement distributions for all three catalogs in the right panels. In general, we find excellent agreement between the measurements. The mean and standard deviation of the difference between this work and S11 (C17) is 0.001 ± 0.052 (0.023 ± 0.031) dex for L1350 (3827 sources), −0.054 ± 0.055 (0.010 ± 0.052) dex for L3000 (56,577 sources) and −0.045 ± 0.104 (0.067 ± 0.122) dex for L5100 (12,967 sources). We note a larger difference in the estimates of L5100 compared to other luminosities between S11, C17, and our work, but mainly for low-luminosity quasars. Our estimates of L5100 lie in between those of S11 and C17. We attribute this difference to differences in the host-galaxy subtraction procedures. For example, S11 did not perform host-galaxy decomposition, thus, their measurements are contaminated by the host-galaxy contribution. On the other hand, C17 used a single 5 Gyr old elliptical galaxy template to subtract the host-galaxy contribution, while we used the PCA method implemented in PyQSOFit to subtract the host galaxy (see Section 2.1). Although the PCA host decomposition method allowed us to systematically decompose the stellar contribution from a large number of spectra, it is a simplistic approach, and in principle, other host-galaxy decomposition methods can be used (e.g., Matsuoka et al. 2015; Rakshit & Woo 2018) that employ different stellar templates (e.g., Bruzual & Charlot 2003; Valdes et al. 2004) to decompose the stellar contribution from the quasar spectra.
Download figure:
Standard image High-resolution imageIn Figure 7 we compare the Hβ (top), Mg ii (middle), and C iv (bottom) line luminosity measurements of all three catalogs. In all cases, we found strong agreement. The mean and standard deviations of the Hβ (13,177 sources) line luminosity between this work and S11 (C17) are −0.045 ± 0.111 (0.044 ± 0.109) dex, while the same for the Mg ii (45,048 sources) and C iv (9,384 sources) line luminosities are 0.094 ± 0.083 (0.067 ± 0.088) dex and −0.014 ± 0.106 (0.043 ± 0.106) dex, respectively. All the line luminosity plots show a strong correlation with the Spearman rank correlation coefficient rs > 0.95 for the Hβ and Mg ii lines, and 0.94 (0.93) for the C iv line luminosity between this work and S ii (C17). We note that compared to the Hβ and C iv line luminosities, the Mg ii line luminosity shows a larger offset. Our estimated Mg ii luminosity is slightly higher than that of S11 and C17. This could be due to the use of different UV Fe ii templates. For example, Shin et al. (2019) found that the Tsuzuki et al. (2006) template provides an average 0.13 dex higher Mg ii flux and 0.10 dex lower UV Fe ii flux than the Vestergaard & Wilkes (2001) template.
Download figure:
Standard image High-resolution imageThe emission line widths in different catalogs are less strongly correlated (Figure 8), having rs = 0.85 (0.75) for Hβ, 0.82 (0.82) for Mg ii, and 0.72 (0.45) for C iv between this work and S11 (C17). This indicates the complexity in measuring the FWHM. The mean and standard deviation between this work and S11 (C17) is −0.045 ± 0.113 (−0.111 ± 0.140) dex for Hβ, 0.011 ± 0.097 (−0.031 ± 0.100) dex for Mg ii, and −0.048 ± 0.119 (−0.118 ± 0.173) dex for the C iv line width measurement. We note that on average, our FWHM measurements are more consistent with S11 than with C17. Although a slight discrepancy between different catalogs is found, measurements are in general agreement with S11 and C17. The discrepancy between different catalogs is due to the use of different spectral decomposition methods, as mentioned above.
Download figure:
Standard image High-resolution imageWe compared our estimated host fraction with that of Vanden Berk et al. (2006), who estamated the host fraction, the ratio of host flux to total flux, in the wavelength range of 4160–4210 Å. To avoid any difference due to the spectral quality between the two catalogs, we only considered objects with the same spectra in the two works (SDSS plate-mjd-fiber). There are 1486 sources with a host contribution >0 in both works. We plot them in Figure 9 (color-coded by continuum S/N). Our results are consistent with them having a median ratio (our ratio to their ratio) of . This shows that our stellar fraction measurements are consistent with those of Vanden Berk et al. (2006).
Download figure:
Standard image High-resolution image4. Impact of the S/N on Spectral Quantities
Although the spectral decomposition of high S/N spectra can be reliable, the decomposition of low S/N spectra is usually difficult. In Figure 10 we plot the density contours of the median continuum S/N with redshift in the upper panel and the distribution of the median continuum S/N at a different redshift range in the lower panel. The tail of the S/N distribution decreases rapidly at higher redshift. At low redshift z < 0.8, the fraction of sources with S/N > 3 pixel−1 is 84%, while for high redshift z > 2.2, the fraction of sources with S/N > 3 pixel−1 is 54%. The total number of sources with S/N > 3 pixel−1 in our catalog is 332,204, i.e., about 63% of the total sample.
Download figure:
Standard image High-resolution imageSeveral authors (e.g., Shen et al. 2011; Denney et al. 2016; Shen et al. 2019) have investigated the impact of the S/N on the spectral decomposition method. They found that for objects with high EW, the FWHMs and EWs are biased by less than ±20% if the line S/N is reduced to as low as about 3, while for low-EW objects, the FWHMs and EWs are biased by >20% for S/N < 5. However, in all cases, even at very low S/N, the continuum luminosity measurements are unbiased. To investigate the impact of the S/N on the measurement of our spectral quantities, we followed an approach similar to that in the previous studies. First, we selected a sample of 30 original spectra with high S/N independently for Hβ in the redshift of z < 0.8, Mg ii in the redshift range of z = 0.8–1.8, and C iv in the redshift range of z = 1.8–3.2. For each spectrum, we then multiplied a constant factor of 2, 3, 4, 6, 8, and 10 to their original flux errors and added a Gaussian random deviate of zero mean and standard deviation given by the new flux errors to the original spectrum. We then repeated our spectral decomposition method as used in the decomposition of high-S/N original spectra, remeasured all the spectral quantities from the degraded spectra, and finally compared them with the high-S/N original spectra. In Figure 11 we plot the ratio of the measurements from degraded spectra to the original high-S/N spectra as a function of the median continuum S/N for all 30 objects for each line (the measurement from the individual spectrum is represented by a unique color). With decreasing S/N, the measurement uncertainties increase, as expected. For example, when the sample median S/N is reduced by a factor of 10 from 36.5 to 3.6, the uncertainty in EW increased from 4.8 Å to 10.2 Å, and the uncertainty in the FWHM increased from 221 km s−1 to 1246 km s−1. The offsets represented by the sample median (stars) are negligible even at very low S/N, suggesting that the measurements are unbiased. However, measurements of any individual spectrum can deviate by 50% or more. The Mg ii EW (top middle) shows a systematic offsets with decreasing S/N. However, this is not present in the case of the Hβ (top left) and C iv (top right) EW. The reason for this offset in Mg ii might be the blending of the UV Fe ii and Mg ii lines. The continuum luminosity is unbiased even at very low S/N.
Download figure:
Standard image High-resolution imageThe above investigation suggests that on average, our spectral decomposition method recovers the measurements of the high S/N ratio spectra, although individual measurements can deviate by 50% or more. For peculiar sources, e.g., those with a double-peak emission line, our decomposition may fail badly. For this purpose, we provide various quality flags for each object, as described in detail in Appendix A, based on several criteria. These quality flags give the reliability of our measurements.
5. Applications
5.1. Correlation Analysis
The spectral catalog generated in this work for a large number of quasars can be used to investigate the correlation between various line and continuum properties in detail. Here, we studied some of the correlations. For this, we considered measurements with a quality flag of zero. The luminosity of the Balmer lines shows a strong correlation with the monochromatic continuum luminosity at 5100 Å over a wide range of redshift and luminosity, suggesting that the physical mechanisms behind the correlation are the same in different AGN across the entire redshift and luminosity range (e.g., Greene & Ho 2005; Jun et al. 2015; Rakshit et al. 2017). In Figure 12 we plot the luminosity of Hα (top panel) and Hβ (middle panel) against L5100. In both cases, a strong correlation is found with rs of 0.93 and 0.86, respectively. We performed a linear regression analysis with measurement errors on both axes using the Bayesian code linmix12 (Kelly 2007) and obtained
with an intrinsic scatter of 0.025 ± 0.001 and 0.035 ± 0.001, respectively. These correlations are shown by the black dashed line in Figure 12. For Equation (5), a linear regression using bces13 (Akritas & Bershady 1996; Nemmen et al. 2012) gives a slope (m) of 1.125 ± 0.005 and intercept (c) of −6.86 ± 0.22 considering as the independent variable (dash-dotted blue line), and c = −12.97 ± 0.27 considering as the independent variable (dotted cyan line) and m = 1.120 ± 0.005 and c = −10.37 ± 0.25 for orthogonal least-squares (dashed magenta line). The same for Equation (6) is found to be 0.937 ± 0.002 and 0.89 ± 0.11 (m, c) for as independent variable, m = 1.177 ± 0.002 and c = −9.75 ± 0.11 considering as the independent variable and m = 1.057 ± 0.002 and c = −4.41 ± 0.10 for orthogonal least-squares. The slopes of the and correlations agree with those in previous studies. For example, using a sample of low-z (z < 0.35) SDSS quasars, Greene & Ho (2005) found a slope of 1.157 ± 0.005 for Equation (5) and 1.133 ± 0.005 for Equation (6). For high-redshift (z = 1.5–2.2) and high-luminosity () quasars, Shen & Liu (2012) obtained a slope of 1.010 ± 0.042 and 1.251 ± 0.067 for Equations (5) and (6), respectively. Jun et al. (2015) investigated the correlation using quasars of z = 0.0–6.2 and luminosity of and obtained a slope of 1.044 ± 0.008.
Download figure:
Standard image High-resolution imageThe correlation between has been widely used to estimate the bolometric luminosity for Type 2 AGN because their host-galaxy contamination prevents a reliable estimation of L5100 (see Kauffmann et al. 2003; Heckman et al. 2004). However, this relation has a large scatter (Heckman et al. 2004; Shen et al. 2011). The L([O iii]λ5007) as a function of L5100 is plotted in the bottom panel (rs = 0.58) of Figure 12. The best-fit linear regression using linmix gives
with an intrinsic scatter of 0.102 ± 0.001. The same using bces is found to be m = 0.693 ± 0.003 and c = 11.24 ± 0.15, however, when is the independent variable, m = 1.347 ± 0.031 and c = −17.74 ± 1.41 when is the independent variable, and m = 0.953 ± 0.015 and c =−0.30 ± 0.69 for orthogonal least-squares. Depending on the treatment of the independent variable, the relation shows a range of slopes of 0.7 to 1.3 due to large scatter. This agrees with Shen et al. (2011), who noted a scatter of ∼0.4 dex around the best-fit relation and a slope of 0.77 when L5100 is considered as the independent variable, and 1.34 for a bisector linear regression fit.
In Figure 13 we plot various such line and continuum quantities, and we performed a correlation analysis between them. The fits to the data are shown by dashed lines in Figure 13, and the results of the fitting are given in Table 2. Most of the correlation agrees with the previous works, which were based on smaller samples. For example, the continuum luminosities at 1350 Å, 3000 Å, and 5100 Å are strongly correlated with each other (e.g., Jun et al. 2015). The line width of Hβ and Mg ii shows a strong correlation (e.g., Wang et al. 2009), however, the correlation is very weak between the Mg ii and C iv line FWHM. All the parameters are uncorrelated with spectral index, except for a weak anticorrelation with L5100 and the Hβ EW (Shen et al. 2011). The mission line FWHM is weakly correlated with the line EW both for Hβ and Mg ii lines, similar to what has been noted for the Mg ii line by Dong et al. (2009). The well-known anticorrelation between continuum luminosity and line EW is found both for Mg ii and C iv (Baldwin 1977), but not for the Hβ EW. Although no correlation between EW of the optical Fe ii (measured from the best-fit optical Fe ii template in the wavelength range of 4435–4685 Å) and of the EW of Hβ is found, the EW of UV Fe ii (measured from the best-fit UV Fe ii template in the wavelength range of 2200–3090 Å) is strongly correlated with the EW of the Mg ii line and C IV (e.g., Kovačević-Dojčinović & Popović 2015).
Download figure:
Standard image High-resolution imageTable 2. Linear Regression Analysis to the y vs. x Correlation for Objects with Quality Flag = 0 Using linmix with a Slope (m), Intercept (c), and Intrinsic Scatter ()
y | x | m | c | σint | rs |
---|---|---|---|---|---|
PL_SLOPE | LOG_FWHM_HB_BR | −0.280 ± 0.018 | −0.239 ± 0.064 | 0.463 ± 0.003 | −0.09 |
PL_SLOPE | LOG_EW_HB_BR | −1.051 ± 0.017 | 0.702 ± 0.030 | 0.445 ± 0.003 | −0.30 |
PL_SLOPE | LOG_L5100 | −0.389 ± 0.006 | 16.100 ± 0.256 | 0.508 ± 0.003 | −0.24 |
PL_SLOPE | LOG_EW_FE_OP | 0.040 ± 0.013 | −1.308 ± 0.021 | 0.492 ± 0.003 | −0.05 |
PL_SLOPE | LOG_FWHM_MG ii_BR | 0.317 ± 0.008 | −2.476 ± 0.030 | 0.311 ± 0.001 | 0.03 |
PL_SLOPE | LOG_EW_MG ii_BR | 0.420 ± 0.005 | −2.042 ± 0.009 | 0.313 ± 0.001 | 0.15 |
PL_SLOPE | LOG_EW_FE_UV | 0.250 ± 0.005 | −1.816 ± 0.010 | 0.297 ± 0.001 | 0.07 |
PL_SLOPE | LOG_L3000 | −0.183 ± 0.002 | 6.972 ± 0.073 | 0.341 ± 0.001 | −0.12 |
PL_SLOPE | LOG_FWHM_C iv | −0.293 ± 0.007 | −0.208 ± 0.024 | 0.258 ± 0.001 | −0.08 |
PL_SLOPE | LOG_EW_C iv | 0.274 ± 0.004 | −1.754 ± 0.008 | 0.254 ± 0.001 | 0.15 |
PL_SLOPE | LOG_L1350 | −0.057 ± 0.002 | 1.407 ± 0.100 | 0.317 ± 0.001 | −0.03 |
LOG_FWHM_HB_BR | LOG_EW_HB_BR | 0.260 ± 0.005 | 3.087 ± 0.008 | 0.031 ± 0.001 | 0.22 |
LOG_FWHM_HB_BR | LOG_L5100 | 0.072 ± 0.002 | 0.338 ± 0.086 | 0.030 ± 0.001 | 0.16 |
LOG_FWHM_HB_BR | LOG_EW_FE_OP | −0.159 ± 0.004 | 3.829 ± 0.006 | 0.030 ± 0.001 | −0.21 |
LOG_FWHM_HB_BR | LOG_FWHM_MG ii_BR | 1.023 ± 0.004 | −0.089 ± 0.014 | 0.005 ± 0.001 | 0.67 |
LOG_FWHM_HB_BR | LOG_EW_MG ii_BR | 0.243 ± 0.004 | 3.171 ± 0.007 | 0.024 ± 0.001 | 0.21 |
LOG_FWHM_HB_BR | LOG_EW_FE_UV | 0.082 ± 0.005 | 3.420 ± 0.010 | 0.024 ± 0.001 | 0.09 |
LOG_FWHM_HB_BR | LOG_L3000 | 0.064 ± 0.002 | 0.730 ± 0.075 | 0.030 ± 0.001 | 0.17 |
LOG_EW_HB_BR | LOG_L5100 | −0.049 ± 0.002 | 3.987 ± 0.091 | 0.034 ± 0.001 | −0.04 |
LOG_EW_HB_BR | LOG_EW_FE_OP | 0.209 ± 0.003 | 1.491 ± 0.005 | 0.027 ± 0.001 | 0.24 |
LOG_EW_HB_BR | LOG_FWHM_MG ii_BR | 0.112 ± 0.007 | 1.433 ± 0.027 | 0.037 ± 0.001 | 0.09 |
LOG_EW_HB_BR | LOG_EW_MG ii_BR | 0.288 ± 0.005 | 1.354 ± 0.008 | 0.035 ± 0.001 | 0.24 |
LOG_EW_HB_BR | LOG_EW_FE_UV | 0.240 ± 0.006 | 1.367 ± 0.012 | 0.041 ± 0.001 | 0.21 |
LOG_EW_HB_BR | LOG_L3000 | −0.005 ± 0.002 | 2.029 ± 0.083 | 0.034 ± 0.001 | 0.04 |
LOG_L5100 | LOG_EW_FE_OP | −0.022 ± 0.006 | 44.438 ± 0.011 | 0.150 ± 0.001 | 0.02 |
LOG_L5100 | LOG_FWHM_MG ii_BR | 0.169 ± 0.009 | 43.874 ± 0.034 | 0.135 ± 0.001 | 0.10 |
LOG_L5100 | LOG_EW_MG ii_BR | −0.752 ± 0.006 | 45.792 ± 0.010 | 0.107 ± 0.001 | −0.42 |
LOG_L5100 | LOG_EW_FE_UV | −0.463 ± 0.007 | 45.396 ± 0.014 | 0.118 ± 0.001 | −0.24 |
LOG_L5100 | LOG_L3000 | 0.806 ± 0.001 | 8.576 ± 0.044 | 0.019 ± 0.001 | 0.93 |
LOG_EW_FE_OP | LOG_FWHM_MG ii_BR | −0.450 ± 0.009 | 3.312 ± 0.033 | 0.053 ± 0.001 | −0.24 |
LOG_EW_FE_OP | LOG_EW_MG ii_BR | −0.068 ± 0.007 | 1.820 ± 0.012 | 0.057 ± 0.001 | −0.13 |
LOG_EW_FE_OP | LOG_EW_FE_UV | 0.094 ± 0.008 | 1.564 ± 0.015 | 0.047 ± 0.001 | 0.00 |
LOG_EW_FE_OP | LOG_L3000 | −0.025 ± 0.003 | 2.792 ± 0.119 | 0.062 ± 0.001 | 0.00 |
LOG_FWHM_MG ii_BR | LOG_EW_MG ii_BR | 0.262 ± 0.001 | 3.207 ± 0.002 | 0.017 ± 0.001 | 0.32 |
LOG_FWHM_MG ii_BR | LOG_EW_FE_UV | 0.103 ± 0.001 | 3.461 ± 0.003 | 0.018 ± 0.001 | 0.12 |
LOG_FWHM_MG ii_BR | LOG_L3000 | 0.014 ± 0.001 | 3.041 ± 0.025 | 0.019 ± 0.001 | 0.06 |
LOG_FWHM_MG ii_BR | LOG_FWHM_C iv | 0.241 ± 0.003 | 2.797 ± 0.013 | 0.013 ± 0.001 | 0.17 |
LOG_FWHM_MG ii_BR | LOG_EW_C iv | 0.064 ± 0.002 | 3.563 ± 0.003 | 0.014 ± 0.001 | 0.06 |
LOG_FWHM_MG ii_BR | LOG_L1350 | −0.031 ± 0.001 | 5.100 ± 0.045 | 0.015 ± 0.001 | −0.02 |
LOG_EW_MG ii_BR | LOG_EW_FE_UV | 0.708 ± 0.002 | 0.342 ± 0.003 | 0.021 ± 0.001 | 0.58 |
LOG_EW_MG ii_BR | LOG_L3000 | −0.214 ± 0.001 | 11.388 ± 0.029 | 0.033 ± 0.001 | −0.49 |
LOG_EW_MG ii_BR | LOG_FWHM_C iv | −0.215 ± 0.006 | 2.511 ± 0.020 | 0.047 ± 0.001 | −0.12 |
LOG_EW_MG ii_BR | LOG_EW_C iv | 0.536 ± 0.002 | 0.781 ± 0.004 | 0.026 ± 0.001 | 0.62 |
LOG_EW_MG ii_BR | LOG_L1350 | −0.338 ± 0.001 | 17.098 ± 0.064 | 0.028 ± 0.001 | −0.65 |
LOG_EW_FE_UV | LOG_L3000 | −0.146 ± 0.001 | 8.516 ± 0.039 | 0.043 ± 0.001 | −0.27 |
LOG_EW_FE_UV | LOG_FWHM_C iv | −0.257 ± 0.006 | 2.882 ± 0.022 | 0.047 ± 0.001 | −0.10 |
LOG_EW_FE_UV | LOG_EW_C iv | 0.515 ± 0.003 | 1.048 ± 0.005 | 0.029 ± 0.001 | 0.54 |
LOG_EW_FE_UV | LOG_L1350 | −0.217 ± 0.002 | 11.805 ± 0.084 | 0.043 ± 0.001 | −0.31 |
LOG_L3000 | LOG_FWHM_C iv | 0.784 ± 0.007 | 42.513 ± 0.027 | 0.173 ± 0.001 | 0.28 |
LOG_L3000 | LOG_EW_C iv | −0.952 ± 0.003 | 47.099 ± 0.006 | 0.120 ± 0.001 | −0.57 |
LOG_L3000 | LOG_L1350 | 0.896 ± 0.001 | 4.624 ± 0.047 | 0.033 ± 0.001 | 0.91 |
LOG_FWHM_C iv | LOG_EW_C iv | −0.179 ± 0.002 | 3.938 ± 0.003 | 0.025 ± 0.001 | −0.21 |
LOG_FWHM_C iv | LOG_L1350 | 0.129 ± 0.001 | −2.278 ± 0.038 | 0.024 ± 0.001 | 0.33 |
LOG_EW_C iv | LOG_L1350 | −0.305 ± 0.001 | 15.749 ± 0.042 | 0.038 ± 0.001 | −0.53 |
Note. The Spearman rank correlation coefficient (rs) is also noted.
Download table as: ASCIITypeset image
Boroson & Green (1992) performed a PCA using a sample of 87 PG quasars (z < 0.5) and found various correlations involving the optical Fe ii, [O iii]5007, and Hβ broad component, the radio-to-optical flux ratio, and the optical to X-ray spectral index. The first PCA eigenvector (Eigenvector 1; hereafter E1) is strongly anticorrelated with (defined by the ratio of the EW of Fe ii (4435−4685 Å) to Hβ broad line) and luminosity of [O iii]. The main parameters of the well-known 4DE1 (Sulentic et al. 2000b, 2002), which can account for the diverse nature of broadline AGN, are the FWHM of the broad Hβ line and . These two quantities are plotted in Figure 14 for objects with quality flag = 0 in the catalog. First, quasars with a wide range of Fe ii strength can be found at a given FWHM(Hβ). Similarly, at a given Fe ii strength, the Hβ line can have a large range. The distribution peaks at ∼0.7, which can be occupied by quasars with an FWHM(Hβ) of ∼1000–10,000 km s−1. This dispersion is suggested to be due to an orientation effect (see Shen & Ho 2014; Sun & Shen 2015). Second, the well-known trend of decreasing FWHM with increasing RFeII is noted, as also shown in previous studies (e.g., Shen et al. 2011; Coffey et al. 2019). Thus, quasars with a very broad Hβ line and strong Fe ii strength are rare, especially in the low-redshift SDSS sample. However, an IR spectroscopic study of high-z quasars shows a systematically larger FWHM(Hβ) than the low-z sources at high (Shen 2016). The dashed line represents the separation of quasars into two populations: population A (FWHM(Hβ, broad) ≤ 4000 km s−1) sources with strong Fe ii and soft X-ray excess, and population B (FWHM(Hβ, broad) > 4000 km s−1) sources with weak Fe ii and a lack of soft X-ray excess (Sulentic et al. 2000a).
Download figure:
Standard image High-resolution imagePrevious studies found an anticorrelation between [O iii] and Fe ii emission, i.e., objects with strong [O iii] are found to be weak Fe ii emitters and vice versa (e.g., Boroson & Green 1992; Grupe et al. 1999; McIntosh et al. 1999; Rakshit et al. 2017), although Véron-Cetty et al. (2001) found this anticorrelation to be very weak. In Figure 15 we plot R5007, i.e., the ratio of EW of [O iii]5007 to the EW of total Hβ (sum of the EW of broad and narrow Hβ components) against (left) and EW of Fe ii (right). We found that the anticorrelation, although weak, is present with a correlation coefficient rs = −0.23 between R5007 and , while the R5007 versus EW (Fe ii) anticorrelation is moderately strong with rs = −0.32. The anticorrelations become stronger with rs = −0.37 for R5007 versus and rs = −0.45 for R5007 versus EW (Fe ii) when sources with continuum S/N > 10 pixel−1 are considered. The correlation found here is consistent with those in previous studies. For example, Boroson & Green (1992) found rs = −0.36 for the R5007 versus relation and −0.52 for the R5007 versus EW (Fe ii) relation. Similarly, McIntosh et al. (1999) found rs = −0.43 for the R5007 versus relation and rs = −0.54 for the R5007 versus EW (Fe ii) relation, while Grupe et al. (1999) found rs = −0.36 for the R5007 versus EW (Fe ii) relation. Therefore, the strong Fe ii sources may have weak [O iii] emission or vice versa.
Download figure:
Standard image High-resolution image5.2. Black Hole Mass Measurement
The value of MBH in an AGN can be estimated using the virial relation from a single-epoch spectrum for which continuum luminosity and line width measurements are available as follows:
where A and B are the constants that have been empirically calibrated by various authors using the reverberation mapping R–L relations of local AGN (Kaspi et al. 2000, 2005; Bentz et al. 2013) as well as internally calibrated based on the availability of multiple strong emission lines such as Mg ii with Hβ (e.g., McLure & Jarvis 2002; McLure & Dunlop 2004; Vestergaard & Peterson 2006; Shen et al. 2011; Woo et al. 2018) and C iv with Mg ii or Hβ (e.g., Vestergaard & Peterson 2006; Assef et al. 2011; Park et al. 2017). In this work, we used the black hole mass calibrations from Vestergaard & Peterson (2006, thereafter VP06), Assef et al. (2011, thereafter A11), Vestergaard & Osmer (2009, thereafter VO09) and S11.
We caution that the derived virial MBH values could have an uncertainty >0.4 dex due to the different systematics (e.g., different line width measures, geometry, and kinematics of the BLR, see Collin et al. 2006; Shen 2013) involved in the calibrations used, which have not been taken into account. The uncertainties in the virial MBH values that are provided in the catalog are only the measurement uncertainties calculated via error propagation of Equation (8). In Figure 16 we compare MBH calculated using different lines. We plot Mg ii based black hole masses against Hβ (upper left) and the ratio of the two masses (lower panels). Both S11 and VO09 provide consistent Mg ii based masses with negligible offsets between Mg ii and Hβ based masses, but with a dispersion of ∼0.3. In the right panel, we compare C iv based masses against Mg ii based masses. Here, we note a larger offset and dispersion in the ratio of C iv to Mg ii based masses compared to Mg ii to Hβ based mass ratio. In the catalog, we provide "fiducial" virial MBH values calculated based on (a) the Hβ line (for z < 0.8) using the calibration of VP06, (b) the Mg ii line (for 0.8 ≤ z < 1.9) using the calibration provided by VO09, and (c) the C iv line (for z ≥ 1.9) using the VP06 calibration.
Download figure:
Standard image High-resolution imageWe estimated the Eddington ratio (REdd), which is the ratio of the bolometric (see Section 2.3) to Eddington luminosity (LEdd). The values of MBH and the Eddington ratios for all quasars are also provided in the catalog. In Figure 17 we plot REdd against MBH for sources with quality flag = 0. First, the accretion rate decreases with increasing MBH, and highly accreting quasars with massive black holes are rare, which is expected considering that REdd is inversely proportional to MBH, which increases with line width. Second, low accreting and low-mass black holes are also rare due to the flux limit of the SDSS. The distribution of REdd and MBH is also plotted. The median of the distribution is , having a range of 7.1–9.9 M⊙ (3σ around the median), and the distribution has a median of with a range of −2.06 to 0.43.
Download figure:
Standard image High-resolution image6. Summary
We have carried out detailed spectral decompositions that include host-galaxy subtraction, AGN continuum, and emission line modeling for more than 500,000 quasar spectra from the SDSS DR14 quasar catalog. We also estimated the spectral properties such as line flux, FWHM, and wavelength shift. We estimated the virial MBH and Eddington ratio for the quasars. We performed various correlation analyses to show the applicability of the measurements presented in this work in a larger context. The strong Fe ii emitters with larger line FWHM and highly accreting high-mass sources are found to be rare in this large sample of quasars. In particular, we found the well-known inverse correlation between EW and continuum flux in C iv and Mg ii, and the strong correlation between Balmer line and continuum luminosity. We provide all the measurements in the form of a catalog, which is the largest catalog containing the spectral properties of quasars so far. This catalog will be of immense use to the community in the study of various quasar properties.
We thank the referee for useful suggestions that significantly improved the clarity of the manuscript. We thank Hengxiao Guo for making the spectral fitting code PyQSOFit publicly available and for useful discussions. J.K. acknowledges financial support from the Academy of Finland, grant 311438. S.R. thanks Neha Sharma (KHU, South Korea) for carefully reading the manuscript. Throughout this work, we have used the FGCI-cluster Titan. This work has made use of SDSS spectroscopic data. Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy Office of Science. The SDSS-III website is http://www.sdss3.org/. SDSS-III is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS-III Collaboration, including the University of Arizona, the Brazilian Participation Group, Brookhaven National Laboratory, Carnegie Mellon University, University of Florida, the French Participation Group, the German Participation Group, Harvard University, the Instituto de Astrofisica de Canarias, the Michigan State/Notre Dame/JINA Participation Group, Johns Hopkins University, Lawrence Berkeley National Laboratory, Max Planck Institute for Astrophysics, Max Planck Institute for Extraterrestrial Physics, New Mexico State University, New York University, Ohio State University, Pennsylvania State University, University of Portsmouth, Princeton University, the Spanish Participation Group, University of Tokyo, University of Utah, Vanderbilt University, University of Virginia, University of Washington, and Yale University.
Software: PyQSOFit (Guo et al. 2018), TOPCAT (Taylor 2005), NumPy (van der Walt et al. 2011), SciPy (Virtanen et al. 2020), Matplotlib (Hunter 2007), Astropy (The Astropy Collaboration et al. 2013), linmix (Kelly 2007), BCES (Akritas & Bershady 1996), and Kapteyn (Terlouw & Vogelaar 2014).
Appendix A: Quality Flag
We provide quality flags on the various spectral quantities in the catalog to access the reliability of the measurements following Calderone et al. (2017). The quality flags are an integer number that is calculated as + + +...+, where "Bits" can have values of 0 (no flag raised) or 1 (flag raised). Therefore, a quality flag of zero means all Bits are zero and the associated quantity is reliable, while a flag >0 means that the associated quantity should be used with caution. Below we tabulate the quality flag statistics and mention the criteria used to set these quality flags for continuum and line quantities as footnotes in Tables A1, A2, and A3. Here we summarize the criteria we used to define the flag.
Table A1. Quality Flag on Host-galaxy Decomposition Using PCA
Quality flag | Number of sources with host decomposition |
---|---|
No PCAa | 513,458 |
PCAb | 12,807 |
Good qualityc | 12,674 (99.0%) |
Bit_0d | 126 (1.0%) |
Bit_1e | 7 (0.1%) |
Notes. The criteria used to set the quality flags are also given in the footnote.
adecomposition is not applied. bdecomposition is applied. call bits set to zero. dfraction of host flux to total flux is higher than 100% at 4200 Å or 5100 Å. ereduced χ2 of host-galaxy decomposition >15 and fraction of host >0.3.Download table as: ASCIITypeset image
Table A2. Quality Flag Statistics for Continuum Luminosity
Quality flag | continuum luminosity | |||
---|---|---|---|---|
L5100 | L4400 | L3000 | L1350 | |
No Conta | 436,243 | 379,399 | 118,583 | 234,287 |
Contb | 90,022 | 146,866 | 407,682 | 291,978 |
Good qualityc | 88821 (98.7%) | 145,586 (99.1%) | 405,085 (99.4%) | 290,334 (99.4%) |
Bit_0d | 103 (0.1%) | 109 (0.1%) | 281 (0.1%) | 211 (0.1%) |
Bit_1e | 6 (0.0%) | 7 (0.0%) | 234 (0.1%) | 285 (0.1%) |
Bit_2f | 149 (0.2%) | 169 (0.1%) | 272 (0.1%) | 171 (0.%) |
Bit_3g | 388 (0.4%) | 425 (0.3%) | 694 (0.2%) | 378 (0.1%) |
Bit_4h | 909 (1.0%) | 947 (0.6%) | 1915 (0.5%) | 1095 (0.4%) |
Bit_5i | 7 (0.0%) | 14 (0.0%) | 39 (0.0%) | 75 (0.0%) |
Notes. The criteria used to set the quality flags are also given in the footnote.
awavelength is outside the observed range. bwavelength is inside the observed range. call bits are set to zero. If PCA flag is >0, then a value of 1000 is added to the final continuum flag. dluminosity or its uncertainty is zero or NaN. erelative uncertainty of luminosity >1.5. fslope or its uncertainty is zero or NaN. gslope hits a limit in the fit. hslope uncertainty >0.3. ireduced χ2 of the continuum fit >50.Download table as: ASCIITypeset image
Table A3. Quality Flag Statistics for Different Emission Lines
Quality Flag | Emission Lines | ||||||
---|---|---|---|---|---|---|---|
Hα Broad | Hβ Broad | Hγ Broad | Mg ii Broad | C iii | C iv | Lyα | |
No Linea | 515,758 | 436,306 | 383,649 | 107,258 | 89,887 | 175,854 | 318,884 |
Lineb | 10,507 | 89,959 | 142,616 | 419,007 | 436,378 | 350,411 | 207,381 |
Good qualityc | 8115 (77.2%) | 63,312 (70.4%) | 51,588 (36.2%) | 309,166 (73.8%) | 312,896 (71.7%) | 290,053 (82.8%) | 150,779 (72.7%) |
Bit_0d | 1112 (10.6%) | 18,179 (20.2%) | 84024 (58.9%) | 60,226 (14.4%) | 85,667 (19.6%) | 20,586 (5.9%) | 9829 (4.7%) |
Bit_1e | 1378 (13.1%) | 3204 (3.6%) | 22,431 (15.7%) | 46,903 (11.2%) | 16,234 (3.7%) | 25,474 (7.3%) | 18,884 (9.1%) |
Bit_2f | 23 (0.2%) | 521 (0.6%) | 5801 (4.1%) | 637 (0.2%) | 134 (0.1%) | 410 (0.1%) | 451 (0.2%) |
Bit_3g | 1378 (13.1%) | 3204 (3.6%) | 22,431 (15.7%) | 46,905 (11.2%) | 16,234 (3.7%) | 25,474 (7.3%) | 18,884 (9.1%) |
Bit_4h | 1406 (13.4%) | 3826 (4.3%) | 35,091 (24.6%) | 47,192 (11.3%) | 36,912 (8.5%) | 31504 (9.0%) | 36,125 (17.4%) |
Bit_5i | 43 (0.4%) | 1639 (1.8%) | 9175 (6.4%) | 15,157 (3.6%) | 27,241 (6.2%) | 6533 (1.9%) | 5005 (2.4%) |
Bit_6j | 22 (0.2%) | 741 (0.8%) | 17,627 (12.4%) | 291 (0.1%) | 146 (0.1%) | 525 (0.2%) | 275 (0.1%) |
Bit_7k | 62 (0.6%) | 2138 (2.4%) | 30,612 (21.5%) | 11,230 (2.7%) | 6846 (1.6%) | 4202 (1.2%) | 8548 (4.1%) |
Bit_8l | 409 (3.9%) | 18,538 (20.6%) | 76,873 (53.9%) | 74,823 (17.9%) | 45,939 (10.5%) | 23637 (6.7%) | 12,964 (6.3%) |
Notes. The criteria used to set the quality flags are also given in the footnote.
aline-fitting window is outside the observed range. bline-fitting window is inside the observed range. call bits are set to zero. drelative uncertainty of peak flux >1/3. eluminosity or its uncertainty is zero or NaN. frelative uncertainty of luminosity >1.5. gFWHM or its uncertainty is zero or NaN. hFWHM ≤910 km s−1. irelative uncertainty of FWHM >2. jvelocity offset or its uncertainty is zero or NaN. kvelocity offset value hits lower or upper limit in the fit. luncertainty in velocity offset >1000 km s−1.Download table as: ASCIITypeset image
The quality flags are assigned based on the measured quantities and their uncertainties. For emission line quantities, if line PEAK_FLUX <3× PEAK_FLUX_ERR, i.e., the relative uncertainty in PEAK_FLUX_ERR/PEAK_FLUX >1/3, a bit is assigned to have a value of 1 and the quantity for a given line is unreliable. Moreover, if the line luminosity, FWHM, and velocity offsets and their uncertainties = 0 or infinite, the associated bits have a value of 1. Moreover, the relative uncertainty (i.e., the ratio of the uncertainty to the reported value) in line luminosity >1.5 and FWHM > 2, and uncertainties in the velocity offsets >1000 km s−1 are considered for the associated bits to have a value of 1. Similarly, for the PCA decomposition, if the host fraction at 4200 Å or 5100 Å is >100, the PCA decomposition is considered to be unreliable and assigned to have a value of 1. Moreover, if the reduced χ2 > 15 and host fraction is >0.3, a bit equal to 1 is assigned. For the AGN continuum luminosity, if the luminosity or continuum slope and their uncertainty is zero or infinite, a bit is assigned to a value of 1. We also considered sources with a relative uncertainty on the continuum luminosity >1.5 and slope >0.3, and reduced- χ2 > 50 in the continuum fit to have a value of 1.
Appendix B: Catalog Format and Column Information
We provide two catalogs14 :
- 1.The main catalog ("dr14q_spec_prop.fits") is based on the spectral information from this study, consisting of 274 columns that are described in Table B1.
- 2.
Table B1. FITS Catalog (dr14q_spec_prop.fits) Description and Column Information of the Spectral Catalog: (1) FITS Column Number, (2) Name of Columns, (3) Format (4) unit, and (5) Description
Number | Column Name | Format | Unit | Description |
---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) |
1 | SDSS_NAME | String | The DR14 object designation as given in the | |
DR14 quasar catalog | ||||
2 | R.A. | Double | Degree | R.A. (J2000) |
3 | Decl. | Double | Degree | decl. (J2000) |
4 | SDSS_ID | String | PLATE-MJD-FIBER | |
5 | PLATE | Long | SDSS plate number | |
6 | MJD | Long | MJD when spectrum was observed | |
7 | FIBERID | Long | SDSS fiber | |
8 | REDSHIFT | Double | Redshift | |
9 | SN_RATIO_CONT | Double | Continuum median S/N per pixel estimated at wavelength | |
around 1350, 2245, 3000, 4210, and 5100 Å depending on the | ||||
spectral coverage | ||||
10 | MIN_WAVE | Double | Å | Minimum wavelength of the rest frame spectrum |
11 | MAX_WAVE | Double | Å | Maximum wavelength of the rest frame spectrum |
12 | PL_NORM | Double | erg s−1 cm | Normalization parameter-AGN power law |
13 | PL_NORM_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PL_NORM |
14 | PL_SLOPE | Double | Slope of AGN power law | |
15 | PL_SLOPE_ERR | Double | Measurement error in PL_SLOPE | |
16 | CONT_RED_CHI2 | Double | Reduced χ2 of the continuum fitting | |
17 | HOST_FR_4200 | Double | Fraction of host-galaxy flux with respect to the total flux | |
at 4200 Å | ||||
18 | HOST_FR_5100 | Double | same as Col. 17, but at 5100 Å | |
19 | PCA_RED_CHI2 | Double | Reduced χ2 of the PCA decomposition | |
20 | QUALITY_PCA | Double | Quality flag of PCA decomposition | |
21 | LOG_L1350 | Double | erg s−1 | Logarithmic continuum luminosity at rest-frame 1350 Å |
22 | LOG_L1350_ERR | Double | erg s−1 | Measurement error in LOG_L1350 |
23 | QUALITY_L1350 | Double | Quality flag of L1350 measurement | |
24 | LOG_L3000 | Double | erg s−1 | Logarithmic continuum luminosity at rest-frame 3000 Å |
25 | LOG_L3000_ERR | Double | erg s−1 | Measurement error in LOG_L3000 |
26 | QUALITY_L3000 | Double | Quality flag of L3000 measurement | |
27 | LOG_L4400 | Double | erg s−1 | Logarithmic continuum luminosity at rest-frame 4400 Å |
28 | LOG_L4400_ERR | Double | erg s−1 | Measurement error in LOG_L4400 |
29 | QUALITY_L4400 | Double | Quality flag of L4400 measurement | |
30 | LOG_L5100 | Double | erg s−1 | Logarithmic continuum luminosity at rest-frame 5100 Å |
31 | LOG_L5100_ERR | Double | erg s−1 | Measurement error in LOG_L5100 |
32 | QUALITY_L5100 | Double | Quality flag of L5100 measurement | |
33 | FBC_FR_3000 | Double | Fraction of Balmer continuum to total continuum at 3000 Å | |
34 | LOGL_FE_UV | Double | erg s−1 | Logarithmic luminosity of the UV Fe ii complex |
within the 2200-3090 Å | ||||
35 | LOGL_FE_UV_ERR | Double | erg s−1 | Measurement error in LOGL_FE_UV |
36 | LOGL_FE_OP | Double | erg s−1 | Logarithmic luminosity of the optical Fe ii complex |
within the 4435-4685 Å | ||||
37 | LOGL_FE_OP_ERR | Double | erg s−1 | Measurement error in LOGL_FE_OP |
38 | EW_FE_UV | Double | Å | Rest-frame EW of UV Fe ii complex |
within the 2200-3090 Å | ||||
39 | EW_FE_UV_ERR | Double | Å | Measurement error in EW_FE_UV |
40 | EW_FE_OP | Double | Å | Rest-frame EW of optical Fe ii complex |
within the 4435-4685 Å | ||||
41 | EW_FE_OP_ERR | Double | Å | Measurement error in EW_FE_OP |
42 | LINE_NPIX_HA | Double | Number of good pixels for the rest-frame 6400-6765 Å | |
43 | LINE_MED_SN_HA | Double | Median S/N per pixel for the rest-frame 6400-6765 Å | |
44 | LINE_NPIX_HB | Double | Number of good pixels for the rest-frame 4750-4950 Å | |
45 | LINE_MED_SN_HB | Double | Median S/N per pixel for the rest-frame 4750-4950 Å | |
46 | LINE_NPIX_HG | Double | Number of good pixels for the rest-frame 4280-4400 Å | |
47 | LINE_MED_SN_HG | Double | Median S/N per pixel for the rest-frame 4280-4400 Å | |
48 | LINE_NPIX_MG ii | Double | Number of good pixels for the rest-frame 2700-2900 Å | |
49 | LINE_MED_SN_MG ii | Double | Median S/N per pixel for the rest-frame 2700-2900 Å | |
50 | LINE_NPIX_C iii | Double | Number of good pixels for the rest-frame 1850-1970 Å | |
51 | LINE_MED_SN_C iii | Double | Median S/N per pixel for the rest-frame 1850-1970 Å | |
52 | LINE_NPIX_C iv | Double | Number of good pixels for the rest-frame 1500-1600 Å | |
53 | LINE_MED_SN_C iv | Double | Median S/N per pixel for the rest-frame 1500-1600 Å | |
54 | LINE_NPIX_LYA | Double | Number of good pixels for the rest-frame 1150-1290 Å | |
55 | LINE_MED_SN_LYA | Double | Median S/N per pixel for the rest -frame 1150-1290 Å | |
56 | LYA_LINE_STATUS | Long | Line-fitting statusa in Lyα fitting | |
57 | LYA_LINE_CHI2 | Double | χ2 in Lyα fitting | |
58 | LYA_LINE_RED_CHI2 | Double | Reduced χ2 in Lyα fitting | |
59 | LYA_NDOF | Long | Degrees of freedom in Lα fitting | |
60 | C iv_LINE_STATUS | Long | Line-fitting status in C iv fitting | |
61 | C iv_LINE_CHI2 | Double | χ2 in Civ fitting | |
62 | C iv_LINE_RED_CHI2 | Double | Reduced χ2 in C iv fitting | |
63 | C iv_NDOF | Long | Degrees of freedom in C iv fitting | |
64 | C iii_LINE_STATUS | Long | Line-fitting status in C iii fitting | |
65 | C iii_LINE_CHI2 | Double | χ2 in C iii fitting | |
66 | C iii_LINE_RED_CHI2 | Double | Reduced χ2 in C iii fitting | |
67 | C iii_NDOF | Long | Degrees of freedom in C iii fitting | |
68 | MG ii_LINE_STATUS | Long | Line-fitting status in Mg ii fitting | |
69 | MG ii_LINE_CHI2 | Double | χ2 in Mg ii fitting | |
70 | MG ii_LINE_RED_CHI2 | Double | Reduced χ2 in Mg ii fitting | |
71 | MG ii_NDOF | Long | Degrees of freedom in Mg ii fitting | |
72 | HG_LINE_STATUS | Long | Line-fitting status in Hγ fitting | |
73 | HG_LINE_CHI2 | Double | χ2 in Hγ fitting | |
74 | HG_LINE_RED_CHI2 | Double | Reduced χ2 in Hγ fitting | |
75 | HG_NDOF | Long | Degrees of freedom in Hγ fitting | |
76 | HB_LINE_STATUS | Long | Line-fitting status in Hβ fitting | |
77 | HB_LINE_CHI2 | Double | χ2 in Hβ fitting | |
78 | HB_LINE_RED_CHI2 | Double | Reduced χ2 in Hβ fitting | |
79 | HB_NDOF | Long | Degrees of freedom in Hβ fitting | |
80 | HA_LINE_STATUS | Long | Line-fitting status in Hα fitting | |
81 | HA_LINE_CHI2 | Double | χ2 in Hα fitting | |
82 | HA_LINE_RED_CHI2 | Double | Reduced χ2 in Hα fitting | |
83 | HA_NDOF | Long | Degrees of freedom in Hα fitting | |
84 | LOGL_HA_NA | Double | erg s−1 | Logarithmic line luminosity of the Hα narrow component |
85 | LOGL_HA_NA_ERR | Double | erg s−1 | Measurement error in LOGL_HA_NA |
86 | EW_HA_NA | Double | Å | Rest-frame EW of Hα narrow component |
87 | EW_HA_NA_ERR | Double | Å | Measurement error in EW_HA_NA |
88 | FWHM_HA_NA | Double | km s−1 | FWHM of Hα narrow component |
89 | FWHM_HA_NA_ERR | Double | km s−1 | Measurement error in FWHM_HA_NA |
90 | LOGL_N ii6549 | Double | erg s−1 | Logarithmic line luminosity of N ii6549 |
91 | LOGL_N ii6549_ERR | Double | erg s−1 | Measurement error in LOGL_N ii6549 |
92 | EW_N ii6549 | Double | Å | Rest-frame EW of N ii6549 |
93 | EW_N ii6549_ERR | Double | Å | Measurement error in EW_N ii6549 |
94 | LOGL_N ii6585 | Double | erg s−1 | Logarithmic line luminosity of N ii6585 |
95 | LOGL_N ii6585_ERR | Double | erg s−1 | Measurement error in LOGL_N ii6585 |
96 | EW_N ii6585 | Double | Å | Rest-frame EW of N ii6585 |
97 | EW_N ii6585_ERR | Double | Å | Measurement error in EW_N ii6585 |
98 | LOGL_S ii6718 | Double | erg s−1 | Logarithmic line luminosity of S ii6718 |
99 | LOGL_S ii6718_ERR | Double | erg s−1 | Measurement error in LOGL_S ii6718 |
100 | EW_S ii6718 | Double | Å | Rest-frame EW of S ii6718 |
101 | EW_S ii6718_ERR | Double | Å | Measurement error in EW_S ii6718 |
102 | LOGL_S ii6732 | Double | erg s−1 | Logarithmic line luminosity of S ii6732 |
103 | LOGL_S ii6732_ERR | Double | erg s−1 | Measurement error in LOGL_S ii6732 |
104 | EW_S ii6732 | Double | Å | Rest-frame EW of S ii6732 |
105 | EW_S ii6732_ERR | Double | Å | Measurement error in EW_S ii6732 |
106 | FWHM_HA_BR | Double | km s−1 | FWHM of Hα broad component |
107 | FWHM_HA_BR_ERR | Double | km s−1 | Measurement error in FWHM_HA_BR |
108 | SIGMA_HA_BR | Double | km s−1 | Line dispersion (second moment) of Hα broad component |
109 | SIGMA_HA_BR_ERR | Double | km s−1 | Measurement error in SIGMA_HA_BR |
110 | EW_HA_BR | Double | Å | Rest-frame EW of Hα broad component |
111 | EW_HA_BR_ERR | Double | Å | Measurement error in EW_HA_BR |
112 | PEAK_HA_BR | Double | Å | Peak wavelength of Hα broad component |
113 | PEAK_HA_BR_ERR | Double | Å | Measurement error in PEAK_HA_BR |
114 | PEAK_FLUX_HA_BR | Double | erg s−1 cm−2 Å−1 | Peak flux of Hα broad component |
115 | PEAK_FLUX_HA_BR_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PEAK_FLUX_HA_BR |
116 | LOGL_HA_BR | Double | erg s−1 | Logarithmic line luminosity of Hα broad component |
117 | LOGL_HA_BR_ERR | Double | erg s−1 | Measurement error in LOGL_HA_BR |
118 | QUALITY_HA | Double | Quality of Hα line fitting | |
119 | LOGL_HB_NA | Double | erg s−1 | Logarithmic line luminosity of Hβ narrow component |
120 | LOGL_HB_NA_ERR | Double | erg s−1 | Measurement error in LOGL_HB_NA |
121 | EW_HB_NA | Double | Å | Rest-frame EW of Hβ narrow component |
122 | EW_HB_NA_ERR | Double | Å | Measurement error in EW_HB_NA |
123 | FWHM_HB_NA | Double | km s−1 | FWHM of Hβ narrow component |
124 | FWHM_HB_NA_ERR | Double | km s−1 | Measurement error in FWHM_HB_NA |
125 | LOGL_O iii4959C | Double | erg s−1 | Logarithmic line luminosity of O iii4959 core component |
126 | LOGL_O iii4959C_ERR | Double | erg s−1 | Measurement error in LOGL_O iii4959C |
127 | EW_O iii4959C | Double | Å | Rest-frame EW of O iii4959 core component |
128 | EW_O iii4959C_ERR | Double | Å | Measurement error in EW_O iii4959C |
129 | LOGL_O iii5007C | Double | erg s−1 | Logarithmic line luminosity of O iii5007 core component |
130 | LOGL_O iii5007C_ERR | Double | erg s−1 | Measurement error in LOGL_O iii5007C |
131 | EW_O iii5007C | Double | Å | Rest-frame EW of O iii5007 core component |
132 | EW_O iii5007C_ERR | Double | Å | Measurement error in EW_O iii5007C |
133 | LOGL_O iii4959W | Double | erg s−1 | Logarithmic line luminosity of O iii4959 wing component |
134 | LOGL_O iii4959W_ERR | Double | erg s−1 | Measurement error in LOGL_O iii4959W |
135 | EW_O iii4959W | Double | Å | Rest-frame EW of O iii4959 wing component |
136 | EW_O iii4959W_ERR | Double | Å | Measurement error in EW_O iii4959W |
137 | LOGL_O iii5007W | Double | erg s−1 | Logarithmic line luminosity of O iii5007 wing component |
138 | LOGL_O iii5007W_ERR | Double | erg s−1 | Measurement error in LOGL_O iii5007W |
139 | EW_O iii5007W | Double | Å | Rest-frame EW of O iii5007 wing component |
140 | EW_O iii5007W_ERR | Double | Å | Measurement error in EW_O iii5007W |
141 | LOGL_O iii4959 | Double | erg s−1 | Logarithmic line luminosity of entire O iii4959 |
142 | LOGL_O iii4959_ERR | Double | erg s−1 | Measurement error in LOGL_O iii4959 |
143 | EW_O iii4959 | Double | Å | Rest-frame EW of entire O iii4959 |
144 | EW_O iii4959_ERR | Double | Å | Measurement error in EW_O iii4959 |
145 | LOGL_O iii5007 | Double | erg s−1 | Logarithmic line luminosity of entire O iii5007 |
146 | LOGL_O iii5007_ERR | Double | erg s−1 | Measurement error in LOGL_O iii5007 |
147 | EW_O iii5007 | Double | Å | Rest-frame EW of entire O iii5007 |
148 | EW_O iii5007_ERR | Double | Å | Measurement error in EW_O iii5007 |
149 | LOGL_HE ii4687_BR | Double | erg s−1 | Logarithmic line luminosity of He ii4687 broad component |
150 | LOGL_HE ii4687_BR_ERR | Double | erg s−1 | Measurement error in LOGL_HE ii4687_BR |
151 | EW_HE ii4687_BR | Double | Å | Rest-frame EW of He ii4687 broad component |
152 | EW_HE ii4687_BR_ERR | Double | Å | Measurement error in EW_HE ii4687_BR |
153 | LOGL_HE ii4687_NA | Double | erg s−1 | Logarithmic line luminosity of He ii4687 narrow component |
154 | LOGL_HE ii4687_NA_ERR | Double | erg s−1 | Measurement error in LOGL_HE ii4687_NA |
155 | EW_HE ii4687_NA | Double | Å | Rest-frame EW of He ii4687 narrow component |
156 | EW_HE ii4687_NA_ERR | Double | Å | Measurement error in EW_HE ii4687_NA |
157 | FWHM_HB_BR | Double | km s−1 | FWHM of Hβ broad component |
158 | FWHM_HB_BR_ERR | Double | km s−1 | Measurement error in FWHM_HB_BR |
159 | SIGMA_HB_BR | Double | km s−1 | Line dispersion (second moment) of Hβ broad component |
160 | SIGMA_HB_BR_ERR | Double | km s−1 | Measurement error in SIGMA_HB_BR |
161 | EW_HB_BR | Double | Å | Rest-frame EW of Hβ broad component |
162 | EW_HB_BR_ERR | Double | Å | Measurement error in EW_HB_BR |
163 | PEAK_HB_BR | Double | Å | Peak wavelength of Hβ broad component |
164 | PEAK_HB_BR_ERR | Double | Å | Measurement error in PEAK_HB_BR |
165 | PEAK_FLUX_HB_BR | Double | erg s−1 cm−2 Å−1 | Peak flux of Hβ broad component |
166 | PEAK_FLUX_HB_BR_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PEAK_FLUX_HB_BR |
167 | LOGL_HB_BR | Double | erg s−1 | Logarithmic line luminosity of Hβ broad component |
168 | LOGL_HB_BR_ERR | Double | erg s−1 | Measurement error in LOGL_HB_BR |
169 | QUALITY_HB | Double | Quality of Hβ line fitting | |
170 | LOGL_HG_NA | Double | erg s−1 | Logarithmic line luminosity of Hγ narrow component |
171 | LOGL_HG_NA_ERR | Double | erg s−1 | Measurement error in LOGL_HG_NA |
172 | EW_HG_NA | Double | Å | Rest-frame EW of Hγ narrow component |
173 | EW_HG_NA_ERR | Double | Å | Measurement error in EW_HG_NA |
174 | LOGL_O iii4364 | Double | erg s−1 | Logarithmic line luminosity of O iii4364 |
175 | LOGL_O iii4364_ERR | Double | erg s−1 | Measurement error in LOGL_O iii4364 |
176 | EW_O iii4364 | Double | Å | Rest-frame EW of O iii4364 |
177 | EW_O iii4364_ERR | Double | Å | Measurement error in EW_O iii4364 |
178 | FWHM_HG_BR | Double | km s−1 | FWHM of Hγ broad component |
179 | FWHM_HG_BR_ERR | Double | km s−1 | Measurement error in FWHM_HG_BR |
180 | SIGMA_HG_BR | Double | km s−1 | Line dispersion (second moment) of Hγ broad component |
181 | SIGMA_HG_BR_ERR | Double | km s−1 | Measurement error in SIGMA_HG_BR |
182 | EW_HG_BR | Double | Å | Rest-frame EW of Hγ broad component |
183 | EW_HG_BR_ERR | Double | Å | Measurement error in EW_HG_BR |
184 | PEAK_HG_BR | Double | Å | Peak wavelength of Hγ broad component |
185 | PEAK_HG_BR_ERR | Double | Å | Measurement error in PEAK_HG_BR |
186 | PEAK_FLUX_HG_BR | Double | erg s−1 cm−2 Å−1 | Peak flux of Hγ broad component |
187 | PEAK_FLUX_HG_BR_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PEAK_FLUX_HG_BR |
188 | LOGL_HG_BR | Double | erg s−1 | Logarithmic line luminosity of Hγ broad component |
189 | LOGL_HG_BR_ERR | Double | erg s−1 | Measurement error in LOGL_HG_BR |
190 | QUALITY_HG | Double | Quality of Hγ line fitting | |
191 | LOGL_MG ii_NA | Double | erg s−1 | Logarithmic line luminosity of Mg ii narrow component |
192 | LOGL_MG ii_NA_ERR | Double | erg s−1 | Measurement error in LOGL_MG ii_NA |
193 | EW_MG ii_NA | Double | Å | Rest-frame EW of Mg ii narrow component |
194 | EW_MG ii_NA_ERR | Double | Å | Measurement error in EW_MG ii_NA |
195 | FWHM_MG ii_NA | Double | km s−1 | FWHM of Mg ii narrow component |
196 | FWHM_MG ii_NA_ERR | Double | km s−1 | Measurement error in FWHM_MG ii_NA |
197 | FWHM_MG ii_BR | Double | km s−1 | FWHM of Mg ii broad component |
198 | FWHM_MG ii_BR_ERR | Double | km s−1 | Measurement error in FWHM_MG ii_BR |
199 | SIGMA_MG ii_BR | Double | km s−1 | Line dispersion (second moment) of MG ii broad component |
200 | SIGMA_MG ii_BR_ERR | Double | km s−1 | Measurement error in SIGMA_MG ii_BR |
201 | EW_MG ii_BR | Double | Å | Rest-frame EW of Mg ii broad component |
202 | EW_MG ii_BR_ERR | Double | Å | Measurement error in EW_MG ii_BR |
203 | PEAK_MG ii_BR | Double | Å | Peak wavelength of Mg ii broad component |
204 | PEAK_MG ii_BR_ERR | Double | Å | Measurement error in PEAK_MG ii_BR |
205 | PEAK_FLUX_MG ii_BR | Double | erg s−1 cm−2 Å−1 | Peak flux of MG ii broad component |
206 | PEAK_FLUX_MG ii_BR_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PEAK_FLUX_MG ii_BR |
207 | LOGL_MG ii_BR | Double | erg s−1 | Logarithmic line luminosity of Mg ii broad component |
208 | LOGL_MG ii_BR_ERR | Double | erg s−1 | Measurement error in LOGL_MG ii_BR |
209 | QUALITY_MG ii | Double | Quality of MG ii line fitting | |
210 | FWHM_C iii | Double | km s−1 | FWHM of entire C iii |
211 | FWHM_C iii_ERR | Double | km s−1 | Measurement error in FWHM_C iii |
212 | SIGMA_C iii | Double | km s−1 | Line dispersion (second moment) of entire C iii |
213 | SIGMA_C iii_ERR | Double | km s−1 | Measurement error in SIGMA_C iii |
214 | EW_C iii | Double | Å | Rest-frame EW of entire C iii |
215 | EW_C iii_ERR | Double | Å | Measurement error in EW_C iii |
216 | PEAK_C iii | Double | Å | Peak wavelength of entire C iii |
217 | PEAK_C iii_ERR | Double | Å | Measurement error in PEAK_C iii |
218 | PEAK_FLUX_C iii | Double | erg s−1 cm−2 Å−1 | Peak flux of C iii |
219 | PEAK_FLUX_C iii_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PEAK_FLUX_C iii |
220 | LOGL_C iii | Double | erg s−1 | Logarithmic line luminosity of entire C iii |
221 | LOGL_C iii_ERR | Double | erg s−1 | Measurement error in LOGL_C iii |
222 | QUALITY_C iii | Double | Quality of C iii line fitting | |
223 | FWHM_C iv | Double | km s−1 | FWHM of entire C iv |
224 | FWHM_C iv_ERR | Double | km s−1 | Measurement error in FWHM_C iv |
225 | SIGMA_C iv | Double | km s−1 | Line dispersion (second moment) of entire C iv |
226 | SIGMA_C iv_ERR | Double | km s−1 | Measurement error in SIGMA_C iv |
227 | EW_C iv | Double | Å | Rest-frame EW of entire C iv |
228 | EW_C iv_ERR | Double | Å | Measurement error in EW_C iv |
229 | PEAK_C iv | Double | Å | Peak wavelength of entire C iv |
230 | PEAK_C iv_ERR | Double | Å | Measurement error in PEAK_C iv |
231 | PEAK_FLUX_C iv | Double | erg s−1 cm−2 Å−1 | Peak flux of C iv |
232 | PEAK_FLUX_C iv_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PEAK_FLUX_C iv |
233 | LOGL_C iv | Double | erg s−1 | Logarithmic line luminosity of entire C iv |
234 | LOGL_C iv_ERR | Double | erg s−1 | Measurement error in LOGL_C iv |
235 | QUALITY_C iv | Double | Quality of C iv line fitting | |
236 | FWHM_LYA | Double | km s−1 | FWHM of entire Lyα |
237 | FWHM_LYA_ERR | Double | km s−1 | Measurement error in FWHM_LYA |
238 | SIGMA_LYA | Double | km s−1 | Line dispersion (second moment) of entire LYA |
239 | SIGMA_LYA_ERR | Double | km s−1 | Measurement error in SIGMA_LYA |
240 | EW_LYA | Double | Å | Rest-frame EW of entire Lyα |
241 | EW_LYA_ERR | Double | Å | Measurement error in EW_LYA |
242 | PEAK_LYA | Double | Å | Peak wavelength of entire Lyα |
243 | PEAK_LYA_ERR | Double | Å | Measurement error in PEAK_LYA |
244 | PEAK_FLUX_LYA | Double | erg s−1 cm−2 Å−1 | Peak flux of LYA |
245 | PEAK_FLUX_LYA_ERR | Double | erg s−1 cm−2 Å−1 | Measurement error in PEAK_FLUX_LYA |
246 | LOGL_LYA | Double | erg s−1 | Logarithmic line luminosity of entire Lyα |
247 | LOGL_LYA_ERR | Double | erg s−1 | Measurement error in LOGL_LYA |
248 | QUALITY_LYA | Double | Quality of LYA line fitting | |
249 | LOGL_NV | Double | erg s−1 | Logarithmic line luminosity of Nv1240 |
250 | LOGL_NV_ERR | Double | erg s−1 | Measurement error in LOGL_NV |
251 | EW_NV | Double | Å | Rest-frame EW of Nv1240 |
252 | EW_NV_ERR | Double | Å | Measurement error in EW_NV |
253 | FWHM_NV | Double | km s−1 | FWHM of Nv1240 |
254 | FWHM_NV_ERR | Double | km s−1 | Measurement error in FWHM_NV |
255 | LOG_MBH_HB_VP06 | Double | M⊙ | Logarithmic single-epoch BH mass estimate based |
on Hβ (VP06) | ||||
256 | LOG_MBH_HB_VP06_ERR | Double | M⊙ | Measurement error in LOG_MBH_HB_VP06 |
257 | LOG_MBH_HB_A11 | Double | M⊙ | Logarithmic single-epoch BH mass estimate based |
on Hβ (A11) | ||||
258 | LOG_MBH_HB_A11_ERR | Double | M⊙ | Measurement error in LOG_MBH_HB_A11 |
259 | LOG_MBH_MG ii_VO09 | Double | M⊙ | Logarithmic single-epoch BH mass estimate based |
on Mg ii (VO09) | ||||
260 | LOG_MBH_MG ii_VO09_ERR | Double | M⊙ | Measurement error in LOG_MBH_MG ii_VO09 |
261 | LOG_MBH_MG ii_S11 | Double | M⊙ | Logarithmic single-epoch BH mass estimate based |
on Mg ii (S11) | ||||
262 | LOG_MBH_MG ii_S11_ERR | Double | M⊙ | Measurement error in LOG_MBH_MG ii_S11 |
263 | LOG_MBH_C iv_VP06 | Double | M⊙ | Logarithmic single-epoch BH mass estimate based |
on C iv (VP06) | ||||
264 | LOG_MBH_C iv_VP06_ERR | Double | M⊙ | Measurement error in LOG_MBH_C iv_VP06 |
265 | LOG_MBH | Double | M⊙ | Logarithmic fiducial single-epoch BH mass |
266 | LOG_MBH_ERR | Double | M⊙ | Measurement error in LOG_MBH |
267 | QUALITY_MBH | Double | Quality of MBH estimation (sum of the quality | |
of the continuum luminosity and the quality of the line FWHM) | ||||
268 | LOG_LBOL | Double | erg s−1 | Logarithmic fiducial bolometric luminosity |
269 | QUALITY_LBOL | Double | Quality of LBOL estimation (quality | |
of the continuum luminosity) | ||||
270 | LOG_REDD | Double | Logarithmic Eddington ratio based on | |
fiducial single-epoch BH mass | ||||
271 | QUALITY_REDD | Double | Quality of REDD estimation (sum of the quality of | |
the continuum luminosity and the quality of the line MBH) | ||||
272 | BI_C iv | Double | km s−1 | BALnicity index of the C IV absorption trough |
from Pâris et al. (2018) | ||||
273 | ERR_BI_C iv | Double | km s−1 | Measurement error in BI_C iv from Pâris et al. (2018) |
274 | BAL_FLAG | BAL flag from Shen et al. (2011) |
Notes. The unmeasurable values are indicated with −999. All the measured continuum and line spectral quantities are from the model and their uncertainties are from the Monte Carlo simulation, as mentioned in the text. The complete table is available on Zenodo (doi:10.5281/zenodo.3878152).
aAn integer number returned by the KMPFIT code (Terlouw & Vogelaar 2014), which is used in PyQSOfit to perform the nonlinear least-squares fitting. Values higher than zero can represent success (but STATUS = 5 may indicate failure to converge). More information about the fitting status can be found in https://idlastro.gsfc.nasa.gov/ftp/pro/markwardt/mpfit.proFootnotes
- 4
- 5
- 6
- 7
, flux and error in flux .
- 8
These objects have MIN_WAVE = 4000 in the catalog Table B4.
- 9
For 91 out of the 526,356 quasars in DR14Q, there are insufficient or no valid data points in the spectrum to perform spectral decomposition. These objects were therefore excluded in this work.
- 10
- 11
- 12
- 13
- 14
Online on Zenodo (doi:10.5281/zenodo.3878152) and at https://www.utu.fi/sdssdr14/.