Mean and Extreme Radio Properties of Quasars and the Origin of Radio Emission

A Thesis

Submitted to the Faculty

Drexel University

Rachael Marie Kratzer

in partial fulﬁllment of the

requirements for the degree

Doctor of Philosophy

Dedications

To Louie, my dedicated companion since 2004 iii

Acknowledgments

First and foremost, I thank my thesis adviser, Gordon Richards, for his guidance and limitless patience over the past seven years. I am truly grateful for the opportunities he has made possible.

Without his support, this thesis would certainly not exist.

I am indebted to David Goldberg for his dedicated mentorship, which extended beyond academia to life, in general. I thank Rick White for providing the FIRST completeness calculations, his object stacking code, and his radio expertise. I am grateful to my remaining committee members,

Michael Vogeley and Brigita Urbanc, for their useful feedback and encouragement throughout the development of this research. I thank my colleagues for comments on the manuscript, particularly

Bob Becker and Amy Kimball. I also acknowledge Adam Myers for help constructing the Master quasar catalog and Coleman Krawczyk for providing corrected black hole masses for my samples.

This work was supported by NSF AAG grant 1108798. Funding for the SDSS and SDSS-II was provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science

Foundation, the U.S. Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, the Max Planck Society, and the Higher Education Funding Council for England. The SDSS was managed by the Astrophysical Research Consortium for the Partic- ipating Institutions. 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 web site 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 Labo- ratory, Carnegie Mellon University, University of Florida, the French Participation Group, the Ger- man Participation Group, Harvard University, the Instituto de Astroﬁsica de Canarias, the Michigan

State/Notre Dame/JINA Participation Group, Johns Hopkins University, Lawrence Berkeley Na- tional Laboratory, Max Planck Institute for Astrophysics, Max Planck Institute for Extraterrestrial iv

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. v

Table of Contents

List of Figures ...... vii

Abstract ...... xi

1. History of Quasars ...... 1

2. Commentary on the Radio-Loud Bimodality ...... 5

3. Data ...... 9

3.1 Sloan Digital Sky Survey...... 9

3.2 FIRST...... 10

3.3 DR7 Quasar Catalog...... 11

3.4 Master Quasar Catalog...... 12

3.5 My Samples...... 13

3.6 Diagnostics ...... 14

3.6.1 Optical Luminosity and Redshift ...... 15

3.6.2 Radio-Loudness...... 16

3.6.3 Radio Luminosity...... 19

3.6.4 Extended Flux Underestimation...... 21

3.6.5 FIRST Detection Limit ...... 22

3.7 Radio Loud Deﬁnition...... 24

4. Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) ...... 27

4.1 The RLF as a function of L and z ...... 27

4.2 The RLF and Accretion Disk Winds: Principal Component and CIV Analyses . . . . 30

4.3 The RLF and Black Hole Mass...... 34

5. Radio Properties in the Mean: Stacking Analysis ...... 39

5.1 Motivation ...... 39

5.2 Image Stacking...... 39 vi

5.3 Median Stacking Diagnostics ...... 40

5.4 The Mean Radio Properties as a Function of L and z ...... 42

5.5 Radio-Optical Spectral Index as a Function of Luminosity...... 46

5.6 The Mean Radio Properties and Accretion Disk Winds: Principal Component and CIV Analyses...... 51

5.7 The Mean Radio Properties and Black Hole Mass...... 53

5.8 The Mean Radio Properties as a Function of Color...... 54

6. Discussion ...... 58

6.1 Luminosity and Redshift Evolution...... 58

6.2 Mass, Accretion Rate, and Spin...... 59

6.3 Accretion Disk Winds ...... 62

6.4 On the Meaning of Radio-Quiet...... 63

7. Conclusions ...... 65

Bibliography ...... 67

Vita ...... 78 vii

List of Figures

3.1 Histograms of the redshift distribution for Sample A (left; yellow), Sample B (left; green), Sample C (right; blue), and Sample D (right; purple). Note that the two plots use diﬀerent y-axis scalings. Sample C ﬁlls in the redshift distribution gap seen near z ∼ 2.7 in Samples A and B, while Sample D vastly increases the sample size by probing deeper than the spectroscopic samples (A,B,C) at the expense of the redshift distribution (and redshift accuracy)...... 15

3.2 Histograms of the i-band magnitude distribution for all four samples (colors are as outlined in Fig. 3.1). Again, note that the two plots use different y-axis scalings. The distribution of Sample A (left; yellow) reflects the different limiting magnitudes for z < 3 and z > 3 quasar selections in SDSS. Sample C (right; blue) demonstrates the extension to fainter magnitudes by the SDSS-III BOSS quasars, while Sample D (right; purple) probes even fainter objects (while eliminating brighter objects where the efficiency of photometric quasar selection is lower due to higher stellar density)...... 16

3.3 L K-corrected to z = 2 as a function of redshift. The cuts made to Sample A 2500 A˚ (yellow) to create Sample B (green) are evident as the removal of the faintest sources with z < 3. After making cuts to Sample D (purple) to create a relatively uniform subsample, photometric redshift degeneracies manifest themselves as bands of missing objects...... 17

3.4 Radio magnitude (t) of detected FIRST sources as a function of i. The dashed diagonal lines represent different values of the radio-loudness parameter, log R, with the boundary between RL and RQ, log R = 1, highlighted in red. It is apparent for all four samples that the radio flux limit significantly restricts the distribution of log R to predominately RL values (for radio detected sources)...... 18

3.5 Radio-loudness, log R, as a function of i for optically conﬁrmed quasars within the FIRST observing area. Quasars undetected by FIRST are assigned a ﬂux of 1 mJy to calculate log R. Quasars undetected by FIRST at i < 18.9 can be considered as RQ, but fainter objects can still be RL despite being undetected by FIRST...... 19

3.6 Radio luminosity as a function of redshift for optically conﬁrmed quasars within the FIRST observing area. Quasars undetected by FIRST are assigned a ﬂux of 1 mJy to calculate Lrad. The horizontal dashed gray line denotes a division between RL and RQ quasars employed by Jiang et al.(2007). It is evident that high-redshift quasars ( z ∼> 3.5) can simultaneously be intrinsically radio-loud and FIRST non-detections. Alternatively, FIRST non-detections for lower redshifts (z ∼< 3.5) strongly indicate that an object is RQ. 21

3.7 Source size, indicated by the ratio of the integrated to peak ﬂuxes (θ), as a function of redshift for FIRST-detected quasars...... 22

3.8 FIRST completeness (provided by R. L. White 2013, private communication) as a function of integrated radio ﬂux (mJy)...... 23 viii

3.9 Spectral energy diagram comparing the distribution of power in the radio, optical, and X-ray regimes. The black lines show the mean radio-to-UV-to-X-ray SED of a quasar −1 −1 with L equal to 30 ergs s Hz . αro and αox give the slope of the SED between 2500 A˚ the radio and optical and the optical and X-ray. αro is not universal for quasars; here I have shown the slope corresponding to the traditional division between RL (steeper slopes) and RQ (ﬂatter slopes). The red and blue lines show the range of radio slopes in that part of the SED. The solid and dashed gray curves show the mean RQ and mean RL SEDs, respectively, from Elvis et al.(1994). The dotted gray line shows a slope equivalent to log R = 3—particularly radio-loud. At the bottom of the panel I show the transmission of the 1.4GHz and i-band bandpasses at z = 0, 2, and 4 (as black, dark gray, and light gray, respectively), demonstrating that, at z = 2, the 1.4GHz and i-band bandpasses are close to 5GHz and 2500A.˚ ...... 25

4.1 RLF as a function of both L and redshift for optically detected quasars within 2500 A˚ the FIRST observing area from Sample B. The plot on the left uses no correction when computing the number of RL objects within a bin while the plot to the right uses the RL completeness correction (see Figure 3.8). Comparing the two, both plots show a declining RLF with increasing redshift and decreasing luminosity with the completeness corrected version (right) showing a less pronounced, yet still present, trend. The data to the right of the dashed red lines in both plots suﬀer the most from selection eﬀects...... 28

4.2 RLF as a function of both L and redshift for optically detected quasars within 2500 A˚ the FIRST observing area for all samples. The RL completeness correction (see Section 4.1) has been applied to all samples. The biases inherent in Samples A (top, left) & C (bottom, left) present themselves as the major jumps in RLF along the diagonal that corresponds to i ≈ 19.1 for z < 3. My least biased Sample B (top, right) conﬁrms the results of Jiang et al.(2007) by demonstrating a decrease in RLF with increasing redshift and decreasing luminosity. My deeper unbiased Sample D (bottom, right) continues the trend found in Sample C to fainter luminosities...... 29

4.3 RLF in low-z EV1 (FWHM Hβ vs. RFe II) parameter space for optically detected quasars within the FIRST observing area for all samples. The RL completeness correction (see Section 4.1) has been applied. The highest RLF bins for all of the sample are concentrated to the high FWHM Hβ-low RFe II corner of this parameter space (in agreement with Sulentic et al. 2000b)...... 32

4.4 RLF as a function of both C IV equivalent width (EQW) and blueshift for optically detected quasars within the FIRST observing area. All plots use the RL completeness correction (see Section 4.1). The RLF primarily decreases from low to high blueshift with no discernible RLF trend in EQW; see also Richards et al.(2011). Objects in the row with the lowest EQWs likely have higher RLFs due to selection eﬀects...... 34

4.5 RLF as a function of BH mass and accretion rate for quasars within the FIRST observing area for all of my samples. As with all of my RLF plots, I have corrected for RL completeness (see Section 4.1). For all of my samples, the highest RLFs are positioned in the corners of parameter space: high BH mass for the lowest accretion rates and low BH mass for the highest accretion rates. The dashed red lines show where L/LEdd = 0.01, 0.1, and 1.0 in order from top-left to bottom right. I assume LBol = 2.75L (Krawczyk 2500 A˚ et al. 2013) to calculate the Eddington ratio...... 35

LIST OF FIGURES ix

4.6 Figure 4.5, Sample B split into different populations based on the spectral line used to calculate the BH masses. Each panel effectively represents quasars within different redshift bins (see Shen et al. 2011, Sec. 3.7). For the two lowest redshift samples (Hβ and Mg II), the highest RLFs belong to the highest BH mass bins for these samples. The lowest BH mass bins in the highest redshift sample (C IV) are (apparently) the most RL. I suggest that this difference is indicative of errors in the C IV BH mass estimates. The dashed red lines from top-left to bottom right show where L/LEdd = 0.01, 0.1, and 1, in order. I assume LBol = 2.75L (Krawczyk et al. 2013) to calculate the Eddington 2500 A˚ ratio...... 37

5.1 Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of redshift (see White et al. 2007, Figure 6). The vertical dashed lines represent z = 2.7, which is the upper limit of eﬃcient SDSS optical selection...... 41

5.2 Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of i-band color (see White et al. 2007, Figure 7). As before, the strongest radio emitters are associated with the optically brightest sources...... 42

5.3 Radio to optical spectral indices (αro) of median stacked quasar bins as a function of i- band color (see White et al. 2007, Figure 12). While Figure 5.2 shows that the brightest sources in the optical are the brightest in the radio, the ratio of radio to optical ﬂux is such that the radio-loudest objects (more negative values of αro ) are associated with the faintest optical sources...... 43

5.4 Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of both redshift and L . The trend roughly follows the i-band magnitude with brighter 2500 A˚ quasars in the optical being brighter in the radio...... 44

5.5 Radio luminosities (erg s−1 Hz−1) of median stacked quasar bins as a function of both redshift and L . These bins are the same as those used in Figures 5.4. As expected, 2500 A˚ the most radio luminous sources are at high-z and have high optical luminosities with radio luminosity decreasing from the upper-right to lower-left...... 45

5.6 Radio to optical spectral indices (αro) of median stacked quasar bins as a function of both redshift and L . These bins are the same as those used in Figures 5.4 and 5.5. This 2500 A˚ trend is completely opposite to that found for the RLF (see Figure 4.2). The median stacking shows stronger radio sources (relative to the optical) with decreasing optical luminosity (at ﬁxed redshift) and with increasing redshift (at ﬁxed optical luminosity). . 46

5.7 Radio luminosity dependence on L within different redshift intervals (see White 2500 A˚ et al. 2007, Figure 10) for median stacked quasar point source bins in Sample B. The dashed red line in each plot shows where Lrad = Lopt. The best fit lines for each redshift range are shown as the different dashed black lines, and the values in the lower-right corner indicate the slopes and intercepts of these lines. Sample B shows a non-linear .90 relationship between Lrad and Lopt, as Lrad ∼ Lopt for its least biased subsample (z < 2, top-middle)...... 48

LIST OF FIGURES x

5.8 Left: The dependence of αro on L (see Steffen et al. 2006, Figure 5, top) for median 2500 A˚ stacked quasar point source bins in Sample B. Recall that more negative values of αro mean more RL; thus, the higher luminosity objects are biased to a more RL SED. Different redshift intervals are highlighted with different colors. Sources to the right of the dashed red line (log L = 30.75) have been removed when computing the linear best fit 2500 A˚ (dashed black line) so as to not artificially skew it with selection effects. Right: The dependence of αro on redshift (see Steffen et al. 2006, Figure 7, top) for the same median stacked bins as the left panel. The discrete appearance of the points in this panel is an artifact of initially binning the samples with respect to redshift. The colors correspond to the different redshift intervals from the left panel. Sources to the right of the dashed red line (z = 2) have been removed when computing the linear best fit (dashed black line). 49

5.9 The relative colors, ∆(g − i), of median stacked quasar bins as a function of both redshift and L . These bins are the same as those used in Figures 5.4, 5.5, and 5.6 and show 2500 A˚ no major diﬀerences in the amount of measured dust reddening/extinction...... 51

5.10 Radio to optical spectral indices (αro) of median stacked quasar bins in low-z EV1 (FWHM Hβ vs. RFe II) parameter space. Just as for the RLF analysis (Figure 4.3), the most RL sources are those in the top-left corners with the broadest Hβ and weakest Fe IIemission lines...... 52

5.11 Radio to optical spectral indices (αro) of median stacked quasar bins in C IV parameter space. Quasars with small C IV blueshifts are more RL, on average, than those with large C IV blueshifts. This trend, although weaker, advances in the same direction as that for the RLF analysis (Figure 4.4)...... 53

5.12 Radio to optical spectral indices (αro) of median stacked quasar bins in BH mass parameter space. The dashed red lines from top-left to bottom right show where L/LEdd = 0.01, 0.1, and 1, in order. I assume LBol = 2.75L (Krawczyk et al. 2013) to calculate the 2500 A˚ Eddington ratio. If the high-redshift BH masses are ignored because of their suspected shortcomings, then the mean radio-loudness increases towards lower Eddington ratios. This trend is consistent with the RLF (Figure 4.5)...... 54

5.13 Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of ∆(g − i). Quasars with colors redder, ∆(g − i) > 0, and bluer, ∆(g − i) < 0, than average have higher radio ﬂux densities. These results match those of White et al.(2007, Figure 14)...... 55

5.14 Radio luminosities (erg s−1 Hz−1) of median stacked quasar bins as a function of ∆(g −i). The intrinsic radio luminosity increases for redder objects...... 56

5.15 Radio to optical spectral indices (αro) of median stacked quasar bins as a function of ∆(g − i). More negative values of αro correspond to more radio-loud. Agreeing with the results of White et al.(2007, Figure 13) who used the R-parameter, radio-loudness increases (αro decreases) for redder quasars...... 57

LIST OF FIGURES xi

Abstract Mean and Extreme Radio Properties of Quasars and the Origin of Radio Emission Rachael Marie Kratzer Gordon T. Richards, Ph.D.

I investigate the evolution of both the radio-loud fraction (RLF) and, using a stacking analysis, the mean radio-loudness of quasars. I consider how these values evolve as a function of redshift and luminosity, black hole (BH) mass and accretion rate, and parameters related to the dominance of a wind in the broad emission line region. I match the FIRST source catalog to samples of luminous quasars (both spectroscopic and photometric) primarily from the Sloan Digital Sky Survey.

After accounting for catastrophic errors in BH mass estimates at high-redshift, I ﬁnd that both the

RLF and the mean radio-loudness increase for increasing BH mass and decreasing accretion rate.

Similarly, both the RLF and mean radio-loudness increase for quasars that are argued to have weaker radiation line-driven wind components of the broad emission line region. In agreement with past work, I ﬁnd that the RLF increases with increasing optical luminosity and decreasing redshift while the mean radio-loudness evolves in the exact opposite manner. This diﬀerence in behavior in the

L-z plane may indicate selection eﬀects that bias an understanding of the evolution of the RLF; deeper surveys in the optical and radio are needed to resolve this discrepancy. Finally, I argue that radio-loud (RL) and radio-quiet (RQ) quasars may be parallel sequences. Only RQ quasars at one extreme of the distribution are likely to become RL through slight diﬀerences in spin and/or merger history.

Chapter 1: History of Quasars

Quasars were ﬁrst imaged in 1959 by the 3rd Cambridge Catalog of Radio Sources (Edge et al.

1959). Although the optical counterparts to these radio objects were point sources just like stars, they possessed “peculiar color indices” and “unusual [spectral] features” that diﬀerentiated them from stars (Matthews and Sandage 1963; Greenstein and Matthews 1963). Schmidt(1963) realized that the broad emission features of 3C 273, one of the ﬁrst documented quasars, were a result of redshifting and proposed that this required these objects to inhabit the centers of other galaxies.

Schmidt and Matthews(1964) used what they thought were the few absolutes of this new class of object (e.g. odd spectral properties, high redshifts, and high luminosities) to categorize its members as quasi-stellar radio sources, which later became “quasars” (Chiu 1964).

Since the brightness of a source varies inversely as the square of its distance, the extragalactic nature of quasars necessitates that they emit massive amounts of energy. Lynden-Bell(1969) first proposed a model for the origin of this tremendous amount of energy. He explained that material falling into a black hole (BH) will do so in the form a flat and rotating accretion disk; even a fraction of the gravitational potential energy from this infalling material would be sufficient enough to supply the immense amount of energy released by quasars. These new and extremely powerful objects warranted further investigation.

Thus, interest in quasars grew, and every known radio-detected quasar corresponded to an optical object. As more optical sources were imaged and analyzed, the same properties characteristic of quasars were being found in objects lacking radio counterparts (Sandage 1965). The extremely high luminosities and odd spectra that were only observed in conjunction with radio emission led

Sandage(1965) to conclude that these objects were just versions of quasars undetected by existing radio surveys. He designated them radio-quiet quasi-stellar galaxies. Recognizing the need to classify both radio detections and non-detections as the same species, a more appropriate designation for quasars, quasi-stellar object (QSO), started being utilized. 2

With the advent of deeper and more diverse sky surveys, the number of known quasars grew.

A standard definition of radio-loud (RL) was needed to compare the population statistics between different surveys since so many optically identified quasars were lacking radio counterparts. Intro- duced by Schmidt(1970), Kellermann et al.(1989) formally defined the degree to which a source is

RL by deﬁning the radio-loudness parameter

f R = 6 cm , (1.1) f 4400 A˚

where f6 cm is the 6 cm (5 GHz) radio flux, f is the 4400 A˚ optical flux, and sources are 4400 A˚ considered RL if log R > 1. Alternatively, if the radio and optical fluxes do not depend on one another, an absolute radio flux or power is a more significant boundary between RL and radio-quiet

(RQ) quasars (Peacock et al. 1986; Miller et al. 1990; Ivezi´cet al. 2002). Goldschmidt et al.(1999)

24 −1 −1 31 −1 −1 used P5GHz = 10 W Hz sr or L20cm = 10 erg s Hz as the limit between RL and RQ.

Regardless of where or how this radio-loudness boundary is imposed, many researchers have argued that the distribution of radio ﬂux and/or radio-loudness (R) exhibits a bimodality (Strittmatter et al. 1980; Kellermann et al. 1989; Miller et al. 1990; Visnovsky et al. 1992; Goldschmidt et al. 1999), with relatively few quasars between the fraction of objects with powerful radio emission and the much larger population with very little radio emission. I will address this issue more in Section2.

Many have speculated that these two classes of QSOs must be governed by similar physical processes (Barthel 1989; Urry and Padovani 1995; Shankar et al. 2010) since they really only diﬀer in the amount of radio emission observed. Some have postulated that the two populations both emit radio waves but that the RQ population has its radio emission blocked by an impenetrable cloud

(Condon et al. 1980). It has also been proposed that the geometry of quasars is solely responsible for the observed radio emission differences (Scheuer and Readhead 1979); if the radio jet is not pointed towards Earth, the quasar could appear as though it is RQ. Still others have suggested that there really are two different types of quasars (Moore and Stockman 1984; Peacock et al. 1986; Miller et al. 1990). More recently, high black hole mass and/or low values of the mass-weighted accretion rate (L/LEdd) have been implicated as being the primary driver of the radio differences (Laor 2000;

Chapter 1: History of Quasars 3

Lacy et al. 2001; Ho 2002).

Wilson and Colbert(1995) proposed that the biggest diﬀerence between RL and RQ quasars is the rate at which the central black hole spins. Since the thermal emission from RL and RQ quasars are so similar (Neugebauer et al. 1986; Sanders et al. 1989; Steidel and Sargent 1991), their black hole masses and accretion rates must also be comparable. Richards et al.(2011) recovers this same basic conclusion after comparing the average spectra of quasars that only diﬀer in the amount of radio emission measured; the congruence of the average spectra of these two classes of quasars suggests similar spectral energy distributions (SEDs) and, thus, similar masses and accretion rates.

The remaining black hole property of spin, which is responsible for the collimation of radio jets in the presence of an accretion disk (Blandford and Znajek 1977; Blandford and Payne 1982; Blandford

1990), is the only plausible explanation for the radio diﬀerence (and may not be independent of BH mass). Unfortunately, it is extremely diﬃcult to accurately determine the spins of distant black holes.

While the exact reasons behind the diversity in radio emission of quasars have yet to be conﬁ- dently explained, many have uncovered valuable properties of these objects that aid in understanding them. My work herein follows and builds upon two of those investigations: one looking at the radio- loud tail of the population (Jiang et al. 2007) and one looking at the mean radio properties of quasars

(White et al. 2007). By presenting a unique synthesis of these two perspectives and by adding new dimensions to these analyses, both by increasing the sample sizes and by considering new parameters, I hope to further constrain the understanding of the nature of both quasars themselves and their (occasional) radio exuberance. Ultimately, the goal is to understand the production of radio emission to the extent that the radio properties of an individual quasar can be predicted from the characteristics of that quasar in other parts of the electromagnetic spectrum.

The structure of the paper is as follows. Although the existence (or lack thereof) of a bimodality in the radio-loudness distribution of quasars is somewhat tangential to my analysis, Section2 provides a thorough overview of the literature concerned with it. Section3 begins with a detailed account of the surveys from which my sources are drawn, but those familiar with the Sloan Digital Sky Survey

Chapter 1: History of Quasars 4 and FIRST data sets are encouraged to skip to Section 3.2. Section4 details the evolution of the extreme radio properties of quasars by presenting the radio-loud fraction of groups of quasars as a function of luminosity and redshift (Section 4.1), Eigenvector 1 and CIV parameters (Section 4.2), and black hole properties (Section 4.3). Section5 details the evolution of the mean radio properties of quasars by presenting the radio-loudness of groups of quasars as a function of luminosity and redshift

(Section 5.4), Eigenvector 1 and CIV parameters (Section 5.6), black hole properties (Section 5.7), and color (Section 5.8). The implications of my ﬁndings are discussed in Section6, and my results and these implications are summarized in Section7.

For the entirety of this paper I employ the accepted cosmology of a ﬂat universe with H0 =

−1 −1 70 km s Mpc ,Ωm = 0.3, and ΩΛ = 0.7 (Spergel et al. 2007). I will use the term “quasar” throughout to describe luminous active galactic nuclei (AGNs), regardless of their radio properties.

Chapter 1: History of Quasars 5

Chapter 2: Commentary on the Radio-Loud Bimodality

Radio-loudness has both been deﬁned in the absolute sense (e.g., Peacock et al. 1986) and in the relative sense (e.g., Kellermann et al. 1989); see Section 3.7. Regardless of how the radio-loud boundary is imposed, many researchers have argued that the distribution exhibits a bimodality

(e.g., Strittmatter et al. 1980; Kellermann et al. 1989; Miller et al. 1990; Visnovsky et al. 1992;

Goldschmidt et al. 1999), with fewer quasars lying between the objects with powerful radio emission and the much larger population with very little radio emission. To set the stage, it is worth spending some time reviewing the literature regarding this argument. In particular, while it would seem that the literature itself is bimodal on the question of a radio bimodality, I will argue that all of these studies are actually in reasonable agreement.

While it had been known that some 10% of AGN and, for that matter, giant elliptical galaxies were strong radio sources, both Strittmatter et al.(1980) and Kellermann et al.(1989) found the radio-loudness distribution to be bimodal; they observed relatively few quasars between the bulk of the quasar population and its radio-loud tail. Kellermann et al.(1989) speciﬁcally analyzed Very

Large Array data of 114 Palomar-Green (PG) quasars (Schmidt and Green 1983), and, while the PG quasars may not be representative of the average Sloan Digital Sky Survey quasar (Jester 2005), they are the best studied quasars, in general. Kellermann et al.(1989) further noted that the bimodality is not obviously due to beaming (see also Barvainis et al. 2005; Ulvestad et al. 2005). Beaming is the process by which electromagnetic radiation appears brighter than it inherently is because it was emitted by matter that is moving relativistically along our line of sight.

White et al.(2000) showed that the depth of the FIRST (Faint Images of the Radio Sky at

Twenty Centimeters; Becker et al. 1995) survey fills in where prior radio samples of quasars were lacking and argued that the historical (apparent) bimodality is not real. Ivezićet al.(2002) pointed out that there are selection effects due to the optical magnitude limit of the Sloan Digital Sky Survey and the radio flux limit of the FIRST survey that must be taken into consideration in such analyses. 6

In short, going deeper in the radio without also going deeper in the optical does, indeed, yield more radio-intermediate sources but without the commensurate ability to ﬁnd more RL sources. When this bias is taken into account, Ivezi´cet al.(2002) demonstrated that the data are consistent with a formal bimodality in the radio-loudness distribution.

In tallying papers for and against a radio bimodality, Cirasuolo et al.(2003a) and Cirasuolo et al.

(2003b) are two of the papers that always appear in the against column. However, the analysis in both papers shows a distribution with two modes. In Cirasuolo et al.(2003b), two components are used in their ﬁt of the distribution, and both a single Gaussian and a ﬂat distribution are rejected.

Thus, these papers would be better categorized as providing evidence in support of a bimodality yet demonstrating that there is not a barren gap between the peaks as some require.

Xu et al.(1999) and Sikora et al.(2007) demonstrate the bimodality in a diﬀerent manner using heterogeneous combinations of multiple subsamples. When radio luminosities are plotted as a function of optical luminosity (Sikora et al. 2007) or [O III] line luminosity (Xu et al. 1999), their samples split into two: a RQ sequence and a RL sequence that is 3–4 order of magnitudes louder

(see Sikora et al. 2007, Figure 1 and Xu et al. 1999, Figure 1). The diﬀerent subsamples give the perception of the sort of gap that Cirasuolo et al.(2003a,b) have argued against; however, this is largely due to the sample selection. The important thing is that, for a given optical or [O III] line luminosity, the radio luminosities span 5 or 7 orders of magnitude, respectively. As described in

Section3, my sample is nicely complementary to these investigations.

Rafter et al.(2009) also present arguments against a radio bimodality. In an attempt to make an unbiased sample, they follow the prescription of Ivezi´cet al.(2002); however, I would argue that their cutting of log R > 2 objects and removal of optical sources whose lack of FIRST radio detections require them to be RQ actually biases their sample. Accounting for these issues, their distribution is consistent with the arguments by Ivezi´cet al.(2002) for bimodality.

Mahony et al.(2012) investigate the radio luminosity distribution of an X-ray selected sample of low-redshift (broad-lined) quasars observed at 20GHz. While they argue that there is “no clear evidence for a bimodal distribution”, their distributions (both in radio luminosity and log R) would

Chapter 2: Commentary on the Radio-Loud Bimodality 7 be poorly fit by a single Gaussian component. While there is no gap in the population, the distributions are consistent with other samples and generally exhibit two modes. Moreover, the findings are consistent with the argument by Xu et al.(1999) that solely using the core radio emission to compute radio-loudness minimizes the differences between the RL and RQ distributions.

Singal et al.(2013) notably take a diﬀerent approach by looking at the optical and radio quasar luminosity functions separately. However, they use the Sloan Digital Sky Survey 7th Data Release quasars without limiting to the “uniform” sample, which was designed for statistical analyses; see

Richards et al.(2006) and Section 3.3. As a result, they include objects selected via the “serendipity” branch of the quasar target selection algorithm (Stoughton et al. 2002); these sources are radio-biased both explicitly by radio-selection and implicitly through X-ray selection (Miller et al. 2011). That issue aside, Singal et al.(2013) ﬁnd that radio-loudness increases with redshift and note that their

ﬁndings are inconsistent with Jiang et al.(2007). However, Jiang et al.(2007) investigate the radio- loud fraction and not the mean radio loudness, so there is no contradiction. Indeed, White et al.

(2007) similarly found an increase in the mean radio-loudness with redshift. I shall explore this point again in Section5.

Singal et al.(2013) further ﬁnd no radio bimodality in the radio-loudness distribution; however, their analysis is limited to radio-detected objects, which severely limits their ability to probe the full

RQ distribution. This restriction to radio-detections would be appropriate where non-detections do not distinguish between RL and RQ. However, Jiang et al.(2007) limit their analysis to i < 18.9 where a non-detection by FIRST is virtually equivalent (modulo incompleteness at the FIRST detection limit, see Section 3.6.5) to being radio-quiet. I would therefore argue that their analysis is not able to accurately test for a bimodality in the radio-loudness distribution.

Balokovi´cet al.(2012) present the best summary of the question of a radio-loud bimodality in quasars. Using Monte Carlo simulations, they show that the radio-loudness probability distribution function is consistent with radio luminosity being dependent upon optical luminosity and inconsistent with a single component model. While they were not able to conﬁrm or reject the hypothesis of the distribution being formally bimodal, the important result is an empirical dichotomy. That is, two

Chapter 2: Commentary on the Radio-Loud Bimodality 8 components are needed to ﬁt the distribution, even if there is not a clear minimum between those components. Indeed, no recent analyses have actually argued for a desert at intermediate radio- loudness; whether the dichotomous distribution is additionally bimodal is a matter of semantics.

I would agree with White et al.(2007) that the radio-loudness distribution is indeed double- peaked but that the dip between the peaks is more modest than the standard binary RL/RQ classification suggests. Arguments contradicting the bimodal nature of the distribution generally are either based on data that actually does show a bimodal distribution or that is analyzed in a biased manner as emphasized by Ivezićet al.(2002, Section 4.2). Nevertheless, the distinction is not very large and is subject to a number of biases due to redshift, limiting magnitudes in both the optical and radio, and inherent selection effects in the quasar population.

I suggest that there is little utility for further discussion about whether the population is bimodal or not without deeper data (over a suﬃcient area) in both the optical and the radio. Going deeper in the optical while maintaining the FIRST depth will artiﬁcially enhance the bimodality as the only new sources will have log R > 2. Similarly going deeper in the radio while maintaining the Sloan

Digital Sky Survey depth will necessarily ﬁll in the radio-intermediate and radio-quiet populations, artiﬁcially reducing any true bimodality. Only by going deeper at both wavelengths can more progress be made. As such, instead of further analysis of the shape of the radio-loudness distribution, in this paper I will focus on extending the demographics used to investigate the radio properties of quasars, providing new constraints on the problem.

Chapter 2: Commentary on the Radio-Loud Bimodality 9

Chapter 3: Data

3.1 Sloan Digital Sky Survey

The Sloan Digital Sky Survey (SDSS; York et al. 2000) is an optical survey that has mapped more than 10,000 square degrees of sky located in the northern galactic hemisphere and partially along the Celestial Equator. Photometry was performed with a wide-ﬁeld 2.5 m telescope (Gunn et al.

2006) in ﬁve magnitude bands between 3,000 and 10,000 A(˚ ugriz; Fukugita et al. 1996; Gunn et al.

1998). Spectra between 3,800 to 9,200 A˚ were also gathered with a pair of double spectrographs.

SDSS collected an unprecedented amount of data and special software procedures were needed to efficiently reduce it. Reducing the massive amounts of photometric data entails: adjusting defective data (i.e. cosmic rays and dead pixels), ascertaining the contributions of noise to the measured brightnesses from within the charge-coupled device (CCD) as well as from the Earth’s atmosphere, calculating the point-spread functions of the CCD array as a function of time and location, pin- pointing objects of interest, combining the data from the five optical bands, fitting simple models to each located object, separating the images of objects that overlap, and determining the positions, magnitudes, and shapes of these objects (York et al. 2000; Hogg et al. 2001; Lupton et al. 2001).

Refinements were made to the astrometry (Pier et al. 2003) and photometry (Ivezićet al. 2004) of sources as the properties specific to this survey became apparent. The photometric data are corrected for Galactic dust reddening using the maps from Schlegel et al.(1998) and are reported in terms of the AB magnitude scale (Oke and Gunn 1983).

The spectroscopic data are automatically reduced by the spectroscopic pipeline (Stoughton et al.

2002) which extracts, corrects, and calibrates the spectra, determines the spectral types, and mea- sures the redshifts. The reduced spectra are then stored in the operational database.

My primary optical data set consists of objects from the SDSS Seventh Data Release (DR7;

Abazajian et al. 2009), but I will also make use of the quasar data from the continuation project

(SDSS-III; Pˆariset al. 2012) as discussed in Section 3.4. 10

3.2 FIRST

The Very Large Array (VLA) FIRST survey (Faint Images of the Radio Sky at Twenty Centimeters;

Becker et al. 1995) covers about the same sky area as SDSS. FIRST radio ﬂuxes were obtained in the

VLA’s B-conﬁguration at 20 cm (1.4 GHz). Images of the radio sky were taken for 165 seconds each with an angular resolution of 500, a typical RMS sensitivity of 0.15 mJy bm−1, and an approximate threshold ﬂux density of 1.0 mJy bm−1. The 2012 February 16 catalog contains over 946,000 sources, but only a fraction of these sources can be matched to known quasars and are studied in

Section4. Additionally, 99.9% of the FIRST pointings are blank sky (White et al. 2007), and these measurements will be used to perform stacking analyses of quasars in Section5.

I initially matched each optically conﬁrmed quasar to the peak ﬂux density of the closest radio source within 1.005, but this technique would only be robust if all of my radio sources were unresolved. Although only about 5% of matched optical-radio sources include lobes and less than 10% of

SDSS-FIRST quasars have radio morphologies other than point sources1, these radio ﬂuxes must be underestimates due to the resolving out of the extended emission at faint magnitudes and/or high redshift; therefore, I will attempt to account for a wider variety of radio emission conﬁgurations

(core, lobe, etc.). In order to avoid systematically underestimating the total luminosity of resolved objects that may have a faint radio core but bright extended radio lobes, I choose to use the total integrated ﬂux of all radio components associated with each optically conﬁrmed quasar for my radio-loud fraction analysis in Section4.

I have followed the approach of Jiang et al.(2007) to find the total integrated radio flux associated with each optical source. Optically confirmed quasars with more than one FIRST source within 30” are assigned total integrated radio fluxes equivalent to the sum of the individual integrated fluxes of the FIRST objects within this matching radius. If only one FIRST detection lies within 30” of an optical source, the matching radius is further constrained to 5” (in order to limit spurious contamination by random matches at >5”). The total integrated radio flux for optical sources with only one radio counterpart within 5” is simply the integrated flux of that matched radio source.

1See Ivezi´cet al.(2002), Section 3.8 for a complete discussion of the demographics of complex radio sources from FIRST.

Chapter 3: Data 11

Expanding the matching radius to 10” for single radio sources would only increase the number of core radio sources by ∼ 2.6%, so I opt for the 5” matching radius to reduce the number of false matches included in my analyses. Finally, optical sources that have only one FIRST match between 5” and 30” are considered radio non-detections. See Section 3.6 for a discussion of possible complications associated with radio measurements and how I plan to address them.

3.3 DR7 Quasar Catalog

My main quasar sample comes from the SDSS DR7 (Abazajian et al. 2009) Quasar Catalog (Schnei- der et al. 2010). It consists of 105,783 spectroscopically conﬁrmed quasars brighter than Mi = −22.0.

The majority of these objects were originally chosen according to the algorithm described by Richards et al.(2002a) for spectroscopic follow-up based on their location in SDSS color space. Low-redshift quasars (z < 3) were limited to i < 19.1 in ugri color space, while high-redshift quasars (z ≥ 3.0) were limited to i < 20.1 in griz color space. Additionally, irrespective of their location in color space, Richards et al.(2002a) included objects with FIRST point sources within 2.000 and eliminated objects with unreliable photometric data.

The goal of the SDSS quasar survey was to construct the largest possible quasar catalog to its given flux limits. As a number of different algorithms were used to select quasars and some of these algorithms changed in the early part of the survey (Stoughton et al. 2002), the quasar sample is not sufficiently uniform for statistical analyses. Section 2 of Richards et al.(2002a) discusses how the sample can be limited to a more uniform selection for the sake of statistical analysis. Approximately

60,000 quasars belong to the uniform sample; these are the objects chosen by the ﬁnal quasar target selection algorithm of Richards et al.(2002a). This ensures a more self-consistent subsample and provides a test as to whether the full quasar catalog results are biased by selection eﬀects.

However, I note that the so-called “uniform” sample was not meant to be radio uniform. The fraction of quasars selected because of their radio properties (as compared to the total number of quasars selected) is non-uniform in situations where the completeness of the optical selection is reduced. For my purposes, this is primarily over redshifts 2.2 < z < 3.5, where optical selection is rather incomplete due to confusion with the stellar locus. In this redshift region, the fraction

Chapter 3: Data 12 selected because of radio properties is artiﬁcially high. Thus, my analyses of the “uniform” sample will need to be further restricted in redshift space in order to avoid biasing the radio properties of the SDSS quasar sample. Nevertheless, the uniform sample is more radio-uniform than the full DR7 quasar sample.

Shen et al.(2011) extend the DR7 Quasar Catalog by improving upon the continuum and emission line measurements calculated by the SDSS pipeline (speciﬁcally Hα,Hβ, Mg II, and C IV). These emission line measurements from Shen et al.(2011) are implemented in my analyses reported in

Sections 4.2 and 5.6. By applying their reﬁned spectral ﬁts, Shen et al.(2011) also estimate the virial masses of the black holes powering these quasars; however, the virial masses based on the

Mg II and C IV spectral lines have since been updated by Raﬁee and Hall(2011) and Park et al.

(2013), respectively. Thus, the virial BH masses utilized for Sections 4.3 and 5.7 are updated from

Shen et al.(2011) according the prescriptions described in Equation 3 of Raﬁee and Hall(2011) and

Equation 3 of Park et al.(2013). Additionally, I used the improved redshifts of Hewett and Wild

(2010) with these samples rather than the redshifts cataloged by the SDSS.

It is worth noting the diﬀerences between the DR7 quasar sample (especially the “uniform” subsample) and the samples analyzed by Xu et al.(1999) and Sikora et al.(2007). Those two papers built samples for analysis that included a very wide range of AGN types: Seyferts, broad- lined radio galaxies, and luminous quasars (including both spiral and elliptical hosts). The SDSS quasars, while spanning the largest redshift range of any monolithic quasar survey, are actually a more homogenous sample of objects and nicely complement these broader analyses. Speciﬁcally, the

SDSS quasars generally only sample the bright end of the luminosity function and are limited to luminosities that distinguish them from lower-luminosity Seyfert AGNs. For further discussion, see

Section6.

3.4 Master Quasar Catalog

Since restricting the DR7 quasar sample to “uniform” quasars considerably reduces the number of objects in my study, I make an attempt to extend my investigations to a larger sample. Thus, the

ﬁnal dataset that I draw sources from is a “Master” Quasar Catalog compiled by Richards et al.

Chapter 3: Data 13

(2014, in prep.). It contains over 1.5 million sources, and over 250,000 of those have conﬁrming spectroscopy. This dataset is a “catalog of catalogs” consisting of sources within the SDSS-I/II/III survey areas and draws objects from the following sources:

• SDSS I/II: Schneider et al.(2010)

• 2QZ: Croom et al.(2004)

• 2SLAQ: Croom et al.(2009)

• AUS: Croom et al. (in prep.)

• AGES: Kochanek et al.(2012)

• COSMOS: Lilly et al.(2007); Trump et al.(2009)

• SDSS-III: Pˆariset al.(2012); Palanque-Delabrouille et al.(2013)

• Richards et al.(2009) Photometric Catalog

• Bovy et al.(2011) Photometric Catalog

• Papovich et al.(2006)

• Glikman et al.(2006)

• Maddox et al.(2012)

This quasar sample is, of course, highly inhomogeneous but does represent nearly every quasar known fainter than i ∼ 16 (including candidate photometric quasars) at the time of the 9th Data

Release of SDSS-III (Ahn et al. 2012) and signiﬁcantly extends the sample in terms of high-z quasars, reddened quasars, and quasars between 2.2 < z < 3.5. Because of this sample’s inhomogeneity, I will also consider more homogeneous subsamples in my analyses.

3.5 My Samples

I deﬁne four subsamples of data (denoted A, B, C, and D) for my analyses of the radio properties of quasars. Sample A is simply the entire DR7 Quasar Catalog, while Sample B is comprised of

Chapter 3: Data 14 objects from the DR7 Quasar Catalog that are flagged “uniform” as discussed above. Sample B is the most robust of my four samples, suffering from the fewest selection effects (especially when limited to z < 2.2 and i < 18.9); however, analyses of the other samples are important in order to expand the total number of sources and the redshift/luminosity ranges covered (at the expense of introducing biases).

Sample C consists of those objects from the Master Quasar Catalog that have spectroscopic redshifts. This is my largest sample of conﬁrmed quasars; however, it has the strongest selection biases and does not include sources as faint as those from Sample D.

Sample D is my attempt to create the largest possible sample while minimizing selection biases.

To increase the size of the sample and extend to fainter limits while maintaining a high level of uniformity, Sample D includes quasar candidates that were identiﬁed by both the KDE algorithm

(Richards et al. 2009) and the XDQSO algorithm (Bovy et al. 2011). Thus, two independent algorithms agreed that these objects are highly likely to be quasars. For the majority of these objects

I must rely on the photometric redshifts reported by these two catalogs; however, if spectroscopic redshifts exist, they are utilized instead. To make Sample D as robust as possible, I further limit it to those objects identiﬁed by the XDQSO algorithm as having only one signiﬁcant peak (exceeding a probability of 80%) in the photometric redshift probability distribution function.

For all of my analyses I exclude quasars that show signs of dust reddening/extinction; I do so by eliminating quasars with ∆(g − i) > 0.5; see (Richards et al. 2003). ∆(g − i) is deﬁned to remove the dependence of color on redshift (due to emission features), making it roughly equivalent to αopt, the underlying continuum, in the optical-UV part of the spectral energy distribution.

Figures 3.1 and 3.2 show histograms of the redshift and i-band magnitude distributions of Samples

A, B, C, and D. Sample B will be used for my primary analyses. The other three samples will provide guidance on how the radio properties change with redshift, luminosity, and apparent magnitude.

3.6 Diagnostics

I present diagnostics of the radio and optical properties of my quasar samples to determine how any biases might complicate an understanding of the physics of quasar radio emission through better

Chapter 3: Data 15 demographical analyses. In this section (and throughout the paper), plots are provided from all four samples for the sake of discussion. For each plot, panels A (top left), B (top right), C (bottom left), and D (bottom right) correspond to Samples A, B, C, and D, respectively.

3.6.1 Optical Luminosity and Redshift

Figure 3.3 shows the relationship between redshift and K-corrected L for all four samples 2500 A˚ described above. Two issues arise with regard to these parameters. First is the ﬂux-limited nature inherent to blind surveys: because of my samples’ ﬁxed magnitude limits, there is an inherent degeneracy between the redshifts and the luminosities of these objects. Thus, some caution is needed to ensure that, for example, an observed characteristic of high-luminosity objects is not instead an inherent characteristic of high-redshift objects. As such, in my analyses in Sections 4.1 and 5.4,I will consider the radio-loudness of quasars as a function of both of these properties simultaneously.

The optical luminosity itself is subject to its own corrections. Specifically, K-corrections need to be applied to ensure that I am comparing fluxes emitted in the same rest-frame wavelength range as opposed to fluxes received in the same observed-frame wavelength range. Traditionally, K- corrections are applied to extrapolate the power emitted by the object in its rest frame within the

ﬁlter’s bandpass. Richards et al.(2006), adopting Wisotzki(2000) and Blanton et al.(2003), argue against K-correcting to z = 0 since most quasars are not found in the local universe. To decrease

Figure 3.1: Histograms of the redshift distribution for Sample A (left; yellow), Sample B (left; green), Sample C (right; blue), and Sample D (right; purple). Note that the two plots use diﬀerent y-axis scalings. Sample C ﬁlls in the redshift distribution gap seen near z ∼ 2.7 in Samples A and B, while Sample D vastly increases the sample size by probing deeper than the spectroscopic samples (A,B,C) at the expense of the redshift distribution (and redshift accuracy).

Chapter 3: Data 16

Figure 3.2: Histograms of the i-band magnitude distribution for all four samples (colors are as outlined in Fig. 3.1). Again, note that the two plots use different y-axis scalings. The distribution of Sample A (left; yellow) reflects the different limiting magnitudes for z < 3 and z > 3 quasar selections in SDSS. Sample C (right; blue) demonstrates the extension to fainter magnitudes by the SDSS-III BOSS quasars, while Sample D (right; purple) probes even fainter objects (while eliminating brighter objects where the efficiency of photometric quasar selection is lower due to higher stellar density).

the errors associated with extrapolating from high redshift objects to z = 0, I follow Richards et al.

(2006) and K-correct closer to the median redshift of my samples, z = 2.

Lastly, I consider the redshift diﬀerences between the four subsamples (as seen in both Figures 3.3 and 3.1). Samples A and B both extend to z ≈ 5.5 with 29 < log L < 33. The cuts made to 2500 A˚ Sample A to create the uniform sample, Sample B, are apparent; most faint sources with z < 2.7 have been removed. Sample C ﬁlls in at z ∼ 2.7 but at the expense of a less homogeneous sample. While

Sample D represents my eﬀorts to create a larger, relatively uniform sample, I do so at the expense of certain redshift regions: the bands of missing objects in Sample D indicate where photometric redshift degeneracies exist.

3.6.2 Radio-Loudness

The standard deﬁnition of radio-loud is based on the ratio of radio to optical ﬂuxes according to

Equation1(Schmidt 1970; Kellermann et al. 1989). As emphasized by Ivezi´cet al.(2002, Section

4.2), the distribution of measured log R values can be significantly affected by both the optical and radio flux limits of a survey. As such, I examine the flux limits of my samples in Figure 3.4. In this plot, the total integrated radio flux (see Section 3.2) of each object is converted to a radio

Chapter 3: Data 17 magnitude, t, in terms of the AB magnitude scale (Oke and Gunn 1983). From Ivezi´cet al.(2002),

f t = −2.5 log int . (3.1) 3631 Jy

The faintest radio magnitude of t = 16.4 corresponds to the 1 mJy detection limit of FIRST.

The SDSS survey limits of i = 19.1 (for low redshift) and i = 20.2 (for high redshift) are readily seen

(Richards et al. 2002a). The lines of constant log R show that these magnitude limits have a direct eﬀect on the possible values of log R that can be measured for a given data set. For this reason,

Ivezićet al.(2002) suggests exploring the histogram of log R values in a parameter space that runs perpendicular to the lines of log R within regions that are not bounded by the apparent magnitude limits (see Figure 19 of Ivezićet al. 2002). Note that only Sample B can be considered reasonably complete to radio-loud sources at the boundary of its optical flux limit (i < 19.1).

Figure 3.3: L K-corrected to z = 2 as a function of redshift. The cuts made to Sample A (yellow) to create 2500 A˚ Sample B (green) are evident as the removal of the faintest sources with z < 3. After making cuts to Sample D (purple) to create a relatively uniform subsample, photometric redshift degeneracies manifest themselves as bands of missing objects.

Chapter 3: Data 18

Figure 3.5 illustrates a further limitation of the radio loudness parameter, log R, by showing how it depends upon optical i-band magnitude for all four samples. I calculated log R using Equation

1. If a particular object is within the FIRST observing area but has not been detected, log R is computed using the FIRST detection threshold flux of 1 mJy; this results in the artificial diagonal lines in the plots. The conventional division between RL and RQ (log R = 1; Kellermann et al. 1989) is plotted as a horizontal dashed gray line. Figure 3.5 illustrates that all of my samples exhibit RL incompleteness for objects fainter than i ≈ 18.9. At fainter magnitudes, it is quite possible for a object to be intrinsically radio-loud but remain undetected by FIRST. On the other hand, objects brighter than i = 18.9 that are not detected by FIRST should be classified as RQ even if they are eventually detected in the radio at a lower flux limit (modulo the incompleteness near the FIRST

ﬂux limits as discussed in Section 3.6.5).

Figure 3.4: Radio magnitude (t) of detected FIRST sources as a function of i. The dashed diagonal lines represent different values of the radio-loudness parameter, log R, with the boundary between RL and RQ, log R = 1, highlighted in red. It is apparent for all four samples that the radio flux limit significantly restricts the distribution of log R to predominately RL values (for radio detected sources).

Chapter 3: Data 19

3.6.3 Radio Luminosity

Alternatively, if the radio and optical fluxes do not depend on one another, an absolute radio flux or power is a more significant boundary between RL and RQ quasars (Peacock et al. 1986; Miller

24 −1 −1 et al. 1990; Ivezi´cet al. 2002). Goldschmidt et al.(1999) used P5GHz = 10 W Hz sr or

31 −1 −1 L20cm = 10 erg s Hz as the limit between RL and RQ. As such, examination of the radio luminosity distributions leads to additional biases in my samples that must be considered. Figure

3.6 illustrates how radio luminosity depends on redshift for all four samples.

As with Figure 3.5, I see that the FIRST ﬂux limit is obvious in this plot and demarcates what redshifts (as opposed to ﬂuxes) beyond which my sample is incomplete to radio-loud quasars. An alternate boundary (instead of log R) between RL and RQ quasars utilized by Jiang et al.(2007),

32.5 −1 −2 −1 L20cm = 10 erg s cm Hz , depends only on radio luminosity and is denoted with a horizontal

Figure 3.5: Radio-loudness, log R, as a function of i for optically conﬁrmed quasars within the FIRST observing area. Quasars undetected by FIRST are assigned a ﬂux of 1 mJy to calculate log R. Quasars undetected by FIRST at i < 18.9 can be considered as RQ, but fainter objects can still be RL despite being undetected by FIRST.

Chapter 3: Data 20 dashed gray line. In all four samples, RL incompleteness exists above z ∼> 3.5. That is, as in Figure

3.5, it is possible for a high-redshift quasar to be intrinsically radio-loud but still not be detected in FIRST. Thus, I should limit my most robust radio-loud fraction analysis to redshifts lower than this. Alternatively, a FIRST non-detection for lower redshifts is a strong indication that the object is radio-quiet. In Section4, I will take advantage of this fact by treating FIRST non-detections

(brighter than i = 18.9 and with z < 3) as conﬁrmed RQ objects.

Furthermore, as with the optical, the spectral indices of quasars in the radio have a fairly large range, spanning at least −1 < αν < 0. Since my samples cover a large range of redshift, they must also span a large range of the rest-frame radio spectrum. As such, K-corrections to the rest-frame wavelength are important to consider. In a manner similar to L (see Figure 3.3), I follow 2500 A˚ Richards et al.(2006) and deﬁne an equivalently z = 2 K-corrected radio luminosity as

αν Lrad −23 (1 + 2) = fint 10 (3.2) 4π LD2 (1 + z)(1+αν )

−1 −1 where Lrad is measured in erg s Hz , luminosity distance (LD) is measured in cm, integrated radio

ﬂux is measured in Jy, and the redshifts were taken from the optically detected objects. αν is the radio spectral index, and I use αν = −0.5 for the entirety of this analysis as I have a combination of ﬂat-spectrum (αν ∼ 0) and steep-spectrum (αν ∼ −1) sources in my samples. This choice of

K-correction means that any sample that uses the radio luminosity to deﬁne RL quasars will be biased towards including ﬂatter spectrum (larger α) sources at z > 2 and steeper spectrum (more negative α) sources at z < 2.

Chapter 3: Data 21

Figure 3.6: Radio luminosity as a function of redshift for optically conﬁrmed quasars within the FIRST observing area. Quasars undetected by FIRST are assigned a ﬂux of 1 mJy to calculate Lrad. The horizontal dashed gray line denotes a > division between RL and RQ quasars employed by Jiang et al.(2007). It is evident that high-redshift quasars ( z ∼ 3.5) can simultaneously be intrinsically radio-loud and FIRST non-detections. Alternatively, FIRST non-detections for < lower redshifts (z ∼ 3.5) strongly indicate that an object is RQ.

3.6.4 Extended Flux Underestimation

A serious issue to consider when using integrated ﬂuxes is that these measurements are underesti- mated for resolved FIRST sources (> 10”) (Becker et al. 1995). The analysis by Jiang et al.(2007) ignores this possible complication, asserting that these highly extended radio sources are rare and so bright that, despite the underestimation of integrated ﬂux, they will undoubtedly be considered

RL.

One way to characterize this eﬀect is to plot the ratio of the integrated to peak ﬂuxes as a

2 function of redshift. Here I use θ = (fint/fpeak) as defined by Ivezićet al.(2002), where θ > 1 for an extended source. In Figure 3.7, the effects of surface brightness dimming, which goes as (1 + z)4, are evident. Some of the apparent fall-off with redshift is simply due to the declining number of sources, but it does appear that at the highest redshifts extended sources are being preferentially

Chapter 3: Data 22 lost. The fall-off of the most extended sources with redshift suggests that my analyses do suffer from a measurable lack of extended sources beyond z ∼ 1.5. However, I emphasize that the (relatively) high-frequency of the FIRST observations already biases the sample towards unresolved objects and reiterate the claim by Ivezićet al.(2002) that the fraction of complex sources is small within the

FIRST sample. Thus, while my analyses are not complete to quasars with extended radio emission, those objects are not dominating my samples even at low redshift and should not dominate any trends with redshift.

3.6.5 FIRST Detection Limit

My ﬁnal demographical analysis involves the FIRST detection limit, speciﬁcally how much this limit varies from the nominal limit of 1 mJy and why.

The depth to which FIRST can detect a source depends on sky position; being in the proximity of a bright object and the systematic increase in noise for lower declinations complicate FIRST’s

Figure 3.7: Source size, indicated by the ratio of the integrated to peak ﬂuxes (θ), as a function of redshift for FIRST-detected quasars.

Chapter 3: Data 23

Figure 3.8: FIRST completeness (provided by R. L. White 2013, private communication) as a function of integrated radio ﬂux (mJy).

sensitivity (Becker et al. 1995). In addition, the radio detection limit for FIRST is calculated using peak fluxes; this makes it difficult to accurately account for extended sources whose radio emission could be distributed throughout various components that may or may not exceed FIRST’s detection threshold (Becker et al. 1995; White et al. 2007). Therefore, the source counts with radio fluxes near the 1 mJy detection limit are incomplete, with extended sources being the most incomplete.

The completeness of FIRST is shown in Figure 3.8 as a function of integrated ﬂux. The dots represent discrete values communicated by R. L. White (2013), and the solid line shows the linear

ﬁt between adjacent points that I used to interpolate completeness percentages. To compute the completeness eﬃciency, R.L. White (2013) used the measured size distribution of detected quasars.

Based on this figure, FIRST suffers significant incompleteness above what is normally considered the “detection limit”.

A concern is that any analysis probing fluxes close to the nominal detection limit will suffer due to the relative uncertainty of the incompleteness correction near this limit. That said, Ivezićet al.

(2002, Section 3.9) found that FIRST is not more than 13% incomplete at the NVSS (NRAO VLA

Sky Survey; Condon et al. 1998) ﬂux limit of ∼2.5 mJy, which is consistent with Figure 3.8. I will further discuss how this incompleteness could aﬀect my results in Section4.

Chapter 3: Data 24

3.7 Radio Loud Deﬁnition

Before I begin my analyses of the data, it is worthwhile to review the mean spectral energy distribution (SED) of quasars and to consider the definition of a radio-loud quasar in the context of the broader quasar SED. Figure 3.9 shows multiple quasar SEDs to help illustrate the difference between RL and RQ. A radio-loud definition based on luminosity would mean simply making a cut

−1 −1 along some constant value of the y-axis. A typical value would be at log Lrad = 32.5 ergs s Hz .

However, as discussed by Ivezićet al.(2002, Appendix C) and Balokovićet al.(2012), the radio luminosity is the best indicator of radio-loudness only if the radio and optical luminosities are not correlated. As Balokovićet al.(2012) demonstrates that these properties are indeed correlated, it is arguably more appropriate to consider the ratio of the radio and optical luminosities.

Indeed, as noted above in Section 3.6.2, the most common criterion used to classify quasars as

RL or RQ is the R parameter (see Equation1). While log R has a long history in the literature and is familiar to radio astronomers, the quasar ﬁeld has become much more dependent on multiwavelength data. As such, it is important to adopt terminology that is not speciﬁc to certain wavebands (e.g., log R in the radio or the energy index, Γ, in the X-ray) and that spans the entire electromagnetic spectrum. Given the common usage of units that are related to ergs s−1 cm−2 Hz−1,

α a logical choice is the slope in log fν vs. log ν space, α, where fν ∝ ν (as shown in Figure 3.9, except in luminosity units).

In my work I will consider the radio-to-optical spectral index, αro , rather than log R. I deﬁne

αro according to

log(f5 GHz/f ) log(f5 GHz/f ) 2500 A˚ 2500 A˚ αro = = = −0.186 log(f5 GHz/f ˚) (3.3) log(ν5 GHz/ν ) log(λ /λ5 GHz) 2500 A 2500 A˚ 2500 A˚ or more practically by considering the ratio of radio luminosity to optical luminosity. I have chosen the wavelength of 2500 A˚ because it is the same as is used in X-ray investigations for comparisons with the optical/UV and represents the i-band at z = 2. I have also chosen the frequency of 5 GHz because it is the historical value used in the radio and roughly corresponds to the frequency of the

Chapter 3: Data 25

Figure 3.9: Spectral energy diagram comparing the distribution of power in the radio, optical, and X-ray regimes. −1 −1 The black lines show the mean radio-to-UV-to-X-ray SED of a quasar with L equal to 30 ergs s Hz . αro 2500 A˚ and αox give the slope of the SED between the radio and optical and the optical and X-ray. αro is not universal for quasars; here I have shown the slope corresponding to the traditional division between RL (steeper slopes) and RQ (ﬂatter slopes). The red and blue lines show the range of radio slopes in that part of the SED. The solid and dashed gray curves show the mean RQ and mean RL SEDs, respectively, from Elvis et al.(1994). The dotted gray line shows a slope equivalent to log R = 3—particularly radio-loud. At the bottom of the panel I show the transmission of the 1.4GHz and i-band bandpasses at z = 0, 2, and 4 (as black, dark gray, and light gray, respectively), demonstrating that, at z = 2, the 1.4GHz and i-band bandpasses are close to 5GHz and 2500A.˚

1.4GHz (20 cm) FIRST data at z = 2 (see below).

The values of αro and log R are effectively equivalent if the frequencies sampled are the same, but using a slope (rise over run) instead of just the flux ratio (rise only) allows the use of data at other wavelengths/frequencies without having to apply significant corrections. In other words, αro is more

ﬂexible than log R. For the sake of backwards compatibility with previous work, the radio-to-optical spectral index, αro, can be related to the traditional log R parameter as follows (e.g. Wu et al. 2012):

Chapter 3: Data 26

f R = 5 GHz (3.4) f 4400 A˚ where

f5 GHz f5 GHz f ˚ = 4400 A (3.5) f f f 2500 A˚ 4400 A˚ 2500 A˚ and !αopt !αopt f ˚ ν ˚ λ ˚ 4400 A = 4400 A = 2500 A (3.6) f ν λ 2500 A˚ 2500 A˚ 4400 A˚ so that λ αopt 2500 A˚ log R λ 4400 A˚ αro = (3.7) log(ν5 GHz/ν ) 2500 A˚ which simpliﬁes to

" !# λ ˚ α = −0.186 log R + α log 2500 A = −0.186 [log R − 0.246α ] . (3.8) ro opt λ opt 4400 A˚

For the mean optical spectral index from Vanden Berk et al.(2001)( αopt = −0.44, needed to extrapolate between 2500 A˚ and 4000 A)˚ this corresponds to

αro = −0.186 log R − 0.020. (3.9)

To help calibrate αro to the log R system, it may help to note that the traditional RL/RQ division

(log R = 1) would be roughly αro = −0.2 and that αro = −0.6 would correspond to a very radio-loud source (log R ∼ 3).

Throughout the rest of this work I will assume that the radio and optical luminosities of quasars are correlated and, as such, will use the radio-to-optical ﬂux ratio as given by αro (rather than log R) to distinguish between RL and RQ sources with αro < −0.2 as the deﬁnition for RL quasars.

Chapter 3: Data 27

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF)

4.1 The RLF as a function of L and z

I begin my analysis by investigating the radio-loud fraction (RLF), which is the the percentage of quasars that have αro < −0.2 (log R > 1). Jiang et al.(2007) used a sample of more than

30,000 quasars to determine that the RLF increases with decreasing redshift and increasing optical luminosity. Their results may mean that the amount of radio emission with respect to that of the optical changes as a function of these two parameters, but it could also suggest that the population densities of RL and RQ quasars evolve with respect to one another.

My aim is to repeat the analysis of Jiang et al.(2007) with larger, deeper, and more uniform samples and to pair my results with my RQ analysis to more fully understand the radio emission in quasars. Since redshift and luminosity are degenerate properties in flux-limited surveys, I divide my samples into equally populated bins within L -z space; this allows me to isolate changes due to 2500 A˚ just one variable. Specifically, I first sort the quasars by redshift, dividing them into a number of slices with an equal population of quasars contained within each slice. Then I order the objects in each redshift slice by luminosity and further separate the objects into equally populated L-z bins.

Sources within a bin are ﬂagged RL if αro < −0.2, and, initially, the RLF for each bin was calculated by dividing the number of RL quasars by the total number of objects for that bin (see Figure 4.1, left). The median z and L values for each bin were used to plot the results, and the color of 2500 A˚ each bin represents the RLF.

In later sections, I will construct similarly binned samples using other observed quantities, plotting some third parameter as a color-scale at the median value of the x and y quantities.

Figure 4.1 (left) shows how the RLF of equally populated bins depends on both redshift and

L for Sample B without accounting for the incompleteness of FIRST (see Section 3.6.5). I ﬁnd 2500 A˚ that the RLF declines with increasing redshift (for a given luminosity) and decreasing luminosity (for 28

Figure 4.1: RLF as a function of both L and redshift for optically detected quasars within the FIRST observing 2500 A˚ area from Sample B. The plot on the left uses no correction when computing the number of RL objects within a bin while the plot to the right uses the RL completeness correction (see Figure 3.8). Comparing the two, both plots show a declining RLF with increasing redshift and decreasing luminosity with the completeness corrected version (right) showing a less pronounced, yet still present, trend. The data to the right of the dashed red lines in both plots suﬀer the most from selection eﬀects.

a given redshift). These results would appear to confirm the findings of Jiang et al.(2007) by using at least twice the number of sources, but I first need to account for the detection incompleteness of

FIRST to be conﬁdent in my ﬁndings.

In order to correct for the incompleteness discussed in Section 3.6.5, I weighted each RL quasar that has a measured integrated ﬂux less than 10 mJy by its corresponding completeness on my best

fit function (Figure 3.8; solid line). For example, a RL object with an integrated radio flux of 1.075 mJy (gcomplete = 0.500) counts as two RL objects since 1/gcomplete = 1/0.500 = 2. Because the completeness function drops off so quickly for integrated fluxes less than 1 mJy, all detected RL objects with values smaller than this are scaled by gcomplete = 0.10. RL quasars with integrated

ﬂuxes greater than or equal to 10 mJy always count as one RL object, and the total number of objects within each bin remains unchanged when computing the RLF.

The plot on the right of Figure 4.1 shows the dependence of the RLF on both redshift and L 2500 A˚ for Sample B after applying the completeness correction. The trend in RLF from the upper-left to the lower-right is reduced but still present. Everything to the right of the vertical red line located at z ∼ 2.5 in both plots denotes where the SDSS optical selection was very ineﬃcient; further care is required to fully understand my analysis beyond this redshift. In short, the optical selection is less complete at this redshift (compare panels a and b in Richards et al. 2006, Figure 6), so the quasars

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 29

Figure 4.2: RLF as a function of both L and redshift for optically detected quasars within the FIRST observing 2500 A˚ area for all samples. The RL completeness correction (see Section 4.1) has been applied to all samples. The biases inherent in Samples A (top, left) & C (bottom, left) present themselves as the major jumps in RLF along the diagonal that corresponds to i ≈ 19.1 for z < 3. My least biased Sample B (top, right) conﬁrms the results of Jiang et al.(2007) by demonstrating a decrease in RLF with increasing redshift and decreasing luminosity. My deeper unbiased Sample D (bottom, right) continues the trend found in Sample C to fainter luminosities.

discovered at z ∼ 2.5 are more likely to be radio sources.

The completeness-corrected RLF is shown for all four of the samples in Figure 4.2. In the left- hand panels (Samples A & C) optical selection eﬀects are apparent and result from the change in target selection strategy at i = 19.1. These two plots demonstrate the radio-loud bias of the faintest quasars in these samples for i > 19.1 and z < 3. The top-right panel (Sample B) shows the gradual trend towards smaller RLF with increasing redshift (at ﬁxed luminosity) and decreasing luminosity

(at ﬁxed redshift) that was seen by Jiang et al.(2007) with a sample that is roughly two times smaller. The bottom-right panel, using Sample D, appears to extend this trend as I have used the same color scale for all four plots.

However, I note that the trends in the right-hand panels are such that the RLF declines in the direction following decreasing i-band magnitude (see also Jiang et al. 2007, Figure 7a and Balokovi´c

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 30 et al. 2012, Figure 10). As it is not clear why an intrinsic quasar property should be a strong function of the apparent magnitude, this result must be taken with a grain of salt. As noted in Section 3.6.5, the completeness correction should be good down to a radio ﬂux of ∼2.5 mJy. However, plugging that value into Equation 5 of Ivezi´cet al.(2002) means my analysis is only robust to i = 17.9.

This is not deep enough to determine if the separate trends in L and z are real or due to 2500 A˚ incompleteness (or some other selection eﬀect). A radio survey covering a signiﬁcant fraction of the

FIRST area and to at least three times the depth of FIRST would be needed to test this eﬀect. I will come back to this issue in Section6 after analyzing stacked radio images in the same L-z parameter space in Section5.

Another issue with this type of analysis is that if the radio distribution does indeed require two components (or if it is bimodal), then it may be the case that the dividing line between the populations should change with luminosity as noted by Laor(2003). Thus, it is possible that I could be under- (or over-) stating the trends with RLF in Figure 4.2. As I cannot establish to what extent these trends are robust, I move on to looking for other demographics to provide further constraints on the nature of radio-loud emission in quasars. I will discuss the interpretation of Figure 4.2 further in Section6.

4.2 The RLF and Accretion Disk Winds: Principal Component and CIV Analyses

As noted by White et al.(2007), the problem is essentially that there is no practical way to identify from optical properties which individual quasars are likely to be radio-loud. I hope that extending my analysis to more detailed spectral properties of quasars in the optical/UV will oﬀer more insight.

Arguably the most in-depth analysis of quasar spectral properties has come from the Principal

Component Analysis (PCA) ﬁrst carried out by Boroson and Green(1992).

Boroson and Green(1992) showed that signiﬁcant new insight could be gained by examining the range of diﬀerences in quasar continua and emission lines using a PCA (or “eigenvector”) analysis.

They found that the properties of the Hβ, O III], and Fe II emission lines were well correlated with other diﬀerences seen in quasar spectra. Moreover, they found that these diﬀerences were

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 31 correlated with the radio continuum in a way that suggested that RL and RQ quasars are not

“parallel sequences” due to a lack of RQs matching the extremes of the RL sample. Boroson(2002) extended this work with a larger sample, and Brotherton and Francis(1999) and Sulentic et al.

(2000a) added additional line and continuum features to this matrix of quasar “eigenvectors”.

Sulentic et al.(2000b) showed that much of the information from the ﬁrst eigenvector of quasar properties is captured by simply looking at the FWHM of Hβ and the strength of optical Fe II emission relative to Hβ,RFe II = W(Fe II λ4570 blend)/W(HβBC). They used this diagram to divide quasars into two populations (A/B). While there is a continuum between the populations, it is useful to think of the extrema in this context. They found that RL quasars are generally isolated to Population B, whereas RQ quasars appear in both. Kruczek et al.(2011) refer to the Population

B objects as “hard-spectrum” sources and Population A objects as “soft-spectrum” sources. I will adopt this terminology especially when referring to RQ quasars where the extrema can be denoted as either hard-spectrum radio quiet (HSRQ) or soft-spectrum radio quiet (SSRQ).

In that context, I consider the radio-loud fraction of the quasars in my samples in this simpliﬁed

“Eigenvector 1 (EV1)” parameter space. Figure 4.3 shows how the RLF of equally populated bins evolves in low-redshift EV1 parameter space (FWHM Hβ vs. RFe II) for all samples. The highest

RLFs are found in the top-left (typically hard spectrum) corner of each panel, consistent with the

ﬁndings of Sulentic et al.(2000b). Importantly, there is no gradient in optical magnitude in this parameter space. As such, this result must be more fundamental than my analysis of the L-z distribution, which was subject to the vagaries of the FIRST completeness corrections.

While EV1 encodes the largest diﬀerences in otherwise similar quasar spectra, the objects which this type of analysis is based upon are necessarily low-redshift (as the spectrum must cover Hβ).

However, the mean SDSS quasar has a redshift closer to z ∼ 1.5, where the EV1 parameters are no longer included in the optical. To this end, it has been shown that the C IV emission line can be used to isolate extrema in quasar properties at high-redshift in a manner similar to EV1 at low-redshift (e.g., Sulentic et al. 2000a; Richards et al. 2002b; Sulentic et al. 2007; Richards et al. 2011). It would appear that high-redshift quasars occupy a broader parameter space than

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 32 low-redshift quasars, presumably due to a larger diversity of black hole masses and accretion rates.

Richards et al.(2011) argue that this diversity can be connected to the ability of a quasar (through its intrinsic SED) to power a strong radiation line-driven wind and that the C IV line represents an

EV1-like diagnostic.

Specifically, Richards et al.(2011) argue that the C IV emission line properties of a quasar, particularly the equivalent width (EQW) and the “blueshift” (the offset of the measured rest-frame line peak from the expected laboratory value), can provide an understanding of the trade-off between the disk and wind parameters of quasars (see Murray et al. 1995; Elvis 2000; Proga et al. 2000; Leighly

2004; Casebeer et al. 2006; Leighly et al. 2007). The C IV emission line is a good diagnostic for a variety of reasons. Aside from Lyα, it is the most conspicuous emission line in high-redshift quasars, which allows for high S/N measurements of this line. More importantly, the EQW and blueshift of

Figure 4.3: RLF in low-z EV1 (FWHM Hβ vs. RFe II) parameter space for optically detected quasars within the FIRST observing area for all samples. The RL completeness correction (see Section 4.1) has been applied. The highest RLF bins for all of the sample are concentrated to the high FWHM Hβ-low RFe II corner of this parameter space (in agreement with Sulentic et al. 2000b).

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 33

C IV have the largest range of emission line properties for all high redshift quasars, making trends more apparent. Additionally, the blueshifting of the C IV line with respect to the quasar’s rest frame (Gaskell 1982; Wilkes 1984) is practically universally present in spectra of luminous quasars

(Sulentic et al. 2000b; Richards et al. 2002b).

In the context of a C IV analysis, Richards et al.(2011) ﬁnd that RQ quasars span the full space occupied by both quasar types—in contrast to the ﬁndings of Boroson and Green(1992) and

Boroson(2002). On the other hand, the RL quasars were largely conﬁned to that part of parameter space with small C IV blueshifts (and large EQWs) (see Richards et al. 2011, Fig. 7)—typically where HSRQs are found. Richards et al.(2011) interpret this result in a disk-wind framework and argue that, on average, RL quasars have weaker radiation line-driven winds than RQs.

Here I take the analysis of the radio properties of quasars in C IV parameter space one step further than Sulentic et al.(2007) and Richards et al.(2011). Speciﬁcally, I have repeated my RLF analysis in C IV parameter space. Figure 4.4 shows that, for all four samples, the RLF primarily decreases from low to high blueshift. No discernible RLF trend exists with respect to EQW1.

Ideally, my goal here is to be able to identify a UV emission line parameter that would predict whether or not an individual quasar is radio-loud. Although I have not accomplished that goal, I can improve the statistical prediction from a blanket ∼ 10% to a fraction that ranges from ∼0% to

∼30% as a function of C IV emission line properties. In one sense this does allow a prediction of radio properties for at least some quasars; it seems that quasars at the extreme end of the C IV blueshift distribution (for a given C IV EQW) are exceedingly unlikely to be radio-loud.

1Objects with very low EQWs were examined by eye. These objects were found to be atypical, being mostly BALs, miniBALs, relatively featureless, highly reddened, etc. It may be that such sources are intrinsically more radio-loud, but it is more likely that such objects appear in the sample due to the bias towards radio-detected quasars in the parts of parameter space where optical selection is ineﬃcient.

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 34

Figure 4.4: RLF as a function of both C IV equivalent width (EQW) and blueshift for optically detected quasars within the FIRST observing area. All plots use the RL completeness correction (see Section 4.1). The RLF primarily decreases from low to high blueshift with no discernible RLF trend in EQW; see also Richards et al.(2011). Objects in the row with the lowest EQWs likely have higher RLFs due to selection eﬀects.

4.3 The RLF and Black Hole Mass

Two of the most important properties that govern how quasars behave are the mass of the central black hole (BH) and its accretion rate. While the mass and accretion rate of a black hole cannot be measured directly, estimates of these values can be derived by using so-called BH mass scaling relations (e.g., Vestergaard and Peterson 2006) in conjunction with emission line and continuum information. I now explore how the RLF behaves as a function these BH mass estimates.

BH masses were compiled by Shen et al.(2011) and have been corrected as described in Sec- tion 3.3. Assuming a bolometric correction of LBol = 2.75L (Krawczyk et al. 2013), L can 2500 A˚ 2500 A˚ be converted to LBol to determine the Eddington ratio, LBol/LEdd, where LEdd is derived directly from the BH mass estimates. Figure 4.5 shows how the radio-loud fraction (of equally populated

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 35

Figure 4.5: RLF as a function of BH mass and accretion rate for quasars within the FIRST observing area for all of my samples. As with all of my RLF plots, I have corrected for RL completeness (see Section 4.1). For all of my samples, the highest RLFs are positioned in the corners of parameter space: high BH mass for the lowest accretion rates and low BH mass for the highest accretion rates. The dashed red lines show where L/LEdd = 0.01, 0.1, and 1.0 in order from top-left to bottom right. I assume LBol = 2.75L (Krawczyk et al. 2013) to calculate the Eddington 2500 A˚ ratio.

bins) depends on L (effectively accretion rate) and black hole mass for all samples. I have pre- 2500 A˚ sented the data in this way instead of plotting L/LEdd directly; Richards et al.(2011) argue that it is best to look at BH mass and accretion rate separately in case there are any threshold effects. Low mass/low accretion rate can have the same L/LEdd as high mass/high accretion rate but potentially very different properties. Nevertheless, L/LEdd appears as dashed red lines in Figure 4.5. The line on the lower-right indicates L/LEdd = 1, or (theoretical) maximal accretion (per mass), while the line on the top left and middle are L/LEdd = 0.01 and L/LEdd = 0.1, respectively.

In Figure 4.5, I ﬁnd that the bins with the highest RLFs are situated in the corners of parameter space, speciﬁcally high BH mass/low accretion rate and low BH mass/high accretion rate. The lowest RLFs exist in a diagonal band that stretches from low BH mass for the lowest accretion rates

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 36 to high BH mass for the highest accretion rates.

Since the estimation of BH masses in high-redshift quasars by way of scaling relations is not an exact science and is dependent on the emission lines used, I also create subsamples based on the origin of these masses (see Figure 4.6). Speciﬁcally, my samples include quasars whose masses are estimated using the Hβ, Mg II and C IV emission lines; this roughly corresponds to z < 0.7,

0.7 ≤ z < 1.9, and z ≥ 1.9, respectively. The mass estimates computed using the Hβ and Mg II emission lines are thought to be reliable as they have been shown to be within a factor of 2.5 of the masses found using reverberation mapping (McLure and Jarvis 2002). Indeed, for the two low- redshift bins, I see behavior that is consistent with what is expected from past work. Namely, that the quasars with the highest masses (at a given luminosity) are the most likely to be radio-loud, though not exclusively radio-loud; most quasars in the RL regions are still RQ (see Lacy et al. 2001).

However, when I examine the high-redshift subsample, I see something quite different. Here the lowest mass quasars appear to be the most radio-loud. There are a number of potential explanations for this observation. One possibility is that there is an actual physical transition at high redshift such that RL quasars are more likely to be found in high L/LEdd systems. This could be magnified (or perhaps caused) by a change in the selection of quasars from UV-excess sources at z ∼< 2 to u-band dropouts at higher redshifts. Alternatively, instead of a physical change, perhaps the high-redshift quasar sample is simply biased against RL quasars with high-mass (at high redshift). However, I consider this unlikely: although the completeness of FIRST drops with redshift, radio sources are explicitly targeted for spectroscopy as quasars candidates in the SDSS surveys. While there are known incompletenesses in the quasar selection at high-z, the most glaring of these has to do with the presence (or lack thereof) of Lyman-limit absorption systems (Worseck and Prochaska 2011) and is independent of the radio properties of the quasars. Indeed, estimates of the completeness of the SDSS quasar survey (e.g., Vanden Berk et al. 2005) are inconsistent with the extreme level of incompleteness that would be required to induce this effect.

Instead I argue that the problem lies in the estimation of BH masses using the C IV emission line.

This could arise from more of the C IV line being emitted in a wind component than has previously

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 37

Figure 4.6: Figure 4.5, Sample B split into different populations based on the spectral line used to calculate the BH masses. Each panel effectively represents quasars within different redshift bins (see Shen et al. 2011, Sec. 3.7). For the two lowest redshift samples (Hβ and Mg II), the highest RLFs belong to the highest BH mass bins for these samples. The lowest BH mass bins in the highest redshift sample (C IV) are (apparently) the most RL. I suggest that this difference is indicative of errors in the C IV BH mass estimates. The dashed red lines from top-left to bottom right show where L/LEdd = 0.01, 0.1, and 1, in order. I assume LBol = 2.75L (Krawczyk et al. 2013) to calculate the 2500 A˚ Eddington ratio.

been thought or in the form of a wind-strength dependence to the proportionality constant in the radius-luminosity relation (Richards et al. 2011). While it is well-known that determining BH mass scaling relations from the C IV lines are the most challenging (e.g., Shen et al. 2011; Assef et al. 2011;

Denney 2012; Runnoe et al. 2013; Park et al. 2013; Denney et al. 2013), the corrections necessary for the radio-loudness trend to match the low redshift BH mass trends are not consistent with the level of “tweaks” to the C IV BH mass scaling relations that are generally advocated. Rather, these

BH mass estimates must be catastrophically wrong. Speciﬁcally, the RL quasar masses at high-z are too low by as much 0.5 dex or more and not simply by ∼ 0.2–0.3 dex.

Technically this statement applies only to the RL quasars. However, as Richards et al.(2011) argue that there is negligible diﬀerence in the emission line properties of RL and RQ quasars with

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 38 small C IV blueshifts, it must also be true that a large fraction of the masses computed for RQ quasars are similarly erroneous. Indeed, it would seem that the C IV BH mass estimates are close to being inverted (large BH mass should be small, and vice versa). As there is only a weak correlation between the BH masses estimated from C IV and Mg II in Shen et al.(2011, Fig. 10), this is perhaps not surprising. I will consider this issue further in Section6.

Chapter 4: Radio Properties in the Extreme: the Radio-Loud Fraction (RLF) 39

Chapter 5: Radio Properties in the Mean: Stacking Analysis

5.1 Motivation

While I have begun my investigation into the physics behind the production of radio emission in quasars by exploring bright radio sources, essentially all quasars are radio emitters when probed to deep enough limits (e.g., Wals et al. 2005; White et al. 2007; Kimball et al. 2011). Thus, an important part of the conversation about the nature of radio emission in quasars is an analysis of the ensemble average of radio properties in quasars that are formally radio-quiet. At the faintest radio luminosities the radio power in quasars is likely due to starburst emission (e.g., Condon et al.

2002; Kimball et al. 2011). However, even if a starburst could produce radio luminosities as high as 1031 ergs s−1 (whereas Kimball et al. 2011 ﬁnd the peak to be 1029 ergs s−1), that still leaves a considerable population of radio-quiet quasars that are neither formally radio-loud nor consistent with a starburst origin (Blundell and Beasley 1998; Jiang et al. 2010; Zakamska and Greene 2014).

Indeed, Ulvestad et al.(2005), using high resolution radio imaging, and Barvainis et al.(2005), using variability studies, both conclude that RQ quasars are just weaker versions of RL quasars.

Alternatively, Blundell and Kuncic(2007) argue that disk winds are responsible for radio emission in RQ quasars. By stacking the radio images of all known quasars covered by the FIRST survey, I hope to learn about the mean radio properties of these objects. I can then contrast these ﬁndings with the properties I have already identiﬁed using formally radio-loud quasars.

5.2 Image Stacking

My stacking analysis follows that of White et al.(2007). For a more detailed explanation, see that paper, but the process is brieﬂy described here.

First, using the optical coordinates of my quasar populations, 0.05 x 0.05 radio images were down- loaded from the FIRST data extraction website1. Since I wish to explore the mean radio characteristics of quasars as a function of various properties, I stack the radio images in bins based on these

1http://third.ucllnl.org/cgi-bin/ﬁrstcutout 40

2-D parameter spaces (e.g., L-z, Section 5.4; C IV and EV1, Section 5.6; BH properties, Section 5.7; color, Section 5.8).

After assigning each quasar to a 2-D parameter bin, all of the FIRST radio images within each bin were added using a median stacking procedure (see White et al. 2007): a pixel in the final stacked image corresponds to the median value of the pixels occupying that same location from the set of radio images within a bin. White et al.(2007) compared the benefits and consequences of median stacking as opposed to mean stacking and concluded that the median stacking procedure best represents the properties of the population. Mean stacking is more susceptible to outliers, making the noise for any mean stack greater than that of the median stack for the same populations. Addition- ally, White et al.(2007) show that the median effectively converges to the mean for distributions such as I consider herein, so I will generally refer to my median stacking results as the mean or average.

After combining the cutouts into stacked images, the peak fluxes of my stacked sources need to be corrected for what White et al.(2007) designate the “snapshot bias”. Learning from the well-known problem of “clean bias” associated with FIRST sources (Becker et al. 1995), White et al.(2007) performed stacking with artificial sources of the same, known flux to be sure that they recovered the correct median values. A systematic underestimation was noticed and the discrepancy was corrected using:

fpeak, corr. = min(1.40 fpeak, fpeak + 0.25 mJy), (5.1)

where fpeak is the peak flux density (mJy) of the median stack. The flux boundary that determines which part of the equation to implement is 625 µJy. As that value is more than 200 µJy greater than the largest median peak flux density I achieve, I only need to multiply my measurements by

1.40 for the entirety of this analysis to correct for this bias.

5.3 Median Stacking Diagnostics

I ﬁrst explore the distribution of mean radio ﬂux density by stacking in redshift bins (Figure 5.1).

Samples A and B were divided into 50 redshift bins while Samples C and D were divided into 100.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 41

After applying the median stacking procedure described above, I retrieve the same basic results as

White et al.(2007): the median ﬂux density declines up to z = 2 (see Figure 5.1). This trend of decreasing ﬂux density with redshift is expected based on inverse square law dimming. Note that even my most restrictive sample (B) includes 10,000 more quasars than considered by White et al.

(2007) (41,295 SDSS DR3 quasars) and should be clean of selection eﬀects up to z ∼ 2.2.

I observe an increase in median flux density starting at roughly z = 2.2 for all samples (typically peaking at z ∼ 2.7). This increase can be attributed to selection effects whereby the SDSS optical selection was very inefficient at z ∼ 2.7 while the radio selection is more complete (compare panels a and b in Richards et al. 2006, Figure 6). As such, the quasars discovered at z ∼ 2.7 are more likely to be radio sources, thus biasing the observed mean flux density and requiring that a robust analysis be limited to z < 2.2. I find an additional increase in flux density for Sample C at z = 1; however, the lack of this feature in Sample D suggests that it is similarly due to selection effects.

Figure 5.1: Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of redshift (see White et al. 2007, Figure 6). The vertical dashed lines represent z = 2.7, which is the upper limit of eﬃcient SDSS optical selection.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 42

I next investigate the mean radio ﬂux density and αro (see Section 3.7) as a function of i-band magnitude to explore the correlation between radio and optical brightness. Figure 5.2 shows that the strongest radio emitters are also the optically brightest, while Figure 5.3 shows that the optically faintest sources are the most radio-loud, consistent with White et al.(2007, Figures 7 and 12).

5.4 The Mean Radio Properties as a Function of L and z

I next split the samples up along L and z as I did for the radio-loud fraction analysis in 2500 A˚ Figure 4.2 in order to break the degeneracy inherent between redshift and brightness in flux-limited samples. As with the RLF, I first divide the samples into equally populated groups according to redshift. Then each group is separated into equally populated bins according to L , and I 2500 A˚ perform the median stacking procedure to extract the stacked peak flux density of each bin. The results from each bin are plotted such that the color of each box represents the median stacked peak

Figure 5.2: Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of i-band color (see White et al. 2007, Figure 7). As before, the strongest radio emitters are associated with the optically brightest sources.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 43

Figure 5.3: Radio to optical spectral indices (αro) of median stacked quasar bins as a function of i-band color (see White et al. 2007, Figure 12). While Figure 5.2 shows that the brightest sources in the optical are the brightest in the radio, the ratio of radio to optical ﬂux is such that the radio-loudest objects (more negative values of αro ) are associated with the faintest optical sources.

ﬂux density of each bin and the location of each box in the L -z plane corresponds to the bin’s 2500 A˚ median values.

The results are shown in Figure 5.4. It can be seen that, at a fixed redshift, quasars that are optically more luminous have higher radio flux densities, and at a fixed optical luminosity, lower redshift quasars have higher radio flux densities. The trend is roughly consistent with the mean radio flux density being primarily dependent on the optical magnitude as it was for the radio-loud fraction: optically brighter quasars are radio brighter, on average; see also Figure 5.2.

Since I really want to understand the intrinsic radio properties of quasars, I now convert from apparent brightness to luminosity, where the conversion from radio ﬂux to radio luminosity is determined according to Equation 3.2 as discussed in Section 3.6.3.

Figure 5.5 shows the results of stacking the radio luminosities in the L -z plane. Again, 2500 A˚

Chapter 5: Radio Properties in the Mean: Stacking Analysis 44

Figure 5.4: Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of both redshift and L . The trend roughly follows the i-band magnitude with brighter quasars in the optical being brighter in the 2500 A˚ radio.

more luminous sources in the optical tend to be more luminous in the radio, but a larger eﬀect is seen with redshift, where a small radio ﬂux at high-z can translate to a high radio luminosity. The most radio luminous sources are at high-z and have high optical luminosities. Objects with roughly equal radio luminosities span a diagonal from the upper-left to the lower-right while radio luminosity decreases from the upper-right to the lower-left.

As noted in Section 3.7, looking at the radio luminosity as a measure of radio-loudness is correct only if there is no correlation between the optical and the radio. If there is, then it is more appropriate to consider the ratio of the two, or equivalently, the spectral index between the radio and optical,

αro , which is deﬁned in Section 3.7.

Figure 5.6 shows the resulting distribution in αro , which is a measure of the shape of the SED between the radio and optical. Normalizing by the optical luminosity has produced a signiﬁcantly diﬀerent trend than was seen in Figure 5.5. That trend is for quasars to be stronger radio sources

Chapter 5: Radio Properties in the Mean: Stacking Analysis 45

Figure 5.5: Radio luminosities (erg s−1 Hz−1) of median stacked quasar bins as a function of both redshift and L . 2500 A˚ These bins are the same as those used in Figures 5.4. As expected, the most radio luminous sources are at high-z and have high optical luminosities with radio luminosity decreasing from the upper-right to lower-left.

(relative to the optical) with decreasing optical luminosity (at ﬁxed redshift) and with increasing redshift (at ﬁxed optical luminosity) — quite the opposite of what was found for the RLF in Section

4.1. Thus, the mean radio properties of quasars are not following the same trends as the extreme

RL population. This trend is perhaps unexpected but is indeed consistent with Figure 5.5, where equal radio luminosities occupied roughly diagonal tracks in L -z space. Along one of those 2500 A˚ diagonals the objects with the lowest optical luminosities will have the largest radio-to-optical ratio, so radio-dominance from the objects along the lower boundary of the distribution is expected. As was the case for the RLF, αro is evolving largely in the direction of changing i-band magnitude

(consistent with Figure 5.3). For the RLF, I was concerned that incompleteness could be causing that dependence. My stacking analysis should be independent of the completeness of FIRST, which argues that the trend with optical magnitude may be real and is related to competing luminosity and redshift eﬀects as noted by Ivezi´cet al.(2002).

Chapter 5: Radio Properties in the Mean: Stacking Analysis 46

Figure 5.6: Radio to optical spectral indices (αro) of median stacked quasar bins as a function of both redshift and L . These bins are the same as those used in Figures 5.4 and 5.5. This trend is completely opposite to that found 2500 A˚ for the RLF (see Figure 4.2). The median stacking shows stronger radio sources (relative to the optical) with decreasing optical luminosity (at ﬁxed redshift) and with increasing redshift (at ﬁxed optical luminosity).

5.5 Radio-Optical Spectral Index as a Function of Luminosity

As there is precedent for the shape of the spectral energy distribution in quasars to be a function of luminosity, it is also important to consider how αro may change with L . In particular, it 2500 A˚ has been repeatedly shown (Avni and Tananbaum 1982; Steﬀen et al. 2006; Just et al. 2007; Lusso et al. 2010) that there is a non-linear relationship between the X-ray and UV luminosity in quasars: quasars with double the UV luminosity do not have double the X-ray luminosity. Failure to correct for any similar systematic trends in αro with optical luminosity could lead to biased conclusions.

As such, I investigate the relationship between Lrad and Lopt in terms of the behavior of αro as a function of L . 2500 A˚

To investigate correlations between Lrad and L , I separate the quasars into bins of optical 2500 A˚ luminosity for each of the four samples. The bins were deﬁned such that each would have equal

Chapter 5: Radio Properties in the Mean: Stacking Analysis 47 populations of about 1,000 quasars.

The top-left plot of Figure 5.7 shows the correlation between Lrad and L (see White et al. 2500 A˚ 2007, Fig. 9) for the entire range of redshifts within Sample B, whereas the other panels show restricted redshift ranges. Here I have limited the analysis to Sample B as using a less homogeneous sample can imprint biases onto the distribution in αro -L parameter space. I have further 2500 A˚ limited the analysis to point sources to avoid optical contributions from the host galaxy and to z < 2 (Figure 5.7, top-middle) to avoid the known bias towards radio sources in the SDSS selection function at higher redshifts.

The best ﬁt line is computed as log Lrad = m log L + b, where the median L value 2500 A˚ 2500 A˚ for each bin was used and the coordinate pairs (m, b) represent the slopes and y-intercepts for the linear best ﬁt models. Just as in White et al.(2007), the radio luminosities for the four samples do not increase linearly with the optical luminosities. For low redshift (z < 2) point-source quasars

0.90 within Sample B, I ﬁnd that the relationship is Lrad ∼ Lopt . This approximately corresponds to a factor of two in radio luminosity between the least and most luminous quasars in the optical, similar to what was found by White et al.(2007). This deviation from a linear relationship is not as strong as it is in the X-ray (exponent of ∼ 0.72 in Steﬀen et al. 2006); however, the lever arm in extrapolating from the optical to the radio is longer than that between the optical and X-ray, and any deviation from linearity is still important to take into account.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 48

Figure 5.7: Radio luminosity dependence on L within different redshift intervals (see White et al. 2007, Figure 2500 A˚ 10) for median stacked quasar point source bins in Sample B. The dashed red line in each plot shows where Lrad = Lopt. The best fit lines for each redshift range are shown as the different dashed black lines, and the values in the lower-right corner indicate the slopes and intercepts of these lines. Sample B shows a non-linear relationship between Lrad and .90 Lopt, as Lrad ∼ Lopt for its least biased subsample (z < 2, top-middle).

The same information from Figure 5.7 is eﬀectively shown in Figure 5.8a, but I have color-coded the diﬀerent redshift regions and am now plotting αro on the y-axis. To compute the stacked

αro values, each of the radio cutouts (in Jy) is ﬁrst divided by its corresponding quasar’s optical

flux. Then, the cutout ratios within a bin are median stacked, and the maximum pixel value in the stacked image is taken to be f5 GHz/f . Finally, the median αro value is found by plugging 2500 A˚ into Equation 3.3. Here I find that the redshift regions occupy wedge-shaped distributions that are consistent with the flux-limited nature of the quasar sample. As such, the best-fit line (for z < 2) removes quasars above log L = 30.75 beyond which there is an artificial bias in the sample. 2500 A˚

Chapter 5: Radio Properties in the Mean: Stacking Analysis 49

The equation reported for the linear best ﬁt model is αro = −0.186(m − 1) log L − 0.186 b, 2500 A˚ such that the values of m and b have the same meaning in both Figures 5.7 and 5.8a. All four of the samples (although only Sample B is pictured) show that αro decreases (gets more radio-loud) as

L increases. This is opposite to what was found in Figure 5.7 and would seem to be due to the 2500 A˚ biased nature of the redshift slices in Figure 5.7 as highlighted by the color-coding of Figure 5.8a.

Indeed, Figure 5.8a suggests that there is a small increase in radio luminosity with optical luminosity

1.059 (consistent with Lrad ∼ Lopt ).

Since the samples are ﬂux limited, any evolution in luminosity could instead be an evolution in redshift. As such, I reproduced Figure 5.8a with redshift instead of L to be sure that there 2500 A˚ were no additional biases. Figure 5.8b shows the dependence of αro on z (see Steﬀen et al. 2006,

Figure 7, top). The coordinate pair (a, B) represents the slope and y-intercept for the linear best

fit model such that αro = a z + B, where I have limited the fitting to data with z < 2 as I did in Figure 5.8a. All four of the samples (Sample B, pictured) show that αro slightly decreases with increasing z. This trend with redshift is larger than that seen in the X-ray (Steffen et al. 2006).

It is an open question as to whether I should be using αro (see Equation 3.3) or ∆αro =

Figure 5.8: Left: The dependence of αro on L (see Steffen et al. 2006, Figure 5, top) for median stacked quasar 2500 A˚ point source bins in Sample B. Recall that more negative values of αro mean more RL; thus, the higher luminosity objects are biased to a more RL SED. Different redshift intervals are highlighted with different colors. Sources to the right of the dashed red line (log L = 30.75) have been removed when computing the linear best fit (dashed black 2500 A˚ line) so as to not artificially skew it with selection effects. Right: The dependence of αro on redshift (see Steffen et al. 2006, Figure 7, top) for the same median stacked bins as the left panel. The discrete appearance of the points in this panel is an artifact of initially binning the samples with respect to redshift. The colors correspond to the different redshift intervals from the left panel. Sources to the right of the dashed red line (z = 2) have been removed when computing the linear best fit (dashed black line).

Chapter 5: Radio Properties in the Mean: Stacking Analysis 50

αro, obs. − αro, best ﬁt (αro corrected for luminosity and/or redshift) in my analysis. The shape of the

SED is measured by αro whether or not αro has any luminosity or redshift dependences. If it is the shape that matters (as it is for the dependence of radiation line-driven winds on αox ), then I should be using αro . In that case, my current analysis will suﬃce. If, on the other hand, the shape relative to the mean at a given luminosity or redshift is most important, then I should be using ∆αro . For example, if dust reddening were causing a trend in αro with luminosity, it might be preferable to use ∆αro . Indeed, absorption is an issue for αox ; however, in this case the relative deﬁcit of optical

ﬂux is for the most luminous sources, not the least luminous. Therefore, it is unlikely that dust reddening is causing the increase in radio-loudness with luminosity. An accurate determination of the luminosity and redshift dependence of αro is a question suitable for its own investigation (e.g.,

Steﬀen et al. 2006). I will leave my analysis in terms of αro noting that the trends could change with ∆αro .

In short, apart from any corrections for the L- and z-dependences of αro , I ﬁnd that the median shape of the radio-to-optical continuum appears to behave in the opposite way as the RLF (see

Figure 4.2). I will discuss this ﬁnding further in Section6.

As with the analysis of the RLF in the L-z parameter space above (Figure 4.2), my analysis of the mean radio properties shows a trend that is strongest in the direction of apparent magnitude.

This is shown explicitly in Figure 5.3 and must give one pause about the reality of any trends in luminosity and redshift separately. One possibility is that bright quasars that are extincted by dust are moving to larger (fainter) magnitudes and, thus, appear to be more RL than they should be.

However, I have excluded quasars that are most likely to be dust reddened/extincted (those with

∆(g − i) > 0.5). Moreover, Figure 5.9 shows the mean relative color in each of the bins. Ignoring the lowest redshift quasars (where host galaxy contamination makes determining the relative colors diﬃcult), I see that there is no strong trend towards redder colors with fainter magnitudes. As such, it would appear that dust is not the explanation, and I am left to conclude that the mean radio-loudness does indeed evolve with both redshift and luminosity in a way that mimics a trend in apparent magnitude.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 51

Figure 5.9: The relative colors, ∆(g − i), of median stacked quasar bins as a function of both redshift and L . 2500 A˚ These bins are the same as those used in Figures 5.4, 5.5, and 5.6 and show no major diﬀerences in the amount of measured dust reddening/extinction.

5.6 The Mean Radio Properties and Accretion Disk Winds: Principal Component and CIV Analyses

In Section 4.2, I found that the RLF was a function of both Eigenvector 1 and C IV parameters; objects likely to have strong radiation line-driven accretion disk winds are least likely to be radio- loud. I repeat that analysis here but with the median radio stacking procedure. Figure 5.10 shows the mean radio-loudness as measured by αro in EV1 parameter space, and Figure 5.11 shows the same but with C IV emission line properties.

There is less of a discernible trend in Figure 5.10 than there was for the RLF analysis (Figure 4.3).

I do note that the most RL sources are still those in the top-left corners with the broadest Hβ and weakest Fe II emission lines. For the C IV analysis, the trend is also weaker but does seem to be in the same direction as it was for the RLF (Figure 4.4): small-blueshift quasars are more RL, on average, than those with large C IV blueshifts.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 52

Figure 5.10: Radio to optical spectral indices (αro) of median stacked quasar bins in low-z EV1 (FWHM Hβ vs. RFe II) parameter space. Just as for the RLF analysis (Figure 4.3), the most RL sources are those in the top-left corners with the broadest Hβ and weakest Fe II emission lines.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 53

Figure 5.11: Radio to optical spectral indices (αro) of median stacked quasar bins in C IV parameter space. Quasars with small C IV blueshifts are more RL, on average, than those with large C IV blueshifts. This trend, although weaker, advances in the same direction as that for the RLF analysis (Figure 4.4).

5.7 The Mean Radio Properties and Black Hole Mass

I now consider the mean radio-loudness as a function of BH mass, accretion rate, and Eddington ratio (L/LEdd) as I did for the RLF in Figure 4.5. Specifically, Figure 5.12 shows the dependence of αro on BH mass and quasar luminosity (effectively BH accretion rate). I find strong similarities between these results and the results from the RLF analysis with the most RL objects being towards the top-left or bottom-right of each panel. If the low-redshift BH masses are reliable and the high- redshift BH masses are compromised as discussed in Section 4.3, then the mean radio-loudness would increase towards lower Eddington ratios, consistent with the RLF.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 54

Figure 5.12: Radio to optical spectral indices (αro) of median stacked quasar bins in BH mass parameter space. The dashed red lines from top-left to bottom right show where L/LEdd = 0.01, 0.1, and 1, in order. I assume LBol = 2.75L (Krawczyk et al. 2013) to calculate the Eddington ratio. If the high-redshift BH masses are ignored 2500 A˚ because of their suspected shortcomings, then the mean radio-loudness increases towards lower Eddington ratios. This trend is consistent with the RLF (Figure 4.5).

5.8 The Mean Radio Properties as a Function of Color

I extend my study of the mean radio properties of quasars by exploring the correlation between the strength of radio emission and optical color. As before, I split the samples into bins based on the colors of optically-detected quasars and apply the median stacking procedure described above.

Similar to White et al.(2007, Figure 14), I use ∆( g − i) for the measure of color. As stated earlier,

∆(g − i) is deﬁned in such a way to remove the dependence of color on redshift. It is roughly equivalent to αopt, the underlying continuum (excluding emission features) in the optical-UV part of the SED.

Figure 5.13 shows how the median radio ﬂux density varies as a function of color. Just as in

White et al.(2007), I ﬁnd that bluer and redder objects have higher radio ﬂux densities, with the reddest objects being the brightest of all. For all four samples, objects with ∆(g − i) > 0.6 have

Chapter 5: Radio Properties in the Mean: Stacking Analysis 55 peak ﬂux densities 2-3 times larger than quasars with average colors.

As in Section 5.4, I wish to analyze the intrinsic properties of quasars. As such, I naturally examine how radio luminosity changes with optical color. Figure 5.14 shows that, for each of the samples, radio luminosity increases for redder objects. Thus, the trend in flux density seen at both color extremes in Figure 5.13 and in White et al.(2007, Figure 14) appears to be artificial: the blue objects simply have very different luminosities than the red objects.

I can quantify the luminosity-corrected trend, instead, by plotting αro as a function of ∆(g − i).

Figure 5.15 shows that radio-loudness increases for redder quasars (recall that more radio-loud corresponds to increasingly negative values of αro). As αro represents another way to measure the

R-parameter (Section 3.7), my results agree with those of White et al.(2007, Figure 13), who found an increase in the R-parameter for objects with redder colors.

These trends match my results from Figures 5.13 and 5.14 since objects with increasing radio

Figure 5.13: Peak radio ﬂux densities (µJy bm−1) of median stacked quasar bins as a function of ∆(g − i). Quasars with colors redder, ∆(g − i) > 0, and bluer, ∆(g − i) < 0, than average have higher radio ﬂux densities. These results match those of White et al.(2007, Figure 14).

Chapter 5: Radio Properties in the Mean: Stacking Analysis 56

Figure 5.14: Radio luminosities (erg s−1 Hz−1) of median stacked quasar bins as a function of ∆(g − i). The intrinsic radio luminosity increases for redder objects.

ﬂux densities and luminosities should be increasingly radio-loud if optical luminosity remains the same. While I expect to measure less optical emission for redder objects because of dust extinction

(which does not reduce observed radio emission), optical extinction by dust does not appear to be the lone cause of this trend as it also applies to relatively blue quasars.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 57

Figure 5.15: Radio to optical spectral indices (αro) of median stacked quasar bins as a function of ∆(g − i). More negative values of αro correspond to more radio-loud. Agreeing with the results of White et al.(2007, Figure 13) who used the R-parameter, radio-loudness increases (αro decreases) for redder quasars.

Chapter 5: Radio Properties in the Mean: Stacking Analysis 58

Chapter 6: Discussion

6.1 Luminosity and Redshift Evolution

In Section 4.1, I reproduced the results of Jiang et al.(2007) and Balokovi´cet al.(2012) by ﬁnding that the RLF increases with increasing optical luminosity and decreasing redshift (Figure 4.2).

Exactly opposite to these ﬁndings, I determined in Section 5.4 that the mean radio-loudness increases with decreasing optical luminosity and increasing redshift (Figure 5.6). Singal et al.(2011) similarly

find increasing radio-loudness with increasing redshift. Their apparent discrepancy with the results of Jiang et al.(2007) can be explained by the difference between the mean radio-loudness (as is shown in Figure 5.6 and in Singal et al. 2011) and the RLF (as is shown in Figure 4.2 and in Jiang et al. 2007). My results are consistent with both papers when considered in this light. Thus, for these two parameters, the mean radio properties of quasars are not following the same trends as the extreme RL population. Indeed, Balokovićet al.(2012) also find that, as redshift increases, quasars become both more RL on average but also less likely to inhabit the formally RL tail of the distribution.

It is curious that, unlike in the L-z parameter space, in the EV1 (Figures 4.3 and 5.10), C IV

(Figures 4.4 and 5.11), and BH (Figures 4.5 and 5.12) parameter spaces, the RLF and mean radio- loudness increase in the same direction. Speciﬁcally, there is reasonable agreement between the directions of evolution of the mean radio-loudness and the RLF when considering parameters that are not a strong function of the apparent magnitude. As a result, an explanation for the diﬀerences seen in the L-z evolution could be the incompleteness of the FIRST survey as discussed in Section 3.6.5.

That is, if the stacking analysis is independent of optical magnitude while the RLF is not and if there is RL incompleteness with fainter optical magnitude, then the stacking results may be less biased than the RLF results. I note that there is no reason that the RLF and the mean radio-loudness have to evolve together in either luminosity or redshift, but the similarity of the trends in the other parameter spaces that I have considered may suggest that the observed RLF evolution is not robust. 59

6.2 Mass, Accretion Rate, and Spin

It is enlightening to consider my results in the context of theoretical work on the relationship between

BH spin, mass, and accretion rate in AGNs. Blandford and Znajek(1977) and Blandford and Payne

(1982) provide the framework for how the spin of the BH and the accretion disk, respectively, can be tapped in the production of radio jets. Wilson and Colbert(1995) argue that BH mass and accretion rate are not that diﬀerent for RL and RQ quasars and that the diﬀerence must be related to BH spin. Richards et al.(2011) also argue for a similarity of properties between RL quasars and at least part of the RQ population.

Previous work has suggested that the radio-loudness is dependent on either the BH mass or the accretion rate. For example, Laor(2000) and Lacy et al.(2001) find that high BH mass is a necessary (if insufficient) condition for being radio-loud and that there is also a (weaker) correlation with the Eddington ratio, L/LEdd. Woo and Urry(2002), on the other hand, argue that there is no dependence on BH mass. Ho(2002) similarly finds little BH mass dependence and argues instead for

L/LEdd as the primary driver; however, Jarvis and McLure(2002) claim that the use of R, rather than Lrad, by Ho(2002) to characterize quasars as RL or RQ led to the uncorrelated results. Jarvis and McLure(2002) also assert that doppler boosting brightens the intrinsic radio luminosity of ﬂat

2.5 spectrum sources and that the corrected data agrees with a dependence of the form Lrad ∝ MBH

(Dunlop et al. 2003). Sikora et al.(2007), using a sample that is largely complementary to the ones utilized for my analyses, found that radio-loudness increases with decreasing L/LEdd but argued that there must also be a secondary parameter (BH mass or spin) having an eﬀect. Based on a PCA analysis, Boroson(2002) presents a schematic in which RL quasars preferentially have both high BH mass and low L/LEdd.

My results (Figures 4.5 and 5.12) lead to a very diﬀerent conclusion than that of Shankar et al.

(2010) (using much of the same data) who find that high-redshift RL quasars have high L/LEdd, while low-redshift RL quasars have low L/LEdd. This difference is because I consider the BH masses of the high-redshift quasars to be in error. Instead, after reconsidering the BH masses estimated using the C IV emission line, I find that my results are in general agreement with most of the

Chapter 6: Discussion 60 investigations above: there is a clear trend towards a higher RLF and a louder mean radio-loudness with decreasing L/LEdd. However, Figure 4.5 suggests that the most RL sources at high luminosities do not have particularly low L/LEdd in the absolute sense: they are simply the objects with the lowest L/LEdd (and thus the highest BH masses) at that luminosity. This ﬁnding may suggest that

BH mass is the dominant eﬀect and that low L/LEdd is a consequence of the mass trend.

Regardless of whatever controls their radio strength, the fact that the vast majority of quasars are still classiﬁed as RQ, even where RL quasars are most prevalent, cannot be ignored. Thus, whether or not the radio-loudness distribution is bimodal, there is still a strong dichotomy between

RL and RQ quasars. This is intriguing as the SDSS quasar targeting and spectroscopic classiﬁcation is such that my samples are expected to be predominantly comprised of massive, bulge-dominated systems (Floyd et al. 2004). Consequently, it is worth noting that my quasar samples still show some of the radio dichotomy of the samples that cover a wider range of luminosity space such as

Sikora et al.(2007).

I note that Floyd et al.(2004) found both the RL and RQ quasars (with MV < −24 and z ∼ 0.4) in their investigation to be bulge-dominated systems and that Sikora et al.(2007) found luminous

RQ quasars hosted by elliptical galaxies. Specifically, while it has long been known that RL quasars live in ellipticals, Floyd et al.(2004) demonstrates that ellipticals can host RQ quasars too. If luminous, high-redshift quasars are predominantly fueled by major mergers, it must be the case that the average quasar hosted by an elliptical galaxy is RQ. As such, the RL/RQ dichotomy among luminous quasars cannot entirely be due to morphology. While I cannot say with certainty that both HSRQ and SSRQ quasars have the same hosts, there is no compelling reason (other than the differences in radio properties) to think that RL quasars and HSRQ quasars have different hosts.

Although it has been found that RL and RQ quasars have diﬀerent environments (Malkan 1984;

Smith et al. 1986; Sikora et al. 2007) with RL quasars living in denser regions, such work has only been done in the context of comparing RL and RQ quasars as a whole. In that case, differences would indeed be expected. What would be of more interest is to redo environmental and clustering studies considering RL quasars in comparison with different subsets of RQ quasars: specifically dividing

Chapter 6: Discussion 61 up RQ quasars into HSRQs (which may have similar properties as RL quasars) and SSRQs (which would be expected to have the largest environmental and clustering diﬀerences in comparison with

RL quasars).

If massive, low L/LEdd systems can be both RL and RQ and if morphology and/or environment are not clearly different between RL quasars and their (putative) parent RQ quasars, then I am left to consider spin as the defining parameter. Major mergers (that are hypothesized to be the dominant cause of the luminous quasars in my investigation) are expected to produce high-spin systems (Wilson and Colbert 1995; Volonteri et al. 2007, 2013). In part, this is because if enough matter is accreted in one feeding (and even if the accretion is initially retrograde) the BH is spun down more efficiently than it is spun up; quasars that have experienced a recent major merger would tend to be driven to high (prograde) spin (Volonteri et al. 2013). This leads to a paradox: if RL quasars have high spin and RQ quasars are not spinning (e.g., Wilson and Colbert 1995), then the accretion efficiency suggested by the Soltan(1982) argument means that RL quasars should not be as rare as they are. This further argues against RL quasars having retrograde spin as in Garofalo

(2009) and Garofalo et al.(2010).

One way to reconcile all of this is if both the RL quasars and those RQ quasars with otherwise similar properties (i.e., HSRQs) are dominated by major mergers, resulting in elliptical hosts with high spins. These are systems with high mass for their luminosity and low L/LEdd. The quasars among these that become RL may be those that have undergone a rare “2nd generation merger” and have been spun up to a value above some threshold (Sikora 2009). However, I note that the spins are not likely to be all that dissimilar; significant differences in the spin would lead to significant changes in the accretion disk properties (in particular the inner radius) that would be expected to produce continuum (and broad emission line region) variations inconsistent with the results of Richards et al.

(2011).

This suggests that RL quasars and HSRQ quasars both have high spin, with the RL quasars being somewhat more extreme. The rest of the RQ population (SSRQs) could be similarly high spin objects, may have decreasing spin with increasing L/LEdd, or could have no spin. Again, however,

Chapter 6: Discussion 62 the Soltan(1982) argument would suggest that having some spin is more likely than having no spin.

In terms of making connections to low-z quasars, it is particularly important to realize that low redshift quasars (that are the frequent focus of detailed observing campaigns and reverberation mapping analysis) and the luminous high redshift quasars that dominate my sample might be rather different creatures, especially if low-z quasars are primarily accreting molecular clouds (Volonteri et al. 2013). This might explain the somewhat different results seen here and by Boroson(2002) in terms of whether or not RL quasars are an extrema of the quasar population in ways other than in the radio. While I agree with Boroson(2002) that RL quasars tend to be high mass, low L/LEdd sources, I do not find them to be unique. Indeed most high mass, low L/LEdd quasars are RQ.

6.3 Accretion Disk Winds

My analyses in the Eigenvector 1 and C IV parameter spaces (both Sections 4.2 and 5.6) along with the stacking in color space (Section 5.8) paint a consistent picture in that I find different quasar properties tracking together. For example, Sulentic et al.(2007) find a correlation between the C IV blueshift and R(Fe II), while Reichard et al.(2003) find a correlation between C IV blueshift and quasar color. Additionally, both Gallagher et al.(2005) and Kruczek et al.(2011) consider the connection between X-ray properties and C IV emission.

In investigating the radio properties of quasars as a function of those parameters, I see quasars (at all redshifts) exhibiting behaviors (emission line and continuum properties) that are consistent with one extreme (large Hβ FWHM, low R(Fe II), small C IV blueshift, red color) being much more likely to host quasars with stronger radio emission than the opposite end of parameter space. Richards et al.

(2011) compared the average RL quasar spectrum to the average spectrum of HSRQs (speciﬁcally

RQs with similar C IV EQWs and low blueshifts) and found little diﬀerence; furthermore, composite spectra from HSRQs and SSRQs possessed quite diﬀerent emission line properties.

This diﬀerence is consistent with my C IV parameter space ﬁndings that a higher RLF (Figure

4.4) and higher mean radio-loudness (Figure 5.11) are biased to low blueshifts. If the C IV blueshift is related to the strength of a radiation line-driven wind, this ﬁnding is very interesting in terms of the long-observed anti-correlation between radio-loud quasars and BALQSOs (Broad Absorption

Chapter 6: Discussion 63

Line QSOs; Stocke et al. 1992). While sources with strong radio lobes tend to avoid BALQSOs (or vice versa) (Stocke et al. 1992; Reichard et al. 2003; Richards et al. 2011), they are not mutually exclusive (Becker et al. 2000; Welling et al. 2014)1. In the Richards et al.(2011) picture, all quasars have some sort of wind; it is just that objects displaying absorption troughs that meet the traditional

BALQSO deﬁnition will have stronger radiation line-driven winds than quasars that do not. Further,

Richards et al.(2011) argue that emission line properties can be used to determine the strength of radiation line-driving and, thus, of seeing BAL troughs along other lines of sight. As high L/LEdd might be most expected to lead to a strong radiation line-driven wind, the general anti-correlation of BALQSOs and RL quasars would be expected.

Second generation quasars in the model of Sikora(2009) could explain the existence of the rare

RL BALQSOs. RL BALQSOs could be those BALQSOs undergoing a “2nd major merger” (Sikora

2009) and getting spun up enough to produce a radio jet or, alternatively, those RL quasars which undergo a signiﬁcant increase in accretion rate that generates a radiation line-driven wind. Another possibility is that RL BALQSOs could be related to radio emission resulting from interactions between their outﬂows and the inter-stellar medium (Jiang et al. 2010; Zakamska and Greene 2014).

6.4 On the Meaning of Radio-Quiet

My goal was to identify non-radio properties of quasars that could be used to predict whether an individual quasar is likely to be a strong radio source or not. In that sense I have failed: RL quasars and at least some RQ quasars do not appear to be significantly different. That said, I have expanded the parameter space over which the RL/RQ dichotomy has been thoroughly investigated and have identified properties that suggest when an optical quasar is very unlikely to be a strong radio source.

To make further progress, it would help to be able to estimate BH spins for large samples of (distant) quasars.

Given the relative similarity of RL and RQ quasars, I emphasize that there is no such thing as a

“radio-quiet” quasar. I mean this literally in that all quasars appear to have some minimum level of radio ﬂux based on direct detections from deep observations (Kimball et al. 2011) and hinted at by

1Though I note that RL BALQSO are often either not RL (due to dust obscuration in the optical) or have relatively narrow absorption troughs that may not be consistent with a strong radiation line driven wind.

Chapter 6: Discussion 64 stacking (Figure 19 and White et al. 2007) and demographic analyses (Condon et al. 2013). However,

I also mean it ﬁguratively in that the RQ population spans a large range of continuum and emission line properties (e.g., Sulentic et al. 2000b and Richards et al. 2011) such that it cannot be considered a single monolithic object class. Investigations that have naively split the quasar population into two

(RL/RQ) should be reconsidered. If a RL sample is compared to a truly representative RQ sample, one would expect to see diﬀerences since the RQ population spans a larger range of parameter space than the RL. Comparisons of RL quasars separately with what I label HSRQ and SSRQs quasars could prove to be extremely interesting.

Chapter 6: Discussion 65

Chapter 7: Conclusions

In Section2, I argued that the seemingly discordant literature on the possible bimodality of the detected quasar population is actually in good agreement. All investigations ﬁnd that the distribution of radio-loudness is poorly ﬁt by a single component and that there is a minority population of radio-loud objects. As noted by Laor(2003), evolution of this RL population may cloud my analyses of it through the use of a single dividing line at all redshifts and luminosities.

Section 3.7 explains why I adopt αro as my measure of radio-loudness instead of log R. Although log R has been commonly used among radio astronomers, I have decided to implement αro in my analyses to make the results accessible to those who do not primarily work within the radio regime.

αro is universal in that it directly describes the shape of a quasar’s spectral energy distribution, speciﬁcally between the radio and the optical.

In Section 4.1 (Figure 4.2), I showed that the radio-loud fraction appears to evolve with both luminosity and redshift, in agreement with Jiang et al.(2007) and Balokovi´cet al.(2012). This evolution is such that the RLF most closely tracks the optical apparent magnitude, suggesting a possible bias. A radio sample covering the area of the FIRST survey to three times its depth is needed to resolve this issue.

In Section 5.4 (Figure 5.6), I found that the mean radio-loudness evolves in the exact opposite sense. Thus, it appears that the mean and extrema of the radio-loudness distribution do not track each other. As the stacking analysis is less sensitive to the ﬂux limit of the FIRST survey, the latter (evolution of the mean) result would appear to be more robust than the evolution of the RLF.

This difference could offer insight into the nature of radio emission in quasars, perhaps suggesting different tracks for the RL and radio-intermediate sources. Alternatively, it could be an indication that FIRST is indeed incomplete in a manner that clouds an understanding of the RLF evolution.

I explored the evolution of radio properties in Eigenvector 1 and C IV parameter spaces in

Sections 4.2 and 5.6 because these properties may trace the relative power of radiation line-driven 66 accretion disk winds (Richards et al. 2011). The RLF is much higher in quasars without emission properties that point to strong radiation line-driven winds. Indeed, the RLF is essentially zero for quasars with the highest C IV blueshifts (Figure 4.4). The mean radio-loudness (Figure 5.11) shows a similar, albeit somewhat weaker, trend. The trends in RLF (Figure 4.3) and mean radio-loudness

(Figure 5.10) in EV1 parameter space are broadly consistent with the C IV results. I further ﬁnd that the mean radio-loudness increases with increasing reddening of the optical continuum (Figure

5.15), which is consistent with these other ﬁndings. Contrary to Boroson(2002), I ﬁnd that RL quasars do not occupy a unique parameter space; thus, it appears that RL and RQ quasars are parallel sequences. Some additional parameter (spin?) must contribute to an object being RL, where that parameter is strongly biased to those quasars without strong radiation line-driven winds.

Sections 4.3 and 5.7 consider the radio properties of quasars as a function of BH mass, accretion rate, and Eddington ratio. I argue that BH mass estimates from survey-quality spectral measurements of C IV have catastrophic errors in luminous quasars. These errors are identiﬁed by a radical change in the radio properties of quasars as a function of BH mass with redshift and have led some previous works to questionable conclusions regarding the redshift evolution of L/LEdd for RL quasars. Ignoring the biased C IV BH mass results (or assuming that the actual C IV BH masses are inverted from their apparent values), I ﬁnd that the radio-loud fraction is highest for the largest BH masses (at a given luminosity) (Figure 4.5). This means that the RLF is a function of L/LEdd, in agreement with past results. The mean radio-loudness shows a similar but somewhat weaker trend

(Figure 5.12).

Further progress must come in the context of the realization that there is no typical RQ quasar with which to contrast the RL population. Rather, RL quasars should be compared to RQ quasars that have similar (non-radio) properties (Population B in Sulentic et al. 2000a,b and HSRQ in

Kruczek et al. 2011) and contrasted with those RQ quasars that exhibit dissimilar (non-radio) properties (Population A in Sulentic et al. 2000a,b and SSRQ in Kruczek et al. 2011). While I ﬁnd little to diﬀerentiate RL quasars and HSRQs, the suggestion by Sikora et al.(2007) of RL quasars being spun-up by second generation mergers is an intriguing one.

Chapter 7: Conclusions 67

Bibliography

K. N. Abazajian, J. K. Adelman-McCarthy, M. A. Agüeros,S. S. Allam, C. Allende Prieto, D. An, K. S. J. Anderson, S. F. Anderson, J. Annis, N. A. Bahcall, and et al. The Seventh Data Release of the Sloan Digital Sky Survey. ApJS, 182:543, June 2009. doi: 10.1088/0067-0049/182/2/543. C. P. Ahn, R. Alexandroff, C. Allende Prieto, S. F. Anderson, T. Anderton, B. H. Andrews, E.´ Aubourg, S. Bailey, E. Balbinot, R. Barnes, and et al. The Ninth Data Release of the Sloan Digital Sky Survey: First Spectroscopic Data from the SDSS-III Baryon Oscillation Spectroscopic Survey. ApJS, 203:21, December 2012. doi: 10.1088/0067-0049/203/2/21. R. J. Assef, K. D. Denney, C. S. Kochanek, B. M. Peterson, S. Kozlowski, N. Ageorges, R. S. Barrows, P. Buschkamp, M. Dietrich, E. Falco, C. Feiz, H. Gemperlein, A. Germeroth, C. J. Grier, R. Hofmann, M. Juette, R. Khan, M. Kilic, V. Knierim, W. Laun, R. Lederer, M. Lehmitz, R. Lenzen, U. Mall, K. K. Madsen, H. Mandel, P. Martini, S. Mathur, K. Mogren, P. Mueller, V. Naranjo, A. Pasquali, K. Polsterer, R. W. Pogge, A. Quirrenbach, W. Seifert, D. Stern, B. Shappee, C. Storz, J. Van Saders, P. Weiser, and D. Zhang. Black Hole Mass Estimates Based on C IV are Consistent with Those Based on the Balmer Lines. ApJ, 742:93, December 2011. doi: 10.1088/0004-637X/742/2/93. Y. Avni and H. Tananbaum. On the cosmological evolution of the X-ray emission from quasars. ApJ, 262:L17–L21, November 1982. doi: 10.1086/183903. M. Baloković,V. Smolˇcić, Z.ˇ Ivezić,G. Zamorani, E. Schinnerer, and B. C. Kelly. Disclosing the Radio Loudness Distribution Dichotomy in Quasars: An Unbiased Monte Carlo Approach Applied to the SDSS-FIRST Quasar Sample. ApJ, 759:30, November 2012. doi: 10.1088/0004-637X/759/ 1/30. P. D. Barthel. Is every quasar beamed? ApJ, 336:606–611, January 1989. doi: 10.1086/167038. R. Barvainis, J. Lehár,M. Birkinshaw, H. Falcke, and K. M. Blundell. Radio Variability of Radio- quiet and Radio-loud Quasars. ApJ, 618:108–122, January 2005. doi: 10.1086/425859. R. H. Becker, R. L. White, and D. J. Helfand. The FIRST Survey: Faint Images of the Radio Sky at Twenty Centimeters. ApJ, 450:559, September 1995. doi: 10.1086/176166. R. H. Becker, R. L. White, M. D. Gregg, M. S. Brotherton, S. A. Laurent-Muehleisen, and N. Arav. Properties of Radio-selected Broad Absorption Line Quasars from the First Bright Quasar Survey. ApJ, 538:72–82, July 2000. doi: 10.1086/309099. R. D. Blandford. Physical processes in active galactic nuclei. In R. D. Blandford, H. Netzer, L. Woltjer, T. J.-L. Courvoisier, and M. Mayor, editors, Active Galactic Nuclei, pages 161–275, 1990. R. D. Blandford and D. G. Payne. Hydromagnetic flows from accretion discs and the production of radio jets. MNRAS, 199:883–903, June 1982. R. D. Blandford and R. L. Znajek. Electromagnetic extraction of energy from Kerr black holes. MNRAS, 179:433–456, May 1977. M. R. Blanton, D. W. Hogg, N. A. Bahcall, J. Brinkmann, M. Britton, A. J. Connolly, I. Csabai, M. Fukugita, J. Loveday, A. Meiksin, J. A. Munn, R. C. Nichol, S. Okamura, T. Quinn, D. P. Schneider, K. Shimasaku, M. A. Strauss, M. Tegmark, M. S. Vogeley, and D. H. Weinberg. The Galaxy Luminosity Function and Luminosity Density at Redshift z = 0.1. ApJ, 592:819–838, August 2003. doi: 10.1086/375776. 68

K. M. Blundell and A. J. Beasley. The central engines of radio-quiet quasars. MNRAS, 299:165–170, August 1998. doi: 10.1046/j.1365-8711.1998.01752.x. K. M. Blundell and Z. Kuncic. On the Origin of Radio Core Emission in Radio-quiet Quasars. ApJ, 668:L103–L106, October 2007. doi: 10.1086/522695.

T. A. Boroson. Black Hole Mass and Eddington Ratio as Drivers for the Observable Properties of Radio-loud and Radio-quiet QSOs. ApJ, 565:78–85, January 2002. doi: 10.1086/324486. T. A. Boroson and R. F. Green. The emission-line properties of low-redshift quasi-stellar objects. ApJS, 80:109–135, May 1992. doi: 10.1086/191661. J. Bovy, J. F. Hennawi, D. W. Hogg, A. D. Myers, J. A. Kirkpatrick, D. J. Schlegel, N. P. Ross, E. S. Sheldon, I. D. McGreer, D. P. Schneider, and B. A. Weaver. Think Outside the Color Box: Probabilistic Target Selection and the SDSS-XDQSO Quasar Targeting Catalog. ApJ, 729:141, March 2011. doi: 10.1088/0004-637X/729/2/141. M. S. Brotherton and P. J. Francis. The Intermediate Line Region and the Baldwin Eﬀect. In G. Ferland and J. Baldwin, editors, Quasars and Cosmology, volume 162 of Astronomical Society of the Paciﬁc Conference Series, page 395, 1999. D. A. Casebeer, K. M. Leighly, and E. Baron. FUSE Observation of the Narrow-Line Seyfert 1 Galaxy RE 1034+39: Dependence of Broad Emission Line Strengths on the Shape of the Pho- toionizing Spectrum. ApJ, 637:157–182, January 2006. doi: 10.1086/498125.

Hong-Yee Chiu. Gravitational collapse. Physics Today, 17(5):21–34, 1964. doi: 10.1063/1.3051610. URL http://link.aip.org/link/?PTO/17/21/1. M. Cirasuolo, A. Celotti, M. Magliocchetti, and L. Danese. Is there a dichotomy in the radio loudness distribution of quasars? MNRAS, 346:447–455, December 2003a. doi: 10.1046/j.1365-2966.2003. 07105.x.

M. Cirasuolo, M. Magliocchetti, A. Celotti, and L. Danese. The radio-loud/radio-quiet dichotomy: news from the 2dF QSO Redshift Survey. MNRAS, 341:993–1004, May 2003b. doi: 10.1046/j. 1365-8711.2003.06485.x. J. J. Condon, S. L. Odell, J. J. Puschell, and W. A. Stein. Radio emission from radio-quiet quasars. Nature, 283:357, January 1980. doi: 10.1038/283357a0.

J. J. Condon, W. D. Cotton, E. W. Greisen, Q. F. Yin, R. A. Perley, G. B. Taylor, and J. J. Broderick. The NRAO VLA Sky Survey. AJ, 115:1693–1716, May 1998. doi: 10.1086/300337. J. J. Condon, W. D. Cotton, and J. J. Broderick. Radio Sources and Star Formation in the Local Universe. AJ, 124:675–689, August 2002. doi: 10.1086/341650.

J. J. Condon, K. I. Kellermann, A. E. Kimball, Z.ˇ Ivezi´c,and R. A. Perley. Active Galactic Nucleus and Starburst Radio Emission from Optically Selected Quasi-stellar Objects. ApJ, 768:37, May 2013. doi: 10.1088/0004-637X/768/1/37. S. M. Croom, R. J. Smith, B. J. Boyle, T. Shanks, L. Miller, P. J. Outram, and N. S. Loaring. The 2dF QSO Redshift Survey - XII. The spectroscopic catalogue and luminosity function. MNRAS, 349:1397–1418, April 2004. doi: 10.1111/j.1365-2966.2004.07619.x. S. M. Croom, G. T. Richards, T. Shanks, B. J. Boyle, R. G. Sharp, J. Bland-Hawthorn, T. Bridges, R. J. Brunner, R. Cannon, D. Carson, K. Chiu, M. Colless, W. Couch, R. de Propris, M. J. Drinkwater, A. Edge, S. Fine, J. Loveday, L. Miller, A. D. Myers, R. C. Nichol, P. Outram, K. Pimbblet, I. Roseboom, N. Ross, D. P. Schneider, A. Smith, C. Stoughton, M. A. Strauss, and D. Wake. The 2dF-SDSS LRG and QSO Survey: the spectroscopic QSO catalogue. MNRAS, 392: 19–44, January 2009. doi: 10.1111/j.1365-2966.2008.14052.x.

Bibliography 69

K. D. Denney. Are Outﬂows Biasing Single-epoch C IV Black Hole Mass Estimates? ApJ, 759:44, November 2012. doi: 10.1088/0004-637X/759/1/44. K. D. Denney, R. W. Pogge, R. J. Assef, C. S. Kochanek, B. M. Peterson, and M. Vestergaard. C IV Line-width Anomalies: The Perils of Low Signal-to-noise Spectra. ApJ, 775:60, September 2013. doi: 10.1088/0004-637X/775/1/60.

J. S. Dunlop, R. J. McLure, M. J. Kukula, S. A. Baum, C. P. O’Dea, and D. H. Hughes. Quasars, their host galaxies and their central black holes. MNRAS, 340:1095–1135, April 2003. doi: 10. 1046/j.1365-8711.2003.06333.x. D. O. Edge, J. R. Shakeshaft, W. B. McAdam, J. E. Baldwin, and S. Archer. A survey of radio sources at a frequency of 159 Mc/s. MmRAS, 68:37–60, 1959. M. Elvis. A Structure for Quasars. ApJ, 545:63–76, December 2000. doi: 10.1086/317778. M. Elvis, B. J. Wilkes, J. C. McDowell, R. F. Green, J. Bechtold, S. P. Willner, M. S. Oey, E. Polom- ski, and R. Cutri. Atlas of quasar energy distributions. ApJS, 95:1–68, November 1994. doi: 10.1086/192093.

D. J. E. Floyd, M. J. Kukula, J. S. Dunlop, R. J. McLure, L. Miller, W. J. Percival, S. A. Baum, and C. P. O’Dea. The host galaxies of luminous quasars. MNRAS, 355:196–220, November 2004. doi: 10.1111/j.1365-2966.2004.08315.x. M. Fukugita, T. Ichikawa, J. E. Gunn, M. Doi, K. Shimasaku, and D. P. Schneider. The Sloan Digital Sky Survey Photometric System. AJ, 111:1748, April 1996. doi: 10.1086/117915. S. C. Gallagher, G. T. Richards, P. B. Hall, W. N. Brandt, D. P. Schneider, and D. E. Vanden Berk. X-Ray Insights into Interpreting C IV Blueshifts and Optical/Ultraviolet Continua. AJ, 129:567–577, February 2005. doi: 10.1086/426913. D. Garofalo. Signatures of Black Hole Spin in Galaxy Evolution. ApJ, 699:L52–L54, July 2009. doi: 10.1088/0004-637X/699/1/L52. D. Garofalo, D. A. Evans, and R. M. Sambruna. The evolution of radio-loud active galactic nuclei as a function of black hole spin. MNRAS, 406:975–986, August 2010. doi: 10.1111/j.1365-2966. 2010.16797.x.

C. M. Gaskell. A redshift diﬀerence between high and low ionization emission-line regions in QSOs - Evidence for radial motions. ApJ, 263:79–86, December 1982. doi: 10.1086/160481. E. Glikman, D. J. Helfand, and R. L. White. A Near-Infrared Spectral Template for Quasars. ApJ, 640:579–591, April 2006. doi: 10.1086/500098. P. Goldschmidt, M. J. Kukula, L. Miller, and J. S. Dunlop. A Comparison of the Optical Properties of Radio-loud and Radio-quiet Quasars. ApJ, 511:612–624, February 1999. doi: 10.1086/306699. J. L. Greenstein and T. A. Matthews. Redshift of the Radio Source 3C 48. AJ, 68:279, 1963. doi: 10.1086/109140. J. E. Gunn, M. Carr, C. Rockosi, M. Sekiguchi, K. Berry, B. Elms, E. de Haas, Z.ˇ Ivezi´c,G. Knapp, R. Lupton, G. Pauls, R. Simcoe, R. Hirsch, D. Sanford, S. Wang, D. York, F. Harris, J. Annis, L. Bartozek, W. Boroski, J. Bakken, M. Haldeman, S. Kent, S. Holm, D. Holmgren, D. Petrav- ick, A. Prosapio, R. Rechenmacher, M. Doi, M. Fukugita, K. Shimasaku, N. Okada, C. Hull, W. Siegmund, E. Mannery, M. Blouke, D. Heidtman, D. Schneider, R. Lucinio, and J. Brinkman. The Sloan Digital Sky Survey Photometric Camera. AJ, 116:3040–3081, December 1998. doi: 10.1086/300645.

Bibliography 70

J. E. Gunn, W. A. Siegmund, E. J. Mannery, R. E. Owen, C. L. Hull, R. F. Leger, L. N. Carey, G. R. Knapp, D. G. York, W. N. Boroski, S. M. Kent, R. H. Lupton, C. M. Rockosi, M. L. Evans, P. Waddell, J. E. Anderson, J. Annis, J. C. Barentine, L. M. Bartoszek, S. Bastian, S. B. Bracker, H. J. Brewington, C. I. Briegel, J. Brinkmann, Y. J. Brown, M. A. Carr, P. C. Czarapata, C. C. Drennan, T. Dombeck, G. R. Federwitz, B. A. Gillespie, C. Gonzales, S. U. Hansen, M. Harvanek, J. Hayes, W. Jordan, E. Kinney, M. Klaene, S. J. Kleinman, R. G. Kron, J. Kresinski, G. Lee, S. Limmongkol, C. W. Lindenmeyer, D. C. Long, C. L. Loomis, P. M. McGehee, P. M. Mantsch, E. H. Neilsen, Jr., R. M. Neswold, P. R. Newman, A. Nitta, J. Peoples, Jr., J. R. Pier, P. S. Prieto, A. Prosapio, C. Rivetta, D. P. Schneider, S. Snedden, and S.-i. Wang. The 2.5 m Telescope of the Sloan Digital Sky Survey. AJ, 131:2332–2359, April 2006. doi: 10.1086/500975. P. C. Hewett and V. Wild. Improved redshifts for SDSS quasar spectra. MNRAS, 405:2302–2316, July 2010. doi: 10.1111/j.1365-2966.2010.16648.x. L. C. Ho. On the Relationship between Radio Emission and Black Hole Mass in Galactic Nuclei. ApJ, 564:120–132, January 2002. doi: 10.1086/324399. D. W. Hogg, D. P. Finkbeiner, D. J. Schlegel, and J. E. Gunn. A Photometricity and Extinction Monitor at the Apache Point Observatory. AJ, 122:2129–2138, October 2001. doi: 10.1086/323103. Z.ˇ Ivezić, K. Menou, G. R. Knapp, M. A. Strauss, R. H. Lupton, D. E. Vanden Berk, G. T. Richards, C. Tremonti, M. A. Weinstein, S. Anderson, N. A. Bahcall, R. H. Becker, M. Bernardi, M. Blanton, D. Eisenstein, X. Fan, D. Finkbeiner, K. Finlator, J. Frieman, J. E. Gunn, P. B. Hall, R. S. J. Kim, A. Kinkhabwala, V. K. Narayanan, C. M. Rockosi, D. Schlegel, D. P. Schneider, I. Strateva, M. SubbaRao, A. R. Thakar, W. Voges, R. L. White, B. Yanny, J. Brinkmann, M. Doi, M. Fukugita, G. S. Hennessy, J. A. Munn, R. C. Nichol, and D. G. York. Optical and Radio Properties of Extragalactic Sources Observed by the FIRST Survey and the Sloan Digital Sky Survey. AJ, 124:2364–2400, November 2002. doi: 10.1086/344069. Z.ˇ Ivezić, R. H. Lupton, D. Schlegel, B. Boroski, J. Adelman-McCarthy, B. Yanny, S. Kent, C. Stoughton, D. Finkbeiner, N. Padmanabhan, C. M. Rockosi, J. E. Gunn, G. R. Knapp, M. A. Strauss, G. T. Richards, D. Eisenstein, T. Nicinski, S. J. Kleinman, J. Krzesinski, P. R. Newman, S. Snedden, A. R. Thakar, A. Szalay, J. A. Munn, J. A. Smith, D. Tucker, and B. C. Lee. SDSS data management and photometric quality assessment. Astronomische Nachrichten, 325:583–589, October 2004. doi: 10.1002/asna.200410285. M. J. Jarvis and R. J. McLure. On the black hole mass-radio luminosity relation for flat-spectrum radio-loud quasars. MNRAS, 336:L38–L42, October 2002. doi: 10.1046/j.1365-8711.2002.05997.x. S. Jester. A Simple Test for the Existence of Two Accretion Modes in Active Galactic Nuclei. ApJ, 625:667–679, June 2005. doi: 10.1086/429812. L. Jiang, X. Fan, Z.ˇ Ivezić,G. T. Richards, D. P. Schneider, M. A. Strauss, and B. C. Kelly. The Radio-Loud Fraction of Quasars is a Strong Function of Redshift and Optical Luminosity. ApJ, 656:680–690, February 2007. doi: 10.1086/510831. Y.-F. Jiang, L. Ciotti, J. P. Ostriker, and A. Spitkovsky. Synchrotron Emission from Elliptical Galaxies Consequent to Active Galactic Nucleus Outbursts. ApJ, 711:125–137, March 2010. doi: 10.1088/0004-637X/711/1/125.

D. W. Just, W. N. Brandt, O. Shemmer, A. T. Steﬀen, D. P. Schneider, G. Chartas, and G. P. Garmire. The X-Ray Properties of the Most Luminous Quasars from the Sloan Digital Sky Survey. ApJ, 665:1004–1022, August 2007. doi: 10.1086/519990. K. I. Kellermann, R. Sramek, M. Schmidt, D. B. Shaﬀer, and R. Green. VLA observations of objects in the Palomar Bright Quasar Survey. AJ, 98:1195–1207, October 1989. doi: 10.1086/115207.

Bibliography 71

A. E. Kimball, K. I. Kellermann, J. J. Condon, Z.ˇ Ivezi´c,and R. A. Perley. The Two-component Radio Luminosity Function of Quasi-stellar Objects: Star Formation and Active Galactic Nucleus. ApJ, 739:L29, September 2011. doi: 10.1088/2041-8205/739/1/L29. C. S. Kochanek, D. J. Eisenstein, R. J. Cool, N. Caldwell, R. J. Assef, B. T. Jannuzi, C. Jones, S. S. Murray, W. R. Forman, A. Dey, M. J. I. Brown, P. Eisenhardt, A. H. Gonzalez, P. Green, and D. Stern. AGES: The AGN and Galaxy Evolution Survey. ApJS, 200:8, May 2012. doi: 10.1088/0067-0049/200/1/8. C. M. Krawczyk, G. T. Richards, S. S. Mehta, M. S. Vogeley, S. C. Gallagher, K. M. Leighly, N. P. Ross, and D. P. Schneider. Mean Spectral Energy Distributions and Bolometric Corrections for Luminous Quasars. ApJS, 206:4, May 2013. doi: 10.1088/0067-0049/206/1/4.

N. E. Kruczek, G. T. Richards, S. C. Gallagher, R. P. Deo, P. B. Hall, P. C. Hewett, K. M. Leighly, C. M. Krawczyk, and D. Proga. C IV Emission and the Ultraviolet through X-Ray Spectral Energy Distribution of Radio-quiet Quasars. AJ, 142:130, October 2011. doi: 10.1088/0004-6256/142/4/ 130. M. Lacy, S. A. Laurent-Muehleisen, S. E. Ridgway, R. H. Becker, and R. L. White. The Radio Luminosity-Black Hole Mass Correlation for Quasars from the FIRST Bright Quasar Survey and a “Uniﬁcation Scheme” for Radio-loud and Radio-quiet Quasars. ApJ, 551:L17–L21, April 2001. doi: 10.1086/319836. A. Laor. On Black Hole Masses and Radio Loudness in Active Galactic Nuclei. ApJ, 543:L111–L114, November 2000. doi: 10.1086/317280.

A. Laor. Some Comments related to AGN Radio Loudness. ArXiv Astrophysics e-prints, December 2003. K. M. Leighly. Hubble Space Telescope STIS Ultraviolet Spectral Evidence of Outﬂow in Extreme Narrow-Line Seyfert 1 Galaxies. II. Modeling and Interpretation. ApJ, 611:125–152, August 2004. doi: 10.1086/422089. K. M. Leighly, J. P. Halpern, E. B. Jenkins, D. Grupe, J. Choi, and K. B. Prescott. The Intrinsically X-Ray Weak Quasar PHL 1811. I. X-Ray Observations and Spectral Energy Distribution. ApJ, 663:103–117, July 2007. doi: 10.1086/518017. S. J. Lilly, O. Le F`evre, A. Renzini, G. Zamorani, M. Scodeggio, T. Contini, C. M. Carollo, G. Hasinger, J.-P. Kneib, A. Iovino, V. Le Brun, C. Maier, V. Mainieri, M. Mignoli, J. Silver- man, L. A. M. Tasca, M. Bolzonella, A. Bongiorno, D. Bottini, P. Capak, K. Caputi, A. Cimatti, O. Cucciati, E. Daddi, R. Feldmann, P. Franzetti, B. Garilli, L. Guzzo, O. Ilbert, P. Kampczyk, K. Kovac, F. Lamareille, A. Leauthaud, J.-F. L. Borgne, H. J. McCracken, C. Marinoni, R. Pello, E. Ricciardelli, C. Scarlata, D. Vergani, D. B. Sanders, E. Schinnerer, N. Scoville, Y. Taniguchi, S. Arnouts, H. Aussel, S. Bardelli, M. Brusa, A. Cappi, P. Ciliegi, A. Finoguenov, S. Foucaud, R. Franceschini, C. Halliday, C. Impey, C. Knobel, A. Koekemoer, J. Kurk, D. Maccagni, S. Mad- dox, B. Marano, G. Marconi, B. Meneux, B. Mobasher, C. Moreau, J. A. Peacock, C. Porciani, L. Pozzetti, R. Scaramella, D. Schiminovich, P. Shopbell, I. Smail, D. Thompson, L. Tresse, G. Vet- tolani, A. Zanichelli, and E. Zucca. zCOSMOS: A Large VLT/VIMOS Redshift Survey Covering 0 < z < 3 in the COSMOS Field. ApJS, 172:70–85, September 2007. doi: 10.1086/516589.

R. Lupton, J. E. Gunn, Z. Ivezić,G. R. Knapp, and S. Kent. The SDSS Imaging Pipelines. In F. R. Harnden, Jr., F. A. Primini, and H. E. Payne, editors, Astronomical Data Analysis Software and Systems X, volume 238 of Astronomical Society of the Pacific Conference Series, page 269, 2001. E. Lusso, A. Comastri, C. Vignali, G. Zamorani, M. Brusa, R. Gilli, K. Iwasawa, M. Salvato, F. Civano, M. Elvis, A. Merloni, A. Bongiorno, J. R. Trump, A. M. Koekemoer, E. Schinnerer, E. Le Floc’h, N. Cappelluti, K. Jahnke, M. Sargent, J. Silverman, V. Mainieri, F. Fiore, M. Bol- zonella, O. Le Fèvre, B. Garilli, A. Iovino, J. P. Kneib, F. Lamareille, S. Lilly, M. Mignoli,

Bibliography 72

M. Scodeggio, and D. Vergani. The X-ray to optical-UV luminosity ratio of X-ray selected type 1 AGN in XMM-COSMOS. A&A, 512:A34, March 2010. doi: 10.1051/0004-6361/200913298. D. Lynden-Bell. Galactic Nuclei as Collapsed Old Quasars. Nature, 223:690–694, August 1969. doi: 10.1038/223690a0.

N. Maddox, P. C. Hewett, C. P´eroux,D. B. Nestor, and L. Wisotzki. The large area KX quasar catalogue - I. Analysis of the photometric redshift selection and the complete quasar catalogue. MNRAS, 424:2876–2895, August 2012. doi: 10.1111/j.1365-2966.2012.21427.x. E. K. Mahony, E. M. Sadler, S. M. Croom, R. D. Ekers, I. J. Feain, and T. Murphy. Is the Observed High-frequency Radio Luminosity Distribution of QSOs Bimodal? ApJ, 754:12, July 2012. doi: 10.1088/0004-637X/754/1/12. M. A. Malkan. The underlying galaxies of quasars. II - Imaging of a radio-loud sample. ApJ, 287: 555–565, December 1984. doi: 10.1086/162714. T. A. Matthews and A. R. Sandage. Optical Identiﬁcation of 3c 48, 3c 196, and 3c 286 with Stellar Objects. ApJ, 138:30, July 1963. doi: 10.1086/147615.

R. J. McLure and M. J. Jarvis. Measuring the black hole masses of high-redshift quasars. MNRAS, 337:109–116, November 2002. doi: 10.1046/j.1365-8711.2002.05871.x. B. P. Miller, W. N. Brandt, D. P. Schneider, R. R. Gibson, A. T. Steﬀen, and J. Wu. X-ray Emission from Optically Selected Radio-intermediate and Radio-loud Quasars. ApJ, 726:20, January 2011. doi: 10.1088/0004-637X/726/1/20. L. Miller, J. A. Peacock, and A. R. G. Mead. The bimodal radio luminosity function of quasars. MNRAS, 244:207–213, May 1990. R. L. Moore and H. S. Stockman. A comparison of the properties of highly polarized QSOs versus low-polarization QSOs. ApJ, 279:465–484, April 1984. doi: 10.1086/161911.

N. Murray, J. Chiang, S. A. Grossman, and G. M. Voit. Accretion Disk Winds from Active Galactic Nuclei. ApJ, 451:498, October 1995. doi: 10.1086/176238. G. Neugebauer, G. K. Miley, B. T. Soifer, and P. E. Clegg. Quasars measured by the Infrared Astronomical Satellite. ApJ, 308:815–828, September 1986. doi: 10.1086/164553.

J. B. Oke and J. E. Gunn. Secondary standard stars for absolute spectrophotometry. ApJ, 266: 713–717, March 1983. doi: 10.1086/160817. N. Palanque-Delabrouille, C. Magneville, C. Yèche, S. Eftekharzadeh, A. D. Myers, P. Petitjean, I. Pâris,E. Aubourg, I. McGreer, X. Fan, A. Dey, D. Schlegel, S. Bailey, D. Bizayev, A. Bolton, K. Dawson, G. Ebelke, J. Ge, E. Malanushenko, V. Malanushenko, D. Oravetz, K. Pan, N. P. Ross, D. P. Schneider, E. Sheldon, A. Simmons, J. Tinker, M. White, and C. Willmer. Luminosity function from dedicated SDSS-III and MMT data of quasars in 0.7 < z < 4.0 selected with a new approach. A&A, 551:A29, March 2013. doi: 10.1051/0004-6361/201220379. C. Papovich, R. Cool, D. Eisenstein, E. Le Floc’h, X. Fan, R. C. Kennicutt, Jr., J. D. T. Smith, G. H. Rieke, and M. Vestergaard. An MMT Hectospec Redshift Survey of 24 µm Sources in the Spitzer First Look Survey. AJ, 132:231–241, July 2006. doi: 10.1086/504598. I. Pâris,P. Petitjean, E.´ Aubourg, S. Bailey, N. P. Ross, A. D. Myers, M. A. Strauss, S. F. Anderson, E. Arnau, J. Bautista, D. Bizyaev, A. S. Bolton, J. Bovy, W. N. Brandt, H. Brewington, J. R. Browstein, N. Busca, D. Capellupo, W. Carithers, R. A. C. Croft, K. Dawson, T. Delubac, G. Ebelke, D. J. Eisenstein, P. Engelke, X. Fan, N. Filiz Ak, H. Finley, A. Font-Ribera, J. Ge, R. R. Gibson, P. B. Hall, F. Hamann, J. F. Hennawi, S. Ho, D. W. Hogg, Z.ˇ Ivezić,L. Jiang, A. E. Kimball, D. Kirkby, J. A. Kirkpatrick, K.-G. Lee, J.-M. Le Goff, B. Lundgren, C. L. MacLeod,

Bibliography 73

E. Malanushenko, V. Malanushenko, C. Maraston, I. D. McGreer, R. G. McMahon, J. Miralda- Escud´e,D. Muna, P. Noterdaeme, D. Oravetz, N. Palanque-Delabrouille, K. Pan, I. Perez-Fournon, M. M. Pieri, G. T. Richards, E. Rollinde, E. S. Sheldon, D. J. Schlegel, D. P. Schneider, A. Slosar, A. Shelden, Y. Shen, A. Simmons, S. Snedden, N. Suzuki, J. Tinker, M. Viel, B. A. Weaver, D. H. Weinberg, M. White, W. M. Wood-Vasey, and C. Y`eche. The Sloan Digital Sky Survey quasar catalog: ninth data release. A&A, 548:A66, December 2012. doi: 10.1051/0004-6361/201220142. D. Park, J.-H. Woo, K. D. Denney, and J. Shin. Calibrating C-IV-based Black Hole Mass Estimators. ApJ, 770:87, June 2013. doi: 10.1088/0004-637X/770/2/87. J. A. Peacock, L. Miller, and M. S. Longair. The statistics of radio emission from quasars. MNRAS, 218:265–278, January 1986.

J. R. Pier, J. A. Munn, R. B. Hindsley, G. S. Hennessy, S. M. Kent, R. H. Lupton, and Z.ˇ Ivezi´c. Astrometric Calibration of the Sloan Digital Sky Survey. AJ, 125:1559–1579, March 2003. doi: 10.1086/346138. D. Proga, J. M. Stone, and T. R. Kallman. Dynamics of Line-driven Disk Winds in Active Galactic Nuclei. ApJ, 543:686–696, November 2000. doi: 10.1086/317154. A. Raﬁee and P. B. Hall. Biases in the quasar mass-luminosity plane. MNRAS, 415:2932–2941, August 2011. doi: 10.1111/j.1365-2966.2011.18910.x. S. E. Rafter, D. M. Crenshaw, and P. J. Wiita. Radio Properties of Low-Redshift Broad Line Active Galactic Nuclei. AJ, 137:42–52, January 2009. doi: 10.1088/0004-6256/137/1/42.

T. A. Reichard, G. T. Richards, P. B. Hall, D. P. Schneider, D. E. Vanden Berk, X. Fan, D. G. York, G. R. Knapp, and J. Brinkmann. Continuum and Emission-Line Properties of Broad Absorption Line Quasars. AJ, 126:2594–2607, December 2003. doi: 10.1086/379293. G. T. Richards, X. Fan, H. J. Newberg, M. A. Strauss, D. E. Vanden Berk, D. P. Schneider, B. Yanny, A. Boucher, S. Burles, J. A. Frieman, J. E. Gunn, P. B. Hall, Z.ˇ Ivezić, S. Kent, J. Loveday, R. H. Lupton, C. M. Rockosi, D. J. Schlegel, C. Stoughton, M. SubbaRao, and D. G. York. Spectroscopic Target Selection in the Sloan Digital Sky Survey: The Quasar Sample. AJ, 123:2945–2975, June 2002a. doi: 10.1086/340187. G. T. Richards, D. E. Vanden Berk, T. A. Reichard, P. B. Hall, D. P. Schneider, M. SubbaRao, A. R. Thakar, and D. G. York. Broad Emission-Line Shifts in Quasars: An Orientation Measure for Radio-Quiet Quasars? AJ, 124:1–17, July 2002b. doi: 10.1086/341167. G. T. Richards, P. B. Hall, D. E. Vanden Berk, M. A. Strauss, D. P. Schneider, M. A. Weinstein, T. A. Reichard, D. G. York, G. R. Knapp, X. Fan, Z.ˇ Ivezić,J. Brinkmann, T. Budavári,I. Csabai, and R. C. Nichol. Red and Reddened Quasars in the Sloan Digital Sky Survey. AJ, 126:1131–1147, September 2003. doi: 10.1086/377014.

G. T. Richards, M. A. Strauss, X. Fan, P. B. Hall, S. Jester, D. P. Schneider, D. E. Vanden Berk, C. Stoughton, S. F. Anderson, R. J. Brunner, J. Gray, J. E. Gunn, Z.ˇ Ivezić,M. K. Kirkland, G. R. Knapp, J. Loveday, A. Meiksin, A. Pope, A. S. Szalay, A. R. Thakar, B. Yanny, D. G. York, J. C. Barentine, H. J. Brewington, J. Brinkmann, M. Fukugita, M. Harvanek, S. M. Kent, S. J. Kleinman, J. Krzesiński, D. C. Long, R. H. Lupton, T. Nash, E. H. Neilsen, Jr., A. Nitta, D. J. Schlegel, and S. A. Snedden. The Sloan Digital Sky Survey Quasar Survey: Quasar Luminosity Function from Data Release 3. AJ, 131:2766–2787, June 2006. doi: 10.1086/503559. G. T. Richards, A. D. Myers, A. G. Gray, R. N. Riegel, R. C. Nichol, R. J. Brunner, A. S. Szalay, D. P. Schneider, and S. F. Anderson. Efficient Photometric Selection of Quasars from the Sloan Digital Sky Survey. II. ˜1,000,000 Quasars from Data Release 6. ApJS, 180:67–83, January 2009. doi: 10.1088/0067-0049/180/1/67.

Bibliography 74

G. T. Richards, N. E. Kruczek, S. C. Gallagher, P. B. Hall, P. C. Hewett, K. M. Leighly, R. P. Deo, R. M. Kratzer, and Y. Shen. Uniﬁcation of Luminous Type 1 Quasars through C IV Emission. AJ, 141:167, May 2011. doi: 10.1088/0004-6256/141/5/167. J. C. Runnoe, M. S. Brotherton, Z. Shang, and M. A. DiPompeo. Rehabilitating C IV-based black hole mass estimates in quasars. MNRAS, 434:848–861, September 2013. doi: 10.1093/mnras/ stt1077. A. Sandage. The Existence of a Major New Constituent of the Universe: the Quasistellar Galaxies. ApJ, 141:1560, May 1965. doi: 10.1086/148245. D. B. Sanders, E. S. Phinney, G. Neugebauer, B. T. Soifer, and K. Matthews. Continuum energy distribution of quasars - Shapes and origins. ApJ, 347:29–51, December 1989. doi: 10.1086/168094. P. A. G. Scheuer and A. C. S. Readhead. Superluminally expanding radio sources and the radio-quiet QSOs. Nature, 277:182–185, January 1979. doi: 10.1038/277182a0. D. J. Schlegel, D. P. Finkbeiner, and M. Davis. Maps of Dust Infrared Emission for Use in Estimation of Reddening and Cosmic Microwave Background Radiation Foregrounds. ApJ, 500:525, June 1998. doi: 10.1086/305772. M. Schmidt. 3C 273 : A Star-Like Object with Large Red-Shift. Nature, 197:1040, March 1963. doi: 10.1038/1971040a0. M. Schmidt. Space Distribution and Luminosity Functions of Quasars. ApJ, 162:371, November 1970. doi: 10.1086/150668. M. Schmidt and R. F. Green. Quasar evolution derived from the Palomar bright quasar survey and other complete quasar surveys. ApJ, 269:352–374, June 1983. doi: 10.1086/161048. M. Schmidt and T. A. Matthews. Redshift of the Quasi-Stellar Radio Sources 3c 47 and 3c 147. ApJ, 139:781, February 1964. doi: 10.1086/147815.

D. P. Schneider, G. T. Richards, P. B. Hall, M. A. Strauss, S. F. Anderson, T. A. Boroson, N. P. Ross, Y. Shen, W. N. Brandt, X. Fan, N. Inada, S. Jester, G. R. Knapp, C. M. Krawczyk, A. R. Thakar, D. E. Vanden Berk, W. Voges, B. Yanny, D. G. York, N. A. Bahcall, D. Bizyaev, M. R. Blanton, H. Brewington, J. Brinkmann, D. Eisenstein, J. A. Frieman, M. Fukugita, J. Gray, J. E. Gunn, P. Hibon, Z.ˇ Ivezi´c,S. M. Kent, R. G. Kron, M. G. Lee, R. H. Lupton, E. Malanushenko, V. Malanushenko, D. Oravetz, K. Pan, J. R. Pier, T. N. Price, III, D. H. Saxe, D. J. Schlegel, A. Simmons, S. A. Snedden, M. U. SubbaRao, A. S. Szalay, and D. H. Weinberg. The Sloan Digital Sky Survey Quasar Catalog. V. Seventh Data Release. AJ, 139:2360, June 2010. doi: 10.1088/0004-6256/139/6/2360. F. Shankar, G. R. Sivakoﬀ, M. Vestergaard, and X. Dai. The relative growth of optical and radio quasars in SDSS. MNRAS, 401:1869–1881, January 2010. doi: 10.1111/j.1365-2966.2009.15764.x. Y. Shen, G. T. Richards, M. A. Strauss, P. B. Hall, D. P. Schneider, S. Snedden, D. Bizyaev, H. Brewington, V. Malanushenko, E. Malanushenko, D. Oravetz, K. Pan, and A. Simmons. A Catalog of Quasar Properties from Sloan Digital Sky Survey Data Release 7. ApJS, 194:45, June 2011. doi: 10.1088/0067-0049/194/2/45.

M. Sikora. Radio bimodality: Spin, accretion mode, or both? Astronomische Nachrichten, 330:291, February 2009. doi: 10.1002/asna.200811178. M. Sikora,L.Stawarz, and J.-P. Lasota. Radio Loudness of Active Galactic Nuclei: Observational Facts and Theoretical Implications. ApJ, 658:815–828, April 2007. doi: 10.1086/511972.

J. Singal, V. Petrosian, A. Lawrence, andL. Stawarz. On the Radio and Optical Luminosity Evolution of Quasars. ApJ, 743:104, December 2011. doi: 10.1088/0004-637X/743/2/104.

Bibliography 75

J. Singal, V. Petrosian,L.Stawarz, and A. Lawrence. The Radio and Optical Luminosity Evolution of Quasars. II. The SDSS Sample. ApJ, 764:43, February 2013. doi: 10.1088/0004-637X/764/1/43. E. P. Smith, T. M. Heckman, G. D. Bothun, W. Romanishin, and B. Balick. On the nature of QSO host galaxies. ApJ, 306:64–89, July 1986. doi: 10.1086/164321.

A. Soltan. Masses of quasars. MNRAS, 200:115–122, July 1982. D. N. Spergel, R. Bean, O. Doré,M. R. Nolta, C. L. Bennett, J. Dunkley, G. Hinshaw, N. Jarosik, E. Komatsu, L. Page, H. V. Peiris, L. Verde, M. Halpern, R. S. Hill, A. Kogut, M. Limon, S. S. Meyer, N. Odegard, G. S. Tucker, J. L. Weiland, E. Wollack, and E. L. Wright. Three-Year Wilkinson Microwave Anisotropy Probe (WMAP) Observations: Implications for Cosmology. ApJS, 170:377–408, June 2007. doi: 10.1086/513700. A. T. Steffen, I. Strateva, W. N. Brandt, D. M. Alexander, A. M. Koekemoer, B. D. Lehmer, D. P. Schneider, and C. Vignali. The X-Ray-to-Optical Properties of Optically Selected Active Galaxies over Wide Luminosity and Redshift Ranges. AJ, 131:2826–2842, June 2006. doi: 10.1086/503627. C. C. Steidel and W. L. W. Sargent. Emission-line and continuum properties of 92 bright QSOs - Luminosity dependence and differences between radio-selected and optically selected samples. ApJ, 382:433–465, December 1991. doi: 10.1086/170732. J. T. Stocke, S. L. Morris, R. J. Weymann, and C. B. Foltz. The radio properties of the broad- absorption-line QSOs. ApJ, 396:487–503, September 1992. doi: 10.1086/171735.

C. Stoughton, R. H. Lupton, M. Bernardi, M. R. Blanton, S. Burles, F. J. Castander, A. J. Con- nolly, D. J. Eisenstein, J. A. Frieman, G. S. Hennessy, R. B. Hindsley, Z.ˇ Ivezić,S. Kent, P. Z. Kunszt, B. C. Lee, A. Meiksin, J. A. Munn, H. J. Newberg, R. C. Nichol, T. Nicinski, J. R. Pier, G. T. Richards, M. W. Richmond, D. J. Schlegel, J. A. Smith, M. A. Strauss, M. SubbaRao, A. S. Szalay, A. R. Thakar, D. L. Tucker, D. E. Vanden Berk, B. Yanny, J. K. Adelman, J. E. Anderson, Jr., S. F. Anderson, J. Annis, N. A. Bahcall, J. A. Bakken, M. Bartelmann, S. Bas- tian, A. Bauer, E. Berman, H. Böhringer,W. N. Boroski, S. Bracker, C. Briegel, J. W. Briggs, J. Brinkmann, R. Brunner, L. Carey, M. A. Carr, B. Chen, D. Christian, P. L. Colestock, J. H. Crocker, I. Csabai, P. C. Czarapata, J. Dalcanton, A. F. Davidsen, J. E. Davis, W. Dehnen, S. Dodelson, M. Doi, T. Dombeck, M. Donahue, N. Ellman, B. R. Elms, M. L. Evans, L. Eyer, X. Fan, G. R. Federwitz, S. Friedman, M. Fukugita, R. Gal, B. Gillespie, K. Glazebrook, J. Gray, E. K. Grebel, B. Greenawalt, G. Greene, J. E. Gunn, E. de Haas, Z. Haiman, M. Haldeman, P. B. Hall, M. Hamabe, B. Hansen, F. H. Harris, H. Harris, M. Harvanek, S. L. Hawley, J. J. E. Hayes, T. M. Heckman, A. Helmi, A. Henden, C. J. Hogan, D. W. Hogg, D. J. Holmgren, J. Holtz- man, C.-H. Huang, C. Hull, S.-I. Ichikawa, T. Ichikawa, D. E. Johnston, G. Kauffmann, R. S. J. Kim, T. Kimball, E. Kinney, M. Klaene, S. J. Kleinman, A. Klypin, G. R. Knapp, J. Korienek, J. Krolik, R. G. Kron, J. Krzesiński,D. Q. Lamb, R. F. Leger, S. Limmongkol, C. Lindenmeyer, D. C. Long, C. Loomis, J. Loveday, B. MacKinnon, E. J. Mannery, P. M. Mantsch, B. Margon, P. McGehee, T. A. McKay, B. McLean, K. Menou, A. Merelli, H. J. Mo, D. G. Monet, O. Naka- mura, V. K. Narayanan, T. Nash, E. H. Neilsen, Jr., P. R. Newman, A. Nitta, M. Odenkirchen, N. Okada, S. Okamura, J. P. Ostriker, R. Owen, A. G. Pauls, J. Peoples, R. S. Peterson, D. Pe- travick, A. Pope, R. Pordes, M. Postman, A. Prosapio, T. R. Quinn, R. Rechenmacher, C. H. Rivetta, H.-W. Rix, C. M. Rockosi, R. Rosner, K. Ruthmansdorfer, D. Sandford, D. P. Schneider, R. Scranton, M. Sekiguchi, G. Sergey, R. Sheth, K. Shimasaku, S. Smee, S. A. Snedden, A. Steb- bins, C. Stubbs, I. Szapudi, P. Szkody, G. P. Szokoly, S. Tabachnik, Z. Tsvetanov, A. Uomoto, M. S. Vogeley, W. Voges, P. Waddell, R. Walterbos, S.-i. Wang, M. Watanabe, D. H. Weinberg, R. L. White, S. D. M. White, B. Wilhite, D. Wolfe, N. Yasuda, D. G. York, I. Zehavi, and W. Zheng. Sloan Digital Sky Survey: Early Data Release. AJ, 123:485–548, January 2002. doi: 10.1086/324741.

P. A. Strittmatter, P. Hill, I. I. K. Pauliny-Toth, H. Steppe, and A. Witzel. Radio observations of optically selected quasars. A&A, 88:L12–L15, August 1980.

Bibliography 76

J. W. Sulentic, P. Marziani, and D. Dultzin-Hacyan. Phenomenology of Broad Emission Lines in Active Galactic Nuclei. ARA&A, 38:521–571, 2000a. doi: 10.1146/annurev.astro.38.1.521. J. W. Sulentic, T. Zwitter, P. Marziani, and D. Dultzin-Hacyan. Eigenvector 1: An Optimal Corre- lation Space for Active Galactic Nuclei. ApJ, 536:L5–L9, June 2000b. doi: 10.1086/312717.

J. W. Sulentic, R. Bachev, P. Marziani, C. A. Negrete, and D. Dultzin. C IV λ1549 as an Eigenvector 1 Parameter for Active Galactic Nuclei. ApJ, 666:757–777, September 2007. doi: 10.1086/519916. J. R. Trump, C. D. Impey, M. Elvis, P. J. McCarthy, J. P. Huchra, M. Brusa, M. Salvato, P. Capak, N. Cappelluti, F. Civano, A. Comastri, J. Gabor, H. Hao, G. Hasinger, K. Jahnke, B. C. Kelly, S. J. Lilly, E. Schinnerer, N. Z. Scoville, and V. Smolˇcić.The COSMOS Active Galactic Nucleus Spectroscopic Survey. I. XMM-Newton Counterparts. ApJ, 696:1195–1212, May 2009. doi: 10. 1088/0004-637X/696/2/1195. J. S. Ulvestad, R. R. J. Antonucci, and R. Barvainis. VLBA Imaging of Central Engines in Radio- Quiet Quasars. ApJ, 621:123–129, March 2005. doi: 10.1086/427426. C. M. Urry and P. Padovani. Unified Schemes for Radio-Loud Active Galactic Nuclei. PASP, 107: 803, September 1995. doi: 10.1086/133630. D. E. Vanden Berk, G. T. Richards, A. Bauer, M. A. Strauss, D. P. Schneider, T. M. Heckman, D. G. York, P. B. Hall, X. Fan, G. R. Knapp, S. F. Anderson, J. Annis, N. A. Bahcall, M. Bernardi, J. W. Briggs, J. Brinkmann, R. Brunner, S. Burles, L. Carey, F. J. Castander, A. J. Connolly, J. H. Crocker, I. Csabai, M. Doi, D. Finkbeiner, S. Friedman, J. A. Frieman, M. Fukugita, J. E. Gunn, G. S. Hennessy, Z.ˇ Ivezić, S. Kent, P. Z. Kunszt, D. Q. Lamb, R. F. Leger, D. C. Long, J. Loveday, R. H. Lupton, A. Meiksin, A. Merelli, J. A. Munn, H. J. Newberg, M. Newcomb, R. C. Nichol, R. Owen, J. R. Pier, A. Pope, C. M. Rockosi, D. J. Schlegel, W. A. Siegmund, S. Smee, Y. Snir, C. Stoughton, C. Stubbs, M. SubbaRao, A. S. Szalay, G. P. Szokoly, C. Tremonti, A. Uomoto, P. Waddell, B. Yanny, and W. Zheng. Composite Quasar Spectra from the Sloan Digital Sky Survey. AJ, 122:549–564, August 2001. doi: 10.1086/321167.

D. E. Vanden Berk, D. P. Schneider, G. T. Richards, P. B. Hall, M. A. Strauss, R. Brunner, X. Fan, I. K. Baldry, D. G. York, J. E. Gunn, R. C. Nichol, A. Meiksin, and J. Brinkmann. An Empirical Calibration of the Completeness of the SDSS Quasar Survey. AJ, 129:2047–2061, May 2005. doi: 10.1086/427856.

M. Vestergaard and B. M. Peterson. Determining Central Black Hole Masses in Distant Active Galaxies and Quasars. II. Improved Optical and UV Scaling Relationships. ApJ, 641:689–709, April 2006. doi: 10.1086/500572. K. L. Visnovsky, C. D. Impey, C. B. Foltz, P. C. Hewett, R. J. Weymann, and S. L. Morris. Radio properties of optically selected quasars. ApJ, 391:560–568, June 1992. doi: 10.1086/171370.

M. Volonteri, M. Sikora, and J.-P. Lasota. Black Hole Spin and Galactic Morphology. ApJ, 667: 704–713, October 2007. doi: 10.1086/521186. M. Volonteri, M. Sikora, J.-P. Lasota, and A. Merloni. The Evolution of Active Galactic Nuclei and their Spins. ApJ, 775:94, October 2013. doi: 10.1088/0004-637X/775/2/94.

M. Wals, B. J. Boyle, S. M. Croom, L. Miller, R. Smith, T. Shanks, and P. Outram. Radio-quiet objects in the 2QZ survey. MNRAS, 360:453–457, June 2005. doi: 10.1111/j.1365-2966.2005. 08926.x. C. A. Welling, B. P. Miller, W. N. Brandt, D. M. Capellupo, and R. R. Gibson. Broad Absorption Line Variability in Radio-Loud Quasars. ArXiv e-prints, March 2014.

Bibliography 77

R. L. White, R. H. Becker, M. D. Gregg, S. A. Laurent-Muehleisen, M. S. Brotherton, C. D. Impey, C. E. Petry, C. B. Foltz, F. H. Chaffee, G. T. Richards, W. R. Oegerle, D. J. Helfand, R. G. McMahon, and J. E. Cabanela. The FIRST Bright Quasar Survey. II. 60 Nights and 1200 Spectra Later. ApJS, 126:133–207, February 2000. doi: 10.1086/313300. R. L. White, D. J. Helfand, R. H. Becker, E. Glikman, and W. de Vries. Signals from the Noise: Image Stacking for Quasars in the FIRST Survey. ApJ, 654:99–114, January 2007. doi: 10.1086/507700. B. J. Wilkes. Studies of broad emission line profiles in QSO. I - Observed, high-resolution profiles. MNRAS, 207:73–98, March 1984. A. S. Wilson and E. J. M. Colbert. The difference between radio-loud and radio-quiet active galaxies. ApJ, 438:62–71, January 1995. doi: 10.1086/175054. L. Wisotzki. Quasar spectra and the K correction. A&A, 353:861–866, January 2000. J.-H. Woo and C. M. Urry. The Independence of Active Galactic Nucleus Black Hole Mass and Radio Loudness. ApJ, 581:L5–L7, December 2002. doi: 10.1086/345944.

G. Worseck and J. X. Prochaska. GALEX Far-ultraviolet Color Selection of UV-bright High-redshift Quasars. ApJ, 728:23, February 2011. doi: 10.1088/0004-637X/728/1/23. J. Wu, W. N. Brandt, S. F. Anderson, A. M. Diamond-Stanic, P. B. Hall, R. M. Plotkin, D. P. Schneider, and O. Shemmer. X-Ray and Multiwavelength Insights into the Nature of Weak Emission-line Quasars at Low Redshift. ApJ, 747:10, March 2012. doi: 10.1088/0004-637X/747/ 1/10. C. Xu, M. Livio, and S. Baum. Radio-loud and Radio-quiet Active Galactic Nuclei. AJ, 118: 1169–1176, September 1999. doi: 10.1086/301007. D. G. York, J. Adelman, J. E. Anderson, Jr., S. F. Anderson, J. Annis, N. A. Bahcall, J. A. Bakken, R. Barkhouser, S. Bastian, E. Berman, W. N. Boroski, S. Bracker, C. Briegel, J. W. Briggs, J. Brinkmann, R. Brunner, S. Burles, L. Carey, M. A. Carr, F. J. Castander, B. Chen, P. L. Colestock, A. J. Connolly, J. H. Crocker, I. Csabai, P. C. Czarapata, J. E. Davis, M. Doi, T. Dombeck, D. Eisenstein, N. Ellman, B. R. Elms, M. L. Evans, X. Fan, G. R. Federwitz, L. Fis- celli, S. Friedman, J. A. Frieman, M. Fukugita, B. Gillespie, J. E. Gunn, V. K. Gurbani, E. de Haas, M. Haldeman, F. H. Harris, J. Hayes, T. M. Heckman, G. S. Hennessy, R. B. Hindsley, S. Holm, D. J. Holmgren, C.-h. Huang, C. Hull, D. Husby, S.-I. Ichikawa, T. Ichikawa, Z.ˇ Ivezi´c,S. Kent, R. S. J. Kim, E. Kinney, M. Klaene, A. N. Kleinman, S. Kleinman, G. R. Knapp, J. Korienek, R. G. Kron, P. Z. Kunszt, D. Q. Lamb, B. Lee, R. F. Leger, S. Limmongkol, C. Lindenmeyer, D. C. Long, C. Loomis, J. Loveday, R. Lucinio, R. H. Lupton, B. MacKinnon, E. J. Mannery, P. M. Mantsch, B. Margon, P. McGehee, T. A. McKay, A. Meiksin, A. Merelli, D. G. Monet, J. A. Munn, V. K. Narayanan, T. Nash, E. Neilsen, R. Neswold, H. J. Newberg, R. C. Nichol, T. Nicinski, M. Nonino, N. Okada, S. Okamura, J. P. Ostriker, R. Owen, A. G. Pauls, J. Peoples, R. L. Peterson, D. Petravick, J. R. Pier, A. Pope, R. Pordes, A. Prosapio, R. Rechenmacher, T. R. Quinn, G. T. Richards, M. W. Richmond, C. H. Rivetta, C. M. Rockosi, K. Ruthmansdor- fer, D. Sandford, D. J. Schlegel, D. P. Schneider, M. Sekiguchi, G. Sergey, K. Shimasaku, W. A. Siegmund, S. Smee, J. A. Smith, S. Snedden, R. Stone, C. Stoughton, M. A. Strauss, C. Stubbs, M. SubbaRao, A. S. Szalay, I. Szapudi, G. P. Szokoly, A. R. Thakar, C. Tremonti, D. L. Tucker, A. Uomoto, D. Vanden Berk, M. S. Vogeley, P. Waddell, S.-i. Wang, M. Watanabe, D. H. Wein- berg, B. Yanny, N. Yasuda, and SDSS Collaboration. The Sloan Digital Sky Survey: Technical Summary. AJ, 120:1579–1587, September 2000. doi: 10.1086/301513. N. L. Zakamska and J. E. Greene. Quasar feedback and the origin of radio emission in radio-quiet quasars. ArXiv e-prints, February 2014.

Bibliography 78

Vita

Rachael M. Kratzer Drexel University, Physics Department, 3141 Chestnut St., Philadelphia, PA 19104 (610) 762-0385 • [email protected] Education Drexel University, Philadelphia, PA Ph.D (2014), Physics Dissertation: Mean and Extreme Radio Properties of Quasars and the Origin of Radio Emission Advisor: Dr. Gordon T. Richards

M.S. (2009), Physics

Cornell University, Ithaca, NY B.A. (2007), Astronomy magna cum laude Advisor: Dr. Joseph Veverka

Academic Employment Drexel University, Philadelphia, PA Research Assistant under Dr. Gordon T. Richards (June 2009 – December 2013)

Teaching Experience Drexel University, Philadelphia, PA Physics 100: Preparation for Engineering Studies, Teaching Assistant (Winter 2014) Physics 223: Modern Physics Laboratory, Lab Instructor (Fall 2013) Physics 101: Fundamentals of Physics I, Head Teaching Assistant (Spring 2009) Physics 201: Fundamentals of Physics III, Lab Instructor (Winter 2009) Physics 154: Introductory Physics III, Teaching Assistant & Lab Instructor (Fall 2008) Physics 102: Fundamentals of Physics II, Teaching Assistant (Spring 2008) Physics 101: Fundamentals of Physics I, Teaching Assistant (Winter 2008) Physics 232: Observational Astrophysics, Teaching Assistant (Fall 2007)

Awards Dean’s Fellowship, Drexel University (September 2007 – August 2009) Excellence in Teaching Award Nominee, Drexel University (2008 and 2009) Oﬃce of Graduate Studies Dissertation Fellowship, Drexel University (2014)

Publications Mean and Extreme Radio Properties of Quasars and the Origin of Radio Emission Kratzer, Rachael M., & Richards, G. T. 2014, ApJ, in preparation

Uniﬁcation of Luminous Type 1 Quasars through CIV Emission Richards, Gordon T., Kruczek, N. E., Gallagher, S. C., Hall, P. B., Hewett, P. C., Leighly, K. M., Deo, R. P., Kratzer, R. M., & Shen, Y. 2011, AJ, 141, 167

Analyzing the Flux Anomalies of the Large-Separation Lensed Quasar SDSS J1029+2623 Kratzer, Rachael M., Richards, G. T., Goldberg, D. M., Oguri, M., Kochanek, C. S., Hodge, J. A., Becker, R. H., & Inada, N. 2011, ApJ Letters, 728L, 18